A logistic regression model predicting malignancy in ... - eJManager

Original Research

Journal of Interdisciplinary Histopathology www.scopmed.org DOI: 10.5455/jihp.20141027123826

A logistic regression model predicting malignancy in follicular thyroid lesions based on CD56 expression and patient’s age Eman Abdelzaher, Azza Mohamed Rizk, Maram Allam ABSTRACT

Department of Pathology, Faculty of Medicine, University of Alexandria, Alexandria, Egypt Address for correspondence: Eman Abdelzaher, Department of Pathology, Faculty of Medicine, University of Alexandria, Alexandria, Egypt. Tel.: +201001236757, E-mail: pathologisteman@ yahoo.com Received: August 06, 2014 Accepted: October 27, 2014 Published: November 17, 2014

Objectives: Follicular thyroid lesions represent a heterogeneous group with variable biological behavior. Discrimination between benign and malignant follicular thyroid lesions by histologic examination poses a significant challenge to pathologists. Recently, CD56 has attracted attention as a potential diagnostic marker in follicular thyroid lesions. Materials and Methods: A retrospective immunohistochemical and statistical analyses of CD56 expression in 68 cases of follicular thyroid lesions were done. Results: CD56 expression was significantly higher in papillary thyroid carcinoma (PTC) (conventional and follicular variants) compared to other categories of follicular thyroid lesions, and it was significantly higher in the follicular variant of papillary carcinoma compared to follicular adenomas (FA) and hyperplastic nodules (HN). A bivariate analysis showed that the malignancy in follicular thyroid lesions was significantly associated with higher CD56 expression and older patients’ age. Both parameters were used to build a logistic regression model and equation for prediction of malignancy in follicular thyroid lesions. On the application of the equation to the studied cases, it was found to conform to their diagnostic nature. Conclusions: We conclude that CD56 could serve as a good positive marker favoring the diagnosis of PTC (conventional and follicular variants) over other categories of follicular thyroid lesions. It also could be of value in differentiating follicular variant of papillary carcinoma from other lesions of benign nature such as FA and HN. CD56 expression is significantly associated with malignant behavior in thyroid follicular lesions, which suggests a role for CD56 in tumorogenesis. More importantly, the developed model and equation based on CD56 expression and patient's age might be of value in the prediction of malignancy in follicular thyroid lesions, but it needs to be validated on a large scale accompanied by follow-up. KEY WORDS: CD56, follicular thyroid lesions, logistic regression model, prediction of malignancy

INTRODUCTION Thyroid nodules of various biological natures occur in approximately 10% of the population worldwide, predominately adult females [1]. Clinically detected thyroid nodules include non-neoplastic lesions such as hyperplastic nodules (HN) and neoplastic lesions such as follicular adenoma (FA), follicular carcinoma (FC) and follicular variant of papillary thyroid carcinoma (FVPC). These lesions histologically share a follicular-patterned morphology and are traditionally grouped under the umbrella term “follicular thyroid lesions” [1-3]. The “gold standard” in the diagnosis of thyroid nodules remains to be the routine histopathological examination of hematoxylin and eosin stained sections. Nevertheless, the histopathological interpretation of thyroid lesions with a follicular growth pattern 205

poses a challenge to pathologists; mainly the distinction between HN and FA, FA and minimally invasive FC as well as FA/FC and the FVPC [1,3,4]. Previous studies have variably employed several ancillary techniques as immunohistochemistry, morphometric studies and molecular methods in order to achieve a definitive diagnosis of different types of follicular thyroid lesions [2-7]. Researchers have investigated the expression of galectin-3, cytokeratin 19, p53, bcl-2, HBME-1, and Ki-67. In addition, certain mutations have been observed such as PET mutations and mutations in the first exon of K-ras in PTC and mutations in codon 61 of H-ras and N-ras in FC [8-12]. Although some of these techniques are useful, they should be used cautiously as they are neither consistent nor reliable [8-12]. This justifies the continuous pursuit of new J Interdiscipl Histopathol 2014; 2(4): 205-212

Abdelzaher, et al.: A model predicting malignancy in thyroid lesions

markers that would ensure an accurate diagnosis of follicular thyroid lesions. CD56, a neural cell adhesion molecule, also known as AntiLeu-19, was originally identified as a glycoprotein on natural killer cells [13]. CD56 is ubiquitously expressed in the central nervous system (CNS), peripheral nerves and skeletal muscles, most types of neuroendocrine cells, and various epithelia (e.g. enterocytes and the newly formed bile ductular cells). It is also expressed in ovarian stromal cells, uterine smooth muscle cells and osteoblasts [14]. CD56 has been used as a diagnostic marker in certain tumors such as acute and chronic myeloid leukemia, ovarian sex cord-stromal tumors, Wilms’ tumor, neuroblastoma and ganglioneuroblastoma, phaeochromocytoma/paraganglioma and certain CNS tumors [13,14]. Furthermore, CD56 was proved to have a prognostic impact in certain malignancies. Loss of CD56 correlates with poor prognosis in colonic and pancreatic carcinomas [15,16]. The aim of the present study was to investigate the diagnostic utility of CD56 protein expression in different follicular thyroid lesions using immunohistochemistry.

MATERIALS AND METHODS

The immunostained slides were evaluated for the extent (percentage) of positive cells (0: 25-50%, and 3+: >50%) and intensity of staining reaction (1 = weak, 2 = moderate, 3 = strong). A total score was calculated by summation of the aforementioned extent and intensity scores [18].

Statistical Analysis Statistical analysis was performed using Statistical Package for Social Sciences (SPSS® Statistics 20, International Business Machines Corporation, New York, USA). Quantitative data were described using median, minimum and maximum, as well as mean and standard deviation (SD). Qualitative data were described using number and percentage. Various statistical tests (Chi-square test [χ2], independent t-test [t], Mann–Whitney [U], Kruskal–Wallis test [H], Pair-wise follow-up analysis, and Spearman’s correlation coefficient [ρ]) were used to evaluate and correlate the results. Significance test results were judged at 5% level. A logistic regression model was built to predict malignancy among thyroid lesions using the patient’s age and the total CD56 staining score. The diagnostic performances of patient’s age, total CD56 staining score and the developed model were evaluated using receiver operating characteristic (ROC) curve analysis. Youden Index was used to estimate the cut-off point at which the highest diagnostic accuracy was reached.

Patients and Tissue Samples

RESULTS

The present study comprised 68 retrospective cases of follicular thyroid lesions retrieved from the archives of the Pathology Department, Faculty of Medicine, University of Alexandria from the year 2005 to the year 2012. Complete clinical and operative data were available for all patients. Histopathological diagnosis was done according to the WHO criteria [17].

Clinicopathological Findings

Immunohistochemistry Tissue sections were deparaffinized in xylene and rehydrated through graded alcohol. Endogenous peroxidase activity was blocked by hydrogen peroxide. Antigen retrieval was done by immersion of tissue sections in a 10 mM citrate buffer (pH 6.0). Immunohistochemical staining was performed using an avidin-biotinylated immunoperoxidase methodology. The primary mouse monoclonal antibody (CD56, Clone 123C3, Dako, Denmark) was applied in a 1:100 dilution. The bound antibody was detected by the Ultra-Vision Detection System anti-polyvalent, HRP/DAB (ready-to-use) (Thermo Scientific™, USA). Negative and positive controls (neuroblastoma) were included in all runs.

Evaluation of Immunohistochemical Staining Cases were considered CD56 positive if they showed cytoplasmic ± membranous staining in ≥10% of the cells [18]. J Interdiscipl Histopathol 2014; 2(4): 205-212

The age of the patients ranged from 20 to 70 years (M = 45, SD = 10.8). 64 patients (94.1%) were females, and four patients (5.9%) were males with a M: F ratio of 1:16. All the studied cases had normal serum levels of T3, T4 and TSH. Ultrasound examination was done for all patients, and pre-operative fine-needle aspiration cytology was performed in 63 patients (92.7%). 52 patients (76.5%) underwent total thyroidectomy, 11 patients (16.2%) underwent hemithyroidectomy, 3 patients (4.4%) underwent subtotal thyroidectomy and 2 patients (2.9%) underwent total thyroidectomy with radical neck dissection. By gross examination, 53 cases (77.9%) showed a single nodule and 15 cases (22.1%) presented as a dominant nodule in a multinodular goiter. The nodules in 39 cases (57.35%) were encapsulated, while in the remaining 29 cases (42.65%) they were non- or partially encapsulated. The size of the nodules (largest dimension) ranged from 0.5 to 5 cm with a mean of 2.75 cm ±SD = 0.33. Cut sections were grayish white in 52 cases (76.47%) and glistening brown in 16 cases (23.53%). Areas of hemorrhage and necrosis were seen in six cases (8.8%). Cystic degeneration was noted in six cases (8.8%), and two cases (2.9%) showed protruding papillae from the cut surface. 206


Microscopically, the studied cases included 30 cases (44.1%) of follicular thyroid lesions with benign behavior (15 cases [22%] of HN, and 15 cases [22%] of FA), and 33 cases (48.5%) of malignant tumors (18 cases [26.6%] of PTC [including 5 cases [7.4%] of conventional PTC and 13 cases [19.2%] of FVPC] and 15 cases [22%] of minimally invasive FC). In addition, 5 cases (7.4%) of well-differentiated tumor of uncertain malignant potential (WDT-UMP) were included, in which questionable capsular invasion and incomplete nuclear features of PTC were detected.

Results of Immunohistochemistry Overall, CD56 was expressed in 31 cases (45.6%). The overall total score ranged from 0 to 6 with a mean of 1.78. CD56 showed a tendency towards a positive correlation with patients’ age (ρ = 0.239, P = 0.050) and was significantly higher expressed in male patients (U = 50.5, P = 0.04). The correlation between

CD56 expression and nodule size was statistically insignificant (ρ = 0.039, P = 0.750). The results of CD56 expression in the studied histopathological types are summarized in Table 1 and illustrated in Figure 1. Total score of CD56 expression was statistically significant among the studied groups of follicular thyroid lesions (H = 41.865, P ≤ 0.001). Pair-wise comparison showed that CD56 expression was significantly higher in PTC (conventional and FVPC) compared to FC, FA and HN (P = 0.008, P < 0.001 and P < 0.001 respectively). In contrast, no statistically significant difference in CD56 expression was found between HN and FA (P = 1.000), HN and FC (P = 0.917), as well as between FA and FC (P = 0.157). Also, no statistically significant difference was found between WDT-UMP and any of the studied types.

Table 1: CD56 expression in the different histopathological types of follicular thyroid lesions studied Studied groups HN FA PTC FC WDT-UMP

No. (%) 15 (22) 15 (22) 18 (26.6) 15 (22) 5 (7.4)

Frequency of CD56 positivity (No. (%))

Cytoplasmic (%)

Pattern of staining Cytoplasmic and membranous (%)

M (SD)

Total staining score Mdn (range)

3 (20) 0 (0) 18 (100) 7 (46.6) 3 (60)

3 (20) 11 (61.1) 3 (42.9) 3 (60)

0 (0) 7 (38.8) 4 (57.1) 0 (0)

0.53 (1.1) 4.5 (1.2) 1.9 (2.2) 1.8 (1.7)

0 (0-3) 5 (2-6) 0 (0-5) 2 (0-4)

Test (P value) H=41.87 P≤0.001

HN: Hyperplastic nodule, FA: Follicular adenoma, PTC: Papillary thyroid carcinoma, FC: Follicular carcinoma, WDT-UMP: Well-differentiated tumor of uncertain malignant potential, M: Mean, SD: Standard deviation, Mdn: Median, H: Kruskal–Wallis test

a

d

c

b

e

f

Figure 1: Immunohistochemical expression of CD56 in follicular thyroid lesions: (a) Focal weak cytoplasmic staining in a case of hyperplastic nodule (×200) (b) Negative staining in a case of follicular adenoma (×40) (c) A case of well-differentiated tumor of uncertain malignant potential showing cytoplasmic immunopositivity of moderate intensity (×400) (d) A case of conventional papillary thyroid carcinoma showing diffuse cytoplasmic staining of moderate intensity (×200) (e) A case of follicular variant of papillary thyroid carcinoma showing cytoplasmic positive staining with membranous accentuation of strong intensity (×200), (f) A case of follicular carcinoma showing cytoplasmic staining of strong intensity (×400) 207

J Interdiscipl Histopathol 2014; 2(4): 205-212


On considering conventional PTC and FVPC as separate groups, there was a statistical significance in the total score of CD56 expression among the studied groups (H = 42.150, P ≤ 0.001). Pair-wise comparison showed that CD56 expression was significantly higher in FVPC compared to FA and HN (P < 0.001 and P < 0.001 respectively). Similarly, CD56 expression was significantly higher in conventional PTC compared to FA and HN (P < 0.001 and P = 0.002 respectively). However, CD56 expression showed no statistically significant difference between PTC and FVPC on one hand and FC on the other hand (P = 0.119 and P = 0.061 respectively). Also, no statistically significant difference in CD56 expression was found between HN and FA (P = 1.000), PTC and FVPC (P = 1.000), HN and FC (P = 1.000), as well as between FA and FC (P = 0.235). In addition, no statistically significant difference was found between WDT-UMP and any of the studied types. In order to assess the relation between CD56 expression and other clinicopathological factors on one hand and tumor’s behavior on the other hand, a bivariate analysis was done and showed that malignancy was significantly associated with higher CD56 expression (U = 140.5, P < 0.001) and older patients’ age (t = 2.72, P = 0.008). Subsequently, a logistic regression model [Table 2] and equation were built to predict malignancy in follicular thyroid lesions:

ROC curve analyses showed that at a cut-off point of score 2 for CD56 expression, 36 years for age, and 0.598 for the developed model, sensitivity and specificity were at their highest levels. On comparing the diagnostic performance of patients’ age, total CD56 score, and the developed model, it was found that the developed model was better than the other two parameters in prediction of malignancy among the studied cases of follicular thyroid lesions [Table 3, Figure 2]. On the application of the built equation to the studied cases, it was found to conform to their diagnostic nature.

DISCUSSION Follicular thyroid lesions represent a heterogeneous group that comprises both non-neoplastic and neoplastic lesions with varying biological behavior [19]. Follicular-patterned thyroid lesions represent a diagnostic grey zone even to experienced pathologists with a significant interobserver variability [5,19]. Accurate pathological diagnosis is essential for optimal clinical management [20]. Most of the patients in our study were adult females, a finding that coincides with the study of Mokhtari et al. [21] who reported that the mean age of their patients was 42.4 years

Probability of malignancy = eY/(1+eY) (e) is a natural number, it is approximately equal to 2.71828 Given that Y = (−0.027) × Age in years +0.942 × Total CD56 score The diagnostic accuracy of CD56 expression, patients’ age and the developed model was further tested by ROC curve analysis in order to detect a cut-off point above which malignancy is predicted. Table 2: The logistic regression model for prediction of malignancy in the studied cases of follicular thyroid lesions Independent predictors

χ2 (P-value)

Nagelkerke

Model equation

Age and total CD56 staining score

32.4 (36

0.598

Cut-off value

(95% CI) (%) Sn

Sp

PPV

NPV

72.73 54.5-86.7 72.73 54.5-86.7 75.76 57.7-88.9

93.33 77.9-99.2 56.67 37.4-74.5 93.33 77.9-99.2

92.3 74.9-99.1 64.9 47.5-79.8 92.6 75.7-99.1

75.7 58.8-88.2 65.4 44.3-82.8 77.8 60.8-89.9

AUC: Area under the curve, 95% CI: 95% Confidence interval, Sn: Sensitivity, Sp: Specificity, PPV: Positive predictive value, NPV: Negative predictive value J Interdiscipl Histopathol 2014; 2(4): 205-212

208


with a female preponderance and a male to female ratio of 1:2.9. We investigated the immunohistochemical expression of CD56 in follicular thyroid lesions. CD56 is cell surface adhesion glycoprotein, of the immunoglobulin gene superfamily [13,14]. It has three isoforms that vary in their cytoplasmic domains. The first (120 kDa) is cell membrane anchored, the second (140 kDa) has a short cytoplasmic domain, and the third (180 kDa) has a long cytoplasmic domain [22]. The used primary antibody in this study was CD56 (Clone 123C3), that recognizes a protein consisting of at least two isoforms [23]. The pattern of CD56 expression varied from cytoplasmic to cytoplasmic with membranous accentuation. Different previous studies [18,24-33], have reported different staining patterns for CD56 that could be attributed to the different antibody clones, and thus isoforms, used. There was a tendency towards a positive correlation between patients’ age and CD56 expression. This might be due to the higher incidence of malignancy with advanced age. In addition, there was a statistically significant association between CD56 expression and male gender. However, this result should be interpreted cautiously due to the small number of male patients. Among benign lesions, CD56 was weakly positive in 20% of HN cases and was negative in FA cases. This contrasts with previous studies that have reported higher frequencies of CD56 expression in HN and FA [18,26,27,29,30]. Shin et al. [26] reported that 58.3% of HN and 100% of FA cases showed immunopositivity for CD56. Park et al. [18] found that 90.5% of HN cases and 93.3% of FA cases were positive for CD56. Abd El Atti and Shash [27] reported that 87.5% of cases of HN and 91.7% of FA cases showed CD56 expression. The studies of both El Demellawy et al. [29] and Migita et al. [30] mentioned that CD56 expression was found in 100% of the studied cases of HN and FA. In the current study, 60% of WDT-UMP cases were positive for CD56. Abd El Atti and Shash [27] recorded a lower frequency (30%) of CD56 expression in WDT-UMP. As regards FC, the present study revealed positivity for CD56 in 46.6% of the studied cases. In contrast, more frequent CD56 expression in FC cases was documented by both Park et al. [18] and El Demellawy et al. [29] who reported CD56 positive staining in 100% and 82.6% of FC cases respectively. All cases (100%) of PTC in our study (conventional PTC and FVPC) showed positive CD56 expression. These results are in general agreement with those of Scarpino et al. [25] who found that 70.4% of the studied PTC cases were positively stained for CD56.

209

On the other hand, Migita et al. [30] reported that only 33% of PTC cases were positively stained for CD56. Less frequent CD56 expressions in papillary carcinomas were reported by other investigators. The studies of Park et al. [18] and Abd El Atti and Shash [27] revealed that only 7.5% and 17.2% of PTC cases (including its follicular variant) were positively stained for CD56 respectively. In addition, Ozolins et al. [28] reported that CD56 expression was detected in only 4% of papillary carcinoma cases. Considering the total score of CD56 expression, the results of our study showed a significantly higher expression of CD56 in PTC (including its FVPC) compared to the other studied categories. This suggests that CD56 expression could be used as a good positive marker distinguishing PTC from other follicular thyroid lesions. Park et al. [18] reported that CD56 expression showed a statistically significant difference between papillary carcinoma cases (as a negative marker) when compared to HN, FA and FC. On the other hand, Etem et al. [34] reported near equal percentages of CD56 positivity in PTC and other follicular lesions including HN, FA and FC, thus concluded that CD56 does not discriminate between these categories. On considering conventional PTC and FVPC as separate groups, our results showed that CD56 expression was significantly higher in each of them when compared to FA and HN. However, CD56 expression showed no statistically significant difference between PTC, FVPC as separate groups when compared to FC. Thus, CD56 proved to be of no value in the discrimination between FVPC and FC. Abd El Atti and Shash [27] reported that CD56 distinguished FVPCs (as a negative marker) from other follicular patterned nodules (FA, HN and WDT-UMP) with a statistically significant difference. Similarly, other studies [18,21,26,27,29,31] have reported that CD56 could serve as a good negative marker for PTC and FVPC distinguishing them from other CD56-positive thyroid lesions with a high sensitivity and specificity. Needless to say, the discrimination between benign and malignant follicular thyroid lesions has a major impact on patients’ management. However, the histological similarities between these lesions blur the distinction. Moreover, the unusual histologic criteria for diagnosing thyroid malignancy, namely capsular and/or vascular invasion for minimally invasive FC and nuclear characteristics for papillary carcinoma, render the discrimination more challenging [5,19]. Much attention has been directed towards the identification of molecular or immunohistochemical markers that can help distinguish follicular thyroid lesions of benign behavior from malignant lesions [35]. J Interdiscipl Histopathol 2014; 2(4): 205-212


In our study, CD56 total score was significantly higher in malignant tumors compared to lesions with benign behavior. Several previous studies, on tissues other than thyroid, revealed that CD56 was exclusively or predominantly expressed in malignant tumors compared to their benign counterparts or normal tissues [36-42]. CD56 was significantly expressed in small cell carcinoma of the lung [36], chronic myeloid leukemia [37], multiple myeloma [38], ovarian serous neoplasms [39], malignant gastrointestinal stromal tumors [40], malignant mesothelioma [41], and colorectal carcinoma [42], compared to their benign counterparts. The findings of our study together with the aforementioned studies, suggest that CD56 plays a role in malignant transformation, a finding that mandates further investigation to clarify the exact role of CD56 in the process of tumorgenesis. In stark contrast to our findings, previous studies employing CD56 on thyroid lesions have shown higher expression of CD56 in benign cases as compared to malignant cases [27,31-33]. The noted discrepancy in the results of CD56 expression in follicular thyroid lesions among different studies could be attributed to the different number of the studied cases, the utilization of different antibody clones with different antibody dilution, and different cut-off values of positive cells used to establish positivity. For example, Shin et al. [26] and Park et al. [18], utilized (Ncl-CD56-1B6), moreover, Shin et al. [26] used a 1:300 dilution. Abd El Atti and Shash [27] used CD56 (MS-204-R7). Migita et al. [30] performed immunostaining on frozen sections, considered 5% as a cut-off value of positivity, and cases were recognized as positive if they demonstrated cytoplasmic and membranous staining. Ozolins et al. [28] performed immunostaining on FNA smears and considered cases as positive if they showed membranous staining pattern. In addition, one should consider the geographical, environmental, and racial differences. In this study, a bivariate and ROC curve analyses showed that CD56 expressions (at score >2) and patients’ age (>36 years) were significant independent predictors of malignancy in thyroid lesions with a high sensitivity and specificity.

All the studied cases of FA and HN had a probability of malignancy below 0.598, a result that favors benignity, and this conforms to the benign nature of FA and HN. For example, in a case of HN from our studied group, the patient’s age was 24 years and the total score of CD56 expression was 0. On the calculation of the Y value, it was found to be −0.648, and the probability of malignancy was 0.343 which is 0.598. Such a result favored malignancy that is supported by the malignant nature of PTC. In addition, in a case FVPC from our study, the age of the patient was 40 years and the CD56 total score was 5. On the calculation of the Y value, it was equal to 3.62 and the probability of malignancy in this case was equal to 0.974 which is >0.598. Such a result favored malignancy that is supported by the malignant nature of FVPC. In a case FC from our study, the age of the patient was 54 years and the CD56 total score was 3. On the calculation of the Y value, it was equal to 1.362 and the probability of malignancy in this case was equal to 0.796 which is >0.598. Such a result favored malignancy that is supported by the malignant nature of FC.

A logistic regression model was developed to predict malignancy in follicular thyroid lesions using CD56 expression and patients’ age. It was found that the sensitivity and capability of the developed model to predict malignancy was higher than that of CD56 expression and patients’ age.

The formulated equation was also applied to all the studied cases of WDT-UMP, and it was found that three cases out of the studied 5 cases (60%) had a probability of malignancy >0.598. Our result suggests that these cases might pursue a malignant course. While the other 2 cases (40%) had a probability of malignancy below 0.598 suggesting that these cases might be of a benign nature. Confirmation of our results requires follow-up of these cases.

Subsequently, an equation was built to detect the probability of malignancy in thyroid lesions. If the probability of malignancy in a given case of follicular thyroid lesions was >0.598, this favors malignancy. Conversely, if the resultant value was ≤0.598, it is more likely for this lesion to be of a benign nature. On the application of the equation to the studied cases, it was found to conform to their diagnostic nature.

For example, in a case diagnosed as WDT-UMP, the age of the patient was 42 years, and the total score of CD56 expression was 4. On the calculation of the Y value, it was equal to 2.634 and the probability of malignancy was equal to 0.933 which is >0.598. This result suggests that this case might pursue a malignant course. Follow-up of this case may disclose its nature.


210


From the aforementioned results, it could be suggested that the developed equation may play an important role in the prediction of malignancy of equivocal cases of follicular thyroid lesions especially cases of WDT-UMP. From the results of the present study it can be concluded that CD56 could be utilized as a good positive marker favoring the diagnosis of PTC (including its FVPC) over other categories of follicular thyroid lesions (HN, FA, and FC). It also could be of value in differentiating FVPC from other lesions of benign nature including FA and HN. CD56 expression is significantly associated with malignant behavior in thyroid follicular lesions, which suggests a role for CD56 in tumorogenesis. More importantly, the developed model and equation based on CD56 expression and patient's age might be of value in the prediction of malignancy in follicular thyroid lesions, but it needs to be validated on a large scale accompanied by follow-up.

15.

16. 17.

18. 19. 20. 21.

22. 23.

REFERENCES 1.

Rosai J, Carcangiou ML, De Lellis RA. Tumors of the thyroid gland. Atlas of Tumor Pathology. 3rd Series, Fascicle 5. 2. Oestreicher-Kedem Y, Halpern M, Roizman P, Hardy B, Sulkes J, Feinmesser R, et al. Diagnostic value of galectin-3 as a marker for malignancy in follicular patterned thyroid lesions. Head Neck 2004;26:960-6. 3. Rezk S, Khan A. Role of immunohistochemistry in the diagnosis and progression of follicular epithelium-derived thyroid carcinoma. Appl Immunohistochem Mol Morphol 2005;13:256-64. 4. Cheung CC, Carydis B, Ezzat S, Bedard YC, Asa SL. Analysis of ret/ PTC gene rearrangements refines the fine needle aspiration diagnosis of thyroid cancer. J Clin Endocrinol Metab 2001;86:2187-90. 5. Lloyd RV, Erickson LA, Casey MB, Lam KY, Lohse CM, Asa SL, et al Observer variation in the diagnosis of follicular variant of papillary thyroid carcinoma. Am J Surg Pathol 2004;28:1336-40. 6. LiVolsi VA, Baloch ZW. Follicular neoplasms of the thyroid: View, biases, and experiences. Adv Anat Pathol 2004;11:279-87. 7. Aiad H, Abdou A, Bashandy M, Said A, Ezz-Elarab S, Zahran A. Computerized nuclear morphometry in the diagnosis of thyroid lesions with predominant follicular pattern. Ecancermedicalscience 2009;3:146. 8. Fischer S, Asa SL. Application of immunohistochemistry to thyroid neoplasms. Arch Pathol Lab Med 2008;132:359-72. 9. Niedobitek C, Niedobitek F, Lindenberg G, Bachler B, Neudeck H, Zuschneid W, et al Expression of galectin-3 in thyroid gland and follicular cell tumors of the thyroid. A critical study of its possible role in preoperative differential diagnosis. Pathologe 2001;22:205-13. 10. Erickson LA, Jin L, Wollan PC, Thompson GB, van Heerden J, Lloyd RV. Expression of p27kip1 and Ki-67 in benign and malignant thyroid tumors. Mod Pathol 1998;11:169-74. 11. Giordano TJ, Kuick R, Thomas DG, Misek DE, Vinco M, Sanders D, et al Molecular classification of papillary thyroid carcinoma: Distinct BRAF, RAS, and RET/PTC mutation-specific gene expression profiles discovered by DNA microarray analysis. Oncogene 2005;24:6646-56. 12. Vasko V, Ferrand M, Di Cristofaro J, Carayon P, Henry JF, de Micco C. Specific pattern of RAS oncogene mutations in follicular thyroid tumors. J Clin Endocrinol Metab 2003;88:2745-52. 13. Lanier LL, Testi R, Bindl J, Phillips JH. Identity of Leu-19 (CD56) leukocyte differentiation antigen and neural cell adhesion molecule. J Exp Med 1989;169:2233-8. 14. Rutishauser U, Acheson A, Hall AK, Mann DM, Sunshine J. The neural cell adhesion molecule (NCAM) as a regulator of cell-cell interactions. Science 1988;240:53-7.

211

24. 25.

26. 27. 28.

29. 30.

31. 32.

33.

34. 35.

36.

Maréchal R, De Schutter J, Nagy N, Demetter P, Lemmers A, Devière J, et al Putative contribution of CD56 positive cells in cetuximab treatment efficacy in first-line metastatic colorectal cancer patients. BMC Cancer 2010;10:340. Aloysius MM, Zaitoun AM, Awad S, Ilyas M, Rowlands BJ, Lobo DN. Mucins and CD56 as markers of tumour invasion and prognosis in periampullary cancer. Br J Surg 2010;97:1269-78. DeLellis RA, Lloyd RV, Heitz PU, Eng C, editors. WHO Classification of Tumors. Pathology and Genetics. Tumours of Endocrine Organs. Lyon, France: International Agency for Research on Cancer (IARC) Press; 2004. p. 52-123. Park WY, Jeong SM, Lee JH, Kang HJ, Sin DH, Choi KU, et al. Diagnostic value of decreased expression of CD56 protein in papillary carcinoma of the thyroid gland. Basic Appl Pathol 2009;2:63-8. Suster S. Thyroid tumors with a follicular growth pattern: Problems in differential diagnosis. Arch Pathol Lab Med 2006;130:984-8. Gharib H, Papini E. Thyroid nodules: Clinical importance, assessment, and treatment. Endocrinol Metab Clin North Am 2007;36:707-35, vi. Mokhtari M, Eftekhari M, Tahririan R. Absent CD56 expression in papillary thyroid carcinoma: A finding of potential diagnostic value in problematic cases of thyroid pathology. J Res Med Sci 2013;18:1046-50. Hildebrandt H, Mühlenhoff M, Gerardy-Schahn R. Polysialylation of NCAM. Adv Exp Med Biol 2010;663:95-109. Schol DJ, Mooi WJ, van der Gugten AA, Wagenaar SS, Hilgers J. Monoclonal antibody 123C3, identifying small cell carcinoma phenotype in lung tumours, recognizes mainly, but not exclusively, endocrine and neuron-supporting normal tissues. Int J Cancer Suppl 1988;2:34-40. El Demellawy D, Nasr A, Alowami S. Application of CD56, P63 and CK19 immunohistochemistry in the diagnosis of papillary carcinoma of the thyroid. Diagn Pathol 2008;3:5. Scarpino S, Di Napoli A, Melotti F, Talerico C, Cancrini A, Ruco L. Papillary carcinoma of the thyroid: Low expression of NCAM (CD56) is associated with downregulation of VEGF-D production by tumour cells. J Pathol 2007;212:411-9. Shin MK, Kim JW, Ju US. CD56 and high molecular weight cytokeratin as diagnostic markers of papillary thyroid carcinoma. Korean J Pathol 2011;45:477-84. Abd El Atti RM, Shash LS. Potential diagnostic utility of CD56 and claudin-1 in papillary thyroid carcinoma and solitary follicular thyroid nodules. J Egypt Natl Canc Inst 2012;24:175-84. Ozolins A, Narbuts Z, Strumfa I, Volanska G, Stepanovs K, Gardovskis J. Immunohistochemical expression of HBME-1, E-cadherin, and CD56 in the differential diagnosis of thyroid nodules. Medicina (Kaunas) 2012;48:507-14. El Demellawy D, Nasr AL, Babay S, Alowami S. Diagnostic utility of CD56 immunohistochemistry in papillary carcinoma of the thyroid. Pathol Res Pract 2009;205:303-9. Migita K, Eguchi K, Kawakami A, Ida H, Fukuda T, Kurata A, et al Detection of Leu-19 (CD56) antigen on human thyroid epithelial cells by an immunohistochemical method. Immunology 1991;72:246-9. Shahebrahimi K, Madani SH, Fazaeli AR, Khazaei S, Kanani M, Keshavarz A. Diagnostic value of CD56 and nm23 markers in papillary thyroid carcinoma. Indian J Pathol Microbiol 2013;56:2-5. Nechifor-Boila A, Catana R, Loghin A, Radu TG, Borda A. Diagnostic value of HBME-1, CD56, Galectin-3 and Cytokeratin-19 in papillary thyroid carcinomas and thyroid tumors of uncertain malignant potential. Rom J Morphol Embryol 2014;55:49-56. Ibrahim RA, Amin MM, Mohammed MF. Value of CD56 immunohistochemistry in distinguishing papillary thyroid carcinoma from other lesions of the thyroid gland. Benha Med J 2011;28:87-91. Etem H, Zekinci S, Mizrak B, Şentürk S. The role of CD56, HBME-1, and p63 in follicular neoplasms of the thyroid. Turk J Pathol 2010;26:238-42. Filie AC, Asa SL, Geisinger KR, Logani S, Merino M, Nikiforov YE, et al Utilization of ancillary studies in thyroid fine needle aspirates: A synopsis of the National Cancer Institute Thyroid Fine Needle Aspiration State of the Science Conference. Diagn Cytopathol 2008;36:438-41. Kontogianni K, Nicholson AG, Butcher D, Sheppard MN. CD56: A


Abdelzaher, et al.: A model predicting malignancy in thyroid lesions useful tool for the diagnosis of small cell lung carcinomas on biopsies with extensive crush artefact. J Clin Pathol 2005;58:978-80. 37. Lanza F, Bi S, Castoldi G, Goldman JM. Abnormal expression of N-CAM (CD56) adhesion molecule on myeloid and progenitor cells from chronic myeloid leukemia. Leukemia 1993;7:1570-5. 38. Chang H, Samiee S, Yi QL. Prognostic relevance of CD56 expression in multiple myeloma: A study including 107 cases treated with highdose melphalan-based chemotherapy and autologous stem cell transplant. Leuk Lymphoma 2006;47:43-7. 39. Cho EY, Choi Y, Chae SW, Sohn JH, Ahn GH. Immunohistochemical study of the expression of adhesion molecules in ovarian serous neoplasms. Pathol Int 2006;56:62-70. 40. Agaimy A, Wünsch PH. Distribution of neural cell adhesion molecule (NCAM/CD56) in gastrointestinal stromal tumours and their intra-abdominal mesenchymal mimics. J Clin Pathol 2008;61:499-503.


41.

Lantuéjoul S, Laverrière MH, Sturm N, Moro D, Frey G, Brambilla C, et al. NCAM (neural cell adhesion molecules) expression in malignant mesotheliomas. Hum Pathol 2000;31:415-21. 42. Fernández-Briera A, García-Parceiro I, Cuevas E, Gil-Martín E. Effect of human colorectal carcinogenesis on the neural cell adhesion molecule expression and polysialylation. Oncology 2010;78:196204. © GESDAV; licensee GESDAV. This is an open access article licensed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/) which permits unrestricted, non-commercial use, distribution and reproduction in any medium, provided the work is properly cited. Source of Support: Nil, Conflict of Interest: None declared.

212