Expression of BCL-2 and p53 in Oncocytic (H0rthle Cell) Tumors of the Thyroid: An Immunohistochemical Study Anupama Kohli, MD Stephen P. Baker, MSC, MPH, Nilima A. Patwardhan, MD, and Ashraf Khan, MD, MRCPATH Abstract Previous studies have established that thyroid follicular neoplasms of higher malignant potential show a high p53 and low bcl-2 expression. This however has not been well studied in Oncocytic (H0rthle cell) neoplasms, the management of which remains controversial. We therefore studied the expression of p53 and bcl-2 in 18 H0rthle cell adenomas (HCA) and 8 H0rthle cell carcinomas (HCC) and compared them with their benign and malignant counterparts, respectively, including 16 follicular adenomas (FA) and 68 papillary carcinomas (PC). All 16 FA were bcl-2 positive, 4 were 2+, and 12 were 3+. On the other hand, 14/18 (78%) HCA showed bcl-2 expression, 5 were 1+, 6 were 2+, and only 3 were 3+. Similarly, HCC showed a weaker bcl-2 staining pattern compared to PC. Only 1 FA showed grade 1, p53 staining, the remaining 15 were negative, and 15/18 HCA showed p53 expression of varying grades. This difference in p53 staining was statistically significant (p = 0.005). A significant p53 overexpression was also seen in HCC compared to PC (p = 0.005). In conclusion, there appears to be an inverse relationship between p53 and bcl-2 expression in thyroid follicular neoplasms. A higher expression of p53 and lower levels bcl-2 in H~rthle cell neoplasms may have biological and clinical implications. This may support a more aggressive surgical treatment for HCA compared to FA. Key Words: H0rthle cell tumors; p53; bcl-2; immunohistochemistry.
Introduction Departments of Pathology (AK, AK), Academic Computing (SPB), and Surgery (NAP), University of Massachusetts Medical Center, Worcester, MA. SPB is now Director of Research and Evaluation, Division of Medical Assistance, Boston, MA. Address correspondence to Ashraf Khan, MD, MRCPath., Department of Pathology, University of Massachusetts Medical Center, 55 Lake Avenue North, Worcester, MA 01655.
Endocrine Pathology,vol. 9, no. 2, 117-123, Summer 1998 9 Copyright 1998 by Humana Press Inc. All rights of any nature whatsoever reserved. 1046-3976/98/9:117-123/$9.75
Oncocytic (Hiirthle cell) neoplasms of the thyroid gland have demonstrated an extremely variable clinical course following surgical intervention. Studies and case reports [1--4] illustrating a higher rate of recurrence and an aggressive clinical course in even the benign Hiirthle cell lesions has further fueled this controversy. A standardized treatment regime has not yet been established owing to the unpredictable biological behavior in these histologically benign lesions. Attempts at predicting their behavior by analyzing the nuclear DNA
content using cytomorphometric methods in cytological  as well as paraffinembedded histological material  have also been made. In both cases, it has a limited value in predicting the course and outcome of these lesions. Mutations or allelic loss of the tumor suppressor gene, p53, has been reported in many tumors and cell lines, including thyroid follicular neoplasms. Dobashi et al.  proposed that p53 alterations in thyroid neoplasms do not participate in the oncogenesis itself, but may significantly contribute to the biological behavior of these tumors. Others [8,9] have claimed these
Volume 9, Number 2
genetic alterations to be associated with dedifferentiation and tumor progression. A significantly higher expression of mutant p53 protein was observed in poorly differentiated and undifferentiated tumors of the thyroid gland by Donghi et al. . Recently, Hosal et al.  concluded that p53 immunolocalization could be used as a prognostic index of clinical behavior in tumors derived from the thyroid follicular cells. Bcl-2 gene, a critical regulator of apoptosis, has a differentiation-related expression . Its expression has been associated with less aggressive tumor behavior and better prognosis in thyroid follicular neoplasms. Pilotti et al.  proposed the use of bcl-2 expression to differentiate poorly differentiated follicular carcinomas from undifferentiated carcinomas of the thyroid gland. In light of these findings, the inverse expression of p53 and bcl-2 has been proposed as a prognostic marker of tumor progression in thyroid tumors by some authors [ 13,14]. However, the expression of p53 and bcl-2 has not been clearly defined in Hiirthle cell neoplasms. In view of the unpredictable biology of Hiirthle cell tumors, we evaluated the immunohistochemical expression of these gene products in Hi~rthle cell tumors of the thyroid gland and compared them with their benign and malignant counterparts, including follicular adenoma and papillary carcinoma.
Materials and Methods Cases and Histological Classification
This retrospective study included 1l0 cases of thyroid follicular tumors, retrieved from the Surgical Pathology files of University of Massachusetts Medical Center between 1989 and 1996. All the sections in each case were reviewed to select one representative block showing the most typical areas
of the tumor for immunohistochemistry. The malignant tumors were classified according to W H O Classification of thyroid tumors, and included 68 papillary carcinomas (PC) and 8 cases of H~irthle cell carcinomas (HCC). The benign lesions included 16 follicular adenomas (FA) and 18 Hiirthle cell adenomas (HCA). Only the follicular neoplasms cytologically comprising more than 75% oncocytic cells were regarded as H/irthle cell adenomas . In addition, 105 cases of both benign and malignant tumors also contained normal thyroid tissue, which was evaluated as normal control. Immunohistochemistry
One representative block selected in each case was evaluated for bcl-2 and p53 expression immunohistochemically. Sections were cut at 4 !li, heated at 60~ for 30 min, then deparaffinized and hydrated through a series of xylenes and alcohols. Optimum pretreatment and dilutions were determined by testing both positive and negative material with an antibody against bcl-2 and a cocktail of two p53 antibodies using an antigen-retrieval method for the best signal-to-noise ratio. The bcl-2 antibody (Dako, Glostrup, Denmark) was used at a dilution of 1:50. A cocktail mixture of two antibodies AB-2 and AB-6 were used for p53 immunostaining (Oncogene Research Products, Cambridge, MA) at a dilution of 1:40 and 1:250, respectively. The slides were microwaved with a proprietary antigen retrieval solution (citrate buffer, Ventana, Tuscon, AR) for 5 min in a microwave oven at 800 W. Following replenishment of this solution, the slides were microwaved again for an additional 5 min and then allowed to cool for 20 min. The staining was performed on a Ventana Solutions TekMate 1000 automated
Fig, 1. Grade distribution of p53 and bcl2 immunoreactive cells in thyroid tumors.
immunostainer using an avidin/biotin complex (ABC) procedure (Ventana). The endogenous peroxidase activity was blocked by hydrogen peroxide followed by a serum blocking step. The slides were then incubated with the primary antibody for 45 rain followed by brief buffer washes. They were then incubated in a cocktail of biotinylated antimouse IgG/IgM and rabbit IgG (Ventana) for 30 min. The sections were washed, incubated in ABC (Ventana) for 30 rain, washed, and then reacted with diaminobenzidine and hydrogen peroxide to visualize the end product. Counterstaining was performed with hematoxylin. A duplicate set of slides was also stained in the exact same manner substituting normal mouse serum for the bcl-2 and p53 antibody to serve as a negative control. Evaluation of Immunostaining
A semiquantitative evaluation was performed for p53 staining. Reactions were graded for percent of tumor cells showing nuclear positivity. Less than of 10% cells with positive staining were classified as grade 1; 11-25% positivity was grade 2; grade 3 included cases with 26-50% positive cells, and more than 50% positivity was grade 4.
Cytoplasmic expression of bcl-2 was examined for intensity of staining qualitatively as negative, weakly positive (1+), intermediate (2+), and strongly positive (3+). Statistical Analysis
Differences in p53 and bcl-2 immunoreactivity among groups (FA, HCA, PC, and HCC) was evaluated using KruskalWallis and Mann-Whitney test. Pairwise comparison between individual groups was performed using Fischer Exact test with Bonferroni adjustment, the latter to compensate for additive type I error owing to multiple comparisons.
Results Figure 1 summarizes the results of p53 and bcl-2 immunostaining in the various groups of thyroid tumors studied. Bcl-2 Expression
Positive bcl-2 immunostaining was observed in all 16 cases of follicular adenoma. The intensity of staining ranged from intermediate (2+) to strongly positive (3+) in these cases. Of the 18 cases of HCA, only 14 (78%) were bcl-2-positive,
Volume 9, Number 2
5 (28%) were 1+, and 6 (33%) were grade 2+ positive. Only 3 (16%) cases were strongly positive. Among the 68 PC, 28 (41%), 24 (35%), and 16 (24%) cases showed 1+, 2+, and 3+ reaction patterns, respectively. HCC (n = 8), on the other hand, had only one case each with intermediate (2+) and strong (3 +) positivity. The remaining 6 (75%) cases were weakly (1+) positive for bcl-2. The normal thyroid tissue sections were all strongly (3+) bcl-2-positive. Figure 2 shows the bcl-2 staining pattern in the lesions evaluated.
p53 Expression Only one of the 16 cases (6%) of FA showed grade 1 (< 10% cells) positivity, whereas the remaining 15 were nonreactive to p53. In contrast, HCA (n = 18) had 15 of the 18 (83.8%) cases showing positive staining of varying grades. Only three cases were p53-negative. Twenty-three of the 68 cases of PC were positive for p53, 18 (26%) were grade 1, and 5 (7%) cases were grade 2. On the other hand, HCC (n = 8) showed a relatively higher grade of expression for p53. One, 2, and 5 cases were grade 2, 3, and 4, respectively. Normal thyroid tissue in all the sections was nonreactive to p53 antibody. Figure 3 shows p53 staining pattern in the various groups of thyroid lesions studied. By pairwise comparison (Table 1), bcl-2 expression was found to be significantly lower in Hiirthle cell adenoma compared to Follicular adenoma (? = 0.015) Also p53 expression was significantly higher in HCA as compared to FA (p = 0.005), in HCC compared to PC (p = 0.005), and HCA compared to PC (p = 0.005).
Discussion For a long time, Endocrine surgeons and pathologists have debated on the biology
of oncocytic (H/.irthle cell) neoplasms. Conflicting results from different studies have further fueled this controversy [1-4]. Thompson et al.  proposed an aggressive surgical treatment for these lesions on the basis of increased incidence of malignant transformation and occurrence of contralateral tumor in cases previously diagnosed as benign. Gundry et al.  proposed a 10-yr follow-up and a total thyroidectomy in these cases. Though numerous reports to this effect have been published, the low incidence of these lesions remains to b~ a confounding factor in all of them. Moreover, the defining criteria for Htirthle cell lesions has been variable in these studies. Current views on pathologic diagnosis of these lesions rely on the same criteria for malignancy as with other follicular neoplasm, namely, capsular and vascular invasion, lymph node, and other distant metastasis [15,16]. Gossain and Clark  addressed the issue of reliability ofhistopathological diagnosis using this criteria in these lesions. Their results suggested a clinicopathological correlation combined with radiological findings could predict the tumor behavior at the time of surgery. Nevertheless, attempts at defining and predicting the course of these lesions have continued. Correlating the nuclear DNA content and the behavior of these tumors by cytomorphological methods have also been inconclusive [5,6]. A number of studies have used the increased p53 and decreased bcl-2 expression in thyroid follicular neoplasms as an indicator of poor prognosis [13,14,17-19]. In this study, we evaluated the biological behavior of Hiirthle cell neoplasms based on p53 and bcl-2 expression in these lesions. We compared this expression in benign as well as malignant H/./rthle cell neoplasms and compared them to follicular adenomas and papillary carcinomas, respectively. To the
p53 and bcl-2 in H0rthle Cell Neoplasms
Fig. 2. bcl-2 expression in (A) FA (grade 3+), (B) PC (grade 2+), (C) HCA (grade 2+), and (D) HCC(grade 1+), irnmunoperoxidase, x200.
Fig. 3. p53 expression in (A) FA (negative), (B) PC (grade I), (C) HCA (grade 3), and (D) HCC (grade 4), immunoperoxidase, x200.
Volume 9, Number 2
Table 1. Pairwise Correlation of 8CL-2 and p53 Immunoreactivity in Different Subgroups of Thyroid
Follicular Lesions Bcl-2 "p" value Diagnosis" HCA vs FA HCA vs PC HCA vs HCC HCC vs PC
best of our knowledge, no such evaluation has been performed that has been directed specifically to the Hiirthle cell neoplasms. Our results indicate a higher expression of p53 and a decreased bcl-2 immunoreactivity in Hiirthle cell adenomas and carcinomas compared to follicular adenomas and papillary carcinomas, respectively. These findings, if taken in isolation, may support the notion that a much more aggressive form of treatment is required in Hiirthle cell neoplasms of the thyroid gland. Oversimplification of the data indicates p53 expression may become a source of error in this case. As previously documented [20,21], p53 point mutation, in addition to abolishing the tumor suppressor activity of the protein, also stabilizes it. This results in accumulation of the stabilized proteins to levels that can be detected by immunohistochemistry. It is possible that a process of secondary stabilization of the p53 protein takes place without the point mutation in some cases . In that case, it becomes even more important to correlate the immunohistochemical data to tumor behavior. In our case group, all five cases of papillary carcinomas with grade 2 p53 expression have not suffered mortality owing to the disease. All of the 18 cases of HCA, despite a high p53 expression, are currently alive with no recurrence of disease over a period of 2-5 yr. On the other hand, two
cases of Hiirthle cell carcinoma (grade 3 and 4) expired because of extensive spread of the disease within a year postdiagnosis. Since both these cases presented with extensive extrathyroidal metastasis, it can be argued that a higher tumor stage was responsible for the aggressive clinical behavior and, consequently, a high p53 expression in these cases. This point, although valid, does not explain the high p53 expression in two minimally invasive HCC (grade 3 and 4), both of whom are alive and well. We believe that a cautious approach should be maintained while evaluating these data, some of the cases being relatively recent. A high p53 expression in HCA should indicate an aggressive behavior, but within the time frame evaluated, none of the cases showed any disease recurrence. A longer follow-up may be helpful in further defining their biological behavior.
Acknowledgments We wish to thank Lou Savas and Teri DiLeo for technical assistance.
References 1. Rosen IB, Luk S, Katz I. Hiirthle cell tumor behavior: Dilemma and resolution. Surgery 98:777-783, 1985.
p53 and bcl-2 in HO~hle Cell Neoplasms 2. Thompson NW, Dunn EL, Batsakis JG, Nishiyama RH. Hiirthle cell lesions of thyroid gland. Surg Gynecol Obstet 139:555560, 1974. 3. Van Heerden J. H/irthle cell neoplasms [discussion]. Arch Surg 119:518-519, 1984. 4. Gossain K, Clark OH. Hiirthle cell neoplasms. Arch Surg 119:515-518, 1984. 5. Bondeson L, Azavedo E, Bondeson AG, Caspersson T, Ljungberg O. Nuclear DNA content and behavior of oxyphil thyroid tumor. Cancer 58:672-675, 1986. 6. Flint A, Davenport RD, Lloyd RV, Beckwith AL, Thompson NW. Cytophotometric measurements of Hiirthle cell tumors of the thyroid gland. Cancer 61:110-113, 1988. 7. Dobashi Y, Sakamoto A, Sugimura H, Mernyei M, Mori M, Oyama T, et al. Overexpression of p53 as a possible prognostic factor in human thyroid carcinomas. Am J Surg Pathol 17:375-381, 1993. 8. Soares P, Cameselle-Teijeiro J, SobrinhoSimoes M. Immunohistochemical detection of p53 in differentiated, poorly differentiated and undifferentiated carcinomas of the thyroid. Histopathology 24:205-210, 1994. 9. Czyz W, Joensuu H, Pylkkanen L, Klemi PJ. p53 protein, PCNA staining and DNA content in follicular neoplasms of the thyroid gland. J Pathol 174:267-274, 1994. 10. Donghi R, Longoni A, Pilotti S, Michieli P, Della Porta G, Pierotti MA. Gene p53 mutations are restricted to poorly differentiated carcinomas of the thyroid gland. J Clin Invest 91:1753-1760, 1993. 11. Hosal SA, Apel RL, Freeman JL, Rosen IB, LiVolsi VA, Asa SL. Immunohistochemical localization of p53 in human thyroid neoplasms: correlation with biological behavior. Endocr Pathol 8:21-28, 1997. 12. ReedJC. Bcl-2 and regulation of programmed cell death. J Cell Biol 124:1-6, 1994.
123 13. Pilotti S, Collini P, Rilke F, Cattoretti G, Del Bo R, Pierotti MA. BCL-2 protein expression in carcinomas originating from the follicular epithelium of the thyroid gland. J Pathol 172:337-342, 1994. 14. Pilotti S, Collini P, Del Bo R, Cattoretti G, Pierotti MA, Rilke E A novel panel of antibodies that segregatesimmunocytochemically poorly differentiated carcinoma from undifferentiated carcinoma of the thyroid gland. Am J Surg Pathol 18:1054-1064, 1994. 15. LiVolsiVA. Htirthle cell lesions. In: Surgical pathology of the thyroid. Philadelphia, PA: WB Saunders, 1990; 275-288. 16. Gundry SR, Bumey RE, Thompson N~, Llyod R. Totalthyroidectomyfor Hiirthle cellneoplasm of the thyroid.Arch Surg 118:529-532, 1983. 17. Matias-Guiu X, Cuatrecasas M, Musulen E, Prat J. p53 expression in anaplastic carcinomas arising from thyroid papillary carcinomas. J Clin Pathol 47:337-339, 1994. 18. Zedenius J, Larsson C, Wallin G, Backdahl M, Aspenblad U, Hoog A, et al. Alteration of p53 and expression of WAF 1tp21 in human thyroid tumors. Thyroid 6:1-9, 1996. 19. FaginJA, Matsuo K, Karmakar A, Chen DL, Tang SH, Koeffler HP. High prevalence of mutation of the p53 gene in poorly differentiated human thyroid carcinomas. J Clin Invest 91:179-184, 1993. 20. Porter PL, Gown AM, Kramp SG, Coltrera MD. Widespread p53 overexpression in human malignant tumors: an immunohistochemical study using methacarn-fixed, paraffin-embedded tissue. Am J Pathol 140:145-153, 1992. 21. JenkinsJR, Rudge K, Chumakov P, Currie GA. The cellular oncogene p53 can be activated by mutagenesis. Nature 317:816-818, 1985. 22. Wynford-Thomas D. p53 in tumor pathology: can we trust immunocytochemistry. J Pathol 166:329,330, 1992.