This study was accepted as a poster for the Stowell–Orbison Award during the 92nd USCAP Meeting held in Washington, DC, March 22–28, 2003.
ThinPrep® 2000 is a registered trademark of the Cytyc Corporation, Boxborough, Massachusetts, and, in this article, is the technology used for thin-layer cytology. The Authors have no significant relationship with the manufacturer of the system used in this study.
Fine-needle aspiration biopsy (FNAB) is the most reliable diagnostic tool for thyroid nodules. A difficult cytologic diagnosis may be supported by an immunocytochemical study. The efficacy of a panel made up of RET, HBME-1, and Galectin-3 antibodies was evaluated in smears processed by thin-layer cytology (TLC).
Thyroid FNABs (n = 99) with both conventional (CS) and thin-layer cytology (TLC) smears were studied. The cases were diagnosed as follows: 5 benign lesions (BL), 13 papillary carcinomas (PC), and 81 follicular proliferations (FP). The category of FP was divided into three subgroups according to nuclear features of follicular cells: Follicular neoplasm (FN NOS), oxyphilic follicular neoplasm (OFN), and folliucular lesion with nuclear pleomorphism (FLWNP). Immunostains for HBME-1, Galectin-3, and RET were carried out on TLC slides.
Among 49 cases undergoing surgery, all 10 PC and 2 BL were histologically confirmed, whereas 15 out of 37 FP (40.5%) were malignant. The complete immunocytochemical panel (ICCP) was positive in 9 of 10 malignancies (90%) and negative in both BLs. Out of 37 FP, the ICCP yielded positive in 15 cases (4 benign, 11 malignant) and negative in 11 (all benign). In the FLWNP subgroup, the ICCP was positive in 11 (84.6%) and negative in 2 histologically benign cases.
Fine-needle aspiration biopsy (FNAB) is the first choice for diagnostic evaluation of thyroid nodules. It can be thought of as a sophisticated screening test for a large population of patients with nodules, a small group of whom should be considered for surgery. FNAB may achieve a correct interpretation of the nature of thyroid lesions in more than 70% of patients and lead to a correct clinical strategy (mostly related to the surgical indication) in more than 90% of patients.1–3 It is simple, safe, and extremely cost effective, especially when performed under sonographic guidance. The diagnostic accuracy of FNAB decreases slightly in thyroid lesions that exhibit a predominant microfollicular pattern. The real nature of these lesions can be assessed only with the detection of capsular and vascular invasion, whereas needle biopsy may sample only the core of the nodule. Although several studies3 have investigated the role of different morphologic features (nuclear atypia and pleomorphism, cell overlapping) that enable a distinction to be made between the most important follicular-patterned neoplasms (follicular adenoma [FA], follicular carcinoma [FC], follicular variant of papillary carcinoma [FVPC]), some other studies have questioned this possibility.4 Immunohistochemistry was introduced to the practice of pathology in the early 1970s, but, in thyroid pathology, its use has been restricted to differential diagnosis between follicular and C-cell derived neoplasms. The detection of new markers of malignancy, which may distinguish malignant from benign lesions regardless of the presence of capsular or vascular invasion, has provided interesting insights into the role of immunohistochemistry in thyroid neoplasms.
In the literature, HBME-1, Galectin-3, and RET protooncogene have shown the best specificity and sensitivity in discriminating benign from malignant differentiated tumors.4, 5 Galectin-3 is a component of the beta-galactoside–binding lectins, whose function is still unclear. It appears to be involved in cell-to-cell and cell-matrix modulation and, according to some authors,6 in cell growth and differentiation. Therefore, this antibody could play a role in the malignant transformation of thyroid cells, and it is expressed in a high proportion of carcinomas, especially the papillary type. HBME-1 is a monoclonal antibody directed against a still unknown membrane antigen of the microvillar surface of the mesothelioma cells. It is positive in neoplastic cells, mostly from papillary carcinoma, and negative in benign lesions.7 RET protooncogene, which is normally expressed in neural-crest–derived tissues, shows no positivity in thyroid follicular cells, unless its rearranged forms, reflecting the RET gene activation, are present. An immunohistochemical positivity for RET antibody has been observed in various thyroid carcinomas, more frequently in the papillary subtype.8 However, no single antibody has shown enough diagnostic accuracy to be used alone in correctly diagnosing an encapsulated thyroid neoplasm.
Immunocytochemistry may find new applications in aspiration cytology, mostly when thin-layer cytologic (TLC) preparations are adopted. This technique has been recently introduced to obtain thin layer slides in cervical cytology. Excellent results in this field have prompted a wider application to almost all cytological branches, including thyroid FNAB.9 The current investigation of a panel comprising three antibodies (HBME-1, Galectin-3, and RET protooncogene) was performed to evaluate its effectiveness in discriminating between benign and malignant thyroid lesions on thin layer slides.10–13
MATERIALS AND METHODS
Between June 2002 and February 2003, 570 thyroid FNABs were performed in the Divisions of Endocrine Surgery and Anatomic Pathology and Histology of the Catholic University, Agostino Gemelli School of Medicine and Hospital of Rome (Italy). Of these, 99 consecutive cases (17.3%) were selected as follows: 1) cytologic diagnosis of follicular proliferation (FP) made by the same pathologist, 2) presence of abundant material in both types of smears,and 3) availability of patients for follow-up. Using the same criteria listed above, 13 papillary carcinomas (PC) and 5 benign lesions (BL) belonging to the same series were selected as the control group. The series included 24 male and 75 female patients with a median age of 32.5 years (range, 13–78 yrs). All aspirations were performed with 25- to 27-gauge needles under sonographic guidance making 2 consecutive passes for each lesion, the first for the conventional smear (CS), and the second for the TLC slides.
CS slides were fixed in 95% ethanol, and TLC slides were fixed with Cytolit™ solution and then processed according to the Thin Prep 2000™ method (Cytyc Co., Marlborough, MA). Both smears were stained with Papanicolaou. The final morphologic diagnosis was made after comparing the two series of slides and observing an overall concordance of > 95%. Those cases where a discordant diagnosis between CS and TLC was found were discussed by three cytopathologists (EDR, GF, and AM) until a final agreement was achieved.
Immunocytochemical staining was carried out with the avidin-biotin–peroxidase complex on the TLC slides using the following antibodies: HBME-1 (Dako, Denmark, 1:100 dilution), Galectin-3 (Ylem, Italy, 1:100 dilution), and RET (Ylem, Italy, 1:100 dilution). The slides were washed 3 times in phosphate-buffered saline (PBS) and then preincubated in normal serum (1:50) in PBS for 20 minutes before incubating overnight at 4 °C with the primary antibody. Slides were then washed 3 times with PBS and incubated with biotinylated secondary antibody conjugated with avidin-biotin–peroxidase complex. The reaction was developed using 3–3′ diamino-benzidine (DAB). All slides were counterstained with hematoxylin for 5 seconds, rinsed in water 3 times, then mounted for microscopic examination. The positivity was assessed, for each cytologic case, when at least 50% of cells showed a convincing cytoplasmic positivity. The cytodiagnostic groups of FP and PC were referred to surgery, whereas the BL underwent surgery only when aesthetic or functional problems were present. The category of FP was divided into three subgroups according to the morphologic classification devised by our group.14, 15 This classification included the following categories (see Table 1).
Table 1. Morphologic Criteria for Classification of Follicular Proliferations
THIN PREP 2000 slides
FN NOS (follicular nodule not otherwise specified)
Scant colloid; microfollicles or small monolayered clusters of medium-sized follicular cells, sometimes with slight nuclear pleomorphism (hyperfunction) and rounded nuclei; fibrovascular tissue; hemorrhage with hemosiderin-laden histiocytes.
Small clusters of medium-sized follicular cells with pleomorphic nuclei, generally with regular outlines; fibrin flakes; none or scanty colloid globules; hemosiderin-laden histiocytes.
Adenomatous nodule in a goiter; follicular adenoma
FLWNP (follicular lesion with nuclear pleomorphism)
Colloid absent; small clusters or tridimensional microfollicles made up of medium- to large-size follicular cells with moderate nuclear pleomorphism, irregular nuclear outlines, clearing and grooves (no pseudoinclusions or papillae); often oxyphilic cytoplasms; hemorrhage; fibrovascular tissue.
Small clusters of medium-sized follicular cells (see FN NOS) mixed with scattered aggregates of large cells with pleomorphic nuclei, clear chromatin and irregular nuclear outlines; no papillae or nuclear pseudoinclusions; fragments of fibrous tissue.
Follicular variant of papillary carcinoma; toxic or hyperplastic adenoma; follicular carcinoma.
OFN (oxyphilic follicular nodule)
Scant colloid; sheets or clusters of oxyphilic cells; hemorrhage; scattered inflammatory cells and reticular tissue in the background
Small aggregates of oxyphilic cells with cytoplasmic granules and large hyperchromatic pleomorphic nuclei; fibrous tissue; scant colloid globules
Oxyphilic neoplasm; oxyphilic hyperplastic nodule in thyroiditis
1Follicular nodule not otherwise specified (FN NOS) (Fig. 1), which is a true indeterminate category made up of medium-sized follicular cells with regular nuclear outlines exhibiting a predominant microfollicular aggregation (Table 1);
2Follicular lesion with nuclear pleomorphism (FLWNP) (Fig. 2), which shares some features with PC (elongation and clearing of follicular nuclei), although papillary structures or nuclear pseudoinclusions, which represent the hallmark of the cytodiagnosis of PC (Fig. 3), cannot be detected;
3Oxyphilic follicular nodule (OFN) showing clusters of cells with pleomorphic, round nuclei and prominent nucleoli with large oxyphilic granular cytoplasms.
For the categories of FLWNP and OFN, which show a relevant risk of carcinomatous occurrence,14, 15 a surgical treatment is routinely advised (see Table 1).
In the current study population, 49 patients underwent surgery. All surgical specimens were fixed in 10% buffered formaldehyde, embedded in paraffin, and cut into 5μm-thick sections that were stained with hematoxylin-eosin.
Immunostaining for the three above-mentioned antibodies, particularly in those cases that did not show a complete positivity of the panel in the smear, was also carried out on the histologic sections. This was performed using the avidin-biotin–peroxidase method with the same antibodies that were used for immunostains on thin layer smears.
Sensitivity, specificity, and overall accuracy of FNAB were calculated as follows: the category of FN NOS was included in the BL; the categories of FLWNP and OFN were included in the malignant lesions. Inadequate specimens and microscopic carcinomas incidentally found in the final pathologic examination were excluded, and the statistical analysis was performed considering only the histologic diagnosis of the sampled lesion.
Benign and malignant lesions were compared for cytologic and immunocytochemical patterns. Statistical analysis was performed by using a commercially available software package (SPSS 10.0; SPSS Inc., Chicago, IL). The chi-square test was used for categoric variables, and the t test was used for continuous variables. A P value less than 0.05 was considered significant.
The 99 cytologic cases were diagnosed as follows: 5 BL; 81 FP including oxyphilic neoplasms; and 13 PC. The 49 cases with histologic examination were cytologically diagnosed as follows: 2 BL including diagnosis of thyroiditis; 37 FP including 11 FN NOS, 5 OFN, and 21 FLWNP; and 10 PC. Among the surgically treated cases, all PCs and BLs were histologically confirmed; 14 out of the 21 (66.7%) FLWNPs were PCs, and 5 (23.8%) were follicular adenomas including 3 toxic adenomas. Among the 16 FN NOSs and OFNs, only 1 case was an oxyphilic carcinoma at histology (Table 2).
The results of the application of immunocytochemical stainings on the TLC slides showed that 37 cases out of 49 (75.5%) resulted in a complete concordant panel (Table 3). In the group of PCs, the complete antibody panel was positive in 9 out of 10 cases (90%), whereas the remaining case was only HBME-1 positive. In FP, the complete panel was positive in 15 cases (4 benign, 26.8%; and 11 malignant, 73.2%) and negative in 11 cases (42.3 %), all resulting BLs. In the FP group, FN NOS showed an entire negative panel in all 8 cases; OFNs were positive in 4 cases (1 oxyphilic carcinoma) and negative in 1 histologically benign case. The subgroup of FLWNP showed panel positivity in 11 cases (84.6%) with only 1 benign case histologically resulting as hyperplastic nodule in Hashimoto thyroiditis (see Figs. 4–6), and negative in 2 benign cases. In the BL group, the complete panel was negative in both cases.
Table 3. Cytologic–Histologic Comparison of Immunocytochemical Resultsa
37 cases with complete panel and 12 with partial positivity are presented.
FP (26 + 11)
FN NOS (8 + 3)
FLWNP (13 + 8)
PC (9 + 1)
Concerning the cases with an incomplete panel, Table 3 shows that the best correlation between cytoplasmic positivity and occurrence of malignancy was observed with HBME-1 and Galectin-3, whereas the RET protooncogene showed a more erratic behavior. HBME-1 and Galectin-3 were negative in 4 out of 5 cases of FP that were benign at histology and in 6 (HBME-1) and 5 (Galectin-3) cases that were positive out of 7 malignant lesions. RET protooncogene was positive in 4 out 5 cases of FP resulting benign and negative in 6 out of 7 cases resulting malignant at histology.
Correlation of the complete panel with its histologic results (Table 4) shows that the simultaneous negativity of the three antibodies is always associated with a benign diagnosis, whereas its complete positivity is associated with a malignant neoplasm in 20 out of 24 cases (83.3%). However, if OFNs (which are easily detected on cytologic examination) and hyperplastic nodules in Hashimoto thyroiditis (clinically diagnosed) are excluded, the percentage of malignant lesions with a positive finding for the complete panel increases to (87.5%).
Table 4. Cytologic–Histologic Comparison for Immunocytochemical Resultsa
P < 0.001.
Only 37 cases with complete panel are presented. The remaining 12 cases without the complete panel are presented in Table 3.
All panel −
All panel +
In Table 5, the correlation between nuclear pleomorphism (NP) of follicular cells and positivity of three antibodies in the FP cases is shown. The simultaneous absence of NP and antibody positivity is described in all 12 BLs, which also include 2 cases of toxic adenoma. The opposite (presence of NP and panel positivity) can be observed in all 20 malignancies. Four cases with simultaneous NP and panel positivity resulted as BLs at histology: 3 exhibited a cytologic picture of OFN, and 1 resulted as hyperplastic nodule in Hashimoto thyroiditis.
Table 5. Cytologic–Histologic Comparison for the Combination of Nuclear Pleomorphism and Immunocytochemistrya
P < 0.001.
NP: nuclear pleomorphism.
All carcinomas and oxyphilic neoplasms are included in the group with nuclear pleomorphism. One case of OFN, showing a complete negative panel, is not included.
All malignant lesions showed a positive immunocytochemical panel, whereas only 4 out of 17 benign lesions exhibited the same positivity (P < 0.001) (Table 4).
All benign lesions showed neither nuclear pleomorphism nor positivity of the panel, whereas nuclear pleomorphism and positivity of the panel were exhibited by almost all (20 out of 24) malignant lesions (P < 0.001). All carcinomas and oxyphilic neoplasms are included in the group with nuclear pleomorphism; however, only one case of OFN, showing a completely negative panel, is not included (Table 5).
In Table 6, the descriptive statistical analysis shows the differences in sensitivity, specificity, diagnostic accuracy, and positive and negative predictive values of the nuclear features of follicular cells, which are shown in Table 6. The same table analyzes the comparison between the NP and the complete immunocytochemical panel and also the combination of both. The exclusion of the OFN from the statistical analysis results in an overall sensitivity of 100%, a specificity of 92.3% and a diagnostic accuracy of 97%. These values are higher when compared to the results of the independent evaluations of NP and panel positivity.
Table 6. Descriptive Statistics for Single and Combined Morphologic and Immunocytochemical Tests
CYT only percentage
ICC only percentage
CYT + ICC (W/O OFN) percentage
CYT + ICC (W/O OFN) percentage
CYT: cytology; ICC: immunocytochemistry; W/O OFN: with oxyphilic follicular neoplasms; W/O OFN: without OFN.
Positive predictive value
Negative predictive value
Among the 49 cases with histologic examination, 18 (36.7%) were selected for a further evaluation of HBME-1, Galectin-3, and RET protooncogene, and all confirmed the cytologic results. The cases selected for immunohistochemistry included all 12 cases that had exhibited a discordance of the three antibodies of the panel, the 2 cytologically benign lesions, 3 papillary carcinomas, and 1 case of hyperplastic nodule in Hashimoto thyroiditis.
Thyroid nodules are a common finding in everyday clinical practice; however only a small percentage is malignant. A cytologic examination is directed toward establishing a precise diagnosis to distinguish malignant or suspicious nodules requiring surgical treatment from those that can be treated pharmacologically. The efficacy of FNAB to make a correct diagnosis of malignant lesions is definitely stated in literature.16–19
However, the cytodiagnostic group of follicular proliferations (FP) is an indeterminate category responsible for a high rate of inappropriate thyroidectomies.20–24 This FP category embraces benign lesions that show focal nuclear pleomorphism (e.g., hyperplastic nodules in Hashimoto thyroiditis, toxic adenomas) and malignant lesions with slight pleomorphism (e.g., minimally invasive follicular carcinoma), a correct cytologic diagnosis.
Therefore, in the current study, the FP group was divided into three subcategories (FN NOS, FLWNP, and OFN) based on the size and the presence of pleomorphism and irregularities of the follicular nuclei.14, 15 These morphologic criteria are summarized in Table 1.
Results on CS and TLC slides of lesions classified as FN NOS, FLWNP, and OFN, stressed the presence of nuclear pleomorphism of follicular cells as a distinctive feature suspicious for papillary carcinoma. However, 7 cases exhibiting nuclear pleomorphism at the preoperative cytology were diagnosed as BLs at thyroidectomy. Three cases were toxic adenomas, and 2 were hyperplastic nodules in a background of Hashimoto thyroiditis, and 2 were oxyphilic adenomas. However, Hashimoto thyroiditis can be easily diagnosed from clinical history and serum analysis (detection of circulating antibodies against thyroglobulin and thyroperoxidase; see Table 2).
To minimize the problem of an indeterminate cytologic diagnosis, immunohistochemical antibodies against antigens associated with malignancy were introduced. Among them, Galectin-3, HBME-1, and RET showed the highest reliability, particularly in differentiated carcinoma.4, 6, 7, 12, 13, 25–28
However, the cytologic material used for CS is, in most instances, unavailable for additional investigation either because of inadequacy of cellularity or because of the presence of blood and debris, which decreases immunoreaction efficacy.
Cytospin preparations are a satisfactory solution for the scant amount of cells and to obtain an adequate number of slides to carry out immunocytochemistry.29 However, this method requires an immediate processing to avoid alterations of cellular morphology and cannot be applied to all cases.
Liquid-based cytology and thin-layer preparation techniques may be effective innovations. Several studies have suggested that TLC may achieve a diagnostic sensitivity as high as that found in conventional preparations.30–35 Moreover, immunocytochemistry can be reliably performed on TLC as accurately as in conventional smears and histology.36–39 The quality of the immunocytochemical reaction on thin layer preparations, as far as morphological details and purity of background are concerned, is sometimes better than conventional smears.31–34, 36, 40
This finding is significant, because additional slides for immunocytochemistry can be easily prepared from stored material. However, the possibility of false positive and false negative results, probably caused by either difficulties in inactivating the endogenous biotin present in large amounts in thyroid tissue9, 11, 12 or by nonperfect diagnostic efficacy of each single antigen, is noted. Therefore, only the concordance of a panel should be considered in the diagnosis of thyroid nodules.
The results of the application of immunocytochemistry in 12 nonconcordant cases highlight (Table 3) the high diagnostic accuracy of HBME-1 and, to a minor extent, of Galectin-3. RET protooncogene was the least accurate of the three. However, in both cytologic and histologic fields, morphologic analysis of each single thyroid neoplasm cannot reliably be replaced by immunocytochemistry. The lack of reliable morphologic criteria for a diagnosis of malignancy in a limited number of thyroid neoplasms requires adherence to strict criteria to avoid unnecessary treatments, even in the preoperative step. In this setting, the use of more than one immunomarker is a further guarantee for a correct diagnostic approach, especially in cases of oxyphilic lesions and hyperplastic nodules in Hashimoto thyroiditis. The distinctive features of oxyphilic cells (rich in cytoplasmic mitochondria) and the influence of inflammation on the follicular cells of the hyperplastic nodule in HT can be responsible for false cytoplasmic positivity, as seen in the cases presented in the current study.
Since the introduction of the processor for TLC slides, there have been only a few reported experiences in literature dealing with immunocytochemistry applied on thin layer slides. Leung and colleagues36 reported satisfactory results with ICC applied on TLC slides from different body sites, except for some lymphoma markers. Dabbs and colleagues40 showed that ICC performed on TLC from a variety of neoplastic and nonneoplastic samples resulted in greater intensity and distribution of proper staining compared with CS. Tabbara and colleagues41 demonstrated that estrogen and progesterone receptor status can be evaluated in FNA specimens from breast carcinoma by useof the thin- layer slide technique and by applying tested immunostaining protocols.
On the basis of ICC alone, the current research showed that 37 lesions (75.5%) had concordant expression of the three antibodies of the panel. All 13 cases exhibiting a negative panel were benign upon histologic examination, whereas among 24 cases with positive panels, 4 (16.7%) were BLs, and 20 (83.3%) were malignant.
The combination of nuclear pleomorphism and positivity of the antibody panel resulted in a slight increase of the specificity and diagnostic accuracy of FNAB (75% and 89% respectively), but this improvement became quite striking when oxyphilic neoplasms, showing a distinctive cytologic picture, were excluded (92.3% and 97% respectively).
Based on results of the present investigation, the management of a thyroid lesion undergoing FNAB includes, 1) accurate examination of clinical history and serum analysis to rule out the possibility of a toxic adenoma or Hashimoto thyroiditis; 2) evaluation of nuclear pleomorphism of follicular cells (including elongation and irregularity of nuclear outlines as described in Table 1) and distinction from nuclear pleomorphism of oxyphilic cells; 3) assessment of the immunocytochemical positivity of cells for the panel made up of HBME-1, Galectin-3, and RET protooncogene. When the above criteria are correctly applied, the overall diagnostic accuracy of cytology in follicular proliferations is 97% with a positive predictive value of 95% and a negative predictive value of 100% as shown in Table 6. Moreover, when thyroiditis is clinically ruled out, the diagnostic accuracy of both morphology and immunocytochemistry increase to 100%.
In conclusion, because it improves accuracy of the conventional cytological evaluation and, thus, can enable a better selection of candidates for surgery, the antibody panel may be a crucial step toward achieving a correct preoperative diagnosis of thyroid nodules. In this perspective, TLC42 may be a valid alternative to CS in achieving a satisfactory result of immunocytochemical stainings.
The authors thank Virginia A. LiVolsi, M.D. and Zubair W. Baloch, M.D. of the University of Pennsylvania, Philadelphia, PA, for their precious suggestions; Federica Cianfrini and Antonella Evangelista of the Division of Anatomic Pathology and Histology of Catholic University of Rome for their technical support; and Louise Memmoli for her linguistic review of the manuscript.