Indian Journal of Pathology and Microbiology

ORIGINAL ARTICLE
Year
: 2019  |  Volume : 62  |  Issue : 1  |  Page : 17--23

Immunohistochemical and morphometrical evaluation of well-differentiated thyroid tumor of uncertain malignant potential


Dareen A Mohamed, Maha M Shamlola 
 Department of Pathology, Tanta University, Egypt

Correspondence Address:
Dareen A Mohamed
Department of Pathology, Tanta Univeristy
Egypt

Abstract

Context: Well-differentiated tumor of uncertain malignant potential (WDT-UMP) were encapsulated follicular pattern tumors with diffuse equivocal or focal unequivocal papillary thyroid carcinoma (PTC) type nuclear changes without definite invasion. These tumors are suggested to be borderline in its nature. Immunohistochemistry by CD56, P63, and CK19 are common markers used in differentiation between benign and malignant thyroid tumor. Computerized nuclear morphometry is an inexpensive objective and reproducible tool to evaluate histological features of thyroid tumors. Aims: To categorize WDT-UMP using combined nuclear morphometry and immunohistochemistry of CD56, P63, and CK19. Subjects and Methods: We used CD56, P63, and CK19 immunostaining and mean nuclear area and perimeter morphometry on 20 cases of WDT-UMP, 25 cases malignant thyroid tumors, and 25 cases benign thyroid lesions. Statistical Analysis Used: Student's t-test (Unpaired), analysis of variance (ANOVA) test, and Chi-square tests by SPSS V. 20. Results: Significant differences were obtained between WDT-UMP and benign group according to the three markers, but no significant difference between WDT-UMP and malignant group. The mean nuclear area and mean nuclear perimeter were significantly higher in WDT-UMP group in comparison with the benign group while there were no significant differences with the malignant group. Conclusion: WDT-UMP are intermediate lesions differing from benign lesions of the thyroid and sharing some criteria with PTC to a certain extent.



How to cite this article:
Mohamed DA, Shamlola MM. Immunohistochemical and morphometrical evaluation of well-differentiated thyroid tumor of uncertain malignant potential.Indian J Pathol Microbiol 2019;62:17-23


How to cite this URL:
Mohamed DA, Shamlola MM. Immunohistochemical and morphometrical evaluation of well-differentiated thyroid tumor of uncertain malignant potential. Indian J Pathol Microbiol [serial online] 2019 [cited 2019 Aug 19 ];62:17-23
Available from: http://www.ijpmonline.org/text.asp?2019/62/1/17/251238


Full Text



 Introduction



The diagnostic criteria in the well-differentiated tumor of uncertain malignant potential (WDT-UMP) were for encapsulated follicular pattern tumors with diffuse equivocal or focal unequivocal papillary thyroid carcinoma type nuclear changes (PTC-Ns) without definite invasion. This group of tumors is suggested to be borderline in its nature.[1]

Studies stated that regardless of whether or not the tumor has an invasive capsule or a papillary growth pattern, PTC-type nuclear features are the most accepted morphological features in the diagnosis of PTC, which include nuclear enlargement, nuclear overlapping, nuclear clearing, nuclear grooving, and cytoplasmic pseudoinclusions.[2]

In follicular-patterned tumor, when PTC nuclear features were ambiguous, such as only nuclear clearing and nuclear grooving without nuclear pseudoinclusions, thyroid lesions will be classified as WDT-UMP. When PTC nuclear features were seen in the entire part of the tumor the diagnosis will be encapsulated follicular variant papillary thyroid carcinoma (FVPTC). When unequivocal PTC nuclear features were found in only part of the tumor, it will be classified as WDT-UMP.[3]

CD56 is a neural cell adhesion molecule and its migration affects the migratory behavior of tumor cells. CD56 can be found in the follicular epithelial cells of the normal thyroid. Some studies have stated that CD56 is not expressed in PTC, while expression is present in most follicular tumors.[4]

P63 is a member of the tumor suppressor gene family and is essential for normal development. P63 acts as a key role in the proliferation and differentiation of epithelial cells and causes defects in epithelial differentiation. Some studies have stated that P63 is expressed in the basal layer of squamous epithelium, in breast and prostate myoepithelial cells, and in the transitional epithelium of the bladder. Some studies stated that P63 is expressed in PTC with no staining in other thyroid tumors.[5]

Cytokeratin 19 is a type I intermediate filament protein and is present in simple epithelial cells. Studies stated that PTC expresses strong and diffuse cytoplasmic reactivity to CK19. However, CK19 is focally stained reactive thyroid follicular epithelium; while in follicular adenomas or carcinomas, most authors have not identified diffuse CK19 positivity.[6],[7]

Computerized nuclear morphometry is an inexpensive objective and reproducible tool to evaluate histological features. It has been stated that nuclear morphometric parameters, such as nuclear perimeters and nuclear areas, help to differentiate between thyroid lesions. Some studies have revealed significant differences between benign and malignant lesions using thyroid aspiration cytology in nuclear areas and perimeters.[8]

So, under the insight of these researches, we are trying in our study to categorize WDT-UMP using combined nuclear morphometry and immunohistochemistry of CD56, P63, and CK19 to predict its clinical behavior.

 Subjects and Methods



Histopathology

Paraffin blocks from 70 cases were selected from the archived files of the pathology department during the period from October 2013 to October 2016. Hematoxylin- and eosin-stained sections were reviewed and classified into three groups. Twenty cases were of WDT-UMP (Group 1), 18 cases were of PTC, and 7 cases were of FVPTC (Group II: malignant cases control group), 10 cases were of follicular adenoma, 10 cases were of Hashimoto's thyroiditis, and 5 cases were of hyperplastic thyroid nodules (Group III: benign cases control group).

Immunohistochemistry

Representative tissue sections were deparaffinized in xylene, rehydrated in descending alcohol grades, and then incubated with the primary antibodies of CK19, CD56, and P63 in dilutions mentioned in [Table 1]. This is done after the antigen retrieval by microwave incubation in 6.1 pH citrate buffer for 20 min and blocking endogenous peroxidase by H2O2. Visualization was obtained by streptavidin-biotin ABC detection kit (Catalog # TA-015-HP, Lab-Vision Corporation Fremont, USA). Color development was done using 3,3'-diaminobenzidine and Mayer hematoxylin as a counterstained. Slides were mounted with DPX and cover slipped. The positive control was PTC while negative control was done by omitting the step of the primary antibody.{Table 1}

Evaluation of immunohistochemistry

By using ×40 objective lenses, >10% positive follicular cells showed membranous and cytoplasmic staining was considered positive for CD56 and CK19, as well as >10% of the nuclear staining of the follicular cells were considered positive for P63.[4],[9]

Morphometric analysis

All slides were submitted to digital image analysis using the Image Leica Q win lab software (USA), and nuclear areas and perimeters were measured. Well-preserved, non-overlapped cells were selected. From the selected areas, an average of 5–10 microscopic fields, at magnification ×400 were captured for each case. At least 100 nuclei were analyzed per slide. The intact whole nuclei from the actual lesion were measured, avoiding stromal cells nuclei. Overlapped and fragmented nuclei were discarded. The mean values were separately compared between the three groups.

Statistical analysis

Statistical presentation and analysis of the present study were conducted, using the mean, standard deviation, Student's t-test (unpaired), analysis of variance (ANOVA) test, and Chi-square tests by SPSS V. 20. Significant level was reported as non-significant if the P value was >0.05, and significant if the P value was <0.05.

 Results



Histopathological and Immunohistochemical results [Table 2]{Table 2}

Group I results [Figure 1]{Figure 1}

The twenty cases included in this group of WDT-UMP showed intact capsule, overlapping, 12 cases showed focal nuclear features of PTC such as nuclear grooving, nuclear ground glass, infrequent pseudoinclusions, and nuclear crowding with the absence of papillae, or abortive papillae. Eight cases showed diffuse but incomplete nuclear changes such as diffuse nuclear clearing, nuclear grooving with the absence of nuclear pseudoinclusion, nuclear overlapping, and papillae formation.

CD56 immunostaining showed positive cytoplasmic and membranous staining in 2/20 cases (10%), P63 immunostaining showed positive nuclear staining in 14/20 cases (70%), as well as CK19 expression showed positive membranous and cytoplasmic staining in 10/20 cases (50%).

Group II results [Figure 2] and [Figure 3]{Figure 2}{Figure 3}

Twenty-five cases were included in this group with 18 cases were diagnosed as PTC, showed papillae, abortive papillae with nuclear grooving, nuclear ground glass, pseudoinclusions, and nuclear overlapping. Seven cases were diagnosed as FVPTC, showed a follicular pattern of tumor cells with nuclear features of PTC including nuclear clearing, nuclear ground glass, nuclear pseudoinclusions, and nuclear grooves.

CD56 immunostaining showed positive cytoplasmic and membranous staining in 3/25 cases (12%) [1/18 (5.6%) of PTC cases and 2/7 (28.6%) of FVPTC cases]. P63 immunostaining showed positive nuclear expression in 20/25 (80%) [15/18 (83.3%) of PTC cases and 5/7 (71.4%) of FVPTC cases], as well as CK19 expression showed positive cytoplasmic and membranous staining in 19/25 (76%) [13/19 (72.2%) of PTC cases and 6/7 (85.7%) of FVPTC cases].

Group III results [Figure 4], [Figure 5], [Figure 6]{Figure 4}{Figure 5}{Figure 6}

Twenty-five cases were included in this group with 10 cases were diagnosed as follicular adenoma, 10 cases were diagnosed as Hashimoto's thyroiditis, and 5 cases were diagnosed as a hyperplastic nodule. CD56 immunostaining showed positive expression in 24/25 cases (96%) [10/10 (100%) cases of follicular adenoma, 9/10 (90%) cases of Hashimoto's thyroiditis, and 5/5 (100%) cases of hyperplastic nodule]. P63 immunostaining showed positive expression in 2/25 (8%) [1/10 (10%) cases of follicular adenoma, 0/10 (0%) cases of Hashimoto's thyroiditis, and 1/5 (5%) cases of hyperplastic nodule]. CK19 immunostaining showed positive expression in 4/25 (16%) [3/10 (30%) cases of follicular adenoma, 0/10 (0%) cases of Hashimoto's thyroiditis, and 1/5 (5%) cases of hyperplastic nodule].

Univariate analysis revealed that statistically significant differences were obtained between the WDT-UMP group and the benign group according to the CD56, P63, and CK19 as P values were <0.001, <0.001, and <0.014, respectively, however, no significant difference was observed between the WDT-UMP group and the malignant group as P values were 0.832, 0.438, and 0.070, respectively.

Morphometric results [Table 3]{Table 3}

Mean nuclear areas and mean nuclear perimeter in WDT-UMP cases were 48.17 ± 12.15 and 28.45 ± 5.61, respectively.

In PTC cases they were 51.67 ± 14.5 and 28.56 ± 4.72, respectively, and in FVPTC they were 50.89 ± 10.67 and 29.61 ± 3.98, respectively. There were no significant differences according to mean nuclear area and mean nuclear perimeter between PC and FVPC cases, as P values were 0.898 and 0.670, respectively.

While in follicular adenoma mean nuclear area and mean nuclear perimeter were 36.54 ± 9.13 and 19.25 ± 2.62, respectively, in Hashimoto's thyroiditis they were 29.78 ± 6.17 and 20.12 ± 2.01, respectively, and in hyperplastic nodule they were 21.56 ± 4.29 and 17.60 ± 2.05, respectively. There was no significant difference according to mean nuclear perimeter between Hashimoto's thyroiditis, follicular adenoma, and hyperplastic nodule cases, as P value was 0.059. There was significant difference according to mean nuclear area between follicular adenoma cases at one hand Hashimoto's thyroiditis and hyperplastic nodule cases on the other hand as P value was 0.004.

The mean nuclear area and mean nuclear perimeter were significantly higher in WDT-UMP group in comparison with the benign group as P values were <0.001 for each of them, while there were no significant differences with the malignant group as P values were 0.407 and 0.788, respectively.

 Discussion



The follicular-patterned lesions which cannot be diagnosed with certainty as benign or malignant were termed as WDT-UMP. This is because of equivocal nuclear features as well as questionable capsular or vascular invasion.[1]

The suspected benefit from this proposed diagnosis is avoidance of aggressive treatment. CD56 is a differentiation antigen for thyroid follicular cells.[10] It is already expressed in normal and benign thyroid tissue but nearly absent in PTC.[11]

In this study, it was found that CD56 was expressed in 96% in the benign group, 12% in the malignant group, and 10% in the WDT-UMP group. This was in accordance with Scarpino et al., who stated that CD56 showed low or absent expression in PTC by either immunohistochemistry or by polymerase chain reaction.[12] This was also comparable to Demellawy et al. who reporting distinctive benefits of CD56 in the differentiation between PTC with its follicular variant and other follicular lesions/neoplasms including Hashimoto's thyroiditis and follicular adenoma.[4]

In this study, there was a significant difference between the WDT-UMP group and benign group in CD56 expression, on the contrary, the malignant group showed no significant difference. The questionable morphological features CK19 proved to be a valuable confirmatory marker differentiating between benign and malignant thyroid lesions in both intensity and distribution.[13],[14]

In this study, it was found that CK19 was expressed in 16% in the benign group, 76% in the malignant group, and 50% in the WDT-UMP group. Others showed intense diffuse CK19 positivity in 80–100% of PTC cases,[15],[16] while in the study of Bukhari and colleagues including a larger number of cases (60 cases of a FVPTC) 95% of cells showed CK19 positivity.[17] The negative expression of CK19 was considered an exclusive diagnosis for PTC.[18] Bukhari and colleagues and Barut et al. demonstrated a strong and diffuse immunoreactivity for CK19 in WDT-UMP and they reported a considerable positivity for CK19 in their 35 cases ranging between 5–95%.[17],[19] On the contrary, the study of Scognamiglio and Hoffmann et al. WDT-UMP showed 63–64% positivity rate for CK19.[20],[21]

The study of Yassin, CK19 was absolutely 100% positive in WDT-UMP cases.[22] Scognamiglio et al. confirmed an intermediate expression for CK19 for WDT-UMP between PTC and showing 5/11 positive cases.[20]

P63 a member of the P53 gene family is uniformly expressed in basal cells, myoepithelial cells, and squamous cells. It is located on a 3q207-29 chromosome and it is expressed in six isoforms. Three of the six isoforms encode protein transactivating P53 inducing cell apoptosis; the other three isoforms encode inhibitory proteins on P53 activity. It is expressed in basal, squamous, and myoepithelial cells.[4],[23],[24]

In this study, it was found that P63 was expressed in 8% in the benign group, 80% in the malignant group, and 70% in the WDT-UMP group. Unger et al. and Demellawy et al. described that P63 was detected in 81.8% of PTC. Accordingly, they postulated that P63 can help in the differentiation of PTC from other thyroid tumors.[4],[25] However, Kim et al. reported that P63 was detected in 12.5% of PTC, 11.1% of poorly differentiated carcinomas, and 71.4% of anaplastic carcinomas, while normal thyroid follicles, hyperplastic thyroid follicles, follicular carcinomas, and medullary carcinomas showed negative expressions for P63. Therefore, Kim et al. concluded that P63 showed a late expression in the course of thyroid tumor progression.[26] One postulated explanation could be the higher positive cut-off value for P63 in our study (at least 10% of tumor cells staining positively) as compared with Demellawy's et al. study, in which any nuclear P63 staining was considered a positive result.[4] However, this method of assessment is difficult in daily practice applications, as positivity in only one nucleus can sometimes be very challenging and difficult to evaluate.

The subjective evaluation of cytological atypia is not a reliable criterion for the diagnosis of malignancy in thyroid lesions since these changes may be encountered in benign lesions such as adenomatous hyperplasia and follicular adenoma. Therefore it is important to differentiate between benign and malignant neoplasms and the non-neoplastic group.[27] To raise the level of precise accuracy of the relatively subjective histopathological diagnosis, objective methods of assessment of nuclear morphology may be beneficial.[28–30]

Aiad et al. stated that it is a matter of importance to use nuclear morphometry in the diagnosis of follicular-patterned lesions. Higher nuclear morphometric values were recorded in a neoplastic group than non-neoplastic lesions with a mean nuclear surface area ≥39.9 μm2, and the nuclear perimeter ≥27.7 μm outweighs the diagnosis of PTC rather than follicular adenoma.[31]

In this study, it was found that the mean nuclear surface area and the mean nuclear perimeter in the WDT-UMP group and the malignant group cases were >39.9 μm2 and 27.7 μm2, respectively, while in the benign group cases they were less than these cut-off values with significant differences between them. This shares comparable results with Liu and his colleagues reporting that the nuclear size of WDT-UMP was 2–4 times that of normal thyroid follicular cells.

So in this study, we found that WDT-UMP shares close features by immunohistochemistry and nuclear morphometry to malignant group than benign group that was in accordance with Coli et al. who reported that tumors showing incomplete nuclear features are suggestive but not diagnostic of PTC and may have started phenotypic changes towards a PTC.[32]

WDT-UMP constitute a true “grey zone” of lesions representing a borderline category necessitating specific curative treatment and postoperative follow-up. Kakudo et al. concluded that WDT-UMP might be biologically borderline lesions representing an early carcinogenesis phase and can be true precursor lesions of PTC.[33]

 Conclusion



We concluded that WDT-UMP are intermediate lesions differing from adenoma and benign lesions of the thyroid sharing some criteria with PTC to a certain extent.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

References

1Williams ED. Guest editorial: Two proposals regarding the terminology of thyroid tumors. Int J Surg Pathol 2000;8:181-3.
2DeLellis RA, LIoyd RV, Heitz PU. World Health Organization classification of tumours. Pathology and Genetics of Tumours of Endocrine Organs. Lyon: IARC Press; 2004.
3Liu Z, Zhou G, Nakamura M, Koike E, Li Y, Ozaki T, et al. Encapsulated follicular thyroid tumor with equivocal nuclear changes, so-called well-differentiated tumor of uncertain malignant potential: A morphological, immunohistochemical, and molecular appraisal. Cancer Sci 2011;102:288-94.
4El Demellawy D, Nasr A, Alowami S. Application of CD56, P63 and CK19 immunohistochemistry in the diagnosis of papillary carcinoma of thyroid. Diagn Pathol 2008,3:5.
5Reis-Filho JS, Schmitt FC. Taking advantage of basic research: p63 is a reliable myoepithelial and stem cell marker. Adv Anat Pathol 2002;9:280-9.
6Cheung C, Ezzat S, Freeman JL, Rosen IB, Asa SL. Immunohistochemical diagnosis of papillary thyroid carcinoma. Mod Pathol 2001;14:338-42.
7Haiyan Liu, Fan Lin. Application of immunohistochemistry in thyroid pathology. Arch Pathol Lab Med 2015;139:67-82.
8Kang Y, Lee YJ, Jung J, Lee Y, Won NH, Chae YS. Morphometric analysis of thyroid follicular cells with atypia of undetermined significance. J Pathol Transl Med 2016;50:287-93.
9Etem H, Özekinci S, Mizrak B, Şentürk S. The role of CD56, HBME-1, and p63 in follicular neoplasms of the thyroid. Turkish Journal of Pathology 2010;26:238-42.
10Norris S, Doherty DG, Curry M, McEntee G, Traynor O, Hegarty JE, et al. Selective reduction of natural killer cells and T cells expressing inhibitory receptors for MHC class I in the livers of patients with hepatic malignancy. Cancer Immunol Immunother 2003;52:53-8.
11Kim JH, Kim YH, Han JH, Lee KB, Sheen SS, Lee J, et al. Silencing of homeobox B9 is associated with down-regulation of CD56 and extrathyroidal extension of tumor in papillary thyroid carcinoma. Hum Pathol 2012;43:1221-8.
12Scarpino 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 down regulation of VEGF-D production by tumor cells. J Pathol 2007;212:411-9.
13Casey MB, Lohse CM, Lloyd RV. Distinction between papillary thyroid hyperplasia and papillary thyroid carcinoma by immunohistochemical staining for cytokeratin 19, galectin-3 and HBME-1. Endocr Pathol 2003;14:55-60.
14Cheung CC, Ezzat S, Freeman JL, Rosen IB, Asa SL. Immunohistochemical diagnosis of papillary thyroid carcinoma. Mod Pathol 2001;14:338-42.
15Barroeta JE, Baloch ZW, Lal P, Pasha TL, Zhang PJ, et al. Diagnostic value of differential expression of CK19, Galectin-3, HBME-1, ERK, RET, and p16 in benign and malignant follicular-derived lesions of the thyroid: An immunohistochemical tissue microarray analysis. Endocr Pathol 2006;17:225-34.
16Sahoo S, Hoda SA, Rosai J, DeLellis RA. Cytokeratin 19 immunoreactivity in the diagnosis of papillary thyroid carcinoma: A note of caution. Am J Clin Pathol 2001;116:696-702.
17Bukhari U, Sadiq S, Kehar SI. Differential expression of CK 19 in follicular adenoma, well-differentiated tumor of uncertain malignant potential (WDT-UMP) and follicular variant of papillary carcinoma. J Pak Med Assoc 2009;59:15-8.
18Nasr MR, Mukhopadhyay S, Zhang S, Katzenstein AL. Immunohistochemical markers in diagnosis of papillary thyroid carcinoma: Utility of HBME1 combined with CK19 immunostaining. Mod Pathol 2006;19:1631-7.
19Barut F, Bektas S, Bahadir B, Kandemir NO, Karadayi N, Ozdamar SO. The value of cytokeratin-19 immunohistochemistry in the differential diagnosis of papillary thyroid carcinomas. Turkiye Klinikleri J Med Sci 2009;29:42-7.
20Scognamiglio T, Hyjek E, Kao J, Chen YT. Diagnostic usefulness of HBME1, galectin-3, CK19, and CITED1 and evaluation of their expression in encapsulated lesions with questionable features of papillary thyroid carcinoma. Am J Clin Pathol 2006;126:700-8.
21Hoffmann V, Lassalle S, Bonnetaud C, Butori C, Loubatier C, Ilie M, et al. Thyroid tumors of uncertain malignant potential: Frequency and diagnostic reproducibility. Virchows Arch 2009;455:21-33.
22Yassin Fel-Z. Diagnostic criteria of well differentiated thyroid tumor of uncertain malignant potential; a histomorphological and immunohistochemical appraisal. J Egypt Natl Cancer Inst 2015;27:59-67.
23Preto A, Reis-Filho JS, Ricardo S, Soares P. P63 expression in papillary and anaplastic carcinomas of the thyroid gland: Lack of an oncogenetic role in tumorigenesis and progression. Pathol Res Pract 2002;198:449-54.
24Levrero M, De Laurenzi V, Costanzo A, Gong J, Wang JY, Melino G. The p53/p63/p73 family of transcription factors: Overlapping and distinct functions. J Cell Sci 2000;113:1661-70.
25Unger P, Ewart M, Wang BY, Gan L, Kohtz DS, Burstein DE. Expression of p63 in papillary thyroid carcinoma and in Hashimoto's thyroiditis a pathobiologic link? Hum Pathol 2003;34:764-69.
26Kim YW, Do IG, Park YK. Expression of the GLUT1 glucose transporter, p63, and p53 in thyroid carcinomas. Pathol Res Pract 2006;202:759-65.
27Baloch ZW, LiVolsi VA. Follicular-patterned lesions of the thyroid: The bane of the pathologist. Am J Clin Pathol 2002;117:143-50.
28Wang SL, Wu MT, Yang SF, Chan HM, Chai CY. Computerized nuclear morphometry in thyroid follicular neoplasms. Pathol Int 2005;55:703-6.
29Murata S, Mochizuki K, Nakazawa T, Kondo T, Nakamura N, Yamashita H, et al. Morphological abstraction of thyroid tumor cell nuclei using morphometry with factor analysis. Microsc Res Tech 2003;61:457-62.
30Minervini A, Lilas L, Minervini R, Selli C. Prognostic value of nuclear grading in patients with intracapsular (pT1-pT2) renal cell carcinoma. Long-term analysis in 213 patients. Cancer 2002;94:2590-5.
31Aiad 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.
32Coli A, Bigotti G, Parente P, Federico F, Castri F, Massi G. Atypical thyroid nodules express both HBME-1 and Galectin-3, two phenotypic markers of papillary thyroid carcinoma. J Exp Clin Cancer Res 2007;26:221-7.
33Kakudo K, Bai Y, Liu Z, Li Y, Ito Y, Ozaki T. Classification of thyroid follicular cell tumors: With special reference to borderline lesions. Endocr J 2012;59:1-12.