Beruflich Dokumente
Kultur Dokumente
a
Department of Imaging, hôpital Central, CHRU de Nancy, 29, avenue du
Maréchal-Lattre-de-Tassigny, 54035 Nancy, France
b
Medical Imaging Department, Oscar Lambret Center, Lille Nord University, 3, rue
Frédéric-Combemale, 59000 Lille, France
c
Department of Pathology, University Hospital CHRU, University Lille-Nord de France, 59000
Lille, France
d
Department of General Oncology, Oscar Lambret Cancer Center, 59000 Lille, France
e
Centre de consultation et d’imagerie de l’appareil locomoteur (CCIAL), CHRU de Lille, 2,
avenue Oscar-Lambret, 59037 Lille cedex France
KEYWORDS Abstract
Magnetic resonance Purpose: To identify quantitative perfusion parameters that are best associated with tumor
imaging (MRI); grade and tumor necrosis at magnetic resonance (MR) imaging at 3-Tesla.
Perfusion MR Methods: MR perfusion studies of 31 patients with a musculoskeletal sarcoma were retrospec-
imaging; tively evaluated by two readers. There were 18 men and 13 women with a mean age of
Musculoskeletal 34.9 ± 24.4 (standard deviation [SD] years) (range: 6—87 years). All patients underwent car-
sarcoma; cinologic tumor resection less than 3 months after MR imaging. For all patients six perfusion
Neoplasm grading; parameters (three semi-quantitative and three permeability parameters) were analyzed. The
Tumor necrosis percentage of tumor necrosis was estimated using MR imaging. Perfusion data were compared
between groups of tumors with different grades and necrosis ratios. Interobserver variability
was calculated using intraclass correlation coefficient (ICC).
Abbreviations: FNCLCC, Fédération nationale des centres de lutte contre le cancer; MSK, Musculoskeletal system; INI, Imaging necrosis
index; ROI, Region of interest; FSE, Weighted fast spin-echo; NEX, Number of excitations; FOV, Field of view; HPF, High power field; SPGR,
Spoiled gradient-echo; AIF, Arterial input function; ICC, Intraclass correlation coefficients; EES, Extravascular extracellular space; CER,
Contrast enhancement ratio; AUC, Area under the curve; Max Slope, Maximum slope of increase; Ktrans , Transfer constant from the plasma
to the extravascular extracellular space; kep , Backflux constant; Ve , Extravascular extracellular space volume.
∗ Corresponding author.
https://doi.org/10.1016/j.diii.2018.02.005
2211-5684/© 2018 Société française de radiologie. Published by Elsevier Masson SAS. All rights reserved.
474 P.A. Gondim Teixeira et al.
Results: Interobserver variability among the perfusion parameters was good to excellent (ICC:
0.72—0.9). The area under the curve and maximum slope values showed a significant association
with the degree of tumor necrosis (P = 0.02—0.04). When tumors with low necrosis ratios were
compared to those with high ratios the former parameter was 80% lower. In the same groups,
the imaging necrosis index was 56.9—59.8% higher in patients with grade 2 necrosis (P = 0.01).
Extracellular space volume (Ve ) was 31.4% to 55.8% lower in tumors with high grade while the
backflow constant (Kep ) was 33.6% to 40.1%% higher in tumors with high grade.
Conclusion: Semi-quantitative MR perfusion parameters have an excellent reproducibility and
are associated with the degree of histologic tumor necrosis in musculoskeletal sarcomas. The
utility of permeability parameters for determining tumor grade needs further investigations.
© 2018 Société française de radiologie. Published by Elsevier Masson SAS. All rights reserved.
• tumor grade: grades 1 and 2 of the FNCLCC or Broder’s extended Tofts pharmacokinetic model with the following
systems were considered to be low grade whereas FNCLCC equation [20]:
grade 3 tumors and Broder’s grade 3 or 4 tumors were
t
considered to be high grade;
• tumor differentiation: score 1 = sarcomas closely Ct (t) = vp Cp (t) + K trans Cp () e−kep (t−) d
0
resembling normal adult mesenchymal tissue; score
2 = sarcomas for which histologic typing is certain; score Application of this pharmacokinetic model yielded three
3 = embryonal and undifferentiated sarcomas, sarco- quantitative perfusion parameters: the volume transfer
mas of doubtful type, synovial sarcomas, soft-tissue constant from the plasma to the extravascular extracellu-
osteosarcomas, primitive neuroectodermal tumors; lar space (EES) (Ktrans in min —1), the rate constant of the
• necrotic index: score 0 = no necrosis; score 1 = less than
backflux from the EES to the plasma (kep in min —1) and the
50% tumor necrosis; score 2 ≥ 50% necrosis. Tumors scored EES volume (Ve ).
0 and 1 were considered to have a low necrosis ratio
whereas those scored 2 were considered to have a high
Slice selection and region of interest (ROI)
necrosis ratio;
• mitotic index. Score 1 = 0—9 mitoses per 10 high power positioning
fields (HPF); score 2 = 10—19 mitoses per 10 HPF; score
Two radiologists, both with 10 years of clinical experience
3 ≥ 20 mitoses per 10 HPF.
with MR imaging, post-processed MR perfusion studies inde-
pendently. The readers were blinded for tumor histologic
MR image acquisition characteristics. Slice selection and ROI positioning was per-
formed on perfusion images. The orthogonal plane and the
®
Images were acquired with a 3T MR scanner (Discovery slice demonstrating the largest tumor diameter was selected
MR750W, General-Electric Healthcare, Milwaukee, WI, USA). for analysis. Three ROI were positioned according to the
Coils were adapted to tumor location. The acquisition proto- following procedures (Fig. 1):
col included conventional a T1-weighted fast spin-echo (FSE) • global tumor ROI: an elliptical ROI was positioned to
sequence, a T2-weighted FSE fat-saturated sequence in at occupy the largest tumor diameter, including areas of
least two different orthogonal planes and a T1-weighted FSE viable and necrotic tumor. The association between per-
fat-saturated sequence post gadolinium. fusion parameters derived from this ROI and histologic
A three-dimensional (3D) spoiled gradient-echo (SPGR) tumor grade and necrosis was evaluated;
sequence was used to evaluate tissue perfusion (TE = 2 ms; • normal appearing muscle ROI: an elliptical ROI was
TR = 3.8; flip angle = 30◦ , bandwidth = 50 Hz; NEX = 0.5, accel- positioned over normal appearing (intermediate signal
eration factor 2.5; matrix from 128 × 128 to 160 × 160; intensity in both T1- and T2-weighted sequences) striated
FOV 320 mm; slice thickness 3.4—5 mm); 28 volumes were muscle in the study region. Intra-muscular vessels were
acquired in 3.2 minutes yielding a temporal resolution of 7 avoided. Normal muscle perfusion data were acquired to
s. In total, 0.2 mL/kg of gadolinium-based contrast material allow for the calculation of tumor/normal muscle ratios
®
(gadoterate meglumine, Dotarem at a dose of 0.5 mmol/ml used to avoid the influence of technical factors (different
Guerbet, Roissy CdG, France) was injected into a periph- coils, signal gain, coil filling rate and variable injection
eral vein. An injection pump (Mallinckrodt, OptistarTM Elite, rates);
Guerbet, Bloomington, IN, USA) was used for all injections. • viable tumor ROI: a free-form ROI was drawn over the
The injection rate was adapted to the patient’s age and viable tumor area. Viable tumor was defined as enhancing
peripheral vein status and varied from 1 mL/s to 5 mL/s areas in the perfusion study. All phases of the perfu-
(mean = 3.3 ± 1.1 [SD] mL/s). The delay between the start of sion acquisition were considered when determining viable
image acquisition and gadolinium injection was 21 seconds. tumor. Small foci of necrosis completely surrounded by
viable tumor were not considered in the determination
of viable tumor zones. This ROI was used to allow the
Perfusion parameters calculation calculation of the INI.
Image post-processing was performed using the AD Volume The areas in mm2 of viable and global tumor ROI were
Share 5 (V.4.6, GE Healthcare) using the GenIQ and the used to calculate the percentage of tumor necrosis on imag-
Readyview modules. The arterial input function (AIF) was ing, designated as imaging necrotic index (INI). Negative
calculated semi-automatically by placing an ROI over vessels INI values were considered to represent 0% necrosis. Post-
adjacent to the tumor. processing time varied from 10 to 15 minutes.
Two types of perfusion parameters were acquired:
semi-quantitative and permeability. The following semi- Statistical analysis
quantitative parameters were calculated: the contrast
enhancement ratio (CER) (peak signal—baseline sig- Statistical analysis was performed with the Stata software
nal/baseline signal), the area under the curve (AUC) and (v.11.2 Statcorp. 2009). Tumor/normal muscle ratios of
the maximum slope of increase defined as the curve slope all perfusion parameters acquired were considered. Uni-
of the steepest part of the curve. A fixed T1 value (900 ms), variate logistic regression analysis was used to assess the
as reported by Bazelaire et al. obtained with similar equip- relationship between perfusion parameters and histological
ment, was used for permeability parameter calculations variables (tumor grade and necrosis rate). Intraclass correla-
[15]. Permeability parameters were calculated using the tion coefficients (ICC) of the perfusion parameters obtained
476 P.A. Gondim Teixeira et al.
Figure 1. An 18-year-old man with osteosarcoma of the proximal fibula. A. Axial post-contrast T1-weighted fat-saturated image demon-
strates an aggressive lesion centered on the proximal fibula (arrowheads) with a necrotic center (area delimitated by the dashed line) and
surrounded by a halo of soft-tissue edema. B. Macroscopic section of the surgical specimen in the sagittal plane of this lesion shows areas of
necrosis (whitish central zone) and viable tissue (peripheral pinkish zone). C—E. ROI positioning procedure. C. A coronal oblique perfusion
image demonstrates the largest tumor diameter is selected. D. A free-form ROI (viable tumor ROI) was drawn to include the largest area
possible of enhancing tumor (red free-form). Small foci of necrosis completely surrounded by viable tumor were not considered in the
analysis (thin arrow). E. An elliptical ROI was drawn to include the largest tumor diameter possible (global tumor ROI) (blue elliptical ROI).
by the two readers were calculated. ICC values less than all patients. The ICC values of the perfusion parameters
0.40 were considered as poor agreement, 0.40—0.59 as fair, acquired varied from 0.72 to 0.9 and were considered good
0.60—0.74 as good, and 0.75—1 was accepted as excellent to excellent. The relation between mean values, SD and vari-
agreement. Quantitative variables were expressed as mean ation percentage of the perfusion parameters with respect
SD and range. Statistical significance threshold was set at to tumor grade and necrosis rate are reported in Table 2.
P < 0.05.
Perfusion versus tumor grade
Results
A total of 7 low grade tumors and 24 high grade tumors
The mean delay between MR examination and histologic were analyzed. For both readers, the Kep values were higher
analysis was 21.8 ± 17.9 (SD) days (range: 2—70 days). The (33.6% to 40.1%) and the Ve values were lower (31.4% to
histologic subtypes and grading of the tumors included are 55.8%) in high grade tumors compared to low grade ones.
presented in Table 1. Perfusion analysis was successful in There was a significant difference in Ve values for reader
Perfusion MR imaging in musculoskeletal sarcomas 477
Table 2 Distribution of perfusion parameter values obtained with total tumor ROIs according to tumor grade and necrosis
rate.
Reader 1 CER AUC Max slope Ktrans Kep Ve
Grade Mean low grade 4.8 ± 2.8 6.5 ± 7.0 3.6 ± 2.7 4.7 ± 2.4 1.1 ± 0.4 4.5 ± 3.4
Mean high grade 3.4 ± 2.8 4.7 ± 7.8 3.1 ± 4.5 5.8 ± 6.3 1.8 ± 1.5 3.1 ± 1.2
Difference —28.3% —26.2% —15.8% 18.9% 36.1% —31.4%
P-value 0.27 0.61 0.74 0.65 0.3 0.06
Necrosis rate Mean low NR 5 ± 3.4 9 ± 10 5.1 ± 5.5 7.1 ± 6.5 2 ± 1.6 3.6 ± 2.1
Mean high NR 2.7 ± 1.7 1.8 ± 1.9 1.7 ± 1.3 4.3 ± 4.6 1.4 ± 1.1 3.3 ± 1.9
Difference —45.4% —79.8% —67.6% —39.8% —29.3% —6.4%
P-value 0.3 0.02* 0.03* 0.2 0.19 0.74
(0 — no necrosis; 1 — less than 50% necrosis; 2 — more than significantly lower in tumors with high necrosis ratios with
50% necrosis), tumor necrosis was underestimated with respect to tumors with low necrosis ratios. The AUC showed
respect to histologic evaluation in 32.2% to 58.1% of tumors the highest variation and is regarded as a robust, widely
for readers 1 and 2. available and reproducible perfusion parameter [21]. The
The MR perfusion parameters studied were well associ- INI was significantly higher when tumors with high grades
ated with histologic tumor necrosis. For both readers, AUC of histologic necrosis were compared to tumors with low
and maximum slope of increase values were significantly grade necrosis. INI presented a fair ICC and tended to under-
lower in tumors with high necrosis ratios for both read- estimate necrosis with respect to histology. This finding is
ers (0.01 < P < 0.046). CER values were only significant for most likely related to the fact that tumor perfusion is an
reader 1 (P = 0.30 and 0.01 for readers 1 and 2). For both indirect indicator of necrosis while, histologically, cellular
readers, perfusion parameters were 44.7—53.7% lower in necrosis can be assessed directly. Non-invasive tumor necro-
patients with high necrosis ratios compared to patients with sis with INI and AUC assessment could be easily integrated on
low necrosis ratios. The AUC yielded greater differences daily practice and could be useful to evaluate the treatment
between tumors with low and high necrosis ratios. The AUC response of patients undergoing adjuvant therapy [1,5].
was 80.3% and 79.8% lower in tumors with high necrosis Permeability parameters presented a week association
ratios for readers 1 and 2 (Fig. 4). with tumor grade. Kep and Ve yielded greater differences
The permeability parameters Ktrans and Kep were also between tumors with low and high grade. This finding is
decreased in tumors with a high necrosis ratio for both read- consistent with literature data and the same perfusion
ers. Ve values also decreased for reader 2 but increased parameters have been shown to correlate with survival rates
for reader 1. The degree of reduction was lesser that what in patients with osteosarcoma by two different authors
was seen with semi-quantitative parameters (6.4 to 49.2% [22,23]. Ve was 31% to 55% lower in tumors with high differ-
for readers 1 and 2) and none of these differences were entiation/grade, and these differences were significant for
statistically significant (0.1 < P < 0.99). one of the readers. This can be at least partially explained
by the higher degree of cellularity of high grade tumors,
which tends to reduce the extravascular extracellular space
Discussion [24]. Although not statistically significant, Kep was 43—64%
higher in tumors with high grade, which could be related to
We have found that semi-quantitative MR perfusion parame- the higher capillary permeability in high grade tumors.
ters provide information with respect to the degree of tumor The lack of association between permeability parame-
necrosis. In addition, we found that MR perfusion parameters ters and tumor grade could be, at least partially, related
are associated with an excellent interobserver reproducibil- to the higher technical variability of permeability param-
ity (ICC > 0.84). The AUC and the maximum slope values were eters [25]. This variation is likely multifactorial, related
Perfusion MR imaging in musculoskeletal sarcomas 479
Figure 3. A 46-year-old woman with a fibromyxoid sarcoma of the left leg. A. Oblique coronal perfusion image in the arterial phase
demonstrating a heterogeneously enhancing mass (arrowheads) with central necrosis (star). The INI was estimated as 34—38.4%. B. Area
under the curve color map demonstrating a clear differentiation of perfusion in the viable (arrowheads) and necrotic tumor zones (star).
C. Macroscopic aspect of the tumor in a coronal section of the surgical specimen. A homogeneous necrotic area is seen at the upper aspect
of the tumor (star). D. Histological section demonstrating tumor necrosis (star) area, which represented less than 50% of the total tumor
volume (× 25).
to higher complexity of the calculations required for these to further explore the relationship between permeability
parameters (tissue gadolinium concentration estimation and parameters and tumor grading.
pharmacokinetic model fitting) [20]. AIF selection could Our study has several limitations. The number of patients
be an important source of variation [26]. Moreover, sig- included was small due to the low prevalence of MSK
nal saturation phenomena can take place in larger vessels sarcomas and the strict inclusion criteria, which limited
leading to AIF underestimation with an impact on calcu- the number of statistically significant variations found. A
lated parameters [27]. T1 variations in pathologic tissue and considerable number of patients were under adjuvant ther-
poor pharmacokinetic model fitting are additional technical apy, which likely influenced the amount of observed tumor
sources of permeability parameter variation [25]. The dis- necrosis. This fact probably explains why semi-quantitative
cordant findings in Ktrans values between readers 1 and 2 perfusion parameters were invariably lower in high grade
when tumor grade was considered can be at least partially tumors a seemingly paradoxical finding. Because the objec-
explained by the previously mentioned technical variation tive of this work was to compare perfusion parameters
issues. Studies, performed with more robust MR perfusion with histologic tumor features (as opposed to non-invasive
protocols (same coils, external reference for T1 value esti- tumor aggressiveness assessment), the influence of adju-
mation, better methods of AIF selection etc.) are needed vant therapy in the presented results is limited. Finally,
480 P.A. Gondim Teixeira et al.
marrow lesions: role of semiquantitative and quantitative [25] Gondim Teixeira PA, Leplat C, Chen B, De Verbizier J, Beau-
dynamic MRI. J Magn Reson Imaging 1999;10:286—94. mont M, Badr S, et al. Contrast-enhanced 3T MR perfusion of
[22] Reddick WE, Wang S, Xiong X, Glass JO, Wu S, Kaste SC, musculoskeletal tumours: T1 value heterogeneity assessment
et al. Dynamic magnetic resonance imaging of regional contrast and evaluation of the influence of T1 estimation methods on
access as an additional prognostic factor in pediatric osteosar- quantitative parameters. Eur Radiol 2017;27:4903—12.
coma. Cancer 2001;91:2230—7. [26] Cutajar M, Mendichovszky IA, Tofts PS, Gordon I. The
[23] Guo J, Reddick WE, Glass JO, Ji Q, Billups CA, Wu J, importance of AIF ROI selection in DCE-MRI renography: repro-
et al. Dynamic contrast-enhanced magnetic resonance imaging ducibility and variability of renal perfusion and filtration. Eur
as a prognostic factor in predicting event-free and over- J Radiol 2010;74:e154—60.
all survival in pediatric patients with osteosarcoma. Cancer [27] Ellinger R, Kremser C, Schocke MF, Kolbitsch C, Griebel J, Fel-
2012;118:3776—85. ber SR, et al. The impact of peak saturation of the arterial
[24] Khoo MMY, Tyler PA, Saifuddin A, Padhani AR. Diffusion- input function on quantitative evaluation of dynamic suscep-
weighted imaging (DWI) in musculoskeletal MRI: a critical tibility contrast-enhanced MR studies. J Comput Assist Tomogr
review. Skeletal Radiol 2011;40:665—81. 2000;24:942—8.