Sie sind auf Seite 1von 233

Cardiovasc Intervent Radiol (2008) 31:14

DOI 10.1007/s00270-007-9231-2

EDITORIAL

Editors Recognition Award for Distinction in Reviewing 2007


Dierk Vorwerk

Received: 18 October 2007 / Accepted: 18 October 2007 / Published online: 22 November 2007
Springer Science+Business Media, LLC 2007

CVIR acknowledges and appreciates all of our contributing Reviewer Acknowledgments


reviewers for another successful year. In 2007, nearly 300
reviewers provided substantial scientific analysis of In order to maintain high scientific standards the Editors
manuscripts submitted for publication in CardioVascular frequently seek the opinions of outside experts. Reviewing
and Interventional Radiology. Several reviewers have scientific papers is a time-consuming and often complex
emerged as our Distinguished Reviewers. These indi- task that is vital to the success of CardioVascular and
viduals have submitted 20 reviews or more in the past year Interventional Radiology. In addition to our fine assembly
(maximum: 32 reviews), and have averaged 16 days or less of Editorial Board Members and Consultants, our Editors
turnaround time (minimum: 5 days). A certificate of rec- are especially grateful to the following individuals for their
ognition is presented to the following individuals. contributions and assistance during the past year.
Aghiad Al-Kutoubi Lebanon
Akira Adachi
Jose Ignacio Bilbao Spain
Andreas N. Adam (Editor Emeritus)
Elias N. Brountzos Greece
Devrim Akinci
Jan H. Gothlin Sweden
Heiko Alfke
Adam A. Hatzidakis Greece
Aghiad Al-Kutoubi
Michael J. Lee Ireland
Ahmad Alomari
Hannu I. Manninen Finland
Poul Erik Andersen
Robert Morgan United Kingdom
J. Fritz Angle
Anthony A. Nicholson United Kingdom
Takeshi Aramaki
Jim A. Reekers Netherlands
Johannes Aufenanger
Satoshi Sawada Japan
Carlos Baccin
Karl Schuermann Germany
Irene Bargellini
Scott O. Trerotola USA
Gabriel Bartal
Dimitrios K. Tsetis Greece
Gary Becker
A. F. Watkinson United Kingdom
Mitchell Benson
Robert Worthington-Kirsch USA
Jean-Paul Beregi
Thank you all, and congratulations for a job very well Jochim Berkefeld
done! Christoph Binkert
Haraldur Bjarnason
Amman Bolia
Lynn Brody
D. Vorwerk (&)
Elias N. Brountzos
Department of Radiology, Klinikum Ingolstadt,
Krumenauerstrasse 25, Ingolstadt D 85049, Germany Dan Brown
e-mail: dierk.vorwerk@klinikum-ingolstadt.de Arno Buecker

123
2 Acknowledgments

Filippo Cademartiri Krassi Ivancev


Mark Callaway Thomas Jahnke
Manfred Cejna Rolf Janka
Nick Chalmers Thorsten Johnson
Ajay Chavan Lowell Kabnick
Jian-ping Chu Tatsumi Kaji
Jin Wook Chung Ihab Kamel
Barbaros Erhan Cil Susumu Kanazawa
Trevor Cleveland Krishna Kandarpa
Sohail Contractor John A. Kaufman
Constantin Cope Alexis Kelekis
Frank Criado Robert Kerlan
Laura Crocetti David Kessel
Michael D. Darcy Joachim Kettenbach
C. J. Davidson Alexander Khandji
Andrew Davis Seung Hyup Kim
Theiny de Baere Milan Kinkhabwala
Christos Dervenis Thomas B. Kinney
Hannes A. Alexander Deutschmann K. Kling
D. D. Do Atsushi Komemushi
Gary Dorfman Anthony Krajina
Peter Drescher Hanno Krieger
B. Durgun Thomas J. Kroencke
Peter Ellis Gabriele Anja Krombach
Ruth E. England Gerrit Krupski-Berdien
Abigail Falk Yoshinobu Kubota
Gudrun Feuchtner Takashi Kumada
Scott Fink Toshiro Kuroiwa
E. E. E. Gabeler Ramazan Kutlu
Peter A. Gaines Hans-Ulrich Laasch
Jeff Geschwind Leo E. H. Lampmann
J. Gillard Peter Lauer
Y. Pierre Gobin Stephen H. Lee
Anthony Dhiantie Goei Byung-Hee Lee
Jafar Golzarian Riccardo Lencioni
Scott Goodwin X. A. Li
Roy L. Gordon G. Lock
Derek Alan Gould Christian Loewe
Alan J. Greenfield Romaric Loffroy
Johan Bastiaan Arie Groeneveld Paul N. M. Lohle
Rolf W. Guenther (Editor Emeritus) Lars Lonn
Patrick Haage Jorge E. Lopera
Neil Halin J. M. Lorenz
Thomas Helmberger Chaim Lotan
Norman Hertzer Sumaira Macdonald
Shozo Hirota Lindsay Machan
J. W. Hoe Martin Mack
Ralf-Thorsten Hoffmann Andreas H. Mahnken
J. Graeme Houston Geert Maleux
David Hunter Andrew J. Mannes
Peter E. Huppert Hannu I. Manninen
S. Imai Kiyoshi Matsueda
Yoshitaka Inaba Osamu Matsui

123
Acknowledgments 3

Naofumi Matsunaga Claudio Schonholz


Kevin McGraw Karl Schuermann
Gordon K. McLean Salvatore Sclafani
Ahmet Memis Hiroshi Seki
Steven Meranze Himanshu Shah
Phillip M. Meyers Sanjiv Sharma
Hidefumi Mimura Ji Hoon Shin
Hiromu Mori Richard Shlansky-Goldberg
Sojiro Morita Paul Singh Sidhu
David Lawson Morris Manrita Sidhu
Jonathan George Moss Yoel Y. Siegel
Stefan Muller-Hulsbeck Horst Sievert
K. Natsis James E. Silberzweig
Joerg Michael Neuerburg Maya Simionescu
Norihisa Nitta Gary Siskin
Kenji Ogawa H. Bob Smouse
Keigo Osuga Y. Soejima
Matthijs Oudkerk Constantinos Thasos Sofocleous
Jae Hyung Park Miyuki Sone
Byeong-Ho Park Hajun Song
Sang-Joon (John) Park Michael C. Soulen
Uday Patel James Spies
Dusan Pavcnik Gundula Staatz
Franco Perona S. William Stavropoulos
Theodore Petsas Christopher Sutton
Radoslaw Pietura Kaya Suzer
Isabel Pinto Pabon Josef Tacke
Vincenzo Pipitone Hiroyuki Tajima
Michael B. Pitton Motoichiro Takahashi
Jeffrey Pollak Masahide Takahashi
Martin R. Prince Haruyuki Takaki
Boris Alexis Radeleff Sumio Takashima
Parvati Ramchandani Yoshito Takeuchi
Thomas Rand Bien Soo Tan
Mahmood Razavi Toshihiro Tanaka
Goetz Martin Richter Ryoichi Tanaka
David Rickards Sumru-Tanju
Alain Roche Gaojun Teng
John D. G. Rose G. Tepe
Volker Ruppert Henrik S. Thomsen
Wael E. A. Saad Ken Thomson
Tarun Sabharwal David Trost
Shunichi Sadaoka Dimitrios K. Tsetis
Vikram Anik Singh Sahni Masakatsu Tsurusaki
Hiroshi Sakaguchi Luc A. E. Turmel-Rodrigues
Akira Sano Raman Uberoi
Morio Sato Michael Uder
Motohiro Sato Renan Uflacker
Yozo Sato Otto M. vanDelden
Baransel Saygi Suresh Vedantham
Peter Schlossberg Thomas Vesely
Thomas Schmitz-Rode Robert L. Vogelzang
Maria Schoder Ajay K. Wakhloo

123
4 Acknowledgments

Michael Wallace Koichiro Yamakado


Judah Weinberger F. Y. Yao
Robert I. White H. Yasuhara
Joachim E. Wildberger Guoqiang Yu
Robert Worthington-Kirsch Soheyla-Zahirifard
Takuji Yamagami Q. Zhang

123
Cardiovasc Intervent Radiol (2008) 31:57
DOI 10.1007/s00270-007-9196-1

INVITED COMMENTARY

Lessons from EVA-3S and SPACE


David Beckett Peter A. Gaines

Received: 5 July 2007 / Accepted: 29 August 2007 / Published online: 23 October 2007
Springer Science+Business Media, LLC 2007

Abstract Carotid endarterectomy (CEA) was established wanted to ensure that the difference in complications at 30
as the gold standard for treatment of carotid occlusive days did not exceed 2% (and in SPACE this was 2.5%).
disease by several landmark papers published in the 1990s.
With the continued trend toward less invasive therapies,
carotid artery stenting (CAS) has challenged CEA for EVA-3S
treatment of significant carotid artery disease. Several trials
have now compared CEA and CAS and a subsequent In this trial the investigators randomized recently symp-
Cochrane review indicated that the 30-day complication tomatic (120 days) patients with carotid stenosis of at least
rates were equivalent. Unfortunately, comparative long- 60% [1] and chose a primary end point of any stroke or
term data are still lacking. Two new trials comparing CAS death within 30 days after treatment. In order to demon-
with CAE in patients with symptomatic internal carotid strate that there was no difference between the two
artery disease have recently been published, but to what treatments (accepting a difference in complications at 30
extent have they usefully added to the available data? days of 2%), they intended to randomize 872 patients. In
fact, after only 527 patients were randomized, the differ-
ence was 5.7% in favor of CEA. The 95% confidence
interval (CI) around that 5.7% difference was 2.1% to
Noninferiority Trials 9.3%. This CI did not include the predefined acceptable
difference of 2% and therefore the trial was closed. At 6
Both the Endarterectomy versus Stenting in Patients with months, the incidence of any stroke or death was 6.1% after
Symptomatic Severe Carotid Stenosis (EVA-3S) trial [1] endarterectomy and 11.7% after stenting (p = 0.02).
and the Stent-Supported Percutaneous Angioplasty of the The natural conclusion from the trial was that CEA had
Carotid Artery versus Endarterectomy (SPACE) trial [2] better outcomes at 30 days and 6 months. But is the answer
were noninferiority trials. In such a trial the investigators that simple? Possibly not. Some insight into the hugely
aim to show that CAS is not worse than CEA. In order to different outcomes lies in the respective skills of the
do so a power calculation is undertaken to decide how operators undertaking CEA and CAS. To undertake CEA
many patients to include and then a decision is made as to the surgeon was required to have performed at least 25
what difference between the two treatments the investiga- endarterectomies before entering the trial. To undertake
tors will allow before they decide that in fact there is a CAS the surgeon was required to have little or no experi-
difference. For example, in EVA-3S the investigators ence. Not only was training severely limited but so was
experience. A total of 30 centers participated. Half the
work was done at five centers, and the other half at 25
centers. The average enrollment per center was therefore
D. Beckett  P. A. Gaines (&)
only 1.7 patients per year. This limitation in training and
Sheffield Vascular Institute, Northern General Hospital, Herries
Road, Sheffield S5 7AU, UK experience explains the high crossover from CAS to CEA
e-mail: p.a.gaines@sheffield.ac.uk and, probably, some of the high 30-day complication rate

123
6 D. Beckett and P. A. Gaines: Lessons from EVA-3S and SPACE

in the CAS arm. Some of the poor CAS outcomes may also addition, SPACE managed 79% of patients taking dual
be explained by the apparent appalling compliance with antiplatelets at the time of CAS.
adjuvant antiplatelet therapy. It is clear that dual anti- These excellent outcomes were managed even though
platelet therapy plays an absolutely major role in the safety the trial did not mandate the use of CPDs, and indeed only
of CAS [3, 4]. In EVA-3S, only 69% of patients were 27% of patients were treated this way. While the numbers
taking some form of antiplatelet therapy at the time of the are necessarily small, the post hoc analysis indicates that
procedure, and that figure fell to 38% after the procedure. CPDs did not confer any benefit.
Not surprisingly, then, the outcomes following CAS were Of some interest is the number of events that occurred
poor. after the procedure but before 30 days. Although figures are
It is difficult to be too clear about the assertion, fol- not given for the SPACE trial, an inspection of the Kaplan-
lowing a post hoc analysis, that stated cerebral protection Meier curves suggests that in fact most adverse events
devices (CPDs) were beneficial since the trial was tempo- following CAS (and CEA) occur after the day of treatment.
rarily halted during recruitment to make CPDs mandatory, In EVA-3S 7 of the 24 strokes that occurred after CAS
even though prior to the suspension there had been no occurred after the first day. Similar data have been pre-
significant difference between those treated with and those sented previously and clearly this is an area for
without the devices. investigation [5].
And what about the comparatorCEA? It appears that
the French surgeons have achieved a better outcome than
Lessons
any other trial conducted in the rest of the world (including
Germany, Austria, and Switzerland over the same time
So what have we learned? How have these trials furthered
period; see below).
our understanding of the use of CAS? and What areas
could be advanced?
SPACE 1. Trials should be powered correctly and should not be
stopped or modified early unless absolutely necessary.
In this trial the investigators randomized recently (180 This is hardly a new message.
days) symptomatic patients with a carotid stenosis of 2. As with any other intervention in any other walk of
[50% and chose a primary end point of ipsilateral stroke life, experience and training matter [6].
or death within 30 days of the procedure. The power 3. SPACE shows that when CAS is performed well, the
calculations by this group resulted in the intention to 30-day major outcomes are comparable with those of
include 1900 patients, accepting a difference in compli- CEA performed well. However, given the poor CAS
cation rate of 2.5% as demonstrating noninferiority. The results in EVA-3S, the question of how generalizable
trial should be stopped, therefore, if the lower limit of the this procedure is should be asked.
90% CIs is above 2.5%. The study was stopped after 1200 4. Given that the intention of both these interventions is
patients had been randomized. At 30 days the primary end to reduce the risk of stroke in the long term, it is clear
point was 6.84% for CAS and 6.34% for CEA. The dif- that we are still in need of good comparative long-term
ference was four events (0.51%) and was not significant. data.
The authors justified closure by indicating that the 90% 5. Major benefit could be obtained by reducing the late
CIs surrounding the difference of 0.51% were 1.89% to events that occur after stent placement. Whether this
2.91% and that the top limit lay outside the 2.5% dif- can be reduced by better adjuvant pharmaceuticals,
ference that they allowed. This means that the better stents, or better monitoring needs to be
predetermined difference of 2.5% was inside the 90% CIs. determined.
Only by recruiting the appropriate number of patients 6. In EVA-3S 30% of the strokes in the surgical group
would the CIs be narrowed and then lie above the 2.5% and 15% in the CAS group were due to intracranial
limit set by the investigators. It is also widely recognized haemorrhage. Again, attention needs to be directed to
that, because the interim analysis suggested that more this.
patients were required, continued funding was not 7. SPACE raises serious questions regarding the benefit
provided. that CPDs may confer.
So the 30-day outcomes were not different between 8. Perhaps the data emphasize the requirement for
CAS and CEA. SPACE recruited more experienced inter- optimal antiplatelet agents.
ventionalists than those participating in EVA-3S, and this 9. These data will further muddy the answer provided
resulted in no patients crossing over from CAS to CEA by a meta-analysis. More data are required to
because of technical inadequacy at the time of CAS. In determine how the two therapies can complement

123
D. Beckett and P. A. Gaines: Lessons from EVA-3S and SPACE 7

each other, justifying continued recruitment to the 3. Beinart SC, Kolm P, Veledar E, et al. (2005) Long-term cost
ICSS and CREST trials. Hopefully, then we will effectiveness of early and sustained dual oral antiplatelet therapy
with clopidogrel given for up to one year after percutaneous
have sufficient information to help us determine coronary intervention results: from the Clopidogrel for the
which technique, with which devices, is best for Reduction of Events During Observation (CREDO) trial. J Am
which patients. Coll Cardiol 46(5):761769
4. McKevitt FM, Randall MS, Cleveland TJ, Gaines PA, Tan KT,
Venables GS (2005) The benefits of combined anti-platelet
treatment in carotid artery stenting. Eur J Vasc Endovasc Surg
References 29(5):522527
5. Hauthe EAM, Drescher R, Jansen C, Gissler M, Schwartz M,
Forsting M, Jaeger HJ, Mathias KD (2006) Complications and
1. Mas JL, Chantellier G, Beyssen B, et al. (2006) Endarterectomy
follow-up after unprotected carotid artery stenting. CardioVasc
versus stenting in patients with symptomatic severe carotid
Interv Radiol 29(4):511518
stenosis. N Engl J Med 355(16):16601671
6. Verzini F, Cao P, De Rango P, et al. (2006) Appropriateness of
2. The SPACE Collaborative Group (2006) Thirty-day results from
learning curve for carotid artery stenting: an analysis of peripro-
the SPACE trial of stent-protected angioplasty versus carotid
cedural complications. J Vasc Surg 44(6):12051211, discussion
endarterectomy in symptomatic patients: a randomised non-
12111212 (Epub 20 October 2006)
inferiority trial. Lancet 368(9543):12391247

123
Springer Science+Business Media, LLC 2007 Cardiovasc Intervent Radiol (2009) 31:813
CardioVascular Published Online: 3 May 2007 DOI: 10.1007/s00270-005-0266-y
and Interventional
Radiology

REVIEW ARTICLE

GP IIb/IIIa Blockade During Peripheral Artery


Interventions
Gunnar Tepe,1 Jakub Wiskirchen,1 Philippe Pereira,1 Claus D. Claussen,1
Stephen Miller,1 Stephan H. Duda2
1
Department of Diagnostic Radiology, University of Tbingen, 72076, Tbingen, Germany
2
Center for Diagnostic Radiology & Minimally Invasive Therapy, The Jewish Hospital, Berlin, Germany

Abstract macological agents [4]. Although encouraging data from


single centers are available, prospective trials have to be
The activation of the platelet GP IIb/IIIa receptor is the final carried out to further define the role of GP IIb/IIIa blockade
and common pathway in platelet aggregation. By blocking in peripheral vascular disease (PVD).
this receptor, platelet aggregation can be inhibited inde- Platelet adhesion, activation, and aggregation occurring
pendently of the stimulus prompted the targeting of this at the site of arterial injury are mediated by local factors,
receptor. Several years ago, three drugs have been approved including thromboxane A2 inhibited by aspirin and adeno-
for coronary artery indications. Since that time, there is sine diphosphonate inhibited by ticlopidine and clopidogrel
increasing evidence that GP IIb/IIIa receptor blockade [5]. Antiplatelet therapy is used for prophylaxis of throm-
might have also an important role in peripheral arterial bembolic complications in acute and chronic periods after
intervention. This article summarizes the action and differ- endovascular procedures. Antiplatelet therapy occurs pri-
ences of GP Ilb/IIIa receptor inhibitors and its possible marily via inhibition of thromboxane A2, ADP release,
indication in peripheral arteries. or blockade of the GP IIb/IIIa receptors. The cascade
Key words: AtherosclerosisPeripheral arterial interven- of platelet activation and aggregation is summarized in
tionary drugEndovascularPlatelet Figure 1.

Recently, there has been increased interest in the role of GP IIb/IIIa Receptor
glycoprotein IIb/IIIa (GP IIb/IIIa) in peripheral arterial
interventions [14]. Thrombembolic and ischemic compli- Knowledge that activation of the platelet GP IIb/IIIa
cations are frequently observed during and after endovas- receptor is the final and common pathway in platelet
cular procedures. With increasing numbers of endovascular aggregation independently of the stimulus prompted the
procedures in peripheral vascular disease, knowledge of the targeting of this receptor and attempts to develop agents
pathophysiological principles that might be followed by with more pronounced antiplatelet effects than aspirin, tic-
early rethrombosis is mandatory. In this article, we will lopidine, or clopidogrel. This approach was supported by the
describe the mode of action of GP IIb/IIIa antagonists, de- results of the EPIC trial with the GP IIb/IIIa receptor
scribe the differences between the available agents, and inhibitor abciximab in high-risk patients undergoing coro-
highlight possible indications for peripheral arterial use. nary angioplasty [6, 7]. The EPIC trial showed reductions in
Because the use of adjunctive therapy with GP IIb/IIIa adverse events with the addition of abciximab to aspirin and
receptor blockade in noncoronary arterial interventions is heparin compared with aspirin and heparin alone. Now, two
new, the reported data of controlled randomized and pro- other GP IIb/IIIa receptor inhibitors, eptifibatide and tir-
spective studies are mainly limited to coronary use. The ofiban, have been approved for clinical use in treatment of
information extrapolated to noncoronary interventions can patients with acute coronary syndromes [4]. Although often
be used to define the side-effect profiles of various phar- mentioned together, there are important distinctions among
the three different agents that have been approved for
clinical practise. All three, however, exhibit rapid binding to
Correspondence to: Gunnar Tepe; email: Gunnar.Tepe@med.uni- platelets in less than 1 minute. Abciximab is a chimeric
tuebingen.de humanmurine monoclonal antibody fragment. Eptifibatide
G. Tepe et al.: Blockade During Peripheral Artery Interventions 9

gies must be used to reverse the activity of each of these


drugs. The short-acting GP IIb/IIIa inhibitors dissociate
from the GP IIb/IIIa receptor. However, return of platelet
function is dependent on the clearance of the drug from the
plasma. Therefore, reversing the effect of these drugs re-
quires cessation of drug infusion. The effects of platelet
transfusion is only very limited because the stoichiometry of
these drugs to the receptor cannot be easily overcome. In
contrast, platelet inhibition by abciximab cannot be rapidly
reversed by stopping the drug infusion, but it can be suc-
cessfully reversed by transfusion of platelets. The reason
that platelet infusions are effective for abciximab and not for
the other agents is that the binding of abciximab to the
GPIIb/IIIa receptor is irreversible, whereas in case of the
other smaller compounds, it is reversible. If a patient who
has already received abciximab gets platelet infusions, these
platelet will not be affected by the previous administration
of abciximab. In contrast to eptifibatide or tirofiban, these
Fig. 1. Summary of the platelet aggregation cascade. The compounds will dissociate from the GPIIb/IIIa receptor and
activation of the GP IIb/IIIa receptor is the last and common bind to the receptor from the platelets that have been in-
pathway before platelet aggregation. fused.
The small and large molecules are both given intrave-
nously starting with a bolus followed by iv infusion for 12
and tirofiban are small-molecule antagonists. However, up 36 h [3]. Due to the different pharmacokinetic profiles, the
to now, no trials have compared the different GP IIb/IIIa amount of total dose in percent as bolus administered with
inhibitors. Comparisons of the agents across the trials have abciximab is much greater than with the eptifibatide and
shown the overall benefit of this class of drugs, while tirofiban. Interestingly, in the coronary trials with the GP
indicating that results with abciximab might differ from IIb/IIIa inhibitors in more than 16,000 patients, the inci-
those of the small molecules. Abciximab [molecular weight dence of cerebral bleeding occurances was not increased.
(MW  50,000)] is approximately two orders of magnitude Most of the bleeding occurrences were seen at the site of
larger than either eptifibatide (MW  800) or tirofiban instrumentation. GP IIb/IIIa administration is associated
(MW  500). Furthermore, although the biologic target, with a low incidence of acute, profound thrombocytopenia
(i.e., blockade of the GP IIb/IIIa receptor) is the same for all defined as platelet count <50 109/L (0.21%). Despite the
of the agents, they differ in other respects. Among the development of profound thrombocytopenia within 1121 h
important differences are receptor selectivity, dissociation of administration, the incidence of hemorrhagic complica-
kinetics, stoichiometry for inhibition, and clearance mech- tion was not significantly increased. Platelet counts re-
anisms. Whereas eptifibatide and tirofiban reacts selectively mained depressed for at least 3 days and returned to baseline
with the GP IIb/IIIa receptor, abciximab binds to the GP IIb/ within 2 weeks. The mechanism of this finding is specula-
IIIa receptor and the vitronectin receptor with equal affinity tive and unknown. From the standpoint of safety, the above
and weaker binding to the Mac-1 receptor. Abciximab is a considerations are paramount. The data favor abciximab
very large molecule and, therefore, it reacts with the GP IIb/ with regard to clinical outcomes in the setting of coronary
IIIa receptor like an umbrella, covering the receptor. Ep- intervention. Because eptifibatide and tirofiban bind only
tifibatide and tirofiban are classical competitive antagonists. transiently to GP IIb/IIIa, a rather hypothetical benefit
The stoichiometry (molecules of drug per GP IIb/IIIa seems to be the overall safety profile. On the other hand, the
receptor) is approximately 100:1 for both agents, with half- major disadvantage of the small molecules might be a de-
lives of 1.52.5 h. In contrast, the stoiechiomety of abcix- creased efficacy, which could be a result of the shorter
imab to the GP IIb/IIIa receptor is approximately 1.5 mol- duration of platelet inhibition. The possible beneficial role
ecules of drug to each receptor. However, in contrast to the of an additional blockade of the vitronectin and the Mac-1
small-molecule inhibitors, there is a slow decrease of receptor with abciximab is still not clearly defined. The
receptor blockade after discontinuation of abciximab treat- pharmacpdynamic characteristics of the different GP IIb/IIIa
ment. Twenty-four hours after ceasing abciximab infusion, receptor antagonists are summarized in Table 1.
there is still 5060% receptor blockade. It has been shown
that approximately 80% of the GP IIb/IIIa receptors on the
surface of the platelets have to be blocked in order to
With Thrombolytics in PVD
achieve a therapeutic platelet aggregation inhibition. On the The GP IIb/IIIa receptor inhibitors are less associated with
basis of the pharmacodynamic differences, different strate- the increased risk of intracranial bleeds (or bleeding in
10 G. Tepe et al.: Blockade During Peripheral Artery Interventions

Table 1. Pharmacodynamics of GPIIb/IIIa receptor inhibitors red blood cells, resulting in an occluding thrombus. The
Abciximab Eptifibatide Tirofiban dissolution of a thrombus formation with lytics is not simply
(Reopro) (Integrilin) (Aggrastat) a progressive process detaching platelets from the clot. It is
Molecule Antibody Peptide Nonpeptide
much more a dynamic process, with formation of a new
Molecular weight (D) 50,000 800 500 thrombus as the older thrombus is dissolved. Inhibition of
Specificity/selectivity new thrombus formation by antiplatelet agents might shift
IIb/IIIa +++ +++ +++
Vitronectin +++
the process to quicker and more effective thrombus disso-
Mac-1 + lution [13, 14].
Stoichiometry molecule/GPIIb/IIIa 11.5:1 100:1 100:1 Second, thrombolytic therapy is associated with platelet
receptor
Reversibility without 12h 24h b
4hb
activation. Administration of thrombolytic agents and sub-
plateletsa sequent dissolution of the thrombus releases thrombin from
Reversibility with platelets Yes No No the clot. Thrombin is a very potent agent for platelet acti-
a
50% return of platelet function. vation, working by direct activation of the GP IIb/IIIa
b
Assumption of normal renal function. receptor. One explanation for the phenomenon of recurrent
thrombosis during initially successful thrombolytic therapy
might be related to this process [4]. Blocking the GP IIb/IIIa
general) than seen with other fibrinolytic agents. In the case receptor is the logical consequence for inhibiting this pro-
of a vessel wall injury, a platelet monolayer forms at the cess. The use of antiplatelet agents might decrease the rate
injury sitethe beginning of hemostasis. The binding is of recurrent thrombosis by limiting the accumulation of new
mediated by GP Ib/IIa receptors and not blocked by platelets into a dissolving thrombus. This had already been
GP IIb/IIIa inhibitors. However, the complete explanation shown by aspirin, which reduced the postintervention re-
for reduced intracranial hemorrhage rate with GP IIb/IIIa thrombosis, but the GP IIb/IIIa inhibition provided a much
inhibitors compared with lytic agents is not fully under- greater degree of platelet inhibition. This is due to the fact
stood. There is a small danger of profound risk of throm- that several ways of platelet activation exist. Inhibition of
bocytopenia with the use of GP IIb/IIIa receptor inhibitors for example, the prostaglandin or the ADP pathway by ei-
[1, 8]. ther aspirin or clopidogrel cannot be as effective compared
Thrombi that form under conditions of slow flow and low to the inhibition of the common final pathway of platelet
shear are composed of mainly red blood cells, and fibrin can aggregation: the GP IIb/IIIa receptor activation.
be assumed to be blood sludge. Conversely, arterial thrombi Third, the phenomenon of dethrombosis, the dissolu-
that form under high shear conditions are made up of tion of thrombus by GP IIb/IIIa inhibitors without con-
platelet aggregates bound together by glycoprotein receptors comitant plasminogen activators, further increased interest
and fibrin strands [9]. in peripheral arterial application of these agents [4].
Thrombolytic therapy has gained acceptance as a mini- The GP IIb/IIIa receptor antagonists had a demonstrable
mally invasive means to clear thrombus from peripheral clinical benefit in the setting of percutaneous coronary ar-
arteries. Catheter-directed thrombolysis has offered an op- tery intervention. In clinical trials of acute myocardial
tion for those individuals with medical comorbidities severe infarction, combination of selected fibrinolytic agents at a
enough that an open surgical revascularization procedure reduced dose with the potent GP IIb/IIIa agent abciximab
would be poorly tolerated. Although the use of catheter- resulted in improved antegrade blood flow in the infarct-
directed thrombolysis has been demonstrated to reduce related artery [7, 1518].
mortality in selected patient subgroups, its use has been Although peripheral arterial occlusions differ from cor-
criticized, citing an unacceptable rate of early thrombotic onary occlusion with respect to the size and age of the
reocclusion. Moreover, present techniques often require occluding thrombus, the success of GP IIb/IIIa inhibition in
protracted periods of administration to achieve satisfactory the setting of acute coronary occlusion might translate to the
recanalization [1012]. peripheral vessels. This background inspired interest in the
Understanding the mechanisms of arterial thrombus for- use of a systemically administered platelet glycoprotein IIb/
mation provides a basis on which to address the shortcom- IIIa inhibitor to augment catheter-directed peripheral arterial
ings of fibrinolytic therapy. There are several reasons for the thrombolysis [19].
use of GP IIb/IIIa receptor antagonists together with fibri- Although there are extensive data for all three GP IIb/IIIa
nolytics. First, the high shear rate causes platelet activation, receptor antagonists available for coronary indication, in
degradation, and release of substances that enhance platelet peripheral artery disease the literature is mainly limited to
activation of surrounding platelets. Active GP IIb/IIIa the use of abciximab. The first use of concomitant throm-
molecules at the surface of platelets serve as an anchor for bolysis and GP IIb/IIIa inhibition was undertaken by the
fibrinogen bridges between the platelets, and platelet Tuebingen group, which investigated the use of abciximab
platelet attachments result in the clumping process known as in conjunction with urokinase for peripheral arterial occlu-
platelet aggregation. Platelet attachment to the vessel wall sion, initially with a group of 14 patients in order to gauge
forms a carpet for deposition of more platelets, fibrin, and safety [19] and later in a randomized, blinded study of 70
G. Tepe et al.: Blockade During Peripheral Artery Interventions 11

patients treated with urokinase thrombolysis with and treatment, the findings of the study supported those of ours
without abciximab [20]. Although the success rate of des- as well as the RELAX trial. Improved outcome was more
obliteration remained unchanged between the groups, the common in the reteplase-plus-abciximab group.
speed of thrombolyisis could be enhanced by adjunctive GP Only recently, we completed the APART study com-
IIb/IIIa blockade. Interestingly, the frequency of major paring reteplase-plus-abciximab and urokinase-plus-abcix-
amputation or open surgery was reduced when abciximab imab in a total of 120 patients with acute peripheral artery
was added to thrombolysis in patients with peripheral arte- occlusion. No major differences between the thrombolytics
rial occlusion. In addition, the results were achieved without were found. Both regimens were safe and the incidence of
a statistically significant increase in bleeding. bleeding complications was much lower than reported in
A smaller, 15-patient registry was published by Drescher other large trials (TOPAS, STILLE) [22] that did not in-
et al. [3]. Whereas the results of the Tuebingen group were clude GP IIb/IIIa antagonists. As in all of the other pro-
obtained with pulsed spray lysis, limiting both the total spective trials with abciximab, the amputation rate was very
amount of administered urokinase and the treatment time, low. Together with the findings during the RELAX trial,
Drescher et al. used catheter-directed reteplase and sys- which showed a significant lower distal embolization rate in
temically administered abciximab in patients with lower the reteplase-plus-abciximab group compared to the retep-
extremity arterial thrombosis and reported no major bleed- lase-alone group, one might speculate that the favorable
ing complications. Complete thrombolsis and limb salvage outcome of patients toward less amputation is due to the
was achieved in 14 of 15 patients [3]. protection from severe embolization during thrombosis
In addition to abciximab, the small peptide eptifibatide was achieved by abciximab.
also used with thrombolytics in acute peripheral artery
occlusion. Yoon and Miller retrospectively reviewed their
experience in 17 patients who also received recombinant
With PTA in Chronic Lesions in PVD
tissue plasminogen activator (rt-PA) and heparin [2]. Four of The theory behind the use of GP IIb/IIIa antagonists during
the 17 patients received their loading dose of eptifibatide by peripheral arterial interventions in chronic lesions is to
direct intraarterial injection, whereas the remaining 13 re- prevent early thrombosis in lesions at high risk for such an
ceived an iv dose. The results were compared with those of 11 occurrence. Such high-risk revascularization procedures are
other patients who received only rt-PA and heparin with re- the treatment of TASC D lesions in the superficial femoral
spect to success, complication rates, and total rt-PA dose. No artery (SFA) and the popliteal artery as well as below-the-
significant differences were found in successful outcome, knee interventions.
major complication, duration of therapy, or total rt-PA be- The use of abciximab in TASC A, B, and C lesions in the
tween those who received eptifibatide and those who did not. SFA, together with stent administration (BLASTER trial,
However, those patients who received an intra-arterial load- presented during TCT 2003 by Gary M Ansel) failed to reduce
ing dose of eptifibatide required substantially less rt-PA to the clinical outcome compared to patients without GP IIb/IIIa
achieve successful thrombolysis. The small size of the study blockade because the clinical end point was mainly driven by
and the retrospective design limit the value of the study to that the restenosis rate. According to the effect of GP inhibition on
of a feasibility trial of eptifibatide and rt-PA. restenosis, these data are in line with results obtained by
A multicenter prospective comparison of reteplase alone several clinical trials in the coronary arteries. We are currently
versus reteplase with abciximab, the RELAX trial, was or- undertaking a multicenter clinical trial (RIO trial [ReoPro and
ganized to evaluate the combination therapy in a larger peripheral Arterial Intervention to Improve Clinical Outcome
number of patients [1]. Due to the fact that this trial was in Patients with Peripheral Artery Disease]) enrolling 420
carried out for safety issues, the Food and Drug Adminis- consecutive patients with TASC D lesions who are at high risk
tration suggested an escalating dose study design beginning for early rethrombosis. The patients randomly receive either
with 0.1 U/h reteplase with and without abciximab. If the abciximab or placebo. Early reocclusion as well as long-term
dose of reteplase was found to be safe, the dose was doubled patency is measured by ancle brachial index (ABI) and duplex
with 1 U/h reteplase (alone and with abciximab) at the end ultrasound after 30 days, 6 months and 1 year. Patients are
of the study. In total, 74 subjects were enrolled in this trial. scheduled to be clinically followed up for 3 years. Currently,
There were no statistically significant differences with re- 320 patients have already been enrolled. The occlusion length
spect to clinical end points such as the duration of throm- ranged between 5 and 60 cm. Most lesions were recanalized
bolytic agent administration, the frequency of open surgery, subintimally. Randomization followed by study drug or pla-
or survival. However, abciximab reduced the incidence of cebo infusion and percutaneous transluminal angioplasty
distal embolization during thrombolysis from 31% to 5% (PTA) was done after successful reentry. Stents had to be used
(p = 0.01). The study of Hull et al. also evaluated reteplase due to unsatisfactory results by the PTA in less than 50% of
with or without abciximab for the same indication in a total the patients. The enrollment is planned to be completed in
of 36 patients [21]. Although not randomized or blinded and early 2006.
complicated by the inclusion of eight patients who received Possible further indications for sufficient platelet inhi-
abciximab after more than 1 day of ineffective reteplase bition are interventions in below-the-knee arteries in pa-
12 G. Tepe et al.: Blockade During Peripheral Artery Interventions

Fig. 2. An 82-year-old female


patient who was admitted to the
hospital with CLI. A nonhealing ulcer
was present for 7 months and carried
a high risk for an amputation. On the
basis of the first angiogram (a), it was
decided to perform a PTA in a high-
grade stenosis in the P III segment.
After the dilatation, an increasing
number of collateral vessels and parts
of the anterial tibial artery could be
seen (b). At that stage, it was decided
to proceed with the intervention in
order to recanalize one below-the-
knee artery. Because the risk of
periprocedural thrombus formation
due to the limited outflow situation
was thought to be high, a bolus of
abciximab was administered. Finally,
the intervention could be successfully
completed (c, d). The ulcer healed
within the next 4 months.

tients with critical limb ischemia (CLI). CLI is often caused Subacute Occlusion in Arteries Below the Knee) investi-
by flow-limiting stenosis or occlusions of a remaining gating the outcome of patients with CLI who present with
below-knee artery. Patients with CLI have a dismal clinical nonhealing ulcerations due to vascular reasons. The inter-
course. During the first year of diagnosis, mortality is ventional procedures will be performed with and without
2030%, and an equivalent number undergo amputations or abciximab. The results will be available in mid-2006.
suffer persistent pain. The surgical options of either ampu-
tation or tibial bypass have changed little since 1966 [23].
Surgical complications include death (1.36%), leg edema,
Summary
and early graft failure rates (649%) requiring repeat sur- Summarizing the studies with GP IIb/IIIa inhibitors in acute
gery or amputation. Angioplasty for CLI was first reported peripheral artery disease, the combination of thrombolytic
nearly 40 years ago [24]. The initial complications were therapy and GP IIb/IIIa platelet inhibitors suggests that the
high and the success rates low. Since that time, special addition of platelet inhibitors offers efficacy benefits with-
devices (both balloon catheters and stents) have been cre- out a significant increase of bleeding risks. The failure to
ated for the use in small arteries with diameters ranging detect both significant clinical benefit and bleeding com-
between 2.5 mm and 4 mm. Nevertheless, complications by plications might have been due to the small sample size of
thrombotic occlusions during the interventions are still each trial. Nevertheless, the current data give level II evi-
feared. Figure 2 shows an example of the use of abciximab dence to support the benefit of GP IIb/IIIa inhibition.
during a below-the-knee intervention at our institution. Even Accordingly, the use of GP Ilb/IIIa inhibitors in high-risk
the vessel size is comparable to coronary arteries, and blood lesions in the SFA and below-the-knee data are encourag-
flow characteristics are limited to the poor runoff situation. ing, but it is still too early to draw any conclusion.
In such a high-risk scenario, Feiring et al. [25] were the first Until data from larger prospective clinical trials are
to describe the use of primary stent-supported angioplasty available, adjunct GP IIb/IIIa inhibitors should be used in
together with GP IIb/IIIa blockade. Unfortunately, they used selected patients with a higher risk of failure during
different stent types and a different GP IIb/IIIa antagonist. thrombolysis or rethrombosis during time-consuming and
Nevertheless, it could be shown that the intervention was technical complicated vascular interventions.
safe. No subacute thrombotic effects and no bleeding
complications were observed. Relief of rest pain and healing
References
of ulcerations and amputations were seen in 96% of patients 1. Ouriel K, Castaneda F, McNamara T, et al. (2004) Reteplase mono-
with CLI who underwent successful intervention. We are therapy and reteplase/abciximab combination therapy in peripheral
currently enrolling patients in the BELOW trial (Balloo- arterial occlusive disease: Results from the RELAX trial. J Vase In-
tervent Radiol 15(3):229238
nangioplastie or Stents with ReoPro for Prevention of
G. Tepe et al.: Blockade During Peripheral Artery Interventions 13

2. Yoon HC, Miller FJ (2002) Using a peptide inhibitor of the glyco- improved ST-segment resolution after thrombolysis: Observations from
protein IIb/IIIa platelet receptor: Initial experience in patients with the TIMI 14 Trial. Am Heart 141(4):592598
acute peripheral arterial occlusions. Am J Roentgentl 178(3):617622 14. Swischuk JL, Fox PF, Young K, et al. (2001) Transcatheter intraarterial
3. Drescher P, McGuckin J, Rilling WS, et al. (2003) Catheter-directed infusion of rt-PA for acute lower limb ischemia: Results and compli-
thrombolytic therapy in peripheral artery occlusions: Combining re- cations. J Vasc Intervent Radiol 12(4):423430
teplase and abciximab. Am J Roentgenol 180(5):13851391 15. Lincoff AM (2000) Potent complementary clinical benefit of abcix-
4. Tepe G, Hahn U, Pusich B, et al. (2001) New strategies in platelet imab and stenting during percutaneous coronary revascularization in
inhibition in noncoronary interventions. Semin Vasc Surg 14(2):143 patients with diabetes mellitus: Results of the EPISTENT trial. Am
149 Heart J 139(2 Pt 2):S46S52
5. Ouriel K (2002) Current status of thrombolysis for peripheral arterial 16. ERASER Study Group (1999) Acute platelet inhibition with abciximab
occlusive disease. Ann Vasc Surg 16(6):797804 does not reduce in-stent restenosis (ERASER study). The ERASER
6. Cho L, Topol EJ, Balog C, et al. (2000) Clinical benefit of glyco- Investigators. Circulation 100(8):799806
protein IIb/IIIa blockade with abciximab is independent of gender: 17. Sacks D, Rundback JH, Martin LG (2000) Renal angioplasty/stent
Pooled analysis from EPIC, EPILOG and EPISTENT trials. Evalu- placement and hypertension in the year 2000. J Vasc Intervent Radiol
ation of 7E3 for the Prevention of Ischemic Complications. Evalu- 11(8):949953
ation in Percutaneous Transluminal Coronary Angioplasty to 18. Marso SP, Lincoff AM, Ellis SG, et al. (1999) Optimizing the percu-
Improve Long-Term Outcome with Abciximab GP IIb/IIIa blockade. taneous interventional outcomes for patients with diabetes mellitus:
Evaluation of Platelet IIb/IIIa Inhibitor for Stent. J Am Coll Cardiol Results of the EPISTENT (Evaluation of platelet IIb/IIIa inhibitor for
36(2):381386 stenting trial) diabetic substudy. Circulation 100(25):24772484
7. Lincoff AM, Tcheng JE, Califf RM, et al. (1999) Sustained suppression 19. Tepe G, SchottU, Erley CM, et al. (1999) Platelet glycoprotein IIb/IIIa
of ischemic complications of coronary intervention by platelet GP IIb/ receptor antagonist used in conjunction with thrombolysis for periph-
IIIa blockade with abciximab: One-year outcome in the EPILOG trial. eral arterial thrombosis. Am J Roentgenol 172(5):13431346
Evaluation in PTCA to Improve Long-term Outcome with abciximab 20. Duda SH, Tepe G, Luz O, et al. (2001) Peripheral artery occlusion:
GP IIb/IIIa blockade. Circulation 99(15):19511958 Treatment with abciximab plus urokinase versus with urokinase
8. Moliterno DJ, Topol EJ (1997) Conjunctive use of platelet glycoprotein aloneA randomized pilot trial (the PROMPT Study). Platelet
IIb/IIIa antagonists and thrombolytic therapy for acute myocardial Receptor Antibodies in Order to Manage Peripheral Artery Thrombo-
infarction. Thromb Haemost 78(1):214219 sis. Radiology 221(3):689696
9. Connors JJ (2004) Pharmacologic agents in stroke prevention, acute 21. Hull JE, Hull MK, Urso JA (2004) Reteplase with or without abcix-
stroke therapy, and interventional procedures. J Vase Intervent Radiol iamb for peripheral arterial occulsions: Efficacy and adverse events. J
15:87101 Vasc Intervent Radiol 15(6):557564
10. Herrmann HC, Moliterno DJ, Ohman EM, et al. (2000) Facilitation 22. Ouriel K, Veith FJ, Sasahara AA (1998) A comparison of recombinant
of early percutaneous coronary intervention after reteplase with or urokinase with vascular surgery as initial treatment for acute arterial
without abciximab in acute myocardial infarction: Results from occlusion of the legs. Thrombolysis or Peripheral Arterial Surgery
the SPEED (GUSTO-4 Pilot) Trial. J Am Coll Cardiol 36(5):1489 (TOPAS) Investigators. N Engl J Med 338(16):11051111
1496 23. Hall KV (1962) The great saphenous vein used in-situ as an arterial
11. Ouriel K, Kandarpa K, Schuerr DM, et al. (1999) Prourokinase shunt after extirpation of vein valves. Surgery 51:492499
versus urokinase for recanalization of peripheral occlusions, safety 24. Rutherford JB (2000) Endovascular interventions in the management of
and efficacy: The PURPOSE trial. J Vase Intervent Radiol chronic lower extremity ischemia. In: JB (ed.) Vascular Surgery 2000;
10(8):10831091 5th ed. WB Saunders, Philadelphia, pp. 10581059
12. Valji K (2000) Evolving strategies for thrombolytic therapy of 25. Feiring A, Wesolowski AA, Lade S (2004) Primary stent-supported
peripheral vascular occlusion. J Vase Intervent Radiol 11(4):411420 angioplasty for treatment of below-knee critcal limb ischemia and se-
13. de Lemos JA, Gibson CM, Antman EM, et al. (2001) Abciximab and vere claudication: early and one-year outcomes. J Am Coll Cardiol
early adjunctive percutaneous coronary intervention are associated with 44(12):23072314
Cardiovasc Intervent Radiol (2008) 31:1422
DOI 10.1007/s00270-007-9211-6

INVITED REVIEW

Distal Embolic Protection for Renal Arterial Interventions


Gregory J. Dubel Timothy P. Murphy

Received: 20 May 2007 / Accepted: 11 September 2007 / Published online: 8 November 2007
Springer Science+Business Media, LLC 2007

Abstract Distal or embolic protection has intuitive DPDs when anatomically feasible. The data available at
appeal for its potential to prevent embolization of materials this time support a beneficial role for these devices.
generated during interventional procedures. Distal protec-
tion devices (DPDs) have been most widely used in the Keywords Embolic protection  Distal protection device 
coronary and carotid vascular beds, where they have Renal artery stenosis  Stent  Renal angioplasty 
demonstrated the ability to trap embolic materials and, in Atheroembolization
some cases, to reduce complications. Given the frequency
of chronic kidney disease in patients with renal artery
stenosis undergoing stent placement, it is reasonable to Introduction
propose that these devices may play an important role in
limiting distal embolization in the renal vasculature. Embolic protection has intuitive appeal for its potential to
Careful review of the literature reveals that atheroemboli- prevent embolization of materials (thrombus, cholesterol,
zation does occur during renal arterial interventions, plaque) generated during interventional procedures from
although it often goes undetected. Early experience with reaching and obstructing distal arterioles. Distal or embolic
DPDs in the renal arteries in patients with suitable anatomy protection devices (DPDs) have undergone rapid evolution
suggests retrieval of embolic materials in approximately and proliferation over the past 10 years, and eight are now
71% of cases and renal functional improvement/stabiliza- approved by the U.S. Food and Drug Administration
tion in 98% of cases. The combination of platelet inhibition (FDA) for marketing in the United States. They have been
and a DPD may provide even greater benefit. Given the most widely used in the coronary and carotid vascular beds,
critical importance of renal functional preservation, it fol- where they have demonstrated the ability to trap embolic
lows that everything that can be done to prevent materials and, in some cases, to reduce complications.
atheroembolism should be undertaken including the use of Given the frequency of chronic kidney disease in patients
with renal artery stenosis (RAS) undergoing stent place-
ment, it is reasonable to propose that these devices may
Dr. Murphy has support as follows: supported by grants HL071556-01
and HL077221 from the National Heart, Lung, & Blood Institute, and
play an important role in limiting distal embolization in the
also received research grants from Boston Scientific, Guidant, Cordis, renal vasculature. However, the case favoring the use of
and Otsuka Pharmaceuticals. embolic or DPDs for routine use in renal artery interven-
tions (RAIs) including percutaneous renal angioplasty and
G. J. Dubel  T. P. Murphy
percutaneous renal artery stenting at this time remains
Department of Diagnostic Imaging, Division of Interventional
Radiology, Brown University Medical School, Providence, RI largely circumstantial due to limited experience with such
02903, USA devices in the renal vasculature. Current clinical evidence
on DPDs for RAI presented here is derived from nonran-
G. J. Dubel (&)
domized cohort studies and one small randomized clinical
RI Hospital, 593 Eddy Street, 3rd Floor Main Building,
Providence, RI 02903, USA trial. Based on these limited data, current limitations of
e-mail: gdubel@lifespan.org DPD technology, and the principle of first, do no harm,

123
G. J. Dubel and T. P. Murphy 15

one could argue that RAI without the use of a DPD should then opened or deployed such that they oppose the wall of
be the standard of care and that DPDs should be used for the vessel but allow antegrade flow during the intervention.
highly selected cases, if at all. However, it may be argued Crossing profiles range from 2.9 to 3.9 Fr and pore size
that the overall benefit of RAI on renal function has been varies between 80 and 209 lm. Upon completion of the
somewhat disappointing. Metanalyses of stent series report procedure the filter is recaptured or collapsed using a
improvement in renal function in 26% [1] and 30% [2] of compatible catheter. The Accunet (Abbott Vascular,
patients, with decline in function in a similar number of Abbott Park, IL), FilterWire (Boston Scientific, Natick,
patients, suggesting that there may be room for improve- MA), Spider/SpideRx (eV3, Minneapolis, MN), Embo-
ment with regard to renal functional outcomes that can be shield (Mednova/Abbott, Abbott Park, IL), and Angioguard
achieved with RAI. Examining each of these topics and the (Cordis, Miami, FL) are examples of filter devices and all
data on DPDs in the renal arteries, one can make a strong carry FDA indication for carotid interventions when used
argument for the implementation of DPDs for RAI in with the appropriate stent. The GuardWire (Medtronic
treatment of patients with RAS when anatomy will allow AVE, Santa Rosa, CA) (Fig. 2) and TriActive FX (Kensey
placement of such devices. A careful review of the litera- Nash, Exton, PA) are distal balloon occlusion systems with
ture suggests that both micro- and macroembolization does FDA indication for use in coronary SVG interventions [3,
occur during RAI and that the effects on the kidney func- 6, 7], and the GuardDog (Possis Medical, Inc., Minneap-
tion could be profound. Furthermore, there is evidence to olis, MN) distal occlusion device is approved for peripheral
support the beneficial role of DPDs in both coronary use. These devices consist of a balloon on a guidewire that
saphenous vein grafts (SVGs) and the carotid arteries in is advanced beyond the lesion and then inflated, ideally
limiting embolization during percutaneous intervention [3 resulting in complete cessation of flow. At the termination
5], both territories, like the renal, where preservation of all of the case the stagnant column of blood is aspirated (i.e.,
parenchyma is highly desirable. through the Export catheter [Medtronic Inc.]) or flushed in
an effort to remove all debris including soluble mediators
that might pass through filter-type devices. Disadvantages
Distal Protection Devices: Designs of occlusive systems, however, are the poor visualization
of the lesion during intervention and ischemia during bal-
There are currently three principal types of DPDs used: loon inflation. Proximal occlusion devices such as the
filter or basket devices, distal balloon occlusion devices, Proxis (St. Jude Medical, Minneapolis, MN) attempt to
and proximal balloon occlusion devices. Filter devices prevent distal embolization by the use of occlusion bal-
consist of a porous windsock mounted on a nitinol loop loons mounted on guiding catheters that are placed
that is attached to or compatible with a guidewire (Fig. 1a). proximal to the lesion. The occluding balloon is deployed
Filter devices must be advanced past the stenotic lesion and in the vessel preceding the stenosis and inflated with the
hope of stopping antegrade flow or achieving flow reversal
without ever crossing the lesion. This would avoid embo-
lization during lesion traversal, which may occur with
either filter or distal occlusion balloon DPD designs. The

Fig. 1 Angioguard filter-type distal protection device. (Photo cour- Fig. 2 GuardWire balloon occlusion system. (Photo courtesy of
tesy of Cordis Endovascular, Warren, NJ.) Medtronic/AVE, Minneapolis, MN.)

123
16 G. J. Dubel and T. P. Murphy

devices are currently limited by their large size and are outcomes are dependent on the dose of microspheres
undergoing further study but are unlikely to prove useful in administered [15]. Only large doses produced proteinuria
the renal arteries, as most stenoses are located close to the and only the largest doses produced acute declines in renal
renal ostium [8]. function. Furthermore, it was shown that the maximum
effect was not seen until 12 weeks [15]. This is in keeping
with the insidiously progressive form of atheroemboliza-
Atheroembolization: Effect on the Kidney tion rather than the acute catastrophic form. It is highly
likely, therefore, that atheroembolization occurs much
Atheroembolization, or cholesterol embolization as it is more often than is recognized clinically and that it has
sometimes called, refers to the release of cholesterol plaque long-lasting effects on renal function.
fragments and crystals dislodged from proximal athero- The true incidence of atheroembolization during RAI
sclerotic plaques or platelet-rich thrombi embolized cannot be known, and very few studies have addressed the
downstream resulting in occlusion of arteries or arterioles. problem. Boisclair et al. reported on renal artery stenting in
The first pathological description of the disease was in 33 patients. Seven patients developed complications,
1945 by Flory [9], who demonstrated occlusion of vessels including four with renal artery emboli [16]. van de Ven
between 55 and 900 lm. Cholesterol crystal emboli cause et al. described two cases of renal insufficiency due to
local mechanical obstruction and induce a significant cholesterol embolism in 24 patients successfully stented for
inflammatory reaction that contributes to tissue ischemia atherosclerotic ostial RAS [17]. Wilms et al. described 2
and necrosis and may lead to irreversible renal functional cases of renal functional decline, including a case of
loss. Although cholesterol crystal embolization can occur massive cholesterol embolism, among 11 patients treated
spontaneously, it is increasingly recognized as an iatro- with renal artery stent placment. Unlike the carotid artery,
genic complication from invasive vascular procedures, where transcranial Doppler can be used to monitor the
such as manipulation of the aorta during angiography or cerebral circulation during an intervention for emboli [18,
vascular surgery, and after anticoagulant and fibrinolytic 19], no such technique is available for monitoring the
therapy [10]. It has been shown that atheroemboli released kidney during RAI. However, an interesting study per-
during revascularization are associated with decreased formed by Hamili et al. recorded Doppler signals over the
survival [11]. The detrimental effects of embolization on femoral arteries in patients undergoing iliac artery percu-
the kidney have been recognized for years [12] but are not taneous transluminal angioplasty, using as the control
always appreciated clinically due to the lag time between group a set of patients undergoing RAI, and demonstrated
the inciting event and the change in renal functional that embolic signals could be detected in some patients
capacity. While a few patients will experience an acute undergoing RAI up to 2 h following the procedure [20]. An
phenomenon that may include acute renal failure, leg or ex vivo study of renal angioplasty and stenting showed that
foot pain (blue toe syndrome), bowel ischemia, and skin large numbers of 60- to 1000-lm particles are released
injury (livedo reticularis), most patients will experience during all phases of intervention (guidewire placement,
a more insidious progressive loss of renal function that balloon dilation, stent placement, etc.) but that even greater
occurs over 312 weeks [10, 13]. numbers of smaller fragments (1040 lm) are also released
While atheroemboli can cause acute vessel obstruction, [21]. Edwards et al. recently reported an analysis of debris
it is their late sequelae, the localized inflammation with retrieved from 20 patients undergoing renal artery stent
intimal thickening and cellular proliferation, that are the placement with the use of the Angioguard device. Visible
most damaging. The perivascular inflammatory response debris was retrieved from 2 of 20 patients, however,
that results from atheroemboli is indistinguishable from the microscopic examination revealed debris in all 20 cases.
late appearance of arteritis [14]. The net effect on kidney An average of 2389 particles was retrieved per patient and
function is highly variable, with some patients experienc- most (96%) of the embolic particles were \100 lm in size
ing mild declines in renal function and others progressing (range, 20 lm to [1 mm) [22]. Correlation analyses
to dialysis [10, 13]. Classically atheroembolization is demonstrated positive trends between increasing collected
associated with eosinophilia; however, this may be seen in debris and impaired renal function response (rank corre-
as few as 22% of patients [13], further complicating the lation with serum creatinine difference, +0.45 [p = 0.06];
diagnosis. rank correlation with estimated glomerular filtration rate
Recent experimental work in a rat model using 20- to difference, 0.40 [p = 0.10]).
30-lm microspheres to mimic atheroemboli provides sup- Although atheroembolization is reportedly uncommon
port for the concept that atheroembolization may often go in percutaneous renal revascularization [2], theoretically a
undetected. Kimura et al. showed that the microspheres number of patients may actually lose a fraction of their
lodge in small arterioles and glomeruli and that renal functional renal parenchyma by atheroembolization, as the

123
G. J. Dubel and T. P. Murphy 17

damage may remain undetected by routine laboratory percutaneous transluminal angioplasty using either a stan-
examination. This is possible because the large functional dard guidewire or a distal occlusion embolic protection
reserve of the kidney allows for a 50% loss of the total system (GuardWire [Percusurge]; Medtronic, Minneapolis,
glomerular filtration before the serum creatinine level rises MN). This trial led to FDA approval for use in SVGs, with
to an abnormal range [23]. In other words, partial loss of a 42% reduction (9.6% vs 16.5%; p = 0.004) in 30-day
functional reserve after angioplasty or stent placement may major adverse clinical events [3]. The trial established
remain undetected unless it is investigated by specific embolic protection as the standard of care for SVG stent-
methods. These data support several important conclusions ing. Subsequent trials using other devices have
regarding atheroembolization and the kidney: (i) athero- demonstrated noninferiority compared with the GuardWire
embolization does occur during aortic interventions and led to their approval for use in SVG interventions
including RAI; (ii) atheroembolization probably occurs (FIRE trial [FilterWire EX] [6] and PRIDE trial [Tri-
more often than is suspected clinically due to the delay in Active, Kinsey-Nash] [25]).
renal effects in most patients and nonspecific clinical Distal embolization has been documented to occur
findings; and (iii) any measure that could reduce the during native coronary artery interventions, especially in
number of embolic particles should have a positive impact the setting of thrombus-laden lesions in acute MI [26],
on long-term kidney function and, by extension, overall however, to date randomized trials have failed to show a
cardiovascular morbidity and mortality. significant benefit in native coronary arteries undergoing
percutaneous coronary intervention (PCI) (EMERALD [7],
PROMISE [27]) despite earlier nonrandomized studies
Renal Artery Interventions Without Distal Embolic showing retrieval of embolic material, improved TIMI3
Protection flow rates, myocardial blush and ST-segment resolution,
and reduction in infarct size compared with conventional
Numerous authors that have described the results for PCI [28, 29]. The reasons that a benefit was not demon-
unprotected RAIs and the results vis-a-vis kidney function strated in the randomized trials are unclear at this time,
outcomes certainly leave room for improvement. Palmaz however, research in the area is ongoing.
published a metanalysis of eight series including 349 Experience in the carotid circulation has also led to the
patients and reported stabilized renal function in 38% and inclusion of distal protection as a routine part of carotid
unchanged decline in 62% [24]. Isles published a met- artery stenting (CAS) despite the fact that few randomized
analysis of 10 series that included 379 patients and reported data exist. EVA-3S was a randomized trial comparing CAS
that renal function (as measured by serum creatinine) was (with or without DPD) to carotid endarterectomy. The stent
improved in 26%, stabilized in 48%, and worsened in 26%, without a DPD arm was stopped after 80 patients were
with 2.4% of patients requiring dialysis following the enrolled because it was associated with a 30-day relative
procedure. The metanalysis by Leertower et al. included 14 risk of stroke of 3.9 compared to stenting with DPD [30].
stent series (678 patients) and 10 percutaneous renal In a large nonrandomized prospective registry a DPD was
angioplasty series (644 patients). The metanalysis reported used in 668 of 1483 patients (45%). Patients treated with a
that percutaneous renal artery stenting resulted in an DPD had a lower rate of ipsilateral stroke (1.7% versus
improvement in renal function in 30%, stabilization in 4.1%; p = 0.007) and a lower rate of all nonfatal strokes
38%, and worsening in 32%. Interestingly, the renal per- and all deaths (2.1% versus 4.9%; p = 0.004) during the
cutaneous transluminal angioplasty series demonstrated an hospital stay [31]. In another registry study, Wholey et al.
improvement in renal function in 38% (p \ 0.001) but had compiled a worldwide experience that included 11,243
a lower technical success rate (77% vs. 98%, respectively; patients (6573 with DPD, 4221 without DPD). The 30-day
p \ 0.001) and higher restenosis rate (17% and 26%, rate of stroke- and procedure-related death was 5.29% for
respectively; p \ 0.001) compared to stenting. unprotected CAS vs. 2.23% for protected CAS [5]. A
metanalysis has also been performed based on single-cen-
ter registry data on CAS. The metanalysis included 3196
DPDs in Other Vascular Territories patients (839 with DPD, 2357 without DPD) and reported a
significant reduction in 39-day stroke and death (1.8% vs.
Most of the data on embolic protection comes from the 5.5%; p = 0.001) for CAS with a DPD compared to
coronary and the carotid vascular territories. The most unprotected CAS. Shortcomings of these data include the
positive data are from randomized prospective trials lack of robust prospective randomized studies, however, it
performed on patients with coronary artery SVGs under- is currently considered standard-of-care to use DPDs for
going percutaneous transluminal angioplasty. The SAFER CAS so it is unlikely that randomized data will ever be
[3] trial included 801 patients randomized to SVG available.

123
18 G. J. Dubel and T. P. Murphy

The large coronary SVG and carotid experiences have with distal occlusive and filter-type devices have been
provided useful information on the ability of various DPDs noted that may preclude the use of the device (i.e., cannot
to prevent embolization of embolic material and provides cross lesion, cannot safely deploy DPD, patient cannot
insight that may be applicable to the renal vasculature. tolerate occlusion of vessel, etc.) in some patients [3, 39,
Early feasibility studies confirmed that both distal occlusion 40]. These experiences support the notion that further
and filter-type devices were capable of allowing capture of improvements will be required in order to make DPDs
potentially embolic materials [32]. In SVG intervention, more effective. This is especially important in the renal
analysis of DPD retrieved material reveals a typical frag- artery, where very proximal branching, steep angulation, or
ment size of \100 lm [33, 34], with particles up to 1 to 3 tortuousity may make DPD use more challenging.
mm [34, 35]. The material is heterogeneous, with a pre- A final intriguing concept is the combination of platelet
dominance of noncellular elements (lipid, cholesterol inhibition using glycoprotein IIb/IIIa inhibitors in con-
crystals, collagen extracellular matrix, microcalcification) junction with a DPD. This idea is based on the fact that
as well as cellular elements (smooth muscle cells, lipid- activated platelet-rich thrombus is commonly retrieved
laden macrophages, and platelets) [3436]. using a DPD despite adequate anticoagulation. Further-
Despite inherent differences in their mechanism of more, there is animal evidence suggesting a role for platelet
protection, a comparative analysis of retrieved debris from activation in ischemia-induced renal dysfunction [41]. In
distal occlusion (GuardWire) and filter (FilterWire) devices animal models platelet activation contributes to ischemic
has revealed remarkably similar amounts and type of glomerular injury, whereas inhibition reduces such injury
retrieved material [33]. Approximately 90% of the particles [42]. In native coronary artery interventions the use of
were \100 lm, confirming that filter-type devices can trap glycoprotein IIb/IIIa inhibitors has been associated with
emboli below their pore size. This is likely due to particles fewer 30-day major adverse clinical events compared with
tendency to clump or strand across filter pores, thus heparin alone [43], however, subgroup analysis from
reducing the effective pore size. This suggests that both multiple studies has not suggested a benefit in SVG inter-
designs may offer clinical efficacy unless soluble mediators ventions [44]. A secondary analysis of the FIRE trial
not detected in the aspirates were to play some important indicated that glycoprotein IIb/IIIa receptor blocker use in
role. Similarly Quan et al. showed no differences in combination with the Filter-Wire DPD was associated with
retrieved material when comparing four different types of better outcomes in high-risk patients [45]. These data are
DPDs used in coronary SVG and carotid interventions. intriguing in that they suggest a beneficial effect in com-
Interestingly they did demonstrate a smaller volume of bining embolic protection and platelet inhibition.
retrieved material in native coronary artery interventions
when using the FilterWire or GuardWire compared with
the Proxis device, however, a greater number of smaller Distal Embolic Protection in the Renal Circulation
fragments were retrieved using the GuardWire [36]. This
study suggests that there are subtle differences in the var- There are limited clinical data on DPDs for renovascular
ious types of DPDs that may have effects on end organ interventions at the present time. These data are summa-
damage and that continued research and refinement will be rized in Table 1. Early work was done by Henry et al. in
required to maximize the benefit of these devices. 2001, who reported experience with the GuardWire
Studies from other vascular territories have demon- (Medtronic/AVE, Minneapolis, MN) device, retrieving
strated that the level of embolic protection afforded by debris from 100% (28 of 28) of patients [46]. Patients with
DPD is not complete and that using these devices may add a renal artery bifurcation \2 cm from the stenosis were
another level of complexity to an intervention. Transcranial excluded from the study, as the DPD could not be used.
Doppler studies of the middle cerebral artery during CAS Mean renal artery occlusion time was 6.58 2.52 min
have shown that both micro- and macroembolic events (range, 2.2913.17 min). Retrieved particles ranged in size
occur despite the use of either distal occlusion [37] or filter- from 38 to 6206 lm (median size, 188 50 lm) and
type [19] DPDs. The Dutch study reported increased included atheromatous plaque, cholesterol crystal, necrotic
transcranial Doppler cerebral events with the use of filter cores, fibrin, thrombi, platelets, and macrophage foam cells
DPDs compared to non-DPD CAS, without any difference [46]. Renal function as measured by serum creatinine was
in clinically evident major or minor stroke [19]. Others stable in 82% and improved in 18% at a mean 6-month
have documented increases in procedural time using DPDs follow-up. Henry updated this experience in 2003 [47] to
[6]. Prolonged ischemic and/or procedural time has been include 56 patients including 10 additional patients (38
postulated as one possible reason that DPDs have not been total) treated with the GuardWire, 26 with the FilterWire,
shown to be beneficial in PCI for acute MI [7, 38]. In both and 1 with the Angioguard. Six-month renal outcomes
the coronary and the carotid circulation technical problems were similar to the original series. Visible debris was

123
G. J. Dubel and T. P. Murphy 19

from stenosis to renal


retrieved from 100% of Angioguard cases and 80% of filter

Lesions dottered prior


to passage of DPD
cases. Hagspiel reported the attempted use of the Filter-

Excluded pts \2 cm

Only visible debris

Only visible debris


WireEX in five renal arteries but achieved success only in
three and in one of the three the seal was felt to be
bifurcation
Comments

counted

counted
incomplete [48]. Interestingly, debris was retrieved from
all three and included cholesterol debris and fibrin. Renal
functional outcomes were not reported. The authors con-
cluded that the DPD needed modification for the renal

35.09% (60) 62.57% (107) 2.3% (4)


Decline

5% (2)

3% (2)
circulation to achieve the full theoretical advantages. In
another series using DPDs and examining aspirated efflu-
0

ent, Edwards et al. treated 32 renal arteries in 26 patients


and successfully deployed the GuardWire in 88% of cases
Stabilized

82% (37)

57% (21)

50% (13)

57% (36)

and retrieved visible debris in 44% [49]. Microscopic


analysis of the size and number of retrieved particles was
not carried out. Renal function was improved (50%) or
stabilized (50%) at short-term follow-up (46 weeks).
Renal outcome Improved

38% (14)

50% (13)

40% (25)
18% (8)

In one of the most widely quoted references on the use


of renal distal embolic protection, Holden and Hill pub-
lished results using the Angioguard in 46 renal arteries in
eGFR 46 wk

36 patients [50]. They noted that 65% of Angioguards


Note. RAs, renal arteries; SCr, serum creatinine; CrCl, creatinine clearance; eGFR, estimated glomerular filtration rate
eGFR 6 mo
CrCl 6 mo

contained debris (Fig. 3) after stent placement, that renal


SCr 6 mo
measure

function stabilized or improved in 95% over 12.5 months


of average follow-up, and that in no patient did renal
dysfunction accelerate after the procedure [50]. They also
71.36% (147)

noted that in a historical cohort of 20 patients before DPDs


Spasm in 2 cases 100% (65)

65% (30)

44% (14)

60% (38)
retrieval

were used routinely, renal function stabilized or improved


Debris

in only 75% [50]. A subsequent report with 63 patients and


w/ GuardWire

Spasm in 1 case
Technical Complications

\1%
success
Table 1 Human experience with distal protection devices (DPDs)

98.1%
GuardWire (38) 100%

100%

100%
88%
Angioguard (1)
FilterWire (26)

Angioguard

Angioguard
GuardWire
Author & year [ref. No.] No. pts/ RAs Device

182/206
56/65

37/46

Edwards et al. 2006 [49] 26/32

63/63
Holden et al. 2003 [50]

Holden et al. 2006 [51]


Henry et al. 2003 [47]

Total

Fig. 3 Embolic material retrieved during renal artery stent procedure


with Angioguard device. (Used with permission from Ref. 51.)

123
20 G. J. Dubel and T. P. Murphy

mean 16-month follow-up confirmed these observations studies results in debate as to the amount of benefit that a
[51]. DPD provides, which cannot be ascertained with certainty at
In the renal bed the only study to address combination this point. Nonrandomized data have proven that DPDs trap
therapy using distal embolic protection and glycoprotein embolic particles in 44%100% of patients [22, 46, 49, 50],
IIb/IIIa inhibition is the RESIST study [52]. The RESIST thus documenting that embolism is occurring during RAI. As
study is a randomized trial designed to assess safety and many of these emboli contain platelet-rich thrombi, the role
efficacy of using protective device (Angioguard) and/or of platelet inhibitors and DPDs is also intriguing. Given the
glycoprotein IIb/IIIa inhibitor (Abciximab; Centocor, critical importance of renal functional preservation, it fol-
Horsham, PA) to prevent renal injury 1 month after renal lows that everything that can be done to prevent
artery stenting. The study is yet unpublished, however, atheroembolism should be undertaken including the use of a
preliminary analysis suggests several important findings. DPD when anatomically feasible.
(i) Abciximab use was associated with a marked reduction
in the occurrence of platelet-rich thrombi captured in the
References
filter baskets (42% vs 7%; p \ 0.01) but no difference was
observed in the capture of atheromatous debris (21 vs 17%; 1. Isles CG, Robertson S, Hill D (1999) Management of renovas-
p = ns) or fibrin-based thrombi. (ii) The use of DPDs alone cular disease: a review of renal artery stenting in ten studies. QJM
was not associated with improved renal functional out- 92:159167
comes compared to unprotected stenting (1 28% vs 2. Leertouwer TC, Gussenhoven EJ, Bosch JL, et al. (2000) Stent
placement for renal arterial stenosis: Where do we stand? A meta-
10 20%; p = 0.09), whereas the use of Abciximab was analysis. Radiology 216:7885
associated with improved renal functional outcomes com- 3. Baim DS, Wahr D, George B, et al. (2002) Randomized trial of a
pared with stenting (0.4 27% vs 10 20%; distal embolic protection device during percutaneous intervention
p \ 0.05). (iii) A significant and positive interaction was of saphenous vein aorto-coronary bypass grafts. Circulation
105:12851290
observed between Abciximab and Angioguard use 4. Kastrup A, Groschel K, Krapf H, Brehm BR, Dichgans J, Schulz
(p \ 0.05). In comparisons between groups an overall JB (2003) Early outcome of carotid angioplasty and stenting with
improvement in renal function was only observed in and without cerebral protection devices: a systematic review of
patients allocated to both treatments (p \ 0.01). This study the literature. Stroke 34:813819
5. Wholey M, Al-Mubarek N, Wholey M (2003) Updated review of
suggests that platelet inhibition may indeed be valuable in the global carotid artery stent registry. Catheter Cardiovasc Interv
preserving renal function, however, final analysis and fur- 60:259266
ther studies will be required to confirm these findings. 6. Stone GW, Rogers C, Hermiller J, et al. (2003) Randomized
The CORAL (Cardiovascular Outcomes with Renal comparison of distal protection with a filter-based catheter and a
balloon occlusion and aspiration system during percutaneous
Atherosclerotic Lesions) Study, the largest National Insti- intervention of diseased saphenous vein aorto-coronary bypass
tutes of Health-sponsored study of RAI, initially included grafts. Circulation 108:548553
the use of DPDs, however, after numerous technical or 7. Stone GW, Webb J, Cox DA, et al. (2005) Distal microcirculatory
angiographic complications in roll-in patients, the device protection during percutaneous coronary intervention in acute
ST-segment elevation myocardial infarction: a randomized con-
was dropped from the protocol [53]. trolled trial. JAMA 293:10631072
8. Kaatee R, Beek FJ, Verschuyl EJ, et al. (1996) Atherosclerotic
renal artery stenosis: Ostial or truncal? Radiology 199:637640
Conclusions 9. Flory C (1945) Arterial occlusions produced by emboli from
eroded aortic atheromatous plaques. Am J Pathol 21:549565
10. Scolari F, Tardanico R, Zani R, et al. (2000) Cholesterol crystal
While the early data on DPDs are promising, much remains embolism: a recognizable cause of renal disease. Am J Kidney
to be learned about DPDs in the renal arteries. Early expe- Dis 36:10891109
rience has demonstrated that DPDs can be used safely and 11. Krishnamurthi V, Novick AC, Myles JL (1999) Atheroembolic
renal disease: effect on morbidity and survival after revasculari-
effectively in patients with appropriate anatomy. As expe- zation for atherosclerotic renal artery stenosis. J Urol 161:1093
rience is gained the relative merits of filter devices versus 1096
balloon occlusion devices should become clearer. Future 12. Harrington JT, Sommers SC, Kassirer JP (1968) Atheromatous
device innovations should assist in achieving technical emboli with progressive renal failure. Ann Intern Med 68:152
13. Thadhani RI, Camargo CA Jr, Xavier RJ, Fang LS, Bazari H
success in a greater number of patients, as newer devices will (1995) Atheroembolic renal failure after invasive procedures.
accommodate renal anatomy better than current devices. Natural history based on 52 histologically proven cases. Medicine
Lower-profile devices should allow device placement with (Baltimore) 74:350358
production of even fewer embolic particles. These 14. Modi KS, Rao VK (2001) Atheroembolic renal disease. J Am Soc
Nephrol 12:17811787
improvements should make RAI procedures using DPDs 15. Kimura M, Suzuki T, Hishida A (1999) A rat model of pro-
even safer and offer the potential to further improve renal gressive chronic renal failure produced by microembolism. Am J
functional outcomes. The lack of randomized prospective Pathol 155:13711380

123
G. J. Dubel and T. P. Murphy 21

16. Boisclair C, Therasse E, Oliva VL, et al. (1997) Treatment of 33. Rogers C, Huynh R, Seifert PA, et al. (2004) Embolic protection
renal angioplasty failure by percutaneous renal artery stenting with filtering or occlusion balloons during saphenous vein graft
with Palmaz stents: midterm technical and clinical results. Am J stenting retrieves identical volumes and sizes of particulate
Roentgenol 168:245251 debris. Circulation 109:17351740
17. van de Ven PJG, Kaatee R, Beutler JJ, et al. (1999) Arterial 34. Webb JG, Carere RG, Virmani R, et al. (1999) Retrieval and
stenting and balloon angioplasty in ostial atherosclerotic reno- analysis of particulate debris after saphenous vein graft inter-
vascular disease: a randomised trial. Lancet 353:282286 vention. J Am Coll Cardiol 34:468475
18. Ackerstaff RGA, Suttorp MJ, van den Berg JC, et al. (2005) 35. Lowe HC, Houser SL, Aretz T, MacNeill BD, Oesterle SN,
Prediction of early cerebral outcome by transcranial Doppler Palacios IF (2002) Significant atheromatous debris following
monitoring in carotid bifurcation angioplasty and stenting. J Vasc uncomplicated vein graft direct stenting: evidence supporting
Surg 41:618624 routine use of distal protection devices. J Invas Cardiol 14:636
19. Vos JA, van den Berg JC, Ernst SMPG, et al. (2005) Carotid 639
angioplasty and stent placement: comparison of transcranial 36. Quan V-H, Huynh R, Seifert PA, et al. (2005) Morphometric
Doppler US data and clinical outcome with and without filtering analysis of particulate debris extracted by four different embolic
cerebral protection devices in 509 patients. Radiology 234:493 protection devices from coronary arteries, aortocoronary saphe-
499 nous vein conduits, and carotid arteries. Am J Cardiol 95:1415
20. Al-Hamali S, Baskerville P, Fraser S, Walters H, Markus HS 1419
(1999) Detection of distal emboli in patients with peripheral 37. Al-Mubarak N, Roubin GS, Vitek JJ, Iyer SS, New G, Leon MB
arterial stenosis before and after iliac angioplasty: a prospective (2001) Effect of the distal-balloon protection system on micro-
study. J Vasc Surg 29:345351 embolization during carotid stenting. Circulation 104:19992002
21. Hiramoto J, Hansen KJ, Pan XM, Edwards MS, Sawhney R, 38. Akbar Ali O, Bhindi R, McMahon AC, Brieger D, Kritharides L,
Rapp JH (2005) Atheroemboli during renal artery angioplasty: an Lowe HC (2006) Distal protection in cardiovascular medicine:
ex vivo study. J Vasc Surg 41:10261030 current status. Am Heart J 152:207216
22. Corriere MA, Pearce JD, Craven TE, et al. (2007) Atheroembo- 39. Powell RJ, Alessi C, Nolan B, et al. (2006) Comparison of
lization during percutaneous renal artery revascularization. In: embolization protection device-specific technical difficulties
Southern Association of Vascular Surgery Meeting, Rio Grande, during carotid artery stenting. J Vasc Surg 44:5661
PR 40. Henry M, Polydorou A, Henry I, Polydorou A, Hugel M (2004)
23. Brenner B, Herbert S (1994) Disturbances of the renal function. Carotid angioplasty under cerebral protection with the percusurge
In: Isselbacher K, Brauwald E, Wilson J, Martin J, Fauci A, guardwire system. Catheter Cardiovasc Interv 61:293305
Kasper D (eds) Harrisons principles of internal medicine. 41. Gianello P, Janssen T, Chatzopoulos C, et al. (1998) Beneficial
McGrawHill, New York effect of piracetam on renal blood flow in ischemically injured
24. Palmaz JC (1998) The current status of vascular intervention in kidneys in the rat. Transplant Proc 20:914916
ischemic nephropathy. J Vasc Interv Radiol 9:539543 42. Masumura H, Kunitada S, Irie K, Ashida S, Abe Y (1991) A
25. Carrozza JJP, Mumma M, Breall JA, Fernandez A, Heyman E, thromboxane A2 synthase inhibitor, DP-1904, prevents rat renal
Metzger C (2005) Randomized evaluation of the TriActiv injury. Eur J Pharmacol 193:321327
balloon-protection flush and extraction system for the treatment 43. The EI (1997) Platelet glycoprotein IIb/IIIa receptor blockade
of saphenous vein graft disease. J Am Coll Cardiol 46:1677 and low-dose heparin during percutaneous coronary revasculari-
1683 zation. N Engl J Med 336:16891697
26. Cameron J, Buchbinder M, Wexler L, Oesterle S (1987) 44. Roffi M, Mukherjee D, Chew DP, et al. (2002) Lack of benefit
Thromboembolic complications of percutaneous transluminal from intravenous platelet glycoprotein IIb/IIIa receptor inhibition
coronary angioplasty for myocardial infarction. Catheter Car- as adjunctive treatment for percutaneous interventions of aorto-
diovasc Diagn 13:100106 coronary bypass grafts: a pooled analysis of five randomized
27. Gick M, Jander N, Bestehorn H-P, et al. (2005) Randomized clinical trials. Circulation 106:30633067
evaluation of the effects of filter-based distal protection on 45. Jonas M, Stone GW, Mehran R, et al. (2006) Platelet glycoprotein
myocardial perfusion and infarct size after primary percutaneous IIb/IIIa receptor inhibition as adjunctive treatment during
catheter intervention in myocardial infarction with and without saphenous vein graft stenting: differential effects after randomi-
ST-segment elevation. Circulation 112:14621469 zation to occlusion or filter-based embolic protection. Eur Heart J
28. Nakamura T, Kubo N, Seki Y, et al. (2004) Effects of a distal 27:920928
protection device during primary stenting in patients with acute 46. Henry M, Klonaris C, Henry I, et al. (2001) Protected renal
anterior myocardial infarction. Circ J 68:763768 stenting with the PercuSurge GuardWire device: a pilot study. J
29. Yip H-K, Wu C-J, Chang H-W, et al. (2003) Effect of the Endovasc Ther 8:227237
PercuSurge GuardWire device on the integrity of microvascula- 47. Henry M, Henry I, Klonaris C, et al. (2003) Renal angioplasty
ture and clinical outcomes during primary transradial coronary and stenting under protection: The way for the future? Catheter
intervention in acute myocardial infarction. Am J Cardiol Cardiovasc Interv 60:299312
92:13311335 48. Hagspiel KD, Stone JR, Leung DA (2005) Renal angioplasty and
30. Mas J, Chatellier G, Beyssen B, Investigators E-S (2004) Carotid stent placement with distal protection: preliminary experience
angioplasty and stenting with and without cerebral protection: with the FilterWire EX. J Vasc Interv Radiol 16:125131
clinical alert from the Endarterectomy Versus Angioplasty in 49. Edwards MS, Craven BL, Stafford J, et al. (2006) Distal embolic
Patients with Symptomatic Severe Carotid Stenosis (EVA-3S) protection during renal artery angioplasty and stenting. J Vasc
Trial. Stroke 35:18e20e Surg 44:128135
31. Zahn R, Mark B, Niedermaier N, et al. (2004) Embolic protection 50. Holden A, Hill A (2003) Renal angioplasty and stenting with
devices for carotid artery stenting: Better results than stenting distal protection of the main renal artery in ischemic nephropa-
without protection? Eur Heart J 25:15501558 thy: early experience. J Vasc Surg 38:962968
32. Henry M, Amor M, Henry I, et al. (1999) Carotid stenting with 51. Holden A, Hill A, Jaff MR, Pilmore H (2006) Renal artery stent
cerebral protection: first clinical experience using the PercuSurge revascularization with embolic protection in patients with ische-
GuardWire system. J Endovasc Surg 6:321331 mic nephropathy. Kidney Int 70:948955

123
22 G. J. Dubel and T. P. Murphy

52. Cooper CJ, Haller ST, Coyler W, et al. (2007) RESIST: embolic events among patients with renal artery stenosis and systolic
protection and platelet inhibition during renal artery stenting. In: hypertension: Rationale and design of the CORAL trial. Am
American College of Cardiology meeting, New Orleans, LA Heart J 152:5966
53. Cooper CJ, Murphy TP, Matsumoto A, et al. (2006) Stent
revascularization for the prevention of cardiovascular and renal

123
Cardiovasc Intervent Radiol (2008) 31:2335
DOI 10.1007/s00270-007-9175-6

CLINICAL INVESTIGATION

Long-Term Follow-Up After Endovascular Treatment of Acute


Aortic Emergencies
M. B. Pitton S. Herber W. Schmiedt A. Neufang B. Dorweiler
C. Duber

Received: 12 May 2007 / Accepted: 29 August 2007 / Published online: 18 October 2007
Springer Science+Business Media, LLC 2007

Abstract cases primary graft extension was done during the initial
Purpose To investigate the long-term outcome and effi- procedure (in 1 case due to distal migration of the graft
cacy of emergency treatment of acute aortic diseases with during stent release, in 2 cases due to the total length of
endovascular stent-grafts. the aortic aneurysm). In 4 cases secondary graft extensions
Methods From September 1995 to April 2007, 37 patients were performedfor new aortic ulcers at the proximal
(21 men, 16 women; age 53.9 19.2 years, range 1885 stent struts (after 5 days) and distal to the graft (after 8
years) with acute complications of diseases of the months) and recurrent aortobronchial fistulas 5 months and
descending thoracic aorta were treated by endovascular 9 years after the initial procedureresulting in a total of
stent-grafts: traumatic aortic ruptures (n = 9), aortobron- 41 endovascular procedures. The 30-day mortality rate was
chial fistulas due to penetrating ulcer or hematothorax (n = 8% (3 of 37) and the overall follow-up was 29.9 36.6
6), acute type B dissections with aortic wall hematoma, months (range 0139 months). All patients with traumatic
penetration, or ischemia (n = 13), and symptomatic aneu- ruptures demonstrated an immediate sealing of bleeding.
rysm of the thoracic aorta (n = 9) with pain, penetration, or Patients with aortobronchial fistulas also demonstrated a
rupture. Diagnosis was confirmed by contrast-enhanced satisfactory follow-up despite the necessity for reinter-
CT. Multiplanar reformations were used for measurement vention and graft extension in 3 of 6 cases (50%). Two
of the landing zones of the stent-grafts. Stent-grafts were patients with type B dissection died due to mesenteric
inserted via femoral or iliac cut-down. Two procedures ischemia despite sufficient mesenteric blood flow being
required aortofemoral bypass grafting prior to stent-graft- restored (but too late). Two suffered from neurologic
ing due to extensive arteriosclerotic stenosis of the iliac complications, 1 from paraplegia and 1 from cerebral
arteries. In this case the bypass graft was used for intro- ischemia (probably embolic), 1 from penetrating ulcer, and
duction of the stent-graft. 1 from persistent ischemia of the kidney. Five of 9 (56%)
Results A total of 46 stent-grafts were implanted: Van- patients with symptomatic thoracic aneurysm demon-
guard/Stentor (n = 4), Talent (n = 31), and Valiant (n = strated endoleaks during follow-up and there was an
11). Stent-graft extension was necessary in 7 cases. In 3 increase in the aneurysm in 1.
Conclusion Endovascular treatment is safe and effective
for emergency treatment of life-threatening acute thoracic
aortic syndromes. Results are encouraging, particularly for
M. B. Pitton (&)  S. Herber  C. Duber traumatic aortic ruptures. However, regular follow-up is
Department of Diagnostic and Interventional Radiology,
mandatory, particularly in the other pathologies, to identify
University Hospital of Mainz, Johannes Gutenberg University of
Mainz, Langenbeckstrasse 1, 55131 Mainz, Germany late complications of the stent-graft and to perform
e-mail: pitton@radiologie.klinik.uni-mainz.de appropriate additional corrections as required.

W. Schmiedt  A. Neufang  B. Dorweiler


Keywords Aortobronchial fistula  Endovascular 
Department of Cardiovascular Surgery, University Hospital of
Mainz, Johannes Gutenberg University of Mainz, Penetrating aortic ulcer  Stent-graft 
Langenbeckstrasse 1, 55131 Mainz, Germany Traumatic aortic rupture  Type B aortic dissection

123
24 M. B. Pitton et al.: Long-Term Follow-Up After Endovascular Treatment of Acute Aortic Emergencies

Introduction of atherosclerotic occlusion of the iliac arteries. These


procedures were continued with a distal anastomosis to the
Emergency surgery in patients with acute aortic injuries femoral artery for completion of an aortofemoral bypass
has demonstrated a high mortality rate [13]. The first graft. There was no procedure-related peri-interventional
patient receiving endovascular treatment for a traumatic mortality; however, the overall 30-day mortality was 8% (3
aneurysm was reported in 1988 [4]. This technique has of 37: two cases with type B dissection and advanced
been refined in recent years and has emerged as an alter- mesenteric ischemia and one case with penetrating aneu-
native to open surgery for acute management of traumatic rysm with cardiac failure). The overall follow-up is 29.9
ruptures, dissections, aortobronchial fistulas, and aneu- 36.6 months (range 0139 months; Table 1) and the
rysms [515]. In our institution, endovascular emergency overall mean survival is 87.9 13.0 months (95% confi-
treatment of thoracic pathologies was introduced in 1995. dence interval 62.4 to 113.3) (Fig. 1). During follow-up,
Patients have been followed-up during the last decade 10 secondary procedures were necessary, including 4
including clinical visits and CT scanning every year. There endovascular procedures and 6 open surgeries. Therefore, a
have been a diversity of clinical findings, with imperative total of 41 interventional procedures were performed, using
secondary interventions and open surgery. This paper general anesthesia in 30 cases (73%) and local anesthe-
summarizes a single-center experience with acute endo- sia in 11 (27%), and 46 stent-grafts were implanted
vascular treatment of life-threatening emergencies of the (Vanguard/Stentor, n = 4; Talent, n = 31; Valiant, n = 11).
thoracic aorta and demonstrates the management of sec- The secondary major open surgeries required general
ondary clinical findings. anesthesia. Patients were classified into four groups with
respect to the leading clinical symptoms or pathologic
findings (Tables 1, 2).
Materials and Methods

From September 1995 to April 2007, 37 patients (21 men, Acute Traumatic Aortic Ruptures
16 women; 53.9 19.2 years, range 1885 years) with
acute aortic emergencies were treated by endovascular Patients with acute traumatic ruptures had been trauma-
stent-grafting. Aortic emergencies covered acute traumatic tized by motorbike or automobile accidents and incomplete
aortic ruptures with hematothorax (n = 9), hematothorax or suicides (n = 9). The mean age was 32.3 12.3 years
hemoptysis because of acute aortobronchial fistulas due to (range 1849 years). The accompanying findings included
penetrating aortic ulcers (n = 6), acute type B dissections hemato-mediastinum and hematothorax (n = 9), pneumo-
with collapse of the true lumen of the thoracic aorta (n = thorax (n = 2), serial rib fractures (n = 7), miscellaneous
13) and consecutive mesenteric ischemia, ischemic legs fractures of the peripheral skeleton (17 fractures in 7
syndrome or renal insufficiency, as well as symptomatic patients) and cranial fractures (n = 7), intracranial trauma
thoracic aneurysms with back pain, penetration, or rupture in 3 cases (edema, n = 1; subarachnoid bleeding, n = 2;
(n = 9). The diagnosis was performed with contrast- epidural hematoma, n = 1), and intra-abdominal bleeding in
enhanced spiral CT until 2000, and thereafter by multide- 3 cases (liver, n = 1; spleen, n = 3; mesenteric trauma, n =
tector (MD) CT (slice thickness 1 mm, reconstruction 2). In cases of acute intracranial or abdominal bleeding,
interval 2 mm, contrast medium 120 ml, flow 4 ml/sec, emergency craniotomy or laparotomy was performed
start delay by bolus trigger (Imeron, Altana Pharma, before endovascular treatment. The other patients were
Konstanz)). Multiplanar reformations were used for mea- treated immediately with stent-grafting. Eight of 9 patients
surement of the prospective landing zones of the stent- had general anesthesia; 1 patient without significant
grafts. Stent-grafts were inserted via femoral or iliac cut- accompanying trauma was treated under local anesthesia.
down and were deployed under fluoroscopic guidance. In 8 of 9 cases the aortic rupture site was located, loco
Follow-up included clinical visits and contrast-enhanced typico, at the junction between the aortic arch and the
CT scans at 3, 6, and 12 months and yearly thereafter. descending aorta. In 1 case, the false aneurysm was located
atypically just above the level of the diaphragm with the
same characteristic imaging features. CT demonstrated
Results hematomas in the dorsal and middle third of the
upper mediastinum and a varying degree of accompanying
All procedures were successfully performed in the angio- hematothorax. Contrast-enhanced CT confirmed the diag-
suite. All but 3 procedures were performed via a femoral or nosis of rupture and demonstrated the false aneurysm with
an external iliac access. In these cases, a Dacron conduit irregular and eccentric diameter increase and the rupture of
was anastomosed to the aorta for stent-graft access because the aortic wall (Fig. 2). Multiplanar reformatted images of

123
M. B. Pitton et al.: Long-Term Follow-Up After Endovascular Treatment of Acute Aortic Emergencies 25

Table 1 Patient details, follow-up and clinical outcome


Total Traumatic rupture Aortobronchial Type B Penetrating
fistulas, aortic ulcers dissections aneurysms

Patients (n) 37 9 6 13 9
Male/female 21/16 8/1 4/2 3/10 6/3
Age, years (range) 53.9 19.2 (1885) 32.3 12.3 (1849) 67.7 11.1 (4877) 54.1 15.8 (2884) 65.9 15.2 (3285)
30-day mortality 3/37 (8%) 2/13 (15%) 1/9 (11%)
30-day morbidity 5/37 (14%) 1/6 (17%)a 4/13 (31%)b
Stent-graft-related 3/37 (8%) 1 2
Not stent-graft-related 2/37 (5%) 2
Follow-up, months (range) 29.9 36.6 (0139) 46.4 43.9 (3113) 63.1 46.7 (6139) 13.0 21.1 (167) 15.7 15.1 (040)
Mean survival 87.9 13.0 months
(62.4 to 113.3, 95%
confidence interval)
Late mortality 6/37 (16%) 1/9 (11%)c 2/6 (33%)d 1/13 (8%)e 2/9 (22%)f
g h i
Late morbidity 8/37 (22%) 1/9 (11%) 2/6 (33%) 5/13 (39%)
Stent-graft-related 3/37 (8%) 2 1
Not stent-graft-related 5/37 (14%) 1 open surgery 1 + 3 open surgery
Endoleaks (EL) 7/37 (19%) 1/9 (11%) proximal 1/6 (17%)j 5/9 (56%)
and perigraft leaks anchor leak proximal-distal 1 type I EL,
anchor leak 3 type II EL,
1 type III EL
a
Aortic ulcer 1 week after stent-grafting treated with stent-graft extension
b
One peri-interventional minor stroke, 1 stent-graft-related aortic ulcer. One persistent ischemia of the left kidney. One additional patient with
onset of paraplegia following a traumatic aortic type B dissection caused by an automobile accident, not related to the stent-graft procedure itself
c
Suicide 2 years after endovascular treatment
d
Cardiac event 2.5 and 5 years after treatment
e
Chronic obstructive pulmonary disease plus pulmonary embolism 1 year after treatment
f
At 6 months perioperative after knee-TEP and at 3 years a cardiac event
g
One open aortic arch surgery (see text)
h
Recurrent hemoptysis at 5 months and 9 years, treated with stent-graft extension
i
Aortic ulcer caused by the stent-graft at 8 months treated with stent-graft extension and 1 case with persisting perfusion deficit of the left
kidney, in 3 patients open surgery for additional aorta ascendens or descendens aneurysms (Table 2)
j
Due to progressive generalized aortic dilatation

the aorta were used for measurement of the intended fix- aneurysm. Therefore, a second stent-graft was implanted to
ation sites of the stent-grafts. In 7 of 9 procedures the overlap the first one.
femoral artery was cut down. In 2 cases the iliac artery was All procedures were successfully completed. There was
exposed: in 1 case because of the small femoral diameters, no procedure-related mortality and no 30-day mortality.
in the other because of the short length of the introducer Follow-up is 46.4 43.9 months (range 3113 months).
sheath of a Stentor device which, in our early experience in One patient was lost to follow-up. In the remainder, stent-
1996, was not dedicated, or of sufficient length, for thoracic graft-related morbidity has not been confirmed. However,
use. With respect to the rupture sites, the covering of the there were some events independent of the stent-graft: One
stent-grafts was placed just distal to the ostium of the patient suffered from adolescent depression and committed
subclavian artery, with the bare stent struts crossing the suicide 20 months after uneventful follow-up. Another
subclavian ostia in 5 of 9 cases. In 2 cases, the covering of developed new-onset arterial hypertension with the need
the stent-graft was placed distal to the ostium of the left for fourfold antihypertensive medication. In this case,
carotid artery and the subclavian orifices were occluded. intensive diagnostic tests have been performed to exclude
No neurologic deficits occurred in these cases. In 1 case the endocrinologic and renovascular causes. Angiography,
stent-graft migrated distally during the deployment proce- however, demonstrated a small aortic arch with some
dure and resulted in insufficient covering of the false buckling of the stent-graft and a systolic pressure gradient

123
26 M. B. Pitton et al.: Long-Term Follow-Up After Endovascular Treatment of Acute Aortic Emergencies

Fig. 1 KaplanMeier analysis


of overall survival after stent- 1,0 Survival
grafting (n = 36). Mean survival censored
is 87.9 13.0 months (62.4 to
113.3, 95% confidence interval),
limited to a maximum follow-up 0,8
of 139 months

0,6

Survival
0,4

0,2

0,0

0,0 20,0 40,0 60,0 80,0 100,0 120,0 140,0


Follow-up (months)

Table 2 Summary of procedures


(n) Total Traumatic Aortobronchial fistulas, Type B Penetrating
ruptures aortic ulcers dissections aneurysms

Procedures (total) 47 10 9 19 9
General anesthesia 36 9 7 18 2
Local anesthesia 11 1 2 1 7
Primary procedures 37 9 6 13 9
Stent-grafts 46 10 9 14 13
Stentor; Talent; Valiant 4; 31; 11 2; 7; 1 2; 7; 0 0; 8; 6 0; 9; 4
Secondary procedures 10/37 (27%) 1/9 (11) 3/6 (50%) 4/13 (31%)
Endovascular 4/37 (11%) 3/6 (50%)a 1/13 (7%)b
Open surgery 6/37 (14%) 1/9 (11%)c 3/13 (23%)d
a
Stent-graft extension because of aortic ulcer at 1 month, and recurrent hemoptysis 5 months and 9 years after treatment
b
Stent-graft extension because of aortic ulcer at 8 months after treatment
c
Aortic arch surgery
d
Three patients with 5 surgeries: 1 aortic arch surgery due to ascendens aneurysm at 2 months, 1 thoracoabdominal aortic replacement plus
revision due to rebleeding at 38 months, 1 aorta ascendens and 1 descendens replacement in the same patient at 6 months

of 15 mmHg between the proximal and distal end of the retrospectively, the arterial hypertension was not caused by
stent-graft. Ultimately, it was proposed that the stent-graft the stent-graft itself. One further case demonstrates a minor
itself might have reduced the elasticity of the aorta and perigraft flow at the proximal fixation site, which has
thereby induced arterial hypertension. In the light of the reduced spontaneously and is still being monitored.
young age of the patient and the amount of antihyperten-
sive drugs needed, aortic arch surgery was performed at 12
months after stent-grafting and the proximal stent struts Aortobronchial Fistulas and Penetrating Aortic Ulcers
were removed. The patient suffered a minor transient
neurologic deficit with paresthesia of the left half of There were 6 patients in this group (4 men, 2 women; age
the face; however, the hypertension persists. Therefore, 67.7 11.1 years, range 4877 years), including 4 with

123
M. B. Pitton et al.: Long-Term Follow-Up After Endovascular Treatment of Acute Aortic Emergencies 27

Fig. 2 Traumatic aortic rupture loco typico, distal to the origin of the Preinterventional angiography shows the typical outpouching of the
left subclavian artery. A CT scan demonstrates aortic laceration with vessel at the site of contained rupture. D Status immediately after
accompanying mediastinal hematoma. B Status after stent-grafting of stent-grafting of the descending aorta shows complete sealing of the
the thoracic aorta, with reduced mediastinal hematoma. C bleeding source

123
28 M. B. Pitton et al.: Long-Term Follow-Up After Endovascular Treatment of Acute Aortic Emergencies

severe hemoptysis caused by penetrating aortic aneurysms Acute Type B Dissections with Ischemic Complications
(n = 2) and aortic ulcers in a normal aortic diameter (n = 2). or Penetration
One of the aneurysms was atherosclerotic, while the second
patient had a history of two previous aortic isthmus cor- This group covers 13 patients (3 men, 10 women; age 54.1
rections. He demonstrated a false aneurysm at the site of 15.8 years, range 2884 years), 10 with atherosclerosis of
anastomosis with aneurysm penetration into the bronchial the aorta, 1 with Marfans disease, and 2 with traumatic
system. Another patient with a penetrating ulcer presented type B dissection. Patients presented with true lumen col-
with hematothorax and severe hemoptysis due to aortic lapse (n = 11), aortic wall hematoma (n = 3), hematothorax
penetration into the 6th segmental bronchus. She had to be (n = 3) including 1 acute aortic rupture (Fig. 4), and
resuscitated and stabilized during CT scanning. In the other mesenteric ischemia (n = 5) including patients with addi-
case, the penetrating aortic ulcer caused severe hemoptysis tional critical limb ischemia (n = 3) and renal ischemia (n =
but was hemodynamically reasonably stable. Another 2 2). Stent-graft deployment achieved restoration of the true
patients with penetrating aortic ulcers suffered from acute aortic lumen at the entry sites immediately distal to the
hematothorax and hypovolemic shock and ulcer penetra- subclavian ostium, with the bare stent struts of the stent-
tion with low back pain, respectively. All procedures were grafts crossing the subclavian ostia in 12 cases and com-
successfully completed in the first instance in terms of plete occlusion of the subclavian artery in 1. In the patient
acute sealing of the bleeding source and there was no with traumatic dissection and critical limb ischemia, the
procedure-related mortality. In 1 case a Dacron conduit to right iliac artery was additionally treated with a self-
the aorta was used for stent-graft access because of expandable stent in order to restore iliac artery blood flow.
occluded iliac arteries. In another case, the stent-graft had In 2 cases, three additional stents were used for blood flow
to be introduced via the left iliac artery because of the short restoration in the renal arteries. In both these patients, renal
length of the introducer system of the Stentor tube graft function improved or completely normalized. No addi-
used. In this early period of endovascular treatment (1995), tional stents were used for the mesenteric or celiac arteries
no dedicated introducer systems for thoracic application because angiography and CT confirmed the restoration of
were available. In all other cases, dedicated thoracic stent- arterial blood flow in the respective organs.
grafts were used. All procedures were technically successful. However, the
The follow-up is 63.1 46.7 months (range 6139 30-day mortality was 15% (2 of 13). Both patients showed
months). However, the stent-graft-related morbidity was pulsatile flow of the mesenteric artery at secondary lapa-
50% (3 of 6) and was associated with operative reinter- rotomy. However, the first demonstrated total necrosis of the
ventions during follow-up in order to maintain an small intestine and died on the same day. The second patient
appropriate treatment result. One patient with a history of was operated on because of an increasing lactate level 2
false aneurysm after two aortic arch surgeries presented weeks after stent-grafting and demonstrated segmental
with recurrent hemoptysis during sports at 9 years after the necrosis of the small intestine. A segment resection was
first endovascular treatment. He again was successfully performed; however, the patient died 1 week later from a
treated with stent-graft extension, an additional stent-graft subdural hematoma. One further patient has had a segment
overlapping the first one for sealing of the bleeding. The resection of the small bowel after restoration of mesenteric
second patient with recurrent hemoptysis presented with a blood flow and is still alive. Moreover, 1 patient suffered a
progression of the underlying atherosclerotic aneurysm at 5 retrograde dissection of the aortic arch during stent-graft
months after primary stent-grafting. He too received a stent deployment which was accompanied by a transient car-
extension and had an uneventful follow-up up to 59 diac arrest. The post-interventional CT demonstrated an
months, when he died of a cardiac event. The third rein- extending dissection into the aortic arch, the brachioce-
tervention occurred 1 week after stent-grafting of an aortic phalic trunk, and the right carotid artery. However, there
ulcer penetration in a generalized dilated aorta. Even was no flow reduction in these vessels and therefore no need
though asymptomatic at this stage, CT scans demonstrated for further intervention. Fortunately, there were also no
a new aortic ulcer at the proximal end of the stent-graft neurologic deficits in this case. The follow-up in this group
caused by penetration of the bare stent struts into the aortic is 13.0 21.1 months (range 067 months). The stent-graft-
wall. This patient was also treated with a proximal graft related morbidity was 23% (3 of 13). One patient suffered
extension (Fig. 3). During follow-up, she presented a from left cerebral ischemia on awakening from general
generalized dilatation of the thoracic and abdominal aorta anesthesia, which might be attributable to the catheteriza-
with new-onset endoleaks at the proximal and distal fixa- tion of the aortic arch during the procedure. One patient with
tion sites. Her symptoms are still under control but surgical a dissection-related aortic rupture suffers from a paraplegia
correction is not possible because of her poor general which was already obvious before stent-grafting and is
condition. therefore not related to the endovascular procedure. Another

123
M. B. Pitton et al.: Long-Term Follow-Up After Endovascular Treatment of Acute Aortic Emergencies 29

Fig. 3 Aortic ulcer with penetration into the left pleural cavity. A, B grafting. E CT control after stent-grafting demonstrates new-onset
Axial CT scans and coronal reformatting of CT demonstrates the aortic wall ulceration at 1 week follow-up caused by penetration of a
penetrating aortic ulcer with penetration through the left dorsolateral bare stent-strut into the aortic wall (arrow). F Confirmation of the new-
wall of the descending aorta with a consecutive complete hematothorax onset penetration by angiography (arrow). Indication for proximal stent
and hemorrhagic shock. C Preinterventional angiography confirms the extension. G Completion angiography immediately after the secondary
penetrating aortic ulcer. D Completion angiography after stent- procedure. H Confirmation of treatment result with CT

123
30 M. B. Pitton et al.: Long-Term Follow-Up After Endovascular Treatment of Acute Aortic Emergencies

Fig. 4 Type B dissection with perforation of the thoracic aorta, Preinterventional angiography with the introducer sheath of the stent-
hemato-mediastinum, and hematothorax with acute severe hemor- graft in situ. D Completion angiography after stent-graft deployment
rhagic shock. A Sagittal oblique reformations of MD-CT in the descending aorta. E, F Confirmation of complete sealing of the
demonstrates the onset of the dissection membrane just distal to the bleeding source with axial CT and sagittal oblique reformations of
orifice of the left subclavian artery. B Source of active bleeding from CT, demonstrating some residual hematothorax on the left side
the descending aorta and the dissection of the descending aorta. C

patient presented at 8 months follow-up with a penetrating range 3285 years) and included 7 symptomatic aneurysms
ulcer 1 cm below the distal end of the stent-graft, which with atherosclerotic origin, 1 with contained rupture
required distal graft extension with an overlapping addi- (Fig. 5), 1 with substantial back pain and aneurysm pene-
tional stent-graft. One further patient presented persistent tration, and 5 with back pain. One woman under
ischemia of the left kidney (CT finding), currently without immunosuppressive therapy after lung transplantation
clinical relevance. In this case, additional stenting of the presented an aneurysm suspicious for inflammatory dis-
respective renal ostium might have been helpful during the ease. Moreover, there was 1 patient with false aneurysms
initial stent-grafting. Three patients had secondary aortic showing a substantial diameter increase after aortic isthmus
surgery for ascendens or descendens aneurysms, with good surgery. Seven of 9 cases were performed with the patient
clinical outcome (Table 1). The clinical course of the under local anesthesia; in 2 cases general anesthesia was
remaining patients is uneventful. used. Eight of 9 patients were treated via femoral cut-
down; in 1 patient with small iliac arteries, a Dacron graft
was anastomosed to the aorta for the stent-graft procedure
Symptomatic Thoracic Aneurysms and was finally completed as an aortofemoral bypass graft.
Three atherosclerotic aneurysms were treated by means of
This group contains 9 patients with aneurysms of the two overlapping stent-grafts, which was the primary
descending aorta (6 men, 3 women; age 65.9 15.2 years, intention in 2 cases. In the other case, the first stent-graft

123
M. B. Pitton et al.: Long-Term Follow-Up After Endovascular Treatment of Acute Aortic Emergencies 31

Fig. 5 Acute aneurysm penetration of the descending thoracic aorta. projection). C Completion angiography immediately after stent-
A CT scan demonstrates the contained aneurysm rupture with contrast grafting without angiographic evidence of an endoleak. D, E Sagittal
extravasation into the posterior mediastinum with a large surrounding and axial contrast-enhanced MR images demonstrate some residual
hematoma and anterior shifting of the heart. B Angiography confirms contrast enhancement around the stent-graft within the aneurysm sac.
the contained rupture site with an adequate landing zone for the stent- The endoleak is still being monitored without additional treatment
graft proximal to the celiac trunk and the mesenteric artery (lateral because the aneurysm sac has shrunk

migrated distally during balloon inflation with the need for stenotic segment of the aortic arch was ligated between the
a proximal extension for complete aneurysm exclusion. orifices of the left carotid artery and the subclavian artery.
One patient was treated after aortic isthmus surgery with a The remaining false aneurysm in the immediate proximity
combined surgical and interventional approach. After a of the subclavian ostium was treated by a custom-made
bypass between the ascending and descending aorta, the tapered stent-graft from the descending aorta to the left

123
32 M. B. Pitton et al.: Long-Term Follow-Up After Endovascular Treatment of Acute Aortic Emergencies

subclavian artery. The remaining patients were treated with traffic accidents, have a considerable life expectancy. Our
only one stent-graft each. data are consistent with the literature, which has reported
The 30-day mortality was 11% (1/9) and was caused by high technical success rates between 92% and 100% [21,
a myocardial infarction and arrhythmia within 6 hr after the 22, 2729] and a low or even no procedure-related mor-
intervention. Follow-up is 15.7 15.1 months (range 040 tality [19, 2830]. Technical failure is predominantly due
months). One 85-year-old patient died 37 months after to a proximal type I endoleak, which can easily be cor-
endovascular treatment because of myocardial infarction. rected with extension (7.8%) [31]. In our early experience,
At 36 months CT follow-up, he showed a persistent type II 1 patient showed such an acute distal migration of the
endoleak, a slight increase in the aneurysm diameter, and stent-graft during its deployment which was treated by an
new lung metastasis of unknown origin. Three further cases additional stent-graft to overlap the first one. Stent-graft
demonstrate type II endoleaks. Aneurysm diameters had migration might be avoided by lowering the systolic blood
shrunk or were stable during follow-up in 5 patients and pressure below 80100 mmHg during deployment, and this
increased in 3. was practiced in the other cases. Moreover, damage to the
femoral or iliac access vessels might occur during an
emergency procedure and has been reported in up to 23%
Discussion of cases [32]. However, in the context of the emergency
situation, this is of minor importance and can be corrected
Acute aortic syndromes covered several life-threatening surgically during the same procedure after sealing of the
aortic pathologies with acute aortic rupture, ischemia or bleeding source.
acute risk of rupture with an imperative need for emer- Patients with aortic ruptures typically suffer from multi-
gency treatment [13]. These patients are often poor level trauma that could cause significant periprocedural
surgical cases with a high mortality rate of between 10% mortality due to multiple organ failure unrelated to the
and 21%, and neurologic deficits in 3.410%, including aortic rupture in up to 14% of cases [22, 23]. Since aortic
paraplegia and hemiparesis, as well as recurrent laryngeal ruptures are contained ruptures in most cases, other
nerve injury in up to 17% [8, 11, 16, 17]. However, there immediate life-threatening non-aortic injuries are usually
have been encouraging reports in the literature that such treated before the aortic stent-grafting procedure, depend-
patients might be favorably treated with endovascular ing on the individuals condition. The treatment of
stent-grafting, with a lower mortality and morbidity. In a contained aortic ruptures secondary to other life-threaten-
review of more than 1,500 thoracic aortic aneurysms, type ing trauma has been favored by several groups [22, 19, 23
B aortic dissections, and traumatic ruptures, the overall 26] and is our current strategy in those cases. However, we
technical success rate of endovascular treatment was 97%, have lost 1 hemodynamically unstable patient, with both
with a 30-day mortality rate of 5.5%. The rates of open thoracic and abdominal trauma, in whom emergency
conversion, neurologic deficits, endoleaks, graft migration, splenectomy was performed as the first-step procedure. He
and graft infection were less than 1.0%, 2.0%, 7.8%, 0.7%, has not been included in our statistics because endovas-
and 0.4%, respectively [18]. Our clinical cases confirm this cular treatment had not been performed. This patient died
experience, with excellent clinical results, particularly for during laparotomy because of an acute swelling of the soft
traumatic ruptures [1215]. However, in addition to the tissue of the cervical neck that was indicative of an acute
favorable immediate results, this paper summarizes the secondary rupture of the false aortic aneurysm that had
single-center results and probable complications during been diagnosed by the primary CT scan and was scheduled
long-term follow-up for a maximum of 13 years. as the second-step procedure after splenectomy. Therefore,
the decision is always an interdisciplinary responsibility
that has to be made on the basis of the individual clinical
Acute Traumatic Aortic Ruptures conditions and CT findings.
The mean follow-up of the reported literature data ran-
Patients with acute traumatic aortic ruptures have shown a ges between 13 and 46 months [1922, 2932] and
favorable outcome compared with historical data. Emer- demonstrates a low rate of procedure-related complications
gency open surgery has been reported to have a high [1921, 2732] with an acceptable rate of secondary pro-
periprocedural mortality of between 17% and 21%, and a cedures. Secondary complications included collapse of the
paraplegia rate of up to 7% [19, 20]. Our data demonstrate stent-grafts (3.6%) and type I endoleaks during follow-up
an excellent clinical outcome and durability of treatment (15.4%) that had to be treated with an additional stent-graft
results without stent-graft-related mortality or morbidity in [21, 31]. To achieve complete sealing of the rupture site
long-term follow-up. This is essential because many of requires the ostium of the left subclavian artery to be
these patients are young healthy men who, apart from their covered in many cases. However, this procedure has been

123
M. B. Pitton et al.: Long-Term Follow-Up After Endovascular Treatment of Acute Aortic Emergencies 33

shown to have an acceptably low rate of clinical claudi- and 1 of unknown causes. However, these authors sup-
cation of the left arm. In symptomatic cases, this might be posed intramural hematomas, ulcer-like lesions, and initial
treated secondarily by transposition of the subclavian aortic diameters of more than 4 cm to be predictors of
artery to the carotid artery or a carotid to subclavian bypass aortic events such as dissection-related death, rupture, new
[27]. In our series, no patient suffered from claudication of type A aortic dissection, and aneurysm formation [35]. The
the arm or other clinical problems in up to 12 years follow- rate of secondary aortic dilatation and aneurysm formation
up, which endorses the endovascular technique not only for might be influenced by the underlying disease, which needs
acute emergency situations but also for its long-term to be recognized in decision-making. In a multicenter trial,
outcome. Gaxotte et al. investigated the course of the aortic diameter
after stent-grafting of aortic dissections in 50 patients. The
rate of false lumen thrombosis at discharge and 15 months
Aortobronchial Fistulas and Penetrating Aortic Ulcers follow-up was 42% and 63%, respectively. At follow-up,
patients with Marfans syndrome demonstrated a lesser
Aortobronchial fistulas and acute bleeding from penetrat- diameter reduction of the false lumen and the entire aorta
ing ulcers were also favorably treated with endovascular [36].
stent-grafting. Our data demonstrate the immediate and The definitive indication and the correct time for en-
effective control of bleeding and acute hemodynamic sta- dovascular treatment of non-complicated type B dissection
bilization of those life-threatening situations. The follow- remain under investigation. The benefit of prophylactic
up is up to 13 years and thereby demonstrates an adequate stent-grafting of uncomplicated type B dissections has not
long-term result in this entity. Eggebrecht et al. reported on yet been proved. Moreover, there is some risk of peripro-
22 patients with penetrating aortic ulcers, 3 of whom had cedural retrograde type A dissection secondary to each
contained aortic rupture [33]. The technical success rate stent-grafting procedure (6.8%) [37]. This occurred in 1 of
was 96%, with 1 patient suffering a minor stroke. During our clinical cases, in which, fortunately, no neurologic
mean follow-up of 27 months, 1 patient needed retreatment complication occurred and no further treatment was nec-
with graft extension because of a type I endoleak. The 1-, essary. Undoubtedly, there is already a consensus for
2-, and 5-year survival rates were 100%, 82.5%, and endovascular stent-grafting of complicated acute type B
61.9%, respectively [33]. Recurrent hemoptysis has also dissections with true aortic lumen collapse and consecutive
been reported in those cases and stent-graft extension has ischemia of visceral organs or legs. This is consistent with
been recommended as adequate minimally invasive treat- our decision-making policy where the decision for endo-
ment option [32, 33]. In our series, secondary interventions vascular treatment was only made in those complications.
with stent-graft extensions were performed in half the These patients are in a compromised general condition with
patients, because of either secondary aortic wall ulcers multi-organ failure, and the peri-interventional complica-
caused by bare stent struts or recurrent hemoptysis due to tion rate might be very high. Other patients without
secondary leaks from the anchor sites. In 1 case, the secondary complications were treated with best clinical
recurrent symptoms occurred nearly a decade after the first care, control of arterial hypertension, and CT follow-up for
treatment, indicating that lifetime follow-up is mandatory secondary complications such as secondary aneurysm
in these patients. development. Dias et al. reported on the results of stent-
grafting in complicated type B dissection with rupture,
intractable pain or hypertension, acute bowel ischemia,
Acute Type B Dissections lower limb and renal ischemia, or paraplegia [38]. This
study found a 30-day mortality rate of 16% (5 of 31) and a
Endovascular treatment of aortic dissections has already late mortality rate of 19% (6 of 31) during a median follow-
been described in the literature with beneficial results [34, up of 22 months. The perioperative morbidity rate was 48%
35]. However, the indications for treatment have not been (15 of 31 patients) and included paraplegia in 1 (3%),
clearly defined. In a recent paper, Winnerkvist et al. stroke in 5 (16%), lower limb ischemia in 3 (10%), myo-
reported on the conservative course of 66 patients after cardial infarction in 3 (10%), pneumonia in 2 (7%), and
acute type B dissection without acute need for surgical colitis in 1 (3%) [38]. This is consistent with our mortality
intervention [35]. During a mean follow-up of 79 months and morbidity data, including 1 case of paraplegia, 1
the actuarial survival rates at 5 and 10 years were 82% and stroke, 1 aortic ulcer, and 1 renal ischemia. However, the
69%, respectively. In this study, 10 patients (15%) devel- paraplegia and the renal ischemia were pre-existing find-
oped aneurysms of more than 6 cm, 3 of whom died ings and therefore not directly related to the procedure,
because of aortic rupture and 2 of whom underwent sur- whereas the stroke and the aortic ulcer were stent-graft
gical repair. Of the remaining 56 patients, 1 died of rupture complications. Both fatal outcomes were due to mesenteric

123
34 M. B. Pitton et al.: Long-Term Follow-Up After Endovascular Treatment of Acute Aortic Emergencies

ischemia. Despite technically successful procedures with aortic syndromes of the descending aorta. Particularly in
sufficient restoration of mesenteric blood flow, the small traumatic aortic ruptures, the long-term results are excel-
intestine was already necrotic at this stage and the patients lent, with confirmed durability of the stent-grafts and no
died despite extensive small bowel resections. Obviously, stent-graft-related morbidity. In the other pathologic enti-
the stent-graft procedures were indicated too late in these ties, recurrent clinical symptoms might be due to stent-
cases. Therefore, as a general rule in such cases, once graft failure. Therefore, regular follow-up is mandatory in
ischemia of visceral organs is evident, a resolute decision order to identify late complications and to perform appro-
for stent-grafting is recommended in order to achieve priate minimally invasive corrections in a timely manner.
adequate restoration of blood flow in time.

References
Symptomatic Thoracic Aneurysms
1. Pate J, Fabian T, Walker W (1995) Acute traumatic rupture of the
aortic isthmus: Repair with cardiopulmonary bypass. Ann Thorac
Endovascular emergency treatment of symptomatic and
Surg 59:9099
penetrating aneurysms of the descending aorta has also 2. Cardarelli M, McLaughlin J, Downing S, Brown JM, Attar S,
been technically successful. However, there was one peri- Griffith BP (2002) Management of traumatic aortic rupture: A 30-
procedural death due to myocardial infarction resulting in a year experience. Ann Surg 236:465470
3. Ahmad F, Cheshire N, Hamady M (2006) Acute aortic syndrome:
30-day mortality rate of 11%. Moreover, our data demon-
Pathology and therapeutic strategies. Postgrad Med J 82:305312
strated a high endoleak rate after stent-grafting and 4. Volodos NL, Karpovich IP, Shekhanin VE, Troian VI, Iakovenko
aneurysm growth in1 case. Appoo et al. reported on the LF (1988) A case of distant transfemoral endoprosthesis of the
results of large thoracic aneurysms treated with stent- thoracic aorta using a self-fixing synthetic prosthesis in traumatic
aneurysm. Grudn Khir 6:8486
grafting in 99 patients and demonstrated a 30-day mortality
5. Dake MD, Miller DC, Mitchell RS, Semba CP, Moore KA, Sakai
rate of 5% and permanent spinal ischemia in 2% [39]. In T (1998) The first generation of endovascular stent-grafts for
this series, 20% of patients received subclavian carotid patients with aneurysms of the descending thoracic aorta. J
transposition or bypass before stent-grafting in order to Thorac Cardiovasc Surg 116:689704
6. Morrissey NJ, Yano OJ, Soundararajan K, Eisen L, McArthur C,
create an adequate landing zone for the stent-graft. The 1-,
Teodorescu V, Kerstein M, Hollier L, Marin ML (2001) Endo-
3-, and 5-year survival were 84.5%, 70.5%, and 52.4%, vascular repair of paraanastomotic aneurysms of the aorta and the
respectively [39]. Similar results have been reported by iliac arteries: Preferred treatment for a complex problem. J Vasc
others; however, there is a considerable rate of endole- Surg 34:503512
7. Nienaber CA, Fattori R, Lund G, Dieckmann C, Wolf W, von
aks [4042]. Most of these data were obtained in elective
Kodolitsch Y, Nicolas V, Pierangeli A (1999) Nonsurgical
endovascular treatments, which might be considerably reconstruction of thoracic aortic dissection by stent-graft place-
different from emergency treatment. We have to consider ment. N Engl J Med 340:15391545
significant comorbidities in such acute cases which might 8. Dake MD, Kato N, Mitchell RS, Semba CP, Ravazi MK, Shi-
mono T, Hirano T, Takeda K, Yada I, Miller DC (1999)
contribute to a considerable morbidity in terms of cardio-
Endovascular stent-graft placement for treatment of acute aortic
pulmonary or vascular events. Those cardiovascular events dissection. N Engl J Med 340:15461552
were responsible for late mortality in our patients [22]. 9. von Oppell UO, Dunne TT, DeGroot MK, Zilla P (1994) Trau-
With respect to long-term outcome, there is a high rate of matic aortic rupture: Twenty year metaanalysis of mortality and
risk of paraplegia. Ann Thorac Surg 58:585593
endoleaks. Whereas type I and III endoleaks have the
10. von Fricken K, Karamanoukian HL, Ricci M, Taheri A, Bergs-
potential for endovascular correction, this is not the case in land J; Salerno TA (2000) Aortobronchial fistula after
type II endoleaks. Because of the possible neurologic endovascular stent graft repair of the thoracic aorta. Ann Thorac
complications, and in contrast to abdominal aortic aneu- Surg 70:14071409
11. von Segesser LK, Genoni M, Kunzli A, Lachat M, Niederhauser
rysms, type II endoleaks are hardly treatable and might be
U, Vogt P, Schonbeck M, Turina M (1996) Surgery for ruptured
responsible for continued aneurysm growth during follow- thoracic thoraco-abdominal aortic aneurysms. Eur J Cardiothorac
up. The outcome of thoracic aortic aneurysms is therefore Surg 10:9961001
less favorable compared with abdominal aortic aneurysms, 12. Duber C, Schmiedt W, Pitton MB, Neufang A, Eberle B, Oelert
H, Thelen M (1996) Endovaskulare Therapie aortaler Aneurys-
where complete aneurysm exclusion and shrinkage can be
men: Erste klinische Ergebnisse. Fortschr Rontgenstr 12:5561
supported by embolization of aortic side branches [43]. Our 13. Dorweiler B, Dueber C, Neufang A, Schmiedt W, Pitton MB,
emergency treatment was, however, effective in terms of Oelert H (2001) Endovascular treatment of acute bleeding com-
prevention of aneurysm penetration and rupture and, in this plications in traumatic aortic rupture and aortobronchial fistula.
Eur J Cardiothorac Surg 19:739745
respect, confirms the experience from the literature.
14. Pitton MB, Schmiedt W, Neufang A, Herber S, Duber C, Thelen
In summary, endovascular treatment is safe and effec- M (2002) Endovaskulare Behandlung akuter Erkrankungen der
tive for emergency treatment of life-threatening acute thorakalen Aorta. Fortschr Rontgenstr 174:593599

123
M. B. Pitton et al.: Long-Term Follow-Up After Endovascular Treatment of Acute Aortic Emergencies 35

15. Pitton MB, Duber C, Neufang A, Schlegel J (2002) Endovascular 30. Saratzis NA, Saratzis AN, Melas N, Ginigs G, Lioupis A, Lyk-
repair of aortic rupture: Acute non-contained aortic rupture of a opoulos D, Lazaridis J, Dimitrios K (2007) Endovascular repair
penetrating aortic ulcer. Cardiovasc Intervent Radiol 25:6467 of traumatic rupture of the thoracic aorta: Single-center experi-
16. Baguley CJ, Sibal AK, Alison PM (2005) Repair of injuries to the ence. Cardiovasc Intervent Radiol 30:370375
thoracic aorta and great vessels: Auckland, New Zealand 1995 31. Neuhauser B, Czermak B, Jaschke W, Waldenberger P, Fraedrich
2004. ANZ J Surg 75:383387 G, Perkmann R (2004) Stent-graft repair for acute traumatic
17. Brandt M, Hussel K, Walluscheck KP, Boning A, Rahimi A, thoracic aortic rupture. Ann Surg 70:10391044
Cremer J (2005) Early and long-term results of replacement of 32. Amabile P, Rollet G, Vidal V, Collart F, Bartoli JM, Piquet P
the descending aorta. Eur J Vasc Endovasc Surg 30:365369 (2006) Emergency treatment of acute rupture of the descending
18. Sayed S, Thompson MM (2005) Endovascular repair of the thoracic aorta using endovascular stent-grafts. Ann Vasc Surg
descending thoracic aorta. Evidence for the change in clinical 20:723730
practice. Vascular 13:148157 33. Eggebrecht H, Herold U, Schmermund A, Lind AY, Kuhnt O,
19. Rousseau H, Dambrin C, Marcheix B, Richeux L, Mazerolles M, Martini S, Kuhl H, Kienbaum P, Peters J, Jakob H, Erbel R,
Cron C, Watkinson A, Mugniot A, Soula P, Chabbert V, Canevet Baumgart D (2006) Endovascular stent-graft treatment of pene-
G, Roux D, Massabuau P, Meites G, Tran Van T, Otal P (2005) trating aortic ulcer: Results over a median follow-up of 27
Acute traumatic aortic rupture: A comparison of surgical and months. Am Heart J 151:530536
stent-graft repair. J Thorac Cardiovasc Surg 129:10501055 34. Sharzaglia P, Lovato L, Buttazzi K, Russo V, Renzulli M, La
20. Kuhne CA, Ruchholtz S, Voggenreiter G, Eggebrecht H, Paffrath Palombara C, Fattori R (2006) Interventional techniques in the
T, Waydhas C, Nastkolb D, AG Polytrauma DGU (2005) Trau- treatment of aortic dissection. Radiol Med (Torino) 111:585596
matic aortic injuries in severely injured patients. Unfallchirurgie 35. Winnerkvist A, Lockowandt U, Rasmussen E, Radegran K (2006)
108:279287 A prospective study of medically treated acute Type B aortic
21. Hoornweg LL, Dinkelman MK, Goslings JC, Reekers JA, Ver- dissection. Eur J Vasc Endovasc Surg 32:349355
hagen HJ, Verhoeven EL, Schurink GW, Balm R (2006) 36. Gaxotte V, Thony F, Rousseau H, Lions C, Otal P, Willoteaux S,
Endovascular management of traumatic ruptures of the thoracic Rodiere M, Negaiwi Z, Joffre F, Beregi JP (2006) Midterm
aorta: A retrospective multicenter analysis of 28 cases in The results of aortic diameter outcomes after thoracic stent graft
Netherlands. J Vasc Surg 43:10961102 implantation for aortic dissection: A multicenter study. J Endo-
22. Agostinelli A, Saccani S, Borrello B, Nicolini F, Larini P, Gherli vasc Ther 13:127138
T (2006) Immediate endovascular treatment of blunt aortic 37. Neuhauser B, Czermak BV, Fish J, Perkmann R, Jaschke W,
injury: Our therapeutic strategy. J Thorac Cardiovasc Surg Chemelli A, Fraedrich G (2005) Type A dissection following
131:10531057 endovascular thoracic aortic stent-graft repair. J Endovasc Ther
23. Desgranges P, Mialhe C, Cavillon A, Alric P, Ovarfordt P, 12:7481
Kobeiter H, Mathieu D, Melliere D, Becquemin JP (1997) En- 38. Dias NV, Sonesson B, Koul B, Malina M, Ivancev K (2006)
dovascular repair of posttraumatic thoracic pseudoaneurysm with Complicated acute type B dissections: An 8 years experience of
stent graft. AJR Am J Roentgenol 169:17431745 endovascular stent-graft repair in a single center. Eur J Vasc
24. Kato N, Dake MD, Miller DC, Semba CP, Mitchell RS, Razavi Endovasc Surg 31:481486
MK, Kee ST (1997) Traumatic thoracic aortic aneurysm: Treat- 39. Appoo JJ, Moser WG, Fairman RM, Cornelius KF, Pochettino A,
ment with endovascular stent-grafts. Radiology 205:657662 Woo EY, Kurichi JE, Carpenter JP, Bavaria JE (2006) Thoracic
25. Cowley RA, Turney SZ, Hankins JR, Rodriguez A, Attar S, aortic stent grafting: Improving results with newer generation
Ahankar BS (1990) Rupture of thoracic aorta caused by blunt investigational devices. J Thorac Cardiovasc Surg 131:10871094
chest trauma: A fifteen-year experience. J Thorac Cardiovasc 40. Greenberg RK, ONeill S, Walker E, Haddad F, Lyden SP,
Surg 100:652661 Svensson LG, Lytle B, Clair DG, Ouriel K (2005) Endovascular
26. Kipfer B, Leupi F, Schuepbach P, Friedli D, Althaus U (1994) repair of thoracic aortic lesions with the Zenith TX1 and TX2
Acute traumatic rupture. Eur J Cardiothorac Surg 8:3033 thoracic grafts. Intermediate-term results. J Vasc Surg 41:589596
27. Lawlor DK, Ott M, Forbes TL, Kribs S, Harris KA, DeRose G 41. Farber MA, Criado FJ (2005) Endovascular repair of nontrau-
(2005) Endovascular management of traumatic thoracic aortic matic ruptures thoracic aortic pathologies. Ann Vasc Surg
injuries. Can J Surg 48:293297 19:167171
28. Michelet P, Roch A, Amabile P, Perrin G, Fulachier V, Piguet P, 42. Eggebrecht H, Schmermund A, Herold U, Baumgart D, Martini
Auffray JP (2005) Endovascular treatment of isthmic aortic S, Kuhnt O, Ling AY, Kuhne C, Kuhl H, Kienbaum P, Peters J,
rupture: Use of second generation stent grafts. Ann Fr Anesth Jakob HG, Erbel R (2005) Endovascular stent-graft placement for
Reanim 24:355360 acute and contained rupture of the descending thoracic aorta.
29. Attia C, Villard J, Boussel L, Farhat F, Robin J, Revel D, Douek Catheter Cardiovasc Interv 66:474482
P (2007) Endovascular repair of localized pathological lesions of 43. Pitton MB (2006) Diagnosis and management of endoleaks after
the descending thoracic aorta: Midterm results. Cardiovasc In- endovascular aneurysm repair: The role of MRI. Abdom Imaging
tervent Radiol May 13 (Epub ahead of print) 31:339346

123
Cardiovasc Intervent Radiol (2008) 31:3642
DOI 10.1007/s00270-007-9167-6

CLINICAL INVESTIGATION

Endovascular Exclusion of Visceral Artery Aneurysms with Stent-


Grafts: Technique and Long-Term Follow-up
Michele Rossi Alberto Rebonato Laura Greco
Michele Citone Vincenzo David

Received: 1 October 2006 / Accepted: 7 March 2007 / Published online: 6 October 2007
 Springer Science+Business Media, LLC 2007

Abstract This paper describes four cases of visceral frequently present as a life-threatening, often fatal, emer-
artery aneurysms (VAAs) successfully treated with endo- gency when associated with rupture and intra- or
vascular stent-grafts and discusses the endovascular retroperitoneal bleeding [2]. The clinical symptoms, natu-
approach to VAAs and the long-term results. Four balloon ral history, and mortality of VAAs vary depending on the
expandable stent-grafts were used to treat three splenic vessels involved, whereas the mortality rate ranges from
artery aneurysms and one bleeding common hepatic artery 10% to 25% and, in pregnant women, up to 75% [2].
pseudoaneurysm. The percutaneous access site and the Before the evolution of minimally invasive endovascular
materials were chosen on the basis of CT angiography techniques such as embolotherapy, VAAs were treated
findings. In all cases the aneurysms were successfully surgically. Currently transcatheter embolization is a widely
excluded. In one case a splenic infarction occurred, with employed endovascular option with a high short-term
nonrelevant clinical findings. At 16- to 24-month follow-up success rate, without the concomitant morbidity of major
three patients had patent stents and complete exclusion and abdominal surgery [3]. Recently the use of stent-grafts to
shrinkage of the aneurysms. One patient died due to pan- exclude aneurysms in the splenic as well as other visceral
creatitis and sepsis, 16 days after successful stenting and arteries has been presented in single case reports [46]. In
exclusion of a bleeding pseudoaneurysm. We conclude that this paper we report four successful endovascular repairs
endovascular treatment using covered stent-grafts is a valid with stent-grafts for aneurysm of the splenic artery in three
therapeutic option for VAAs. Multislice CT preoperative patients and for a common hepatic bleeding pseudoaneu-
study helps in planning stent-graft positioning. rysm in one patient. With the exception of one who died 16
days after the procedure, all patients were followed up for a
Keywords Stent-grafts  Endovascular procedure  period ranging from 16 to 24 months after intervention.
Visceral aneurysm  CT angiography

Materials and Methods


Introduction
Four patients were treated between February 2004 and July
Although rare, aneurysms of the visceral arteries (VAAs) 2005. Table 1 lists the main data regarding the patients and
represent an important vascular disease. The most common pathological findings. The common hepatic aneurysm was
sites are the splenic, hepatic, superior mesenteric, gastro- discovered during angiography for assessment of upper
duodenal, and small pancreatic arteries [1]. They gastrointestinal tract bleeding. The pathogenesis of the
three splenic aneurysms was unknown, and on admission
the patients were asymptomatic. CT angiography using 16-
M. Rossi  A. Rebonato (&)  L. Greco  M. Citone  V. David slice multirow detector computed tomography (MDCT;
Interventional Radiology Unit, Radiology Department,
Philips MDT MX 8000) with volume rendering (VR),
S. Andrea, University Hospital La Sapienza,
1035 Via di Grottarossa, Rome 00189, Italy multiplanar reconstruction (MPR), and advanced vessel
e-mail: albertorebonato@libero.it analysis (AVA) was performed before the procedure for the

123
M. Rossi et al.: Endovascular Exclusion of Visceral Artery Aneurysms with Stent-Grafts 37

three splenic aneurysms. The investigation provided aneurysm. MDCT scan at 16 months showed patency of the
information about the sizes of the aneurysms (diameter, stent and shrinkage of the aneurysm.
length of the neck, distance from the aorta, diameter of the
afferent and efferent arteries) as well as the relationship
with surrounding tissues, anatomy, and possible additional Case 2
pathologies of the celiac trunk or its branches. At the start
of the vascular procedure a bolus of 2500 U of heparin was The MDCT study showed a saccular aneurysm located at
given and a second 2500 U was preloaded in a continuous the origin of the splenic artery (Fig. 2A). Right transfe-
saline flush and used as necessary. No heparin was given to moral artery access was chosen and celiac trunk
the patient with the bleeding pseudoaneurysm. The mean catheterization and multiple angiograms were performed
fluoroscopy time was 14 min, and the maximum amount of with a reshaped MP 8-Fr sheath (Fig. 2B). With a regular
contrast medium administered was 180 ml. All patients Terumo guide wire and a 5-Fr MP catheter (Cordis),
received prophylactic antibiotic therapy for infection to catheterization of the efferent artery was performed, and
stent-graft or to necrotic tissue in the case of embolic then the hydrophilic guide wire was exchanged for a long
infarcts, and a regimen of lifelong acetylsalicylic acid stiff wire (Amplatz; Boston Scientific) to give stability to
administration (300 mg/day) after the procedure. At the the system. A 38-mm-long Jostent peripheral stent-graft
start of the vascular procedure the patients were informed (Abbott) was advanced and inflated, bridging the aneu-
about the procedure and written consent was obtained. The rysmal neck (Fig. 3A). After removal of the stiff guide
follow-up of the patients was based on clinical, CT, and wire, moderate abdominal pain felt previously by the
ultrasound (US) data. patient decreased definitively, but a persistent tenderness
was felt for about 12 h. MDCT scan documented segmental
infarction of the spleen. No changes in blood chemistry
Case 1 were found following the procedure and the patient was
discharged after 3 days. MDCT angiography was per-
The origin of the celiac trunk was stenotic with a postste- formed 1 month later (Fig. 3B) and demonstrated precise
notic dilatation (Figs. 1A and B). The diameter of the stent localization, complete aneurysm thrombosis, and
splenic artery was *5 mm. On the basis of the CT findings normal splenic artery branching. At 12- and 24- month CT
we used a left axillary approach. A stable carrier system follow-up these findings were unchanged and the stentgraft
with a 6-Fr guiding catheter and a 0.014-in. wire was was patent. The aneurysm almost disappeared and the
placed, allowing contrast medium injection for angio- splenic infarctual alterations were not evident anymore
graphic evaluation. A coronary stent-graft (26 5-mm (Figs. 3C and D).
Graft Master; Jostent) was positioned, covering the length
of the neck (Fig. 1D). Perfusion of the spleen was pre-
served and no complication occurred during or after the Case 3
procedure. MDCT scan confirmed the angiographic suc-
cessful findings. The patient was discharged uneventfully 3 MDCT confirmed the presence of a saccular 3-cm aneu-
days later. MDCT scan after 4 months and US at 12 months rysm in the intermediate segment of the splenic artery
showed complete exclusion and thrombosis of the (Figs. 4A and B). A 7-Fr introducer sheath (Flexor; Cook)

Table 1 Patient data and pathologic findings


Case 1 Case 2 Case 3 Case 4

Age/gender 76/F 55/F 67/female 65/M


Aneurysm diameter 3.5 cm 3 cm 3 cm 3 cm
Location Splenic Splenic Splenic Hepatic
Approach Transhumeral Transaxillary Transaxillary Transaxillary
Carrier system 6-Fr guiding catheter 8-Fr introducer sheath 7-Fr introducer sheath 8-Fr introducer sheath
Guide wire Choice, 0.014 in. Terumo, 0.035 in. Terumo, 0.035 in. Terumo, 0.035 in.
Stent-graft Graft Master Jostent, 5 26 mm Jomed, 38 mm Advanta, 7 59 mm Jomed, 38 mm
Follow-up 18 mo 28 mo 18 mo 16 days exitus
Patency Yes Yes Yes Yes
Aneurysm thrombosis Yes, complete Yes, complete Yes, complete Yes, complete

123
38 M. Rossi et al.: Endovascular Exclusion of Visceral Artery Aneurysms with Stent-Grafts

Fig. 1 Case 1. MDCT scan and VR reconstruction in lateral (A) and is above the aneurysm and a 5-Fr catheter is withdrawn, leaving a
anteroposterior (B) view. The 3.5-cm aneurysm with a 2-cm neck is 0.14-in. guide wire in the efferent artery. (D) Angiography after
located in the midtract of the splenic artery. A stenosis of the celiac placement of the coronary stent-graft (26 5-mm Graft Master;
axis with a poststenotic dilatation and tortuousity is also present Jostent) shows exclusion of the aneurysm and preserved blood flow
(arrow). Two additional subcentimetric aneurysms were present more through the stent-graft and the splenic artery. There is some residual
distally. (C) A selective angiogram shows that the guiding catheter tip contrast medium in the sac indicating lack of washout

Fig. 2 Case 2. VR MDCT scan (A) and transfemoral selective celiac trunk arteriography (B) confirm the large aneurysm located just at the
origin of the splenic artery, with a very short proximal neck

was placed through a left transaxillary approach in the aneurysmal segment. Final angiograms confirmed that the
proximal splenic artery. The regular guide wire was aneurysm was excluded, the splenic artery was patent, and
exchanged first for a stiff guide wire (Terumo) and then for the splenic parenchymal vascularization was normal
an Amplatz superstiff guide wire (Boston Scientific) that (Fig. 4D). The patient was discharged 3 days later. MDCT
provided enough rigidity to the carrier system. A pre- at 1 and 18 months demonstrated complete thrombosis of
mounted, balloon expandable, 7 59-mm Advanta stent- the aneurysm and preserved normal splenic flow and per-
graft system (V12; Atrium) was advanced and placed in the fusion of the spleen (Figs. 4E and F).

123
M. Rossi et al.: Endovascular Exclusion of Visceral Artery Aneurysms with Stent-Grafts 39

Fig. 3 Case 2. (A)


Postprocedural angiography
documents complete exclusion
of the aneurysm. (B) Axial
MDCT at 1 month shows
optimal positioning of the stent,
with exclusion of the aneurysm.
Residual contrast is still seen in
the aneurysm from the
injections during the procedure.
(C), (D) MPR and AVA MDCT
reconstruction 24 months
document optimal placement of
the stent just at the origin of the
splenic artery, without
protrusion into the celiac trunk
or the hepatic artery, and
confirm the patency of the stent,
with regular splenic flow and
shrinkage of the aneurysm

Case 4 graft were assessed, without evidence of bleeding. Subse-


quently the patients clinical condition worsened. He died
A 65-year-old man with pancreatic head malignancy of sepsis and pancreatitis 16 days later.
underwent pancreatic-duodenal resection. In the postoper-
ative period he developed pancreatitis and sepsis and was
discharged after 1 month of medical therapy. Three months Discussion
later the patient was readmitted for gastrointestinal bleed-
ing with hematemesis and hypovolemic shock and was Splenic artery aneurysms comprise approximately 60% of
referred for angiographic evaluation. Contrast injection all VAAs and have been reported as incidental findings in
showed a pseudoaneurysm in the distal tract of the com- 0.78% of arteriograms as well as 0.1% to 10.4% of
mon hepatic artery, probably at the origin of the residual, autopsies [7]. The causes of visceral aneurysm include
surgically ligated gastroduodenal artery (Fig. 5A). After atherosclerosis, medial degeneration, infection, fibro-
selective catheterization with a 4-Fr MP catheter (Cordis), muscolar dysplasia, and congenital anomalies [8].
the catheter was exchanged over an Amplatz guide wire Although arteriosclerosis is a common histological finding
(Boston Scientific) with an 8-Fr introducer sheath (Super of splenic artery aneurysm, degeneration of the media of
Arrow; Flex PA). Subsequently a 38-mm-long Jostent the splenic artery secondary to hormonal changes and
peripheral, balloon expandable stent-graft, 6/12 mm increased splenic blood flow associated with pregnancy is
(Abbott), was mounted over a 5 40-mm Wanda angio- usually an important factor [9]. False aneurysms of the
plasty balloon (Boston Scientific) and advanced to the neck splenic artery seem to be less frequent than aneurysms and
of the pseudoaneurysm. The stent-graft was expanded and may be posttraumatic or mycotic, but are most often sec-
the angiogram showed exclusion of the bleeding stump, ondary to pancreatitis [1].Independent of the etiology, the
with incomplete visualization of the right hepatic artery, natural history of most VAAs appears to be expansion and
probably due to procedure-related vessel injury or distal eventual rupture, with life-threatening consequences. Pre-
embolization. Angiograms showed good flow through the dictive factors for rupture have been described in the
stent. The bleeding was stopped (Fig. 5B).Over the next literature: absence of calcification in the wall, young age,
days the hemoglobin level increased from 4 to 9 mg/dl. A and patients not taking b-blockers [10]. The mortality after
CT evaluation was performed 6 days after the procedure rupture rises to 35% for hepatic aneurysms and 50% for
for clinical evidence of pancreatitis. CT showed findings of gastroduodenal aneurysms [7]. Because of the unpredict-
peripancreatic fluid collection and a small infarct in the able course of these lesions and the high mortality rate in
right liver. Thrombosis of the aneurysm and patency of the the event of rupture, we believe, according to other authors,

123
40 M. Rossi et al.: Endovascular Exclusion of Visceral Artery Aneurysms with Stent-Grafts

Fig. 4 Case 3. MDCT with VR


reconstruction in lateral (A) and
axial (B) view showing a
saccular aneurysm located in the
intermediate segment of the
tortuous splenic artery. Note the
broad-based neck (18 mm) and
the perspective of the efferent
and afferent artery of the
aneurysm. C Preoperative
transbrachial selective
arteriography of the splenic
artery. After placement of a
balloon expandable stent-graft
(Advanta) the control (D) shows
exclusion of the aneurysm, with
flow in the distal splenic artery.
Residual contrast from the
previous injections is still seen.
CT scan with VR reconstruction
at 1 month (E) and AVA
reconstruction at 18 months (F)
after the procedure, showing
exclusion of the aneurysm and
patency of the stent-graft

Fig. 5 Case 4. (A) The


emergent celiac angiogram
showed a pseudoaneurysm in
the common hepatic artery
approximately at the origin of
the residual gastroduodenal
artery. (B) The control after
placement of a Jostent stent-
graft showed complete
exclusion of the
pseudoaneurysm

that an aggressive therapeutic approach is indicated, pregnant, or measures 2 cm. A threshold diameter of 2 cm
especially in young patients [6]. Trastek et al. [11] advo- is most often referred to in the literature [12]. The treat-
cate treating any aneurysm that is symptomatic, has grown, ment of false aneurysms must be undertaken without delay
is found in a woman who is pregnant or expecting to be whatever their size, even if there is no bleeding, because

123
M. Rossi et al.: Endovascular Exclusion of Visceral Artery Aneurysms with Stent-Grafts 41

the risk of spontaneous rupture is very high (mortality rate embolization is planned, and it is always challenging. A
close to 100% due to risk of rupture) [1]. Surgical bipolar hydrophilic guide wire is of extreme help before engaging
ligature is a treatment option [7]. However, surgery carries the efferent artery and then advancing more distally. Other
a considerable morbidity and mortality of 1%2% and major technical considerations for proper repair of VAAs
10%, respectively. During the last decade, endovascular by stent-graft are (a) to have a stable carrier system for
techniques have offered different alternatives for treating safely advancing the stent-graft, (b) to choose a stent-graft
these aneurysms: coil embolization, detachable balloons, with the appropriate dimensions and mechanical properties,
gelfoam, thrombin, or a combination of the above [12]. On and (c) to release the stent in the proper position. Our
the basis of the literature reports the coil-embolization feelings about these technical aspects are as follows. (a) A
technique is the most common and the most successful stiff guide wire is mandatory to reduce tortuosities and
[12]. Precise choice of the site of occlusion is necessary in ensure that the delivery system is stable. (b) In our expe-
order to preserve splenic artery collateral branches and rience the preprocedural MDCT angiography evaluation
spare the spleen. The principal consists of packing the with VR, MPR, and AVA provides fundamental informa-
lumen of the aneurysm with embolizing agents or exclud- tion for planning the treatment strategy. With the AVA
ing the neck. Guillon et al. describe two methods: (a) the analysis we obtain the precise dimension of the aneurysm
sandwich method, with an upstream and downstream (length of the neck, distance from the aorta, diameter of the
occlusion, equivalent to an exclusion-ligature; and (b) the afferent and efferent arteries), tortuosity of the arteries, and
packing method, filling of the aneurysmal sac with coils angulations of the aneurysmal tract, so that we can select a
and then proximal occlusion of the trunk of the splenic stent-graftcompatible with the anatomic situation. (c) The
artery. best approach, from above or below, is also suggested by
Coil embolization has been an accepted therapeutic three-dimensional CT imaging, which allows us also to
option in the past because the risk of splenic infarction and/ understand the spatial relationship between the aneurysm
or abscess formation is relatively low [6]. Reduced blood and the splenic artery tract. The ideal stent-graft for
supply to the spleen is compensated by collaterals from the VAA repair should be low-profile, relatively flexible (to
gastroepiploic arcade. Sometimes a postembolization syn- advance in small-caliber carrier devices), and adaptable to
drome (abdominal pain, elevation of pancreatic enzymes) different vessel diameters and aneurysm lengths. The rel-
is observed [12].One of the major problems after coil ative stiffness of the balloon expandable stent-grafts used
embolization is the occurrence of recanalization, with an in our cases did not cause relevant anatomic deformations
incidence of 9% to 42.9% [13, 14]. Guillon et al. reported of the vessels. For small vessels and short aneurysmal
that in 9 of 12 cases (75%), the intervention was considered necks we believe that using a low-profile, balloon
entirely satisfactory (aneurysm excluded, spleen still per- expandable stent-graft on a 6-Fr guiding catheter is a safe,
fused): 7 were complete immediate successes and 2 low-invasive, and very effective approach (case 1). In
successes with a second embolization, after 1 and 4 conclusion, this technique is shown to be a safe and
months. One treatment was unsuccessfully performed with effective option for treatment of VAAs, fully respective of
a stent-graft and required surgical intervention. the vascular anatomy of that area. We believe that this new
More recently the use of a stent-graft to completely therapeutic approach of stent-graft for VAAs is an effec-
exclude the aneurysm has been reported as a valid alter- tive, safe, and durable treatment option. Complete stent-
native to the other endovascular technique [5, 6].The main graft patency in these vascular territories has been shown at
advantage of using a stent-graft, in comparison with coil mid- to long-term follow-up of our patients. Longer follow-
packing, is the preservation of the splenic artery and its up data need to be collected. However, if a stenosis or
abundant collateral branches. The risk of splenic emboli- obstruction of the stent-graft should occur, it would have
zation with this technique is linked to the manipulation of the same effect as a bipolar percutaneous embolization.
the devices (clot formation or air embolism on the carrier
sheath), especially when inaccurate maneuvers are per-
References
formed. We had this kind of complication in one patient,
clearly shown by CT scan. She experienced only moderate 1. McDermott VG, Shlansky-Goldberg R, Cope C, et al. (1994)
abdominal pain for about 12 h, easily controlled by anal- Endovascular management of splenic artery aneurysm and
gesic drugs, without other physical or biohumoral pseudo-aneurysm. Cardiovasc Interv Radiol 17(4):179184
alterations. Her symptoms had completely regressed at the 2. Chiesa R, Astore D, Guzzo G, et al. (2005) Visceral artery
aneurysms. Ann Vasc Surg 19(1):4248
1-month CT control. Technically the preliminary step is to 3. Salam TA, Lumsden AB, Martin LG, et al. (1992) Non-operative
advance guide wires and catheters distally into the efferent management of visceral aneurysms and pseudoaneurysms. Am J
artery. This maneuver is equally required if a sandwich Surg 164(3):215219

123
42 M. Rossi et al.: Endovascular Exclusion of Visceral Artery Aneurysms with Stent-Grafts

4. Venturini M, Angeli E, Salvioni M, et al. (2002) Haemorrhage 10. Stone WM, Abbas M, Cherry KJ, et al. (2002) Superior mesen-
from a right hepatic artery pseudoaneurysm: endovascular treat- teric artery aneurysms: Is presence an indication for intervention?
ment with a coronary stent-graft. J Endovasc Ther 9(2):221224 J Vasc Surg 36(2):234237
5. Hyun-Ki Y, Mats L, Petr U, et al. (2001) Stent-graft repair of a 11. Trastek VF, Pairolero PC, Joyce JW, et al. (1982) Splenic artery
splenic artery aneurysm. Cardiovasc Interv Radiol 24(3):200203 aneurysms. World J Surg 91(6):694699
6. Gabelmann A, Gorich J, Merkle EM (2002) Endovascular treat- 12. Guillon R, Garcier JM, Abergel A, et al. (2003) Management of
ment of visceral artery aneurysms. J Endovasc Ther 9(1):3847 splenic artery aneurysms and false aneurysms with endovascular
7. Stanley JC, Wakefield TW, Graham LM, et al. (1986) Clinical treatment in 12 patients Cardiovasc Interv Radiol 26(3):256260
importance and management of splanchnic artery aneurysms. J 13. Sofocleous CT,Hinrichs CR, Hubbi B, et al. (2005) Embolization
Vasc Surg 3:836840 of isolated lumbar artery injuries in trauma patients. Cardiovasc
8. Grego FG,Lepidi S, Ragazzi R, et al. (2003) Visceral artery Interv Radiol 28(6):730735
aneurysm: a single center experience. Cardiovasc Surg 11:1925 14. Sessa C, Tinelli G, Porcu P, et al. (2004) Treatment of visceral
9. Motohiro S, Izumi A, Masayuki Y, et al. (2006) Splenic artery artery aneurysms: description of a retrospective series of 42
aneurysm of the anomalous splenomesenteric trunk: successful aneurysms in 34 patients. Ann Vasc Surg 18(6):695703
treatment by transcatheter embolization using detachable coils.
Cardiovasc Interv Radiol 29(3):432434

123
Cardiovasc Intervent Radiol (2008) 31:4348
DOI 10.1007/s00270-007-9222-3

CLINICAL INVESTIGATION

Primary Self-Expandable Nitinol Stent Placement in Focal


Lesions of Infrarenal Abdominal Aorta: Long Term Results
Jarmila Lastovickova Jan H. Peregrin

Received: 8 February 2007 / Accepted: 28 September 2007 / Published online: 30 October 2007
Springer Science+Business Media, LLC 2007

Abstract Introduction
Purpose To evaluate the technical and clinical success,
safety and long term results of percutaneous transluminal Percutaneous transluminal angioplasty (PTA) and stent
angioplasty/self-expandable nitinol stent placement of in- placement are accepted as a standard method of treatment
frarenal abdominal aorta focal lesions. for patients with significant stenoses and occlusions in the
Materials and Methods Eighteen patients underwent PTA iliac, femoropopliteal, and infrapopliteal regions. Reports
of focal atherosclerotic occlusive disease of distal abdom- on endovascular treatment of atherosclerotic lesions of the
inal aorta. Two symptomatic occlusions and 16 stenoses in abdominal aorta are rare. Focal stenoses or short occlusions
10 male and 8 female patients (mean age 68.2 years) were of the infrarenal aorta are a relatively infrequent entity;
treated with primary self-expandable nitinol stent atherosclerotic lesions often involve aortic bifurcation and
placement. iliac arteries [1, 2]. Treatment of these complex and more
Results Primary self-expandable nitinol stent placement extensive lesions is usually surgical.
was technically successful in all 18 procedures; clinical Surgical endarterectomy or aortobifemoral (aortobiiliac)
success was achieved in 100% of patients. No complica- bypass graft as a conventional method of treatment has also
tions associated with the procedure occurred. During the been used for patients with focal aortic stenoses and short
49.4 months of mean follow up (range 396, 4 months) all occlusions. The first PTA of the abdominal aorta was
treated aortic segments remained patent. performed in 1980 [3, 4] using the kissing-balloon
Conclusions Endovascular treatment (primary self- technique. During the 1980s, a limited number of patients
expandable nitinol stent placement) of focal atherosclerotic with occlusive disease of the distal abdominal aorta were
lesions of distal abdominal aorta is a safe method with treated with PTA and with stent implantation when PTA
excellent primary technical and clinical success rates and failed. Primary stent placement was later used to treat
favourable long term results. irregular or calcified lesions [4, 5]. The number of patients
with distal aortic occlusive disease treated with PTA and
Keywords Aorta  stenosis or obstruction  stent placement is still limited, and reports in the literature
Aorta transluminal angioplasty  Stents of the long-term results of primary stent implantation to the
distal abdominal aorta are rare [3].

Materials and Methods


J. Lastovickova  J. H. Peregrin
Department of Diagnostic and Interventional Radiology,
Institute for Clinical and Experimental Medicine, Eighteen patients with focal occlusive disease of the distal
Prague, Czech Republic abdominal aorta underwent endovascular treatment in our
institution from January 1999 to October 2006. No patients
J. Lastovickova (&)
Vdenska 1958/9, 140 21, Prague 4, Czech Republic with lesions involving aortic bifurcation were included.
e-mail: jala@medicon.cz The 10 men and 8 women ranged in age from 51 to 84

123
44 J. Lastovickova and J. H. Peregrin: Primary Self-Expandable Nitinol Stent Placement

years (mean 68.2). All lesions were caused by atheroscle- gradients were measured after stent placement and ranged
rosis; 2 patients had short occlusion of the distal abdominal from 0 to 5 mm Hg. Endovascular treatment was clinically
aorta, and the remaining 16 patients had focal stenoses not successful in all patients: Claudications and at-rest pain
involving aortic bifurcation. Patients were referred for resolved; the ulcer healed; and the blue-toe syndrome
claudications in both limbs (with walking distance from 5 disappeared within days after the procedure. Immediate
to 200 m): 1 patient had at-rest pain; 1 patient had a non- technical and clinical success was 100%.
healing ulcer; and 1 patient had blue-toe syndrome.
Diagnostic angiography of the abdominal aorta and both
lower extremities performed after clinical examination Follow-Up
diagnosed (additionally to aortic occlusive disease) sig-
nificant stenoses of the iliac arteries in 7 patients and All patients were followed-up on an outpatient basis.
lesions in the femoropopliteal region in 4 patients. The Clinical examination, which occurred at 1 and 6 months
aortic diameter ranged from 10 to 18 and aortic length from and then every year after the endovascular procedure,
20 to 50 mm. included patient history to evaluate claudication, rest pain,
Endovascular treatment of the abdominal aorta was or trophic changes of the lower extremities, examination of
performed in all patients by way of femoral approach peripheral pulses, measurement of anklebrachial pressure
(bypass limb in 1 patient with previous aortobifemoral index (ABPI), and colour-coded duplex ultrasonography of
bypass) using a 7F to 10F introducer sheath with the patient the abdominal aorta. Mean follow-up of the 18 patients
under local anaesthesia. A self-expandable nitinol stent was with primary nitinol self-expandable stent implantation
implanted in all 18 patients as a primary measure. In 2 was 49.4 months (range 3 to 96.4).
patients with occlusions and in 1 patient with a tight cal- Angiography was performed in nine patients during the
cified lesion, predilatation was achieved using a smaller follow-up period because of postprocedural claudication.
balloon (8-mm diameter). No restenosis of the abdominal aorta was detected in any of
Stent diameter ranged from 10 to 18 and length from 40 these patients. All aortic stents were well-patent, but de
to 60 mm. Sinus stents (Optimed, Ettlingen, Germany) novo iliac, femoropopliteal, or infrapopliteal lesions
were used in 7 patients, and SMART stents (Cordis, occurred as atherosclerosis progressed. Such patients were
Johnson and Johnson, New Jersey, USA) were used in 11 successfully treated with PTA/stents, one with repeat PTA
patients. All stents were dilated using 10- to 16-mm bal- and thrombolytic therapy of the infrapopliteal arteries
loons to achieve good expansion (Figs. 1 and 2). PTA of during follow-up (Table 1).
the iliac arteries with primary stenting was performed Two patients died during follow-up (one from cardiac
during the same procedure in patients with significant disease and one from pulmonary carcinoma) without
stenoses of the iliac arteries. Patients with femoropopliteal symptoms of limb ischemia. All of the remaining 16
lesions were treated either by endovascular technique or by patients were symptom free (including patients with PTA
surgery in following days (Table 1). who underwent intervention of the iliac or femoropopliteal
Bolus heparin, 5000 IU, was administered intra-arteri- arteries during follow-up) at follow-up.
ally during PTA. In the evening and in the morning after
the procedure, 5000 IU heparin was administrated subcu-
taneously. Antiaggregation was achieved with 100 mg Discussion
acetylsalicylic daily starting the day before the procedure.
Patients were dismissed from the hospital on postprocedure Occlusive disease of the distal abdominal aorta is caused
day 1. mostly by atherosclerosis, which sometimes may be asso-
ciated with hypoplastic aortoiliac syndrome. This most
likely congenital hypoplastic vascular disorder occurs
Results almost always in women and is characterised by a small-
size aortoiliac system (13 mm for the infrarenal aorta and
Technical success was defined as residual stenosis \30% 10 mm for the abdominal aorta above the bifurcation), high
and/or transstenotic systolic pressure gradient \5 mm Hg abdominal aorta bifurcation, and straight course of the iliac
after stent placement. Clinical patency was defined as the arteries with an acute angle of the aortic bifurcation. The
absence or improvement of symptoms as supported by a combination of hypoplasia and risk factors, such as ath-
favourable Doppler ultrasound examination. erosclerosis or heavy smoking, accounts for patients with
All endovascular procedures were technically success- occlusive atherosclerotic disease having symptoms at an
ful. No major or minor periprocedural or postprocedural earlier age than seen in other populations [6, 7]. Another
complications occurred. In 12 patients, systolic pressure possible cause of occlusive disease of the abdominal aorta

123
J. Lastovickova and J. H. Peregrin: Primary Self-Expandable Nitinol Stent Placement 45

Table 1 Summary of Cases


Patient Age Stenosis Stent type Fontaine Stage Follow-up Intervention
Additional Intervention during follow-up
No. (y) (%) Before After Months Fontaine
Treatment Treatment Stage

1 84 80 Sinus 14/40 comm.il. a. PTA Rt II b I 96.4 I


2 57 100 Sinus 14/40 II b I 93.5 I
3 69 80 Sinus 14/40 II b I 79.4 I
4 63 80 Smart 10/40 II b I 78.4 I FP bypass Rt
5 67 90 Sinus 12/40 comm.il. a. PTA Rt III I 75.3 I comm.il. a. rePTA Rt
6 62 80 Smart 14/40 II b I 71.3 I
7 77 90 Smart 14/60 II b II a 65.3 II a CHSE Rt
8 63 70 Sinus 14/40 FP bypass Rt II b I 65.3 I FP bypass Lt
9 72 100 Smart 14/40 comm.il. a. PTA Rt IV I 8,1D I
10 63 80 Sinus 16/60 ext.il.a. PTA Rt II b I 50.2 I comm.il. a. PTA Rt, FP bypass Lt
11 75 80 Smart 14/60 comm.il. a. II b I 44.3 I comm.il. a. PTA Rt, FP bypass PTA
PTA Lt, FP bypass Lt
12 75 70 Smart 14/40 FP bypass Lt III I 44.3 I FP bypass Lt TL
13 74 80 Sinus 18/40 III I 36.2D II a infrapopl.a. rePTA Rt,
Lt, FP bypass Rt, Lt
14 51 80 Smart 10/40 III I 40.2 I
15 70 80 Smart 14/40 AFS PTA Rt IV I 19/1 I AFS PTA Lt
16 59 80 Smart 14/40 II b I 11 I comm.il. a., ext. il. a. PTA Lt
17 77 90 Smart 14/40 ext. il. a. PTA Rt II b I 8.1 I
18 70 80 Smart 14/60 II b I 3 I
Note: No complications occurred in any case.

is arteritis. Takayasu arteritis is chronic inflammatory preservation or improvement of sexual function in male
arteriopathy, which affects the aorta and its major branches patients [5].
as well as the pulmonary arteries. Stenoses caused by this Angioplasty and, later, stent implantation became stan-
type of arteritis can be very rigid [8]. All occlusive lesions dard therapy for the interventional treatment of large
of the distal abdominal aorta in our 18 patients were caused vessels (e.g., common iliac artery) with 5-year patency rates
by atherosclerosis. [70% to 80%. Excellent long-term stenting results in iliac
Patients with aortic occlusive disease are usually refer- arteries portends good results of stenting in abdominal
red for treatment because of bilateral intermittent calf, aortas. Long-term experience with PTA and stenting of the
thigh, or gluteal claudications but rarely for rest pain or a infrarenal abdominal aorta remains limited because of the
nonhealing ulcer [9, 10]. Visualization of both limb arteries limited number of published studies, which usually com-
to the periphery is essential and must be accompanied by prised small numbers of patients. Few data exist in the
subsequent treatment of aortic lesions as well as stenoses literature regarding the primary use of stents. During the
and occlusions occurring in the iliac, femoropopliteal, or last 2 decades, endovascular therapy results were reported
infrapopliteal regions. in \600 patients, but some of the lesions reported involved
Favourable long-term results of traditional surgical stenoses extending to or below the aortic bifurcation [3].
abdominal aorta enarterectomy/aortobifemoral (aorto- The type of procedure (PTA alone with the kissing-balloon
biiliac) bypass have been well documented in the literature, technique, primary or secondary stent placement), type of
with a patency rate of 90% at 5 years and 75% at 10 years stent (balloon expandable or self expandable), and techni-
[1, 2]. However, it may be associated with significant cal and clinical outcomes were not specified in detail in
mortality and morbidity (5% to 10% of early major com- many studies [3]. Stoeckelhuber et al. [1] published a ret-
plications), including sexual dysfunction in male patients rospective study reporting nine patients with infrarenal
[2]. Endovascular therapy offers many potential advantages aortic occlusive disease, all of whom were treated with
compared with surgical treatment, including low mortality Palmaz stent and Wallstent after predilatation. The primary
and major complication rates, avoidance of general technical end clinical success rates were both 100%, and
anaesthesia and abdominal incision, short hospital stay, and the long-term results were favourable. Yilmaz et al. [3]

123
46 J. Lastovickova and J. H. Peregrin: Primary Self-Expandable Nitinol Stent Placement

reported primary self-expandable stent implantation against the vessel wall. Predilatation with a smaller balloon
(Wallstent and Sinus stents) in 13 patients. Schedel et al. was used in 2 patients with occlusions and in 1 patient with
[2] reported 15 patients having aortic stenoses extending to a very tight stenosis. Primary stent placement without
the iliac arteries treated with Palmaz stent implantation and predilatation is recommended to decrease restenosis caused
kissing-balloon technique dilation. Sheeran et al. [5] used at by intimal hyperplasia [11].
balloon-expandable stent in 6 patients, and Nyman et al. [4] The radiopacity of nitinol stents is not ideal, but markers
reported primary stenting with both balloon- and self- on stents and delivery systems indicate even minimal
expandable stents in 30 patients with aortic and aortoiliac shortening during implantation, thus allowing accurate
stenoses and occlusions. placement. Self-expandable stents employ good radial
All 18 patients were included in our study because they force to the vessel wall, and nitinol stents achieve further
all had focal stenoses of the infrarenal aorta that did not expansion to their nominal range even after the procedure.
involve aortic bifurcation. All of them were treated with The balloon diameter used to dilate a stent after implan-
primary of self-expandable nitinol stent placement. Pri- tation may therefore be smaller in size to minimize vessel
mary stenting allows treatment of high-risk lesions (e.g., trauma. Another advantage of self-expandable stents is
heavily calcified, irregular, eccentric, or ulcerated). Pri- their small delivery system.
mary stenting may decrease (1) the risk of distal Complications of endovascular therapy of the abdominal
embolization by covering the lesion with the stent before aorta have been well documented in the literature [4] as
balloon dilation (such that the atheroma and the clot are follows: aortoiliac dissection, distal embolisation, iliofem-
trapped between the stent and the aortic wall) and (2) the oral thrombosis, pseudoaneurysm, access-site haematoma,
risk of vessel rupture by better distributing dilating forces and puncture-site infection. Meticulous technique, along

Fig. 1 Female patient, 57 years old, with 50 m claudications in both abdominal aorta was achieved (b), and the patient has been free of
limbs caused by short occlusion of the distal abdominal aorta (a). claudications during 90-month follow-up
After primary Sinus stent placement, excellent patency of the

123
J. Lastovickova and J. H. Peregrin: Primary Self-Expandable Nitinol Stent Placement 47

Fig. 2 Male patient, 75 years old, with 10 m claudications in both implantation yielded a good result: His pressure gradient is 0 mm
limbs as well as significant stenoses of the distal and abdominal aorta Hg (b), and the patient has been free of claudications during 41-month
(pressure gradient 50 mm Hg) (a). Primary SMART stent follow-up

with adequate antiaggregation and periprocedural heparin- popliteal, and infrapopliteal arteries and were treated
isation, can minimize the risk of complications. mostly endovascularly and sometimes by femoropopliteal
In our small study of 18 patients, a 100% technical bypass (Table 1).
success rate without any complications was achieved. A Stenting the distal abdominal aorta can cause inferior
100% clinical success rate was achieved after all patients mesenteric artery occlusion [1]. We did not experience
with additional significant lesions of the iliac and fem- clinical evidence of acute or chronic mesenteric ischemia
oropopliteal arteries were successfully treated. during follow-up in our series of patients, although the
No consistent approach to long-term follow-up has been inferior mesenteric artery was overstented in 16 of 18
published in the literature. Long-term results are based patients.
mostly on clinical examinations using ABPI measurement Endovascular therapy can be recommended as a treat-
or Doppler ultrasonography. ment of choice for patients with focal atherosclerotic
All of our patients were periodically investigated during stenoses or short occlusions of the distal abdominal aorta
follow-up. Clinical assessment was paramount. Doppler not extending to the aortic bifurcation. It is safe and
ultrasound of the abdominal aorta did not find any signif- effective, has high technical and clinical success rates,
icant restenosis in any patient with a stented abdominal shows favourable long-term clinical results, and confers a
aorta. ABPI measurement was sometimes misleading in low rate of complications. Primary stent placement is
patients who had additional atherosclerotic lesions of the recommended in patients with eccentric, irregular, and
iliac or femoropopliteal arteries. Nine patients became calcified lesions and occlusions to prevent major compli-
symptomatic during follow-up because of progression of cations (mainly peripheral embolisation).
their atherosclerosis. In all of these patients, angiogra-
phy of the stented aorta showed well-patent stents
Acknowledgment Support was provided by the Institute for Clini-
without significant restenosis. De novo lesions caused by cal and Experimental Medicine financial support (Grant No. MZO
atherosclerotic progression were found in pelvic, femoro- 00023001)

123
48 J. Lastovickova and J. H. Peregrin: Primary Self-Expandable Nitinol Stent Placement

References 6. Audet P, Therasse E, Oliva VL, et al. (1998) Infrarenal aortic


stenosis: Long-term clinical and hemodynamic results of percu-
1. Stoeckelhuber BM, Stoeckelhuber M, Gellisen J, et al. (2006) taneous transluminal angioplasty. Radiology 209:357363
Primary endovascular stent placement for focal infrarenal aortic 7. Raso AM, Varetto G, Bellan A, et al. (2001) Small aorta syn-
stenosis: Long-term results. J Vasc Interv Radiol 17:11051109 drome: Hypothesis or reality? Minerva Cardioangiol 49:21120
2. Schedel H, Wissgott Ch, Rademaker J, et al. (2004) Primary stent 8. Tyagi S, Sethi KK, Khalilullah M (1992) Balloon angioplasty of
placement for infrarenal aortic stenosis: Immediate and midterm the aorta in Takayasu arteritis: Initial and long-term results. Am
results. J Vasc Interv Radiol 15:353359 Heart J 124:876882
3. Yilmaz S, Sindel T, Yegin A, et al. (2004) Primary stenting focal 9. Yakes WF, Kumpe DA, Brown SB, et al. (1989) Percutaneous
atherosclerotic infrarenal aortic stenoses: Long-term results in 13 transluminal aortic angioplasty: Techniques and results. Radiol-
patients and literature review. Cardiovasc Intervent Radiol ogy 172:965970
27:121128 10. Therasse E, Cote G, Oliva VL, et al. (2001) Infrarenal aortic
4. Nyman U, Uher P, Lindh M, et al. (2000) Primary stenting in stenosis: Value of stent placement after percutaneous translumi-
infrarenal aortic occlusive disease. Cardiovasc Intervent Radiol nal angioplasty failure. Radiology 219:655662
23:97108 11. Harnek J, Zouca E, Stenram W (2002) Insertion of self-expand-
5. Sheeran SR, Hallisey MJ, Ferguson D (1997) Percutaneous able nitinol stents without previous balloon angioplasty reduces
transluminal stent placement in the abdominal aorta. J Vasc In- restenosis compared with PTA before stenting. Cardiovasc In-
terv Radiol 8:5560 tervent Radiol 25:430436

123
Cardiovasc Intervent Radiol (2008) 31:4955
DOI 10.1007/s00270-007-9214-3

CLINICAL INVESTIGATION

Transluminal Angioplasty of Peroneal Artery Branches


in Diabetics: Initial Technical Experience
Lanfroi Graziani Antonio Silvestro Luca Monge Gian Mario Boffano
Francesco Kokaly Ilaria Casadidio Francesco Giannini

Received: 15 March 2007 / Accepted: 27 June 2007 / Published online: 25 October 2007
 Springer Science+Business Media, LLC 2007

Abstract The present study aimed to report the technical that PTA of the collateral branches from distal peroneal to
feasibility of percutaneous transluminal angioplasty (PTA) foot arteries is a feasible technique. Future studies are
of obstructed or insufficient collateral branches (anterior required to define the clinical role of this novel approach.
and posterior perforating branches) from distal peroneal to
foot arteries in diabetic patients with chronic critical limb Keywords Percutaneous transluminal angioplasty 
ischemia (CLI) and chronic noncrossable occlusion of the Critical limb ischemia  Diabetes mellitus  Diabetic foot 
anterior and posterior tibial arteries. Twenty-four diabetic Collateral circulation
CLI patients (age, 67 8 years; 87% males) undergoing
collateral PTA were included. Baseline clinical angio-
graphic and follow-up data were retrospectively reviewed. Introduction
Collateral PTA was associated with a concomitant PTA of
other sites in 21 (83%) cases. In 15 cases the treated col- Percutaneous transluminal angioplasty (PTA) has emerged
lateral linked the peroneal with the plantaris communis; in as a valid alternative to bypass surgery for the treatment of
9 cases, the peroneal with the dorsalis pedis. Angiographic critical limb ischemia (CLI) [1, 2]. This is particularly true
results of collateral PTA were good in 13 cases (\30% in CLI patients with diabetes mellitus [3, 4], as well as in
residual stenosis), whereas the result was considered those in chronic dialysis [5], in whom PTA has proven to
moderate (30%49% residual stenosis) in the remaining be safe and effective in avoiding major amputation. Nev-
cases. Neither perforation nor acute occlusion of the treated ertheless, it must be considered that PTA may be
collaterals or other relevant complications were observed. technically challenging in diabetic subjects with CLI, as
Mean follow-up was 32 17 months. Major amputation they mostly have infrapopliteal disease, usually character-
was necessary for two (8.3%) patients. Cumulative limb ized by long chronic occlusions of anterior and posterior
salvage rates at 2 and 4 years were 96% and 87%, tibial arteries [6].
respectively. In conclusion, this initial experience shows The importance of providing direct straight-line flow to
the foot for thus obtaining limb salvage in diabetic patients
with CLI has already been emphasized by vascular sur-
L. Graziani (&)  A. Silvestro  F. Kokaly geons who reported better results with distal bypass
Servizio di Emodinamica, Istituto Clinico Citta di Brescia, Via compared to femoropopliteal graft or profundoplasty [7].
Gualla 15, 25123 Brescia, Italy However, improvement of collateral circulation by surgical
e-mail: langrazi@tin.it
or endovascular procedures has also been reported. In the
L. Monge  G. M. Boffano case of noncrossable superficial femoral artery (SFA)
SC Malattie Metaboliche e Diabetologia, Centro Traumatologico occlusion, an alternative to increasing the blood supply to
Ortopedico CTO, Torino, Italy the foot is to improve the collateral circulation by the
treatment of ostial stenosis of the profunda femoris.
I. Casadidio  F. Giannini
Servizio di Diabetologia e Malattie del Metabolismo, Ospedale However, the clinical efficacy of this procedure in diabetic
di Lucca, Lucca, Italy patients with CLI is doubtful [8, 9], probably due to the

123
50 L. Graziani et al.: Collateral PTA in CLI Patients

reduced development and efficiency of collaterals in dia-


betic subjects, as already seen in the coronary circulation
[10]. The use of collateral circulation has also been
described in coronary interventions, consisting of recana-
lization of the totally occluded native coronary artery by
retrograde approach, with the passage of the wire and
balloons through collateral branches supplying the occlu-
ded artery, but without direct dilatation of collaterals [11].
We usually attempt to provide, by endovascular treat-
ment, direct straight-line flow to the foot through a native
tibial artery, selecting, whenever possible, the anterior
tibial artery for ischemic forefoot lesions and the posterior
tibial for calcanear lesions. If neither the anterior nor the
posterior tibial artery can be treated despite several intra-
luminal and subintimal crossing attempts, the alternative
treatment may consist of providing direct flow along the
peroneal artery. As shown in Fig. 1, from the distal pero-
neal artery some collateral branches arise for foot arteries.
In particular, the posterior perforating branch connects the
peroneal with the plantaris communis, whereas the anterior
perforating branch links the peroneal to the dorsalis pedis.
In this case we evaluate the condition of those branches,
which are frequently occluded or stenosed. In the case of
suitable anatomy, with the collateral branch communicat-
ing with the corresponding foot artery, we used to treat this
collateral vessel. In the present study, we retrospectively
report data on a consecutive series of diabetic patients
treated for CLI and ischemic foot lesions, in whom direct
or improved flow to the foot was attempted through PTA of
collateral branches from the distal peroneal artery to the
foot arteries. Noteworthily, the vascular anatomy of the
ankle may vary in some subjects, in whom the anterior
tibial artery or posterior tibial artery is hypoplasic and the
dorsalis pedis or plantaris communis, respectively, is a
continuation of the peroneal artery. Patients with this Fig. 1 Well-developed anterior and posterior collateral branches
anatomic variance are not included in the present study. from distal peroneal to pedal arteries in a case of chronic total
occlusion of both the anterior and the posterior tibial arteries in a
nondiabetic patient. APB, anterior perforating branch; PPB, posterior
perforating branch. Sites of collateral PTA were site 1 (APB) and site
Methods 2 (PPB)

The study population included 24 consecutive subjects branches from peroneal artery to foot arteries was per-
with diabetes mellitus, treated with distal peroneal collat- formed, associated or not with concomitant treatment of
eral branch PTA. They were referred to our Catheterization other obstructions along the ipsilateral iliac-femoral-pop-
Laboratory between 2002 and 2005 for the presence of liteal and/or infrapopliteal arterial segments. Baseline
chronic CLI, defined according to TASC criteria [12]. All angiographic and clinical data as well as follow-up infor-
of them had foot ulcer and/or gangrene, not healing after 4 mation were retrospectively reviewed.
weeks of conservative treatment, including local dressing,
or presented typically ischemic rest pain. TcPO2 at the
dorsum of the foot was \30 mm Hg in all cases. Ankle Patient Categorization
pressure measurements were not performed due to the high
prevalence of medial layer calcification and falsely high Patients have been classified according to the angiographic
values [13]. We consecutively included patients, treated in severity of peripheral artery disease using a morphologic
the study period, in whom a PTA of diseased collateral score based on seven classes of progressive vascular

123
L. Graziani et al.: Collateral PTA in CLI Patients 51

involvement [6]. Figure 2 shows examples of the adopted artery. None of the included subjects had iliac disease. A 5-
classification. to 6-Fr sheath (Super Sheath; Medikit) was positioned to
perform a preliminary angiography. In the case of signifi-
cant obstruction (50% diameter reduction), sodium
Endovascular Procedure heparin (5000 IU) was given intra-arterially and a PTA
procedure attempted. The aim of the procedure was to treat
In this series, all the procedures were performed by a single each significant femoropopliteal and below-the-knee
experienced operator (L.G.) under local anesthesia through (BTK) obstruction, in order to improve or obtain direct
an antegrade puncture of the ipsilateral common femoral flow to the foot. Most patients had obstructions at the level
of both the anterior and the posterior tibial artery, which
could not be successfully treated with a catheter-based
technique, including a subintimal approach, whereas the
anatomy of distal peroneal and collateral branches to the
pedal artery was considered suitable for PTA. In some
cases, the patency of one tibial artery was present, but not
directly supplying the ischemic lesion. Anyway, in those
cases a possible improvement in blood flow to the ischemic
foot region was tried through balloon dilatation of collat-
eral branches, from distal peroneal to foot arteries. PTA
was performed only at the level of the collateral branches
in four patients (17%); in the remaining cases, besides
collateral PTA, the procedure included PTA of another
BTK site(s) in 14 patients, PTA of another femoropopliteal
site(s) in 2 patients, and PTA of other femoropopliteal plus
BTK sites in 4 cases. As shown in Fig. 1, the collateral
PTA site was defined as site 1 when PTA was performed
at the collateral branch level from the peroneal to the
dorsalis pedis (9 cases) and as site 2 when PTA was
performed at the level of the collateral branch from the
peroneal to the plantaris communis (15 cases).
Superficial femoral and popliteal artery lesions were
usually crossed using 0.035in. wires (Starter Newton;
Boston Scientific), whereas dilatation was performed using
4- to 6-mm-diameter balloon catheters (Synergy; Boston
Scientific). To achieve a good angiographic result, a pro-
longed dilatation time (180 s) was used. In infrapopliteal
arteries, including collateral branches, intraluminal or
subintimal crossing of lesions was performed using 0.014
in. coronary-type hydrophilic stiff guide wires (PT Graphix
Super Support; Boston Scientific-Scimed). Dilatation bal-
loon sizes were in the range of 1.53.0 mm (either Sprinter,
Medtronic, or Amphirion Deep; Invatec). In the case of
unsatisfactory results (residual stenosis [30% and/or flow-
limiting dissection), a new prolonged high-pressure (up to
20-atm) dilatation was performed. If angiography still did
not show a good result, stent implantation was considered
only for the superficial femoral artery (self-expandable
nitinol stent; Protege; EV3) and rarely (2% of cases) for
only proximal infrapopliteal vessels (Driver; Medtronic).
In all treated limbs, technical success was defined by the
Fig. 2 Examples of below-the-groin arterial involvement corre-
absence of residual obstructions after treatment: residual
sponding to different classes of the adopted morphologic
classification. (From European Journal of Vascular Endovascular stenosis \30% in the absence of flow-limiting dissections
Surgery 33:453460, 2007.) on the final angiogram. Intra-arterial local injection of

123
52 L. Graziani et al.: Collateral PTA in CLI Patients

nitroglycerin was used in the case of spasm. On the basis of Results


the final angiogram at the level of collateral, the procedure
was defined as obtaining a good angiographic result Table 1 shows clinical characteristics and procedural data
(\30% residual stenosis) or a moderate angiographic result of the 24 included subjects. Seven patients (29%) were in
(30%49% residual stenosis). All patients were treated Rutherford category 4, whereas the remaining 17 patients
with antiplatelet drugs after the procedure: for 2 months were in category 5. With respect to angiographic findings,
after the procedure with a thienopyridine (either Ticlopi- the most common morphologic class was 4 (two arteries
dine, 250 mg b.i.d., or Clopidogrel, 75 mg/day), then with occluded plus other multiple stenoses of the femoropop-
aspirin, 100300 mg/day, if not contraindicated. liteal and BTK arteries). In 13 cases collateral PTA
Study end points were the occurrence of procedure- produced a good angiographic result; in the remaining
related complications, major amputation, and target cases a moderate angiographic result was obtained.
extremity revascularization. Major amputation was defined Peripheral runoff at the end of the procedure consisted of
as limb loss above the metatarsal level. Minor amputation the isolated peroneal artery refilling a foot artery in 19
was defined as transmetatarsal amputation or removal of cases and of the peroneal artery + one of the two tibial
more distal parts of the lower extremity. arteries in 5 cases. One patient had undergone a previous
femoropopliteal bypass, with patency of the graft at the
time of the endovascular intervention. The adopted mor-
Statistical Analysis phologic score was not applicable in this particular case.
No clinically relevant periprocedural complication was
Demographic and clinical characteristics of the study observed, including perforation/rupture or acute occlusion
population are reported as mean SD for continuous of the treated collateral. No clinical worsening of foot
variables and as number (percentage) for categorical vari- ischemia following collateral PTA was observed. Fig-
ables. Cumulative rates of limb survival without major ures 35 show some examples of PTA of collateral
amputation and reintervention were calculated with the branches.
Kaplan-Meier survival life table. During a follow-up of 32 17 months, major amputa-
tion was necessary for two (8.3%) patients. Among the four
patients with isolated PTA of a peroneal collateral branch
for dorsalis pedis, in one case a major amputation was
necessary 36 months after the PTA. The other major
amputation was necessary within 1 month after a PTA of
Table 1 Characteristics of the study population
the peroneal plus the collateral branch for dorsalis pedis. In
Age 67 8 yr both cases, collateral PTA had produced a moderate
angiographic result.
Male gender 21 (87%)
During follow-up two (8.3%) patients died, one from
Smoking habit
myocardial infarction and one from accidental trauma.
Never 11 (46%)
Former 6 (25%)
Current 7 (29%)
Discussion
Hypertension 14 (58%)
CAD 10 (42%)
Patients with diabetic ischemic foot have a very high risk
Foot ulcer/gangrene 17 (71%)
of major amputation, unless a revascularization procedure
Baseline angiographic score [6]
is performed [12]. Endovascular treatment of CLI in these
Class 3 3 (12%)
patients is a low-invasive procedure with proven efficacy
Class 4 15 (63%) and a very good safety profile [3, 4]. Noteworthily, the
Class 5 2 (8%) avoidance of major amputation implies an improvement in
Class 6 3 (12%) the quality of life as well as the life expectancy [14]. In
Other PTA site these patients, the aim of the treatment should be to provide
None 4 (2%) adequate blood flow to the bone and soft tissues of the foot
Femoropopliteal 2 (8%) to promote the healing process, and therefore a direct
BTK 14 (58%) straight-line flow to a pedal artery is recommended [10,
Femoro-politeal + BTK 4 (2%) 15]. To achieve this result it is often necessary to cross
Note. CAD, coronary artery disease; PTA, percutaneous transluminal long, challenging tibial artery occlusions, also using sub-
angioplasty; BTK, below the knee intimal guide-wire passage, which is a technique with

123
L. Graziani et al.: Collateral PTA in CLI Patients 53

Fig. 3 Example of percutaneous transluminal angioplasty (PTA) of diseased collateral branch from peroneal to dorsalis pedis artery: before PTA
(left) and after PTA (right)

Fig. 4 Example of percutaneous transluminal angioplasty (PTA) of diseased collateral branch from peroneal to plantaris communis: before PTA
(left), during wire passage (middle), and after PTA (right)

proven safety and efficacy [16]. However, when tibial branches) of the distal peroneal artery must be considered,
artery recanalization is not possible, the possibility of which, according to our initial experience, seems to be a
obtaining improved or direct flow by balloon angioplasty feasible technique. Indeed, in this series of 24 patients,
of collateral branches (anterior or posterior perforating PTA of distal peroneal collateral produced a good

123
54 L. Graziani et al.: Collateral PTA in CLI Patients

Fig. 5 Example of percutaneous transluminal angioplasty (PTA) of diseased collateral branch from peroneal to plantaris communis: before PTA
(left), during wire passage (middle), and after PTA (right)

angiographic result (defined as \30% residual stenosis) in Super Support; Boston Scientific-Scimed) hydrophilic
about half (13 of 24) of the patients. In the remaining cases coronary-type wires for the highest possible crossability
a moderate angiographic result was achieved, with \50% and support. This stiff wire, however, may present some
residual stenosis. No clinically relevant periprocedural problems, such as the perforation of such thin vessels.
complication has been observed. In particular, dilatation of Thus, these procedures should be performed by operators
collateral branches in no case led to perforation or occlu- familiar with thin wires and small vessels. Dilatation is
sion of the vessel. During follow-up, only two patients usually performed with coronary-type, rapid-exchange,
required major amputation, one patient among those with low-compliant balloon catheters of 1.5- to 2.0-mm diam-
isolated collateral PTA. Both major amputations were eter (Sprinter; Medtronic). Alternatively, an over-the-wire
performed in cases of a mild result of the collateral PTA. balloon catheter (Amphirion Deep; Invatec) may be used,
However, in this regard, it must be emphasized that the aim especially in cases with long segments to dilate. In the
of the present study is purely to report on the technical presence of tight and calcified stenosis, it is also possible to
feasibility and the safety of PTA of the peroneal collateral adopt a balloon catheter with particularly low compliance
branches. To date, no conclusion can be drawn regarding (PowerSail; Guidant-Abbott), which can be inflated at
the clinical efficacy of this novel alternative approach, due higher pressures (up to 22 atm). For this purpose, it must be
to the relatively limited size of the study population and the noted that the occurrence of diabetic macroangiopathy
concomitant recanalization of other below-the-groin seg- responsible for obstruction in collaterals may imply some
ments in most of the included subjects. Therefore future advantages during balloon dilatation. Indeed, one of the
investigations are necessary to determine the clinical role shortcomings of using collaterals during percutaneous
of collateral PTA. interventions is that these vessels have small lumens and
weak walls and, thus, are more prone to perforation and
rupture. However, the typical pathologic changes of dia-
Technical Considerations betic macroangiopathy [17] with medial layer uniform
calcification may indeed offer protection against such
To perform PTA of distal ankle-foot vessels, an antegrade complications.
femoral approach is advisable to optimize angiographic In conclusion, our results show that PTA of collateral
images and have good support. In the particular case of branches for pedal arteries is a feasible and safe technique.
dilatation of collateral branches it must be taken into In cases in which it is not possible to obtain a direct flow
account that balloon passage through stenotic and tortuous through tibial arteries, angioplasty of collateral branches
vessels may be somewhat troublesome. To overcome this from distal peroneal to foot arteries could be a new option
problem, we commonly use stiff (0.014 in. PT Graphix in the presence of suitable anatomic conditions.

123
L. Graziani et al.: Collateral PTA in CLI Patients 55

References role in chronic critical limb ischemia? J Endovasc Ther 11:119


124
1. Kudo T, Chandra FA, Ahn SS (2005) The effectiveness of per- 9. Dick P, Mlekusch W, Sabeti S, et al. (2006) Outcome after en-
cutaneous transluminal angioplasty for the treatment of critical dovascular treatment of deep femoral artery stenosis: results in a
limb ischemia: a 10 year experience. J Vasc Surg 41:423435 consecutive patient series and systematic review of the literature.
2. Hynes N, Mahendran B, Manning B, et al. (2005) The influence J Endovasc Ther 13:221228
of subintimal angioplasty on level of amputation and limb sal- 10. Waltenberger J (2001) Impaired collateral vessel development in
vage rates in lower limb critical ischemia: a 15-year experience. diabetes: potential cellular mechanisms and therapeutic implica-
Eur J Vasc Endovasc Surg 30:291299 tions. Cardiovasc Res 49:554560
3. Faglia E, Mantero M, Caminiti M, et al. (2002) Estensive use of 11. Surmely JF, Tsuchikane E, Katoh O, et al. (2006) New concept
peripheral angioplasty, especially infrapopliteal, in treatment of for CTO recanalization using controlled antegrade and retrograde
ischemic foot ulcer: clinical results of a multicentric study of 221 subintimal tracking: the CART technique. J Invas Cardiol
consecutive diabetic subjects. J Intern Med 252:225232 18:339340
4. Faglia E, Dalla Paola L, Clerici G, et al. (2005) Peripheral 12. Dormandy JA, Rutherford RB (2000) Management of peripheral
angioplasty as the first-choice revascularization procedure in arterial disease (PAD). TASC Working Group. TransAtlantic
diabetic patients with critical limb ischemia: prospective study of Inter-Society Consensus (TASC). J Vasc Surg 31 (Suppl):1296
993 consecutive patients hospitalized and followed between 1999 13. Quigley FG, Faris IB, Duncan HJ (1991) A comparison of
and 2003. Eur J Vasc Endovasc Surg 29:620627 Doppler ankle pressures and skin perfusion pressure in subjects
5. Graziani L, Silvestro A, Bertone V, et al. (2007) Percutaneous with and without diabetes. Clin Physiol 11:2125
transluminal angioplasty is feasible and effective in patients on 14. Tentolouris N, Al-Sabbagh S, Walker MG, et al. (2004) Mortality
chronic dialysis with severe peripheral artery disease. Nephrol in diabetic and nondiabetic patients after amputations performed
Dial Transplant 22:11441149 from 1990 to 1995: a 5-year follow-up study. Diabetes Care
6. Graziani L, Silvestro A, Bertone V, et al. (2007) Vascular 27:1598604
involvement in diabetic subjects with ischemic foot ulcer: a new 15. ACC/AHA (2006) ACC/AHA guidelines for the management of
morphologic categorization of disease severity. Eur J Vasc En- patients with peripheral arterial disease (lower extremity, renal,
dovasc Surg 33:453460 mesenteric, and abdominal aortic). Circulation 113:e463e654
7. Aulivola B, Pomposelli FB. (2004) Dorsalis pedis, tarsal and 16. Bolia A (2005) Subintimal angioplasty in lower limb ischaemia.
plantar artery bypass. J Cardiovasc Surg 45:203212 J Cardiovasc Surg (Torino) 46:385394
8. Diehm N, Savolainen H, Mahler F, et al. (2004) Does deep 17. Andresen JL, Rasmussen LM, Ledet T (1996) Diabetic macro-
femoral artery revascularization as an isolated procedure play a angiopathy and atherosclerosis. Diabetes 45:S91S94

123
Cardiovasc Intervent Radiol (2008) 31:5665
DOI 10.1007/s00270-007-9221-4

CLINICAL INVESTIGATION

Cost Determinants of Percutaneous and Surgical Interventions for


Treatment of Intermittent Claudication from the Perspective of
the Hospital
Bertrand Janne dOthee Michael F. Morris
Richard J. Powell Michael A. Bettmann

Received: 16 October 2005 / Accepted: 25 September 2007 / Published online: 1 November 2007
Springer Science+Business Media, LLC 2007

Abstract Results The linear regression model shows that proce-


Purpose To identify pretreatment predictors of proce- dural costs per patient are 25% lower in percutaneous
dural costs in percutaneous and surgical interventions for patients (versus surgical), 42% lower for patients without
intermittent claudication due to aortoiliac and/or femoro- rest pain than for those with, 28% lower if treated lesions
popliteal disease. are unilateral (versus bilateral), 12% lower if the treated
Methods A retrospective study was conducted in 97 lesion is stenotic rather than occlusive, 34% higher in
consecutive patients who underwent percutaneous or sur- sedentary patients, and 11% higher in patients with a his-
gical interventions over 15 months at a tertiary care center. tory of cardiac disease. After a mean clinical follow-up [2
Nineteen clinical predictive variables were collected at years, between-group differences between percutaneous
baseline. Procedural costs (outcome) were assessed from and surgical patients were small and of limited significance
the perspective of the hospital by direct calculation, not in all dimensions of clinical outcome.
based on ratios of costs-to-charges. A multivariable Conclusion Predictors of clinical outcome are different
regression model was built to identify significant cost from predictors of costs, and one should include both types
predictors. Follow-up information was obtained to provide of variables in the decision-making process. The choice of
multidimensional assessment of clinical outcome, includ- percutaneous versus surgical strategy, the presence of rest
ing technical success (arteriographic score) and clinical pain, and the bilaterality of the culprit lesions were the
result (changes in ankle-brachial pressure index; cumula- main pretreatment determinants of procedural costs. When
tive patency, mortality, and complication rates). possible choices of treatment strategy overlap, percutane-
ous treatment should provide an acceptable result that is
less expensive (although not equal to surgery).
B. Janne dOthee (&)
Department of Radiology, Cardiovascular Imaging & Keywords Angiography  Arteries  Arteriosclerosis 
Intervention, Massachusetts General Hospital, Harvard Medical Cost-effectiveness  Interventional radiology 
School, 55 Fruit Street GRB-290, Boston 02114-2696, USA Stents and prostheses  Transluminal angioplasty
e-mail: bjannedothee@partners.org

M. F. Morris
Banner Good Samaritan Hospital, 1111 E. McDowell Road, Introduction
Phoenix, AZ 85006, USA
Cost data on the diagnosis or treatment of intermittent
R. J. Powell
Department of Surgery, Dartmouth-Hitchcock Medical Center, claudication are critically sparse, in both the medical and
One Medical Center Drive, Lebanon, NH 03756, USA the economic literature. A few prior economic evaluations
have focused on peripheral arterial obstructive disease
M. A. Bettmann
(PAD) in general, but not on intermittent claudication
Department of Radiological Science, Wake Forest University
Baptist Medical Center, Medical Center Boulevard, specifically [1, 2]. Also, most cost data available so far
Winston-Salem, NC 27157, USA were obtained by calculations based on the ratio of cost-to-

123
B. Janne dOthee et al.: Costs and Outcomes in Claudication 57

charges (RCC = costs/charges for a given service), which following inclusion criteria were then applied: (1) non-acute
typically entail approximations [3], and have been inter- lower limb ischemia presenting initially with lifestyle-lim-
preted in terms of incremental cost-effectiveness ratios iting intermittent claudication with or without ischemic rest
(ICERs), which carry inherent problems for statistical pain (classes II, III or IV of the classification of the Society
analysis and interpretation [4]. of Vascular Surgery/International Society of Cardiovascular
Extensive literature over the last decades has shown the Surgery (SVS-ISCVS)) [2, 6]; and (2) intervention(s) per-
clinical advantages of endovascular therapy [1, 2, 5] for formed for revascularization of stenotic and/or occlusive
treating PAD. These merits include the less invasive nature lesions located between the level of the infrarenal aorta and
of these procedures compared with surgery, with decreased the below-knee popliteal artery. Exclusion criteria were: (1)
morbidity and mortality, shorter duration of hospitaliza- patients undergoing invasive procedures unrelated to PAD;
tion, shortened postprocedural recovery time, and patient (2) more severe stages of PAD such as ulcer, tissue loss,
preference. In addition, the cost of percutaneous proce- necrosis or gangrene (classes VVI); (3) acute lower limb
dures has always been perceived as lower. However, the ischemia or blue toe syndrome; (4) previous invasive vas-
cost argument has come under criticism, as the beneficial cular intervention for PAD performed at an outside
effects of surgical alternatives are thought to be longer institution or for which the corresponding operative or
lasting. radiologic reports were not available; and (5) distal surgical
In the present study, we first calculated the actual bypasses (i.e., originating from the distal third of the
operating expenses (costs) of performing percutaneous and superficial femoral artery or below). Bypasses from the
surgical interventions for PAD by using a cost calculation common femoral artery to the below-knee arteries were
method that was not based on RCCs. We subsequently included. A total of 97 patients met these criteria and are the
explored potential determinants of these procedural costs scope of this study. Among these, a percutaneous group was
by building a multivariable regression model based on defined by the absence of arteriotomy. This study aimed at
predictive variables available clinically before treatment. providing IR physicians a better understanding of the
Better identifying cost determinants may not only be useful decision-making process at the hospital level for patients
for interventional radiologists and vascular specialists to with intermittent claudication; therefore, we chose the
predict or reduce costs; it may also help them to understand perspective of the hospital administrator.
the perspective of hospital administrators who face strate-
gic decisions regarding invasive treatments for PAD.
Cost Data Collection: Outcome Variable

Material and Methods The cost accounting methodology in use at our institution
recognizes three basic types of costs: direct costs, transfer
This was a retrospective study on costs and outcomes in 97 costs, and institutional overhead costs. Direct costs corre-
consecutive patients who underwent invasive interventions spond to expenses generated by so-called revenue
(percutaneous or surgical treatment) for lifestyle-limiting departments (e.g., a surgical unit) to provide a given service.
intermittent claudication due to PAD at the aortoiliac and/ The hospital Finance Department calculates these costs by
or femoropopliteal levels in the Interventional Radiology using departmental cost analyses. Transfer costs are those
(IR) and Vascular Surgery departments of our tertiary care incurred in support departments that are specific to patient
center over a 15-month period. Our institutional Committee procedures or other departments (e.g., phlebotomy/IV team).
for the Protection of Human Subjects from Research Risks The hospital uses a cost transfer algorithm to allocate these
approved the study. transfer costs from support departments to revenue and
overhead departments. The overhead costs category thus
includes not only institutional overhead (administration,
Patient Selection and Study Perspective liabilities, building maintenance, etc.) but also part of the
transfer costs that was allocated to them. By allocating back
A database query was conducted by the Medical Record these overhead costs to the revenue departments, the hospital
Department to retrieve the identification logs of all patients is able to attribute all its expenses to revenue departments,
who underwent an invasive vascular procedure for PAD hence to patients chargeable procedures.
during the study period. The request was performed on a The total cost for a patients procedure was calculated
wide selection of CPT 4 codes (Common Procedural Ter- by adding all the direct, transferred, and overhead costs for
minology) related to invasive procedures performed by both a procedure.
the Radiology and the Vascular Surgery departments. The Procedure time and all items used in a given intervention
results of this query provided a list of 401 patients. The were encoded into a support software system and

123
58 B. Janne dOthee et al.: Costs and Outcomes in Claudication

transmitted by the technicians supervisor to the hospital First, the distribution of procedural costs was explored to
Financial Department. The cost of those items corresponded select an appropriate regression model. Second, the rela-
to how much the hospital paid for the products or services it tionships between each of the 19 potential predictors and
needed to ensure its activity and provide healthcare to the logarithm of procedural costs were studied by univar-
patients. Thus, this cost corresponds to hospital expenses, as iate analysis (correlation coefficients for continuous
opposed to the charges billed to third-party payers predictors, t-test assuming equal variances for binary pre-
(including hospital and technical fees), which are part of the dictors). Third, a multivariable linear regression was built.
hospital income and, therefore, not included here. Of note, The strategy for model building was based on the goal of
labor cost incurred in revenue departments included only the analysis, i.e., trying not to miss any significant predictor
the salaries of paramedical staff involved in the procedure, or any confounder and obtaining precise effect estimates
not the salaries of the attending physicians (which fall in a for each of them. To build the kernel of the model, only the
private practice group, not the hospital itself). covariates found to be significantly associated with the
Initial costs included all inpatient and outpatient costs outcome in the univariate analyses were entered in the
over the first month (including procedure and hospital model because such an approach would include all possible
stay); follow-up costs included all activities related to confounders. No covariate was forced into the model based
patient care beyond that, both as outpatient and inpatient. on prespecified clinical presumption. Terms found to be
Global costs were the sum of initial and follow-up costs. confounders (i.e., effect estimate changing by [20%
between crude and adjusted models) were to be included in
the model, whereas collinear terms (standard error of effect
Clinical Data Collection: Predictive Variables estimate changing by [20%) would be studied by appro-
priate diagnostics for possible exclusion. Interactions of
The clinical files of all patients were reviewed in detail clinical interest (specified a priori) to be explored during
using the digital clinical information system of the hospital, model building involved (1) intention to treat and SVS
including all operative and radiological reports related to class, and (2) intention to treat and lesion type. Regression
vascular interventions and reports of outpatient visits during diagnostics were used to verify (1) the robustness of the
follow-up. For each patient, 19 pretreatment predictors of final model in terms of stability to outliers (Cooks D,
interest (i.e., reflecting information obtainable by the studentized and jackknifed residuals) and (2) the appro-
treating physician before choosing the therapeutic strategy) priateness of the model assumptions (normal distribution
were collected. They included a time variable reflecting the and equal variance of the residuals). In case of missing
date of the procedure (continuous variable), patients mean data, the sample size was left floating, in order to maximize
age (continuous) and gender (binary), and a series of six power for each analysis (sample size was only 97) and
cardiovascular risk factors and four markers of the extent of minimize bias, thus increasing generalizability.
cardiovascular disease that were scored according to the
SVS-ISCVS classification [2, 6]. Risk factors included
heredity, sedentarity (both binary), diabetes, smoking, Clinical and Cost Data Collection During Follow-up
hypertension, and hyperlipidemia (multi-category); markers
of the extent of disease (multi-category) included cardiac, Because this study was not a prospective randomized trial,
carotid, renal, and lung disease history. Other collected variables describing clinical outcome during follow-up
predictive variables included baseline ankle-brachial index were also collected in an attempt to detect potential sub-
(ABI), patients class under the SVS classification system group differences that may help to better explain
(binary: class 23 versus 4), intention to treat (binary: a posteriori the findings of the linear regression model.
percutaneous or surgical), level of the lesion(s) treated These follow-up variables were not included in the model,
(aortic, iliac, femoral, popliteal, or their combinations), however, as they carry information that is unavailable to
unilaterality versus bilaterality of lesion(s) of interest clinicians before treatment.
(binary), and lesion type (binary: stenosis versus occlusion). The immediate technical result of the procedure was
assessed by comparing arteriographic films before and after
the procedure when available, based on another scoring
Analysis of Cost Determinants: Building the system proposed by the SVS-ISCVS [2, 6]. By these cri-
Multivariable Regression Model teria, a low score indicates no or minimal disease and high
scores correspond to extensive lesions.
The goal of the model was to explore retrospectively what Clinical outcome was assessed as a multidimensional
predictors could have potentially influenced the cost of the entity including the changes in ABI, the occurrence of
procedure (i.e., the outcome of interest) in our 97 patients. major complications or events during the follow-up period,

123
B. Janne dOthee et al.: Costs and Outcomes in Claudication 59

patency rates, and mortality rates. A 0.15 change in ABI data for costs incurred in hospital (i.e., all hospital cost data
occurring during follow-up was chosen as the unitary could be tracked down by the Finance Department for each
objective measure to define treatment success as recom- patient).
mended in the SVS-ISCVS guidelines [2]. In patients with Procedure costs were not normally distributed (an
unilateral symptoms, only the value corresponding to the unsurprising finding with cost data). To satisfy the linear
limb of interest was retained. In patients in whom the initial regression model assumptions, the outcome variable was
intention was bilateral treatment, the average ABI for both modified using a logarithmic transformation, as this allows
limbs was used (see explanation below). Major complica- better interpretability than other types transformations.
tions (i.e., those resulting in unanticipated need for This successfully restored a normal distribution of the
significant care such as hospital admission or additional outcome variable; no further transformation was deemed
invasive procedure) and reportable events occurring during necessary and a linear regression model was chosen.
follow-up were recorded. Reportable events reflected all
visits to the hospital and clinics during follow-up, thus
including not only complications and subsequent vascular Cost and Clinical Data (Univariate Analysis)
events (e.g., carotid or coronary) but also unrelated non-
vascular events (e.g., cancer). Patency of the treated From the list of 19 potential candidate predictors, six were
segments was categorized into primary, primary assisted, found to be significantly associated with outcome in the
secondary, and failure, according to ad hoc guidelines [2] univariate analyses. Procedural cost in the entire study
and assessed by Kaplan-Meier survival analysis. The same cohort (n = 97) was $8,475 $46 per patient (mean 95%
method was used to analyze mortality rates. confidence interval). Variable ($3,345 $19) and fixed
($3,670 $20) cost components were the main cost
drivers, while overhead costs were less important quanti-
Statistical Considerations tatively ($1,495 $8). Costs and charges of the procedures
were highly correlated (Spearmans q = 0.96, p \ 0.001),
Statistical analyses were performed on a standard spread- as were costs and service counts (q = 0.88, p \ 0.001).
sheet or on SAS 9 (SAS Institute, Cary, NC, USA). Unless Procedural costs were significantly higher (p = 0.001) in
a variable transformation could successfully restore nor- surgical patients ($12,422 $109) than in percutaneous
mality, nonparametric tests were used for non-normally patients ($6,502 $35). That is, the procedural cost of
distributed data. In case of missing data, only available data surgical interventions was, on average, $5,920 (91%)
were analyzed. Data are presented as the mean and its 95% higher per patient than that of percutaneous procedures.
confidence interval (CI) to represent how close our sample Significant differences (p = 0.035 or lower) were also
observations are to the actual population mean. A p value found between these two groups for variable, fixed, and
\0.05 was considered significant in two-tailed analyses. overhead costs. Percutaneous patients were treated under
local anesthesia (n = 64) in angiography suites, versus
epidural (n = 16) or general (n = 17) anesthesia in the
Results operating room for surgical patients.
Other than intention to treat, five more variables were
Description of Study Group and Outcomes found to be significantly associated with procedural costs
during univariate analyses, including SVS class (p =
In all 97 patients, PAD appeared to be due to atheroscle- 0.001), sedentarity (p = 0.017), cardiac status (p = 0.003),
rosis, based on clinical, laboratory, and angiographic unilaterality of lesions (p = 0.006), and lesion morphology
criteria. Among them, 64 (66%) were treated percutane- (p = 0.001).
ously on an intention-to-treat basis while the remaining 33
(34%) were directed to surgical treatment. Procedures
performed included a wide range of techniques, with Cost Determinants (Multivariable Linear Regression
numbers of procedure components per patient that were Model)
similar between both groups (Table 1). The surgical group
had significantly worse baseline arteriographic scores (p = The linear regression model includes all six predictors (all
0.001), lower ABI values (0.13 mean difference, p = 0.021) binary) found to be significant in the univariate analyses.
(Fig. 1) and more occlusions to treat (88%) than the per- This number of predictors remains below the maximal p
cutaneous group (23%). Otherwise, the distribution of risk desirable number of covariates given our sample size ( 97
factors and extent of cardiovascular disease were similar = 9.8). The model has an r2 of 0.45 (F statistic = 10.59,
among these two groups (Table 2). There were no missing p \ 0.001). Three of the predictors (intention to treat, SVS

123
60 B. Janne dOthee et al.: Costs and Outcomes in Claudication

Table 1 Type and number of interventions performed in each patient


group
Procedures performed Total no. performed

Percutaneous group 111


(64 patients)
Balloon angioplasty (PTA) 52
Stenting 42
PTA and stent placement 3
In situ thrombolysis 4
PTA and in situ thrombolysis 1
Atherectomy and PTA 1 Fig. 1 ABI results over time. Mean values of the ankle-brachial
Balloon-wire recanalization 3 index (ABI) are shown for the percutaneous (left, lighter columns)
and surgical groups (right, darker columns) at three different time
Kissing balloons 1 points. Between-group differences were associated with p values of
Kissing stents 4 0.021 before treatment, 0.140 during the first month of treatment, and
Surgical group 68 0.180 at the end of the follow-up period
(33 patients)
Aortobifemoral bypass 10 Table 2 Between-group comparisons on initial characteristics and
Iliofemoral bypass 2 predictors
Femorofemoral bypass 1 Percutaneous Surgical p values
Above-knee femoropopliteal bypass 6 group group
Below-knee femoropopliteal bypass 4 (n = 64) (n = 33)
Femorodistal bypass 6 Risk factors
Iliac endarterectomy 1 Age Mean age 62.7 63.6 NS
Common femoral endarterectomy 14 Sex % of males 63% 73% NS
Profunda femoris endarterectomies 7 Cardiovascular Present 15% 19% NS
Patch angioplasty 7 heredity
Surgical thrombectomy 5 Sedentarity Present 11% 13% NS
Aneurysm repair 2 Diabetes Score = 0 76% 76% NS
Arterial reimplantation 1 Tobacco Score = 0 22% 25% NS
Arteriovenous fistula ligation 1 Hypertension Score = 0 32% 34% NS
Complementary balloon angioplasty 1 Hyperlipidemia Score = 0 59% 53% NS
Runoff Mean score 3.0 3.2 NS
In the percutaneous group, a total of 111 initial procedures were
performed in 64 patients (i.e., several procedures could be performed Indicators of extent of cardiovascular disease
in a given patient at the time of study inclusion, on average 1.7 Heart Score = 0 52% 63% NS
procedure per patient). In the surgical group, 68 procedures were Carotid Score = 0 67% 50% NS
performed in 33 patients (2.1 per patient; p = 0.07)
Kidneys Score = 0 91% 88% NS
Lungs Score = 0 57% 53% NS
class, and unilaterality of lesions) remain significant in the
Initial parameters
multivariate analysis. The other three covariates (sedenta-
Lesion type Frequency of 23% 88% \0.001
rity, heart status, and type of lesion), although significant in occlusions
the univariate analyses, were nonsignificant in the multi- (vs. stenoses)
variate analysis (p = 0.106, 0.069 and 0.353, respectively). ABI Mean ABI 0.67 0.54 0.021
These nonsignificant covariates were kept in the model as Arteriographic Mean score 2.097 2.536 \0.001
they caused confounding of the other, significant predictors score
already included. Removing them from the model would NS, nonsignificant
have caused too much ([20%) imprecision in our estimates For risk factors, Score = 0 indicates the absence of the risk factor of
of the effect of the other predictors already included in the interest. For indicators of the extent of cardiovascular disease, Score =
model. No other confounder needed to be added to the 0 means that patients were asymptomatic or without known disease
predictors already included in the model. Collinearity did for the territory under consideration. Runoff resistance score: decimal
score equal to a base resistance of 1 added to the sum of the weighted
not appear to be a problem. When forced into the model in resistances for each major runoff artery. The contribution of each
addition to their corresponding baseline covariates, the two occlusive lesion to runoff resistance depends on its arterys relative
prespecified interaction terms were not significant and were weight [6]

123
B. Janne dOthee et al.: Costs and Outcomes in Claudication 61

therefore not included in the final model. Regression significant after treatment (either early after or at the end of
diagnostics confirmed the assumptions of normality of the follow-up) (Table 3). The pretreatment differences in
residuals and of equal variance. The number of outliers arteriographic score and ABI values (both worse in the
(i.e., studentized or jackknifed residuals \-2 or [+2) was surgical group) were no longer significant after treatment
close to 5% (5/97), thus similar to what would be expected (Fig. 1). For example, the initial, pretreatment 0.13 dif-
in a normal distribution, and their influence on the model ference (0.67 0.54) in mean ABIs (p = 0.021) dropped to
effects estimates were minimal (Cooks D markedly \1). 0.07 soon after the procedure and 0.05 at the end of the
The outcome variable did not suffer from missing data follow-up, and these last two differences were not statis-
(i.e., all hospital cost data could be tracked down by the tically significant (p = 0.140 and 0.180, respectively). A
Finance Department for each patient). For the six signifi- similar conclusion applied to between-group differences in
cant predictors included in the final model, there were no arteriographic score. Thus, although initial patient selection
missing data for intention to treat, for bilaterality of lesions, for one therapeutic strategy versus the other may have been
and for lesion type, and information regarding SVS class based on clinical criteria (mainly ABIs and runoff), fol-
was missing in only 1 of 97 patients. The number of lowing intervention there was no residual difference related
missing data was 10 for sedentarity (n = 87) and 11 for to the pretreatment patient characteristics. The differences
heart status (n = 86). in procedure-related complication rates were not statisti-
To allow interpretability of the model, the logarithmic cally significantly at 30 days (14% in the percutaneous
transformation of the outcome was reverted, giving the group versus 27% in the surgical group). Reportable events
following equation: Procedure costs = $8,526 + (0.75  within 30 days after treatment occurred with similar fre-
Intention to treat) + (0.58  SVS Class) + (0.72  Uni- quency in the two groups. Cumulative primary, primary
laterality) + (1.34  Sedentarity) + (1.11  Heart) + (0.88  assisted, and secondary patency rates (Table 4, Fig. 2)
Lesion type) + e, where e represents the error term. That were similar between the groups. Despite the fact that the
is, the procedural costs will be 25% lower in a patient mean clinical follow-up exceeded 2 years in both groups,
considered for percutaneous treatment than in another the time scope of the present series may qualify as midterm
patient similar in all respects except for being intended only (as opposed to long-term results), as the Kaplan-Meier
for surgery. Similarly, procedural costs will be 42% lower projections had standard errors exceeding 10% at the 2-
for patients without rest pain (SVS class 23) than for
those with rest pain (class 4), and 28% lower if treated Table 3 Technical and clinical outcomes (comparisons before versus
lesions are unilateral rather than bilateral. Regarding after treatment)
nonsignificant terms that were kept in the model to Mean values Percutaneous Surgical
maintain precision of estimates of the other (i.e., signifi- group group
cant) predictors, procedural costs were 34% higher in (n = 64) (n = 33)
sedentary than in nonsedentary patients, 11% higher for a
Mean clinical follow-up period 941 1,138
patient with history of cardiac disease than for a patient (days)
without, and 12% lower if lesion was a stenosis rather Arteriographic score
than an occlusion. The reader may remember that these Before treatment 2.097 2.536
last three predictors were nonsignificant and were kept in Early after treatment 0.483 0.200
the model only to obtain precise estimates of the effect of p value (for differences before \0.001 \0.001
other predictors. vs. after treatment)
ABI
Early changes (before treatment vs. 1 month after treatment)
Clinical and Cost Data During Follow-up Before treatment 0.67 0.54
Early after treatment (within 1 0.81 0.88
Both groups were followed up for similar mean periods of month)
time: the mean duration of the clinical follow-up (941 p value (for differences before \0.001 \0.001
days) and imaging follow-up (536 days) in the percutane- vs. after treatment)
ous group did not differ significantly from that observed in Global changes (before treatment vs. at end of follow-up)
the surgical group (1,138 (p = 0.093) and 747 days (p = Before treatment 0.67 0.54
0.102), respectively). Ankle-brachial values (ABIs) at the At end of follow-up 0.81 0.86
end of follow-up were available in 83% of percutaneous p value (for differences before \0.001 \0.001
and 91% of surgical patients. vs. after treatment)
Most baseline clinical differences observed between the Within-group comparisons on clinical outcomes before and after
percutaneous and surgical groups were no longer treatment (paired t-tests). p values \ 0.05 are in bold

123
62 B. Janne dOthee et al.: Costs and Outcomes in Claudication

Table 4 Between-group comparisons on clinical outcomes


Percutaneous Surgical p values
group group
(n = 64) (n = 33)

Initial parameters (before treatment)


Mean ABI 0.67 0.54 0.021
Mean arteriographic 2.097 2.536 \0.001
score
Early results (during first month after initial treatment)
Technical success: arteriographic score
Mean score 0.483 0.200 NS Fig. 2 Kaplan-Meier survival analysis: cumulative primary patency
Mean change in score 1.586 2.121 0.005 in each group during the clinical follow-up. Kaplan-Meier survival
(i.e., after before) analysis for primary patency of the treated segment in each group
Clinical outcome shows similar curves during the entire follow-up period
ABI
Mean ABI 0.81 0.88 NS arteriographic score after treatment, nor with the change in
Mean change in ABI 0.15 0.34 NS arteriographic score between pre- and postprocedural
(i.e., after before) assessments. Similarly, costs were not correlated with the
Patency rates ABIs before or after procedure, nor with the difference
Primary patency 98% 100% NS between these. The costs per capita incurred later during
Primary assisted 100% 100% NS follow-up (Table 5) were $1,816 higher in percutaneous than
patency in surgical patients and the between-group difference was
Secondary patency 100% 100% NS statistically significant (p = 0.034). Accordingly, patients in
30-day mortality 0% 0% NS the percutaneous group experienced a higher frequency and
rates an overall greater number of reportable clinical events
Procedure-related 14% 27% NS (including both complications and unrelated visits to the
complication rates at
hospital and clinics) during follow-up compared with the
30 days of follow-up
surgical patients. However, this did not translate into wors-
Final results (at end of follow-up)
ened ABIs. The average global cost (i.e., including both early
Clinical outcome
and follow-up costs) for the percutaneous group (mean =
ABI
$11,849 per patient, 95% CI = $49) was lower than for the
Mean ABI 0.81 0.86 NS
surgical group (mean = $15,577 per patient, 95% CI = $113),
Mean global change in -0.15 -0.32 NS
ABI (before final) but the difference did not reach significance (p = 0.064).
Mean late change in \0.01 0.02 NS
ABI (early after final)
Patency rates at 1 Discussion
year of follow-up
Primary patency 94% 95% NS The present study aimed at identifying cost determinants
Primary assisted patency 97% 100% NS for invasive procedures to treat intermittent claudication
Secondary patency 97% 100% NS with or without rest pain. The main baseline characteristics
Late mortality rate 9.4% 15.2% NS that can be helpful for predicting costs were treatment
Reportable events 213% 200% NS strategy (percutaneous versus surgical, on an intention-to-
during entire treat basis), the presence or absence of ischemic rest pain
follow-up (i.e., class 23 versus 4), and whether the patient had to
NS, nonsignificant undergo unilateral or bilateral treatment. Other covariates
also played a role, although likely less important, in
determining procedural costs. Based on our findings, pre-
year time point [2]. The 30-day mortality was nil. The dictors of clinical outcome are different from predictors of
difference in late mortality rate between the two groups costs, and one should include all these patient character-
(9% in the percutaneous group, 15% in surgical group) was istics in the decision-making process.
not statistically significant. This study also confirmed findings from previous reports
The costs per capita incurred within 30 days after the [1, 7, 8] which showed that the initial procedural cost of
procedure (early costs) did not correlate with the percutaneous treatment for intermittent claudication was

123
B. Janne dOthee et al.: Costs and Outcomes in Claudication 63

Table 5 Cost outcomes


Early costs (30 days) Late costs (follow-up) Global costs
Whole Percutaneous Surgical p value Whole Percutaneous Surgical p value Whole Percutaneous Surgical p value
cohort group group cohort group group cohort group group
Variable 97 64 33 97 64 33 97 64 33

Total costs
Mean $8,475 $6,502 $12,422 0.002 $4,729 $5,347 $3,531 0.034 $13,117 $11,849 $15,577 0.064
95% CI $46 $35 $109 $29 $38 $36 $51 $49 $113
Variable costs
Mean $3,345 $2,785 $4,465 0.035 $2,284 $2,499 $1,867 0.106 $13,117 $5,284 $6,196 0.274
95% CI $19 $16 $46 $13 $17 $17 $22 $23 $46
Fixed costs
Mean $3,670 $2,595 $5,821 \0.001 $1,713 $1,978 $1,201 0.021 $5,346 $4,573 $6,845 0.009
95% CI $20 $15 $45 $11 $15 $14 $22 $20 $48
Overhead costs
Mean $1,459 $1,121 $2,136 0.003 $732 $870 $464 0.008 $2,176 $1,991 $2,535 0.121
95% CI $8 $6 $19 $5 $7 $6 $9 $9 $20
CI, confidence interval
Significant p values (\0.05) are in bold

approximately half that of open bypass surgery. However, Obviously, some patients are clear surgical candidates
our cost calculation method was very different, providing and the cost increase is justified in such cases (e.g., severe
cost values in another range; it is perhaps fortuitous that the acute embolic ischemia). However, there are cases where
cost ratio between the two strategies was similar in these the relative indications for each strategy overlap and are
prior studies and ours. These studies, as well as other subject to discussion on a case-by-case basis. Furthermore,
reports, used ratios of costs-to-charges (RCCs) as a proxy some clinical indications of these therapies are evolving:
for costs, a common but controversial [3] method of cost for example, an increasing number of long occlusions of
data collection. We applied a microeconomic, cost the femoral arteries (a classical indication of surgical
accounting method that calculates the actual operating bypass) are now being treated by percutaneous intentional
expenses at the level of the health care service provider. extraluminal recanalization (PIER) [911] or other tech-
In the present study, the main differences between the niques. Lastly, turf wars between specialties may also
percutaneous and surgical groups before treatment were the influence the choice of treatment in a given patient, which
ABIs, type of lesion, and runoff. Thus, these three may sometimes be different from RCT-based recommen-
parameters appear to be the main factors that influenced the dations and practice guidelines. Thus, there may be cases
choice of therapeutic strategy (a decision made by or in where both therapies enter in competition with each other,
agreement with the referring physician). Typically, those of for example when limited hospital resources have to be
our patients who had worse ABIs, occluded vessels, and shared and both options are reasonable indications in a
poor runoff scores were sent to surgery. The purpose of this given patient. In such cases, one needs to carefully evaluate
study was not to determine whether the percutaneous their respective roles, which is a difficult decision that is
option was better than its alternative in terms of a given best addressed by cost-effectiveness analyses. If we assume
measure of clinical outcome (e.g., patency rate). This has that there are many cases where possible choices of treat-
been the scope of multiple previous reports, and is best ment strategy overlap, percutaneous treatment should
addressed by a prospective randomized controlled trial provide an acceptable result (although not equal to surgery,
(RCT). Our results suggest that, after the preferred thera- by this study) that is less expensive.
peutic option had been determined on the basis of runoff, One could infer that open surgery was the only way to
lesion type, and ABIs, the post-treatment clinical outcome achieve greater ABI improvement and to provide durable
was similar between the two groups, but costs were clearly long-term patency. However, assessing the success of an
different. The commonly held opinion that percutaneous intervention based on these two parameters only is too
treatment of PAD is cheaper than open surgery was verified simplistic, and many additional parameters should be
here, but the critical caveat is that the results may be less considered. Clinical outcome measures such as walking
durable and possibly even less perfect. distance, ABIs, and patency have proved to correlate

123
64 B. Janne dOthee et al.: Costs and Outcomes in Claudication

poorly with quality-of-life (QoL) outcome parameters [12]. intervention, and bypass. Thus comparing costs seems
ABIs do not correlate well with the degree of exercise appropriate. To compare technical outcomes in groups with
impairment, with changes in walking distance after ther- identical lesions remains a good aim for a prospective
apy, or with descriptive health status questionnaires. Such randomized trial. Another limitation of this study comes
questionnaires also correlate imperfectly with angiographic from our use of the average ABI value in patients under-
findings and valuational QoL measures [2]. The wide going bilateral treatment. However, the alternative (i.e.,
variety of QoL metrics used in clinical studies published on study each case on a per limb basis, rather than on a per
PAD [1316] illustrates the lack of agreement among patient basis) carries its own problems too, given that cost
authors and, overall, there is no general consensus on the data were available per patient, not per limb. Had we opted
question of what is the most appropriate method to use for the latter approach, a splitting rule would then have
[1724]. Our findings support these observations: the ABIs been necessary in bilateral cases, to divide costs between
and arteriographic score within 1 month after treatment both limbs in a given patient (several criteria could be used
were typically not correlated. Thus, the success of any to do so). This rule would probably have been somewhat
treatment for PAD should be considered and reported as a arbitrary and would have done just the opposite of what we
multidimensional entity rather than being focused on did for the ABIs. Thus, there was no perfect option here. A
patency rates only. fourth limitation is that costs of surgical synthetic and vein
In addition to using a different and direct cost calculation grafts are lumped together in the surgical intention-to-treat
technique, this study uses another significant variation in group, although costs differences do exist between these
methodology: the cost evaluation was based on the per- materials. Similarly, the percutaneous group includes both
spective of the hospital administrator, not society. balloon angioplasties and stent placements. Although we
Guidelines for cost-effectiveness and cost-utility analyses did not stratify for these technical factors, the analysis still
have recommended that economic evaluations be done using showed significant differences based on intention to treat
the perspective of society and based on the reference case alone. Again, this probably reflects the fact that, regardless
(i.e., a theoretical situation chosen to be as neutral as possible of any technical factor, surgery costs more due to general
in order to allow valid comparisons across different studies). anesthesia and open, invasive intervention. Fifth, the study
However, cost-effectiveness analyses (CEAs) from the focused on predictive variables that are obtainable by
societal perspective are difficult to perform reliably, in par- physicians before starting the procedure (so that the model
ticular because of methodologic difficulties related to cost could be used as a basis for future clinical prediction rule),
accounting. For several reasons, other points of view are also therefore excluding intraprocedural parameters (such as
of interest [25, 26], such as those of the hospital adminis- procedure duration) that may be strongly associated with
trator, the patient, the physician, the insurer, and others. Each costs [28]. Lastly, the costs and events occurring during
of these requires its own specific approach in calculating follow-up included those for which patients returned to our
costs and effects of therapy [26, 27]. In CEAs based on the institution, potentially excluding costs incurred elsewhere.
perspective of the departmental or hospital administrator, However, our institution is located in a rural environment
clinical and economic endpoints may be easier to delineate and is the main tertiary referral center for a wide area; its
and accurate data easier to collect than with the societal geographic position thus generally leads patients to return
approach. By choosing the present approach, we attempted to our hospital rather than seeking care at remote institu-
to provide physicians and other decision-makers with an tions. Accordingly, the use of health care in the United
additional point of view that enhances understanding of the States has been shown to be highly compartmentalized,
economic implications of these invasive treatmentsa field with most patients being treated in facilities close to home
that critically lacks cost data. [29, 30]. Furthermore, such possibly forgotten costs
These data must be interpreted in the context of the incurred elsewhere would be extremely difficult to track,
study design and the applicability of the present study to and their relevance may be questionable when using the
other academic centers remains to be determined. Although perspective of the care provider.
a purpose of this report was to provide cost data by a direct In conclusion, predictors of clinical outcome are dif-
cost calculation method, future multicenter prospective ferent from predictors of costs, and one should include all
studies remain necessary to confirm our findings. Another these patient characteristics in the decision-making pro-
limitation is that ABIs were worse and occlusions more cess. The choice of percutaneous versus surgical strategy,
frequent in the surgical group. It is unlikely, however, that presence of rest pain, and bilaterality of the culprit lesions
the cost of surgery would have been different from the were the main pretreatment determinants of procedural
costs calculated for the surgical group if percutaneous costs. When possible choices of treatment strategy overlap,
patients underwent surgery instead. This is because surgery percutaneous treatment should provide an acceptable result
would still have required general anesthesia, open surgical (although not equal to surgery) that is less expensive.

123
B. Janne dOthee et al.: Costs and Outcomes in Claudication 65

Acknowledgments The authors thank their patients who filled the 14. Khaira HS, Hanger R, Shearman CP (1996) Quality of life in
quality-of-life questionnaires, and Shirley A. Henry and Charles H. patients with intermittent claudication. Eur J Vasc Endovasc Surg
Allen for their help in collecting cost data. The insightful help of 11:6569
Jeffrey Hoch, PhD, was also greatly appreciated during preparation of 15. Whyman MR, Fowkes FG, Kerracher EM, et al (1997) Is inter-
the manuscript. Bertrand Janne dOthee was supported by a Research mittent claudication improved by percutaneous transluminal
Grant from the French Society of Radiology (Societe Francaise de angioplasty? A randomized controlled trial. J Vasc Surg 26:551
Radiologie) and by Dartmouth College. 557
16. Zannetti S, LItalien GJ, Cambria RP (1996) Functional outcome
after surgical treatment for intermittent claudication. J Vasc Surg
24:6573
References
17. Weinstein MC, Stason WB (1982) Cost-effectiveness of coronary
artery bypass surgery. Circulation 66:III5666
1. Hunink MG, Cullen KA, Donaldson MC (1994) Hospital costs of 18. Albers M, Fratezi AC, De Luccia N (1996) Walking ability and
revascularization procedures for femoropopliteal arterial disease. quality of life as outcome measures in a comparison of arterial
J Vasc Surg 19:632641 reconstruction and leg amputation for the treatment of vascular
2. Dormandy JA, Rutherford RB (2000) Management of peripheral disease. Eur J Vasc Endovasc Surg 11:308314
arterial disease (PAD). TASC Working Group. TransAtlantic 19. Gibbons GW, Burgess AM, Guadagnoli E, et al (1995) Return to
Inter-Society Consensus (TASC). J Vasc Surg 31:S1S296 well-being and function after infrainguinal revascularization. J
3. Janne dOthee B, Langdon DR, Bell GK, Bettmann MA (2006) Vasc Surg 21:3544; discussion 4445
Operating expenses for the diagnosis and treatment of peripheral 20. Jenkinson C, Gray A, Doll H, Lawrence K, Keoghane S, Layte R
vascular disease in an academic interventional radiology depart- (1997) Evaluation of index and profile measures of health status
ment: Cost calculations according to a microeconomic method. J in a randomized controlled trial. Comparison of the Medical
Vasc Interv Radiol 17:8594 Outcomes Study 36-Item Short Form Health Survey, EuroQol,
4. Hoch JS, Briggs AH, Willan AR (2002) Something old, some- and disease specific measures. Med Care 35:11091118
thing new, something borrowed, something blue: A framework 21. Pell JP (1995) Impact of intermittent claudication on quality of
for the marriage of health econometrics and cost-effectiveness life. The Scottish Vascular Audit Group. Eur J Vasc Endovasc
analysis. Health Econ 11:415430 Surg 9:469472
5. Dacey LJ, Dow RW, McDaniel MD, Walsh DB, Zwolak RM, 22. Dorman P, Slattery J, Farrell B, Dennis M, Sandercock P (1998)
Cronenwett JL (1988) Cost-effectiveness of intra-arterial throm- Qualitative comparison of the reliability of health status assess-
bolytic therapy. Arch Surg 123:12181223 ments with the EuroQol and SF-36 questionnaires after stroke.
6. Rutherford RB, Baker JD, Ernst C, et al (1997) Recommended United Kingdom Collaborators in the International Stroke Trial.
standards for reports dealing with lower extremity ischemia: Stroke 29:6368
Revised version. J Vasc Surg 26:517538 23. Busschbach JJ, McDonnell J, Tangelder MJ, et al (1999) EuroQol
7. Jansen RM, de Vries SO, Cullen KA, Donaldson MC, Hunink values for economic modeling quality of life after infrainguinal
MG (1998) Cost-identification analysis of revascularization pro- bypass grafting surgery: a rectification. J Vasc Surg 30:1162
cedures on patients with peripheral arterial occlusive disease. J 1163
Vasc Surg 28:617623 24. Hiatt WR (1997) Quality of life assessment in peripheral arterial
8. Doubilet P, Abrams HL (1984) The cost of underutilization. disease. Atherosclerosis 131(Suppl):S3536
Percutaneous transluminal angioplasty for peripheral vascular 25. Meltzer D, Johannesson M (1999) Inconsistencies in the societal
disease. N Engl J Med 310:95102 perspective on costs of the Panel on Cost-Effectiveness in
9. Bolia A (1998) Percutaneous intentional extraluminal (subinti- Health and Medicine. Med Decis Making 19:371377
mal) recanalization of crural arteries. Eur J Radiol 28:199204 26. Powe NR (1994) Economic and cost-effectiveness investigations
10. Spinosa DJ, Leung DA, Harthun NL, et al (2003) Simultaneous of radiologic practices. Radiology 192:1118
antegrade and retrograde access for subintimal recanalization of 27. Powe NR (1995) Cost-effectiveness analyses. N Engl J Med
peripheral arterial occlusion. J Vasc Interv Radiol 14:14491454 332:123124; discussion 124125
11. Treiman GS, Whiting JH, Treiman RL, McNamara RM, Ashrafi 28. Janne dOthee B, Faintuch S, Reedy AW, Nickerson CF, Rosen
A (2003) Treatment of limb-threatening ischemia with percuta- MP (2008) Retrievable versus permanent caval filter procedures:
neous intentional extraluminal recanalization: A preliminary When are they cost-effective for Interventional Radiology? J
evaluation. J Vasc Surg 38:2935 Vasc Interv Radiol: in press
12. Sheiman RG, Janne dOthee BJ (2007) Screening carotid 29. Huber TS, Seeger JM (2001) Dartmouth Atlas of Vascular Health
sonography before elective coronary artery bypass graft surgery: Care review: Impact of hospital volume, surgeon volume, and
Who needs it? AJR Am J Roentgenol 188:W475479 training on outcome. J Vasc Surg 34:751756
13. Currie IC, Wilson YG, Baird RN, Lamont PM (1995) Treatment 30. Dartmouth Medical School CfECS (2000) The Dartmouth Atlas
of intermittent claudication: The impact on quality of life. Eur J of Vascular Health Care. AHA Press, Health Forum, Chicago
Vasc Endovasc Surg 10:356361

123
Cardiovasc Intervent Radiol (2008) 31:6672
DOI 10.1007/s00270-007-9187-2

CLINICAL INVESTIGATION

Infarction of Uterine Fibroids After Embolization: Relationship


Between Postprocedural Enhanced MRI Findings and Long-Term
Clinical Outcomes
Tetsuya Katsumori Toshiyuki Kasahara
Yoko Kin Taiki Nozaki

Received: 25 May 2007 / Accepted: 11 September 2007 / Published online: 18 October 2007
 Springer Science+Business Media, LLC 2007

Abstract Keywords Embolization  Fibroid  MRI  Outcome 


Purpose To retrospectively evaluate the relationship Uterus
between the degree of infarction of uterine fibroids on
enhanced MRI after embolization and long-term clinical
outcomes. Introduction
Methods During 92 months, 290 consecutive patients
with symptomatic uterine fibroids were treated with Uterine artery embolization has been recognized as a safe
embolization; 221 who underwent enhanced MRI before and effective treatment which can provide long-term
embolization and 1 week after embolization were included symptom control for symptomatic uterine fibroids [16].
in this study. The infarction rates of all fibroid tissue were MRI with no administration of gadolinium (unenhanced
assessed using enhanced MRI after embolization. Patients MRI) before embolization is a useful method for predicting
were divided into three groups according to the infarction short- and mid-term outcomes by evaluating the signal
rates: group A (100% infarction, n = 142), group B (90 intensity of uterine fibroids on T1-weighted images and T2-
99% infarction, n = 74), group C (\90% infarction, n = 5). weighted images, as previously reported in the literature [7,
The cumulative rates of clinical outcomes were compared 8]. Recently, Pelage et al. [9] found that, in a study of MRI
among groups using the Kaplan-Meier limited method. with administration of gadolinium (enhanced MRI) after
Results Group A had a significantly higher rate of embolization, the degree of infarction of uterine fibroids was
symptom control than groups B and C. The cumulative a useful indicator of long-term clinical outcomes. They
rates of symptom control at 5 years were 93%, 71%, and reported that incompletely infarcted fibroid tissue after
60% in groups A, B, and C, respectively. Group A had a embolization might not affect short-term clinical outcomes
significantly lower rate of gynecologic intervention after but could regrow, causing symptom recurrence in the long
embolization than groups B and C. The cumulative rates of term [9]; however, their preliminary study dealt with a small
additional gynecologic intervention at 5 years were 3%, number of cases (n = 20) and had a limited follow-up period.
15%, and 20% in groups A, B, and C, respectively. The objective of this study was to retrospectively assess the
Conclusions The degree of infarction of uterine fibroids relationship between the degree of infarction of uterine fib-
after embolization on enhanced MRI was related to long- roids after embolization observed on enhanced MRI and
term clinical outcomes. Complete infarction of all fibroid long-term clinical outcomes, including symptom control and
tissue can induce a higher rate of symptom control, with a additional gynecologic intervention, using a larger number
lower rate of additional gynecologic intervention in the long of cases and much longer follow-up.
term compared with incomplete infarction of fibroid tissue.

Materials and Methods


T. Katsumori (&)  T. Kasahara  Y. Kin  T. Nozaki
Department of Radiology, Saiseikai Shiga Hospital,
Ohashi 2-4-1, Ritto, Shiga 520-3046, Japan During 92 months, from December 1997 to July 2006, 290
e-mail: katsumo@eurus.dti.ne.jp consecutive women underwent uterine artery embolization

123
T. Katsumori et al.: Fibroids Infarction after UAE on Enhanced MRI 67

using only gelatin sponge particles for symptomatic uterine embolization; 203 patients (92%) underwent the MRI at
fibroids at our hospital. We selected 221 patients who our hospital and the remaining 18 patients (8%) at out-
underwent unenhanced and enhanced MRI before emboli- side hospitals.
zation and at 1 week after embolization. The remaining 69 All patients underwent unenhanced and enhanced MRI
women did not meet these criteria: 59 patients underwent 1 week after embolization at our hospital. The initial 154
only unenhanced MRI before embolization at outside hos- patients underwent MRI 1 week after embolization at our
pitals; 8 patients underwent only unenhanced MRI before hospital using a 1.0 T system (Magnetom Impact, Sie-
embolization at our hospital; and the remaining 2 women did mens Medical System, Germany) using a CP-body coil.
not undergo enhanced MRI at 1 week after embolization due T1-weighted images before and after contrast adminis-
to allergy to the contrast medium. We excluded the patients tration were obtained by breath-hold two-dimensional
who had not undergone enhanced MRI before embolization fast low-angle shot sequences and T2-weighted images
because it was difficult to assess the viability of tumors by breath-hold fast spin-echo sequences. T1-weighted
without enhanced MRI. The 221 women enrolled were pre- images were acquired with the following parameters: TR/
menopausal Asians, with a mean age of 42.8 3.8 years TE,165/5.0; field of view, 30 cm; matrix, 256 154;
(range 3054 years), and had fibroid-related symptoms: 180 section thickness, 10 mm; inter-section gap, 2 mm. T2-
women had menorrhagia, 80 women had pain during men- weighted images were acquired with the following
struation, and 174 had bulk-related symptoms. Their mean parameters: 3685/128; echo train, 23; field of view, 30
body weight was 54.4 3.8 kg (range 4095 kg). Seventy-six cm; matrix, 256 138; section thickness, 10 mm; inter-
patients had previously undergone Gn-RH analog treatment, section gap, 2 mm. Enhanced MR images were obtained
27 patients myomectomy, 14 patients had taken birth control approximately several minutes after intravenous infusion
pills, and 1 patient had had embolization. All women gave of 10 ml of gadolinium (Magnevist, Schering, Germany)
oral and written informed consent to undergo embolization by manual operation. The same protocol was used for
and MRI. The institutional ethics committee approved the unenhanced and enhanced MRI before embolization at
present study. our hospital. Images could only be reviewed using con-
ventional films.
The subsequent 67 women underwent MRI 1 week
Embolization Procedure
after embolization at our hospital using a 1.5 T system
(Gyroscan Intera, Philips Medical Systems, Best, The
Uterine artery embolization was performed via a unilateral
Netherlands) with a SENSE body coil, because the MR
femoral approach using local anesthesia, as previously
unit at our hospital was replaced in June 2004. The
reported by us [10]. A microcatheter was used in all but 4
following sequences were obtained: axial T1-weighted
patients. The embolic agent used was only gelatin sponge
spin-echo (TR/TE, 550/10; echo-train length, 3; slice
particles of approximately 5001,000 lm, which were made
thickness, 6 mm; gap, 1 mm; excitations, 2; field of
from gelatin sponge sheets (Yamanouchi, Tokyo, Japan)
view, 25 cm; matrix, 256 512; and respiratory com-
using a scalpel and small scissors by the operators. The mi-
pensation), axial T2-weighted turbo spin-echo (TR/TE,
crocatheter was advanced into the transverse segment of the
4151/100; slice thickness, 6 mm; gap, 1 mm; excitations,
uterine artery in most cases, and the mixture of the embolic
3; field of view, 25 cm; and respiratory compensation),
agent, saline and contrast medium was slowly infused. The
and sagittal T2-weighted turbo spin-echo (TR/TE, 4151/
angiographic endpoint was complete or near stasis of the
100; echo-train length,15; slice thickness, 6 mm; gap, 1
ascending uterine artery. Bilateral uterine artery emboliza-
mm; excitations, 3; field of view, 25 cm; and respiratory
tion was successfully performed in 99% (219/221) of all
compensation). Enhanced MRI was performed within
patients. Of the 2 remaining patients, the contralateral uterine
approximately 5 min after intravenous administration of
artery was absent in 1 patient and one uterine artery could not
10 ml of gadolinium using a power injector or by
be technically catheterized in the other. Of the 219 patients,
manual operation. The sequences were as follows: axial
the accessory uterine artery could not be embolized in 2
T1-weighted spin-echo (TR/TE, 550/10; slice thickness,
patients, because of catheterization failure. No ovarian artery
6 mm; gap, 1 mm; excitations, 2; field of view, 25 cm;
embolization was performed in any patient. No repeat
and respiratory compensation), and sagittal T1-weighted
angiography or embolization was performed in any patient.
spin-echo (TR/TE, 550/10; slice thickness, 6 mm; gap, 1
mm; excitations, 2; field of view, 25 cm; and respiratory
Imaging Technique compensation). The same protocol was used for unen-
hanced and enhanced MRI before embolization at our
A total of 221 women underwent unenhanced and hospital. Images could be reviewed using liquid crystal
enhanced axial and sagittal pelvic MRI before display monitors.

123
68 T. Katsumori et al.: Fibroids Infarction after UAE on Enhanced MRI

Clinical Follow-up questionnaire or at the time of the last routine or unantic-


ipated hospital visit.
We routinely followed up patients at 1 week, 4 months, and
1 year after embolization. They underwent unenhanced and
enhanced MRI to assess uterine fibroids at the time of the Data Analysis
routine follow-up. We also instructed all the women to
inform us of any adverse reactions should these occur. All data were collected in August 2006. The initial 154
After 1 year, when problems related to uterine fibroids or MRI series was reviewed using all available images printed
embolization, including failure of symptom control or on conventional films, whereas the subsequent 67 MRI
adverse reactions occurred and the patients wanted to visit series were reviewed manually by scrolling through all
our hospital, we examined them and performed unen- available images shown on liquid crystal display monitors,
hanced and enhanced MRI as necessary. It had been more by experienced radiologists. The information before
than 1 month since embolization in all patients. Thirty-six embolization was obtained from a prospectively acquired
women (16%, 36/221) were lost to follow-up during this database of uterine artery embolization for symptomatic
period. Of these 36 patients, 2 were lost to follow-up at 4 uterine fibroids performed at our hospital. The largest
months, 15 at 1 year, 13 at 2 years, 5 at 4 years, and 1 at 5 tumor volume and uterine volume before embolization
years. The mean follow-up period was 29.6 months (range were measured with baseline unenhanced and enhanced
0.2584 months). MRI using the formula of a prolate ellipse (length depth
width 0.5233). In this retrospective study, we defined
fibroid tissue as all uterine fibroids of 1 cm or larger in size.
Questionnaire Coexisting adenomyosis was excluded from the fibroid
tissue. We defined infarction of fibroid tissue as when the
Baseline clinical symptoms were assessed in all women intensity on enhanced MRI did not change compared with
using an oral questionnaire before embolization. Written that on unenhanced MRI. The information of complete or
questionnaires were obtained from each patient at 4 partial enhancement of all fibroid tissue before emboliza-
months, 1 year and annually after embolization to evaluate tion was obtained by the prospectively acquired database
changes in fibroid-related symptoms compared with the with baseline unenhanced and enhanced MRI. Next, two
baseline status and additional gynecologic intervention. masked experienced radiologists independently interpreted
Symptom changes were classified into three grades all images of axial and sagittal enhanced MRI obtained 1
improved, no change, and worsenedon the basis of week after embolization and subsequently estimated the
responses to serial written questionnaires. During the fol- subjective impression of infarction rates of all fibroid tissue
low-up period, when patients underwent additional (range 0100%; interval 10%) after embolization. Any
gynecologic treatments, including hysterectomy, myo- differences were resolved by consensus in a subsequent
mectomy, hysteroscopic intervention, dilatation and discussion. Thereafter, the infarction rates after emboliza-
curettage, embolization, and angiography, we did not tion were divided for each patient into three categories:
obtain subsequent questionnaires because these gyneco- group A, 100%; group B, 9099%; and group C, less than
logic treatments could have affected the outcomes of 90% (Fig. 1).
embolization. We defined patients as lost to follow-up The characteristics of each group were measured and
when responses to questionnaires could not be obtained. compared among the three groups using the v2 test and
These patients were considered to be censored at this point, Wilcoxon rank sum test. The cumulative rates of symptom
and their last responses on serial questionnaires were control and additional gynecologic intervention were
considered to indicate patient status. Symptom control was measured for each group using the Kaplan-Meier product
defined in the present study when women responded that limit method and compared using the Log-rank test. A
symptoms had improved on the last questionnaire or at the probability value of less than 5% (p\0.05) was considered
time of the last routine or unanticipated hospital visit, and significant.
had not undergone any further gynecologic intervention
during the follow-up period. Additional gynecologic
intervention was defined as when patients underwent hys- Results
terectomy, myomectomy, hysteroscopic intervention,
dilatation and curettage, embolization, or angiography due MRI before embolization showed the largest tumor volume
to fibroid-related symptoms, including failure of symptom and the uterine volume had a mean of 321 330 ml (range
control and complication-related symptoms, or new uterine 41,972 ml) and 889 545 ml (range 1512,853 ml),
disease. Information was obtained from the last respectively. The largest tumor was located in the

123
T. Katsumori et al.: Fibroids Infarction after UAE on Enhanced MRI 69

Fig. 1 AC Three groups


categorized by infarction rates
of all fibroid tissue on enhanced
MRI 1 week after embolization.
A Group A, 100% infarction of
all fibroid tissue after
embolization. B Group B, 90
99% infarction of all fibroid
tissue after embolization. C
Group C, less than 90%
infarction of all fibroid tissue
after embolization

submucosal area in 70 patients, in the intramural area in 0.0177; group A vs. C, p = 0.0483). There was no signif-
120 patients, and in the subserosal area in 31 patients. The icant difference in additional gynecologic intervention
number of uterine fibroids 1 cm or larger per patient had a rates between groups B and C (p = 0.6065). The cumulative
mean of 6.4 5.9 (range 132). Of 221 patients, 13 had rates of additional gynecologic intervention at 5 years after
coexisting adenomyosis. Enhanced MRI before emboliza- embolization were 3%, 15%, and 20% in groups A, B, and
tion showed that all fibroid tissue was enhanced in 196 C, respectively (Fig. 4).
patients and the vast majority of all fibroid tissue was
enhanced in the remaining 25 patients.
There were 142 patients in group A (100% infarction of Discussion
all fibroid tissue after embolization), 74 patients in group B
(9099% infarction), and 5 patients in group C (less than Our results suggest that patients with complete infarction
90% infarction). Table 1 shows the characteristics of each of all fibroid tissue after embolization had significantly
group. higher rates of symptom control at long-term follow-up
Fig. 2 shows the cumulative rates of symptom control than those with incomplete infarction of tumors after
by the infarction grade of all fibroid tissue after emboli- embolization. The results corresponded with those reported
zation. Group A had a significantly higher rate of symptom by Pelage et al. [9], who found that complete infarction of
control than groups B and C (group A vs. B, p = 0.0296; fibroids could provide long-term symptom control whereas
group A vs. C, p = 0.0014). There was no significant dif- a residual perfusion area in the fibroid could be a cause of
ference in symptom control rates between groups B and C symptom failure or recurrence in the long term on the basis
(p = 0.2495). The cumulative rates of symptom control at 5 of serial enhanced MRI after embolization. In addition, our
years were 93%, 71%, and 60% in group A, B, and C, results suggest that patients with complete infarction of all
respectively. fibroid tissue after embolization have a lower rate of
Fig. 3 shows the cumulative rates of additional gyne- additional gynecologic intervention than those with
cologic intervention after embolization by the infarction incomplete infarction of tumors after embolization; there-
grade of all fibroid tissue. Group A had a significantly fore, we believe that it is necessary for long-term symptom
lower rate of additional gynecologic intervention after control to induce infarction of all fibroid tissue as much as
embolization than groups B and C (group A vs. B, p = possible, without complications.

123
70 T. Katsumori et al.: Fibroids Infarction after UAE on Enhanced MRI

Table 1 Characteristics of each


Group A Group B Group C p value
group
Number 142 74 5
Age (years) 43.2 3.4 41.8 4.0 44.2 5.7 0.0404
Previous surgery 11.3% (16/142) 12.2% (9/74) 40% (2/5) 0.2748
Fibroid
Dominant tumor volume (ml) 259 249 441 427 316 275 0.0084
Numbers 6.0 5.8 7.2 6.2 4.6 3.9 0.3383
Uterine volume (ml) 770 438 1112 656 988 519 0.0006
Symptoms
Menorrhagia 83.8% (119/142) 78.4% (58/74) 60% (3/5) 0.3292
Pain during menstruation 30.3% (43/142) 48.6% (36/74) 20% (1/5) 0.0222
Bulk-related symptoms 73.2% (104/142) 89.2% (66/74) 80% (4/5) 0.0178
Coexisting adenomyosis 5.6% (8/142) 4.1% (3/74) 40% (2/5) 0.0645
Observed period (months) 33.1 21.5 22.3 21.5 37.5 33.5 0.0003

Fig. 2 Cumulative rates of symptom control after embolization. The Fig. 3 Cumulative rates of additional gynecologic intervention after
cumulative rates of symptom control at 5 years after embolization embolization. The cumulative rates of additional gynecologic inter-
were 93% in group A, 71% in group B, and 60% in group C. Group A vention at 5 years after embolization were 3% in group A, 15% in
versus B, p = 0.0296; group A versus C, p = 0.0014; group B versus group B, and 20% in group C. Group A versus B, p = 0.0177; group A
C, p = 0.2495 versus C, p = 0.0483; group B versus C, p = 0.6065

Kroencke et al. [11] reported, in a prospective study been influenced by several factors, including study design
dealing with 115 women to assess the relationship between (prospective versus retrospective), timing of enhanced MRI
devascularization of all fibroid tissue on enhanced MRI 72 hr obtained (72 hr versus 1 week after embolization), length of
after embolization and clinical outcomes followed up to 36 the follow-up period (up to 36 months vs. up to 84 months),
months, that the 100% devascularization group (n = 60) had and embolic agents (spherical polyvinyl alcohol or Tris-acryl
similar rates of improvement in menorrhagia and bulk- microsphere versus gelatin sponge particles). With regard to
related symptoms compared with the 90-99% devasculari- the 089% devascularization group, because the sample size
zation group (n = 32), and had significantly higher rates than of our study was too small (n = 5), we think it might have been
the 089% devascularization group (n = 23). Their results are difficult to assess the detailed outcomes of this group.
likely to be somewhat different from ours because in the With regard to the group with 9099% infarction of all
present study the 100% devascularization group (n = 142) fibroid tissue after embolization, the cumulative rate of
had significantly higher rates of symptom control than the symptom control was 71% at 5 years, which was signifi-
9099% devascularization group (n = 74). The difference in cantly lower than that (93%) in the group with 100%
results between Kroencke et al. [9] and ourselves might have infarction of all fibroid tissue after embolization. In

123
T. Katsumori et al.: Fibroids Infarction after UAE on Enhanced MRI 71

Fig. 4 AD A 41-year-old
woman with menorrhagia and
bulk-related symptoms due to
multiple uterine fibroids. She
had previously undergone
abdominal myomectomy and
treatment with Gn-RH analog.
A Enhanced MRI before
embolization showed the uterus
was markedly enlarged due to
multiple uterine fibroids, which
were completely enhanced. B
Enhanced MRI 1 week after
embolization showed more than
90% infarction of all fibroid
tissues (group B). C Enhanced
MRI 1 year after embolization
showed uterus was decreased in
volume with residual enhancing
fibroids. The primary symptoms
markedly improved. D
Enhanced MRI 2 years and 6
months after embolization
showed an enlarged uterus due
to viable fibroid tissues, which
markedly increased in volume.
She had recurrent symptoms
and underwent abdominal
hysterectomy at 4 years after
embolization

addition, the cumulative rate of additional gynecologic have affected the findings of enhanced MRI at 1 week after
intervention in this group was 15% at 5 years, which was embolization. In addition, the former images were
significantly higher than that (3%) in the group with 100% reviewed using conventional films, while the latter were
infarction of all fibroid tissue after embolization. However, reviewed using liquid crystal display monitors, which
the cumulative rate of additional gynecologic intervention might induce a bias. Third, masked experienced radiolo-
in this group was very similar to that after myomectomy, gists retrospectively estimated the infarction rates of all
reported to be 10% by 5 years using life-table analysis [12]. fibroid tissue, but a prospective study would be more
Symptom control was assessed on the basis of serial appropriate. Fourth, the infarction rates of all fibroid tissue
questionnaires obtained in the present study, but the long- were estimated from the subjective impressions of each
term anatomic change of fibroid tissue in the long term was reader, which might have induced a potential bias; however
not systematically assessed on imaging, including we had no objective method to calculate the rates of all
enhanced and unenhanced MRI. As a result, we could not fibroid tissue after embolization.
evaluate which factors, including a residual perfusion area We conclude that the degree of infarction of uterine fib-
in the fibroids, development of new fibroids, or occurrence roids after embolization on enhanced MRI is related to long-
of other disease, were related to the failure of symptom term clinical outcomes. Complete infarction of all fibroid
control. Thus, a further detailed study will be required to tissue after embolization can induce higher rates of symptom
assess the relationship between anatomic change and control, with lower rates of additional gynecologic inter-
symptom control in the long term. vention in the long term, compared with incomplete
There were several limitations of this study. First, this infarction of all fibroid tissue after embolization.
was a retrospective study performed at a single institute.
Second, the initial 154 patients underwent enhanced MRI 1
References
week after embolization with a 1.0 T system, whereas the
subsequent 67 patients underwent enhanced MRI 1 week 1. Ravina JH, Herbreteau D, Ciraru-Vigneron N, et al. (1995)
after embolization with a 1.5 T system. As the two MRI Arterial embolisation to treat uterine myomata. Lancet 346:671
units had different resolutions and protocols, this might 672

123
72 T. Katsumori et al.: Fibroids Infarction after UAE on Enhanced MRI

2. Worthington-Kirsch R, Spies JB, Myers ER, et al. (2005) The 8. Jha RC, Ascher SM, Imaoka I, et al. (2000) Symptomatic fibro-
Fibroid Registry for Outcomes Data (FIBROID) for uterine leiomyomata: MR imaging of the uterus before and after uterine
embolization: Short-term outcomes. Obstet Gynecol 106:5259 arterial embolization. Radiology 217:228235
3. Spies JB, Myers ER, Worthington-Kirsch R, et al. (2005) The 9. Pelage JP, Guaou NG, Jha RC, et al. (2004) Uterine fibroid
Fibroid Registry: Symptom and quality-of-life status 1 year after tumors: Long-term MR imaging outcome after embolization.
therapy. Obstet Gynecol 106:13091318 Radiology 230:803809
4. Spies JB, Bruno J, Czeyda-Pommersheim F, et al. (2005) Long- 10. Katsumori T, Nakajima K, Mihara T, et al. (2002) Uterine artery
term outcome of uterine artery embolization of leiomyomata. embolization using gelatin sponge particles alone for symptom-
Obstet Gynecol 106:933939 atic uterine fibroids: Midterm results. AJR Am J Roentgenol
5. Walker WJ, Barton-Smith P (2006) Long-term follow up of 178:135139
uterine artery embolisation: An effective alternative in the 11. Kroencke TJ, Scheuring C, Schink T, et al. (2006) Devasculari-
treatment of fibroids. Br J Obstet Gynaecol 113:464468 zation of fibroids after UAE determines clinical outcome: A
6. Katsumori T, Kasahara T, Akazawa K (2006) Long-term outcomes prospective study. J Vasc Interv Radiol 17:743
of uterine artery embolization using gelatin sponge particles alone 12. Fauconnier A, Chapron C, Babaki-Fard K, et al. (2000) Recur-
for symptomatic fibroids. AJR Am J Roentgenol 186:848854 rence of leiomyomata after myomectomy. Hum Reprod Update
7. Burn PR, McCall JM, Chinn RJ, et al. (2000) Uterine fibroleio- 6:595602
myoma: MR imaging appearances before and after embolization
of uterine arteries. Radiology 214:729734

123
Cardiovasc Intervent Radiol (2008) 31:7385
DOI 10.1007/s00270-007-9195-2

CLINICAL INVESTIGATION

Midterm Clinical and First Reproductive Results of a


Randomized Controlled Trial Comparing Uterine Fibroid
Embolization and Myomectomy
Michal Mara Jana Maskova Zuzana Fucikova David Kuzel
Tomas Belsan Ondrej Sosna

Received: 8 March 2007 / Accepted: 13 September 2007 / Published online: 18 October 2007
Springer Science+Business Media, LLC 2007

Abstract The purpose of this study was to compare the women after myomectomy and 26 after UAE have tried to
midterm results of a radiological and surgical approach to conceive, and of these we registered 50 gestations in 45
uterine fibroids. One hundred twenty-one women with women. There were more pregnancies (33) and labors (19)
reproductive plans who presented with an intramural and fewer abortions (6) after surgery than after emboliza-
fibroid(s) larger than 4 cm were randomly selected for tion (17 pregnancies, 5 labors, 9 abortions) (p \ 0.05).
either uterine artery embolization (UAE) or myomectomy. Obstetrical and perinatal results were similar in both
We compared the efficacy and safety of the two procedures groups, possibly due to the low number of labors after UAE
and their impact on patient fertility. Fifty-eight emboliza- to date. We conclude that UAE is less invasive and as
tions and 63 myomectomies (42 laparoscopic, 21 open) symptomatically effective and safe as myomectomy, but
were performed. One hundred eighteen patients have fin- myomectomy appears to have superior reproductive out-
ished at least a 12-month follow-up; the mean follow-up in comes in the first 2 years after treatment.
the entire study population was 24.9 months. Embolized
patients underwent a significantly shorter procedure and Keywords Comparison  Myomectomy  Reproduction 
required a shorter hospital stay and recovery period. They Uterine fibroid embolization
also presented with a lower CRP concentration on the
second day after the procedure (p \ 0.0001 for all
parameters). There were no significant differences between Embolization of uterine fibroids, which was first introduced
the two groups in the rate of technical success, symptom- by Ravina et al. [1] in 1995, is now a well-established
atic effectiveness, postprocedural follicle stimulating therapeutic procedure. Uterine artery embolization (UAE)
hormone levels, number of reinterventions for fibroid is mostly performed in women with symptomatic fibroids
recurrence or regrowth, or complication rates. Forty who do not plan further pregnancy. Its use in women with
reproductive plans is still controversial. Evidence exists
that there is a possible risk of infection, which may lead to
M. Mara  Z. Fucikova  D. Kuzel  O. Sosna sepsis and hysterectomy and premature ovarian failure in
Department of Obstetrics and Gynecology, General Faculty
Hospital and First Medical Faculty of Charles University,
these patients [24].
Apolinarska 18, 128 00 Prague 2, Czech Republic Myomectomy is considered to be a standard treatment
procedure for the removal of myomas in patients of fertile
J. Maskova  T. Belsan age [5]. This procedure can also be associated with dra-
Department of Radiology, Central Military Hospital, U Vojenske
nemocnice 1200, 169 02 Prague 6, Czech Republic
matic complications such as perioperative bleeding and
uterine rupture during subsequent gravidity, which may
M. Mara (&) require hysterectomy. We thus decided to compare the
Department of Obstetrics & Gynecology, 1st Medical Faculty of efficacy and safety of these two approaches in women with
Charles University, Apolinarska 18, 128 00 Prague 2, Czech
Republic
reproductive plans who presented with intramural uterine
e-mail: michal.mara@quick.cz fibroids. The first results of this trial have already been

123
74 M. Mara et al.: Embolization Versus Myomectomy

published [6]. The aim of this paper is to present midterm symptoms and 10 = maximal intensity of symptoms. The
clinical and first reproductive results obtained in a larger following symptoms were evaluated: (i) menorrhagia and/
group of patients. or hypermenorrhea, (ii) dysmenorrhea, (iii) dyspareunia,
(iv) pelvic pain, (v) dysuria and/or urinary frequency, and
(vi) pressure symptoms.
Materials and Methods The following inclusion criteria were set: (i) US-verified
intramural fibroid at least 4 cm at its largest diameter (in
Trial Design the case of multiple fibroids, at least one with a size of 4
cm), (ii) age \40 years, (iii) serum FSH concentration \30
The trial was approved by the Ethics Committee of the First IU/L (on the third day of menstrual cycle), and (iv) planned
Medical Faculty of Charles University. It was designed as pregnancy. Intramural fibroids were defined using US as a
prospective and randomized. Every newly recruited patient uterine wall expansion of typical echo structure with the
was randomly assigned a computer-generated integral prevalent part of its volume inside the myometrium.
number from 1 to 100 (using a random number generating Exclusion criteria included (i) nonintramural localiza-
program available at http://www.random.org, by Mads tion of the main fibroid (submucosal and subserous); (ii)
Haahr, Distributed Systems Group, Department of Com- size of the dominant myoma [12 cm in its largest diameter
puter Science, University of Dublin, and Trinity College, (according to US) or uterus enlarged to the size corre-
Ireland). This was always done at the point of randomiza- sponding to [4 months of pregnancy (according to
tion, so that no researcher could know or predict any bimanual pelvic examination); (iii) previous myomectomy,
subsequent number. Patients who were assigned an odd embolization, or hormonal therapy of fibroids with GnRH
number were included in group E (embolization), and agonists or Danazol; (iv) suspected uterine sarcoma or
patients who were assigned an even number in group M diffuse adenomyosis (according to US or MRI); and (v)
(myomectomy). The researcher in charge of patient serious disease contraindicating gravidity.
recruitment and detailed instruction has always been dif-
ferent from the researcher accomplishing randomization.
This second doctor did not know any patient details at the Embolization Procedure
time of randomization and was only notified that the patient
met the criteria for trial entry. Patients included in group E underwent bilateral UAE. The
access for the procedure was from the right groin via the
right common femoral artery. The aim was to bilaterally
Study Population embolize the ascending branches of the uterine artery
supplying the fibroid in order to achieve a complete loss of
One hundred forty-nine women with uterine fibroid or fibroid perfusion and, at the same time, leave free flow in
fibroids and unfinished reproductive plans were examined the main stems and in cervico-vaginal branches of both
from November 2001 to December 2005. All patients uterine arteries. We refrained from embolization of sites
entering the trial underwent gynecological examination and displaying significant utero-ovarian anastomoses of type III
abdominal and transvaginal ultrasonography (US) exami- (main ovarian blood supply arises from the uterine artery)
nation of the small pelvis, including Doppler examination [7], which could not be overcome by microcatheter.
of uterus and fibroids. Serum levels of follicle stimulating The technique of free flow embolization was
hormone (FSH) luteinizing hormone (LH), and estradiol employed to perform all procedures, using a 5-Fr catheter
were measured on the third day of the cycle in all patients. (RUC, COOK; William Cook Europe, Bjeeverskov, Den-
Additionally, pelvic magnetic resonance imaging (MRI) mark) and always with the aid of a coaxially introduced
was performed (i) when clinical examination or US was microcatheter (Embocath; BioSphere Medical Inc., Rock-
suspicious for adenomyosis or uterine sarcoma, (ii) in land, MA, USA). Trisacryl gelatin microspheres
virgins where vaginal US was impossible, and (iii) starting (Embospheres; BioSphere Medical, S.A., Roissy, France)
in April 2004, in all patients before UAE. Other possible were used for embolization in all cases. At the start of the
causes of infertility were systematically investigated in study (first five patients), we chose particles 300 to 900 lm
couples with primary or secondary sterility or a history of in diameter. Later, in accordance with data published in the
consecutive abortions. literature, particles larger than 500 lm were used exclu-
All patients completed a questionnaire relating to sively, to prevent possible nontargeted ovarian
myoma-related symptoms before the start of the thera- embolization via utero-ovarian anastomoses [8].
peutic procedure. Each patient had to define the intensity of A single dose of antibiotics (sultamicillin; 1.5 g intra-
symptoms on a scale from 1 to 10, where 1 = absence of venously) was administered to every patient 30 min before

123
M. Mara et al.: Embolization Versus Myomectomy 75

the embolization. For pain management during the first 24 (iii) unexpected penetration of the uterine cavity (in women
h after the procedure patients were given either epidural in whom fibroid prominence in the cavity was not previ-
analgesia (10 ml of 0.5% bupivacain plus 5 lg of sufen- ously detected by hysteroscopy), (iv) unplanned conversion
tanyl in 50 ml of normal saline, administered continually at from LM to laparotomy (during myoma enucleation), and
5 to 10 ml per hour) or intravenous analgesia (5 lg of (v) early reoperation because of uterine bleeding, hemo-
sufentanyl plus 0.15 mg of clonidin in 50 ml of normal peritoneum, or hematoma. The same two surgeons
saline as a continual infusion at 5 to 10 ml per hour). performed all myomectomies.
Nausea and postembolization discomfort during sub-
sequent days were treated by thiethylperazin, diclofenac,
and paracetamol. The minimal length of hospitalization Follow-up
after the procedure was 48 h.
Embolization leading to bilateral occlusion of ascending The occurrence of the following early postoperative com-
branches of urinary arteries and a complete loss of fibroid plications were monitored during the first 30 days: fever,
perfusion, as detected by angiography, was considered signs of pelvic infection, severe vaginal bleeding, severe
technically successful. Dissection or spasm of uterine pain not responsive to analgesics, prolonged hospital stay
arteries, adverse reaction to administered drugs, hematoma ([48 h after UAE, 72 h after LM, and [144 h after OM),
in the groin, and other complications of angiography were the necessity for antibiotics or blood transfusion, rehospi-
considered periprocedural complications. All procedures talization, allergic reactions, wound complications after
were performed by the same interventional radiologist. myomectomy, ischemic phenomena after UAE, surgical
intervention due to hematoma (pelvic, subfascial, retro-
peritoneal,or inguinal) or infection, thromboembolic
Myomectomy complications, and hysterectomy.
Patients were examined (clinically and using ultraso-
The myomectomy procedure was always initiated with nography) 1 month and 6 months postprocedurally and
hysteroscopy: a finding of a submucous fibroid of type 0 or subsequently every 6 months. The levels of FSH, LH, and
type I (according to the classification of European Society estradiol (on the third day of the cycle or at another time in
for Hysteroscopy) would eliminate the relevant patient the case of amenorrhea) were measured and myoma-related
from the study. Hysteroscopy was followed by laparoscopy symptoms were again evaluated (the same questionnaire) 6
and the access for myomectomy was chosen according to months after the procedure. The FSH level was monitored
predefined criteria. Open myomectomy (OM) was pre- in the subsequent course of follow-up in women with signs
ferred when a fibroid was larger than 8 cm, in the case of a of ovarian failure. Patients were examined immediately in
finding of multiple intramural fibroids, and in the case of a the case of difficulties, complications, or signs of
very unfavorable localization of a fibroid (e.g., in uterine pregnancy.
edges reaching the pelvic wall or deep in the posterior The following late complications were assessed more
uterine wall reaching the insertions of sacro-uterine liga- than 30 days after the procedure: (i) signs of uterine
ments). In all other cases myomectomy was performed by infection or sepsis; (ii) permanent or transient signs of
laparoscopy. The suture of the uterine wall defect required ovarian failure (clinical, i.e., amenorrhea not related to
after myoma enucleation was performed using atraumatic pregnancy, with or without vasomotor symptoms of men-
stitches in two layers (vicryl 2/0, polyglactin 910; Ethicon, opause, requiring hormone replacement therapy [HRT]; or
Brussels, Belgium). Myomectomy was also covered by a laboratory, in the case of FSH increase by [5 IU/L, com-
single dose of antibiotics (sultamicillin, 1.5 g i.v. 30 min pared to pretreatment values); (iii) ischemic phenomena
before the procedure), by the corresponding protocol of after UAE; (iv) chronic pelvic pain or dyspareunia; (v)
continuous intravenous analgesia (sufentanyl plus clonidin sudden severe uterine bleeding; (vi) chronic malodorous
for the first 24 h after the procedure), and by the same vaginal discharge; (vii) loss of libido; (viii) emergency
symptomatic therapy (antiemetics, antipyretics, analgesics) myomectomy or hysterectomy; (ix) vaginal fibroid expul-
such as UAE. The minimal length of hospital stay was 48 h sion; and (x) uterine rupture.
after LM and 120 h (5 days) after OM. Great emphasis was placed on ultrasonographical scan
Myomectomy was evaluated as successful when all of the uterus 6 months after the procedure. The following
detectable fibroids larger than 4 cm were completely outcomes were evaluated as favorable, with regard to
removed. The following were considered to be periopera- planned gravidity: (i) absence of a fibroid larger than 5 cm
tive complications: (i) injury of organs in the abdominal or a fibroid deforming the uterine cavity and (ii) absence of
cavity (fallopian tube, ovaries, urinary bladder, intestines) a hematoma or thinning of the myometrium in place of
or major pelvic vessels, (ii) blood loss exceeding 1000 ml, previous myomectomy. MRI was performed 6 months after

123
76 M. Mara et al.: Embolization Versus Myomectomy

embolization in patients who underwent MRI before UAE based on the elimination criteria; the remaining 121 women
(all patients starting from April 2004) and also in patients were randomized into one of the two groups (58 for UAE
who showed no sign of a fibroid decrease by US. The main and 63 for myomectomy) and they underwent appropriate
benefit of MRI in these patients was confirmative deter- therapy (Table 1). Of this count, 120 patients have finished
mination of fibroid reperfusion or its insufficient infarction. a 6-month follow-up and 3 patients have dropped out of the
trial: 1 patient did not turn up for the checkup at 6 months
after myomectomy and 2 patients did not turn up at 12
Reproductive Follow-up and Reinterventions months after UAE.
Of the total number of 121 patients, 110 patients were
All patients with reproductive plans were recommended to symptomatic (90.9%). The mean age of women was 32.4
wait for at least 6 months after the therapy. Women with a and 32.0 years in groups E and M, respectively. Sixty-six
history of infertility who had a favorable outcome of a US patients were nulligravidae (54.5%), 35 were sterile
uterus scan 6 months after the procedure were referred to a (28.9%; 11 in group E and 24 in group M; p \ 0.05), 18
center of assisted reproduction to undergo causal infertility had miscarried in the past (14.9%), and 51 had another
treatment (according to associated factors), including in subfertility factor other than myoma (42.1%). Mean FSH
vitro fertilization (IVF). History of reproductive attempts levels before the procedure were 6.98 2.9 IU/L in group
and results of all patients were systematically recorded E and 6.73 1.9 IU/L in group M. Six patients had a FSH
during regular checkups (every 6 months). In the case of value [10 IU/L before embolization, and four women
pregnancy, prenatal monitoring and delivery at our hospital before myomectomy. Except for the rate of sterility, there
were offered to all patients. The type of delivery and were no statistically significant differences in any afore-
possible indications for operative delivery were subjected mentioned factors or in other entry parameters (size of
to standard rules: cesarean section was not primarily indi- dominant fibroid, number of fibroids) between the two
cated (only if other indications were also present) except groups. A detailed summary of US scan findings prior to
for cases where persisting fibroid formed an obstetric both procedures is given in Table 2.
obstruction and in patients who underwent intrauterine Periprocedural results are shown in Table 3. The rate of
penetration during myomectomy. technical failures was about 10% in both groups. Six
Secondary myomectomy was recommended in the case patients were embolized unilaterally (four due to atypical
of undetectable fibroid shrinkage at 6 months after UAE branching or spasms of uterine arteries not responding to
and/or in the case of a persisting fibroid [5 cm. Similar vasodilators and two because of large utero-ovarian anas-
reintervention was recommended anytime later (except tomoses). In five women myomectomy was incomplete due
during pregnancy) when fibroid regrowth after UAE (over to unfavorable localization, resulting in the retention of a
5 cm) or recurrent fibroid [5 cm after myomectomy was fibroid larger than 4 cm. Forty-two (67%) myomectomies
detected. The intervals between checkups after surgical of 63 were performed by LM. The frequency of laparo-
reintervention were also 6 months, and patients were rec- conversions, i.e., myomectomies started by LM which had
ommended to delay their pregnancy plans for at least to be completed as an open surgery because of complica-
another 6 months. tions (i.e., bleeding and/or difficult tumor enucleation from
the uterus), was 4.5%. The following complications have
occurred: in group E, one case of artery dissection and
Analysis of the Results three cases of uterine artery spasms; and in group M, three
myomectomies with unexpected intrauterine penetration
The results were analyzed on an intent-to-treat basis. For and two nonelective laparoconversions.
statistical comparison of qualitative parameters from both There were no significant differences in most monitored
groups (e.g., rehospitalization: yes or no), chi-square test parameters of early postprocedural results between the two
and Fischers test were used. For comparison of quantita- groups (Table 4). The average length of hospitalization and
tive parameters (e.g., FSH level), Students t-test and postoperative recovery were significantly longer and the
Mann-Whitney test were used. p \ 0.05 was determined to mean serum C-reactive protein concentration was signifi-
be statistically significant. cantly higher in the group of women treated by surgery.
There were no significant differences in the frequency of
early complications between the two groups. All these
Results complications can be considered as mild or moderately
serious. It was the febrile status in most cases which
Twenty-one patients (15.4%) of 149 refused to participate required therapy with antibiotics (eight patients in group E,
in the trial and 7 patients were excluded from the study five patients in group M). After UAE 1 patient required a

123
M. Mara et al.: Embolization Versus Myomectomy 77

Table 1 Flowchart of 142 patients (pt) eligible (according to inclusion criteria) to enter the trial
Pt no. Pt initials Year of birth Date of randomization Randomized into group Date of procedure Type of procedure

1 Z.I. 1969 2.11.2001 E 20.11.2001 UAE


2 M.H. 1967 2.11.2001 E 22.11.2001 UAE
K.S. 1965 Refused
3 L.N. 1973 7.12.2001 M 9.1.2002 OM
4 D.D. 1965 14.12.2001 E 24.1.2002 UAE
5 D.K. 1964 11.1.2002 M 21.2.2002 OM
J.B. 1973 Refused
6 M.M. 1976 8.2.2002 E 28.3.2002 UAE
B.C. 1970 Refused
7 M.L. 1974 15.3.2002 M 11.4.2002 OM
8 D.L. 1973 22.3.2002 M 15.4.2002 OM
9 V.H. 1976 22.3.2002 E 3.5.2002 UAE
10 K.S. 1968 12.4.2002 M 22.5.2002 LM
P.K. 1964 Refused
M.P. 1976 Refused
11 M.M. 1972 31.5.2002 M 4.7.2002 OM
L.H. 1977 Refused
12 L.B. 1971 21.6.2002 M 7.8.2002 OM
13 D.M. 1975 6.9.2002 E 11.10.2002 UAE
14 E.M. 1976 13.9.2002 E 15.10.2002 UAE
15 M.K. 1973 27.9.2002 M 27.11.2002 OM
16 J.C. 1974 4.10.2002 E 14.11.2002 UAE
17 L.S. 1975 18.10.2002 M 27.11.2002 LM
18 A.H. 1970 25.10.2002 M 16.12.2002 OM
19 I.K. 1970 25.10.2002 E 21.11.2002 UAE
H.S. 1966 Refused
20 M.K. 1975 13.12.2002 M 22.1.2003 OM
21 M.C. 1966 20.12.2002 E 9.1.2003 UAE
22 D.H. 1976 20.12.2002 M 28.1.2003 LM
23 D.V. 1967 10.1.2003 E 30.1.2003 UAE
24 J.B. 1971 24.1.2003 E 6.3.2003 UAE
S.B. 1968 Refused
25 V.H. 1970 7.2.2003 E 20.3.2003 UAE
26 M.O. 1967 21.2.2003 M 2.4.2003 OM
27 I.T. 1968 28.2.2003 M 14.5.2003 OM
28 H.A. 1964 28.3.2003 E 30.4.2003 UAE
29 L.V. 1969 4.4.2003 E 7.5.2003 UAE
30 J.M. 1963 25.4.2003 M 3.6.2003 OM
31 M.K. 1974 25.4.2003 E 19.5.2003 UAE
32 L.T. 1968 2.5.2003 M 17.6.2003 OM
33 S.B. 1971 2.5.2003 E 29.5.2003 UAE
34 E.T. 1968 23.5.2003 E 3.7.2003 UAE
35 B.P. 1973 13.6.2003 M 9.9.2003 OM
36 L.K. 1974 13.6.2003 E 3.7.2003 UAE
I.S. 1964 Refused
37 M.S. 1966 10.10.2003 M 26.11.2003 LM
38 M.H. 1964 24.10.2003 E 4.12.2003 UAE
A.H. 1965 Refused

123
78 M. Mara et al.: Embolization Versus Myomectomy

Table 1 continued
Pt no. Pt initials Year of birth Date of randomization Randomized into group Date of procedure Type of procedure

39 K.K. 1975 19.12.2003 M 10.2.2004 LM


40 L.J. 1967 19.12.2003 E 9.1.2004 UAE
41 I.F. 1971 9.1.2004 E 23.1.2004 UAE
42 D.L. 1977 16.1.2004 M 17.2.2004 LM
43 J.V. 1977 16.1.2004 E 17.2.2004 UAE
44 E.S. 1972 30.1.2004 M 2.3.2004 LM
45 R.K. 1966 6.2.2004 E 5.3.2004 UAE
46 J.H. 1972 13.2.2004 E 12.3.2004 UAE
47 M.M. 1967 13.2.2004 M 23.3.2004 LM
48 M.Z. 1970 20.2.2004 E 12.3.2004 UAE
49 E.S. 1970 5.3.2004 M 20.4.2004 LM
50 M.M. 1965 12.3.2004 M 20.4.2004 OM
51 K.K. 1968 12.3.2004 E 2.4.2004 UAE
52 D.M. 1968 19.3.2004 M 21.4.2004 OM
53 N.V. 1969 19.3.2004 M 21.4.2004 LM
54 M.T. 1970 26.3.2004 M 8.6.2004 LM
55 I.S. 1966 26.3.2004 E 29.4.2004 UAE
E.Z. 1971 Refused
56 M.P. 1974 23.4.2004 M 7.7.2004 LM
57 E.K. 1972 23.4.2004 E 14.5.2004 UAE
58 P.S. 1967 7.5.2004 M 19.7.2004 OM
59 L.D. 1974 14.5.2004 M 21.7.2004 LM
60 J.S. 1968 14.5.2004 E 11.6.2004 UAE
61 H.K. 1965 21.5.2004 M 7.9.2004 OM
62 M.M. 1972 28.5.2004 M 12.10.2004 LM
63 A.M. 1970 28.5.2004 M 9.11.2004 LM
64 V.H. 1978 31.5.2004 E 11.6.2004 UAE
65 K.W. 1970 4.6.2004 E 9.7.2004 UAE
L.S. 1975 Refused
66 K.H. 1973 18.6.2004 E 20.8.2004 UAE
67 E.B. 1983 25.6.2004 E 20.8.2004 UAE
68 K.F. 1968 2.7.2004 E 9.9.2004 UAE
69 M.R. 1973 9.9.2004 M 16.11.2004 LM
70 M.H. 1967 9.9.2004 E 1.10.2004 UAE
J.P. 1976 Refused
D.R. 1965 Refused
71 K.F. 1976 23.9.2004 M 29.11.2004 LM
72 P.C. 1975 23.9.2004 E 22.10.2004 UAE
73 P.P. 1972 30.9.2004 E 22.10.2004 UAE
74 B.S. 1972 7.10.2004 M 19.1.2005 LM
75 P.H. 1969 14.10.2004 E 3.11.2004 UAE
76 J.L. 1979 14.10.2004 M 26.1.2005 LM
77 J.E. 1968 21.10.2004 M 27.1.2005 LM
78 L.A. 1969 21.10.2004 E 19.11.2004 UAE
79 M.D. 1975 4.11.2004 M 2.2.2005 LM
80 M.R. 1972 11.11.2004 E 10.12.2004 UAE
81 M.Z. 1966 18.11.2004 M 17.2.2005 LM
82 S.Z. 1971 9.12.2004 M 17.2.2005 LM

123
M. Mara et al.: Embolization Versus Myomectomy 79

Table 1 continued
Pt no. Pt initials Year of birth Date of randomization Randomized into group Date of procedure Type of procedure

83 H.S. 1968 16.12.2004 E 21.1.2005 UAE


84 M.S. 1974 6.1.2005 E 18.2.2005 UAE
E.K. 1965 Refused
S.D. 1970 Refused
85 L.L. 1972 27.1.2005 E 18.3.2005 UAE
86 L.M. 1972 3.2.2005 M 21.3.2005 LM
87 I.C. 1970 17.2.2005 M 23.3.2005 LM
88 P.Z. 1971 24.2.2005 E 8.4.2005 UAE
89 M.K. 1969 24.2.2005 M 11.4.2005 LM
M.T. 1975 Refused
90 H.K. 1967 3.3.2005 M 26.4.2005 OM
91 B.Z. 1974 10.3.2005 E 6.5.2005 UAE
92 M.K. 1965 24.3.2005 M 4.5.2005 LM
93 A.G. 1977 31.3.2005 M 16.5.2005 LM
94 L.P. 1972 7.4.2005 M 17.5.2005 LM
95 L.H. 1977 7.4.2005 M 18.5.2005 OM
96 J.K. 1976 14.4.2005 E 13.5.2005 UAE
97 M.L. 1976 21.4.2005 M 23.5.2005 LM
H.V. 1965 Refused
98 M.K. 1974 12.5.2005 M 6.6.2005 LM
99 P.L. 1978 26.5.2005 E 1.8.2005 UAE
100 E.P. 1971 23.6.2005 E 1.8.2005 UAE
101 K.K. 1977 30.6.2005 M 7.9.2005 LM
102 L.F. 1969 14.7.2005 E 21.9.2005 UAE
E.S. 1966 Refused
103 J.R. 1980 8.9.2005 M 26.9.2005 LM
104 E.H. 1972 8.9.2005 E 23.9.2005 UAE
105 O.S. 1967 15.9.2005 M 5.10.2005 LM
106 I.K. 1970 15.9.2005 M 12.10.2005 LM
H.A. 1966 Refused
107 P.P. 1966 22.9.2005 E 14.10.2005 UAE
108 M.A. 1977 29.9.2005 M 26.10.2005 LM
109 E.Z. 1973 29.9.2005 M 2.11.2005 LM
110 I.S. 1979 29.9.2005 M 2.11.2005 LM
111 D.M. 1967 6.10.2005 E 21.10.2005 UAE
112 J.B. 1967 6.10.2005 M 2.11.2005 LM
113 A.G. 1983 13.10.2005 M 9.11.2005 OM
114 J.P. 1973 13.10.2005 E 15.11.2005 UAE
115 S.S. 1969 20.10.2005 M 16.11.2005 LM
116 H.D. 1974 27.10.2005 M 1.12.2005 LM
117 L.B. 1973 27.10.2005 E 2.12.2005 UAE
D.V. 1976 Refused
118 K.K. 1970 10.11.2005 E 2.12.2005 UAE
119 M.J. 1972 10.11.2005 M 1.12.2005 LM
120 R.L. 1977 10.11.2005 E 9.12.2005 UAE
121 D.R. 1975 24.11.2005 E 9.12.2005 UAE
Note. LM, laparoscopic myomectomy; OM, open myomectomy; UAE, uterine artery embolization. Dates given as day.month.year

123
80 M. Mara et al.: Embolization Versus Myomectomy

Table 2 Ultrasound findings for 121 patients prior to uterine artery embolization (UAE) or myomectomy
UAE (n = 58) Myomectomy (n = 63) p

Average size of dominant fibroid, mm SD (range) 62.3 19.1 (42107) 59.8 16.5 (41110) NSa
No. women with myoma [80 mm 8 (13.8%) 7 (11.1%) NSb
No. myomas [2 cm SD (range) 1.95 1.8 (18) 1.84 2.1 (116) NSa
No. women with Solitary myomas 39 (67.2%) 40 (63.5%) NSb
25 fibroids 15 (25.9%) 21 (33.3%) NSb
[5 fibroids 4 (6.9%) 2 (3.2%) NSc
Uterine cavity deviated by fibroid(s) 30 (51.7%) 22 (34.9%) NSb
a b c
Note. NS, nonspecific difference; SD, standard deviation. Tested by: Mann-Whitney test; chi-square test; Fishers test

Table 3 Periprocedural evaluation


UAE (n = 58) Myomectomy (n = 63) P

Technical success 52 (89.7%) 58 (92.1%) NSb


Average time of procedure, min SD (range) 59.2 23.1 (30140) 108.9 27.7 (54173) \0.0001a
Fluoroscopy time, min SD (range) 16.8 6.3 (533)
No. women with utero-ovarian anastomoses 25 (43.1%)
Average perioperative blood loss, ml SD (range) 296.0 122.6 (100700)
Type of analgesia 38 EDA/20 IVA 63 IVA
Complications 4 (6.9%) 5 (7.9%) NSc
Note. UAE, uterine artery embolization; NS, nonspecific difference; SD, standard deviation; EDA, epidural analgesia; IVA, intravenous
analgesia. Tested by: a Mann-Whitney test; b chi-square test; c Fishers test

Table 4 Early postprocedural results (from day 1 to day 30 after procedure)


UAE (n = 58) Myomectomy (n = 63) P

Average hospital stay, h SD (range) 60.2 32.3 (36216) 86.1 40.4 (48192) \0.0001a
*
Prolonged hospital stay 9 (15.5%) 6 (9.5%) NSb
Hospital stay [7 days 1 (1.7%) 3 (4.8%) NSc
Readmissions to hospital 2 (3.4%) 1 (1.6%) NSc
Febrile morbidity 7 (12.1%) 7 (11.1%) NSb
Day 2 serum CRP, mg/L SD (range) 23.8 27.3 (3168) 36.9 28.1 (6137) \0.0001a
Need for antibiotics 8 (13.8%) 6 (9.5%) NSb
Necessity for transfusion 0 2 (3.2%) NSc
Recovery period, days SD (range) 11.9 5.9 (330) 22.1 12.3 (765) \0.0001a
Disability [2 wk 13 (22.4%) 36 (57.1%) \0.0001b
Complications 12 (20.7%) 10 (15.9%) NSb
a
Note. UAE, uterine artery embolization; NS, nonspecific difference; SD, standard deviation; CRP, C-reactive protein. Tested by: Mann-
Whitney test; b chi-square test; c Fishers test
*Longer hospital stay than usual (i.e., more than 48 h after UAE, 72 h after laparoscopic myomectomy, and 144 h after open myomectomy)

prolonged hospital stay due to severe vaginal bleeding right groin and was managed by conservative therapy.
(treated by pharmacotherapy), one patient developed a rash After myomectomy two patients (one after LM and one
as a probable reaction to analgesics, and one woman was after OM) required transfusion due to severe anemia, one
treated for postpuncture headache after epidural anesthesia. woman developed a urinary tract infection, and one woman
Additionally, a subcutaneous hematoma of approximately a wound infection. One patient underwent surgical evacu-
5 cm developed in one woman at the site of puncture in the ation of subfascial hematoma 1 day after OM.

123
M. Mara et al.: Embolization Versus Myomectomy 81

Table 5 Late postprocedural results


UAE (n = 58) Myomectomy (n = 62) P

Mean follow-up, mo SD (range) 26.2 14.2 (655) 23.7 13.9 (654) NSa
Relief from symptoms* 88.5% (46/52) 87.9% (51/58) NSb
6 mo after procedure Serum FSH, IU/L SD (range) 7.89 6.0 (4.048.9) 6.49 2.0 (3.612.3) NSa
FSH [10 IU/L 8 (13.8%) 2 (3.2%) \0.05c

Significant elevation of FSH 3 (5.2%) 0 NSc
Regrowth or recurrence of fibroid(s) 6 (10.3%) 5 (8.1%) NSb
Reinterventions 19 (32.8%) 2 (3.2%) \0.0001b
Reinterventions for regrowth or recurrence of fibroid(s) 4 (6.9%) 2 (3.2%) NSc
Complications 8 (13.8%) 5 (8.1%) NSb
Note. UAE, uterine artery embolization; NS, nonspecific difference; SD, standard deviation. Tested by: a Mann-Whitney test; b chi-square test; c
Fishers test
*Lower sum of six followed symptoms at 6 months after procedure than preprocedurally (based on the questionnaire)

At least 5 IU/L increase at 6 months after procedure (compared with preprocedural level)

Tables 5 and 6 summarize late clinical results and Table 6 Results of imaging studies 6 months after uterine artery
results of imaging studies at 6 months after the procedure. embolization (UAE)
The mean length of follow-up was approximately 2 years Mean shrinkage of diameter of 31.7%
in both groups; 81% of women after embolization and 79% dominant fibroid on US (58 patients)
after surgery have been followed for more than 1 year, and Mean shrinkage of volume of 58.7%
50% of patients in group E and 43% in group M have been dominant fibroid on MRI*
followed for at least 2 years. Both methods were similarly Mean volume of dominant 69 cm3
fibroid on MRI* (pre-UAE: 166 cm3)
effective when cumulative evaluation of all symptoms at 6
Partial infarction of dominant 28.9% (11 / 38)
weeks after the procedure was considered. While it seemed fibroid on MRI*
at preliminary evaluation of results that representation of No infarction of dominant 2.6% (1 / 38)
patients with total relief from all myoma-related symptoms fibroid on MRI*
would be significantly higher in group M, it now appears
*Calculated from only 38 patients with both (pre- and post-UAE)
that their number is only insignificantly higher (32 of 58 MRI scans
women, i.e., 55 %) in this group than after embolization
(26 of 52 women, i.e., 50%). serious or life-threatening complications have occurred
With respect to the main goal of the therapy (gravidity) after UAE either, but in four women there were compli-
and trial design, both groups differed significantly in the cations related to ovarian function. One patient with 6
frequency of reinterventions. Secondary myomectomy was weeks of amenorrhea had no further menopausal symp-
performed in 19 cases in group E: 9 by LM, 8 by lapa- toms, had a normal FSH level (6.4 IU/L) at 6 months after
rotomy, and 2 by hysteroscopy. The mean interval from UAE, and is pregnant now. A transitory, but significant
UAE to reintervention was 12.4 months (12.2 months; FSH elevation, to 15.0, 30.4, and 48.9 IU/L, respectively,
median, 7.0 months). The persistence of a large fibroid ([5 occurred in three women with preprocedural FSH \10 IU/
cm) at 6 months after UAE was the most frequent indica- L (Table 5). One of these women suffered amenorrhea for
tion (15 cases, including 5 patients with UAE technical 2 months and did not respond to progesterone, but her cycle
failure), and in 4 cases we reintervened due to regrowth of adjusted spontaneously. This woman has not tried to con-
a fibroid (see Table 5). In group M, the reintervention was ceive to date. Another two patients with FSH elevation
indicated only in cases of clinically significant recurrence. were clinically asymptomatic (from the time of emboliza-
Remyomectomy was performed in two women (one by LM tion until now, which is 53 and 35 months, respectively),
and one by open surgery) at a 15- and 30-month interval but they both underwent unsuccessful therapy for sterility,
from primary therapy, respectively. including two and three IVF cycles, respectively, to date.
We recorded a relatively low frequency of late com- Low response to ovarian stimulation with gonadotropins
plications. All these incidents (dyspareunia, pelvic pain, (high consumption of FSH, low number of reacting folli-
endometritis, and one episode of metrorrhagia) after cles) and poor development of most embryos were
myomectomy were regarded as mild and nonserious. No common for their cycles.

123
82 M. Mara et al.: Embolization Versus Myomectomy

35 was 50% to date, delivery rate 19%, and abortion rate 64%,
32
while after myomectomy the pregnancy rate was 78%,
30
delivery rate 48%, and abortion rate 23%. The differences
25 in all these parameters were statistically significant
20 19 (p \ 0.05, v2 test). Relative risk (RR) of women treated
17
with UAE not to get pregnant was 2.22 (95% confidence
15
interval, 1.11\RR\4.44); not to deliver, 1.54 (1.08\RR
10 9 \ 2.18); and to abort, 2.79 (1.25 \ RR \ 6.22).
5 6 5
5
1 1 1 1 1
0
pregnancy delivery abortion ectopic termination pregnant Discussion
now
UAE Myomectomy
Comparison of myomectomy and UAE, two very different
therapeutic approaches, is difficult and could appear mis-
Fig. 1 Reproductive results of 26 women after uterine artery
embolization (UAE) and 40 women after myomectomy. Statistical
leading, particularly in some parameters (invasiveness,
difference between the groups (p value): pregnancy, NSb; delivery, complication rate, reinterventions). Nevertheless, the
\0.05b; abortion, \0.05b; ectopic gestation, NSc; pregnancy termi- question of possible use of uterine fibroid embolization in
nation, NSc; pregnant now, NSc. Tested by: bchi-square test; cFishers young women with active reproductive plans is very
test. NS, nonsignificant
important from the point of view of gynecology and
reproductive medicine. The surgical therapy is, especially
Overall there were eight patients with a FSH concen- in some patients, technically difficult, invasive, and risky,
tration [10 IU/L at 6 months after UAE (including four and therefore comparison of UAE with myomectomy as a
women with a pre-UAE value [10 IU/L and three with an possible alternative to existing standard therapy is most
increase of [5 IU/L) and two patients with FSH [ 10 IU/L desirable [9].
at 6 months after myomectomy (including one with pre- There are many reports about gravidities and repro-
procedural FSH [ 10 IU/L and no case of significant ductive results after embolization in the literature today
increase). No patients had climacteric vasomotor symp- [1015]. Surprisingly, there is an apparent nearly absolute
toms and no patient required HRT. Two patients lack of prospective studies comparing not only myomec-
complained of temporarily decreased sexual appetence tomy and UAE, but also different fibroid treatments in
after embolization (this improved within 12 months): 1 of relation to fertility (e.g., myomectomy with expectation).
them in connection with noticeable hypomenorrhea and the Only four papers [1619] compare clinical results of
other patient in connection with the feeling of vaginal embolization and myomectomy. But they are not ran-
dryness. No case of uterine rupture, urgent hysterectomy domized studies, only one is prospective [19], and only one
or other emergency surgery, or hospitalization for late is aimed at obstetrical, not just reproductive results [17].
complications after embolization or myomectomy has In this study we confirmed in a midterm time horizon
occurred in the whole cohort to date. and in a larger cohort of patients most of the results from
Existing reproductive and first perinatal results are our preliminary evaluation [6]. Myomectomy and emboli-
summarized in Fig. 1 and Table 7. From the limited cohort zation were comparable as far as technical success rate,
of patients who have already tried to conceive (26 after frequency of early and late complications, and symptom-
UAE and 40 after myomectomy), 13 from group E (mean atic effectiveness are concerned. We also verified lower
age, 32.8 years; range, 22 to 40 years) and 31 from group invasiveness of the radiological approach compared to
M (mean age, 34.3 years; range, 2742 years) have already myomectomy (hospital stay, recovery period, acute phase
become pregnant. Four women after embolization and one markers). The rate of serious complications was very low
patient after myomectomy have already become pregnant in both groups. The trial does not give the answer to the
two times; one woman after myomectomy gave birth to management of large or recurrent fibroids (see exclusion
twins (after IVF). The mean interval between the procedure criteria) but our goal was to keep the study group as
and gravidity was 18 and 13 months in groups E and M, homogeneous as possible, which is always difficult in
respectively. This difference could be partially influenced uterine fibroid patients. Unlike other studies, we did not
by the rate of secondary myomectomy in the two groups of focus on economical comparisons between the two treat-
pregnant women: it was performed in 5 of 13 pregnant ment methods [20], mainly because the cost of UAE is
women after UAE but in 1 only patient of 31 pregnant many times higher than that of myomectomy at the site of
patients in group M. Presenting the results in the language this study, and there are significant differences between
of reproductive medicine, the pregnancy rate after UAE open and laparoscopic myomectomy.

123
M. Mara et al.: Embolization Versus Myomectomy 83

Table 7 Perinatal results


UAE (n = 5) Myomectomy (n = 19) p

Mean age of mothers, yr SD (range) 32.2 4.1 (2939) 34.4 4.4 (2742) NSa
Mean birth weight, g SD (range) 3042 316 (28303600) 3035 538 (16003940) NSa
Mean completed gestational weeks SD (range) 38.8 0.6 (3840) 37.5 2.4 (3140) NSa
Preterm delivery (\37th week) 0 5 (26.3%) NSb
Cesarean section 3 (60.0%) 13 (68.4%) NSb
Postpartum hemorrhage 1 (20.0%) 0 NSb
Perinatal hypoxia of neonate 0 0
Pre-eclampsia 0 2 (10.5%) NSb
Fetal intrauterine growth restriction 0 2 (10.5%) NSb
a b
Note. UAE, uterine artery embolization; NS, nonspecific difference; SD, standard deviation. Tested by: Mann-Whitney test; Fishers test

The relatively high rate of LMs (two-thirds) might seem and Doppler US evidence of infarcted fibroid, this number
surprising and it certainly does not represent a typical would have been expected to be significantly lower.
picture of a broad gynecological practice. But we believe Existing reproductive results could be partially influ-
that these patients should always be referred to centers of enced by the short duration of the follow-up and mainly by
reproductive surgery with experienced endoscopic sur- the unequal number of patients who tried to conceive in each
geons. With respect to certain limits for LM (mentioned group: 40 after myomectomy and only 26 after embolization
under Materials and Methods), many studies have already (p \ 0.05, v2 test). Nevertheless, the statistically significant
proved the equal safety and potency of the laparoscopic differences observed in the number of successful deliveries
approach in comparison to the open procedure [21, 22]. (19 after myoma enucleation and only 5 after UAE) and in
The technical failure rate of UAE was higher than the the number of early pregnancy losses (6 after myomectomy
rates reported in other studies [23, 24], mainly as a result of and 9 after embolization; in all cases spontaneous or missed
the unusual or unfavorable anatomy (anastomoses to miscarriage in the first trimester) were in support of the
ovarian artery) present. Another reason for the higher surgical approach. We can only speculate whether these
failure rate could be the nature of our patient group, reproductive results are due to an error of small numbers,
comprised of young women desiring pregnancy. This whether they reflect the influence of UAE on ovarian
might account for a particularly careful approach to their function, uterine perfusion, and implantation quality, or
treatment. Only the results of the EMMY trial showed a whether they reflect a direct influence of embolization on the
similar frequency of technical failures [25], but in their uterine cavity and endometrium [2628].
case, in addition to difficult anatomy, the absence of one of The fact that the rate of abortions after UAE was higher
the uterine arteries was the most common reason. than 60% (in contrast with 23% after myomectomy) is the
In accordance with our expectations, there were far most alarming result of the study to date, in contrast to
more reinterventions in the group treated by embolization existing reports from other authors [10, 12, 17]. The post-
(Table 5). But these numbers do not reflect the frequency UAE abortion rate was 16.7% in the Ontario multicenter,
of method failure or the rate of recurrence. Instead, they prospective trial (24 pregnancies in 21 women of mean age
seem to be the logical consequences of the different char- 34 years), 27% in the retrospective trial of Carpenter and
acter of the two procedures (UAE leaves fibroids in situ), of Walker (26 pregnancies; mean age of patients, 37 years),
the main goal of the therapy (to optimize the uterine con- and 24% after UAE and 15% after laparoscopic myomec-
dition before planned conception), and of methods adapted tomy in the controlled retrospective multicenter trial of
to it (specific strategy for indications of reinterventions). Goldberg et al. (53 pregnancies; mean age, 38 years), but
The fact that in the initial study postprocedural MRI was the difference was not statistically significant. In our cohort,
performed only in patients where significant shrinkage of the mean age of pregnant women after UAE was lower
fibroid had not occurred (including 6 patients with unilateral (32.8 years) than that of pregnant women after myomec-
embolization only) could account for the relatively high tomy (34.3 years), and at the same time, it was lower than in
number of cases with at least partially maintained fibroid the aforementioned trials. The mean age of women who
perfusion at 6 months after UAE (12 patients of 38 evalu- aborted after embolization was only slightly higher (33.0
ated; see Table 6). If we had performed this examination in years), and that is why the age factor does not explain the
all women, including those with a good clinical response frequency of abortions after UAE. The abortion rate was

123
84 M. Mara et al.: Embolization Versus Myomectomy

expressed as a quotient of all abortions in all begun preg- 3. Stringer NH, Grant T, Park J, et al. (2000) Ovarian failure after
nancies, thus including two patients with terminations uterine artery embolization for treatment of myomas. J Am Assoc
Gynecol Laparosc 7:395400
in early gravidity (one termination due to extrauterine 4. Vashisht A, Studd JW, Carey AH, et al. (2000) Fibroid emboli-
gravidity and one termination in the eighth week of gesta- sation: a technique not without significant complications. Br J
tion). Even when excluding these two cases from the overall Obstet Gynaecol 107:11661170
abortion rate, this number would still be notably higher 5. Olive DL, Lindheim SR, Pritts EA (2004) Non-surgical man-
agement of leiomyoma: impact on fertility. Curr Opin Obstet
(53%). On the other hand, the pregnancy success rate after Gynecol 16:239243
UAE is significantly higher in our cohort (50%) than the 6. Mara M, Fucikova Z, Maskova J, et al. (2006) Uterine fibroid
rate reported by Carpenter and Walker, where only 26 embolization versus myomectomy in women wishing to preserve
(33%) of 79 women conceived after embolization. fertility: preliminary results of a randomized controlled trial. Eur
J Obstet Gynecol Reprod 126:226233
It is difficult to compare obstetrical results at the 7. Razavi MK, Wolanske KA, Hwang G, et al. (2002) Angiographic
moment because only a small number of patients after classification of ovarian artery to uterine artery anastomoses:
embolization have passed the first trimester of pregnancy. initial observations in uterine fibroid embolization. Radiology
The statistical analysis did not prove any significant dif- 224:707712
8. Pelage JP, Le Dref O, Beregi JP, et al. (2003) Limited uterine
ferences in any of the parameters studied between the two artery embolization with tris-acryl gelatin microspheres for
groups, however, Table 6 provides an interesting summary uterine fibroids. J Vasc Interv Radiol 14:1520
of existing perinatal results and complications. It is 9. Fauconnier A, Pelage JP, Lacombe P, et al. (2004) Embolization
remarkable to point out that, provided a pregnant woman of uterine fibroids and infertility: Is a clinical trial conceivable?
Gynecol Obstet Fertil 32:81824
after UAE successfully passed the first trimester, we sub- 10. Carpenter TT, Walker WJ (2005) Pregnancy following uterine
sequently did not record any of the serious pregnancy artery embolisation for symptomatic fibroids: a series of 26
complications (e.g., gestational hypertension, fetal growth completed pregnancies. Br J Obstet Gynaecol 112:321325
retardation, malpresentation, or prematurity) repeatedly 11. McLucas B, Goodwin S, Adler L, et al. (2001) Pregnancy fol-
lowing uterine fibroid embolization. Int J Gynaecol Obstet 74:17
described in other trials [10, 12, 17]. 12. Pron G, Mocarski E, Bennett J, et al. (2005) Pregnancy after
At the beginning of this trial we asked ourselves the uterine artery embolization for leiomyomata: the Ontario multi-
question whether the less invasive method of fibroid center trial. Obstet Gynecol 105:6776
embolization in women with reproductive plans is as 13. Ravina JH, Vigneron NC, Aymard A, et al. (2000) Pregnancy
after embolization of uterine myoma: report of 12 cases. Fertil
effective and safe as myomectomy. After more than 4 years Steril 73:12411243
of the trial duration it can be concluded that both methods 14. Walker WJ, Pelage JP (2002) Uterine artery embolization for
are comparable in terms of technical success rate, safety, symptomatic fibroids: clinical results in 400 women with imaging
and symptomatic efficacy. UAE is a less invasive follow up. Br J Obstet Gynecol 109:12621272
15. Kim MD, Kim NK, Kim HJ, et al. (2005) Pregnancy following
approach, but also, as it appears at this midterm following, uterine artery embolization with polyvinyl alcohol particles for
it is less definitive if the aim is to maximally eradicate patients with uterine fibroid or adenomyosis. CardioVasc Interv
fibroids before gravidity. For a definitive case comparison Radiol 28:611615
of reproductive and perinatal results we need to analyze 16. Broder MS, Goodwin S, Chen G, et al. (2002) Comparison of
long-term outcomes of myomectomy and uterine artery emboli-
more patients who try to conceive after the procedures and zation. Obstet Gynecol 100:864868
use a longer follow-up. However, the existing results 17. Goldberg J, Pereira L, Berghella V, et al. (2004) Pregnancy
clearly indicate that myomectomy is a method with a outcomes after treatment for fibromyomata: uterine artery
greater chance of success in women who plan to get embolization versus laparoscopic myomectomy. Am J Obstet
Gynecol 191:1821
pregnant early after the procedure. 18. Razavi MK, Hwang G, Jahed A, et al. (2003) Abdominal myo-
mectomy versus uterine fibroid embolization in the treatment of
Acknowledgments The study was supported by a grant from the symptomatic uterine leiomyomas. AJR 180:15711575
Internal Grant Agency (IGA) of the Ministry of Health of the Czech 19. Siskin GP, Shlansky-Goldberg RD, Goodwin SC, et al. (2006)
Republic (NR/8099-3). The authors wish to thank Mrs. Alena Doh- A prospective multicenter comparative study between myo-
nalova for statistical consultations and analysis of the results, Pavel mectomy and uterine artery embolization with polyvinyl alcohol
Dundr, M.D., for histopathological examinations of all excised myo- microspheres: long-term clinical outcomes in patients with
mas, and Petr Kriz, M.D., for pain management of all treated patients. symptomatic uterine fibroids. J Vasc Interv Radiol 17:1287
1295
20. Goldberg J, Bussard A, McNeil J, et al. (2007) Cost and reim-
bursement for three fibroid treatments: abdominal hysterectomy,
References abdominal myomectomy, and uterine fibroid embolization. Car-
dioVasc Interv Radiol 30:5458
1. Ravina JH, Herbreteau D, Ciraru-Vigneron N, et al. (1995) Arterial 21. Dubuisson JB, Fauconnier A, Fourchotte V, et al. (2001) Lapa-
embolization to treat uterine myomata. Lancet 346:671672 roscopic myomectomy: predicting the risk of conversion to an
2. Payne JF, Robboy SJ, Haney AF (2002) Embolic microspheres open procedure. Hum Reprod 16:17261731
within ovarian arterial vasculature after uterine artery emboliza- 22. Seracchioli R, Rossi S, Govoni F, et al. (2000) Fertility and
tion. Obstet Gynecol 100:883886 obstetric outcome after laparoscopic myomectomy of large

123
M. Mara et al.: Embolization Versus Myomectomy 85

myomata: a randomized comparison with abdominal myomec- fibroid tumors (EMMY trial): periprocedural results and com-
tomy. Hum Reprod 15:26632668 plications. J Vasc Interv Radiol 17:471480
23. Pron G, Bennett J, Common A, et al. (2003) Technical results and 26. Honda I, Sato T, Adachi H, et al. (2003) Uterine artery emboli-
effects of operator experience on uterine artery embolization for zation for leiomyoma: complications and effect on fertility.
fibroids: the Ontario Uterine Fibroid Embolization Trial. J Vasc Nippon Igaku Hoshasen Gakkai Zasshi 63:294302
Interv Radiol 14:545554 27. Tropeano G, Litwicka K, Di Stasi C, et al. (2003) Permanent
24. Kroncke TJ, Gauruder-Burmester A, Gronewold M, et al. (2004) amenorrhea associated with endometrial atrophy after uterine
Technical success rate, peri-interventional complications and artery embolization for symptomatic uterine fibroids. Fertil Steril
radiation exposure of the transarterial embolization for leio- 79:132135
myomas of the uterus. Rofo 176:580589 28. Tulandi T, Sammour A, Valenti D, et al. (2002) Ovarian reserve
25. Volkers NA, Hehenkamp WJ, Birnie E, et al. (2006) Uterine after uterine artery Embolization for leiomyomata. Fertil Steril
artery embolization in the treatment of symptomatic uterine 78:197198

123
Cardiovasc Intervent Radiol (2008) 31:8690
DOI 10.1007/s00270-007-9189-0

CLINICAL INVESTIGATION

Hepatic Arterial Infusion Chemotherapy through a Port-Catheter


System as Preoperative Initial Therapy in Patients with Advanced
Liver Dysfunction due to Synchronous and Unresectable Liver
Metastases from Colorectal Cancer
Toshihiro Iguchi Yasuaki Arai Yoshitaka Inaba Hidekazu Yamaura
Yozo Sato Masaya Miyazaki Hiroshi Shimamoto

Received: 5 June 2007 / Accepted: 11 September 2007 / Published online: 10 October 2007
Springer Science+Business Media, LLC 2007

Abstract remaining 2 patients died because extrahepatic metastases


Purpose We retrospectively evaluated the safety and developed and their performance status worsened; thus,
efficacy of preoperative initial hepatic arterial infusion surgery could not be performed. The median survival times
chemotherapy (HAIC) through a port-catheter system in of all patients and the operated patients were 309 and 386
patients with liver dysfunction due to synchronous and days, respectively.
unresectable liver metastases. The aim of HAIC was to Conclusion Initial HAIC administration is a safe and
improve patients clinical condition for later surgical efficacious method for improving liver function prior to
removal of primary colorectal cancer. operative resection of primary colorectal cancer in patients
Methods Port-catheter systems were placed radiologi- with liver dysfunction due to synchronous and unresectable
cally in 21 patients (mean age 58.6 8.1 years) with liver liver metastases.
dysfunction due to synchronous liver metastases from
colorectal cancer. Initial HAIC of 1,000 mg/m2 5-fluoro- Keywords Colorectal cancer 
uracil was administered weekly as a 5 hr continuous Hepatic arterial infusion chemotherapy  Liver metastasis 
infusion through this system. Surgical removal of the pri- Port-catheter system
mary lesion was planned after HAIC improved the liver
function.
Results Port-catheter system placement was successful in Introduction
all patients without severe complications. Patients were
followed up for a median of 309 days (range 51998 days). Colorectal cancer is the fourth most commonly diagnosed
After starting HAIC, no severe adverse events that caused malignant disease worldwide [1], and synchronous liver
drug loss and treatment postponement or suspension were metastases are identified in 1020% of cases [2]. However,
observed in any of the patients. HAIC was performed a the treatment protocol for patients with stage IV colorectal
mean of 4.5 3.0 times and the liver function improved in cancer with synchronous liver metastases has not been
all patients. Curative (n = 18) or palliative (n = 1) surgical firmly established [2, 3]. In such patients, the choice of
removal of the primary lesion was performed. The treatment strategy differs based on various factors such as
liver function, the patients condition, the urgency of
operating on the primary lesion, and the institutions pro-
T. Iguchi  Y. Inaba (&)  H. Yamaura  Y. Sato  M. Miyazaki tocols for dealing with liver metastases and primary
 H. Shimamoto
lesions. For the primary lesion, it is desirable that surgical
Department of Diagnostic and Interventional Radiology, Aichi
Cancer Center Hospital, 1-1 Kanokoden, Nagoya, Chikusa-ku removal is selected to improve the quality of life of the
464-8681, Japan patients, because colorectal cancer may cause obstruction,
e-mail: 105824@aichi-cc.jp perforation, bleeding, or pain [3]. Additionally it has been
reported that stage IV patients who underwent resection of
Y. Arai
Department of Diagnostic Radiology, National Cancer Center their asymptomatic primary lesions had prolonged median
Hospital, 511, Tsukiji, Tokyo, Chuo-ku 1040045, Japan and 2-year survival periods compared with stage IV

123
T. Iguchi et al.: Initial Hepatic Arterial Infusion Chemotherapy in Stage IV Colorectal Cancer Patients 87

patients who did not undergo resection [3]. However, metastases in both the right and left lobes of the liver. All
patients with advanced liver dysfunction due to synchro- hepatic lesions were unresectable. All patients had
nous liver metastases are not good candidates for surgical advanced liver dysfunction due to liver metastases, with
removal of the primary lesion. In such circumstances, increased levels of aspartate aminotransferase (AST; mean
surgeons and anesthetists usually hesitate to perform sur- 110 109 IU/l, range 21549 IU/l), alanine aminotrans-
gical removal of the primary lesion, mainly because the ferase (ALT; mean 59 43 IU/l, range 17183 IU/l), total
patients condition is too poor to perform surgery and the bilirubin (T-BIL; mean 1.0 0.5 mg/dl, range 0.32.2 mg/
liver is seen to be the prognosis-limiting factor. As a result dl), lactate dehydrogenase (LDH; mean 1242 1002 IU/l,
insufficient and palliative systemic chemotherapy might be range 2213,870 IU/l), alkaline phosphatase (ALP; mean
selected without performing surgical removal of the pri- 874 570 IU/l, range 4162660 IU/l), and gamma-glut-
mary lesion in many cases. amyl transpeptidase (GTP; mean 393 433 IU/l, range
With the recent advances in interventional radiology 1302,023 IU/l). Since the liver dysfunction in these
techniques, radiological placements of port-catheter system patients had already progressed, we decided to initially
are increasingly being used in Japan [4, 5]. Repeated administer HAIC instead of the standard systemic chemo-
hepatic arterial infusion chemotherapy (HAIC) that is therapy in order to improve their liver function. Even in
performed through an implanted port-catheter system is an patients with extrahepatic metastases, we initially admin-
effective therapy employed for unresectable advanced liver istered HAIC because we judged that liver metastasis was
malignancies [68]. In particular, many reports have indi- the prognosis-limiting factor. In 6 of 21 patients, extrahe-
cated that HAIC is effective for liver metastases from patic metastases were observed in organs such as the lung
colorectal cancer [68]. It reported that, compared with (n = 5), bone (n = 1), and lymph nodes (n = 1). Only 1
systemic chemotherapy, HAIC increased the possibility of patient showed evidence of hepatitis B and C virus infec-
tumor response and might improve liver function [6]. tion; no other patient had a history of hepatitis. Usually, we
The purpose of this study was to retrospectively evaluate consider T-BIL levels [3.0 mg/dl or an Eastern Co-oper-
the safety and efficacy of the initial administration of HAIC ative Oncology Group performance status [9] of 4 as the
through a port-catheter system in patients with advanced exclusion criteria for HAIC administration. However, in
liver dysfunction due to synchronous and unresectable liver this retrospective study, despite conforming to the exclu-
metastases from colorectal cancer. The aim of HAIC was sion criteria, 4 of 212 patients underwent HAIC; these
to improve their clinical condition for the later surgical patients were not included in the analysis because these
removal of the primary lesion. were not planned surgeries.

Materials and Methods Port-Catheter System Placement and HAIC

Approval from the institutional review board of our hos- All procedures for the placement of port-catheter systems
pital and informed consent from all the patients were were performed by interventional radiologists in the angi-
obtained before performing any procedure. ography suite with the patients under local anesthesia. The
procedure was performed as follows. All patients under-
went angiography before catheter placement, which was
Patients performed using a 5 Fr angiographic catheter (Clinical
Supply, Gifu, Japan) inserted from the right femoral artery
Between January 2000 and October 2004, 212 patients with to allow arterial mapping and to prevent extrahepatic influx
unresectable liver metastases from colorectal cancer of the anticancer agents. The extrahepatic arteries branch-
underwent radiological placement of port-catheter systems ing from the hepatic artery, such as the right gastric artery,
at our institution. In this study, 21 of 212 patients (4 men, posterior superior pancreatoduodenal artery, and superior
17 women; age 3977 years, mean 58.6 8.1 years) ini- duodenal artery, were embolized with microcoils (Tornado;
tially received HAIC through this system to prepare for the Cook, Bloomington, IN, USA or Trufill; Cordis, Miami
surgical removal of the primary lesion later; these patients Lakes, FL, USA) through a 2.5 Fr microcatheter (Jamiro;
had liver dysfunction due to synchronous liver metastases. Kaneka, Osaka, Japan or Sniper; Clinical Supply, Gifu,
The primary sites of malignancy were as follows: the Japan) inserted coaxially [10, 11]. In patients with more
cecum (n = 2), ascending colon (n = 5), transverse colon (n than two hepatic arteries, these arteries were converted into
= 3), sigmoid colon (n = 7), and rectum (n = 4). With the a single arterial supply by microcoil embolization so that
exception of 1 patient who had a large metastasis in the drugs could be distributed to the entire liver using a single
right lobe of the liver, all patients had diffuse or multiple indwelling catheter [10]. Next, a 5 Fr angiographic catheter

123
88 T. Iguchi et al.: Initial Hepatic Arterial Infusion Chemotherapy in Stage IV Colorectal Cancer Patients

was inserted from the left subclavian artery and advanced Placement of the Port-Catheter System
to the common hepatic artery via the celiac artery. Then an
indwelling catheter (Anthron P-U catheter; Toray Medical, The radiological placement of the port-catheter system was
Tokyo, Japan or W spiral catheter; PIOLAX, Yokohama, successful in all 21 patients. During and after the proce-
Japan) with a side hole was inserted using the catheter- dure, there were no complications such as hematoma,
exchange method. The catheter tip was inserted into the subclavian or vertebral artery thrombosis, infections,
deep segment of the gastroduodenal artery so that the side hepatic artery occlusions, and catheter malfunctions.
hole was placed into the common hepatic artery. The
gastroduodenal artery around the tip of the indwelling
catheter was embolized using microcoils and a mixture Clinical Course after Starting HAIC
(1:1.5) of n-butyl cyanoacrylate (NBCA; Histoacryl;
Braun, Melsungen, Germany) and iodized oil (Lipiodol After starting HAIC, no severe adverse events that caused
Ultrafluide; Laboratoire Guerbet, Roissy, France) through a drug loss and treatment postponement or suspension were
microcatheter inserted coaxially via the 5 Fr angiographic observed in any of the patients. HAIC was performed a
catheter inserted from the right femoral artery. Finally, the mean of 4.5 3.0 times (range 115 times) and the liver
proximal end of the indwelling catheter was connected to a function improved in all 21 patients. In particular, the AST,
port implanted in the subcutaneous pocket created in the ALT, LDH, ALP, and GTP levels were improved signifi-
left chest wall. cantly (Table 1). In 19 of 21 patients, curative (n = 18) or
Digital subtraction angiography and CT were performed palliative (n = 1) surgical removal of the primary lesion
during injection of contrast medium through the implanted was performed. In the remaining 2 patients, although the
port-catheter system within a few days of implantation to liver function had improved after HAIC was administered
confirm that the catheter was not dislodged and that the 15 times and 5 times, respectively, extrahepatic metastases
entire liver was perfused adequately. Thereafter, HAIC was in the lung, bone or peritoneum developed rapidly and their
administered through this system: 1,000 mg/m2 of 5-fluo- performance status worsened. Though systemic chemo-
rouracil (5-FU) weekly by continuous 5 hr infusion [7]. therapy was administered with or instead of HAIC
After administration of the chemotherapeutic agent, the afterward, they died 186 and 51 days, respectively, after
implanted port-catheter system was flushed and filled with the placement of port-catheter system; thus, surgery could
2 ml of heparin solution (1,000 IU/ml). not be performed.
Among the 19 patients who underwent surgery, HAIC
was administered a mean of 3.9 1.8 times (range 19
Statistical Analysis times), and the median period between placement of the

The success rate and the complications of the placement Table 1 Liver function before and after HAIC administration
of the port-catheter system were evaluated. After starting
Before starting After p value
HAIC the clinical course, including improvement in liver HAIC HAIC
function tests, performance of surgery, and survival were
evaluated. In patients who underwent surgical removal of AST (IU/l) Mean 110 109 56 55
the primary lesion, the frequency of HAIC administration, Range 21549 21273 0.0001*
time between the placement of the port-catheter system ALT (IU/l) Mean 59 43 31 22
and surgery, details of the surgery, postoperative therapy, Range 17183 11101 0.0005*
and survival were evaluated. The Wilcoxon signed rank TBIL (mg/dl) Mean 1.0 0.5 1.2 1.1
test was used to compare the liver functions before sur- Range 0.32.2 0.34.4 0.717
gery with those before starting HAIC. The cumulative LDH (IU/l) Mean 1242 1002 551 501
survival rate was calculated using the Kaplan-Meier Range 2213870 1742050 \0.0001*
method. ALP (IU/l) Mean 874 570 663 526
A p value of less than .05 was considered significant. Range 4162660 1242335 0.0046*
GTP (IU/l) Mean 393 433 207 169
Range 1302023 9602 0.0061*
Results
AST, aspartate aminotransferase; ALT, alanine aminotransferase; T-
BIL, total bilirubin; LDH, lactate dehydrogenase; ALP, alkaline
After placement of the port-catheter system, patients phosphatase; GTP, gamma-glutamyl transpeptidase; HAIC, hepatic
were followed up for a median of 309 days (range 51998 arterial infusion chemotherapy
days). *Significant at p \ 0.05

123
T. Iguchi et al.: Initial Hepatic Arterial Infusion Chemotherapy in Stage IV Colorectal Cancer Patients 89

c
Fig. 2 AC. A 55-year-old man with multiple liver metastases from
rectal cancer. A Contrast-enhanced CT scan obtained before starting
HAIC shows unresectable multiple liver metastases in both the right
and left lobes. B An arteriogram via the port obtained before starting
HAIC shows that all hepatic arteries are well visualized. The catheter
tip was inserted into the deep segment of the gastroduodenal artery and
embolized using microcoils and a mixture of n-butyl cyanoacrylate
and iodized oil. The side hole was placed into the common hepatic
artery (arrow). The accessory left hepatic artery, which branched from
the left gastric artery, was embolized with microcoils (arrowhead) in
order to establish hepatic arterial supply from a single vessel. C
Contrast-enhanced CT scan obtained after five HAIC administrations
shows slightly smaller multiple liver metastases. With the exception of
T-BIL, the patients liver function improved (AST improved from 83
to 26 IU/l, ALT improved from 49 to 18 IU/l, LDH improved from
1,155 to 458 IU/l, and ALP improved from 950 to 502 IU/l)

port-catheter system and surgery was 29 days (range 1468


days). Of 13 patients who had no extrahepatic metastases
prior to the surgery, 10 developed extrahepatic metastases.
Among 16 of 19 patients, systemic chemotherapy with or
instead of HAIC was administered after the surgery.
The overall median survival time of all the patients was
309 days and that of the patients who underwent surgery
was 386 days (Fig. 1). At present, 20 patients have died.
A representative case is shown in Fig. 2.

Discussion

Many studies have reported the effectiveness of HAIC


administration through a port-catheter system for liver
metastases from colorectal cancer [68]. In Western
countries, it has been reported that HAIC is effective in
treating liver metastases; however, it does not improve the
prognosis [6]. On the other hand, in Japan, good results
have been reported after intermittent hepatic arterial infu-
sion of a high dose of 5-FU: the response rate is reportedly
78% and the median survival time is 25.8 months [7].

In general, systemic chemotherapy is usually selected


for colorectal cancer with distant metastases [2]. Recently,
the standard regimens such as FOLFILI (5-FU plus leu-
covorin with oxaliplatin) and FORFOX (5-FU plus
leucovorin and irinotecan) are used, and the median sur-
vival after FOLFILI and FORFOX has been reported to be
12.621.5 months [12]. In many cases, systemic chemo-
therapy might be the first choice of treatment for patients
with primary colorectal cancer and synchronous distant
Fig. 1 Overall survival time metastases, and we usually select systemic chemotherapy

123
90 T. Iguchi et al.: Initial Hepatic Arterial Infusion Chemotherapy in Stage IV Colorectal Cancer Patients

as an initial therapy for such patients. Although it is because it is only recently that such standard regimens have
doubtful whether the initial HAIC administration is effec- been employed in practice in Japan. The survival period
tive in patients who have not undergone any therapy for the might have been prolonged if we had employed the currently
primary lesion, HAIC was administered initially in order to used standard systemic chemotherapy.
improve or control liver metastases. We judged that liver In conclusion, initial HAIC administration is a safe and
metastasis was the prognosis-limiting factor, because the efficacious method for improving liver function prior to
liver dysfunction in these patients had already progressed operative resection of primary colorectal cancer in patients
due to liver metastases. Additionally, we aim to surgically with liver dysfunction due to synchronous and unresectable
remove the primary lesion later, if possible, because pri- liver metastases.
mary colorectal cancer may cause obstruction, perforation,
bleeding, or pain [3]. Based on the results of this study, we
believe that initial HAIC administration is effective References
because, in 19 of 21 patients, surgery was possible after the
liver function had been improved by HAIC administration. 1. Ferlay J, Bray F, Pisani P, Parkin DM (2004) GLOBOCAN 2002:
In the 19 patients who underwent surgery, HAIC Cancer incidence, mortality and prevalence worldwide. IARC
CancerBase no. 5, version 2.0. IARC Press, Lyon, France
administration was terminated 1 week before surgery to 2. Alexander HR, Kemeny NE, Lawrence TS (2000) Metastatic
prevent its effect on surgery. The wide range of the fre- cancer to the liver. In: DeVita VT (ed) Cancer, 7th edn. Williams
quency of HAIC administration was due to the fact that & Wilkins, Baltimore, pp 23532368
surgery was not performed until, in the surgeons opinion, 3. Ruo L, Gougoutas C, Paty PB, et al. (2003) Elective bowel
resection for incurable stage IV colorectal cancer: Prognostic
the patients liver function had improved. We observed that variables for asymptomatic patients. J Am Coll Surg 196:722
the liver functions before surgery had improved signifi- 728
cantly after HAIC administration compared with those 4. Tanaka T, Arai Y, Inaba Y, et al. (2003) Radiologic placement of
before starting HAIC. It has been reported that HAIC has side-hole catheter with tip fixation for hepatic arterial infusion
chemotherapy. J Vasc Interv Radiol 14:6368
fewer side effects than systemic chemotherapy [13] and, in 5. Yamagami T, Iida S, Kato T, et al. (2002) Using n-butyl cya-
fact, we observed that surgeries could be performed with- noacrylate and the fixed-catheter-tip technique in percutaneous
out any adverse effects arising due to HAIC. We usually implantation of a port-catheter system in patients undergoing
consider T-BIL [3.0 mg/dl or a performance status of 4 as repeated hepatic arterial chemotherapy. AJR Am J Roentgenol
179:16111617
the exclusion criteria for HAIC administration because, 6. Meta-Analysis Group in Cancer (1996) Reappraisal of hepatic
based on our experience, it is difficult to reduce such liver arterial infusion in the treatment of nonresectable liver metastases
dysfunction and also improve performance status in from colorectal cancer. J Natl Cancer Inst 88:252258
patients. Further, the liver dysfunction of such patients may 7. Arai Y, Inaba Y, Takeuchi Y, et al. (1997) Intermittent hepatic
arterial infusion of high-dose 5FU on a weekly schedule for liver
be adversely affected by HAIC administration. Based on metastases from colorectal cancer. Cancer Chemother Pharmacol
our results, there were no severe adverse events after HAIC 40:526530
administration when these exclusion criteria were used for 8. Link KH, Sunelaitis E, Kornmann M, et al. (2001) Regional
the selection of the candidates. In 2 of 21 patients, although chemotherapy of nonresectable colorectal liver metastases with
mitoxantrone, 5-fluorouracil, folinic acid, and mitomycin C may
the liver function improved after HAIC administration, prolong survival. Cancer 92:27462753
surgery could not be performed because they developed 9. Oken MM, Creech RH, Tormey DC, et al. (1982) Toxicity and
extrahepatic metastases in the lung, bone or peritoneum, response criteria of the Eastern Cooperative Oncology Group.
and their performance status worsened. Unfortunately, we Am J Clin Oncol 5:649655
10. Arai Y, Inaba Y, Takeuchi Y (1997) Interventional techniques for
cannot expect HAIC administration to have an anticancer hepatic arterial infusion chemotherapy. In: Castaneda-Zuniga
effect on the entire body [7]. WR (ed) Interventional radiology, 3rd edn. Williams & Wilkins,
There were some limitations in our retrospective study. Baltimore, pp 192205
Firstly, the liver dysfunction of our patients was already 11. Inaba Y, Arai Y, Matsueda K, et al. (2001) Right gastric artery
embolization to prevent acute gastric mucosal lesions in patients
advanced; therefore, we hesitated to administer systemic undergoing repeat hepatic arterial infusion chemotherapy. J Vasc
chemotherapy when malignancy was first identified. Sec- Interv Radiol 12:957963
ondly, in many patients, other distant metastases were 12. Kelly H, Goldberg RM (2005) Systemic therapy for metastatic
present or developed and systemic chemotherapy was started colorectal cancer: Current options, current evidence. J Clin Oncol
23:45534560
after the surgery. We could not administer standard systemic 13. Collins JM (1984) Pharmacokinetic rationale for intra-arterial
chemotherapy such as FOLFILI and FORFOX, and our therapy. In: Howell SB (ed) Intra-arterial and intracavitary cancer
regimens of systemic chemotherapy were not established, chemotherapy. Martinus Nijhoff, Boston, pp 110

123
Cardiovasc Intervent Radiol (2008) 31:9197
DOI 10.1007/s00270-007-9170-y

CLINICAL INVESTIGATION

Impact of Multislice CT Angiography on Planning of Radiological


Catheter Placement for Hepatic Arterial Infusion Chemotherapy
Miyuki Sone Kenichi Kato Atsuo Hirose Tatsuhiko Nakasato
Makiko Tomabechi Shigeru Ehara Takao Hanari

Received: 30 December 2006 / Accepted: 10 September 2007 / Published online: 10 October 2007
 Springer Science+Business Media, LLC 2007

Abstract The objective of this study was to assess pro- in the right gastric artery, and 2.7 0.8/2.6 0.8 in the left
spectively the role of multislice CT angiography (MSCTA) gastric artery. No statistically significant differences exist
on planning of radiological catheter placement for hepatic between the two modalities. Interobserver agreement for
arterial infusion chemotherapy (HAIC). Forty-six patients MSCTA was equivalent to that for DSA. Two patients
with malignant liver tumors planned for HAIC were showed stenosis of the celiac trunk on both modalities.
included. In each patient, both MSCTA and intra-arterial Based on these imaging findings, technical success was
digital subtraction angiography (DSA) were performed, accomplished in all patients. In conclusion, MSCTA is
except one patient who did not undergo DSA. Comparison accurate in assessing arterial anatomy and abnormalities.
of MSCTA and DSA images was performed for the MSCTA can provide adequate information for planning of
remaining 45 patients. Detectability of anatomical variants radiological catheter placement for HAIC.
of the hepatic artery, course of the celiac trunk, visuali-
zation scores of arterial branches and interobserver Keywords Liver cancer 
agreement, presence of arterial stenosis, and technical Hepatic arterial infusion chemotherapy  Multislice CT 
outcome were evaluated. Anatomical variations of the CT angiography
hepatic artery were detected in 19 of 45 patients (42%) on
both modalities. The course of the celiac trunk was dif-
ferent in 12 patients. The visualization scores of celiac Introduction
arterial branches on MSCTA/DSA were 3.0 0/2.9 0.2
in the celiac trunk, 3.0 0/2.9 0.3 in the common Hepatic arterial infusion chemotherapy (HAIC) is an
hepatic artery, 2.9 0.2/2.9 0.3 in the proper hepatic established treatment for unresectable malignant hepatic
artery, 2.9 0.3/2.9 0.4 in the right hepatic artery, neoplasm [13]. Although a few randomized control trials
2.8 0.4/2.9 0.4 in the left hepatic artery, 2.9 0.2/ have shown a survival benefit of HAIC [4, 5], HAIC still
2.9 0.3 in the gastroduodenal artery, 2.1 0.8/2.2 0.9 remains an important treatment option because of the
higher response rate and lower systemic toxicities com-
pared with systemic chemotherapy [15]. For the
M. Sone (&)  K. Kato  T. Nakasato  M. Tomabechi 
S. Ehara implantation of a catheter and port system, a percutaneous
Department of Radiology, Iwate Medical University, approach using techniques of interventional radiology is
19-1, Uchimaru, Morioka 0208505, Japan now widely used instead of surgical methods [1, 3].
e-mail: msone@athena.ocn.ne.jp
Radiological catheter placement consists of serial proce-
A. Hirose dures including unification of multiple hepatic arteries,
Department of Radiology, Morioka Red-Cross Hospital, embolization of gastrointestinal branches arising from the
6-1-1, Sanbonyanagi, Morioka 0208560, Japan hepatic artery, and implantation of a catheter and port
system. All these procedures are needed to accomplish the
T. Hanari
Center for Radiological Sciences, Iwate Medical University, adequate distribution of anticancer agents to the entire liver
19-1, Uchimaru, Morioka 0208505, Japan without extrahepatic perfusion through the implanted

123
92 M. Sone et al.: Impact of Multislice CT Angiography on Catheter Placement for Chemotherapy

catheter and port system [1, 3, 6]. Incorrect evaluation of Tokyo). For 3-D image reconstruction, multiplanar volume
vascular anatomy sometimes leads to inadequate situations reformation (MPVR) images were obtained. When small
such as a defect of drug distribution in the liver, gastro- vessels, including the right gastric artery, were not visu-
intestinal complication, or catheter-related arterial alized on conventional MPVR images, additional images
occlusion resulting in treatment failure. Thus, vascular using targeted MPVR technique were used.
mapping is essential for planning of radiological catheter
placement. Although conventional angiography has been
used for such assessment, multislice CT angiography Digital Subtraction Angiography
(MSCTA) may be suitable for noninvasive abdominal
vascular evaluation [710]. However, the specific need for Intra-arterial digital subtraction angiography (DSA) using
radiological catheter and port placement by interventional Seldingers technique was performed just before implan-
radiologists, such as variations of the hepatic artery, gas- tation of the catheter and port system, within 2 weeks after
trointestinal branches arising from hepatic arteries, and the MSCTA. A transfemoral 5-Fr simple curved catheter was
course of the celiac trunk, has not been well evaluated. The inserted selectively into the celiac trunk. Twenty to twenty-
purpose of this study was to assess prospectively the role of five milliliters of nonionic contrast material (iopamidol;
MSCTA in planning of radiological catheter placement for Iopamiron 300; Bayer-Schering Pharma, Tokyo) was
HAIC. administered at a rate of 45 mL/s. Images were obtained
during expiration.

Materials and Methods


Image Interpretation and Evaluation
Patients
Four radiologists evaluated MSCTA and DSA images
Between August 2003 and March 2007, eligible patients independently, blinded to the other observers results or
were enrolled in this study. Eligibility criteria were as patient history. A 4-week interval was set between the
follows: (a) patients planned to undergo radiological sessions for evaluation of two modalities to eliminate the
catheter and port placement for HAIC, (b) patients without awareness of the findings of another modality. Arterial
prior catheter placement for HAIC or any kind of embo- anatomy and abnormalities were assessed regarding the
lotherapy to the celiac or hepatic arterial region, (c) following items.
patients without a history of adverse reaction to contrast
1. Anatomic variation of the hepatic artery
material and renal dysfunction (serum creatinine 2.0),
2. Course of the celiac trunk: cranial, horizontal, or
and (d) written informed consent obtained. This study
caudal
protocol was approved by the institutional review board of
3. Visualization of the following vessels: celiac trunk,
our institution.
common hepatic artery (CHA), proper hepatic artery
(PHA), right hepatic artery (RHA), left hepatic artery
(LHA), gastroduodenal artery (GDA), right gastric
MSCT Image Acquisition and Postprocessing
artery (RGA), and left gastric artery (LGA). Visual-
ization scores were on a scale of 13 (1, not visualized;
A 16-channel multislice CT scanner (Aquilion-16; Toshiba
2, fair visualization [i.e., only origin or proximal
Medical Systems, Tokyo) was used for the evaluation.
portion of the artery recognizable]; 3, excellent
MSCTA was performed as part of a pretreatment routine
visualization (i.e., entire course of the artery recogniz-
CT evaluation of the abdomen. Ninety to one hundred
able). Interobserver agreement was also evaluated.
milliliters of nonionic contrast material (iohexol; Omni-
4. Stenosis of 50% in any segment of the hepatic
paque 300 syringe; Daiichi-Sankyo, Inc., Tokyo) was
arteries
administered into the antecubital vein at a rate of 3 mL/s
with an automated injector (Dual Shot/ A-300; Nemoto
Kyorindo, Tokyo). For MSCTA, arterial-phase image
scanning was started within 8 s after reaching the bolus of Estimation of Radiation Dose
contrast material into the descending aorta (100 HU), using
a bolus-tracking method. Images were obtained during Radiation exposure was estimated by measuring the vol-
inspiration. A collimation of 0.5 mm with a helical pitch of ume computed tomography dose index (CTDIvol) and dose
15 was used. Three-dimensional (3-D) reconstruction was length product (DLP). CTDIw (mGy) was automatically
obtained on a dedicated workstation (Zio M900; Amin, calculated and displayed on the CT scanner using particular

123
M. Sone et al.: Impact of Multislice CT Angiography on Catheter Placement for Chemotherapy 93

parameters for the specific imaging protocol mentioned Table 1 Anatomic variations in hepatic artery on MSCTA and DSA
above. DLP (mGy  cm) was calculated by the following (number of arteries)
formula [11]: MSCTA DSA

DLP CTDIvol  scan length Total 19 19


RHA from SMA 12 12
where scan length (cm) is the length of the scan for LHA from LGA 6 6
individual patients. RHA from CHA 1 1
Note. MSCTA, multislice CT angiography; DSA, digital subtraction
angiography; RHA, right hepatic artery; SMA, superior mesenteric
Technical Outcome artery; LHA, left hepatic artery; LGA, left gastric artery; CHA,
common hepatic artery
Technical success was evaluated regarding unification of
multiple hepatic arteries, embolization of gastric arteries,
and implantation of catheter and port system. vascular injury and spasm resulting in possible hepatic or
mesenteric ischemia, due to this anatomical variation.
Therefore, 45 of 46 patients underwent DSA and entered in
the analysis.
Statistical Analysis Anatomical variations of the hepatic artery were present
in 19 of 45 patients (42%) on both MSCTA and DSA
Sensitivity and specificity were calculated for each arterial (Table 1, Figs. 1 and 2). The sensitivity of MSCTA was
segment. Visualization scores between MSCTA and DSA 100% and the specificity was 100% when DSA was used as
were compared by two-sided Wilcoxons signed-rank test. the gold standard. On MSCTA and DSA, the course of the
The interobserver agreement for visualization score of proximal portion of the celiac trunk was cranial in 9
MSCTA and DSA was evaluated using the j value of two patients on MSCTA/13 patients on DSA, horizontal in 7 on
readers among four (six pairs each). Perfect agreement MSCTA/9 on DSA, and caudal in 29 on MSCTA and 23 on
resulted in a j value of 1; agreement expected on the basis DSA. So, the courses of the celiac trunk in two modalities
of chance alone resulted in a j value of 0. All calculations were not corresponding in 12 patients. Regarding the
were performed using statistical analysis software (Dr. visualization scores of each arterial segment, no statisti-
SPSS II 11.0.1.J; SPSS Inc., Chicago, IL). cally significant differences were noted between MSCTA
and DSA (Table 2, Figs. 3 and 4). j values, which were
used to assess interobserver agreement for visualization
Results scores, ranged from 0.47 to 0.50 (mean, 0.49) for MSCTA
and from 0.41 to 0.57 (mean, 0.48) for DSA. Stenosis of
Patient Characteristics the celiac trunk was noted in two patients on both MSCTA
and DSA (Fig. 5).
Forty-six patients with unresectable malignant liver
tumors, 29 males and 17 females, ranging in age from 36 to
83 years (median, 63 years), were enrolled in this study. Technical Outcome
Thirty-three patients had metastatic liver tumors, seven had
hepatocellular carcinoma, three had gallbladder carcinoma, Except for one patient who did not undergo DSA and
and three had cholangiocellular carcinoma. catheter placement due to MSCTA findings mentioned
above, 45 patients in this study received interventional
proceduresunification of multiple hepatic arteries,
Image Evaluation embolization of gastric arteries arising from hepatic arter-
ies, and catheter placementwhich were successful in all
In all patients, MSCTA examination was technically ade- patients.
quate and was suitable for analysis. In one patient, DSA
was not performed, because MSCTA showed severe ste-
nosis of the common celiac trunk and the superior Estimated Radiation Dose
mesenteric artery (i.e., celiacomesenteric trunk), and we
judged that insertion of an angiographic catheter for DSA Calculated CTDIvol for this MSCTA acquisition protocol
and placement of an arterial infusion catheter should be was 33.8 mGy. Mean DLP standard deviation was
avoided to prevent catheter-related complications such as 773 95 mGy.

123
94 M. Sone et al.: Impact of Multislice CT Angiography on Catheter Placement for Chemotherapy

Fig. 2 Replaced left hepatic artery in a 73-year-old man with liver


metastases. A MSCTA shows the left hepatic artery originating from
the left gastric artery (arrows). B DSA of the celiac trunk confirms the
Fig. 1 Replaced right hepatic artery in a 46-year-old man with liver CT findings
metastases. A MSCTA shows the right hepatic artery originating from
the superior mesenteric artery (SMA; arrows). B Digital subtraction
angiogram (DSA) of the SMA confirms the CT findings Table 2 Visualization scores for each arterial segment
MSCTA DSA p
Discussion
Celiac trunk 3.0 0 2.9 0.2 0.080
CHA 3.0 0 2.9 0.3 0.220
This prospective study shows that MSCTA provides
PHA 2.9 0.2 2.9 0.3 0.079
important information for radiological catheter and port
placement for HAIC. Anatomical variation of the hepatic RHA 2.9 0.3 2.9 0.4 0.371
artery was demonstrated in 50% of patients on MSCTA, LHA 2.8 0.4 2.9 0.4 0.392
with a sensitivity of 100% and a specificity of 100%. GDA 2.9 0.2 2.9 0.3 0.076
Visualization of each vascular segment showed no statis- RGA 2.1 0.8 2.2 0.9 0.318
tically significant differences between MSCTA and DSA. LGA 2.7 0.8 2.6 0.8 0.236
Interobserver agreement for visualization score of MSCTA Note. Score: 1, not visualized; 2, fair; 3, excellent. CHA, common
was equivalent to that of DSA. hepatic artery; PHA, proper hepatic artery; RHA, right hepatic artery;
The hepatic vessels have a complicated anatomy, and LHA, left hepatic artery; GDA, gastroduodenal artery; RGA, right
gastric artery; LGA, left gastric artery
we frequently face anatomical variants during the inter-
ventional procedure of hepatic arterial catheter placement.
Such variants of the hepatic artery affect the technical that MSCTA is accurate in assessing hepatic and gastric
outcome of our interventional procedure. If there are any arteries [710]. The improvement in longitudinal resolu-
replaced or accessory hepatic arteries, arterial redistribu- tion with MSCT is a substantial advantage over single
tion to unify hepatic arteries is required to achieve the helical or nonhelical CT [713]. The shorter acquisition
delivery of chemotherapeutic agents to the entire liver via time may also be advantageous for obtaining a pure
an indwelling catheter [1, 3]. Recent studies have shown arterial phase by eliminating contamination of venous or

123
M. Sone et al.: Impact of Multislice CT Angiography on Catheter Placement for Chemotherapy 95

Fig. 3 A 79-year-old man with liver metastases. A MSCTA shows


the hepatic and gastrointestinal branches of the hepatic arteries. B
DSA confirms the CT findings

portal venous structures [1314]. In previous studies,


MSCTA demonstrated anatomical variation in 35% to
69% of patients [8, 9, 12, 14]. With regard to preoperative
evaluation for surgical HAIC pump placement using
laparotomy, two studies demonstrated an advantage of
MSCTA [15, 16]. However, there is no report on CTA
regarding radiological catheter and port placement. In
radiological catheter and port placement, gastrointestinal
arteries arising from hepatic arteries need to be embolized
to avoid possible adverse gastrointestinal events due to
exposure to high-dose chemotherapeutic agents [1, 3, 17].
Among the branches, the right gastric artery, arising from Fig. 4 A 75-year-old man with liver metastases. A MSCTA shows
hepatic arteries, is relatively difficult to visualize on CTA the right gastric artery (RGA), however, its origin is not clear. B, C
MPVR image, targeted to the interval shown in B, demonstrates the
because it tends to be small in diameter [10, 17]. Ta-
origin (arrow) of the RGA (arrowheads)
kahashi et al. reported the usefulness of MSCTA for
depiction of small hepatic artery branches [10]. They 2.9 0.3 for GDA and 2.1 0.8/2.2 0.9 for RGA.
reported that MSCTA detected 50 RGAs in 56 patients. Visualization scores for the RGA were lower than for
The previous studies of preoperative assessment for HAIC other hepatic arterial segments in both modalities, prob-
pump under laparotomy did not include assessment of the ably because of its small size and complex course.
gastroduodenal branches [15, 16]. In our study, mean However, the overall visualization score of MSCTA was
visualization scores on MSCTA/DSA were 2.9 0.2/ equivalent to that of DSA. MSCTA is advantageous in

123
96 M. Sone et al.: Impact of Multislice CT Angiography on Catheter Placement for Chemotherapy

The course of the celiac axis also affects the technique


of catheter placement. When a transfemoral or transepi-
gastric approach is used, a caudal course of the celiac axis
tends to be more difficult than a cranial course because
multiple inflection points result in a reduction in torque of
the catheter and guide wire. When the course is caudal, use
of a specific J-type long sheath is considered for implan-
tation of the catheter and port system [18]. Evaluation of
the celiac axis prior to catheter and port implantation is
important for planning the procedure. In this study, respi-
ratory phase was consistent with routine examination and
different between MSCTA and DSA. Lee et al. reported
that there was no significant difference between inspiration
and expiration in the angles between the celiac trunk and
the aorta on MRI [19]. Factors other than respiration, such
as redundant and tortuous celiac axis, may have affected
the difference in our study. Even considering the effect of
respiration, MSCTA showed an advantage in detecting the
origin of the celiac trunk from the aorta constantly. Celiac
DSA did not always demonstrate the origin of the celiac
trunk. The 3-D capability of MSCTA is also effective for
visualization of the celiac trunk from various directions.
Another important indication of preoperative vascular
evaluation for HAIC is to detect vascular abnormalities
such as stenosis or occlusion along the implantation route.
Stenosis or occlusion may result not only in complications
during procedures of angiography and catheter implanta-
tion, but also in early occlusion of the hepatic artery after
catheter placement. In a postprocedure setting, MSCTA
with intra-arterial administration of contrast material from
an implanted catheter and port system help detect arterial
stenosis [20]. In the present study, MSCTA with intrave-
nous administration of contrast material demonstrated
stenosis in two patients. In one patient with severe stenosis
of the common stem of the celiac and mesenteric arteries,
in which even diagnostic angiography poses a risk of
complication, we could preclude angiography and catheter
placement based on MSCTA findings.
In our series, interventional radiological placement of
the catheter and port system was successful in all patients
with preprocedural planning by MSCTA. Previously, we
performed two sessions of DSA occasionally, one for
evaluation of vascular anatomy and partial embolization
and the other for placement of the catheter and port system
and embolization. After implementation of MSCTA, this
Fig. 5 Stenosis of the celiac trunk in a 57-year-old man with liver procedure has been done in one session, which may result
metastases. A, B Coronal (A) and sagittal (B) MPVR images show in a shorter hospital stay.
stenosis of the origin of the celiac trunk (arrows) and dilatation of the Optimization of the radiation dose is crucial for
inferior pancreatic artery and the gastroduodenal artery (arrowheads)
MSCTA. In the current study, the radiation dose repre-
sented by CTDIvol and DLP was equivalent to the criteria
giving information in advance of catheter and port from the European Guidelines on Quality Criteria for CT:
placement, which may reduce the procedural time and weighted CTDI of 35 mGy and weighted DLP of 780 mGy
avoid complications.  cm [21]. Although the radiation dose was not as low as in

123
M. Sone et al.: Impact of Multislice CT Angiography on Catheter Placement for Chemotherapy 97

the previous report of a spiral and conventional CT study 6. Seki H, Kimura M, Yoshimura N, et al. (1999) Hepatic arterial
[22], our result is acceptable because this examination is infusion chemotherapy using percutaneous catheter placement
with an implantable port: assessment of factors affecting patency
designed for single, not repeated, preprocedural assessment of the hepatic artery. Clin Radiol 54:221227
including evaluation of liver tumor, nodal, and peritoneal 7. Guiney MJ, Kruskal JB, Sosna J, et al. (2003) Multi-detector row
tumor extent. Furthermore, MSCTA images provide ade- CT of relevant vascular anatomy of the surgical plane in split-
quate information on vascular mapping, and they lead to an liver transplantation. Radiology 229:401407
8. Sahani D, Saini S, Pena C, et al. (2002) Using multidetector CT
overall reduction of radiation dose. Nonetheless, further for preoperative vascular evaluation of liver neoplasms: tech-
radiation dose reduction should be made, while maintain- nique and results. AJR 179:5359
ing image quality. Techniques to reduce the radiation dose 9. Byun JH, Kim TK, Lee SS, et al. (2003) Evaluation of the hepatic
include reducing the milliampere-second value, increasing artery in potential donors for living donor liver transplantation by
computed tomography angiography using multidetector-row
the pitch, adjusting the milliampere-second value accord- computed tomography: comparison of volume rendering and
ing to the patients size, and reducing the beam energy [11, maximum intensity projection techniques. J Comput Assist To-
23]. More important is to eliminate unnecessary CT mogr 27:125131
examination or excessive multiphase study. 10. Takahashi S, Murakami T, Takamura M, et al. (2002) Multi-
detector row helical CT angiography of hepatic vessels: depiction
Limitations of our study include the small number of with dual-arterial phase acquisition during single breath hold.
patients and lack of comparison with other noninvasive Radiology 222:8188
modalities such as duplex US. At our institute, contrast- 11. McNitt-Gray MF (2002) AAPM/RSNA physics tutorial for resi-
enhanced CT is routinely used for pretreatment evaluation of dents: topics in CT: radiation dose in CT. Radiographics
22:15411553
malignant hepatic tumors and additional MSCTA in the 12. Erbay N, Raptopoulos V, Pomfret EA, et al. (2003) Living donor
same study was considered to be adequate. Another limita- liver transplantation in adults: vascular variants important in
tion is that we did not evaluate diverse reformation surgical planning for donors and recipients. AJR 181:109114
techniques for MSCTA, such as volume rendering (VR). We 13. Fishman EK (2001) From the RSNA refresher courses: CT
angiography: clinical applications in the abdomen. Radiographics
chose the MPVR technique based on our previous experience 21:316
and reports by other investigators [9, 18, 24]. The application 14. Winter TC III, Nghiem HV, Freeny PC, et al. (1995) Hepatic
of target VR eliminated partial volume averaging on MPVR. arterial anatomy: demonstration of normal supply and vascular
This technique was useful in the visualization of small variants with three-dimensional CT angiography. Radiographics
15:771780
arteries such as the RGA, however, other reformation tech- 15. Sahani DV, Krishnamurthy SK, Kalva S, et al. (2004) Multide-
niques should be taken into consideration. tector-row computed tomography angiography for planning intra-
In conclusion, MSCTA is accurate in the detection of arterial chemotherapy pump placement in patients with colorectal
abdominal arterial anatomy, variations, and abnormalities. metastases to the liver. J Comput Assist Tomogr 28:478484
16. Kapoor V, Brancatelli G, Federle MP, et al. (2003) Multidetector
MSCTA is suitable for planning of the catheter and port CT arteriography with volumetric three-dimensional rendering to
system implantation for HAIC. evaluate patients with metastatic colorectal disease for placement
of a floxuridine infusion pump. AJR 181:455463
Acknowledgment We are grateful to the Japanese Society of 17. Yamagami T, Nakamura T, Iida S, Kato T, Nishimura T (2002)
Implantable Port Assisted Regional Treatments (JSIPART) for Embolization of the right gastric artery before hepatic arterial
awarding and supporting this study. infusion chemotherapy to prevent gastric mucosal lesions:
approach through the hepatic artery versus the left gastric artery.
AJR 179:16051610
18. Katoh K, Sone M, Nakasato T, et al. (2006) A new method using
References J-type long sheath for implantation of indwelling catheters for
trans-femoral hepatic arterial infusion. Radiat Med 24:8083
1. Arai Y, Inaba Y, Takeuchi Y (1997) Interventional techniques for 19. Lee VS, Morgan JN, Tan AGS, et al. (2003) Celiac artery com-
hepatic arterial infusion chemotherapy. In: Casterneda-Zuniga E pression by the median arcuate ligament: a pitfall of end-
(ed) Interventional radiology. Williams & Wilkins, Baltimore, pp expiratory MR imaging. Radiology 228:437442
192205 20. Sone M, Kato K, Nakasato T, et al. (2002) Multislice CT angiography
2. Kemeny NE, Ron IG (1999) Hepatic arterial chemotherapy in through an implantable catheter and port system: early experience in
metastatic colorectal patients. Semin Oncol 26:524535 detection of vascular complications during hepatic arterial infusion
3. Tanaka T, Arai Y, Inaba Y, et al. (2003) Radiologic placement of chemotherapy. J Comput Assist Tomogr 26:515519
side-hole catheter with tip fixation for hepatic arterial infusion 21. European guidelines on quality criteria for computed tomogra-
chemotherapy. J Vasc Interv Radiol 14:6368 phy. Available at: http://www.drs.dk/guidelines/ct/quality/.
4. Meta-Analysis Group in Cancer (1996) Meta-Analysis Group in Accessed December 12, 2006
Cancer: reappraisal of hepatic arterial infusion in the treatment of 22. Hidajat N, Wolf M, Nunnemann A, et al. (2001) Survey of
nonresectable liver metastases from colorectal cancer. J Natl conventional and spiral CT doses. Radiology 218:395401
Cancer Inst 88:252258 23. Karla MK, Maher MM, Toth TL, et al. (2004) Strategies for CT
5. Kemeny N, Niedzwiecki D, Hollis DR (2006) Hepatic arterial radiation dose optimization. Radiology 230:619628
infusion versus systemic therapy for hepatic metastases from 24. Mallouhi A, Rieger M, Czermak B, et al. (2003) Volume-ren-
colorectal cancer: a randomized trial of efficacy, quality of life, and dered multidetector CT angiography: noninvasive follow-up of
molecular markers (CALGB 9481). J Clin Oncol 24:13951403 patients treated with renal artery stents. AJR 180:233239

123
Springer Science+Business Media, LLC 2007 Cardiovasc Intervent Radiol (2008) 31:98106
CardioVascular Published Online: 23 June 2007 DOI: 10.1007/s00270-004-0232-0
and Interventional
Radiology

CLINICAL INVESTIGATION

Rationale for a Male Lumpectomy, a Prostate


Cancer Targeted Approach Using
Cryoablation: Results in 21 Patients with at
Least 2 Years of Follow-Up
Gary Onik
Division of Surgical Imaging, Center for Surgical Advancement, Department of Radiology and Urology, Celebration Health/Florida
Hospital, Celebration, FL 34747, USA

Abstract Key words: Prostate cancerMale lumpectomy

Background: Prostate cancer in men raises many of the


Current management of prostate cancer covers both ends of
same issues that breast cancer does in women. Complica-
the treatment spectrum. Patients can elect no treatment at
tions of prostate cancer treatment, including impotence and
all, i.e., watchful waiting [1], or aggressive whole-gland
incontinence, affect the self-image and psyche of a man no
treatments such as radical prostatectomy, with no middle
less than does the loss of a breast in a woman. We present a
ground available. Both management approaches leave
pilot study in which 21 patients were treated with a focal
something to be desired. In watchful waiting a patient
cryoablation procedure.
may miss his window of curability, particularly if his cancer
Methods: Focal cryoablation was performed using biplane
is under-staged to begin with. Aggressive whole-gland
transrectal ultrasound if the tumor was confined to only one
treatments can have significant lifestyle-limiting morbidity
prostate lobe. Preoperative PSA values were recorded.
such as incontinence and impotence. The literature indicates
Cryoablation was planned to encompass the area of known
that nerve-sparing radical prostatectomy and radiation have
tumor. PSA values were obtained every 3 months for 2 years
similar long-term potency rates of approximately 50% [2,
and every 6 months thereafter. Potency and continence
3]. Incontinence is also a significant risk of both radiation
status was obtained at the same intervals. Routine biopsy
and radical prostatectomy [4, 5]. Prostate cryoablation of the
was obtained at 1 year.
whole gland has become a standard treatment for primary
Results: Twenty-one patients had focal cryoablation. Fol-
prostate cancer, having been approved for Medicare cov-
low-up ranged from 24 to 105 months with a mean of 50
erage in the United States approximately 5 years ago [6].
months. Twenty of 21 (95%) patients have stable PSA
Whole-gland cryoablation has been shown to have excellent
values with no evidence for cancer, despite 10 patients being
long-term cancer control rates with an 88% 7-year disease-
at medium to high risk for recurrence. All patients biopsied
free survival in low-risk as well as high-risk patients [7].
(n = 19) were negative for tumor. Potency was maintained
While whole-gland cryoablation has resulted in a lower
in 17 of 21 patients (80%). No other complications,
incontinence rate (2%) than radical prostatectomy a high
including incontinence or fistula formation, were noted.
degree of impotence has been its limiting factor [8]. Cryo-
Conclusion: These preliminary results indicate a male
ablationusing imaging guidance unlike traditional treat-
lumpectomy, in which the prostate tumor region itself is
ments such as radical prostatectomy and radiation
destroyed, appears to preserve potency in a majority of pa-
however, is technically very well suited to a lumpectomy-
tients and limits other complications, without compromising
type approach in which only the cancer in the prostate is
cancer control. If these results are confirmed by further stu-
targeted. It stands to reason that if no treatment at all can be
dies and long-term follow-up, this treatment approach could
advocated for a subset of prostate cancer patients (i.e.,
have a profound effect on prostate cancer management.
watchful waiting) then the compromise of attempting to
destroy just the focus of cancer in the gland could be a
Correspondence to: Gary Onik M.D.; email: onikcryo@aol.com viable option as well.
G. Onik: Rationale for a Male Lumpectomy 99

The main theoretical objection to a lumpectomy-type 3. Later in our experience we used a transperineal mapping
approach to prostate cancer is the realization that prostate biopsy in which biopsies were taken every 5 mm throughout
cancer is often a multifocal disease within the prostate. The the length of the gland using a standard brachytherapy grid.
prostate cancer pathology literature clearly shows that many Each sample was separately labeled as to its location. The
patients do not have multifocal prostate cancer. In a study positive results could then be correlated back to the ultra-
examining radical prostatectomy specimens Djavan et al. [9] sound image with the grid overlay to use in targeting the
showed that patients with unifocal disease constituted one cancer.
third of the cases studied and could be reliably differentiated The second criterion was that the patient was potent, based on
from patients with multifocal disease with a sensitivity of their history, and that maintenance of potency was a primary
90% using the prostate specific antigen (PSA) density of the concern of the patient. Standardized questionnaires for potency
transition zone and percent free to total PSA. In addition, were not obtained. Cryosurgical informed consent was given
Villiers et al. [10], showed that 80% of multifocal tumors describing the known risks of cryosurgery including bleeding,
are less than 0.5 ml in volume, indicating they may not be of infection, rectal fistula, and incontinence. All patients were in-
clinical significance. This finding was confirmed by the formed of the additional risk of tumor being left untreated in any
studies of Rukstalis et al. [11] and Noguchi et al. [12] in tissue not frozen. Patients treated after February 1999 signed
which pathologic examination showed that unifocal tumors consent forms approved by our institutional review board. Patients
in whom these biopsies were negative were offered a focal tumor
were present in 20% and 25% of patients, respectively, and
treatment.
that using the size criterion of 5 mm or less as an insig-
nificant tumor, an additional 60% and 39% of patients,
respectively, might be candidates for a focal treatment ap-
proach. Clearly, then, based on the known pathology of Procedure
prostate cancer an opportunity exists to investigate a The ultrasound-guided percutaneous prostate cryoablation proce-
lumpectomy-type approach. dure was the same as that described originally by Onik et al. [14].
The rationale for a focal destruction of prostate cancer The following changes were made to the procedure to accommo-
would therefore be to decrease overall patient morbidity date the concept of tumor targeting and to increase the safety and
while maintaining adequate cancer control. An earlier report efficacy of the procedure: The extent of freezing was tailored to the
by Onik et al. [13] on 9 patients treated with focal cryoab- particular patient and was determined by the patients clinical
parameters, which included: Gleason grade, stage, PSA level, and
lation indicated that good cancer control with high potency
extent and location of cancer on preoperative biopsies. The PSA
and continence rates could be achieved using a focal cancer level indirectly influenced the amount of freezing, as a high PSA
treatment approach. In this paper we present an updated (over 10 ng/ml) is one of the criteria for a patient with a high risk
retrospective analysis of 21 patients, of that earlier report, in of recurrence: we therefore tended to freeze more tissue if this
whom a cancer-targeted cryoablation was carried out. criterion was present. In all patients an attempt was made to spare
one neurovascular bundle on the side opposite the tumor. The
Materials and Methods neurovascular bundle was destroyed on the side of the patients
tumor in all patients.
Patient Selection The cryoprobes were 3.4 mm blunt-tipped probes placed using a
Seldinger technique. From one to eight probes could be used
Patients were initially considered for cancer-targeted cryoablation
depending on the size of the patients gland and the extent of freezing
if, using the following methods, it was determined that their cancer
required. Cryoprobes were placed approximately 1 cm apart in the
was confined to one prostate lobe:
regions to be destroyed and within 5 mm of the capsule on the side of
1. Original transrectal ultrasound (TRUS) biopsy showed dis- the tumor. It is known that the spread from the prostate into the
ease confined to one prostate lobe. seminal vesicles occurs in the midline where the seminal vesicles
2. A repeat TRUS biopsy confirmed the original biopsy find- enter the prostate. Tumor involvement of the lateral centimeter of the
ings, with the repeat biopsy being carried out with the fol- seminal vesicle is therefore unusual except in the most advanced
cases. We therefore treated this region prophylactically in all but one
lowing method:
patient to prevent seminal vesicle involvement.
(a). Re-biopsy was carried out on the previously negative Tissue temperature monitoring was carried out in critical
prostate lobe. locations such as the apex of the gland and the neurovascular
(b). Color Doppler ultrasound examination was carried out in bundle on the side of the tumor to ensure adequate tumor-
all patients. Any areas of abnormal increased flow and or destructive freezing temperatures of )35C. Freezing was carried
out using copper/constantan thermocouples place by ultrasound
hypoechogenicity were biopsied under TRUS guidance.
guidance. A continuous reading of the thermocouple temperature
(c). In addition, the re-biopsy protocol included five cores ta-
was provided by the cryosurgical equipment (Endocare, Irvine,
ken from the lateral peripheral zone evenly spaced from CA, USA).
base to apex and two or three cores taken from the medial To insure adequate destruction the neurovascular bundle was
peripheral zone at base, mid-gland, and apex. identified by color Doppler imaging and frozen at least twice. The
(d). Prostate MRI or MRI spectroscopy were not routinely temperature of the neurovascular bundle opposite the tumor was
obtained. also monitored to prevent bundle destruction.
100 G. Onik: Rationale for a Male Lumpectomy

In order to decrease the chance for urethro-rectal fistula, just high risk, 5 patients were categorized as medium risk and
prior to the start of freezing a 22 gauge spinal needle was placed 5 patients as high risk with the remainder as low risk. All
into Denonvilliers fascia via a transperineal route. Once the needle procedures were technically successful and the intended
was noted to be in an adequate position normal saline was injected target was covered in all cases. Preoperative androgen
into the space separating the rectum from the prostate. Approxi-
ablation therapy of 6 months was carried out in 11 pa-
mately 50 ml of saline was injected. The space was maintained by
tients. Four of the patients, with tumors involving the mid-
continued downward pressure of the TRUS transducer and the
advancement of the ice-ball was continuously monitored using line, had bilateral gland freezing with an attempt to spare
TRUS to insure that the freezing was stopped prior to reaching the only the neurovascular bundle (Table 2). One patient had
rectum. only the area of the tumor, with an appropriate margin
An argon gas based system was used in all cases to carry out the around the tumor, frozen. We found that due to shadowing
freezing (Endocare, Irvine, CA, USA), replacing the earlier origi- caused by the ice formation and the fact that frozen tissue
nal liquid nitrogen freezing equipment originally used for whole- does not necessarily die, the advantages of the ultrasound
gland cryosurgery. imaging were limited to confirmation of cryoprobe and
A Foley catheter was left in place after the procedure, for a thermocouple placement and the approach of the ice-ball
variable amount of time depending on the extent of freezing. The to the rectum (Fig. 1) The remainder of the patients had
Foley catheter was needed due to the postprocedural swelling of
half of the gland frozen.
the prostate which prevented normal voiding.
All patients were potent preoperatively, although one
Patients on combined hormonal therapy prior to cryoablation
had the therapy stopped immediately after treatment. patient, who had had a previous transurethral resection of
the prostate (TURP), was primarily treated with this tech-
nique to decrease the chance of postoperative incontinence.
Patient Follow-Up All but one of the patients have stable PSA values at this
All patients were removed from combined hormonal therapy after time based on the ASTRO criteria, with the postoperative
the procedure. A PSA level was obtained every 3 months for the PSA stabilizing at some fraction of the preoperative value
first 2 years and then every 6 months thereafter. Patients were depending on the extent of the gland freezing (Fig. 2). The
considered to have a stable PSA if they met the ASTRO criteria one patient with an unstable PSA was a high-risk patient,
(three consecutive rises in PSA after a post-treatment nadir with a Gleason grade 8 adenocarcinoma, PSA of 11.4 ng/ml
defining a failure) of being biochemically disease-free. All patients and stage T2b. Postoperative saturation biopsy of this pa-
were requested to have routine TRUS biopsies at 1 year after the tient using a brachytherapy grid and a transperineal ap-
procedure (as defined by our repeat biopsy protocol above) proach (45 cores taken) revealed no evidence for cancer on
regardless of their PSA stability, to include both the treated and
either the side treated or the opposite side, and he is now
untreated side. Patients were followed up by written questionnaire
considered to have micro-metastatic disease. One additional
and telephone.
The questionnaire asked the following questions: patient initially had an unstable PSA with a negative post-
operative TRUS biopsy. Spectroscopic MRI of the prostate
1. Latest PSA value. was carried out indicating cancer in the untreated side of his
2. Whether the patient was having erections sufficient for gland. He was re-treated with cryoablation to that side and
vaginal penetration. now has an undetectable PSA.
3. Whether the patient was satisfied with his sexual function- Of the 21 patients, 19 had postoperative biopsies at 1
ing. year, with 2 patients with stable PSA values refusing biopsy.
4. Number of pads the patient was wearing for incontinence. All 19 patients had negative biopsies on both the treated and
5. Any other complications the patient experienced. untreated side. Biopsies of the treated areas in all patients
Patients were considered potent if they had erections sufficient showed coagulative necrosis, benign stroma, and inflam-
for vaginal penetration and they were satisfied with their sexual matory infiltrates. Biopsies in untreated areas showed a
functioning. Patients were considered incontinent if they used any benign normal prostate in all patients with some showing
pads at any time. areas of chronic prostatitis. The mean and median preop-
The mean, median, and standard deviation were calculated for erative PSA values were 7.6 ng/ml and 6.0 ng/ml, respec-
pre- and postoperative PSA values. tively. The mean and median postoperative PSA values
were 2.2 ng/ml and 1.4 ng/ml, respectively, with the post-
operative PSA reflecting the last PSA obtained for each
Results patient.
Between June 1995 and March 2002, 22 patients had Potency was maintained in the 17 of 20 patients (85%)
cancer-targeted cryoablation. One patient was lost to fol- who had a single treatment. The 1 patient re-treated on the
low-up, leaving 21 patients for evaluation (Table 1). Using opposite side of the original freeze is impotent as expected.
the traditional criteria with which to stratify patients as to One patient who was impotent after the procedure (the first
risk of recurrence (PSA >10 ng/ml, Gleason score >6, patient treated) had the neurovascular bundle on the nerve-
stage >T2a), in which patients with one risk factor are sparing side frozen to 20C, probably accounting for his
considered medium risk and patients with two risk factors postoperative impotence. The second impotent patient was
G. Onik: Rationale for a Male Lumpectomy 101

Table 1. Pre-operative Patient Characteristics


Patient Age Months F/U Gleason Pre-op PSA Positive cores Stage Pre-op ADT

1 62 105 5 10.5 1 T1c No


2 64 95 8 12.9 1 T2a Yes
3 58 94 5 4 1 T2a Yes
4 61 93 6 1.5 1 T2a No
5 67 74 6 4.5 1 T1c No
6 72 62 6 10.8 1 T1c Yes
7 56 55 6 5.5 1 T2a No
8 67 52 7 1.5 1 T2a No
9 64 40 3 6 1 T1c No
10 65 36 7 6 1 T1c Yes
11 62 36 6 7.3 1 T1c No
12 64 36 6 0.9 1 T2a No
13 71 35 6 5.7 1 T1c Yes
14 63 35 7 18 1 T1c Yes
15 74 34 7 13 3 T2b Yes
16 61 34 ? 7.2 1 T1c No
17 61 33 6 8.5 2 T1c Yes
18 51 32 6 16 3 T1c Yes
19 59 30 6 4.4 1 T1c Yes
20 68 30 7 4.8 2 T2a No
21 75 24 8 11.4 2 T2b Yes
50.7142857 Mean 7.638095238
36 Median 6
STDev 4.030508653

ADT, androgen deprivation therapy; PSA, prostate specific antigen

Table 2. Post Operative Results


Patient Area frozen PSA Cur PSA stable Biopsy Potency change

1 Bilateral / 1NVB spared 0.1 Yes Negative Impotent


2 Bilateral / apex only 0.5 Yes Negative Potent with viagra
3 Half 0.7 Yes Negative Potent
4 Bilateral / 1NVB spared 0.5 Yes Negative Potent
5 Half 1.04 Yes Refused Potent
6 Half 3.6 Yes Negative Impotent
7 Half 1.8 Yes Negative Potent
8 Half 0.5 Yes Negative Potent
9 Tumor + 1.5 cm margin 4.5 Yes Negative Potent
10 Half 0.8 Yes Negative Potent with viagra
11 Half 0.01 Yes Negative/Positive SMRI Impotent/refrozen bilaterally
12 Half 0.4 Yes Negative Potent
13 Half 2.6 Yes Negative Potent
14 Half 1.4 Yes Negative Potent with viagra
15 Half 5.1 Yes Negative Potent
16 Half 2.9 Yes Refused Potent
17 Half 1.7 Yes Negative Potent
18 Bilateral /1 NVB spared 5.5 Yes Negative Potent
19 Half 1.4 Yes Negative Potent
20 Half 1.6 Yes Negative Impotent
21 Half 10.1 No Negative Potent
Mean 2.22619
Median 1.4
STDev 1.06066

ADT, androgen deprivation therapy; PSA, prostate specific antigen

72 years old with multiple medical problems and a history incontinence which resolved at 6 months after the proce-
of a previous TURP, who was on chronic steroids for dure. The third impotent patient has no recognizable reason
asthma. Focal cryoablation was originally carried out in this for the postoperative impotence.
patient to minimize the chance of incontinence rather than No other instances of complications previously described
potency sparing. This patient required a postoperative with cryoablation, such as permanent incontinence, penile
TURP to remove sloughed tissue and had transient stress numbness or fistula formation, have occurred.
102 G. Onik: Rationale for a Male Lumpectomy

Fig. 1. A T2-weighted MR image of the prostate. A small area of decreased intensity approximately 4 mm in diameter is seen
in the left peripheral zone of the prostate just abutting the surgical capsule (black arrow). The location of the neurovascular
bundle is indicated by NVB. B Transrectal ultrasound (TRUS) of the prostate shows a hypoechoic area (arrow) that corre-
sponds exactly to the location of the area of decreased intensity on the MR image. A TRUS-directed biopsy revealed this to be
a Gleason grade 3 carcinoma. C A bright dot representing the cryoprobe can be seen in the hypoechoic area (white arrow). D
Freezing is seen covering the area of the previous hypoechoic tumor. Freezing appears as a hyperechoic rim with posterior
acoustic shadowing.

Discussion cancer adequately enough to form the basis for recognizing


this patient population [16]. In addition, we do believe that
The use of breast-sparing surgery. i.e., lumpectomy, to the traditional sextant biopsy would probably be inadequate
treat breast cancer has revolutionized the local control of for patient selection in focal therapy. Nevertheless, it has
that disease. Lumpectomy has shown that the quality of life been demonstrated that optimization of biopsy results by a
of the individual patient can successfully be integrated into second set of biopsies and improved gland sampling can
the equation of cancer treatment without major loss of greatly diminish the chances of missing a significant mul-
cancer treatment efficacy [15]. Prostate cancer in men raises tifocal tumor [17]. In addition, demonstration of negative
many of the same issues that breast cancer does in women. biopsies on the nerve-sparing side is an excellent predictor
Complications of prostate cancer treatment producing of negative margins at nerve-sparing radical prostatectomy
impotence and incontinence affect a mans self-image and and therefore presumably minimal disease on the side
psyche no less than the loss of a breast affects a woman. As opposite the index cancer [18]. Newer biopsy techniques
has already been stated, a significant number of men have (not carried out in our original series) now being used, in
unifocal cancer and therefore from a pathologic standpoint which the gland is biopsied every 5 mm using a brachy-
would be amenable to a lumpectomy treatment approach. therapy-type grid, could also have an impact on excluding
However, the question has to be raised whether these patients with significant multifocal disease. Our preliminary
patients with unifocal disease can be identified clinically. It results in 25 patients undergoing three-dimensional global
is well known that there is no radiographic imaging cur- mapping biopsies on the side opposite the index cancer
rently available which can determine the extent of prostate showed that cancer could be demonstrated in 50% of pa-
G. Onik: Rationale for a Male Lumpectomy 103

Fig. 2. The comparative pre- and postoperative PSA values for each patient.

tients who previously had negative biopsies in the suppos- were low, medium or high risk for biochemical failure.
edly uninvolved prostate lobe [19]. Based on this analysis the range of results for cryoablation
Since the anatomy of the prostate gland does not make it was equivalent to that of all other treatments in low- and
amenable to partial removal or lumpectomy, tumor medium-risk patients and appeared to be superior in high-
destruction by another modality is needed to achieve a risk patients. Overall complication rates were similar with
lumpectomy in a man. We chose cryoablation because it has all the modalities. The only article directly comparing
a long history of effective tumor treatment in various parts cryoablation with radical prostatectomy, by Gould et al.
of the body. The early negative results that prostate cryo- [21], showed cryoablation to be equivalent to radical pro-
ablation experienced have been largely mitigated by major statectomy in low-risk patients but, as the patients preop-
technical advances in the procedure such as improved ure- erative PSA level increased, cryoablation results were
thral warmer design, and it has been shown to be an superior to radical prostatectomy. The basis for this apparent
effective and safe alternative in treating prostate cancer superiority in high-risk patients may be the ability of
involving the whole gland. Approximately 5 years ago cryoablation to treat extracapsular extension of cancer and
prostate cryoablation was approved by Medicare in the to be repeated without any additional morbidity if needed.
United States as a treatment for primary prostate cancer Based on these results one can conclude that cryoablation is
(removing it from the investigational category), and is the a safe and effective treatment for treating prostate cancer
only treatment specifically approved for treating patients and that its inherent ability to be tailored to the extent of a
with recurrent cancer after radiation therapy. Long-term 5- patients disease makes it a platform upon which a treatment
year and 7-year data recently published by Donnelly et al. such as lumpectomy can be based.
[8] and Bahn et al. [7] are now available that confirm Our earlier results previously reported in the journal
cryoablation is a treatment competitive with both surgery Urology [13] indicated that patients with unifocal prostate
and radiation in treating prostate cancer. A recent article by cancer could be successfully identified and treated with
Katz and Rewcastle [20] reviewed the 5-year biochemical limited cryoablation. The additional patients and longer
disease-free survival of patients treated with brachytherapy, follow-up that we present in this paper confirm our earlier
CT conformal radiation therapy, radical prostatectomy, and results. Within the context of our mean follow-up of
cryoablation in every study published in the last 10 years. approximately 4 years, this approach has been successful in
The results were stratified based on whether the patients local cancer control and results have been equivalent to
104 G. Onik: Rationale for a Male Lumpectomy

those cited for whole-gland cryotherapy, as has been evi- of tissue and neurovascular bundle was carried out (Fig. 1).
denced by stable PSA results in 95% of patients and nega- The reward for this minimal treatment was that the patient
tive postoperative biopsies in all 19 patients. One patient has did not need a Foley catheter postoperatively and was sex-
had cancer demonstrated by spectroscopic MRI in the pre- ually active 1 week after the procedure. The remainder of
viously unfrozen region. He has been successfully treated the patients had half of the gland frozen.
with repeat cryoablation. Additionally 1 patient has an Lastly, known areas of potential extracapsular extension
unstable PSA with negative biopsies in both frozen and can be prophylactically treated by extending the freezing to
unfrozen regions and presumably has micro-metastatic dis- encompass the peri-prostatic tissue in the area of the neu-
ease. These results are particularly impressive in that 10 of rovascular bundle and seminal vesicles. This locally
the 21 patients reported fell into the medium- and high-risk extensive freezing, which makes this procedure a unique
categories for cancer recurrence. This is a very different combination of an aggressive yet conservative treatment, is
situation from that regarding brachytherapy, the other facilitated by the technical advance of injecting saline into
minimally invasive procedure for prostate cancer, in which Denonvilliers fascia to separate the rectum from the pros-
medium- and high-risk patients are no longer treated with tate. Increasing this space eliminates the chance for rectal
brachytherapy alone due to high local recurrence rates. It is damage, allowing greater freezing into extracapsular re-
also of great significance in that unlike in breast cancer, gions.
where adjuvant radiation therapy is considered a require- The two major treatment side effects that prostate cancer
ment in lumpectomy patients, thus adding to the compli- patients fear the most are impotence and incontinence.
cations and the cost of treatment, our results were obtained Certainly, any minimally invasive prostate cancer treatment
without any additional adjuvant therapies. However, the must minimize the incidence of these complications if it is
need for adjuvant radiation therapy, perhaps at a lower dose to claim an advantage over the present whole-gland treat-
than primary radiation therapy, will have to be assessed in ments. The procedure we describe appears to fulfill the goal
the future. of a lumpectomy-type procedure by having extremely low
One of the difficulties of our treatment approach is morbidity. None of our patients had significant long-term
defining a successful result from the cancer recurrence point incontinence after the procedure, with all but 1 patient being
of view since the usual criterion of a negligible PSA level fully continent immediately. One patient, who had a previ-
used for a successful result following radical prostatectomy ous TURP and was on chronic steroids for asthma, had a
is not valid. In this procedure variable amounts of prostatic focal cryoablation in an attempt to decrease the incidence of
tissue on the side opposite the tumor are knowingly left urinary complications. He had mild stress incontinence after
untreated. Depending on the degree of tissue untreated we a TURP to remove sloughed tissue that eventually resolved.
expect to see a postoperative PSA reading above 0.2 ng/ml. Even in total gland cryosurgical ablation, incontinence is
Due to the retrospective nature of the study and the use of seen in less than 2% of patients [8]. Incontinence, with our
preoperative hormone therapy which artificially decreases more minimal cryosurgical approach, would be expected to
both prostate volume and preoperative PSA, correlation of be negligible since only a portion of each sphincter has a
the gland volume with preoperative and postoperative PSA potential to be damaged. We see this as a positive secondary
was not recorded, and the potential success of the procedure effect of our attempt to improve potency, as nerve-sparing
correlated with postoperative volume and serum PSA levels radical prostatectomy can still have incontinence rates as
was not analyzed. As in patients without prostate cancer, high as 6%, with many other patients gaining urinary control
however, we would expect PSA stability, i.e., no rise in PSA only after an extended period of time [4].
over time in patients adequately treated. This criterion of The preservation of potency associated with focal cryo-
PSA stability coupled with negative biopsy results is con- ablation is better than we expected. Of the 20 patients who
sistent with common sense clinical practice and resulted in could be evaluated for potency, 17 remained potent (85%)
19 of our 21 patients needing no further investigation or and satisfied with their sexual functioning. Nonetheless, this
treatment for a cancer recurrence. being a retrospective study, without the use of standard
These encouraging results can in part be attributed, we sexual functioning questionnaires, investigator bias, as well
believe, to one of the strengths of the procedure, which is as patient inclination to please the treating physician, should
the ability to tailor the tissue destruction on the basis of the be considered a possible factor affecting the results reported.
threat of extracapsular extension and the risk of disease in In order to at least account for the vagaries of potency we
the opposite prostate lobe. Four of our patients considered to included in our success criteria patient satisfaction with his
be at high risk for multifocal disease had bilateral prostate sexual functioning. Two patients required Viagra postoper-
freezing, with only neurovascular bundle sparing on the side atively to meet our success definition, while 1 patient using
opposite the cancer. On the other hand, one patient had a Viagra preoperatively continued its use postoperatively. To
well-defined (on both MRI and US) 4 mm tumor of Gleason avoid local recurrences only one neurovascular bundle was
grade 3. Biopsies surrounding the visible abnormality spared in all our patients, making these results particularly
showed no cancer and, therefore, a minimal, focal procedure interesting. The literature on nerve-sparing radical prosta-
directed only to the visible tumor and a surrounding margin tectomy shows a significant decrease in potency rates when
G. Onik: Rationale for a Male Lumpectomy 105

one neurovascular bundle is spared rather than two [4], with long-term efficacy in treating prostate cancer. The important
usual potency rates for one bundle in the range of 30%. concept, however, is not which ablation technology is used
Achieving high potency rates with our unilateral nerve but that a population of prostate cancer patients can be
sparing, without risking a positive margin on the tumor side, identified and successfully treated with a lumpectomy ap-
is an advantage of this procedure compared with bilateral proach. Undoubtedly if long-term viability of such a treat-
nerve-sparing radical prostatectomy. ment is demonstrated then other forms of thermal ablation
These superior results for cryoablation in unilateral will be attempted such as radiofrequency, microwave, high-
nerve-sparing may be explained by cryoablations minimal intensity focused ultrasound and laser ablation [2326], to
vascular disruption and the lack of nerve manipulation and accomplish the same end. A less satisfactory approach we
trauma compared with radical prostatectomy. believe would be to attempt a focal treatment with radiation
Another difference in our results compared with nerve- therapy, since the lack of real-time feedback to guide ther-
sparing radical prostatectomy appears to be in the rate of apy, the limitations of dose threshold, and the inherent
potency return between the two procedures. Potency rates in nature of radiation scatter make it a less than optimal
nerve-sparing radical prostatectomy procedures are often modality for this purpose.
reported at 18 months after the procedure. In our series the In conclusion, our preliminary results show that a pro-
return to function was relatively rapid. All patients had re- cedure which targets the cancerous portion of the prostate
turn to sexual functioning within 1 year of the procedure. gland (i.e., a male lumpectomy) can provide good cancer
The rapidity of return seemed in part related to whether the control while limiting patient morbidity. Further prospective
patient had been on preoperative hormone therapy, as well studies, with larger numbers of patients and longer follow-
as the extent of freezing that was carried out. up, is needed to evaluate the full implications of such a
Nerve-sparing cryoablation seems to have advantages treatment approach. At a time when patients must choose
over brachytherapy and external beam radiotherapy, as well. between watchful waiting or high-morbidity whole-gland
Unlike brachytherapy, which is limited to patients with low- treatments, a lumpectomy treatment would be a welcome
volume disease of low Gleason grade, our procedure is addition to the treatment options for the male cancer pop-
limited by whether the disease is confined to one side of the ulation.
gland, rather than other clinical parameters. Based on their
Gleason grade, PSA levels, and extent of disease, nearly half References
the patients we treated would not have been candidates for 1. Messing EM, Thompson I (2003) Follow-up of conservatively man-
brachytherapy alone. aged prostate cancer: Watchful waiting and primary hormonal therapy.
Urol Clin North Am 30:687702
Unfortunately, radiation therapy does not appear to 2. Catalona WJ, Basler JW (1993) Return of erections and urinary con-
maintain its initial potency advantage over the long term. tinence following nerve sparing radical retropubic prostatectomy. J
Potency rates after 2 years are essentially equivalent to those Urol 150:905907
3. Sanchez-Ortiz RF, Broderick GA, Rovner ES, et al. (2000) Erectile
after nerve-sparing radical prostatectomy [3]. Brachythera- function and quality of life after interstitial radiation therapy for
pys urinary tract complications can have a significant effect prostate cancer. Int J Impot Res 12 (Suppl 3):S18S24
on patient lifestyle [5] and rectal complications, which are a 4. Talcott JA, Reiker P, Propert KJ, et al. (1989) Patient reported impo-
major concern with radiation therapy, have been virtually tence and incontinence after nerve-sparing radical prostatectomy. J
Natl Cancer Inst 16:11171123
eliminated in our procedure by separation of the rectum and 5. Zelefsky MJ, Hollister T, Raben, et al. (2000) Five year biochemical
prostate with saline injection into Denonvilliers fascia prior outcome and toxicity with transperineal CT-planned permanent I-125
to freezing. In addition, brachytherapy patients who fail prostate implantation for patients with localized prostate cancer. Int J
Radiat Oncol Biol Phys 47:12611266
locally have limited curative options available. Lastly, a 6. Bagley GP (1999) Cryosurgical ablation of the prostate. Decision
major drawback to radiation therapy is that patients who fail memorandum 2/01/99. Issue #CAG-00031
radiation show a significant increase in Gleason grade and 7. Bahn DK, Lee F, Bandalament R, et al. (2002) Seven-year outcomes in
tumor aggressiveness in the recurrent cancer that adversely the primary treatment of prostate cancer. Urology 60(Suppl 1):311
8. Donnelly BJ, Salikan JC, Ernst DS, et al. (2002) Prospective trial of
effect patient survival [22]. Certainly this is not a favorable cryosurgical ablation of the prostate: Five year results. Urology 60:645
characteristic in a procedure being applied to possibly a 9. Djavan B, Susani M, Bursa B, et al. (1999) Predictability and signifi-
younger patient population. cance of multi-focal prostate cancer in the radical prostatectomy
specimen. Techn Urol 5:139142
Finally, a level of comfort with a focal cryoablation is
10. Villiers A, McNeal JE, Freiha FS, et al. (1992) Multiple cancers in the
possible, unlike with nerve-sparing radical prostatectomy prostate: Morphologic features of clinically recognized vs. incidental
and radiation, since cryoablation has the unique ability to tumors. Cancer 70:23122318
allow repeat treatments without added morbidity. The one 11. Rukstalis DB, Goldknopf JL, Crowley EM, et al. (2002) Prostate
cryoablation: A scientific rationale for future modifications. Urology
patient in whom disease was discovered on the side opposite 60:1925
the focal cryoablation treatment had a repeat treatment on 12. Noguchi M, Stamey TA, McNeal JE, et al. (2003) Prognostic factors
the other half of his gland with no complications and a for multifocal prostate cancer in radical prostatectomy specimens: Lack
of significance of secondary cancers. J Urol 170:459463
negligible postoperative PSA level.
13. Onik G, Narayan P, Vaughan D, et al. (2002) Focal nerve sparing
At present we are using cryoablation to carry out this cryoablation for the treatment of primary prostate cancer: A new ap-
cancer-targeted treatment since it has the demonstrated proach to preserving potency. Urology 60:109114
106 G. Onik: Rationale for a Male Lumpectomy

14. Onik GM, Cohen JK, Reyes GD, et al. (1993) Transrectal ultrasound- 20. Katz A, Rewcastle JC (2003) The current and potential role of cryo-
guided percutaneous radical cryosurgical ablation of the prostate. ablation as a primary treatment for prostate cancer. Curr Oncol Rep
Cancer 72:12911299 5:231238
15. Santiago RJ, Wu L, Harris E, Fox K, Schultz D, Glick J, Solin LJ 21. Gould RS (1999) Total cryoablation of the prostate versus standard
(2004) Fifteen-year results of breast-conserving surgery and definitive cryoablation versus radical prostatectomy: Comparison of early results
irradiation for Stage I and II breast carcinoma: The University of and the role of trans-urethral resection in cryoablation. J Urol
Pennsylvania experience. Int J Radiat Oncol Biol Phys 58:233 162:16531657
240 22. OSullivan DC, Barrett DM, Colby TV, et al. (1992) Effect of external
16. Brassel SA, Rosner IL, Mcleod DG (2005) Update on magnetic reso- beam radiation therapy on prostatic carcinoma DNA content as mea-
nance imaging, prostascint and novel imaging in prostate cancer. Curr sured by static image cytometry. Eur Urol 21:235239
Opin Urol 15:233240 23. Zlotta AR, Djavan B, Matos C, et al. (1998) Percutaneous transperineal
17. Levine MA, Ittman M, Melamed J (1998) Two consecutive sets of radiofrequency ablation of the prostate tumour: Safety, feasibility, and
transrectal ultrasound guided sextant biopsies of the prostate for the pathological effects on human prostate. Br J Urol 81:265275
detection of prostate cancer. J Urol 159:471475 24. Chen JC, Moriarty JA, Derbyshire JA (2000) Prostate cancer: MR
18. Graefen M, Hammerer P, Michl U, et al. (1998) Incidence of positive imaging and thermometry during microwave thermal ablation-initial
surgical margins after biopsy-selected nerve sparing radical prostatec- experience. Radiology 214(1):290297
tomy. Urology 51:437442 25. Boni RA, Sulser T, Jochum W (1997) Laser ablation indices changes in
19. Onik GM (2004) 3D global mapping biopsies: A more efficacious the prostate: Findings at endorectal imaging with histologic correlation.
method of determining the extent of prostate cancer. Presented at Radiology 202:232236
Society of Uroradiology annual meeting, Scottsdale, AZ, March 7, 26. Uchida T (2005) High intensity focused ultrasound for localized
2004 prostate cancer. Nippon Rinsho 63:345349
Cardiovasc Intervent Radiol (2008) 31:107115
DOI 10.1007/s00270-007-9208-1

CLINICAL INVESTIGATION

Factors Limiting Complete Tumor Ablation by Radiofrequency


Ablation
Erwan Paulet Christophe Aube Patrick Pessaux Jerome Lebigot
Emilie Lhermitte Frederic Oberti Anne Ponthieux Paul Cales
Catherine Ridereau-Zins Philippe L. Pereira

Received: 22 February 2007 / Accepted: 25 September 2007 / Published online: 30 October 2007
Springer Science+Business Media, LLC 2007

Abstract The purpose of this study was to determine lesions were significantly different for the recurrence and
radiological or physical factors to predict the risk of residual mass groups compared to the recurrence-free
residual mass or local recurrence of primary and secondary group (p \ 0.05). The presence of an enhanced ring on the
hepatic tumors treated by radiofrequency ablation (RFA). magnetic resonance control was more frequent in the
Eighty-two patients, with 146 lesions (80 hepatocellular recurrence and residual mass groups. In the group of
carcinomas, 66 metastases), were treated by RFA. Mor- patients with residual lesions, analysis of physical param-
phological parameters of the lesions included size, eters showed a significant increase (p \ 0.05) in the time
location, number, ultrasound echogenicity, computed necessary for the temperature to rise. In conclusion, this
tomography density, and magnetic resonance signal study confirms risk factors of recurrence such as the size of
intensity were obtained before and after treatment. the tumor and emphasizes other factors such as a post-
Parameters of the generator were recorded during radio- treatment enhanced ring and an increase in the time
frequency application. The recurrence-free group was necessary for the rise in temperature. These factors should
statistically compared to the recurrence and residual mass be taken into consideration when performing RFA and
groups on all these parameters. Twenty residual masses during follow-up.
were detected. Twenty-nine lesions recurred after a mean
follow-up of 18 months. Size was a predictive parameter. Keywords Liver tumor  Radiofrequency ablation 
Patients sex and age and the echogenicity and density of Magnetic resonance imaging  Recurrence

E. Paulet (&)  C. Aube  J. Lebigot  C. Ridereau-Zins Introduction


Department of Radiology, University Hospital Angers,
4 rue Larrey, 49933 Angers, France
e-mail: erwanpaulet@yahoo.fr Radiofrequency (RF) thermal ablation has been extensively
performed for the treatment of malignant hepatic lesions.
P. Pessaux  E. Lhermitte Although surgical resection of hepatic colorectal metasta-
Department of Visceral Surgery, University Hospital Angers,
ses (CRM) and hepatocellular carcinoma (HCC) is still
49933 Angers, France
considered the best hope for cure, it is important to
F. Oberti  P. Cales remember that only 30% of patients with HCC and 20%
Department of Hepato-Gastroenterology, University Hospital with CRM are qualified for surgical resection [1, 2].
Angers, 49933 Angers, France
Although liver transplantation gives the best result for the
A. Ponthieux treatment of HCC, only 3% of patients with HCC can be
Clinical Research Center, University Hospital Angers, transplanted and there is a need for a bridge therapy until
49933 Angers, France transplantation [3].
Several studies have shown that RF ablation (RFA) is a
P. L. Pereira
Department of Diagnostic Radiology, Eberhard-Karls simple, effective, and less expensive technique with a low
University, Tubingen, Germany morbidity compared with surgical treatment [49].

123
108 E. Paulet et al.: Factors Limiting Complete Tumor Ablation by RFA

However, one of the major challenges with this technique diameter) or XL starbust (nine expandable electrodes; 5-cm
is residual tumor tissue and local recurrence after local maximal diameter) electrodes was used. The power deliv-
treatment. Improved survival in patients treated with RFA ered by this generator is permanently modulated according
could be obtained by early detection of these recurrences. to the temperature recorded by thermocouples located at
Patients could then be treated again before the size or shape the end of five of the seven or nine expandable electrodes
of the lesion made RFA impossible [10, 11]. and by impedance. Following the RITA protocol the pro-
Several studies have reported factors influencing local cedure was begun with a 2-cm deployment of the
recurrence of hepatic lesions treated by RFA such as size, expandable electrode using a maximal power 80 W, until a
histology, and proximity of large vessel. Pre- and post- temperature of 104C was reached. Then the electrodes
therapeutic patterns of these lesions have never been were deployed centimeter by centimeter, increasing pro-
considered as factors predictive of recurrence. gressively the power. The number of steps depended on the
The aim of this study was to analyze the patterns of the final diameter wanted. The last step consisted of 7 to 14
lesions before and after treatment and to record physical min at maximal power (150 W). In addition to the venders
parameters during RFA to determine potential factors protocol, the last step was systematically doubled.
predictive of recurrence of lesions treated by RFA.

Parameters Measured
Patients and Methods
Before treatment, the following radiological patterns were
Patients noted by only one radiologist (E.P.) for each lesion: size,
contour, proximity (\3 mm) to vessels (portal vein and
Between January 2002 and June 2005, 82 patients with 146 hepatic veins), US echogenicity, CT density, MR signal
lesions were treated with percutaneous RF at our institu- intensity on T1, T2, out-of-phase weighted sequences, and
tion. Mean age was 65 years (range, 2583 years). There enhancement of the lesion with iodine contrast medium or
were 75% men and 25% women; 57% (n = 47) of patients gadolinium. The patients were examined before treatment
had cirrhosis and the etiology of cirrhosis was alcohol in by MR (n = 47) and/or CT (n = 44) and/or US (n = 68).
85% and viral in 15%. The size of each lesion was noted pre- and posttreatment.
The diagnosis was obtained either by biopsy or on the To take into account the shape, we measured the surface of
basis of clinical, biological, and radiological results. Fifty- the lesions [S = p 9 (D1 9 D2)/4] and calculated the dif-
five percent (n = 80) of the 146 treated lesions, were pri- ference in the surfaces before and after treatment.
mary hepatic tumors, and 45% metastases (n = 66). The technical data of the generator were recorded during
Ninety-eight percent (n = 79) of the primitive lesions were the procedure for 83 of the 146 treated lesions. For each
HCC and 2% cholangiocarcinoma (n = 1). The origin of lesion we calculated the time for an increase in temperature
the metastasis was colorectal cancer in 77% of the cases and the difference between the desired and the obtained
(n = 51), neuroendocrine carcinoma in 12% (n = 8), epi- temperature during treatment (Fig. 1). We also calculated
dermoid carcinoma in 9% (n = 6), and pancreatic the duration of treatment at the optimal temperature, the
adenocarcinoma in 2% (n = 1). standard deviation between the temperature of each ther-
mocouple, and the average power delivered by the
generator during RF application.
RF Technique Follow-up was performed by MRI (Echo Speed Signa
and Excites; Medical GE Systems; Fremont, CA, USA) at
RF was performed by two radiologists (C.A., J.L.), by 6 weeks, 3, 6, and 9 months, then every 6 months after
percutaneous (n = 97) or surgical route (n = 49), with treatment. The radiological patterns of the treated lesions
ultrasound (US; n = 141) or computed tomography (CT; were obtained during the first control at 6 weeks. A ring of
n = 5) guidance. Procedures were performed under general enhancement present around the ablation zone was classi-
anesthesia during surgery and under neuroleptanalgesia or fied as thin (B2 mm) or thick ([2 mm).
general anesthesia by percutaneous route. Surgical proce- A residual mass was defined as a remaining mass of
dures were conducted by the same operator (C.A., J.L.) pathological tissue or a nodular or peripheral enhancement
after the surgeon had exposed the liver. If surgical resec- thicker than 5 mm. A local recurrence was defined as an
tion was associated, it was performed either before or after increase in the surface of a treated lesion or as the occur-
the RF procedure. A RITA 1500X generator (RITA Med- rence of nodular or peripheral enhancement or the presence
ical System, Inc., Mountain View, CA), with expandable of tissue nodules, enhanced or not, in contact with the
Starbust (seven expandable electrodes; 3-cm maximal ablation zone. The development of new lesions at a

123
E. Paulet et al.: Factors Limiting Complete Tumor Ablation by RFA 109

tumors. Local recurrence occurred within an average of 9


months (1 to 30 months). Recurrence was statistically more
frequent in women (p = 0.009), based on a multivariate
odds ratio (OR) analysis (OR = 0.39; 95% confidence
interval [CI] = [0.16, 0.92]). Age and the presence of cir-
rhosis, or its etiology, did not significantly affect
recurrence.
The frequencies of metastases and HCC were identical
in the recurrence-free and recurrence groups. But there was
a tendency between the recurrence-free and the residual
groups regarding histology (68% of HCC in the residual
group and 54% in the recurrence-free group).
There was no significant difference between patients
treated by percutaneous route and those treated by surgical
route for the occurrence of residual tumors or recurrence.

Lesion Size (Tables 1 and 2)

The size of the treated lesions was larger, and the differ-
Fig. 1 Graph 1 shows the temperatures recorded at the tips of the ence in the lesion surfaces pre- versus posttreatment were
electrodes; Graph 2, the recorded power and impedance during RF greater, in the residual and recurrence groups (p \ 0.001).
application. In graph 1, part 1 is the rise in temperature (37 to 105 or
110C according to the protocol) and part 2 is considered the time of
efficient treatment. 80 9 87 mm (300 9 300 DPI)
Lesion Patterns Before Treatment (Table 3)
distance from the ablation zone was not considered a
recurrence [12, 13]. Residual tumors were more frequent when lesions were
not well circumscribed, but this was not significant
(p = 0.06). There was no significant difference in recur-
Statistical Analysis rence between lesions close to (\3 mm) and lesions
farther from large vessels (portal vein and hepatic veins)
Statistical analyses were performed with the SPSS software and vascular structures. Hyperechoic lesions (p = 0.04)
version 12.0 (SPSS Inc). Chi-square test (or Fisher test) and and lesions with T2W hyperintensity (p = 0.06) were
Mann-Whitney test were performed to test the difference more frequent in the residual tumor group. Isoattenuated
between the metastasis and the HCC groups, between the lesions (p = 0.001) and lesions with T2W isointensity
recurrence and the recurrence-free groups, and between the (p = 0.06) were more frequent in the recurrence-free
recurrence-free and the residual groups. Multiple regres- group. Among the HCCs, an enhancing border was more
sion analysis was used to identify determinants of frequent in the residual group, but this was not significant
recurrence and residual mass such as age, sex, contours (p = 0.07).
(regular/irregular), ring of enhancement (absent/thin/thick),
T2 (iso/hyper), and rise in temperature. Significance level
was taken as p \ 0.05. Lesion Patterns on MRI Performed 6 Weeks After
Treatment (Table 4)
Results
Six weeks after treatment, metastases were more frequently
Results are reported in Tables 1 to 5 and shown in Figs. 24. found in hypointense T2W, and HCC in hypointense T1W
(p = 0.001). However, there was no statistical difference in
intensity between the recurrence-free group and the
Population Characteristics (Table 1) recurrence or residual tumor group.
A ring of enhancement was significantly less frequent in
Among the 82 treated patients, 19 (23%) had recurrence the recurrence-free group than in the recurrence
and 11 (13%) had residual tumors. Of 146 treated lesions, (p = 0.002) and residual tumor (p = 0.001) groups. This
29 (20%) had recurrence and 20 (13.7%) had residual ring was either thin or thick. In multivariate analysis, the

123
110 E. Paulet et al.: Factors Limiting Complete Tumor Ablation by RFA

Table 1 Comparison of the metastasis and HCC groups regarding age, sex, cirrhosis, and size of the lesion and comparison among the three
groupsrecurrence-free, recurrence, and residual tumorregarding age, sex, cirrhosis, histology and maximum diameter of the lesiona
Metastasis HCC pb Recurrence-free Recurrence Residual pc pd

Age (yr) 64 : 14 67 : 8 64 : 10 67 : 11 68 : 17  **
Gender (%)
Male 50 90 *** 78 52 65 **
Female 50 10 *** 22 48 35 **
Cirrhosis (%) 11 89 *** 55 41 58
Histology (%)
Metastasis 46 52 32
HCC 54 48 68
Maximum diameter (mm) 21.8 : 10.8 23.3 : 10 20 : 9 25 : 12 35 : 15 *** ***
a
Arithmetic mean : SD or percentage
b
Comparison between metastasis group and HCC group using Mann-Whitney test or chi-square test. ***p \ 0.001; , not significant
c
Comparison between recurrence-free group and recurrence group using Mann-Whitney test or chi-square test. 0.05 B p \ 0.1; **p \ 0.01;
***p \ 0.001; , not significant
d
Comparison between recurrence-free group and residual group using Mann-Whitney test or chi-square test. **p \ 0.01; ***p \ 0.001; , not
significant

presence of a thick ring was a risk factor for residual


tumors (OR = 0.018; CI = [0.001, 0.234]).

Physical Parameters During Treatment (Table 5)

None of the recorded electrical parameters were statisti-


cally significantly different between the metastasis and the
HCC groups, or between the recurrence-free and the
recurrence groups, except for the time for increase in
temperature. This was significantly longer in the residual
than in the recurrence-free group (565 : 423 and 366 :
233 s, respectively; p = 0.036). However, in multivariate
analysis the OR was not significant.

Discussion

In this study recurrence and residual tumor rates (20% and


13.7%, respectively) were similar to those described in the
literature, which ranged from 2% to 47% [1419] and from
5% to 12% [3, 20, 22, 23]. The fairly high rate of recurrence
in the present study is probably due to the inclusion of pri-
mary hepatic tumors and metastatic lesions; the studies with
the best results have generally included only HCC, which is
known to respond better to RF treatment [20, 21]. Moreover,
our study had a long follow-up (18 : 12 months) to detect
late recurrence [5]. Thus, two lesions recurred after 18
months (18 and 30 months). The learning curve of this
Fig. 2 (A) Enhanced CT at the portal phase shows a metastasis of a
technique and the recent clarification of RF indications
colonic adenocarcinoma. (B) Recurrence 5 months after RFA with a
peripheral tissue enhanced after gadolinium injection on T1-weighted undoubtedly played a role in these results. Finally, follow-
gradient-echo images up by MRI could also explain this rate of recurrence because

123
E. Paulet et al.: Factors Limiting Complete Tumor Ablation by RFA 111

surgery) and different guidance modes (US vs. CT) could


influence the ablation and thus could be a methological
problem. However, as all ablations were performed by only
two operators, the above-mentioned factors are systemic
mistakes which probably do not influence the results.
In the literature there are not many well-identified fac-
tors predictive of the efficacy of RF treatment. The main
predictive factor of recurrence or residual tumors is the size
of the lesions [19, 2630], with a limit of 3 cm for optimal
treatment [19]. A second factor is the heat sink effect [5,
29, 30], and a third is the difference in electric conductivity
between the tumor and surrounding tissue [31] which
probably explains, like the oven effect, the better
effectiveness of RF ablation in HCC [4, 3237].
Our series reconfirms the size of the lesion as the main
predictive factor of recurrence; the rate of recurrence was
18% for lesions B3 cm and 28% for lesions [3 cm.
However, we can also discuss new predictive factors and
findings in this study. The presence of a thick ring of
enhancement was more frequent in the recurrence and
residual groups. But recurrences occurred in 34.5% of
patients with a thin ring, therefore they occurred in only
5% of patient without any ring of enhancement. Thus the
existence of a thin ring, which is usually explained as a
benign physiological response to thermal injury, consti-
tuted a risk factor for recurrence. It should be noted that
although the absence of a ring of enhancement always
means that there is no residual tumor, it does not mean that
there will be no recurrence, whatever the type of lesion.
PET-CT does not seem to distinguish a benign physio-
logical response from recurrence in the first posttreatment
months [14]. Recurrence or residual tissue is thought to
have the same pattern as the original tumor [11, 38].
Fig. 3 (A) Patient treated by RFA for HCC on alcoholic cirrhosis. However, in our study 2 of 10 hypovascular metastases had
The 6-week control MRI shows a well-delimited ablation zone with
hypointensity on the gadolinium-enhanced T1-weighted gradient- a thick ring and 5 had a thin ring of enhancement. More-
echo image. (B) One hundred fifty days afterward, the recurrence over, in the residual group, five hypovascular metastases
appears like a peripheral nodular enhancement had nodular or peripheral irregular enhancement on MRI
follow-up. These results may indicate that the initial
most published studies use CT. Indeed MRI seems to be reactive hyperemia due to thermal injury could lead to
more sensitive than CT for detecting recurrence [3, 24, 25]. hypervascularization of the residual tumor or promote
In our series the different access routes (percutaneous vs. implantation of new metastatic cells.

Fig. 4 (A) T2-weighted MR


image. (B) T1-weighted MR
image with gadolinium at the
arterial phase. (C) T1-weighted
MR image with gadolinium at
the portal phase. Patient treated
for HCC on alcoholic cirrhosis.
The 6-week MR control shows a
residual tumor enhanced in the
arterial phase (arrow) in the
anterior part of the ablation
area. 178 9 64 mm (72 9 72
DPI)

123
112 E. Paulet et al.: Factors Limiting Complete Tumor Ablation by RFA

Certain traditional factors favoring recurrence found in 3. In our study the proximity of a vascular structure does
the literature were not found in our study. not appear to be a factor predictive of treatment
failure. Since the heat sink effect is well known, these
1. Influence of histological type. This may be in part lesions were treated more intensively. This is sup-
because there were too few patients to show this ported in our series by the use of larger electrodes for
difference. It could also be due to the characteristics of equivalent-sized lesions if they were close to vessels.
our population. In our study the etiology of cirrhosis
Two factors of recurrence related to population character-
was alcoholic in 85% of cases, whereas in most studies
istics were not explained. First, recurrence was more
the etiology of cirrhosis is viral, which emphasizes that
frequent in women (p = 0.009). It is true that there were
HCC responds better to RFA. But the assumption of a
more women in the metastasis group (p = 0.001), but
different electrical conduction between the liver for
unlike in the literature, this histological factor was not, in
these two etiologies needs to be demonstrated.
our series, a significant factor of recurrence. We have not
2. The lack of difference between the percutaneous and
established any other correlation between factor of recur-
the surgical route may be because the RF procedure
rence and gender. Thus we have no hypothesis to explain
was always performed by the same two operators,
this finding. Second, patients were significantly older in the
whatever the route.
recurrence and residual groups (p = 0.068 and p = 0.009,
respectively), although the size of lesions was not corre-
Table 2 Size of lesions by group: recurrence-free, recurrence, and lated with age.
residual The pattern of lesions before treatment as a risk factor
Group Maximum diameter, Lesions Lesions for residual tumors or recurrence is probably related to the
average (cm) B3 cm (%) [3 cm (%) pathology. Lesions that were hyperechoic on US or heter-
Recurrence-free 20 73 38 ogeneous on CT were significantly more frequent in the
Recurrence 25 18 28 residual group (p = 0.04 and p = 0.001, respectively). This
Residual 35 9 34
could be because these lesions were bulkier than hypo- or
isoechoic lesions (p = 0.03) and it is well known that

Table 3 Comparison among the three groupsrecurrence-free, recurrence, and residualregarding contours, vessel proximity, echogenicity,
CT density, and MR signal intensity in T1, T2, and outphase weighted images recorded on preablation examination
Recurrence-freea Recurrencea Residuala p1b p2c

Contours % Irregular 18 (17) 21 (6) 40 (8) 


Vessels % Close 10 (9) 17 (5) 20 (4)
Echogenicity Iso 19 (14) 7 (2) 0 (0) **
CT density Hypo 71 (45) 67 (8) 71(5) ***
MR signal intensity T2 Hyper 49 (23) 58 (7) 67 (8) 
Enhancement Absent 46 (44) 34 (10) 25 (5) 
a
Percentage (effective)
b
Comparison between recurrence-free group and recurrence group using Mann-Whitney test or chi-square test. , not significant
c
Comparison between recurrence-free group and residual group using Mann-Whitney test or chi-square test. , not significant; 0.05 B
p \ 0.1; **p \ 0.05; ***p \ 0.001

Table 4 Comparison among the three groupsrecurrence-free, recurrence, and residualfor the parameters of MR signal intensity in T1- and
T2-weighted images and the presence of a ring of enhancement on MR control at 6 weeksa
Recurrence-free (%) Recurrence (%) Residual (%) p1b p2c

T2 Iso/hyper/ hypo 14(8)/37(21)/49(28) 13(2)/20(3)/67(10) 10(1)/30(3)/60 (6)


T1 Iso/hyper/hypo 4(2)/32(19)/64(38) 0(0)/40(6)/60(9) 0(0)/42(5)/58(7)
Ring of enhancement Absent/thin/thick 51(47)/42(39)/7(6) 34.5(10)/34.5(10)/31(9) 0(0)/5(1)/95(19) ** ***
a
Percentage (effective)
b
Comparison between recurrence-free group and recurrence group using Mann-Whitney test or chi-square test. , not significant; **p \ 0.002
c
Comparison between recurrence-free group and residual group using Mann-Whitney test or chi-square test. , not significant; ***p \ 0.001

123
E. Paulet et al.: Factors Limiting Complete Tumor Ablation by RFA 113

Table 5 Comparison among recurrence, residual, and recurrence-free group regarding physical parameters (rise in temperature, difference in
temperature between desired temperature and obtained temperature, standard deviation of temperature curves, average power)a
Recurrence-free Recurrence Residual p1b p2c

Rise in temp (s) 366 : 233 417 : 221 565 : 423 


Time at optimal temp (s) 629 : 336 909 : 586 780 : 520
Variation type enters curves (C) 7.7 8.7 6.0
Difference between desired & obtained temp (C) 9.8 : 13 9.9 : 15 14.4 : 30
Power (W) 560 : 268 495 : 228 731 : 390
a
Arithmetic mean : SD or percentage
b
Comparison between recurrence-free group and recurrence group using Mann-Whitney test or chi-square test. , not significant
c
Comparison between recurrence-free group and residual group using Mann-Whitney test or chi-square test. , not significant; 0.05 B p\0.1

HCCs that are hypoechoic when they are small become Conclusion
hyperechoic (due to their keratinization) when they grow
[3941]. In the same way, residual tumors were more If factors predictive of recurrence or residual tumors can be
frequently (p = 0.06) hyperintense in T2W lesions in our established for lesions treated by RFA, more intensive
series. This hyperintensity is related to poorly differenti- treatment of these lesions will be possible. In the same
ated HCCs, which are known to have a higher rate of way, residual tumors or recurrences can be detected and
recurrence [43]. treated earlier if patients with these factors are monitored
The analysis of the physical parameters was somewhat more closely. Thus well-known predictive factors such as
disappointing. Indeed, the only identified risk factor for a the size of the lesions, the heat sink effect, and differences
residual tumor was a long time for the rise in temperature. in electric conductivity between the tumor and surrounding
This could be because of poor electrical conduction or a tissue should be taken into account when RFA is performed
small difference in electrical conduction between the lesion and during follow-up of these patients.
and the surrounding tissue [31]. The heterogeneity of the Our series revealed new factors. The presence of a ring
temperatures at the end of the expandable electrodes, as of enhancement on an MRI performed 6 weeks after
well as the difference between the desired and the obtained treatment is a predictive factor for recurrence, even if the
temperatures, was not significantly different between the ring is thin. Reactive hyperemia and post-RFA inflamma-
groups. This was probably because even if the optimal tion, which is normally considered to be a benign
temperatures were not always reached, a temperature of physiological response to thermal injury, could play a role
60C applied for 5 min is enough to obtain effective in recurrence. This hypothesis could be the subject of
coagulation [43, 44]. future studies. In clinical practice, lesions which show a
Thus size is still the major limitation to the effectiveness ring of enhancement on post-RFA MRI should be closely
of RFA. For lesions larger than 3 cm several applications or followed up. The age and gender of the patients are more
adjuvant therapies should be considered to increase the surprising predictive factors because we cannot explain
diameter of coagulation: hyperthermic cellular sensitiza- them, and thus they should be considered with caution.
tion by chemotherapy [8, 45], vascular clamping [46], or Finally, the study of the physical parameters recorded on
cellular ischemia (chemoembolization) [47] and alteration the generator during RF application was disappointing and
of electric and thermal conductivity (instillation of saline will not help in daily practice.
injection) [48]. Recent progress in bipolar electrodes could
provide better control of the shape of large coagulations
[50]. References
Establishing the date of the first postablation control is
1. Fan ST, Lo CM, Liu CL, et al. (1999) Hepatectomy for hepato-
not clear-cut; it can vary from the first week [5, 9, 33, 50]
cellular carcinoma: toward zero hospital deaths. Ann Surg
to 6 weeks [51, 52] after RFA. But the real challenge is 229:322330
having a tool for precise monitoring so that a second RF 2. Fong Y, Sun RL, Jarnagin W, Blumgart LH (1999) An analysis of
application can be made during the same procedure if 412 cases of hepatocellular carcinoma at a Western center. Ann
Surg 229:790799, discussion 799800
necessary. Contrast-enhanced US [5355] or, more prob-
3. Lu DS, Yu NC, Raman SS, et al. (2005) Percutaneous radiofre-
ably, MRI [56, 57] could be the most effective technique quency ablation of hepatocellular carcinoma as a bridge to liver
for monitoring for this purpose. transplantation. Hepatology 41:11301137

123
114 E. Paulet et al.: Factors Limiting Complete Tumor Ablation by RFA

4. Solbiati L, Ierace T, Tonolini M, Osti V, Cova L (2001) Radio- 24. Dromain C, de Baere T, Elias D, et al. (2002) Hepatic tumors
frequency thermal ablation of hepatic metastases. Eur J treated with percutaneous radio-frequency ablation: CT and MR
Ultrasound 13:149158 imaging follow-up. Radiology 223:255262
5. Rhim H, Goldberg SN, Dodd GD 3rd, et al. (2001) Essential 25. Mazzaferro V, Battiston C, Perrone S, et al. (2004) Radiofre-
techniques for successful radio-frequency thermal ablation of quency ablation of small hepatocellular carcinoma in cirrhotic
malignant hepatic tumors. Radiographics 21:1735 patients awaiting liver transplantation: a prospective study. Ann
6. Livraghi T (2003) Radiofrequency ablation, PEIT, and TACE for Surg 240:900909
hepatocellular carcinoma. J Hepatobil Pancreat Surg 10:6776 26. Lu DS, Yu NC, Raman SS, et al. (2005) Radiofrequency ablation
7. Lencioni R, Cioni D, Crocetti L, Bartolozzi C (2004) Percuta- of hepatocellular carcinoma: treatment success as defined by
neous ablation of hepatocellular carcinoma: state-of-the-art. Liver histologic examination of the explanted liver. Radiology
Transpl 10:S91S97 234:954960
8. Goldberg SN, Charboneau JW, Dodd GD 3rd, Dupuy DE, 27. Kim YS, Rhim H, Cho OK, Koh BH, Kim Y (2006) Intrahepatic
Gervais DA, Gillams AR, Kane RA, Lee FT Jr, Livraghi T, recurrence after percutaneous radiofrequency ablation of hepa-
McGahan JP, Rhim H, Silverman SG, Solbiati L, Vogl TJ, Wood tocellular carcinoma: analysis of the pattern and risk factors. Eur
BJ (2003) Image-guided tumor ablation: proposal for standardi- J Radiol 59(3):432441
zation of terms and reporting criteria. International Working 28. Lu DS, Yu NC, Raman SS, et al. (2005) Radiofrequency ablation
Group on Image-Guided Tumor Ablation. Radiology 228(2):335 of hepatocellular carcinoma: treatment success as defined by
345 histologic examination of the explanted liver. Radiology
9. Goldberg SN, Gazelle GS (2001) Radiofrequency tissue ablation: 234(3):954960
physical principles and techniques for increasing coagulation 29. Lu DS, Raman SS, Vodopich DJ, Wang M, Sayre J, Lassman C
necrosis. Hepatogastroenterology 48:359367 (2002) Effect of vessel size on creation of hepatic radiofrequency
10. Curley SA (2001) Radiofrequency ablation of malignant liver lesions in pigs: assessment of the heat sink effect. Am J
tumors. Oncologist 6:1423 Roentgenol 178:4751
11. McGhana JP, Dodd GD 3rd (2001) Radiofrequency ablation of 30. de Baere T (2001) Radiofrequency ablation of the liver. Am J
the liver: current status. Am J Roentgenol 176:316 Roentgenol 177:12131215
12. de Baere T, Elias D, Dromain C, et al. (2000) Radiofrequency 31. Solazzo SA, Liu Z, Lobo SM, et al. (2005) Radiofrequency
ablation of 100 hepatic metastases with a mean follow-up of more ablation: importance of background tissue electrical conductiv-
than 1 year. Am J Roentgenol 175:16191625 ityan agar phantom and computer modeling study. Radiology
13. Chopra S, Dodd GD 3rd, Chintapalli KN, Leyendecker JR, Ka- 236:495502
rahan OI, Rhim H (2001) Tumor recurrence after radiofrequency 32. Livraghi T, Goldberg SN, Lazzaroni S, Meloni F, Solbiati L,
thermal ablation of hepatic tumors: spectrum of findings on dual- Gazelle GS (1999) Small hepatocellular carcinoma: treatment
phase contrast-enhanced CT. Am J Roentgenol 177:381387 with radio-frequency ablation versus ethanol injection. Radiology
14. Veit P, Antoch G, Stergar H, Bockisch A, Forsting M, Kuehl H 210:655661
(2006) Detection of residual tumor after radiofrequency ablation 33. Solbiati L, Goldberg SN, Ierace T, et al. (1997) Hepatic metas-
of liver metastasis with dual-modality PET/CT: initial results. Eur tases: percutaneous radio-frequency ablation with cooled-tip
Radiol 16(1):8087 electrodes. Radiology 205:367373
15. Rossi S, Buscarini E, Garbagnati F, et al. (1998) Percutaneous 34. Berber E, Senagore A, Remzi F, Rogers S, Herceg N, Casto K,
treatment of small hepatic tumors by an expandable RF needle Siperstein A (2004) Laparoscopic radiofrequency ablation of
electrode. Am J Roentgenol 170:10151022 liver tumors combined with colorectal procedures. Surg Laparosc
16. Curley SA, Izzo F, Delrio P, et al. (1999) Radiofrequency abla- Endosc Percutan Tech 15:472544
tion of unresectable primary and metastatic hepatic malignancies: 35. Van Duijnhoven FH, Jansen MC, Junggeburt JM, van Hilleg-
results in 123 patients. Ann Surg 230:18 ersberg R, Rijken AM, van Coevorden F, van der Sijp JR, van
17. Curley SA, Izzo F, Ellis LM, Nicolas Vauthey J, Vallone P Gulik TM, Slooter GD, Klaase JM, Putter H, Tollenaar RA
(2000) Radiofrequency ablation of hepatocellular cancer in 110 (2006) Factors influencing the local failure rate of radiofrequency
patients with cirrhosis. Ann Surg 232:381391 ablation of colorectal liver metastases. Ann Surg Oncol 13:651
18. Livraghi T, Solbiati L (2001) [Percutaneous treatment: radiofre- 658
quency ablation of hepatic metastases in colorectal cancer]. 36. Abdalla EK, Vauthey JN, Ellis LM, et al. (2004) Recurrence and
Tumori 87:S69 outcomes following hepatic resection, radiofrequency ablation,
19. Livraghi T, Goldberg SN, Lazzaroni S, et al. (2000) Hepatocel- and combined resection/ablation for colorectal liver metastases.
lular carcinoma: radio-frequency ablation of medium and large Ann Surg 239(6):818827
lesions. Radiology 214:761768 37. Berber E, Rogers S, Siperstein A (2005) Predictors of survival
20. Siperstein A, Garland A, Engle K, et al. (2000) Laparoscopic after laparoscopic radiofrequency thermal ablation of hepatocel-
radiofrequency ablation of primary and metastatic liver tumors. lular cancer: a prospective study. Surg Endosc 19(5):710714
Technical considerations. Surg Endosc 14:400405 38. Goldberg SN, Gazelle GS, Mueller PR (2000) Thermal ablation
21. Siperstein A, Garland A, Engle K, et al. (2000) Local recurrence therapy for focal malignancy: a unified approach to underlying
after laparoscopic radiofrequency thermal ablation of hepatic principles, techniques, and diagnostic imaging guidance. Am J
tumors. Ann Surg Oncol 7:106113 Roentgenol 174:323331
22. Lencioni RA, Allgaier HP, Cioni D, et al. (2003) Small hepato- 39. Inoue T, Shiga J, Machinami R (1992) Direct correlation between
cellular carcinoma in cirrhosis: randomized comparison of radio- sonography and histology of minute malignant hepatic nodule.
frequency thermal ablation versus percutaneous ethanol injection. Pathol Res Pract 188:330339
Radiology 228:235240 40. Sheu JC, Chen DS, Sung JL, et al. (1985) Hepatocellular carci-
23. Choi D, Lim HK, Rhim H, et al. (2007) Percutaneous radiofre- noma: US evolution in the early stage. Radiology 155:463467
quency ablation for early-stage hepatocellular carcinoma as a 41. Ignee A, Weiper D, Schuessler G, Teuber G, Faust D, Dietrich
first-line treatment: long-term results and prognostic factors in a CF (2005) Sonographic characterisation of hepatocellular carci-
large single-institution series. Eur Radiol 17(3):684692 noma at time of diagnosis. Z Gastroenterol 43:289294

123
E. Paulet et al.: Factors Limiting Complete Tumor Ablation by RFA 115

42. Amano S, Ebara M, Yajima T, et al. (2003) Assessment of cancer 50. Rhim H, Dodd GD 3rd (1999) Radiofrequency thermal ablation
cell differentiation in small hepatocellular carcinoma by com- of liver tumors. J Clin Ultrasound 27:221229
puted tomography and magnetic resonance imaging. J 51. Solbiati L, Livraghi T, Goldberg SN, et al. (2001) Percutaneous
Gastroenterol Hepatol 18:273279 radio-frequency ablation of hepatic metastases from colorectal
43. Cady B, Jenkins RL, Steele GD Jr, et al. (1998) Surgical margin cancer: long-term results in 117 patients. Radiology 221:159
in hepatic resection for colorectal metastasis: a critical and 166
improvable determinant of outcome. Ann Surg 227:566571 52. Gazelle GS, Goldberg SN, Solbiati L, Livraghi T (2000) Tumor
44. Pereira PL, Trubenbach J, Schmidt D (2003) [Radiofrequency ablation with radio-frequency energy. Radiology 217:633646
ablation: basic principles, techniques and challenges]. Rofo 53. Choi D, Lim HK, Lee WJ, et al. (2004) Radiofrequency abla-
175:2027 tion of liver cancer: early evaluation of therapeutic response
45. Christophi C, Winkworth A, Muralihdaran V, Evans P (1998) with contrast-enhanced ultrasonography. Korean J Radiol 5:
The treatment of malignancy by hyperthermia. Surg Oncol 7:83 185198
90 54. Nouso K, Shiraga K, Uematsu S, Okamoto R, Harada R, Ta-
46. Denys AL, De Baere T, Mahe C, et al. (2001) Radio-frequency kayama S, Kawai W, Kimura S, Ueki T, Okano N, Nakagawa M,
tissue ablation of the liver: effects of vascular occlusion on lesion Mizuno M, Araki Y, Shiratori Y (2005) Prediction of the ablated
diameter and biliary and portal damages in a pig model. Eur area by the spread of microbubbles during radiofrequency abla-
Radiol 11:21022108 tion of hepatocellular carcinoma. Liver Int 25(5):967972
47. Yamakado K, Nakatsuka A, Ohmori S, et al. (2002) Radiofre- 55. Choi D, Lim HK, Lee WJ, Kim SH, Kim YH, Kim SH, Lim JH
quency ablation combined with chemoembolization in (2003) Early assessment of the therapeutic response to radio
hepatocellular carcinoma: treatment response based on tumor size frequency ablation for hepatocellular carcinoma: utility of gray
and morphology. J Vasc Interv Radiol 13:12251232 scale harmonic ultrasonography with a microbubble contrast
48. Livraghi T, Goldberg SN, Monti F, et al. (1997) Saline-enhanced agent. J Ultrasound Med 22(11):11631172
radio-frequency tissue ablation in the treatment of liver metas- 56. Clasen S, Boss A, Schmidt D, Fritz J, Schraml C, Claussen CD,
tases. Radiology 202:205210 Pereira PL (2006) Magnetic resonance imaging for hepatic
49. Haemmerich D, Staelin ST, Tungjitkusolmun S, Lee FT Jr, radiofrequency ablation. Eur J Radiol 59(2):140148
Mahvi DM, Webster JG (2001) Hepatic bipolar radio-frequency 57. Mahnken AH, Buecker A, Spuentrup E, et al. (2004) MR-guided
ablation between separated multiprong electrodes. IEEE Trans radiofrequency ablation of hepatic malignancies at 1.5 T: initial
Biomed Eng 48:11451152 results. J Magn Reson Imaging 19:342348

123
Cardiovasc Intervent Radiol (2008) 31:116121
DOI 10.1007/s00270-007-9220-5

CLINICAL INVESTIGATION

Temperature Measurements of the Low-Attenuation


Radiographic Ice Ball During CT-Guided Renal Cryoablation
Sompol Permpongkosol Richard E. Link
Louis R. Kavoussi Stephen B. Solomon

Received: 21 September 2006 / Accepted: 14 May 2007 / Published online: 7 November 2007
Springer Science+Business Media, LLC 2007

Abstract During renal cryoablation a low-attenuation images showed the thermocouple tip at the edge of the low-
area on CT develops around the cryoprobe. Knowledge of attenuation ice ball. At lower temperatures the tip was within
the temperature of the growing low-attenuation area can the low-attenuation ice ball. We conclude that knowledge of
guide therapy and ensure lethal temperatures. Herein, we the temperature at the ice ball edge during cryoablation can
report thermocouple results and correlating CT images be used to predict the extent of tissue necrosis and thus
during the development of the low-attenuation radio- provide an estimate of cryotherapy effectiveness during the
graphic ice ball. Five patients who underwent percutaneous procedure. Further work is necessary to establish a firm
CT-guided renal cryoablation were identified who had relationship between the thermal conditions and the zone of
thermocouples inserted and serial intraprocedural CT ima- damage.
ges that included images with thermocouple measurements
of 0 and sub-0C. Thermocouples had been percutaneously Keywords Computed tomography 
placed just beyond the edge of the tumors either to ensure Temperature determination  Ice ball edge  Cryoablation 
adequate cooling or to ensure safety to adjacent critical Renal tumor
structures. Renal cryotherapy under CT guidance produced a
growing low-attenuation area corresponding to the radio-
graphic ice ball. When the thermocouple measured 0C, CT Introduction

Image-guided thermal ablation of tumors is an emerging


S. Permpongkosol technique for small, accessible tumors [1, 2]. The ideal
James Buchanan Brady Urological Institute, Johns Hopkins ablation tool, be it heat ablation or cold ablation, would
University School of Medicine, Johns Hopkins Hospital,
allow intraprocedural monitoring to permit the user to be
Baltimore, MD, USA
certain that the targeted tissue is being adequately treated
S. Permpongkosol with appropriate temperatures. While magnetic resonance
Department of Surgery, Faculty of Medicine, Ramathibodi imaging (MRI) thermometry can provide intraprocedural
Hospital, Mahidol University, Bangkok, Thailand
monitoring, it is infrequently used due to cost, patient
R. E. Link accessibility, and instrument compatibility issues [1].
Scott Department of Urology, Baylor College of Medicine, Computed tomography (CT) and ultrasound are more com-
Houston, TX, USA monly used tools to guide thermal ablation [2, 3]. However,
while these imaging tools are helpful in guiding ablation
L. R. Kavoussi
Institute for Urology, North Shore LIJ Health System, applicator placement, they have had limited utility in pro-
Long Island, NY, USA viding important intraprocedural ablation feedback [46].
CT imaging relies on differences in density to create
S. B. Solomon (&)
clinically useful images. While tissue density changes
Department of Radiology, Memorial Sloan-Kettering Cancer
Center, 1275 York Avenue, New York, NY 10021, USA related to temperature alone may be difficult to ascertain
e-mail: Solomons@mskcc.org [7], the tissue density changes associated with solid ice ball

123
S. Permpongkosol et al.: Temperature Measurements of the Low Attenuation Radiographic Ice Ball 117

formation during cryoablation are quite dramatic. In 1998, Table 1 Demographic data of patients who underwent CT-guided
Sandison et al. [8] described the theoretical changes in percutaneous renal cryoablation
Hounsfield unit (HU) values from unfrozen to completely Characteristic Mean
frozen tissue. The temperature in the density change region
Age, yr (range) 69.4 (5877)
of the water-ice interface can be calibrated to allow iso-
therm contour overlays on the ice ball images. Clinically Gender (n), male/female 4/1
these observations have not been documented. Recently, Tumor size on CT, mm (range) 20.5 (1533)
Gupta et al. [9] demonstrated that percutaneous CT-guided Kidney side (n), right/left 3/4
cryoablation of the kidney is technically feasible. During Tumor position (n), anterior/lateral/posterior 0/0/5
these clinical studies low-attenuation CT density (0-HU) Tumor location (n)
changes related to ice ball formation were noted. Exophytic/endophytic 4/1
However, while the low-attenuation area visualized on Central/noncentral 1/4
CT appears during the ice ball formation phase of cryo-
ablation, it is unclear what temperature the actual edge of
the low-attenuation area actually represents [8]. It is criti-
cally important to know the temperature at the edge if this Continuous monitoring of heart rate, electrocardiogram,
information is to be useful in guiding and monitoring the oxygen saturation, and respiratory rate was performed. The
cryoablation procedure [10]. Reports suggest that -20C is skin overlying the lesion was marked, prepared, and draped
considered the temperature necessary for cell death in renal in normal sterile fashion. The puncture site was anesthe-
cryoablation [11]. Knowing what the temperature is at the tized with 1% lidocaine. A 3- to 4-mm incision was made
edge of the low-attenuation area is therefore important. The at the site of local anesthesia. Prior to cryoablation treat-
current report correlates CT images during low-attenuation ment, three CT-guided diagnostic biopsies were performed
ice ball development with thermocouple measurements. using an 18-gauge core biopsy needle. Under CT guidance,
a 2.4-mm cryoprobe (Endocare Inc., Irvine, CA) was
placed into the tumor. A 16-gauge thermometer (TempP-
Materials and Methods robe; Endocare Inc.) was inserted at least 10 mm away
from the cryoprobe to ensure adequate tumor ablation or
Our study received Institutional Review Board approval and ensure safety to adjacent critical structures.
clinical informed consent was obtained in all cases. Intraprocedural CT was used to confirm accurate
Guidelines regarding the standardization of terms and positioning of the cryoprobe and thermometer and to
reporting criteria for ablation technologies have been pub- monitor the freezing process. For cryoablation, an argon
lished previously and were used in reviewing this study [12]. gas-based cryosurgical unit (Endocare, Irvine, CA) was
As part of the contemporary renal cryoablation proce- used and the system has an integrated temperature mon-
dure, patients had percutaneous thermocouples placed itoring system. The patients underwent a double freeze-
when the physician was concerned about injuring nearby thaw cycle cryoablation with 12- and 10-min freeze
structures or when the physician wished to confirm ade- cycles as well as 8-min thawing protocols. Temperatures
quately low temperatures at the tumor border. This was an from the cryoprobe and thermometer probe were mea-
observational study of five patients for whom serial CT sured at various times during the freezing process. CT
images during the cryoablation were obtained. All patients images were obtained when the thermometer probe
were candidates for nephron-sparing surgery on the basis of measured roughly 10, 0, and -10C and at the end and
the tumor characteristics and/or the individual patients beginning of freezing/thawing. A variable number of
clinical situation. The exclusion criteria were the inability cryoprobes were placed directly into the renal mass
of patients to stop anticoagulation therapy before the pro- positioned under CT fluoroscopic guidance so that the
cedure or the presence of a bleeding dyscrasia. A relative zone of cryogen-induced tissue destruction would
contraindication to percutaneous cryotherapy was anterior encompass the entire mass with an ice ball diameter that
tumor location. Patients were included in this study if they was at least 1.0 cm larger than the tumor diameter. The
had a 0C CT image acquired and a sub-0C CT image location of the thermometer tip was assessed in compar-
acquired and a radiologist (S.S.) evaluated CT images. ison to the low-attenuation ice ball seen on the CT
The demographic data are listed in Table 1. After the images. After completion of the final freeze cycle, the
patients were positioned prone in the CT unit and short- probe was thawed and removed. Follow-up for each
acting intravenous sedation was administered (1 mg of patient included contrast-enhanced CT or MRI and serum
midazolam and 50 lg of fentanyl), scanning was performed creatinine at least 13 months and then every 6 months
through the abdomen to localize the kidney lesion. after the procedure.

123
118 S. Permpongkosol et al.: Temperature Measurements of the Low Attenuation Radiographic Ice Ball

Results temperature probe was far from the tumor edge. The mean
diameter of the low-attenuation area beyond the tumor in
In this observational study, five patients with enhancing the final radiographic was 20 mm (range, 10 to 35 mm).
renal masses underwent CT-guided percutaneous renal The first patient (Fig. 1) was a 64-year-old man with a
biopsy and cryoablation. Percutaneous biopsy at the time of history of right renal cell carcinoma, who had undergone a
ablation yielded four tumors of renal cell carcinoma (80%) right radical nephrectomy and adrenalectomy 3 years ear-
and one tumor of nondiagnostic preablation pathological lier. The patient had a 14-mm left anterior endophytic renal
findings. The mean ice ball size (low-attenuation area) was mass (Fig. 1a). A single 2.4-mm cryoprobe was used for
40.5 mm (range, 24 to 53 mm). A single and a couple of ablation. At 0C of the thermometer probe, the image
2.4-mm cryoprobes were used in two and three cases, showed the low-attenuation area corresponding to the
respectively. In all patients, the thermometer probes were radiographic ice ball to be 19 mm in diameter and con-
placed outside the edge of the tumors and CT images firmed that the thermometer tip was on the edge of the low-
obtained at temperatures above 0C demonstrated the attenuation area (Fig. 1b). At the end of the second freeze
thermometer probe outside of the growing low-attenuation cycle, the ice ball diameter was 24 mm. The temperature of
area. Images obtained when the temperature probes mea- the thermometer at the end of the second freeze was -
sured 0C demonstrated the edge of the growing low- 11C, and the thermocouple tip was inside the low-atten-
attenuation or dark area at the tip of the temperatures uation ice ball (Fig. 1c).The dark area was roughly 10 mm
probes. Finally, the tips of the thermometer probes were beyond the tumor edge. A 3- and 6-month follow-up con-
completely within the dark zone at the end of cryoablation trast-enhanced CT scan demonstrated no residual
therapy when the mean temperatures were -12C (range, enhancement (Fig. 1d).
-10 to -13C). Of note, this temperature did not repre- The second patient (Fig. 2) is a 58-year-old man with a
sent the temperature at the edge of the tumor, since the history of esophageal cancer, status post esophageal

Fig. 1 A 64-year-old man with a history of right renal cell carcinoma showing the thermometer tip (white arrow) at the low-attenuation
was 3-year status post right nephrectomy and adrenalectomy. The edge (black arrow) at 0C. (c) The thermocouple tip was in the low-
patient had a 1.4-cm left endophytic and central renal tumor and attenuation ice ball at -11C and was demonstrated at the end of the
underwent percutaneous CT-guided cryoablation with temperature second freeze cycle. (d) Follow-up CE-CT 6 months postcryoablation
guidance. (a) Nonenhanced axial CT in the prone position and before in the supine position demonstrated a nonenhancing area of medial
cryoablation showed the demonstrated tumor (arrow). (b) Contrast- renal parenchyma of the left kidney
enhanced (CE) CT obtained during cryoablation in the prone position

123
S. Permpongkosol et al.: Temperature Measurements of the Low Attenuation Radiographic Ice Ball 119

Fig. 2 A 57-year-old man with


a history of esophageal cancer
and bilateral renal tumors
underwent percutaneous CT-
guided cryoablation of the left
renal tumor with temperature
guidance. (a) Contrast-enhanced
(CE) CT in the supine position
before cryoablation showed an
enhanced solid renal mass
(white arrow) measuring 2.0 cm
in the posterior and lower pole
of the left. (b) Nonenhanced
axial CT in the prone position
obtained at 0C demonstrated
that the thermometer tip (white
arrow) was at the ice ball edge
(black arrow). (c) At the end of
cryoablation, the thermocouple
tip (white arrow) was in the
low-attenuation ice ball (black
arrow) at -14C. (d) Three-
month follow-up CE-CT after
cryoablation showed
postablation changes and no
evidence of residual tumor
(arrow)

resection with a gastric pull-through. The patient had Of the three remaining cases, the thermometer tip
bilateral renal masses. The left mass was a lower-pole positioned at the radiographic ice ball edge was demon-
exophytic 20-mm lesion (Fig. 2a). Two 2.4-mm cryop- strated on CT in the same fashion at the time the
robes were used simultaneously to perform cryoablation. temperature reading was 0C. Table 2 demonstrates the
At 0C, the CT image demonstrated an ice ball diameter of correlation between the tumor size and the cryoablation
30 mm and showed the thermometer tip at the edge of the results. No intraoperative or postoperative complications
dark area (Fig. 2b). Maximal ice ball diameter during occurred in any patient. Serum creatinine level did not
freeze was 45 mm. The temperature of the thermocouple differ significantly following cryoablation (baseline of 1.1
tip at the end of the second freeze was -14C and the tip of and 1.6 mg/dl to a level of 1.1 and 1.7mg/dl, respectively).
the thermometer probe was within the dark zone (Fig. 2c).
The ice ball was 25 mm beyond the tumor edge. One-
month follow-up contrast-enhanced CT scan showed the Discussion
region of ablation to measure 26 9 29 mm, encompassing
the previously documented solid renal lesion, and no Cryotherapy provides a means for the focal destruction of
contrast enhancement was observed at 3-month (Fig. 2d) malignant tissue while preserving the majority of the sur-
and 6-month follow-up imaging. rounding normal tissue [13]. The mechanisms of tissue

Table 2 The correlation between tumor size and cryoablation results


Original tumor Probes Cryoablation: Technique effectiveness Enhancement/time of follow-up
ice ball (cm) (complete ablation)a after procedure (mo)
No. Size (cm) Size (mm) No. of probes

1 1.4 2.4 1 2.4 Yes No/9


2 2.0 2.4 2 4.5 Yes No/6
3 1.5 2.4 2 4.0 Yes No/6
4 2.5 2.4 2 4.1 Yes No/9
5 3.3 2.4 2 5.3 Yes No/6
Mean 1.8 2.4 2 4.1
a
A tumor that is treated accordingly to protocol and covered completely by the ice ball with a diameter[1.0 cm larger than the tumor diameter

123
120 S. Permpongkosol et al.: Temperature Measurements of the Low Attenuation Radiographic Ice Ball

injury during cryotherapy are (i) intracellular ice forma- This study is limited by being a small feasibility study.
tion, (ii) dehydration of cells, and (iii) stagnation of the The inexact correlation of the location of the thermometer
microcirculation [14]. Recently, percutaneous cryotherapy tip with the location of the edge of low-attenuation area on
has been increasingly applied to treat renal masses. Based CT makes the exact temperature at the ice ball edge
on the wide experience with CT-guided percutaneous imprecise. Since the thermal map gradient is steep, the
procedures, a CT-guided approach to renal cryotherapy has temperature observed could be very different without much
been popular, although other modalities such as MRI and observational difference on the CT image. These concerns
ultrasound have also been applied to the guidance problem. are somewhat mitigated by the fact that the observations
Gilbert et al. [15] suggested that CT has significant correlate with the theoretical and experimental observa-
advantages over MRI for cryotherapy guidance, but the tions of Sandison et al. [8]. The utility of thermometers
main disadvantage of CT is its inability to measure tem- during cryotherapy is somewhat controversial. Although
perature gradients within the ice ball. temperature monitoring with thermocouples provides
The issue of tissue temperature is a complex and critical important information, it comes at the cost of additional
one for percutaneous cryotherapy. Several anatomic issues risk related to additional needle placements. Ideally a
including the proximity of vascular structures and the noninvasive method of intraprocedural ablation monitoring
collecting system may limit the cooling effect from the would be possible. Cryoablation has the potential of
cryoprobe, potentially compromising efficacy. Blood ves- offering a noninvasive method of intraprocedural ablation
sels continuously bring 37C blood to an area being cooled. monitoring without the added risk of thermocouple place-
This effect is known as the thermal (heat /cold) sink[16]. ment. Unlike heat-based ablation techniques, during
Since, due to these factors, a single protocol creates vary- cryotherapy a low-attenuation ice ball is visible on CT. The
ing ablation zones, having a means for intraprocedural results presented here provide some understanding of the
monitoring of the cryotherapy to ensure adequate treatment meaning of this low-attenuation area and provide some
is essential [8]. In 1983, Reiser et al. [6] first indicated that guidance to its use during cryotherapy. Results here con-
during cryotherapy under CT guidance, a low-attenuation firm the animal work by others and suggest that the edge of
area surrounding the cryoprobe expands and correlates with the low-attenuation area likely represents 0C. This is only
the growing ice ball based on in vivo experiments on an partly helpful since 0C is insufficient to cause reliable cell
anesthetized pig. Density changes induced by cooling were death. Nonetheless, this is still very useful information and
visualized on CT and the corresponding temperature allows physicians to necessarily create ice balls that are
changes were measured using thermocouples. In 1998, larger than targeted tumors.
Sandison et al. [8] described a phantom study to determine
the potential of CT for the direct monitoring of temperature
within the ice ball and to define the complete three- Conclusion
dimensional boundary of an ice ball by the signature
change in HU value from unfrozen to completely frozen Knowledge of the temperature at the ice ball edge during
tissue. No clinical experience has correlated temperature cryoablation can be used to predict the extent of tissue
findings with CT imaging findings during ice ball necrosis and thus provide an estimate of cryotherapy
formation. effectiveness during the procedure. Further work is nec-
Note that in the particular patients reviewed here, the essary to establish a firm relationship between the thermal
mean nadir temperature measured at the thermometer tip conditions and the zone of damage.
which was within the final ice ball edge was -12C. In all
of these cases, the radiographic ice ball extended several
millimeters beyond the tumor edge. At follow-up, contrast- References
enhanced CT demonstrated an absence of contrast
enhancement in the patients with subsequent follow-up 1. Silverman SG, Tuncali K, Morrison PR (2005) MR imaging-
imaging, suggesting successful ablation of renal tissue. The guided percutaneous tumor ablation (1). Acad Radiol
12(9):11001109
data presented here stress that the radiographic ice ball 2. Dupas B, Frampas E, Leaute F, Bertrand-Vasseur A, Lerat F
must be significantly larger than the tumor to achieve (2005) [Complications of fluoroscopy-, ultrasound-, and CT-
consistent tissue necrosis, since the temperature at the ice guided percutaneous interventional procedures]. J Radiol 86(5; Pt
ball edge is only 0C. In these cases, we assumed that the 2):586598
3. Santambrogio R, Bianchi P, Pasta A, Palmisano A, Montorsi M
temperature at the edge of the tumor was below -20C, (2002) Ultrasound-guided interventional procedures of the liver
since the tumor edge was several millimeters closer to the during laparoscopy: technical considerations. Surg Endosc
cryoprobes than even the thermometer tips. 16(2):349354

123
S. Permpongkosol et al.: Temperature Measurements of the Low Attenuation Radiographic Ice Ball 121

4. Hinshaw JL, Lee FT Jr (2004) Image-guided ablation of renal cell 10. Chosy SG, Nakada SY, Lee FT Jr, Warner TF (1998) Monitoring
carcinoma. Magn Reson Imaging Clin N Am 12(3):429447, vi renal cryosurgery: predictors of tissue necrosis in swine. J Urol
5. Popken F, Seifert JK, Engelmann R, Dutkowski P, Nassir F, 159(4):13701374
Junginger T (2000) Comparison of iceball diameter and tem- 11. Campbell SC, Krishnamurthi V, Chow G, Hale J, Myles J,
perature distribution achieved with 3-mm accuprobe cryoprobes Novick AC (1998) Renal cryosurgery: experimental evaluation of
in porcine and human liver tissue and human colorectal liver treatment parameters. Urology 52(1):2933, discussion 3324
metastases in vitro. Cryobiology 40(4):302310 12. Goldberg SN, Grassi CJ, Cardella JF, et al. (2005) Image-guided
6. Reiser M, Drukier AK, Ultsch B, Feuerbach S (1983) The use of tumor ablation: standardization of terminology and reporting
CT in monitoring cryosurgery. Eur J Radiol 3(2):123128 criteria. Radiology 235(3):728739
7. Rewcastle JC, Sandison GA, Hahn LJ, Saliken JC, McKinnon JG, 13. Jang TL, Wang R, Kim SC, Troe T, Pins MR, Nadler RB (2005)
Donnelly BJ (1998) A model for the time-dependent thermal Histopathology of human renal tumors after laparoscopic renal
distribution within an iceball surrounding a cryoprobe. Phys Med cryosurgery. J Urol 173(3):720724
Biol 43(12):35193534 14. Harada J, Mogami T (2004) [Minimally invasive therapy under
8. Sandison GA, Loye MP, Rewcastle JC, Hahn LJ, Saliken JC, image guidanceemphasizing MRI-guided cryotherapy]. Rinsho
McKinnon JG, Donnelly BJ (1998) X-Ray CT monitoring of Byori 52(2):145151
iceball growth and thermal distribution during cryosurgery. Phys 15. Gilbert JC, Rubinsky B, Wong ST, Brennan KM, Pease GR,
Med Biol 43(11):33093324 Leung PP (1997) Temperature determination in the frozen region
9. Gupta A, Allaf ME, Kavoussi LR, Jarrett TW, Chan DY, Su LM, during cryosurgery of rabbit liver using MR image analysis.
Solomon SB (2006) Computerized tomography guided percuta- Magn Reson Imaging 15(6):657667
neous renal cryoablation with the patient under conscious 16. Ladd AP, Rescorla FJ, Baust JG, Callahan M, Davis M, Grosfeld
sedation: initial clinical experience. J Urol 175(2):447452, dis- JL (1999) Cryosurgical effects on growing vessels. Am Surg
cussion 452443 65(7):677682

123
Cardiovasc Intervent Radiol (2008) 31:122130
DOI 10.1007/s00270-007-9225-0

CLINICAL INVESTIGATION

Frequency and Risk Factors of Various Complications After


Computed TomographyGuided Radiofrequency Ablation
of Lung Tumors
Tomohisa Okuma Toshiyuki Matsuoka Akira Yamamoto Yoshimasa Oyama
Masami Toyoshima Kenji Nakamura Yuichi Inoue

Received: 30 July 2007 / Accepted: 9 October 2007 / Published online: 6 November 2007
Springer Science+Business Media, LLC 2007

Abstract after the procedure. Regarding major complications, three


Objective To retrospectively determine the frequency and patients developed fever[38.5C; three patients developed
risk factors of various side effects and complications after abscesses; two patients developed pneumothorax requiring
percutaneous computed tomographyguided radiofre- chest tube insertion; and one patient had air embolism and
quency (RF) ablation of lung tumors. was discharged without neurologic deficit. Univariate and
Methods We reviewed and analyzed records of 112 multivariate analyses suggested that a lesion located B1 cm
treatment sessions in 57 of our patients (45 men and 12 of the chest wall was significantly related to pain (p\0.01,
women) with unresectable lung tumors treated by ablation. hazard index 5.76). Risk factors for pneumothorax
Risk factors, including sex, age, tumor diameter, tumor increased significantly with previous pulmonary surgery
location, history of surgery, presence of pulmonary (p \ 0.05, hazard index 6.1) and presence of emphysema
emphysema, electrode gauge, array diameter, patient (p \0.01, hazard index 13.6).
position, maximum power output, ablation time, and min- Conclusion The total complication rate for all treatment
imum impedance during ablation were analyzed using sessions was 58%, and 25% of patients did not have any
univariate and multivariate analyses. complications after RF ablation. Although major compli-
Results Total rates of side effects and minor and major cations can occur, RF ablation of lung tumors can be
complications occurred in 17%, 50%, and 8% of treatment considered a safe and minimally invasive procedure.
sessions, respectively. Side effects, including pain during
ablation (46% of sessions) and pleural effusion (13% of Keywords Complication  Lung tumor 
sessions), occurred with RF ablation. Minor complications, Radiofrequency ablation 
including pneumothorax not requiring chest tube drainage Computed axial tomographyguided
(30% of sessions), subcutaneous emphysema (16% of
sessions), and hemoptysis (9% of sessions) also occurred
Introduction

T. Okuma (&)  T. Matsuoka  A. Yamamoto  K. Nakamura 


Radiofrequency (RF) ablation is a minimally invasive
Y. Inoue
Department of Radiology, Osaka City University Graduate procedure for the percutaneous treatment and local control
School of Medicine, 1-4-3 Asahi-machi, Abeno-ku, of hepatocellular carcinoma and metastatic liver cancer.
Osaka 545-8585, Japan Recently, it has also been used to treat kidney, adrenal
e-mail: o-kuma@msic.med.osaka-cu.ac.jp
gland, breast, bone, and soft-tissue tumors [1, 2]. Several
Y. Oyama reports have established the clinical benefits of RF ablation
Department of Radiology, Yodogawa Christian Hospital, for patients with unresectable lung tumors [39]. Com-
2-9-26 Awaji, Higashiyodogawa-ku, Osaka 533-0032, Japan puted axial tomography (CT)guided radiofrequency (RF)
ablation for lung tumors is generally a safe procedure that
M. Toyoshima
Department of Radiology, Kobe Nishi-shimin Hospital, requires a short hospital stay, and complications are usually
2-4 Ichibancho Nagata-ku, Kobe 653-0013, Japan minor. Complications associated with RF ablation of lung

123
T. Okuma et al.: Frequency and Risk Factors of Various Complications After Computed Axial TomographyGuided RF 123

tumors have been described in many studies; however, cancer; 9 treatment sessions in 6 patients for rectal cancer;
there have been few analyses of risk factors [7, 8]. Early 8 treatment sessions in 1 patient for malignant mixed soft-
detection and proper management of complications are tissue tumor; 7 treatment sessions in 3 patients for hepa-
essential to the successful performance of RF ablation of tocellular carcinoma; 7 treatment sessions in 1 patient for
lung tumors. The purpose of this study was to retrospec- esophageal cancer; 6 treatment sessions in 1 patient for
tively determine the frequency of complications, as well as pancreatic carcinoma; 5 treatment sessions in 1 patient for
the risk factors for various side effects and complications, renal cancer; 5 treatment sessions in 1 patient for leiomy-
occurring in our hospital after RF ablation of lung tumors. osarcoma; 4 treatment sessions in 2 patients for bladder
cancer; 2 treatment sessions in 2 patients for thyroid can-
cer; 1 treatment session in 1 patient for uterine cancer; 1
Materials and Methods treatment session in 1 patient for uterine cervical cancer; 1
treatment session in 1 patient for laryngeal cancer; and 1
Patients treatment session in one patient for transitional cell carci-
noma of renal pelvic carcinoma) (Table 1). Mean tumor
RF ablation for lung tumors was performed after obtaining diameter at the time of RF ablation was 2.0 : 1.1 cm (:
approval from our hospitals Ethics Committee as well as SD).
written informed consent from the patients. Primary lung
tumors were diagnosed based on radiographic and clinical
findings and confirmed by pathologic examination obtained RF Ablation Method
by the most effective (sputum cytology, transbronchial
needle biopsy, or fine-needle aspiration). The presence of a Before RF ablation, malignancy was confirmed; physical
metastatic lung tumor on follow-up radiographic imaging examination was performed; and blood tests, chest radio-
suggested that a patient, who had been treated for cancer in graph, electrocardiogram, arterial blood gases, pulmonary
another organ, had developed a new pulmonary tumor and/ function, bleeding tendency, pulmonary infections, and use
or progression of nodule size. The procedure was per- of anticoagulant drugs or bronchodilators were checked
formed in patients whose cancer was considered and noted. All procedures were performed with the patient
unresectableor who were deemed unable to tolerate che- under local anesthesia; 5 mg diazepam was administered
motherapy because of previous pulmonary resection, poor
pulmonary function, multiple tumors, or other reasonsafter
Table 1 Characteristics of the 112 treatment session in 57 patients
discussion with thoracic surgeons. None of the patients
who underwent RF ablation had previously refused sur- Sex
gery. The following patients were considered unsuitable Men 45
candidates for lung RF ablation: patients with a lesion women 12
located B1 cm of the heart, mediastinum, or hilum; patients Age (mean : SD) 70.4 : 10.9
who were unable to care for themselves; patients who Tumor type
required the equivalent of institutional or hospital daily Primary lung cancer 9 sessions (4 patients)
care; patients whose disease was considered to be pro- Metastatic
gressing rapidly; patients with active inflammation or lung cancer 33 sessions (25)
infection; patients who had difficulties with activities of Colon 13 sessions (6)
daily living caused by previous cerebrovascular disease; Rectum 9 sessions (6)
and patients with pulmonary hypertension, bleeding ten- malignant mixed soft tumor 8 sessions (1)
dencies, or coagulation abnormalities (bleeding time [ 5 hepatocellular ca. 7 sessions (3)
minutes or prothrombin timeinternational normalized ratio esophagus 7 sessions (1)
[1.5). To evaluate complications after RF ablation, we pancreas 6 sessions (1)
retrospectively analyzed the records of 57 patients (45 men renal cell carcinoma 5 sessions (1)
and 12 women; age 70.4 : 10.9 years [mean : SD]) who leiomyosarcoma 5 sessions (1)
underwent 112 RF ablation treatment sessions during 6 bladder 4 sessions (2)
years (June 2000 through January 2007). The treated thyroid 2 sessions (2)
lesions in these patients included primary lung cancer (9 uterine corpus 1 sessions (1)
sessions in 4 patients); recurrence of primary lung cancer uterine cervix 1 sessions (1)
(33 sessions in 25 patients); and lung metastasis from larynx 1 sessions (1)
various organs (70 sessions in 28 patients, including 13 transitional cell ca. of renal pelvix 1 sessions (1)
treatment sessions in 6 patients for metastases from colonic

123
124 T. Okuma et al.: Frequency and Risk Factors of Various Complications After Computed Axial TomographyGuided RF

orally before the procedure as premedication. During the scanner (Asteion Multi; Toshiba, Tokyo, Japan) with a
early period (i.e., June 2000 through March 2005), the slice thickness of 2 mm and a pitch of 5.5. CT scan was
procedures were performed with the patient under local performed using lung window settings (window width
anesthesia, and conscious sedation was achieved with 1800 to 2000 HU and window level 500 to -600 HU).
injection of 15 to 45 mg pentazocine hydrochloride.
However, 3 patients with subpleural tumor complained of
severe pain during ablation, which interrupted treatment. Recorded Variables and Statistical Analysis
After that (from April 2005 onward), we used conscious
sedation and achieved pain control with intravenous Minor complications were defined as those requiring no or
injection of 0.1 to 0.2 ml fentanyl and 5 mg midazolam. nominal therapy and without consequence. Major compli-
We used an RF 2000 generator (Boston Scientific, Natick, cations were defined as those that, if left untreated, may
MA) and 15- (until March 2003) or 17-gauge (from April have threatened the patients life, lead to substantial dis-
2003 onward) LeVeen electrodes (Boston Scientific). A 2 ability and morbidity, resulted in hospital admission, or
cmdiameter array was used for lesions\2 cm in diameter, substantially lengthened hospital stay according to previ-
and a 3 cmdiameter array was used for tumors [2 cm in ously described guidelines [10]. Clinical records and CT
diameter. A grounding pad was placed on each of the scans were compared retrospectively with the frequency of
patients thighs, and the patient was placed in the CT various complicationsincluding pneumothorax, subcuta-
scanner in a supine or prone position, depending on the neous emphysema, fever-up, hemoptysis, and side effects,
location of the tumor. CT scan was performed using X- including pain and pleural effusionafter RF ablation of lung
Vigor (Toshiba, Tokyo, Japan), with a slice thickness of 2 tumors as determined on the basis of treatment sessions. To
mm and a pitch of 1. After locally anesthetizing the sub- assess pain, we used a three-point scale: grade 0 = no pain;
cutaneous tissue with 5 ml 1% procaine, we introduced the grade 1 = mild, self-limited pain that required no sedation;
electrode in a manner similar to that used in CT-guided and grade 2 = severe pain that made it difficult to continue
biopsy, i.e., with nonreal time gradual increments and the procedure without sedating the patient.
correlation of the electrode to the lesion site. We confirmed To determine underlying contributing factors contrib-
that the electrode was correctly inserted into the center of uting to various side effects and complications, we assessed
the lesion and that the fully opened electrode tip (umbrella- the following: (1) patient- and tumor-related factors (sex,
shaped array) covered the tumor. RF power was applied age, tumor diameter, tumor location [upper vs. lower and
initially between 10 and 30 W and was increased in 5-W whether the lesion was located B1 cm of the chest wall],
increments every 2 minutes. To complete ablation of large history of lung surgery, presence vs. absence of pulmonary
tumors, several overlapping ablations were performed at emphysema) and (2) RF ablation procedurerelated factors
different sites in the tumor to establish a safety margin. (electrode gauge [17 vs. 15], array diameter [2 vs. 3 cm],
Maximum output was B80 W. As a rule, ablation was patient position [prone vs. supine], maximum power out-
continued until impedance reached maximum, and then the put, ablation time, and minimum impedance during
current was automatically stopped (called roll-off). ablation). Before RF ablation three radiologists determined
Impedance and duration of ablation were automatically per CT scan the presence or absence of pulmonary
recorded. During the procedure, the patients heart rate and emphysema, tumor diameter, and whether the lesion was
rhythm (electrocardiogram monitoring), arterial oxygen located B1 cm of the chest wall. Data are presented as
saturation (oximetry), and arterial blood pressure were mean : SD. For univariate analysis, Mann-Whitney U test
monitored. To prevent infection, prophylactic antibiotic (2 was used to assess age, tumor size, maximum power out-
g cefazorin; Astellas Pharma, Tokyo, Japan) was admin- put, ablation time, and minimum impedance factors. Other
istered for 3 days after the procedure. The presence of factors were assessed using Fishers exact test. Multivari-
pneumothorax, subcutaneous emphysema, and pleural able analysis using logistic regression was performed for
effusion was evaluated on CT scan after RF ablation. various factors related to side effects and complications.
Statistical analysis was performed using commercially
available software (StatView version 5.0; SAS, ), and p
Follow-Up values \0.05 were considered statistically significant.

Follow-up consisted of CT scans, blood tests, and physical


examination performed at 1 week, 1 month, and every 2 to Results
3 months thereafter. All patients also underwent nonen-
hanced CT scans. Follow-up CT scan after RF ablation of Total rates of side effects and minor and major complica-
the tumor was performed using a multidetector-row CT tions of treatment sessions were 17%, 50%, and 8%,

123
T. Okuma et al.: Frequency and Risk Factors of Various Complications After Computed Axial TomographyGuided RF 125

respectively. Side effects of the treatment sessions, suggested that the frequency of pain was significantly
including pain during ablation and pleural effusion, higher in cases involving a lesion located B1 cm of the
occurred in 46% and 13% patients, respectively. Minor chest wall, older age, minimum impedance, and total
complications, including pneumothorax not requiring chest ablation time (p \ 0.05). No statistical difference was
tube drainage (30% of patients), mild subcutaneous observed for the other factors (Table 2). In contrast, mul-
emphysema (16% of patients), and small hemoptysis (9% tivariate analysis suggested that a lesion located B1 cm of
of patients), occurred after the procedure. 8 patients had a the chest wall was significantly related to pain during RF
prolonged hospital stay to treat the following complica- ablation (p \ 0.01, hazard index 5.76) (Table 3).
tions: pneumothorax (2% of patients required drainage),
fever [38.5C (3% of patients were treated by prolonging
administration of antibiotics), and cavity formation at the Pneumothorax
ablation site (3% of patients were readmitted for suspected
infection requiring antibiotics). We encountered one Pneumothorax was the next most common complication
patient who developed air embolism during electrode and occurred during and/or after 30% of sessions (34 of
insertion into a metastatic lung tumor, but he was dis- 112). In 32 sessions (32 patients), pneumothorax was
charged without neurologic deficit. No procedure-related minor, and no treatment was required (Figs. 2 and 3).
death occurred. These 32 patients were clinically asymptomatic. In this
study, chest tube drainage was required in only 2 patients
with massive pneumothorax. Two patients complained of
Side Effects and Complications dyspnea or decreased oxygen saturation. In the univari-
able and multivariate analyses, risk factors for
Pain pneumothorax increased significantly with previous pul-
monary surgery and presence of emphysema (p \0.05).
Pain during ablation was the most frequent side effect and No significant differences were noted in the other factors
occurred in 46% of sessions (51 of 112). Thirty-seven of 51 (Tables 2 and 3).
patients with subpleural tumor complained of pain during
ablation (Fig. 1). Pain was reported as grade 0 in 61
treatment sessions, as grade 1 in 12 treatment sessions, and Subcutaneous Emphysema
as grade 2 in 12 treatment sessions. Univariate analysis
Subcutaneous emphysema occurred during and/or after
16% of sessions (18 of 112) (Fig. 3). No patient required

Fig. 1 Pain during RF ablation in a 77-year-old man with lung tumor Fig. 2 Pneumothoraces in a 56-year-old man who underwent RF
that metastasized from bladder cancer. Transverse CT images taken ablation for recurrence after surgery for primary lung cancer. CT
during the ablation show tumors located near the chest wall. The images taken during RF ablation show mild pneumothoraces (arrow),
frequency of pain is significantly higher in patients who undergo RF but insertion of a chest tube was not required. Pneumothorax is one of
ablation for a lesion located B1 cm of the chest wall. We expected the most frequent complications and occurred in 30% of treatment
pain to occur and administered midazolam and fentanyl before the sessions in our series, although chest tube drainage was required in
procedure only 2% of patients

123
Table 2 Results of univariate analysis to determine risk factors for side effects and complications
126

Factors Pain No pain P value Pneumothorax No pneumothorax P value SE* No SE* P value

123
Male/Female 42/9 47/14 0.639 29/5 60/18 0.446 15/3 72/22 0.759
Age (Y r : SD) 74.2 : 10.9 67.2 : 9.6 P \ 0.01 72.5 : 7.6 69.0 : 12.3 0.410 71.2 : 11.3 70.2 : 10.6 0.618
Size (cm : SD) 2.2 : 1.3 1.9 : 1.1 0.170 2.1 : 0.9 2.0 : 1.2 0.420 2.1 : 1.0 2.0 : 1.1 0.909
Upper / Lower lobe 11/40 21/40 0.148 7/27 25/53 0.260 7/11 25/69 0.393
Located within 1 cm of chest wall (yes / no) 37/14 24/37 P \ 0.001 18/16 46/35 0.840 7/11 54/40 0.198
History of lung surgery (yes / no) 25/26 33/28 0.705 10/24 48/30 P \ 0.01 6/12 52/42 0.122
Pulmonary emphysema (yes / no) 12/39 7/54 0.129 12/22 7/71 P \ 0.01 6/12 13/81 0.079
Electrode Gauge; 17G / 15G 32/19 27/34 0.059 19/15 40/38 0.685 10/8 49/45 1.000
Diameter of the array; 2cm / 3cm 22/29 27/34 1.000 12/22 37/41 0.301 7/11 42/52 0.797
Patient position (prone / supine) 35/26 39/22 0.578 7/27 47/31 0.054 10/8 63/32 0.429
Maximum power out put (W : SD) 50.1 : 19.1 47.9 : 21.1 0.520 49.2 : 16.5 48.9 : 21.7 0.740 54.3 : 21.1 47.9 : 20.0 0.208
Ablation time (min : sec) 53:28 : 23:41 41:26 : 27:14 P \ 0.05 51:19 : 26:39 44:16 : 26:09 0.250 46:57:18:31 46:16:27:56 0.658
Minimum impedance (X : SD) 88.3 : 21.2 103 : 33.2 P \ 0.05 95.1 : 22.7 97.0 : 31.9 0.797 89.7 : 19 97.6 : 30.9 0.457
*Note-SE means subcutaneous emphysema

Table 2 Results of univariate analysis to determine risk factors for side effects and complications
Pleural effusion No effusion P value Hemoptysis No hemoptysis P value

12/2 76/22 0.730 7/3 82/20 0.426


68.1 : 4.0 70.8 : 11.5 0.137 62.3 : 8.1 71.6 : 10.6 P \ 0.01
2.2 : 1.3 2.0 : 1.1 0.540 3.1 : 1.2 1.9 : 1.1 P \ 0.01
3/11 29/69 0.754 4/6 28/74 0.468
8/6 53/45 1.000 3/7 58/44 0.181
7/7 51/47 1.000 8/2 8/94 P \ 0.001
2/12 17/81 1.000 0/10 19/83 0.207
4/10 55/43 0.084 3/7 56/46 0.187
6/8 43/55 1.000 2/8 47/55 0.181
11/3 63/35 0.375 5/5 69/33 0.303
47.3 : 16.1 49.2 : 20.7 0.902 67.8 : 19.2 47.0 : 19.2 P \ 0.01
44:02 : 32:04 46:38 : 25:37 0.551 52:39 : 27:03 45:41 : 26:21 0.332
94.6 : 23.5 96.6 : 30.0 0.959 86.6 : 18.2 97.5 : 30.0 0.199
*Note-SE means subcutaneous emphysema
T. Okuma et al.: Frequency and Risk Factors of Various Complications After Computed Axial TomographyGuided RF
T. Okuma et al.: Frequency and Risk Factors of Various Complications After Computed Axial TomographyGuided RF 127

treatment for this complication. All patients with mild analyses to detect factors associated with hemoptysis
subcutaneous emphysema were clinically asymptomatic, indicated the relevance of tumor location (p \ 0.05)
and this complication resolved during careful follow-up. (Tables 2 and 3).
Statistical analyses showed no significant differences in
various factors (Tables 2 and 3). In contrast, analysis by
Fishers exact test suggested that occurrences of pneumo- Discussion
thorax and multiple lesions were significantly related to
subcutaneous emphysema (p \ 0.05). Since the first application of RF ablation for lung tumors by
Dupuy et al. [3], many groups have reported its clinical
benefits as a safe and minimally invasive procedure for
Pleural Effusion patients with unresectable lung tumors [49]. Some of
these studies also reported the frequency and factors con-
Pleural effusion after ablation was also a common side tributing to complications during and/or after RF ablation
effect and occurred during and/or after 13% of sessions (14 of lung tumors [7, 8]. Pain was the most frequent side
of 112). Most cases were self-limiting, and two patients effect in our series. RF ablation is typically painful [9], and
with severe pleural effusion complained of dyspnea on we believe that control of pain during ablation is important
exertion and required chest tube drainage. Statistical to successfully achieve treatment. The possible merits of
analyses showed no significant differences in various fac- general anesthesia include pain control; however, we need
tors (Tables 2 and 3). the help of an anesthesiologist to accomplish this. In Japan,
general anesthesia during RF ablation of lung tumors is
uncommon because anesthesiologists are few; instead,
Self-Limiting Hemoptysis local anesthesia and strong analgesics are usually admin-
istered during the procedure to control pain. Hoffman et al.
Hemoptysis occurring after ablation is not uncommon, and reported that RF ablation of lung tumors with the patient
the prevalence of this was 9% (10 of 112 sessions). Mild under general anesthesia or consciously sedation did not
hemoptysis (\10 mL) usually lasted for 3 days and did not result in different complication rates. They concluded that,
require any treatment. None of the patients required a except in anxious patients, conscious sedation should be
blood transfusion. In the univariate analysis, risk factors for used during RF ablation [11]. In this series, which included
hemoptysis increased significantly with older patient age, a relatively small number of patients who were adminis-
larger tumor size, maximum power output, and previous tered fentanyl and midazolam during RF ablation, we could
pulmonary surgery (p \ 0.01). In contrast, multivariate not determine that differences in sedation affected the

Table 3 Results of multivariate analysis to determine risk factors for side effects and complications
Factors Pain Pneumothorax Subcuaneous emphysema Pleural effusion Hemoptysis
P value Hazard P value Hazard P value Hazard P value Hazard P value Hazard
ratio ratio ratio ratio ratio

Male / Female 0.892 0.69 0.2551 2.885 0.2730 3.978 0.999 1 0.1774 0.036
Age (Y r : SD) 0.781 1.597 0.9391 0.997 0.7687 1.013 0.6306 1.035 0.5219 0.949
Size (cm : SD) 0.643 1.277 0.6488 0.822 0.0813 0.416 0.1367 2.962 0.1865 4.489
Upper / Lower lobe 0.406 0.566 0.2104 0.323 0.5422 1.561 0.4532 0.452 P \ 0.05 29.42
Located within 1 cm of chest P \ 0.01 5.755 0.737 1.277 0.9491 0.951 0.8171 0.787 0.1886 0.054
wall (yes / no)
History of lung surgery (yes / no) 0.735 0.795 P \ 0.01 6.06 0.1757 0.316 0.6683 1.61 0.0796 90.745
Pulmonary emphysema (yes / no) 0.288 2.239 P \ 0.01 13.569 0.2817 2.493 0.2448 4.125 1 1
Electrode Gauge; 17G / 15G 0.308 1.93 0.536 0.640 0.6749 1.388 0.9994 1 0.0985 0.026
Diameter of the array; 2cm / 3cm 0.784 1.261 0.1495 0.254 0.4902 1.996 0.6697 1.687 0.6401 0.304
Patient position (prone / supine) 0.278 0.473 0.6625 1.419 0.8826 0.897 0.665 1.622 0.7645 1.694
Maximum power out put (W : SD) 0.814 1.005 0.3324 1.026 0.3017 1.031 0.6415 0.984 0.2742 1.075
Ablation time (min : sec) 0.758 1 0.5783 1.000 0.4912 1 0.8477 1 0.94 1
Minimum impedance (X : SD) 0.146 0.974 0.3706 1.014 0.4902 0.983 0.4206 1.016 0.9589 0.998

123
128 T. Okuma et al.: Frequency and Risk Factors of Various Complications After Computed Axial TomographyGuided RF

Fig. 3 Subcutaneous emphysema and pneumothoraces (arrow) in a


72-year-old man who underwent RF ablation for lung tumors that
metastasized from hypopharyngeal cancer. CT image taken during RF
ablation shows pneumothoraces. This patient did not complain of
dyspnea and did not show decreased oxygen saturation. He was
asymptomatic, and insertion of a chest tube was not required

degree of pain that occurred during the procedure. There


have only been reports of pain during RF ablation of lung
tumors. Yasui et al. described a patient with severe pain
who was under conscious sedation for ablation of apical
subpleural tumor, and they noted controllable intraproce-
dural pain in 29% of treatment sessions [12]. It appears that
pain may increase with ablation of tumors near the chest
wall. Multivariate analysis showed that lesion located B1
cm of the chest wall was significantly related to pain during
RF ablation (p \ 0.01, hazard index 5.76). Further study is
necessary to evaluate how pain during ablation is influ-
enced by different types of sedation. The reported
complication type and rate associated with RF ablation
includes pneumothorax, which occurred at a frequency of
9% to 63% of sessions and required the use of a chest tube
in some patients [311, 2026]. The frequency of pneu-
mothorax during RF ablation appears to be approximately
the same as that occurring during needle biopsy [1319].
Many reports of CT-guided lung biopsy have suggested Fig. 4 Cavity formation with infection in a 57-year-old man with
that risk factors for pneumothorax correlate with the lung metastases from colorectal cancer who was referred for RF
ablation. (A) During the procedure, no complications (e.g., pneumo-
presence of chronic obstructive pulmonary disease, number thorax, pain) were encountered, and he was discharged a few days
of electrode passes, older patient age, lesion depth, sub- after ablation. (B) Three weeks after ablation, however, the patient
pleural lesion, and smaller lesion size [1319]. However, developed fever, and abscess formation was shown on chest CT scan.
determination of the factors contributing to increased risk He was treated with intravenous and oral antibiotics
of pneumothorax after CT-guided lung biopsy remains
controversial. According to Yamagami et al., pneumotho-
rax during RF ablation occurred in 29.3% of cases, and a ablated tumors, involvement of the middle or lower lobe,
contributing factor for pneumothorax was emphysema (p\ and increased length of the aerated lung transversed by the
0.05) [7]. Hiraki et al. reported a 51% prevalence of electrode (p \ 0.05) [8]. In our series, the frequency of and
pneumothorax, and risk factors for development included factors contributing to pneumothorax support these two
male sex, no history of pulmonary surgery, large number of reports.

123
T. Okuma et al.: Frequency and Risk Factors of Various Complications After Computed Axial TomographyGuided RF 129

reported that risk factors for bleeding after needle biopsy


include lesion depth and smaller lesion size [18]. A few
studies have also reported that 2.7% to 12% of patients
experienced hemoptysis after RF ablation of lung tumors
[12, 24, 26]. In the current study, statistical analyses to
detect factors associated with hemoptysis was not in
accordance with the results of CT-guided needle biopsy.
Few studies have investigated the risk factors for hem-
optysis after RF ablation of lung tumors. A larger study
may be necessary to explain the exact mechanisms.
Some reports exist of serious and life-threatening com-
plications-such as air embolism [27, 28], abscess formation
[29], tumor seeding [30], massive hemorrhage [31], cere-
bral infarction [32], and procedure-related death [33, 34]-
occurring during RF ablation of lung tumors. In our series of
patients, major complications included high fever, abscess,
Fig. 5 Air embolism in a 74-year-old man with lung metastasis from pneumothorax requiring drainage, and air embolism
hepatocellular carcinoma in the right lower lobe who was referred for (Fig. 5). Regarding high fever and abscess formation, some
RF ablation. First, we used the electrode to penetrate the pleura, and, reports exist that included a few patients who developed
under CT guidance, advanced the electrode toward the tumor. We high fever or abscess formation after RF ablation [5, 6, 12,
then attempted to expand the electrode. At that stage, the patient
coughed; suddenly became unresponsive, right hemiparetic, and 25, 29]. In our hospital, to prevent infection, prophylactic
pulseless; and went into respiratory arrest. CT scan shows air antibiotics were administered for 3 days after the procedure.
embolism in the descending aorta (arrow). The patient recovered 30 However, the prophylactic use of antibiotics to prevent
minutes later, and the air embolism resolved spontaneously. Diffu- fever remains controversial [9]. Regarding air embolism
sion-weighted image and fluid-attenuated inversion recovery of the
brain (not shown) on the day after the complication showed no during RF ablation, Ghaye et al. and Okuma et al. reported
abnormal signal of acute infarction. The patient was discharged a few cases of air embolism during insertion of an 18-gauge
days after the event without neurologic deficit LeVeen electrode [27, 28]. These patients were discharged
a few days after the event; there were no mortalities.

Subcutaneous emphysema and pleural effusion occurred


in 16% and 13% of sessions, respectively. In previous Conclusion
reports, the prevalence of subcutaneous emphysema and
pleural effusion occurring during and/or after RF ablation In conclusion, the rates of side effects and minor and major
was 0% to 19% and 4% to 19% of patients, respectively [3 complications were 17%, 50%, and 8% of treatment ses-
12, 2026]. Hiraki et al. reported that pleural effusion was sions, respectively. RF ablation of lung tumors is thought
seen in 19% of patients and that cluster electrode, to be a safe and minimally invasive procedure. Knowledge
decreased distance to the nearest pleura, and decreased of the diverse complications that can occur during RF
length of aerated lung transversed by the electrode resulted ablation of lung tumor, as well as their radiologic and
in a significantly higher incidence of pleural effusion [8]. clinical appearance, may facilitate their early diagnosis and
In this study, statistical analyses to detect factors associated prompt treatment.
with subcutaneous emphysema and pleural effusion
showed no significant differences in various factors. Most
patients with pleural effusion were clinically asymptom- References
atic. Pleural effusion is rarely treated with surgical
insertion of a chest tube. We believe that in patients with 1. Gazelle GS, Goldberg SN, Solbiati L, et al. (2000) Tumor abla-
subcutaneous emphysema and pleural effusion after RF tion with radiofrequency energy. Radiology 217:633646
2. Dupuy DE, Goldberg SN (2001) Image-guided radiofrequency
ablation, watchful follow-up is usually sufficient. tumor ablation: Challenges and opportunitiesPart 2. J Vasc Interv
In our series, the prevalence of hemoptysis after RF Radiol 12:11351148
ablation was 9%, and all cases were self-limiting; small 3. Dupuy DE, Zagoria RJ, Akerley W, et al. (2000) Percutaneous
hemoptyses lasted only a few days. Westcott et al. reported radiofrequency ablation of malignancies in the lung. AJR Am J
Roentgenol 174:5759
that minor hemoptysis sometimes occurred (usually in 4. Nishida T, Inoue K, Kawata Y, et al. (2002) Percutaneous
B10% of patients) with CT-guided needle biopsy and that radiofrequency ablation of lung neoplasms: A minimally invasive
most patients required no treatment [13]. Some studies strategy for inoperable patients. J Am Coll Surg 195:426430

123
130 T. Okuma et al.: Frequency and Risk Factors of Various Complications After Computed Axial TomographyGuided RF

5. Steinke K, King J, Glenn D, et al. (2003) Radiologic appearance cancer and metastases: Preliminary report. Radiology 230:125
and complications of percutaneous computed tomography-guided 134
radiofrequency-ablated pulmonary metastases from colorectal 21. Suh RD, Wallace AB, Sheehan RE, et al. (2003) Unresectable
carcinoma. J Comput Assist Tomogr 27:750757 pulmonary malignancies: CT-guided percutaneous radiofre-
6. Akeboshi M, Yamakado K, Nakatsuka A, et al. (2004) Percuta- quency ablation-preliminary results. Radiology 229:821829
neous radiofrequency ablation of lung neoplasms: Initial 22. Gadaleta C, Mattioli V, Colucci G, et al. (2004) Radiofrequency
therapeutic response. J Vasc Interv Radiol 15:463470 ablation of 40 lung neoplasms: Preliminary results. AJR Am J
7. Yamagami T, Kato T, Hirota T, et al. (2006) Pneumothorax as a Roentgenol 183:361368
complication of percutaneous radiofrequency ablation for lung 23. Belfiore G, Moggio G, Tedeschi E, et al. (2004) CT-guided
neoplasms. J Vasc Interv Radiol 17:16251629 radiofrequency ablation: A potential complementary therapy for
8. Hiraki T, Tajiri N, Mimura H, et al. (2006) Pneumothorax, patients with unresectable primary lung cancerA preliminary
pleural effusion and chest tube placement after radiofrequency report of 33 patients. AJR Am J Roentgenol 183:10031011
ablation of lung tumors: Incidence and risk factors. Radiology 24. de Baere T, Palussiere J, Auperin A, et al. (2006) Midterm local
241:275273 efficacy and survival after radiofrequency ablation of lung tumors
9. Rose SC, Thistlethwaite PA, Sewell PE, et al. (2006) Lung cancer with minimum follow-up of 1 year: Prospective evaluation.
and radiofrequency ablation. J Vasc Interv Radiol 17:927951 Radiology 240:587596
10. Goldberg SN, Grassi CJ, Cardell JF, et al. (2005) Image-guided 25. Bojarski JD, Dupuy DE, Mayo-Smith WW (2005) CT imaging
tumor ablation: Standardization of terminology and reporting findings of pulmonary neoplasms after treatment with radiofre-
criteria. Radiology 235:728739 quency ablation: Results in 32 tumors. AJR Am J Roentgenol
11. Hoffmann RT, Jakobs TF, Lubienski A, et al. (2006) Percuta- 185:466471
neous radiofrequency ablation of pulmonary tumorsIs there a 26. Herrera LJ, Fernando HC, Perry Y, et al. (2003) Radiofrequency
difference between treatment under general anaesthesia and ablation of pulmonary malignant tumors in nonsurgical candi-
under conscious sedation? Eur J Radiol 59:168174 dates. J Thorac Cardiovasc Surg 125:929937
12. Yasui K, Kanazawa S, Sano Y, et al. (2004) Thoracic tumors 27. Ghaye B, Bruyere PJ, Dondelinger RF (2006) Nonfetal system-
treated with CT-guided radiofrequency ablation: Initial experi- atic air embolism during percutaneous radiofrequency ablation of
ence. Radiology 231:850857 a pulmonary metastasis. AJR Am J Roentgenol 187:W327W328
13. Westcott JL, Rao N, Colley DP (1997) Transthoracic needle 28. Okuma T, Matsuoka T, Tutumi S, et al. (2007) Air embolism
biopsy of small pulmonary nodules. Radiology 202:97103 during needle placement for CT-guided radiofrequency ablation
14. Covey AM, Gandhi R, Brody LA, et al. (2004) Factors associated for unresectable lung metastatic lesion. J Vasc Interv Radiol [in
with pneumothorax and pneumothorax requiring treatment after press]
percutaneous lung biopsy in 443 consecutive patients. J Vasc 29. Okuma T, Matsuoka T, Yamamoto A, et al. (2007) Factors
Interv Radiol 15:479483 contributing to cavitation after computed tomography-guided
15. Saji H, Nakamura H, Tsuchida T, et al. (2002) The incidence and percutaneous radiofrequency ablation for lung tumors. J Vasc
the risk of pneumothorax and chest tube placement after percu- Interv Radiol 18:399404
taneous CT-guided lung biopsy. Chest 121:12511526 30. Yamakado K, Akeboshi M, Nakatsuka A, et al. (2005) Tumor
16. Kazerooni EA, Lim FT, Mikhail A, Martinez FJ (1996) Risk of seeding following lung radiofrequency ablation: A case report.
pneumothorax in CT-guided transthoracic needle aspiration Cardiovasc Interv Radiol 28:530532
biopsy of the lung. Radiology 198:371375 31. Vaughn C, Mychaskiw G, Swell P (2002) Massive hemorrhage
17. Laurent F, Michel P, Latrabe V, et al. (1999) Pneumothoraces during radiofrequency ablation of pulmonary neoplasm. Anesth
and chest tube placement after CT-guided transthoracic lung Analg 94:11491151
biopsy using a coaxial technique: Incidence and risk factors. AJR 32. Jin GY, Lee JM, Lee YC, Han YM (2004) Acute cerebral
Am J Roentgenol 172:10491053 infarction after radiofrequency ablation of an atypical carcinoid
18. Yeow KM, Su IH, Pan KT, et al. (2004) Risk factors of pneu- pulmonary tumor. AJR Am J Roentgenol 182:990992
mothorax and bleeding: Multivariate analysis of 660 CT-guided 33. Fernando HC, De Hoyos A, Landreneau RJ, et al. (2005)
coaxial cutting needle lung biopsies. Chest 126:748754 Radiofrequency ablation for the treatment of non-small cell lung
19. Jin GY, Lee JM, Lee YC, et al. (2004) Primary and secondary cancer in marginal surgical candidates. J Thorac Cardiovasc Surg
lung malignancies treated with percutaneous radiofrequency 129:639644
ablation: evaluation with follow-up helical CT. AJR Am J 34. Steinke K, King J, Glenn DW, et al. (2004) Percutaneous
Roentgenol 183:10131020 radiofrequency ablation of lung tumors with expandable needle
20. Lee JM, Jin GY, Goldberg SN, et al. (2004) Percutaneous electrodes: Tips from preliminary experience. AJR Am J
radiofrequency ablation for inoperable non-small cell lung Roentgenol 183:605611

123
Cardiovasc Intervent Radiol (2008) 31:131134
DOI 10.1007/s00270-007-9091-9

CLINICAL INVESTIGATION

Does Prebiopsy, Nonsterile Ultrasonography Gel Affect


Biopsy-Site Asepsis?
Kamil Gurel Oguz Karabay Safiye Gurel
Charles Hildebolt

Received: 14 February 2007 / Revised: 18 April 2007 / Accepted: 22 April 2007 / Published online: 3 November 2007
Springer Science+Business Media, LLC 2007

Abstract biopsy-site location (head and neck or breast and abdo-


Purpose The purpose of this study was to determine the men), and local factors (skin fold, skin tag, and hair).
extent to which the use of nonsterile gel, prior to antiseptic Results The following odds ratios (adjusted for the other
procedures in ultrasonography (US)-guided percutaneous variables) and their 95% confidence intervals were calcu-
biopsies, results in contamination of the biopsy site. lated: (1) group (2.9 [0.811.1]; p = 0.10); (2) gender (1.2
Materials and Methods Patients referred for US-guided [0.35.2]; p = 0.78); (3) coincidental disease (7.6 [0.9
percutaneous biopsies were included in this study. Trans- 166.7]; p = 0.09); (4) biopsy site location (6.2 [1.431.3];
mission material used for US evaluation before biopsy-site p = 0.02); and (5) local factors (7.0 [1.636.0]; p = 0.01).
antiseptic procedures were performed was either nonsterile No bacterial growth occurred with swabs obtained from
gel or sterile saline. Patients were randomly assigned to gel, povidine-iodine, or saline.
two groups: nonsterile gel (n = 30) and sterile saline Conclusion We conclude that nonsterile gel used prior to
(n = 30). Before the transmission material was used and percutaneous biopsy does not affect biopsy-site asepsis.
after antiseptic procedures were performed, microbial
swabs of a 10-cm2-diameter area were obtained at the Keywords Ultrasonography  Needle biopsy  Asepsis
biopsy site. Swabs were also obtained from the gel, saline,
and povidine-iodine. Inoculated specimen plates were
incubated at 37C under aerobic conditions, and the num- Ultrasonography (US) gels are water-based viscous prod-
bers of colony-forming units recorded. Nominal logistic ucts that aid in the transmission of ultrasonic wavelengths
regression analysis was used to calculate the odds of pos- from patient surfaces to transducer surfaces. Even though
tantisepsis bacterial growth (after antiseptic procedures they are most commonly stabilized with antimicrobial
were performed) based on group, gender, coincidental preservatives, they are not sterile [1]. US probes, machines,
disease (diabetes, chronic renal failure, and malignancy), and gels could become sources of infection if they are
contaminated [25]. To help prevent the US probes
spreading of infection, it is recommended that between
routine uses probes be cleaned twice with nonsterile paper
K. Gurel (&)  S. Gurel
Department of Radiology, Abant Izzet Baysal University, Izzet wipes and that, for uses with patients who are at high risk
Baysal School of Medicine, 14280 Golkoy/Bolu, Turkey for contracting infection (neonates and patients in the
e-mail: kamilgurel@hotmail.com intensive care units), before use the probes be cleaned with
nonsterile paper wipes and alcohol wipes [4, 6, 7]. Such
O. Karabay
Department of Clinical Microbiology and Infectious Diseases, preventive measures are important because hospital out-
Abant Izzet Baysal University, Izzet Baysal School of Medicine, breaks of infection have been traced to US gels [1, 6, 7].
14280 Golkoy/Bolu, Turkey Single-use containers are recommended for use with non-
sterile gels. If reusable containers are used, for nonsterile
C. Hildebolt
Mallinckrodt Institute of Radiology, Washington University, St. gels, they should be thoroughly cleaned between uses and
Louis, MO 63110, USA discarded if cracked [8].

123
132 K. Gurel et al.: Ultrasonography Gel and Biopsy-Site Asepsis

To determine the areas where antiseptic procedures need The distributions of ages for Group 1 (nonsterile gel)
to be performed and the exact locations of biopsy sites, most and Group 2 (sterile saline) were tested for normality with
interventional radiologists perform US examinations just Shapiro-Wilk W tests. The differences in ages between
before antiseptic procedures are performed for US-guided Group 1 and Group 2 were tested for significant differences
percutaneous biopsy. Nonsterile gel is routinely used for this with a Students t test. The effect of group membership on
US examination prior to the US-guided percutaneous biopsy. the numbers of CFU through time was analyzed with a
This nonsterile gel is usually removed with nonsterile paper repeated-measures analysis of variance, with the numbers
towels before antiseptic procedures are performed at the of pre- and postbiopsy CFU as the repeated (within-sub-
biopsy site. There is neither any reported serious complica- ject) factor. Nominal logistic regression analysis was used
tion due to nonsterile gel usage nor any controlled study of to calculate the odds of postbiopsy bacterial growth based
the safety of nonsterile gel usage. The aim of this study is to on the independent variables of (1) group (1 or 2), (2)
determine the extent to which nonsterile coupling gel, used gender (male or female), (3) coincidental disease (diabetes,
just prior to antiseptic procedures in US-guided percutane- chronic renal failure, and malignancywhich could
ous biopsies, affects biopsy-site asepsis. affect immunity), (4) biopsy-site location (the head and
neck were considered exposed areas, and the breast and
abdomen nonexposed areas), and (5) local factors (skin
Materials and Methods fold, skin tag, and hair). In the analysis, we controlled for
the effect of each independent variable. Alpha was set at
Between January 2004 and March 2005, 65 patients were 0.05. Statistical analyses (including the calculations of
referred for US-guided percutaneous biopsies. These odds ratios and 95% confidence intervals) were performed
patients were randomized into two groups: Group 1, those with JMP 6 Statistical Software (SAS Institute, Inc., Cary,
for whom nonsterile gel was used; and Group 2, those for NC).
whom sterile saline was used. This study was approved by our universitys research
A single, newly opened, 1-liter, gel bottle (Konix, ethics board, and informed consent was obtained from all
Thelessaloniki, Greece) was used throughout the study. patients.
Nonsterile gel, povidine-iodine, and sterile saline were
sampled 12 times (at approximately monthly intervals)
throughout the course of the study. Sterile cotton-tipped Results
swabs were used to sample 10-cm2-diameter areas at the
biopsy sites prior to the application of gel or sterile saline. The swabs for three patients were contaminated during
Prebiopsy US examinations were performed and any sampling, and the swabs for two other patients did not
remaining gel or saline was removed by wiping the biopsy reach the microbiology laboratory within 1 h. The swabs
sites twice with soft, nonsterile paper. for these 5 patients were excluded, leaving data for 60
Antiseptic procedures at the biopsy sites were performed patients, who had been randomly assigned to two groups.
twice with 10% povidine-iodine. After these procedures Group 1 (nonsterile gel) consisted of 18 females and 12
were performed, the biopsy sites were again sampled with males, with an age range of 22 to 78 years. The mean
sterile cotton-tipped swabs. Swabs that were obtained age standard deviation was 44. 8 13.9 years. The
before and after antiseptic procedures were coded with biopsy sites for Group 1 were the abdomen (n = 7), breast
random numbers and sent to the microbiology laboratory. (n = 2), thyroid (n = 20), and parotid (n = 1). Five patients
For swabs to be used in our study, they had to be delivered had local factors, and five patients had coincidental disease.
to the laboratory within 1 h of collection. Sampling and Group 2 (sterile saline) consisted of 22 females and 8
biopsies were performed by the same radiologist. Percu- males, with an age range of 22 to 78 years. The mean
taneous US-guided biopsies were performed with a age standard deviation was 49.5 13.9 years. There
standard, free-hand method, with the US probes covered was no significant difference in the ages for Groups 1 and 2
with sterile surgical gloves. After the biopsies, patients (p = 0.19). The biopsy sites for Group 2 were the abdomen
were recalled for follow-up examinations at 3 and 5 days. (n = 6), breast (n = 1), and thyroid (n = 23). Ten patients
A microbiologist, who was blinded to identifying had local factors, and three patients had coincidental
information, used the swabs to inoculate eosine-methylene disease.
blue plates and 5% sheep blood-agar plates. The plates A repeated-measures analysis of variance demonstrated
were incubated at 37C and examined after 24 and 48 h. that the mean standard deviation and (median) per-
Plates were considered positive for bacterial growth if they centage decreases through time in CFU were highly
contained C1000 colony-forming units (CFU) per cubic significant (Group 1, 90 39 [100]; Group 2, 89 23
millimeter. [100]; p \ 0.01; Table 1) but that there was not a

123
K. Gurel et al.: Ultrasonography Gel and Biopsy-Site Asepsis 133

Table 1 Mean, standard deviations, median values for colony-forming attributable to a mechanical cleaning effect of gel, which
units for Groups 1 (nonsterile gel) and 2 (sterile saline) has a more viscous consistency than saline.
Group 1 Group 2 Health Canada has recommended that sterile gel be used
for US-guided biopsies, including transrectal biopsies [8].
Prebiopsy 73,833 66,034 (45,500) 72,767 63,096 (58,000) It is, however, unclear whether the recommendation is for
Postbiopsy 5,067 20,846 (0) 13,533 31,367 (0) sterile gel to be placed between the probes sterile cover
Change 90% 39% (100) 89% 23% (100) and the disinfected skin at the biopsy site or between the
probes sterile cover and the probe itself. The use of sterile
gel between the probes sterile cover and the disinfected
significant difference between groups (p = 0.58). Twenty- skin is warranted and should be considered obligatory. In
three percent (7 of 30) subjects for whom nonsterile gel biopsies other than transrectal, the covers integrity is not
was used (Group 1) had postbiopsy bacterial growth, and disturbed; therefore, the use of sterile gel between the
47% (14 of 30) subjects for whom sterile saline was used probe and the probes sterile cover is not necessary, as the
(Group 2) had postbiopsy bacterial growth. Thirty-three gel does not contact the skin or affect the asepsis of the
percent (13 of 40) females had postbiopsy bacterial growth, biopsy site. In transrectal biopsies, however, the use of
and 40% (8 of 20) males had postbiopsy bacterial growth. sterile gel between the probes sterile cover and the probe
Thirty-eight percent (20 of 52) of subjects with no coin- does not seem necessary. Core biopsies of prostates are
cidental disease had postbiopsy bacterial growth, and 13% usually obtained from multiple quadrants, with the covers
(1 of 8) of subjects with coincidental disease had postbi- integrity being broken by the multiple needle passes, which
opsy bacterial growth. Fifty percent (8 of 16) of subjects contaminate the sterile gel and the needle with rectal flora.
whose biopsies were in the breast and abdomen nonex- Although our study concerned the use of nonsterile gel
posed areas had postbiopsy bacterial growth, and 30% (13 prior to biopsy, rather than during the biopsy procedure, we
of 44) of subjects whose biopsies were in the head and neck think that nonsterile gel applied directly to the skin prior to
exposed areas had postbiopsy bacterial growth. Fifty-six biopsy would be more important in achieving asepsis than
percent (9 of 16) of subjects whose biopsies involved local would be using gel between the probe and sterile cover.
factors (skin fold, skin tag, and hair) had postbiopsy bac- Because of this and the results of our study (which dem-
terial growth, and 27% (12 of 44) of subjects whose onstrated that nonsterile gel usage before biopsy does not
biopsies did not involve local factors had postbiopsy bac- affect asepsis), we think that the use of sterile gel between
terial growth. The following odds ratios (adjusted for the the probe and the sterile cover for biopsies needs to be
other variables) and their 95% confidence intervals were reevaluated. The usage of sterile gel in urologic procedures
calculated: (1) Group (2.9 [0.811.1]; p = 0.10); (2) gender either for anesthesia or for prevention of urinary tract
(1.2 [0.35.2]; p = 0.78); (3) coincidental disease (7.6 infection remains controversial and is beyond the scope of
[0.9166.7]; p = 0.09); (4) biopsy-site location (6.2 [1.4 our study. If nonsterile gel could be used between the probe
31.3]; p = 0.02); and (5) local factors (7.0 [1.636.0]; and the sterile cover for biopsies, this would reduce the
p = 0.01). No bacterial growth resulted from swabs of time and cost required for performing these biopsies.
nonsterile gel, povidine-iodine, and sterile saline. No ery- In our study, there was more postbiopsy bacterial growth
thema, edema, or focal pain was observed at the biopsy in the nonexposed areas of the breast and abdomen than in the
sites 3 and 5 days after the biopsies had been performed. exposed areas of the head and neck. There were also more
bacteria at biopsy sites, where there were local factors such
as hair, skin folds, and skin tags and in nonexposed areas such
Discussion as the breast and abdomen. Our results are in agreement with
the findings of others, who demonstrated that there were
The aim of preinterventional antiseptic procedures is to more bacteria at axillary and inguinal regions and that
reduce the bacterial count as far as possible, as complete alcohol wipes of wet and hairy sites before US examination
eradication of bacteria is impossible with currently avail- were more effective in reducing bacterial counts than were
able skin-disinfecting methods [9].This study has alcohol wipes of probes [5].
determined the effect of prebiopsy nonsterile gel usage on Previous studies have mainly focused on probe cleaning
biopsy site asepsis. We demonstrated that the use of non- methods [25] and gel contamination [1, 6, 7]. We are
sterile ultrasonography gel prior to biopsy does not affect aware of no publication about effective skin-cleaning
the asepsis of the biopsy site. There was a slight but non- methods to remove the residual US gel before asepsis of
significant tendency for there to be less bacterial growth the biopsy site. In our study, we used double paper wipes to
associated with nonsterile gel usage than with sterile saline remove residual, nonsterile gel from the skin. The double-
usage. This tendency, which might seem odd, may be paper wipe method that we used was adapted from probe-

123
134 K. Gurel et al.: Ultrasonography Gel and Biopsy-Site Asepsis

cleaning methods, which reduce bacterial transmission by a 3. Muradali D, Gold WL, Phillips A, Wilson S (1995) Can ultrasound
factor of about 100 [4]. probes and coupling gel be a source of nosocomial infection in
patients undergoing sonography? An in vivo and in vitro study.
There are some weaknesses to our study. First, our AJR 164(6):15211524
sample size was small. Second, we did not test the effect of 4. Fowler C, McCracken D (1999) US probes: risk of cross infection
mechanical cleaning. Third, in our study, coincidental and ways to reduce it-comparison of cleaning methods. Radiology
disease (diabetes, chronic renal failure, and malignancy) 213(1):299300
5. Karadeniz YM, Kilic D, Kara Altan S, Altinok D, Guney S
did not affect biopsy-site sterility. This finding is contra- (2001) Evaluation of the role of ultrasound machines as a
dictory to that of others (1, 7) and may be attributable to the source of nosocomial and cross-infection. Invest Radiol
small number of our patient who had coincidental disease. 36(9):554558
In conclusion, the use of noncontaminated, nonsterile 6. Gaillot O, Maruejouls C, Abachin E, et al. Nosocomial outbreak of
Klebsiella pneumoniae producing SHV-5 extended-spectrum beta-
ultrasonography gel prior to biopsy does not affect the lactamase, originating from a contaminated ultrasonography
asepsis of the biopsy site. coupling gel. J Clin Microbiol 36(5):13571360
7. Weist K, Wendt C, Petersen LR, Versmold H, Ruden H (2000) An
outbreak of pyodermas among neonates caused by ultrasound gel
References contaminated with methicillin-susceptible Staphylococcus aureus.
Infect Control Hosp Epidemiol 21(12):761764
8. Health Canada (2004) Notice to hospitals: important safety
1. Hutchinson J, Runge W, Mulvey M, et al. (2004) Burkholderia information on ultrasound and medical gels. Available at:
cepacia infections associated with intrinsically contaminated http://www.hc-sc.gc.ca/dhp-mps/alt_formats/hpfb-dgpsa/pdf/
ultrasound gel: the role of microbial degradation of parabens. medeff/ultrasound_nth-ah_e.pdf; accessed April 4, 2005
Infect Control Hosp Epidemiol 25(4):291296 9. Arata T, Kamitani M, Miyai T, Ito M (1997) Antiseptic effects at
2. Spencer P, Spencer RC (1988) Ultrasound scanning of post- injection sites. Dermatology 195(Suppl 2):107110
operative woundsthe risks of cross-infection. Clin Radiol
39(3):245246

123
Cardiovasc Intervent Radiol (2008) 31:135141
DOI 10.1007/s00270-007-9197-0

CLINICAL INVESTIGATION

Empyema and Effusion: Outcome of Image-Guided Small-Bore


Catheter Drainage
A. N. Keeling S. Leong P. M. Logan
M. J. Lee

Received: 8 May 2007 / Accepted: 11 September 2007 / Published online: 18 October 2007
 Springer Science+Business Media, LLC 2007

Abstract Empyema and complicated pleural effusion Keywords Small-bore chest drain  Empyema 
represent common medical problems. Current treatment Pleural effusion  Image-guided chest catheter
options are multiple. The purpose of this study was to
access the outcome of image-guided, small-bore catheter
drainage of empyema and effusion. We evaluated 93 small- Introduction
bore catheters in 82 patients with pleural effusion (n = 30)
or empyema (n = 52), over a 2-year period. Image guid- Empyema and complicated pleural effusions represent
ance was with ultrasound (US; n = 56) and CT (n = 37). common medical problems, with a mortality of up to 20%
All patients were followed clinically, with catheter dwell [1]. Hippocrates first described rib resection and open
times, catheter outcome, pleural fluid outcome, reinsertion intercostal drainage as a treatment modality for empyemas
rates, and need for urokinase or surgery recorded. Ninety- in 400 BC [2]. Since then, current chest drainage tech-
three small-bore chest drains (mean=10.2 Fr; range, 8.2 niques have become less invasive, avoiding the need for
12.2 Fr) were inserted, with an average dwell time of 7.81 thoracotomy and general anesthesia.
days for empyemas and 7.14 days for effusions (p [ 0.05). Difficulty in differentiating empyema from simple
Elective removal rates (73% empyema vs 86% effusions) pleural effusion with imaging alone often occurs in the
and dislodgement rates (12% empyema vs 13% effusions) clinical setting. As a result, needle aspiration with or
were similar for both groups. Eight percent of catheters without formal chest drainage may be necessary in many
became blocked and 17% necessitated reinsertion in em- circumstances [1]. Traditionally, surgeons inserted large-
pyemas, with no catheters blocked or requiring reinsertion bore chest drains (2830 Fr) with underwater seals, to
in effusions (p \ 0.05). Thirty-two patients (51%) required achieve decompression of pleural fluid collections. Recent
urokinase in the empyema group, versus 2 patients (6%) in literature suggests that small-bore drains may be adequate,
the effusion group (p \ 0.05). All treatment failures, certainly for treatment of pleural effusions [3, 4], with
requiring surgery, occurred in the empyema group (19%; image guidance playing an important role [5]. However,
n = 12; p \ 0.05). In conclusion, noninfected pleural col- the debate continues on the effectiveness of small-bore
lections are adequately treated with small-bore catheters, catheters in the setting of documented empyemas, with
however, empyemas have a failure rate of 19%. The many authors not advocating their use at all [3, 6, 7].
threshold for using urokinase and larger-bore catheters Therefore the purpose of this study was to retrospec-
should be low in empyema. tively review all patients referred to the Interventional
Radiology Department for image-guided percutaneous
small-bore chest drain insertion, for both simple pleural
effusions and infected empyemas, in order to determine
success rates, procedure complications, catheter dwell
A. N. Keeling  S. Leong  P. M. Logan  M. J. Lee (&)
times, and clinical outcome. Our aim was to determine if
Department of Academic Radiology, Beaumont Hospital, Dublin
9, Ireland small-bore catheters have any role in the empyema setting
e-mail: mlee@rcsi.ie as a first-line treatment modality.

123
136 A. N. Keeling et al.: Outcome of Image-Guided Small-Bore Catheter Drainage

Materials and Methods indications of empyema (n = 52; 63%) and simple pleural
effusion (n = 30; 37%), with US guidance used in 60%
Patients (n = 56) of drainages and CT guidance employed in the
remainder (40%; n = 37). Diagnosis of empyema versus
We evaluated the outcome of 93 small-bore catheters in 82 simple noninfected pleural effusion was determined on the
patients with pleural effusion or empyema, retrospectively, basis of pleural fluid sampling either by prior tap or at the
over 2-year period from April 1, 2004, to March 31, 2006. time of fluid drainage. Standard parameters were used,
All consecutive patients, with pleural fluid collections, including pH\7.2, glucose\2.2 mmol/L, and LDH[1000
referred to our Interventional Radiology service for small- IU/L, to diagnose empyema [6]. Simple nonpurulent pleural
bore catheter chest drainage were included. Patients were effusions were due to a variety of causes including conges-
sourced from the Interventional Radiology procedure log- tive cardiac failure (CCF; n = 12; 15%), malignant effusion
book, with follow-up data obtained from the patients (n = 8; 10%), trauma (n = 5; 6%), mesothelioma (n = 1;
medical notes. The inclusion criteria consisted of patients 1%), and bile leak (n = 1; 1%) (see Fig. 1). Patients initially
who had either simple pleural effusion or empyema, presented with a combination of clinical symptoms, com-
causing an impaired clinical condition as a result of prising dyspnea (n = 72; 88%), cough (n = 65; 79%), fever
hypoxia or sepsis, and thus were deemed to necessitate (n = 49; 59%), sputum production (n = 31; 38%), pleuritic
catheter drainage with an image-guided small-bore cathe- chest pain (n = 26; 32%), and weight loss (n = 12; 14%).
ter. The exclusion criteria were patients who were referred Males accounted for 61% (n = 50) of the study population,
for catheter drainage but only underwent needle thoraco- with an overall average age of 66 years and the wide age
centesis, with no drain inserted. Prior to chest catheter range of 2784 years. Average white cell count was higher in
insertion, a full clinical history was taken, including current the empyema group (17.2) versus the effusion group (12.1),
symptoms with duration, clinical examination focusing on ESR values were similar (empyema, ESR = 65; effusion,
respiratory system, laboratory indexes, particularly full ESR=52), and patients in the empyema group had higher
blood count (FBC), arterial blood gases (ABGs), erythro- average temperatures prior to chest drain insertion (empy-
cyte sedimentation rate (ESR), and coagulation screen, ema, 38.1C; effusion, 37C).
along with all chest imaging including chest x-ray (CXR),
chest ultrasound (US), and chest computed tomography
(CT). Following chest catheter insertion, patients were Catheter Insertion Technique
routinely followed up on the inpatient hospital ward by the
respiratory team and interventional radiologists. Clinical Prior chest imaging (CXR, US, or CT) was available in all
outcomes, including resolution or improvement in pleural patients referred for chest catheter insertion; this, in com-
fluid collection and catheter dwell times, were documented bination with clinical examination, determined the side to
prior to discharge. Following catheter removal and patient be drained. Image guidance, with both US and CT, was
discharge, all patients were seen in the respiratory outpa- used for all drainages to ensure accurate catheter placement
tient department as part of routine follow-up. within the fluid collection or to ensure placement within the
Eighty two patients underwent 93 small-bore chest cath- largest fluid pocket, if the collection was loculated. US was
eter insertions during this 24-month period for the clinical the primary guidance modality employed (60%), with CT
reserved for more difficult cases (small empyemas, large
Mesothelioma, patients) and as a result of one operators preference.
Malignant Effusions, 10% 1% TB, 4% Trauma, Two catheter models were used during the course of the
6% study period, according to operator preference: the UreSil
pigtail catheter (812 Fr, UreSil, Hydrophilic Coated GP
CCF,
15%
General-Purpose Drainage Catheter; UreSil, L.P., Skokie,
IL, USA) and the Mac-Loc pigtail catheter (8.212.2 Fr,
Ultrathane Mac-Loc Multipurpose Drainage Catheter;
Bile Leak, 1% Cook, William Cook Europe, Bjaeverskov, Denmark).
Empyema, Small-bore chest catheters were defined as those 12.2 Fr.
63%
Patients gave written informed consent themselves, or if
they were not deemed fit to give their own consent, this was
obtained from the next of kin, along with a witness. All
catheters were inserted under sterile technique by experi-
Fig. 1 Clinical indications necessitating small-bore catheter chest enced interventional radiologists or under their supervision
drainage. by specialist radiology registrars. Catheters were inserted

123
A. N. Keeling et al.: Outcome of Image-Guided Small-Bore Catheter Drainage 137

either in a dedicated IR suite or in the CT department, Table 1 Procedure details


depending on the image guidance method elected. Patients Empyema Effusion
were placed in the sitting position for US guidance, being (n = 52; 63%) (n = 30; 37%)
supine/oblique for CT guidance, with continuous oxygen
saturation, pulse, and blood pressure monitoring. US or CT No. of patients 52 30
was used to mark the skin at the most appropriate intercostal No. of catheters 63 30
space for catheter insertion. Patients, skin was cleansed with Catheter size 10.2 Fr 10.2 Fr
bethadine (chlorhexadine, if allergic to iodine) and patients Pleural space: side Right: 27 (52%) Right: 15 (50%)
were draped with sterile surgical drapes as standard. Local No. complications 2 (3%) 1 (3%)
anesthesia (1% lignocaine) was administered subcutane-
ously above the rib below, within the intercostal space. A
small incision was made and blunt dissection was performed, Results
prior to insertion of a vascular access needle (19 G 7 cm;
Cook, William Cook Europe) under imaging guidance. In Over the course of the 2-year study period, 93 small-bore
large patients a Mitty Pollack needle (18 G 14 cm, outer chest catheters were inserted using imaging guidance, into
core; 22 G 22 cm, inner core; Cook Urological Inc., 82 patients with both empyema (63%) and simple nonin-
Spencer, IN, USA) was used for access. A small amount of fected effusion (37%). Sixty-three catheters were placed in
pleural fluid was aspirated for later analysis. A regular J 52 patients with empyemas, while 30 catheters were
guidewire (Standard Wire Guide, 145 cm, 0.035 in.; Cook, inserted in 30 patients with simple pleural effusions
William Cook Europe) was inserted into the pleural fluid (Table 1). The average catheter size for both groups was
collection via the vascular access needle to enable serial 10.2 Fr (range, 812.2 Fr), with the right side drained to a
dilatation of the access point. Catheter size was largely slightly higher, nonsignificant, rate (Table 1).
determined by operator preference, however, if frank pus
was aspirated, some operators elected to insert a larger (12-
Fr) drain. All catheters were sutured in situ, stuck down with Insertion Complications (Table 1)
a Drain Fix (Unomedical Ltd., Stonehouse, Great Britian),
and the drainage bag (Cook Urological Inc., Spencer, IN, An overall minor insertion complication rate of 3% (n = 3)
USA) pinned to the patients clothing. Catheters were left on was found, with two insertion complications in the empyema
free drainage until a maximum of 1 L drained in a 24-h period group and one in the effusion group. One of the patients
(if [1 L/day, the catheter was clamped until the next 24-h developed a small hematoma at the insertion site of the
period, to avoid potential pulmonary edema). Any catheters catheter for an empyema. This subcutaneous hematoma
that failed to drain were flushed with normal saline and required no blood transfusion, did not cause a significant
aspirated under aseptic technique by the nursing staff on the drop in hemoglobin, and most likely occurred due to a co-
ward on an 8-hourly basis. If saline flushing was not suc- agulopathic state as a result of sepsis (International
cessful, intracatheter fibrinolytic therapy, in the form of Normalized Ratio [INR] =2.1). A small pneumothorax
100,000 U urokinase once daily, was instituted. occurred following CT drainage of another empyema. The
In all cases there was close physicianradiologist liaison patient was somewhat dyspneic but was managed with the
with respect to catheter management and catheter dwell small-bore catheter alone. The final complication occurred in
times. At our institution, there is a dedicated weekly Chest a large patient with an effusion drained under US guidance.
Conference for case discussion and follow-up, with a Minimal fluid was obtained despite a large effusion. CT was
dedicated interventional fellow for daily team liaison and performed on the following day and the drain was malposi-
patient review if deemed necessary. All catheter dwell tioned within the subcutaneous tissues. This drain was
times, fluid collection outcome, further catheter insertion resited under CT guidance.
(small or large bore), use of fibrinolytic, necessity for
surgery, and clinical outcome were recorded in the
patients medical notes. Small-Bore Catheter Outcome Secondary to Catheter
Size (Tables 2 and 3)

Statistical Analysis Three catheter sizes were employed: 8 Fr (n = 33; 36%),


10 Fr (n = 44; 47%), and 12 Fr (n = 16; 17%). While there
The statistical software package used for all of the analysis was no significant difference in catheter or pleural fluid
was Stata (version 8; College Station, TX, USA). A p value outcome with respect to catheter size (small numbers in
\0.05 was deemed to be significant. each group), there was a trend to shorter catheter duration,

123
138 A. N. Keeling et al.: Outcome of Image-Guided Small-Bore Catheter Drainage

reduced blockage, and better fluid resolution with 12-Fr more effusions demonstrating complete resolution on fol-
catheters in empyemas. However, catheter size did not low-up CXR than empyemas: 57% versus 40%
have any relation with the need for surgical decortication in respectively (p \ 0.05). Overall, 35% of patients had a
the empyema group. small residual costophrenic angle blunting on CXR, likely
representing pleural thickening and not pleural fluid, with
38% of empyemas and 30% of effusions yielding this
Catheter Outcome for Empyema versus Effusion residual costophrenic blunting. Overall small-bore catheter
(Table 2) failure rate was 13% for effusions (no change in size of
effusion), versus the much higher rate of 23% for empy-
Similar catheter dwell times were obtained for both emas (no change in size or worse). While a similar number
empyema and effusion groups, with the average dwell time of empyemas and effusions remained unchanged in size
being 7.81 and 7.14 days, respectively. Elective removal following drainage, significantly more empyemas deterio-
rates demonstrated no significant differences in the two rated despite catheter drainage. Repeat drainages were
groups, with 73% of empyema catheters and 86% of necessitated in the empyema group only. Ultimate percu-
effusion catheters intentionally removed, following ade- taneous drainage failure requiring surgical decortication
quate fluid drainage. Inadvertent catheter dislodgement occurred only in the empyema group and accounted for
remained a problem for both groups but showed no sig- 19% of patients, with these 10 patients needing surgical
nificant difference. Significantly more catheters blocked in intervention to adequately treat their empyema. Of these 10
the empyema group compared with the effusion group, 8% patients with catheter failures, 50% (n = 5) had their
(n = 5; n = 3 were 8 Fr, n = 2 were 10 Fr) versus 0%, catheters upsized in an attempt to facilitate further drain-
respectively (p \ 0.05). A number of catheters required age, with 3 patients receiving large-bore (14-Fr) chest
replacement during the course of the study, with all of the catheters (1 patient had a surgical drain inserted) and the
repeat drainages occurring in the empyema group (17%; other 2 patients having 12-Fr catheters placed. The
n = 11; n = 5 secondary to blockage, n = 6 secondary to remaining 50% of failures did not have any catheter up-
dislodgement)). Most patients required only one more sizing performed and proceeded to surgery due to
drain, however, one patient required two further catheters deteriorating clinical condition as a result of blocked
and one patient required three more catheters. Just over catheters failing to drain the pus (four catheters were 10 Fr,
half of the empyema group required the fibrinolytic uro- with one being 8 Fr). Forty percent of patients who failed
kinase to break down loculations and assist in pleural fluid catheter drainage had loculated empyemas. No patients
drainage, a significantly higher number than for nonin- with effusions required surgical decortication.
fected pleural fluid collections (p \ 0.05).

Discussion
Pleural Fluid Outcome (Table 3)
The British Thoracic Society (BTS) Guidelines have rec-
Overall, 46% pleural fluid collections resolved completely ommended a number of clinical indications where chest
following small-bore catheter drainage, with significantly tube drainage is warranted in the setting of pleural fluid
collections, both infected and noninfected [6]. However, at

Table 2 Small-bore catheter outcome following image-guided


insertion into pleural fluid collection Table 3 Pleural fluid collection outcome, following small-bore chest
catheter drainage
8Fr/10Fr/12Fr
8 Fr/10 Fr/12 Fr
Empyema Effusion
(n = 63) (n = 30) Empyema Effusion
(n = 52) (n = 30)
No. catheters 15/33/15 18/11/1
Catheter dwell time 8.22/8.02/7.19 7.59/6.99/6.84 No. patients 12/27/13 18/11/1
days days Resolved 3/10/8 (40%) 9/7/1 (57%) p \ 0.05
Removed electively 6/27/13 (73%) 16/9/1 (86%) p [ 0.05 Residual fluid 4/12/4 (38%) 6/3/0 (30%) p [ 0.05
Dislodged 2/4/2 (12%) 2/2/0 (13%) p [ 0.05 Unchanged 2/3/1 (11.5%) 3/1/0 (13%) p [ 0.05
Blocked 3/2/0 (8%) 0/0/0 p \ 0.05 Worse 3/2.0 (10%) 0/0/0 p \ 0.05
Reinserted 3/6/2 (17%) 0/0/0 p \ 0.05 Repeat drainage 4/3/1 (15%) 0/0/0 p \ 0.05
Urokinase 9/21/2 (51%) 1/1/0 (6%) p \ 0.05 Surgical decortication 3/5/2 (19%) 0/0/0 p \ 0.05

123
A. N. Keeling et al.: Outcome of Image-Guided Small-Bore Catheter Drainage 139

present there is no consensus among the BTS, the Ameri- documented in the literature, thus this is an accepted
can College of Chest Physicians, and other groups on the established technique [1416, 21, 23].
optimal chest drain size for treating both pleural effusions Dwell times for pigtail catheters, in this study, demon-
and empyemas in adults [1, 6]. The guidelines for children strated no significant difference in the pleural effusion
do state that small-bore drains, including pigtail catheters, group compared with the empyema group. Pus, by nature,
should be used whenever possible, in order to minimize is more viscous than simple fluid, thus one would expect a
patient discomfort, thus allowing better ambulation and more viscous fluid to flow through a small tube at a slower
increasing speed of recovery [8]. In the literature, there is a rate. This fact was reported by Park et al., who found that a
lack of large randomized controlled trials comparing large- smaller tube drains thick secretions more slowly [24].
bore chest drains with smaller-bore drains for pleural fluid Neither our study nor that by Horsley et al. found this to be
collection evacuation. There are a number of case series true, as we demonstrated no significant correlation between
but numbers for empyemas remain low. Therefore, our catheter size and failure to drain viscous empyema col-
study is useful in examining patients with empyema to lections [3]. In our population, the similar catheter dwell
determine the outcome of small-bore catheters. times may be explained by the aggressive approach to
Pigtail catheter insertion success rate was 99% in our follow-up of chest drains within empyemas, with further
patient group. We feel that this success rate was due to the management, in the form of saline flushing, fibrinolytic, or
image guidance methods employed, namely, US and CT. upsizing the catheter to 12 or 14 Fr, instituted if pleural
Prior chest imaging also plays a vital role in catheter place- fluid drainage was inadequate.
ment, as the correct diagnosis of a pleural fluid collection Intrapleural fibrinolytic use is recommended by the BTS
must be made. Plain radiographs, both frontal and lateral, guidelines, for both adults and children, and has been
may be sufficient to enable localization in both adults and shown to improve radiological outcome, however, it is not
children [911]. US of the pleural space enables diagnosis, clearly documented whether fibrinolytic use results in
fluid characterization [12], and has the added advantage of decreased mortality or reduction in the need for surgery [6,
guiding aspiration [9, 13] and pigtail catheter placement [14 8]. Tillett et al. first described the use of streptokinase
16]. Lack of radiation exposure for the patient, its portable within the pleural space to lyse fibrinous septations in 1949
nature, and its availability are other useful features of US. It [25]. A number of small studies have all reported improved
has been found that US will not reliably predict the outcome pleural fluid drainage, assessed radiologically, following
of chest drainage or determine those patients who will intrapleural streptokinase or urokinase, however, end
eventually need surgical decortication [17]. CT, often points such as mortality rates and thoracotomy frequency
reserved for the more challenging diagnostic dilemmas, were not assessed [2629]. Reported complications of
yields very eloquent, often classical, images [1820], but at a fibrinolytics are difficult to assess due to the underlying
cost, in terms of finance, radiation dose, and accurate diag- sepsis from the empyema, as fever is the most common
nosis of empyema versus effusion, as not all cases yield immunological adverse effect [30]. Various case reports of
classical imaging features. CT is also used as a modality to other adverse effects exist in the literature, such as local
guide catheter placement [21], to diagnose any coexisting pleural hemorrhage [31, 32], systemic hemorrhage [33],
lung parenchymal pathology, and to outline pleural space and cardiac arrhythmia [34], however, the true incidence of
anatomy prior to any surgical procedures [8]. In our study, the adverse effects is difficult to calculate in the absence of
while predrainage CT thorax was obtained on patients with a large-scale randomized trial. A recent review by Bouros
more complicated clinical presentations, i.e., postoperative, et al. states that to date there is no high-grade evidence,
diagnostic dilemma, trauma, and malignancy, CT guidance from currently available literature, to support the wide-
was used as an operator preference. spread use of intrapleural fibrinolytics in patients with
We report the low insertion complication rate of 3%, parapneumonic effusions or unilocular empyemas, how-
which is in concordance with previously published rates [3, ever, there is a role for use in poor surgical candidates who
21, 22], confirming that small-bore catheters are safe and fail chest drain therapy [35]. The BTS and the Medical
technically feasible to insert. Interestingly, image guidance Research Council are currently conducting a multicenter
used in our study yielded a much lower complication rate, trial to assess the efficacy of intrapleural streptokinase [6];
3%, than blind insertion, which has a higher total compli- this study may then allow for an evidence-based set of
cation rate of 22% [3]. This result is not surprising, as with guidelines. At present, it is our clinical practice to use
US or CT guidance one can directly visualize the needle tip intrapleural urokinase in the empyema setting with failure
during insertion into the pleural fluid collection, thus of fluid drainage before surgical referral is sought; this
avoiding misplacement, pneumothorax, injury to adjacent perhaps reflects our elderly population.
vessel, or failure of insertion. Accurate catheter placement Pigtail catheter failure remains a significant problem in
as a result of image guidance has been previously the treatment of both infected and noninfected pleural fluid

123
140 A. N. Keeling et al.: Outcome of Image-Guided Small-Bore Catheter Drainage

collections, and our population is no exception, with a 2. Hipppocrates (1847) Genuine works of Hippocrates. Anderer F,
failure rate of 13% for effusions (no change in size of trans. Syndeham Society, London, UK: 1847
3. Horsley A, Jones L, White J, Henry M (2006) Efficacy and
effusion), versus the much higher rate of 23% for empy- complications of small-bore, wire guided chest drains. Chest
emas (no change in size or worse). These figures are more 130:18571863
favorable than those of Horsley et al., at 37% [3], and 4. Marom EM, Palmer SM, Erasmus JJ, et al. (2003) Pleural effu-
Shankar et al., at 37.5% [23], but worse than those quoted sions in lung transplant recipients: image-guided small-bore
catheter drainage. Radiology 228:241245
by Moulton et al., at 6% [21]. Small-bore catheter failure 5. Silverman SG, Mueller PR, Saini S, et al. (1988) Thoracic
necessitating a thoracic surgical procedure, VATS or tho- empyema: management with image-guided catheter drainage.
racotomy, accounted for 19% of empyemas in our series. Radiology 169:59
Surgical rates from 6% to 20% are quoted in the literature 6. Davies CW, Gleeson FV, Davies RJ (2003) BTS guidelines for
the management of pleural infection. Thorax 58 (Suppl 2):ii18
[3, 21]. ii28
Despite suturing all catheters to the patients skin, 7. Hyde J, Sykes T, Graham T (1997) Reducing morbidity from
securing with an adhesive dressing, and pinning the chest drains. BMJ 314:914915
drainage bag to the patients clothing, there was a catheter 8. Balfour-Lynn IM, Abrahamson E, Cohen G, et al. (2005) BTS
guidelines for the management of pleural infection in children.
dislodgement rate of 13% for effusions and 12% for em- Thorax 60:121
pyemas, which is higher than the 8% quoted by a group 9. Stavas J, Van Sonnenberg E, Casola G, et al. (1987) Percutaneous
who only suture the catheter in situ, with no other drain drainage of infected and noninfected pleural thoracic fluid col-
fixation [3]. This fact highlights the need for thorough lections. J Thorac Imaging 2:8087
10. Light RW, Girard WM, Jenkinson SG, et al. (1980) Parapneu-
patient and nursing staff education to avoid inadvertent monic effusions. Am J Med 69:507512
catheter dislodgement. 11. King S, Thompson A (2002) Radiological perspectives in
The main limitation of our study is its retrospective empyema. Br Med Bull 61:203214
nature and the fact that chest catheter size was determined 12. Yang PC, Luh KT, Chang DB, et al. (1992)Value of sonography
in determining the nature of pleural effusion: analysis of 320
based on operator preference rather than simply on con- cases. AJR 159:2933
sistency of pleural fluid aspirate, which may explain the 13. Eibenberger KL, Dock WI, Ammann ME, et al. (1994) Quanti-
average size of 10.2 Fr in both the empyema group and the fication of pleural effusions: sonography versus radiography.
effusion group. The fact that there was no significant dif- Radiology 191:681684
14. Ulmer JL, Choplin RH, Reed JC (1991) Image-guided catheter
ference in catheter size between the two groups strengthens drainage of the infected pleural space. J Thorac Imaging 6:6573
the argument for a trial of small-bore catheters in treating 15. Westcott JL (1985) Percutaneous catheter drainage of pleural
empyema, as if catheters larger than 12 Fr were used as effusion and empyema. AJR 144:11891193
first line, if the pleural aspirate was frank pus, then perhaps 16. Merriam MA, Cronan JJ, Dorfman GS, et al. (1988) Radiologi-
cally guided percutaneous catheter drainage of pleural fluid
the failure rate would decrease. A larger prospective ran- collections. AJR 151:11131116
domized trial of catheter size in treating empyema is 17. Kearney SE, Davies CWH, Davies RJO, et al. (2000) Computed
warranted to further evaluate the role of small-bore cath- tomography and ultrasound in parapneumonic effusions and
eters in this important clinical scenario. empyema. Clin Radiol 55:542545
18. Aquino SL, Webb SR, Gushiken BJ (1994) Pleural exudates and
In conclusion, pleural space fluid collections are com- transudates: diagnosis with contrast-enhanced CT. Radiology
mon clinical entities that interventional radiologists can 192:803808
accurately diagnose as well as successfully treat. Nonin- 19. Waite RJ, Carbonneau RJ, Balikian JP, et al. (1990) Parietal
fected pleural fluid collections are adequately treated with pleural changes in empyema: appearances at CT. Radiology
175:145150
small-bore catheters, however, empyemas have a definite 20. Takasugi JE, Godwin JD, Teefey SA (1991)The extrapleural fat
failure rate. The threshold for using urokinase and larger- in empyema: CT appearance. Br J Radiol 64:580583
bore catheters should be low in empyema. We recommend 21. Moulton JS, Benkert RE, Weisiger KH, Chambers JA (1995)
using an 8-Fr catheter as first line for simple effusions, with Treatment of complicated pleural fluid collections with image-
guided drainage and intracavitary urokinase. Chest 108:1252
a 10-Fr catheter as first line for empyemas, with scope to 1259
upsize to 12-Fr or larger if drainage is poor. The use of 22. Parulekar W, Di Primio G, Matzinger F, Dennie C, Bociek G
large-bore chest catheters ([12 Fr) versus small-bore (2001) Use of small-bore vs large-bore chest tubes for treatment
catheters in empyema should be addressed in a large-scale of malignant pleural effusions. Chest 120:1925
23. Shankar S, Gulati M, Kang M, Gupta S, Suri S (2000) Image-
randomized controlled trial. guided percutaneous drainage of thoracic empyema: Can
sonography predict the outcome? Eur Radiol 10:495499
24. Park JK, Kraus FC, Haaga JR (1993) Fluid flow during percu-
References taneous drainage procedures: an in vitro study of the effects of
fluid viscosity, catheter size, and adjunctive urokinase. AJR
1. Colice GL, Curtis A, Deslauriers J, et al. (2000) Medical and 160:165169
surgical treatment of parapneumonic effusions: an evidence- 25. Tillett WS, Sherry S (1949) The effect in patients of streptococcal
based guideline. Chest 118:11581171 fibrinolysin (streptokinase) and streptococcal desoxyribonuclease

123
A. N. Keeling et al.: Outcome of Image-Guided Small-Bore Catheter Drainage 141

on fibrinous, purulent, and sanguinous pleural exudations. J Clin 30. Chapman SJ, Davies RJ. (2004) The management of pleural
Invest 28:173190 space infections. Respirology ;9:411
26. Bilaceroglu S, Cagerici U, Cakan A (1997) Management of 31. Temes RT, Follis F, Kessler RM, et al. (1996) Intrapleural fib-
complicated parapneumonic pleural effusions with image-guided rinolytics in management of empyema thoracis. Chest 110:102
drainage and intrapleural urokinase or streptokinase: a controlled 106
randomized trial. Eur Respir J 10:325S 32. Porter J, Banning AP (1998) Intrapleural streptokinase. Thorax
27. Davies RJO, Traill ZC, Gleeson FV (1997) Randomized con- 53:720
trolled trial of intra-pleural streptokinase in community acquired 33. Godley PJ, Bell RC (1984) Major hemorrhage following
pleural infection. Thorax 52:416421 administration of intrapleural streptokinase. Chest 86:486487
28. Bouros D, Schiza S, Tzanakis N, et al. (1999) Intrapleural uro- 34. Alfageme I, Vazquez R (1997) Ventricular fibrillation after
kinase versus normal saline the treatment of complicated intrapleural urokinase. Intens Care Med 23:352
parapneumonic effusions and empyema. Am J Respir Crit Care 35. Bouros D, Tzouvelekis A, Antoniou KM, Heffner JE (2007)
Med 159:3742 Intrapleural fibrinolytic therapy for pleural infection. Pulm
29. Thompson AH, Hull J, Kumar R, et al. (2002) A randomized trial Pharmacol Ther 20:616626
of intrapleural urokinase in the treatment of childhood empyema.
Thorax 57:343347

123
Cardiovasc Intervent Radiol (2008) 31:142148
DOI 10.1007/s00270-007-9179-2

CLINICAL INVESTIGATION

Fluoroscopy-Guided Pull-Through Gastrostomy


M. B. Pitton S. Herber C. Duber

Received: 21 May 2007 / Accepted: 29 July 2007 / Published online: 16 October 2007
Springer Science+Business Media, LLC 2007

Abstract The purpose of this study was to simplify a Five minor complications occurred: one tube loss during
fluoroscopy guided gastrostomy technique using pull-type the passage of the hypopoharynx because of a torn thread,
tubes which are traditionally introduced with gastroscopic one transient small leakage alongside the tube (both suc-
assistance. The stomach was transorally probed with a 5-Fr cessfully treated), and three cases of transient moderate
catheter and a guidewire. A second access was performed local pain without leakage (symptomatic treatment). We
percutaneously through the anterior abdominal and gastric conclude that this fluoroscopy-guided pull-through gas-
wall using an 8-Fr sheath and an 8-Fr guiding catheter. A trostomy technique is easy and safe to perform and may be
duplicated guidewire was introduced through the guiding suggested as a standard procedure for radiological gas-
catheter in order to result in a great custom-made loop trostomies. It combines the ease of the radiological
within the stomach. The transoral guidewire was captured approach with the advantages of the pull-type tube devices,
and tightened with this loop and the guiding catheter, and particularly the benefits of the typical retention plates.
both were subsequently pulled by the transoral guidewire
until the tip of the guiding catheter exited the mouth. A Keywords Enteral feeding 
thread was fed through the guiding catheter for fixation of Fluoroscopy-guided percutaneous gastrostomy 
the pull-type gastrostomy tube. Finally, the fixed tube was Pull-type gastrostomy
pulled through the esophagus into the stomach and through
the abdominal wall until the anterior gastric wall fixed the
retention plate of the tube. Thirty-seven patients (28 male, Introduction
9 female; age, 65.1 14.4 years) with miscellaneous
indications for percutaneous gastrostomies were supplied During the last decades, several percutaneous gastrostomy
with pull-type gastrostomy catheters in a fluoroscopy techniques with either radiological or endoscopic guidance
technique without endoscopic assistance. Twenty-five of have been developed. Percutaneous gastrostomy is partic-
the 37 patients (67.6%) had undergone unsuccessful pre- ularly indicated in head and neck tumors and in
ceding gastroscopically guided PEG attempts because of unresectable tumors of the esophagus or the cardia. It is
tumor stenosis (n = 12) or impossible transillumination of indicated in radiation therapy if dysphagia or mucositis is
the abdominal wall (n = 13). All procedures were techni- anticipated or additional caloric intake is required. More-
cally successful, without major complications. Particularly, over, gastrostomy might be indicated in a variety of
all patients with frustrating gastroscopic attempts were neurological disorders (cerebrovascular accidents, intra-
successfully provided with pull-type gastrostomy tubes. cranial hemorrhage, hypoxic brain injury, or cerebral
palsy). Another indication is the decompression of the
stomach in cases with small bowel obstruction. Compara-
M. B. Pitton (&)  S. Herber  C. Duber tive data have suggested some advantage of the endoscopic
Department of Diagnostic and Interventional Radiology,
technique compared to fluoroscopy-guided techniques,
University Hospital, Johannes Gutenberg University of Mainz,
Langenbeckstrasse 1, 55131 Mainz, Germany which is possibly due to the fact of the retention plates of
e-mail: pitton@radiologie.klinik.uni-mainz.de such pull-type tubes [16]. These retention plates might

123
M. B. Pitton et al.: Fluoroscopy-Guided Pull-Through Gastrostomy 143

a b

2. 4.

1. 3.

Fig. 1 Scheme of the radiological pull-through gastrostomy. a Status tip of the guiding catheter exits the mouth (2). b Passage of the
after percutaneous access to the stomach and placement of a sheath attached gastrostomy tube through the esophagus into the stomach by
and a guiding catheter. Capture of the transoral guidewire and fixation pulling the thread (3) until the cone of the tube exits the abdominal
with the loop using the duplicated transabdominal guidewire (1). wall (4)
Retrograde pulling of the guidewire and the guiding catheter until the

assure improved sealing and better prevention of tube and a hydrophilic guidewire (0.035 in; Radiofocus; Terumo
dislocation compared to the tubes traditionally used for Europe, Leuven, Belgium). Butylscopolaminiumbromide
radiological gastrostomies. This paper describes a (20 mg; Carinopharma, Gronau, Germany) was adminis-
straightforward technique for fluoroscopy-guided percuta- tered intravenously to facilitate gastroparesis and to
neous gastrostomy using a pull-type gastrostomy tube, prevent peristalsis and decompression of the stomach. The
which was traditionally implanted with endoscopic stomach was inflated through the 5-Fr orogastric catheter
assistance. using environmental air until the gastric wall was in the
immediate vicinity of the abdominal wall (Fig. 1).
Posterior-anterior and lateral fluoroscopy assured that no
Methods other organs were interposed at the dedicated puncture site.
Two punctures were performed using a cannula and anchor
Prior to the procedure, all patients or their authorized set (Suture Anchor Set; William Cook Europe, Bjaervers-
persons gave informed consent. Previous abdominal sur- kov, Denmark) in order to achieve transient fixation of the
gery was excluded by patient history. In cases with adverse gastric wall during the procedure. After positioning of the
anatomy after abdominal surgery, CT was used to assure anchors, a 0.035-in., stiff guidewire was introduced
that an appropriate area of the gastric wall could be shifted through the introducer needle and an 8-Fr sheath and a 90-
to the anterior abdominal wall. In other cases, a CT-guided cm, 8-Fr guiding catheter (STR; Cordis, Miami FL, USA)
jejunostomy was performed. Stenosis of the hypopharynx were introduced into the stomach (Fig. 2a). An extralong
or the esophagus, history of failed endoscopic attempts hydrophilic guidewire (0.035 in. [260 cm]; Terumo) was
because of impossible endoscopic passage into the stom- folded and doubled in the midway. The duplicate hydro-
ach, and failed transillumination of the abdominal wall philic guidewire was then advanced through the 8-Fr
were not contraindications. The pull-through gastros- guiding catheter (Fig. 2b) in order to build a large custom-
tomy was performed using the 15-Fr polyurethane tube of made loop within the stomach after deployment from the
a PEG set for endoscopic use (Freka PEG; Fresenius Kabi tip of the guiding catheter (Fig. 2c). The orogastrically
AG, Bad Homburg, Germany). This tube consists of a cone introduced guidewire was captured as the loop and the
with an integrated fixation loop at the proximal end for guiding catheter were tightened on the oropharyngeal
fixation and for passage through the gastric and abdominal guidewire (Fig. 2d). After secure fixation, the wire-loop-
wall. A silicone retention plate at the distal end of the tube guiding catheter combination was retrogradely pulled out
provides fixation at the inner gastric wall. by the orogastric guidewire in order to retrogradely pass
All patients fasted overnight for a minimum of 6 h and the esophagus and hypopharynx until the tip of the guiding
the coagulation status had to be adequately balanced for the catheter exited the mouth (Fig. 2e). After the guidewires
procedure. The posterior wall of the oropharynx was were removed, the thread of the PEG device was easily
anesthetized with an oral xylocaine spray. Intravenous pushed through the lumen of the guiding catheter without
midazolam (25 mg i.v.; Ratiopharm, Ulm, Germany;) and any further support until it exited the tip of the catheter
piritramide (510 mg i.v.; Dipidolor; Jannsen-Cilag, Neuss, (Figs. 2f and g). The loop of the thread was then passed
Germany) provided conscious sedation. The stomach was through the fixation loop of the tube and joined together to
catheterized by the orogastric placement of a 5-Fr catheter achieve secure fixation (Fig. 2h). The tip of the cone was

123
144 M. B. Pitton et al.: Fluoroscopy-Guided Pull-Through Gastrostomy

pulled into the end hole of the guiding catheter and was the guiding catheter and the PEG tube exited the abdominal
fixed in this position. This provided a smooth transition wall (Figs. 1 and 2j and i) and the inner retention plate of
from the tube cone to the guiding catheter and enables a the tube adhered to the anterior gastric wall. A slight
nontraumatic passage even in higher-grade stenosis. For resistance was felt in most cases when the retention plate
the passage, the whole construct (guiding catheter plus passed the proximal and distal esophagus sphincters and
fixed PEG tube) was pulled through the esophagus into the during the passage of higher-grade tumor stenosis. After
stomach and through the gastric and abdominal wall until the tube cone exited the abdominal wall, the cone was cut

123
M. B. Pitton et al.: Fluoroscopy-Guided Pull-Through Gastrostomy 145

b Fig. 2 a After gastroparesis and inflation with environmental air, the technique decribed here was used (28 male, 9 female; age,
stomach has been punctured and an 8-Fr sheath with an 8-Fr guiding 65.1 14.4 years; range, 20.495.2 years). Patient data,
catheter is already introduced. b Lateral projection: orogastric 5-Fr
catheter with hydrophilic guidewire (arrowhead). Introduction of a
underlying diseases, and indications for gastrostomy are
duplicated guidewire through the transabdominal guiding catheter summarized in Table 1. Twenty-five of these 37 patients
(arrow) in order to build a large loop within the stomach. c Large loop (67.6%) had frustrating preceding endoscopic attempts,
from the duplicated guidewire (arrow). Two anchor sutures in situ either because the endoscopes could not be passed through
(arrowhead). d Capture of the transorally introduced guidewire and
tightening of the guiding catheter and the guidewire loop on the
high-grade tumor stenosis into the stomach (n = 12) or
orogastric guidewire. e Retrograde passage of the fixed transoral because transillumination of the abdominal wall was
guidewire and the tightened guiding catheter-guidewire combination impossible (n = 13). The fluoroscopy-guided probing of
through the esophagus until it exits the mouth. f, g Introduction of the the stenosis and the passage of the tubes was successful in
thread of the device into the guiding catheter and pushing of the thread
until it exits the end hole of the guiding catheter (arrows). h, i Fixation of
all cases, even in those cases with high-grade tumor ste-
the tube cone of the gastrostomy catheter and firm adaptation to the end nosis. During follow-up, there was no evidence of tumor
hole of the guiding catheter (arrow). j Assisting the retention plate to seeding at the puncture sites. In one case with neurological
enter the mouth and passing the gastrostomy tube through the esophagus indication (without stenotic tumor), the tube was lost in the
into the stomach by pulling the thread and the fixed devices (guiding
catheter and pull-type gastrostomy tube fixed by the thread) until the
hypopharynx during the procedure because the thread of
cone of the tube exits the abdominal wall and the inner retention plate is the device was torn. In this case, the tube was removed by a
fixed at the anterior gastric wall. k Tube fixation with the outer retention forceps and a new tube was connected and introduced in
plate and fixation of the Luer Lock connector. Fixation of the anchors by the intended way without clinical complications. All
sutures for 510 days (arrowhead). l Confirmation of the result by
contrast injection
intended pull-type gastrostomies were successfully com-
pleted by the technique described here. The mean
fluoroscopy time was 8.1 3.2 min (range, 3.114.3 min)
off and the tube was fixed with the outer fixation plate in and the overall radiation dose was 55.4 40.0 Gycm2
order to achieve a tight connection between the gastric and (range, 3.7153.8 Gycm2) including the learning curve.
the abdominal walls (Fig. 2k). The tube was finally There was no peri-interventional mortality and no severe
equipped with the tube clamp and the Luer Lock connector procedure-related problems. Postinterventionally, however,
for use. In the first 21 cases, the suture anchors were fixed
for 5 to 10 days and cut thereafter. In the remaining
Table 1 Fluoroscopy-guided pull-through gastrostomy
patients, the anchors were detached immediately after the
procedure when we felt that the retention plate would No. Patients 37
provide adequate fixation by itself. The proper position of Age 65.1 14.4
the tube end and retention plate was confirmed by contrast (range, 20.495.2) yr
injection (Fig. 2l). After the procedure, the patients Male/female 28/9
remained fasting for the next 12 h. Feeding began on the Diagnosis and indication
following day after repeated confirmation of the correct Head and neck tumors 7
tube position and if there was no evidence of contrast Esophagus and cardia tumors 5
leakage or any other clinical symptoms. Neurological disorders 9
Radiation therapy 6
Decompression of the stomach 3
Results Others 7
Total 37
Between August 2005 and April 2007 a total of 54 patients Fluoroscopy time 8.1 3.2
were supplied with radiological gastrostomies. Ten of the (range, 3.114.3) min
54 patients (18.5%) were served with the conventional Radiation exposure 55.4 40.0
radiological gastrostomy technique using Cope-loop devi- (range, 3.7153.8) Gy  cm2
ces. In these patients, the gastrostomy served for short-term Follow-up 1.3 2.6
nutrition during radiation therapy of the esophagus or as (range, 0.19.5) mo
decompression of the stomach in cases with duodenal Intrainterventional
technical problem
stenoses or peritoneal carcinosis. Seven of the 54 patients
Loss of tube because 1
(13%) demonstrated adverse anatomy after previous of a torn thread
abdominal surgery or peritoneal carcinosis and were trea- Postinterventional complications
ted with CT-guided direct jejunostomy using Cope-loop
Transient leakage 1
devices for nutrition or for decompression. In 37 of the 54
Moderate local pain 3
patients (68.5%), the fluoroscopy-guided pull-type

123
146 M. B. Pitton et al.: Fluoroscopy-Guided Pull-Through Gastrostomy

one patient demonstrated local abdominal pain and some retention device have been reported in the literature and
contrast leak adjacent to the tube. In this case, the external were used at our institution until introduction of this pull-
retention plate was ineffectively fixed and thereby pre- through gastrostomy technique. T-Fasteners are often used
vented firm adaptation of the gastric wall to the abdominal for gastropexy to facilitate the gastrostomy procedure [14,
wall. After both retention plates were secured, the patient 15]. They were also used for the puncture procedure in our
became asymptomatic during the next few hours. Three technique. Such anchors have to be cut 510 days postin-
patients suffered from moderate local pain at the puncture sertion in order not to induce a foreign body reaction,
sites, one of them with transient slight fever. In these cases, inflammation, or damage to the gastric mucosa. Peristomal
CT scans showed no fluid retention or leakages around the leakages, skin infections, and persistent drainage after
tubes and no contrast extravasation was seen in the fluo- removal of the gastrostomy catheters have been attributed
roscopy control. Those discomforts disappeared within to retained T-fasteners [16]. Meanwhile, we have aban-
some hours with symptomatic treatment alone. The follow- doned suture fixation with gastropexy devices but we
up was 1.3 2.6 months (range, 0.19.5 months), until esteem the transient use of such anchors to prevent loss of
patients were discharged from the hospital or were trans- guidewire position and inadvertent entry into the peritoneal
ferred to peripheral hospitals or nursing homes. space [17]. We did not see any complication in this respect.
The fluoroscopy-guided percutaneous pull-through gas-
trostomy described in this paper combines the advantages
Discussion of both the ease and speediness and the high technical
success rate of the radiological technique with the benefi-
Percutaneous gastrostomy techniques with either radiolog- cial design of the gastrostomy tubes that were traditionally
ical or endoscopic guidance have replaced surgical used with endoscopic assistance. The outstanding benefit of
gastrostomies because of the risk of anesthesia and the the radiological approach is the fact that almost every
increased morbidity associated with laparotomy [26]. stenosis of the hypopharynx or the esophagus can be easily
Radiological percutaneous gastrostomy, gastrojejunostomy, probed with hydrophilic guidewires and the respective
and direct jejunostomy are technically possible in the catheters, which is the essential precondition for the pro-
majority of patients and complex anatomical preconditions cedure. On the other hand, the design of the pull-type
can be managed by the use of additional CT guidance for gastrostomy tubes with the large bores and the typical
the puncture [7, 8]. Therefore, percutaneous radiological retention plates comprise two advantages that were tradi-
access is technically feasible in nearly all of those cases in tionally restricted to the endoscopic technique only. The
whom an endoscopic attempt is not possible, because of large bores have been demonstrated to reduce the occur-
either stenotic tumor, altered anatomical conditions after rence of tube occlusions and the typical retention plates
gastrectomy, or other conditions that prevent transillumi- improve the sealing of the puncture sites and better prevent
nation of the abdominal wall [9]. However, comparative tube dislocation [1]. The results presented describe our
data have shown some advantages of the endoscopic tech- initial experience with this technique. In our department,
nique, which might be determined by the specific design of most requests for this procedure include patients with
such pull-type tubes with the typical retention plates [1]. either stenosis of the hypopharynx or esophagus who have
Those retention plates assure better sealing of the puncture experienced one or more frustrating endoscopic attempts
sites and improved prevention of tube dislocations com- and cases in which endoscopic transillumination of the
pared to the traditionally used radiological devices. abdominal wall was not possible for any reason. Moreover,
For fluoroscopy-guided gastrostomy, multiple gastros- it is self-evident that this technique might serve as a simple
tomy devices have been described in the literature [6]. The alternative to the endoscopic method in all other cases,
initially used Foley catheters have been demonstrated to with the respective low rates of tube dislocation, leakage,
have an increased morbidity rate including tube leakage, pain, and infection [1]. Our data have demonstrated a rather
breakage, migration, proximal small bowel obstruction, low rate of acute peri-interventional complications with
and gastric wall penetration [10, 11]. Gastrostomy buttons respect to local peritonitis and pain, pneumoperitoneum, or
consist of a low-profile catheter with a feeding hub, which tube dislocation. However, reliable comparative data with
can be disconnected between meals and therefore might be the traditional radiologic gastrostomy are not available.
more convenient. Such devices may be designed as bal- Our mean follow-up is 1.3 months, with a maximum fol-
loon-type or mushroom-type, with either a retention low-up of 9.5 months without evidence of complications.
balloon or a mushroom tip for fixation of the device. Both The procedure is straightforward, with a short fluoroscopy
types have been designed for mature tracts, however, a time and an acceptable overall radiation exposure.
primary insertion has also been described [12, 13]. Most Several other attempts have been described for guide-
commonly, 10- to 14-Fr catheters with a Cope loop wire placement by transabdominal access into the stomach

123
M. B. Pitton et al.: Fluoroscopy-Guided Pull-Through Gastrostomy 147

and via the esophagus through the mouth, which is the gastrostomy: a comparison of indications, complications and
essential precondition for push-type and pull-type tube outcomes in 370 patients. Eur J Radiol 56:8490
2. Gauderer MWL, Ponsky JL, Izant RJ (1980) Gastrostomy without
gastrostomies [8, 18, 19]. With these techniques, the laparotomy: a percutaneous endoscopic technique. J Pediatr Surg
guidewires are captured within the stomach or at the gas- 15:872875
troesophageal junction using different types of snares or 3. Pershaw RM (1981) A percutaneous method for inserting a
baskets with given diameters, which can be quite chal- feeding gastrostomy tube. Surg Gynecol Obstet 152:658660
4. Cosentini ER, Sautner T, Gnant M, et al. (1998) Outcomes of
lenging. Alternatively, the esophagus has been cannulated surgical, percutaneous endoscopic, and percutaneous radiological
retrogradely from a transabdominal sheath through the gastrostomies. Arch Surg 133:10761083
esophagus until the guidewire exits the mouth. Subse- 5. Stern JS (1986) Comparison of percutaneous endoscopic gas-
quently, these guidewires were also used for pull-type or trostomy with surgical gastrostomy at a community hospital. Am
J Gastroenterol 207:598603
push-type gastrostomy tubes [20, 21]. Retrograde cannu- 6. Given MF, Hanson JJ, Lee MJ (2005) Interventional radiology
lation of the esophagus, however, can also be very time- techniques for provision of enteral feeding. CardioVasc Interv
consuming and the puncture sites may be oriented toward Radiol 28:692703
the gastroesophageal junction. The technique described 7. Sanchez RB, van Sonnenberg E, DAgostino HB, Goodacre BW,
Moyers P, Casola G (1992) CT-guidance for percutaneous gas-
here works without any dedicated expensive capture trostomy and gastroenterostomy. Radiology 184:201205
devices. The duplicated guidewire loop can be arbitrarily 8. Tsukuda T, Fujita T, Ito K, Yamashita T, Matsunaga N (2006)
extended to the gastric walls, which affords quick capture Percutaneous radiologic gastrostomy using the push-type gas-
and fixation of the orogastric guidewire, and it is therefore trostomy tubes with CT and fluoroscopic guidance. AJR
186:574576
our current standard procedure. 9. Thornton FJ, Varghese JC, Haslam PJ, McGrath FP, Keeling F,
There have been reports on tumor seeding of the per- Lee MJ (2000) Percutaneous gastrostomy in patients who fail or
cutaneous tract after percutaneous endosopic gastrostomies are unsuitable for endoscopic gastrostomy. CardioVasc Interv
during longer follow-up. Cappell et al. [22] analyzed the Radiol 23:279284
10. Trerotola SO, Shah N, Johnson MS, et al. (1998) Single-step
risk factors and supposed an increased risk, particularly for dilatation for a large bore percutaneous gastrostomy and gastro-
pharyngoesophageal primary cancer, squamous cell his- jejunostomy. J Vasc Interv Radiol 9:579582
tology, less well-differentiated cancer, large size, and 11. Hicks ME, Surratt RS, Picus D, Marx MV, Lang EV (1999)
advanced cancer stage. These authors reviewed the litera- Fluoroscopically gui