Articles

Galectin-3-expression analysis in the surgical selection of

follicular thyroid nodules with indeterminate ne-needle
aspiration cytology: a prospective multicentre study
Armando Bartolazzi, Fabio Orlandi, Enrico Saggiorato, Marco Volante, Federico Arecco, Ruth Rossetto, Nicola Palestini, Ezio Ghigo, Mauro Papotti,
Gianni Bussolati, Marco Paolo Martegani, Federico Pantellini, Angelo Carpi, Maria Rosaria Giovagnoli, Salvatore Monti, Vincenzo Toscano,
Salvatore Sciacchitano, Gian Maria Pennelli, Caterina Mian, Maria Rosa Pelizzo, Massimo Rugge, Giancarlo Troncone, Lucio Palombini,
Gennaro Chiappetta, Gerardo Botti, Aldo Vecchione, Rino Bellocco, for the Italian Thyroid Cancer Study Group (ITCSG)

Summary
Background In the USA, about 30 200 well-dierentiated thyroid carcinomas were diagnosed in 2007, but the Lancet Oncol 2008; 9: 54349
prevalence of thyroid nodules is much higher (about 5% of the adult population). Unfortunately, the preoperative Published Online
characterisation of follicular thyroid nodules is still a challenge, and many benign lesions, which remain indeterminate May 19, 2008
DOI:10.1016/S1470-
after ne-needle aspiration (FNA) cytology are referred to surgery. About 85% of these thyroid nodules are classied
2045(08)70132-3
as benign at nal histology. We aimed to assess the diagnostic eect of galectin-3 expression analysis in distinguishing
See Reection and Reaction
preoperatively benign from malignant follicular thyroid nodules when FNA ndings were indeterminate. page 508
Department of Pathology,
Methods 544 patients were enrolled between June 1, 2003, and Aug 30, 2006. We used a puried monoclonal antibody St Andrea University Hospital,
to galectin-3, a biotin-free immunocytohistochemical assay, and a morphological and phenotypic analysis of FNA- Rome, Italy (A Bartolazzi MD,
derived cell-block preparations. Galectin-3-expression analysis was applied preoperatively on 465 follicular thyroid M P Martegani PhD); Cellular
and Molecular Tumour
proliferations that were candidates for surgery, and its diagnostic accuracy was compared with the nal histology. Pathology Laboratory,
Karolinska Cancer Centre,
Findings 31 patients were excluded because they had small galectin-3-negative thyroid nodules; we did not have data for Karolinska Hospital,
47 patients; and one patient with an oncocytic nodule was excluded. 331 (71%) of the assessable 465 preoperative Stockholm, Sweden
(A Bartolazzi); Department of
thyroid FNA samples did not express galectin-3. 280 (85%) of these galectin-3-negative lesions were classied as Pathology, St Giovanni Battista
benign at nal histology. Galectin-3 expression was detected, instead, in 134 of 465 (29%) thyroid proliferations, Hospital, Turin, and
101 (75%) of which were conrmed as malignant. The overall sensitivity of the galectin-3 test was 78% (95% CI 7482) Department of Endocrinology,
Azienda University Hospital,
and specicity was 93% (9095). Estimated positive predictive value was 82% (7986) and negative predictive value
St Luigi, Orbassano, University
was 91% (8893). 381 (88%) of 432 patients with follicular thyroid nodules who were referred for thyroidectomy were of Turin, Italy (F Orlandi MD,
correctly classied preoperatively by use of the galectin-3 test. However, 29 (22%) of 130 cancers were missed by the E Saggiorato, M Volante MD,
galectin-3 method. F Arecco MD, R Rossetto MD,
N Palestini MD, Prof E Ghigo MD,
Prof M Papotti MD,
Interpretation Our ndings show that if the option of surgery was based theoretically on galectin-3 expression alone, Prof G Bussolati MD);
only 134 thyroid operations would have been done in 465 patients; therefore a large proportion (71%) of unnecessary Department of Endocrinology,
thyroid surgical procedures could be avoided, although a number of galectin-3-negative cancers could be potentially University La Sapienza of
Rome, Italy, and St Peter
missed. The galectin-3 test proposed here does not replace conventional FNA cytology, but represents a complementary
Hospital Research Centre,
diagnostic method for those follicular nodules that remain indeterminate. Fatebenefratelli Association for
Research (AFaR), Italy;
Funding Compagnia di San Paolo, Turin, Italy, and Italian Association for Cancer Research (AIRC), Rome, Italy. F Pantellini MD, A Carpi MD,
M R Giovagnoli PhD,
S Monti MD, Prof V Toscano MD,
Introduction thyroid carcinoma, oncocytic carcinoma, and follicular S Sciacchitano MD); Department
Thyroid nodules are a common clinical problem. The variant of papillary carcinoma)1,37 FNA fails to distinguish of Pathology, Veneto Institute
prevalence of palpable thyroid proliferations in adults between benign and malignant disease in about 1530% of Oncology, Istituto
Oncologico Veneto (IOV-
increases with age (mean prevalence 47% in the USA1), of tests, depending on the diagnostic centre.1,37 Con- IRCCS), and Department of
but is much higher when subclinical nodules are also sequently, many patients with follicular proliferations are Endocrinology, University of
counted.2 About 8085% of these lesions are benign, there- referred for thyroidectomy without real therapeutic Padua, Padua, Italy
fore a reliable and systematic approach to their character- necessity.7 Furthermore, nal histology often conrms (G M Pennelli MD, C Mian MD,
M R Pelizzo MD,
isation is needed. malignancy in about 1015% of excised lesions.810 Prof M Rugge MD);
Thyroid ne-needle aspiration (FNA) has substantially To improve the diagnostic accuracy of thyroid FNA Departments of Pathology and
improved the characterisation of thyroid nodules and has cytology, we proposed a test method (galectin-3 thyrotest) Cytology, University Federico
led to decreases in health-care costs and improved diag- based on expression analysis of galectin-3 on FNA-derived II, Naples, Italy (G Troncone MD,
Prof L Palombini MD);
nosis of malignant lesions when thyroidectomy is done. cell blocks.11 Galectin-3 is a -galactosyl-binding molecule Departments of Pathology and
However, FNA has intrinsic limitations in distinguishing in the lectin group, and is involved in dierent biological Endocrinology, National Cancer
between benign (nodular hyperplasia, follicular adenoma functions, including cell adhesion, cell-cycle regulation, Institute, Fondazione Pascale,
and its variants) and malignant follicular lesions (follicular apoptosis, and tumour progression.12 We used the Naples, Italy (G Chiappetta MD,

http://oncology.thelancet.com Vol 9 June 2008 543

 Articles
G Botti MD,
Prof A Vecchione MD); and
Department of Medical
Epidemiology and
Biostatistics, Karolinska
Institute, Stockholm, Sweden,
and Department of Statistics,
University of Milano Bicocca,
Milan, Italy (R Bellocco ScD)
Correspondence to:
Dr Armando Bartolazzi,
Department of Pathology,
St Andrea Hospital, Via di
Grottarossa 1035, 00189 Rome,
Italy
Armando.Bartolazzi@ki.se
See Online for webgure
Thy3 proliferations
Follicular lesions or suspected
follicular tumours
Lesions
analysed
Group 1 168
Group 2 103
Group 3 49
Group 4 49
Group 5 29
Others 67
FA=follicular adenoma (all variants). H=nodular hyperplasia. FC=follicular carcinoma. FVPC=follicular variant of
papillary carcinoma. OC=oncocytic carcinoma (considered as follicular carcinoma variants by WHO, 2004). Other
tumours were poorly dierentiated thyroid tumours. Borderline tumours were follicular tumours with undened
malignant potential.22,23
Table 1: Distribution and histological typing (post-operatively) of follicular thyroid nodules with
indeterminate FNA cytology
544
galectin-3 test method for the preoperative characterisation
of thyroid nodules for several reasons: rst, in normal
conditions galectin-3 is not expressed in the cytoplasm of
thyroid cells and its forced expression (by galectin-3 cDNA
transfection) generates a transformed phenotype;1214 con-
versely, inhibition of galectin-3 expression has been
shown to revert the transformed phenotype in dierent
tumour models.1416 Second, the aberrant expression of
galectin-3 blocks the apoptotic programme, a feature that
favours the development of cancer.12,17 Third, we had
previously shown18 that galectin-3 is a physiological target
of P53 transcriptional activity, and that P53-mediated
down-regulation of galectin-3 is needed for P53-induced
apoptosis. Fourth, published studies showed that well-
dierentiated thyroid carcinomas almost always express
galectin-3, whereas healthy thyroid tissue and most
benign thyroid proliferations do not.11,19
The main reason that conventional FNA cytology fails to
characterise reliably follicular thyroid nodules in a third of
tests is because cytological criteria for distinguishing
benign from malignant follicular proliferations do not
exist, despite progress in FNA cytology.1,37,20 To assess the
diagnostic accuracy of the galectin-3 test method we aimed
to undertake a national, prospective, multicentre study of
465 patients in collaboration with 11 specialised thyroid
institutions. The nal histological diagnosis (considered
as the gold standard) was compared with the preoperative
diagnostic ndings of the test method.
Methods
Patients and procedures
Candidate working groups were selected on the basis of
their recorded clinical activity (eg, number of ne-needle
aspirations of the thyroid (FNAB) each year; number of
thyroid ultrasonographies each year; number of thyroid
procedures each year) and pertinent publications on
thyroid research. 11 specialised thyroid institutions in
northern, central, and southern Italy were selected from
22 applications (constituting two groups in northern Italy,
one group in central Italy, two groups in southern Italy,
Prevalence of lesions at nal histology
Benign (n=302) Malignant (n=130)
105 (59 FA; 46 H) 51 (4 FC; 31 FVPC; 14 OC; 2 other)
59 (17 FA; 42 H) 35 (3 FC; 26 FVPC; 4 OC; 2 other)
33 (25 FA; 8 H) 13 (2 FC; 8 FVPC; 1 OC; 2 other)
36 (30 FA; 6 H) 7 (1 FC; 6 FVPC)
24 (12 FA; 12 H) 3 (1 FC; 1 FVPC; 1 OC)
45 (33 FA; 12 H) 21 (4 FC; 14 FVPC; 2 OC; 1 other)
and an other group of institutions in northern and
southern Italy; webgure). Operative protocols for patient
selection, ultrasonography-guided thyroid FNA, cell-block
preparation, and immunohistocytochemistry were
provided in advance to all participating centres during two
workshops. Patients were enrolled into the study between
June 1, 2003, and Aug 30, 2006. All patients had a follicular
thyroid nodule cytologically classied as Thy3.21 465 of
544 patients enrolled in this study were referred to surgery
based on clinical reasons and multidisciplinary assessment.
For the other 79 patients who were initially enrolled:
31 were not considered for surgery because they had small
galectin-3-negative thyroid nodules (ie, nodules <1 cm in
size or suspected colloid nodules that were present for
several years without evidence of modications), which
were deemed benign after a clinicopathological consensus
meeting. These patients are in active follow-up without
signs of thyroid disease. We do not have information on
47 patients who received FNA cytology and galectin-3
assessment in our institutions (preoperative diagnoses),
but were treated elsewhere in regional hospitals. One
patient with an oncocytic nodule was removed from the
study after the central blind revision of histological
diagnosis because of inadequate sampling, where the
sample was deemed necrotic and undened.
70 (15%) of 465 thyroid proliferations showed atypical
cells at FNA cytology, whereas 395 (85%) of 465 did not
show cytological atypia. 98 (21%) of 456 follicular thyroid
nodules showed oncocytic features (ie, presence of
oxyphilic cells).
The study group included 355 (76%) women and
110 (24%) men, mean age 50 years (range 2176). Mean
size of the single or dominant nodule object of study was
28 cm (range 1055). To obtain adequate FNA cytology
and cell-block preparations, and to increase the possibility
of surgical removal of the lesion, participating centres were
discouraged in the workshops from enrolling patients with
thyroid nodules measuring less than 1 cm in diameter. The
nal histological assessment of the 465 excised thyroid
lesions was done by at least two experienced pathologists
in each specialised thyroid centre (no substantial internal
diagnostic disagreement was reported).
A blind central review of 294 unselected histological
Borderline (n=33) preparations was done by two expert and independent
12 pathologists (Juan Rosai, Centro Diagnostico Italiano,
9 Milan, Italy, and Virginia LiVolsi, Department of Pathology,
3 University of Pennsylvania, PA, USA) for quality control
6
of the gold standard (nal histology). The lesions included
2
in this study were: thyroid carcinomas (follicular types and
1
follicular variants of papillary carcinoma) in 130 (28%) of
465 patients and benign thyroid proliferations (adenomas
and hyperplasias) in 302 (65%) of 465 patients. 33 (7%) of
465 lesions were histologically classied as follicular
tumours of uncertain malignant potential (FT-UMP)22,23
(table 1). The proportions of thyroid lesions are those
expected according to national trends, with the exception
of FT-UMP. This specic cluster of lesions was initially
http://oncology.thelancet.com Vol 9 June 2008

preferentially diagnosed in central Italy, but after the blind
central review (of the internal histological diagnosis), we
noted that such lesions were well distributed across the
participating centres in all regions.
The aim of this prospective study was to assess the
diagnostic accuracy of galectin-3 expression analysis on
the preoperative characterisation of 465 patients with
follicular thyroid nodules that FNA cytology could not
distinguish as benign or malignant, and who were
candidates for partial or total thyroidectomy. These thyroid
proliferations were classied as Thy3 according to guide-
lines of the British Thyroid Association, Royal College of
Physicians, London, UK.21 This study was done according
to the ethical guidelines of the Declaration of Helsinki.
Written informed consent was obtained from each enrolled
patient. Specic approval was also obtained from each
Institutional Scientic Board.
Galectin-3 expression analysis and immunocytohistochemistry
Operative protocols and homogeneous reagents for
immunostaining were provided in advance to all centres.
Immunocytohistochemistry was done as described else-
where.11,19 Briey, antigen-retrieval microwave treatment
(001 M citrate buer, pH 60) was applied for three cycles
of 5 min each at 750 W. Endogenous peroxidase activity
was quenched with methanol-hydrogen peroxide (3%) for
15 min. After blocking with unrelated antiserum, slides
were incubated with the primary rat monoclonal antibody
to galectin-3 (Mabtech, Nacka, Sweden) at a concentration
of 10 g/mL in a moist chamber at 4C. The use of
formalin-xed and paran-embedded tissues as target
and a biotin-free immunoperoxidase staining method
were imperative for this study. These were obtained by use
of FNA-derived, formalin-xed, and paran-embedded
cell-block preparations and a galectin-3 thyrotest kit for
direct immunoperoxidase (Mabtech, Space Import Export
Srl, Milan, Italy) or Envision System (DakoCytomation,
Glostrup, Denmark) with the same galectin-3-specic
unconjugated reagent (Mabtech, Nacka, Sweden) as the
primary monoclonal antibody, in indirect immuno-
peroxidase. The two galectin-3 tests were completely
interchangeable. The enzymatic activity was visualised by
use of 3,3-diamino-benzidine (DakoCytomation). Usually,
one FNA-derived cell block was tested for each patient.
The galectin-3-expression scoring system has been
described elsewhere.19,23 We considered cell-block
preparations as positive if more than 5% of thyroid cells
showed an evident cytoplasm immunostaining. Thyroid
foamy macrophages, often detected in histological and
cytological thyroid preparations, express galectin-3
constitutively, and were deemed internal positive controls.
Immunocytohistochemistry and nal histology were
assessed by at least two pathologists in each centre. All
pathologists and cytologists participating in this study
were asked to follow strictly the diagnostic criteria for
denition of follicular thyroid proliferation reported in the
current published studies1,3,4,21,24 More specically, adequate
http://oncology.thelancet.com Vol 9 June 2008
Articles
FNA-derived cell blocks were dened as containing at least
four or ve groups of follicular arranged thyroid cells with
1020 cells in each group.
Follicular carcinoma was histologically diagnosed when
capsular penetration or vascular invasion, or both, were un-
equivocally shown by adequate tissue sampling.1 Papillary
thyroid carcinoma and its variants were diagnosed on the
basis of nuclear features (nuclear clearing, groves, and
pseudoinclusions) and papillary or follicular architecture
(or both).1,3,4,20,24
As for the histological diagnosis, a blind central review
of the cytological preparations was considered, but was
extremely dicult for several reasons. Most importantly,
there were legal issues in collecting and sending the slides
for central review because they could not be duplicated.
Specic authorisation from each patient and cytology
services were needed, but these tasks were too time-
consuming.
Website for data collection and analysis
Clinical, cytological, and pathological data, and the ndings
of galectin-3 expression analysis were collected for
544 patients over a period of 35 years. A dedicated web-
based system was developed for data collection and
analysis. The system consists of a website connected to a
centralised database that uses PHP (Hypertext Pre-
processor) and HTML (HyperText Markup Language)
scripting languages. Data were stored by use of Microsoft
Access and MySQL.25 Password-protected access to this
website enabled each centre to enter patients data in one
or more sessions by completing specic electronic forms.
A JavaScript program ltered the data entered before their
online submission. All data were hosted and protected on
a secure server of an internet provider. At the end of the
study, the entire set of data on 465 eligible thyroid
proliferations was downloaded and analysed.
Statistical analysis
Sensitivity, specicity, and positive and negative predictive
values of galectin-3 thyrotest were assessed as reported
elsewhere.11 Areas under the Receiver Operating Charac-
teristics (ROC) curves were calculated to describe the
overall performance of the test by use of Stata 9.2 (version
9.0). Formal tests of intercentre variability, based on
Fishers exact test for heterogeneity with ve degrees of
freedom, were done separately for sensitivity and
specicity, and by analysis of two homogeneous groups of
lesions: thyroid carcinomas and benign thyroid follicular
proliferations. The 33 patients with FT-UMP were not
included in the statistical analysis because they were
indeterminate at the nal histology.
Role of the funding source
The funding source had no role in study design, data
collection, analysis, or interpretation, or in the writing of
the report. AB had full access to all data and had nal
responsibility to submit the report for publication.
545
 Articles

Results (including four oncocytic variants), and two poorly

Galectin-3 expression was absent in 331 (71%) of dierentiated carcinomas.
465 thyroid nodules assessed preoperatively (table 2). 8 (28%) of 29 false-negative carcinomas showed variable
280 (85%) of these galectin-3-negative lesions were galectin-3 expression when the test method was applied
diagnosed as benign at nal histology, whereas 29 (9%) postoperatively to the corresponding excised tumours
were diagnosed as thyroid cancer. These false-negative (data not shown). This observation probably suggests
tumours included 19 follicular variants of papillary technical inaccuracies with FNA sampling or cell-block
carcinomas (FVPCs), eight follicular carcinomas immunostaining, or both. However, the remaining
21 galectin-3-negative carcinomas (at preoperative and
postoperative assessment) were undetectable with the
Lesions analysed Galectin-3 negative Malignant Benign Borderline
(n=465) (n=331) (n=29)* (n=280) (n=22)
proposed immuno-diagnostic approach.
Galectin-3 expression on preoperative FNA-derived cell
Group 1 168 116 7 99 10
blocks was noted instead in 134 (29%) of 465 patients
Group 2 103 57 2 53 2
(table 3). 101 (75%) of these 134 tumours were conrmed
Group 3 49 41 8 30 3
as malignant. In 22 patients, galectin-3-positive nodules
Group 4 49 40 4 31 5
were categorised as benign, whereas 11 patients had
Group 5 29 27 1 24 2
nodules that were nally classied as borderline lesions
Others 67 50 7 43 0 (ie, FT-UMP). 19 of the 22 patients had galectin-3-
*8 of 29 samples showed a variable galectin-3 expression at nal histology. expressing benign thyroid proliferations classied as
adenomas, and only three patients had nodular hyperplasia
Table 2: Comparison of galectin-3-negative ndings on preoperative FNA-derived cell blocks with nal
at nal diagnosis.
histological assessment
The possibility that these galectin-3-positive follicular
proliferations (19 adenomas of 176 tested and 11 FT-UMP)
Lesions analysed Galectin-3 positive Malignant Benign Borderline could represent early carcinomas, in which the histological
(n=465) (n=134) (n=101) (n=22)* (n=11) hallmarks of malignant disease (ie, capsule and vascular
Group 1 168 52 44 6 2 invasion) is not evident yet has been previously reported
Group 2 103 46 33 6 7
elsewhere.11,23 We plan to assess this hypothesis at
Group 3 49 8 5 3 0
phenotypic and molecular levels. As expected, only three of
Group 4 49 9 3 5 1
126 histologically conrmed hyperplasic nodules were
deemed galectin-3 false-positive lesions (tables 13). The
Group 5 29 2 2 0 0
diagnosed thyroid malignancies were mostly follicular
Others 67 17 14 2 1
variants of papillary carcinoma (86 of 130 [66%] of the
*Included 19 follicular adenomas and three nodular hyperplasias. tumours), 37 (28%) were follicular carcinomas (15 [12%]
Table 3: Comparison of galectin-3-positive ndings on preoperative FNA-derived cell blocks with nal
were conventional and 22 [17%] were oncocytic variants)
histological assessment and seven (5%) were poorly dierentiated carcinomas. The
unexpected large number of follicular variants of papillary

Preoperative galectin-3 Central blind histological Summary

immunocytochemistry review
Follicular adenoma (2 lesions) Positive PTC Preoperative cytology and positive galectin-3 suggested surgery
FV-PTC (4 lesions) Positive Follicular adenomas Preoperative cytology and positive galectin-3 suggested surgery
Follicular adenoma (2 lesions) Negative FV-PTC Missed cancer diagnosis
Follicular hyperplasia (6 lesions) Negative FV-PTC Missed cancer diagnosis
Follicular adenoma (2 lesions) Negative Follicular carcinomas Missed cancer diagnosis
Follicular adenoma (1 lesion) Negative Poorly dierentiate Missed cancer diagnosis
carcinoma
Follicular adenoma or FT-UMP Positive Follicular carcinoma Preoperative cytology and positive galectin-3 suggested surgery
(1 lesion)
FV-PTC (1 lesion) Negative Follicular adenoma Preoperative cytology and negative galectin-3 suggested follow-up
Follicular carcinoma (1 lesion) Negative Hyperplasia Preoperative cytology and negative galectin-3 suggested follow-up
Oncocytic carcinoma* (1 lesion) Positive Oncocytic nodule, necrotic Preoperative cytology and positive galectin-3 suggested surgery

PTC=papillary thyroid carcinoma. FV-PTC=follicular variant of papillary thyroid carcinoma. FT-UMP=follicular tumour with uncertain malignant potential. *This lesion was
excluded from the study because histologically undened. If thyroidectomy was theoretically based on galectin-3-expression analysis alone, this adenoma was not selected
for surgery during a rst clinicopathological assessment.

Table 4: Summary of major discrepancies noted at central blind review of a representative group of unselected histological diagnoses (n=294) and
potential diagnostic accuracy of galectin-3 test, by histological ndings (n=21)

546 http://oncology.thelancet.com Vol 9 June 2008

 Articles

carcinoma highlights the fact that conventional cytology

Group 1 Group 2 Group 3 Group 4 Group 5 Others Total
fails to correctly identify these lesions for a large number of (n=51) (n=35) (n=13) (n=7) (n=3) (n=21) (n=130)
patients.6,19 381 (88%) of 432 patients with follicular thyroid
Negative 7 (14) 2 (6) 8 (62) 4 (57) 1 (33) 7 (33) 29 (22)
nodules who were referred for thyroidectomy were correctly
Positive 44 (86) 33 (94) 5 (39) 3 (43) 2 (67) 14 (67) 101 (78)
classied preoperatively by use of the galectin-3 test.
During the central histological review of unselected thyroid Data are n (%). Fishers exact p=0001.
lesions for quality-control purposes, the external reviewers
Table 5: Intercentre variability based on galectin-3 test in patients with cancer
noted 21 major discrepancies in our diagnoses of thyroid
cancer, which were promptly discussed and revised
independently of immunoreactivity ndings. These thyroid Group 1 Group 2 Group 3 Group 4 Group 5 Others Total
proliferations that were not correctly classied at both (n=105) (n=59) (n=33) (n=36) (n=24) (n=45) (n=302)
preoperative cytology and at nal histology probably Negative 99 (94) 53 (90) 30 (91) 31 (86) 24 (100) 43 (96) 280 (93)
represented a real diagnostic variable in thyroid pathology. Positive 6 (6) 6 (10) 3 (9) 5 (14) 0 2 (4) 22 (7)
We noted that the preoperative use of galectin-3 immuno-
Data are n (%). Fishers exact p=0313.
staining could resolve about 50% of these lesions (table 4).
There were 29 false-negative lesions and 22 false-positive Table 6: Intercentre variability based on galectin-3 test in patients with benign lesions
lesions. The overall sensitivity of the galectin-3 test was
78% (95% CI 7482) and specicity was 93% (9095). With
a prevalence of 30% (ie, 130 of 432; 95% CI 2634), the in eight (28%) of these thyroid proliferations, which were
estimated positive predictive value was 82% (7986) and analysed after thyroidectomy, galectin-3 expression was
the negative predictive value was 91% (8893). The nally shown by immunohistochemistry (table 2, and data
estimated ROC was 85 (8188), which suggested good not shown). Furthermore, we noted that a centre
overall accuracy of the test, and 381 (88%) of 432 lesions contributing 20 follicular thyroid proliferations (six folli-
were correctly classied. cular variants of papillary thyroid carcinoma, six oncocytic
The variability between centres was assessed separately adenomas, and eight follicular adenomas) provided only
for positive and negative thyroid nodules at the nal galectin-3-negative ndings. This suggests technical
histology (tables 5 and 6). In patients with cancer, positive failure during the galectin-3 thyrotest occurred in this
ndings varied signicantly between centres (range from centre. However, specic training and dedicated workshops
39% to 94%); however, the proportions of galectin-3- on use of galectin-3 expression techniques might lower
positive tests in thyroid cancers was fairly consistent for the number of galectin-3-positive cancers that are missed.
those contributing centres that reported more than Nonetheless, at least 21 (16%) of 130 thyroid carcinomas
30 thyroid lesions (94% and 86%). Fishers exact test analysed here do not express galectin-3; this is probably
conrmed the variability between centres (p=0001). When because of additional molecular alterations that aect
the capability of each centre to correctly identify galectin-3- LGALS3 gene transcription, and we are currently studying
negative nodules was assessed, we recorded little variation this possibility.
(range 86100), and Fishers exact test conrmed this As expected, we noted 19 galectin-3-positive follicular
nding (p=0313). adenomas of 176 tested adenomas, and 11 galectin-3-
positive follicular proliferations histologically classied as
Discussion FT-UMP.22,23 We believe that surgical excision of these
The high prevalence of thyroid nodules in the adult specic lesions might be advisable. By contrast, we
population (1967% of randomly selected individuals2) conrmed the paucity of galectin-3 expression in thyroid
and the low prevalence of thyroid cancers makes the hyperplasia (tables 2 and 3).
diagnosis of thyroid cancer very dicult.2,26,27 Our ndings Although the diagnostic accuracy of the test method in
show that 381 (88%) of 432 patients with follicular thyroid some of the participating centres was not as good as in the
nodules who were referred for thyroidectomy were other centres and could be improved by more workshops,
correctly classied preoperatively by use of the galectin-3 the cumulative ndings presented here clearly show that if
test. Therefore, many unnecessary thyroid operations the option of surgery was theoretically based on galectin-3
could be avoided.11,19,28,29 The overall sensitivity of the test expression alone, only 134 thyroid operations would have
was 78% (95% CI 7482) and the specicity was 93% been donea decrease of 71%. This observation rep-
(9095). The estimated positive predictive value was 82% resents an important achievement considering that most
(7986) and the negative predictive value was of these lesions were benign, especially because patients
91% (8893). undergoing thyroidectomy need substitute hormone
Notably, in 29 (22%) of 130 patients with thyroid treatment which can be stressful for them and aect their
carcinoma, the diagnostic test did not detect malignant subsequent quality of life. However, of note, such a strategy
cells preoperatively, but technical issues were probably would have missed 29 of 130 cancers6% of the
responsible for the diagnostic failure in some of these 465 studied. However, patients are only referred for
patients. This observation is substantiated by the fact that thyroidectomy after careful pathological and clinical

http://oncology.thelancet.com Vol 9 June 2008 547

 Articles
assessment of each specic thyroid lesion, and some
suspicious, but galectin-3-negative, nodules could also be
referred for surgery for clinical or cytological reasons (or
both). Furthermore, the thyroid lesions analysed in this
study represent a highly selected sample, from which
patients with obvious cancer and obvious benign lesions
have already been removed. Therefore, the 29 galectin-3-
negative thyroid cancers we noted represent a group of
missed lesions in a sample of thyroid nodules in the
participating centres that was about ve-times the size of
the sample studied here. We believe that an estimated
false-negative proportion of around 12% might be
acceptable for a second-level diagnostic test such as the
galectin-3 test.
Although more than 120 studies on galectin-3
expression in thyroid cancer have been published in the
past decade with general agreement on its potential
diagnostic usefulness in thyroid cancer, some researchers
noted conicting ndings with use of galectin-3-
expression test for diagnostic purposes.2831 To our
knowledge, this is the rst prospective multicentre study
to assess galectin-3 expression for diagnostic purposes
and because diagnostic variability in the denition of
benign and malignant thyroid lesions can occur, even
between thyroid experts, some of the published studies
might also have beneted from a central review of the
histological diagnosis. Additionally, major methodological
issues were noted in these reports.32 Furthermore,
optimum galectin-3-expression analysis needs formalin-
xed and paran-embedded cytological preparations and
a biotin-free immunohistocytochemical detection
method. Galectin-3 expression should not be assessed on
conventional cytological thyroid smears for diagnostic
purposes, and only puried galectin-3-specic monoclonal
antibodies should be used.
The paucity of cells obtained from thyroid FNA is an
important limitation for cell-block preparation and
galectin-3 immunostaining.28 We proposed a strategy to
resolve this issue.29 Selected patients who are candidates
for surgery, for whom FNA cytology is inadequate (Thy1)
or inconclusive (Thy3) could be considered for preoperative
ultrasonography-guided large-needle aspiration biopsy
(LNAB), which uses 20-gauge needles. The diagnostic
performance of the galectin-3 thyrotest on LNAB-derived
substrates is excellent.29 Furthermore, LNAB-derived cell-
block preparations allow a comparative immunocyto-
chemical assessment, on the same cytological slides, of
dierent antigens associated with thyroid cancer.
Of these markers, the expression of human mesothelial
antigen (HBME-1) and cytokeratin-19 (CK19), and paucity
of thyroperoxidase immunoreactivity have been reported
to be useful for this purpose.33,34 Although the biological
rationale for the preferential expression (HBME-1 and
CK19) or downregulation (thyroperoxidase) of these
molecules in thyroid cancer is unclear, good overall
diagnostic accuracy has been reported with the combined
use of some of these markers.33 The quality of cell-block
548
preparations (number of thyroid cells processed) is crucial
for multiple and reliable immunostaining.
In conclusion, our ndings show that the galectin-3 test
can be used in the clinical setting, at least in specialised
thyroid hospitals and cancer centres. Specic training and
technical workshops are still necessary in some of the
institutions involved in this study and other interested
specialised hospitals, to optimise the accuracy of the
galectin-3 test.
The galectin-3 test proposed here does not replace
conventional FNA cytology, but represents a complementary
diagnostic method for those follicular nodules that remain
indeterminate. The correct approach for this preoperative
characterisation of thyroid nodules always needs careful
multidisciplinary assessment of each patient, according to
published guidelines.20,21,24,35 When benign and malignant
follicular lesions can be reliably distinguished pre-
operatively, a large proportion of unnecessary thyroid sur-
gical procedures will then nally be avoided, thereby
decreasing health-care costs.
Contributors
AB was the study coordinator. All the authors participated equally
(according to their medical competences: endocrinology, cytopathology,
thyroid surgery, and surgical pathology) in the clinical and diagnostic
activity of characterising preoperatively the follicular thyroid
proliferations noted in each thyroid centre or university hospital. This
activity included the application of the galectin-3 test and its independent
assessment. RB and MPM were responsible for data collection and
statistical analysis.
Others contributors
Departments of Pathology and Endocrinology, University of Pisa, Italy
Aldo Pinchera, Paolo Vitti, Fulvio Basolo.
Department of Pathology St Andrea Hospital, Rome, Italy
Alessandra Gasbarri, Fabrizio del Prete, Luca Lavra.
Department of Clinical and Biological Sciences, University of Turin,
ItalySusanna Cappia, Francesca Garino, Morena Martinese.
Departments of Pathology and Endocrinology University of Cagliari,
ItalyMaria Letizia Lai, Stefano Mariotti.
Departments of Pathology and Endocrinology University of Insubria,
Circolo Hospital, Varese, ItalyLuigi Bartalena, Carlo Capella,
Mariella Audi Grivetta, Annamaria Chiaravalle.
Department of Endocrinology University of Padua, Italy
Benedetto Busnardo, Davide Naccamulli.
Department of Pathology and Surgery, National Cancer Institute,
Fondazione Pascale, Naples, ItalyFranco Fulciniti, Pezzullo Luciano.
Also: Departments of Pathology and Endocrinology University of Pavia
and IRCCS Policlinico San Matteo, Pavia, Italy; Departments of
Pathology Regina Apostolorum Hospital, Albano Laziale, Rome, Italy;
Departments of Pathology V. Cervello Hospital, Palermo, Italy.
Conicts of interest
AB has received a translational research grant from AIRC (Italian
Association for Cancer Research), Rome, Italy. The other authors
declared no conicts of interest. The authors do not benet nancially
from the galectin-3 test.
Acknowledgments
This study was funded by Compagnia di San Paolo, Progetto Oncologia
20022006, Turin, Italy, and Italian Association for Cancer Research
(AIRC), Rome, Italy. The authors are grateful to Juan Rosai and
Virginia A LiVolsi for reviewing the microscopic section of 294 thyroid
follicular lesions from this series and providing their consensus
diagnoses on them. We thank Sandra Villani, Carlo Falasca, and
Clorinda Marchetti for their technical assistance; Francesca Mariani and
Annamaria Sarotto for their skilful organisational assistance, and Elena
Polizzi for the excellent administrative support.
http://oncology.thelancet.com Vol 9 June 2008

References
1 Rosai J, Carcangiu ML, De Lellis RA. Atlas of tumor pathology:
tumors of the thyroid gland. Third series, fascicle 5. Washington,
DC: Armed Force Institute of Pathology, 1992: 1343.
2 Tan GH, Gharib H. Thyroid incidentalomas: management
approaches to no palpable nodules discovered incidentally on
thyroid imaging. Ann Intern Med 1997; 126: 22631.
3 Kini SR. Thyroid. In: Kline TS, ed. Guide to clinical aspiration
biopsy, volume 3. New York: Igaku Shoin, 1987.
4 Baloch ZW, Sack MJ, Yu GH, et al. Fine-needle aspiration of
thyroid: an institutional experience. Thyroid 1998; 8: 56569.
5 Hall TL, Layeld LJ, Philippe A, et al. Sources of diagnostic error in
ne needle aspiration of the thyroid. Cancer 1989; 63: 71825.
6 Lloyd RV, Erickson LA, Casey MB, et al. Observer variation in the
diagnosis of follicular variant of papillary thyroid carcinoma.
Am J Surg Pathol 2004; 28: 133640.
7 Bartolazzi A. Improving accuracy of cytology for nodular thyroid
lesions. Lancet 2000; 355: 166162.
8 Mazzaferri EL. Management of a solitary thyroid nodule.
N Engl J Med 1993; 328: 55359.
9 Castro MR, Gharib H. Continuing controversies in the
management of thyroid nodules. Ann Intern Med 2005;
142: 92631.
10 Schlumberger MJ, Pacini F. Thyroid nodule. In: Thyroid tumors,
second edition. Paris: Editions Nucleon, 2003: 1131.
11 Bartolazzi A, Gasbarri A, Papotti M, et al. Application of an
immunodiagnostic method for improving preoperative diagnosis of
nodular thyroid lesions. Lancet 2001; 357: 164450.
12 Liu F, Rabinovich GA. Galectins as modulators of tumor
progression. Nature Rev Cancer 2005; 5: 2941.
13 Takenaka Y, Inohara H, Yoshii T, et al. Malignant transformation of
thyroid follicular cells by galectin-3. Cancer Lett 2003; 30: 11119.
14 Yoshii T, Inohara H, Takenaka Y, et al. Galectin-3 maintains the
transformed phenotype of thyroid papillary carcinoma cells.
Int J Oncol 2001; 18: 78792.
15 Honjo Y, Nangia-Makker P, Inohara H, et al. Down regulation of
galectin-3 suppresses tumorigenicity of human breast carcinoma
cells. Clin Cancer Res 2001; 7: 66168.
16 Kim HR, Lin HM Biliran H, et al. Cell cycle arrest and inhibition of
anoikis by galectin-3 in human breast epithelial cells. Cancer Res
1999; 59: 414854.
17 Akahani S, Nangia-Makker P, Inohara H, et al. Galectin-3: a novel
antiapoptotic molecule with a functional BH1 (NWGR) domain of
Bcl-2 family. Cancer Res 1997; 57: 527276.
18 Cecchinelli B, Lavra L, Rinaldo C, et al. Repression of the
antiapoptotic molecule galectin-3 by homeodomain-interacting
protein kinase 2-activated p53 is required for p53-induced
apoptosis. Mol Cell Biol 2006; 26: 474657.
19 Saggiorato E, Cappia S, De Giuli P, et al. Galectin-3 as a presurgical
immunocytodiagnostic marker for minimally invasive follicular
thyroid carcinoma. J Clin Endocrinol Metab 2001; 86: 515258.
http://oncology.thelancet.com Vol 9 June 2008
Articles
20 Gharib H. Fine-needle aspiration biopsy of thyroid nodules:
advantages, limitations and eects. Mayo Clin Proc 1994; 69: 4450.
21 British Thyroid Association, Royal College of Physicians of London.
Fine needle aspiration cytology. In: Guidelines for the management
of thyroid cancer, second edition. Lavenham Press: Lavenham,
2007: 910.
22 William ED. The Chernobyl Pathologist Group. Guest editorial:
two proposals regarding the terminology of thyroid tumors.
Int J Surg Pathol 2000; 8: 18183.
23 Papotti M, Rodriguez J, De Pompa R, et al. Galectin-3 and HBME-1
expression in well-dierentiated thyroid tumors with follicular
architecture of uncertain malignant potential. Mod Pathol 2005;
18: 54146.
24 Thyroid ne needle aspiration (FNA) and cytology. Thyroid.
Consensus guidelines for thyroid testing in the new millennium.
In: Laboratory medicine practice guidelines: laboratory support for
the diagnosis and monitoring of thyroid disease. American Thyroid
Association: Falls Church, 2003; 13: 8086.
25 Gerner J, Owens M, Naramore E, et al. Professional LAMP: Linux,
Apache, MySQL and PHP web development. Indianapolis,
IN: Wrox, 2006.
26 Sherman SI. Thyroid carcinoma. Lancet 2003; 361: 50111.
27 Jemal A, Murray T, Wards E, et al. Cancer statistics 2005.
CA Cancer J Clin 2005; 55: 1030.
28 Herrmann MF, LiVolsi VA, Pasha TL, et al. Immunohistochemical
expression of galectin-3 in benign and malignant thyroid lesions.
Arch Pathol Lab Med 2002; 126: 71013.
29 Carpi A, Naccarato AG, Iervasi G et al. Large needle aspiration
biopsy and galectin-3 determination in selected thyroid nodules
with indeterminate FNA-cytology. Br J Cancer 2006; 95: 20409.
30 Niedziela M, Maceluch J, Korman E. Galectin-3 is not an universal
marker of malignancy in thyroid nodular disease in children and
adolescents. J Clin Endocrinol Metabol 2002; 87: 441115.
31 Mehorotra P, Okpokam A, Bouhaidar R, et al. Galectin-3 does not
reliably distinguish benign from malignant thyroid neoplasms.
Histopathology 2004; 45: 493500.
32 Bartolazzi A, Bussolati G. Galectin-3 does not reliably distinguish
benign from malignant thyroid neoplasms. Histopathology 2006;
48: 21213.
33 Saggiorato E, De Pompa R, Volante M, et al. Characterization of
thyroid follicular neoplasms in ne-needle aspiration cytological
specimens using a panel of immunohistochemical markers:
a proposal for clinical application. Endocr Relat Cancer 2005;
12: 30517.
34 Weber KB, Shroyer KR, Heinz DE, et al. The use of combination of
galectin-3 and thyroid peroxidase for the diagnosis and prognosis of
thyroid cancer. Am J Clin Pathol 2004; 122: 52431.
35 Cooper DS, Doherty GM, Haugen BR, et al. Management
guidelines for patients with thyroid nodules and dierentiated
thyroid cancer. Thyroid 2006; 16: 133.
549