Comparative immunohistochemical studies of bcl-2 and p53 proteins in benign and malignant ovarian endometriotic cysts
Abstract
BACKGROUND
A number of histologic and epidemiologic studies have suggested an association between endometriosis and ovarian carcinoma. Some reports have described a transition from endometriosis to atypical endometriosis to carcinoma. Using immunohistochemistry, the authors compared staining patterns in benign endometriotic cysts with ovarian tumors and the endometriotic cyst lining from which they arose, in an attempt to identify sequential or etiologic correlations.
METHODS
One hundred thirteen formalin-fixed, paraffin-embedded sections were studied (30 benign ovarian endometriotic cysts, 24 endometriotic cysts containing endometrioid carcinomas, 19 endometriotic cysts harboring clear cell carcinomas, and 40 ovarian papillary serous cystadenocarcinomas). All sections were immunostained with anti-bcl-2 and anti-p53 antibodies using the streptavidin-biotin method.
RESULTS
bcl-2 was reported to stain 23% of benign endometriotic cysts, 67% of endometrioid carcinomas, 73% of clear cell carcinomas, and 50% of papillary serous carcinomas. Approximately 42% of benign endometriotic lesions adjacent to the endometrioid carcinoma and 73% adjacent to clear cell carcinomas were found to stain for bcl-2 (p = 0.274 [not significant (NS)] and P = 0.008, respectively). p53 staining was negative in the benign endometriotic cyst group and was positive in 37–55% of the group with tumors. p53 staining was positive in 25% of the benign endometriotic lesions next to the endometrioid carcinoma and in 9% of the benign endometriotic lesions next to clear cell carcinoma (P = 0.014 and P = 0.239 [NS], respectively).
CONCLUSIONS
The results of the current study suggest that alterations in bcl-2 and p53 may be associated with the malignant transformation of endometriotic cysts. Cancer 2002;94:2935–40. © 2002 American Cancer Society.
DOI 10.1002/cncr.10566
Ovarian carcinoma is the number one cause of death from gynecologic malignancy, and the fourth leading cause of cancer death in women in the U.S. 1 To our knowledge there currently is no adequate method for the early detection of ovarian carcinoma and, in spite of intensive research in this area, the mechanisms of ovarian carcinogenesis remain largely unknown. One approach to understanding the development of ovarian carcinoma further may be through a study of endometriosis.
A number of histologic and epidemiologic observations have suggested an association between endometriosis and ovarian carcinoma. Vercellini et al. 2 found that approximately 25% of patients undergoing surgery for ovarian carcinoma of endometrioid and clear cell types had associated endometriosis. In a study that involved the long-term follow-up of 20,686 Swedish women who initially were hospitalized for endometriosis, Brinton et al.3 reported a nearly two-fold excess risk of ovarian carcinoma that increased to nearly four-fold in patients with ≥ 10 years of follow-up. Mostoufizadeh and Scully4 detected 8 definite cases of carcinoma arising in 950 cases of ovarian endometriosis, a malignancy rate of 0.8%. Fukunaga et al.5 found atypia in 1.7% of cases of benign ovarian endometriosis, and in 61% of cases of endometriosis associated with ovarian carcinoma. Ogawa et al.6 described a transition from endometriosis to atypical endometriosis to carcinoma in a proportion of cases of ovarian endometriosis associated with carcinoma, and found a corresponding increase in Ki-67 immunohistochemical staining in these groups. In a study of three cases of carcinoma arising in a setting of endometriosis, Han et al.7 found no expression of p53 protein in benign endometriosis, an intermediate level in a borderline adenofibroma, and high levels of expression in the carcinoma case. They postulated that p53 protein may be involved in the malignant transformation of endometriosis. Similarly, we believe that through a study of ovarian carcinoma cases arising in endometriotic cysts, it might be possible to identify proteins involved in the early steps of tumor formation. Our hypothesis is that endometriosis adjacent to malignancy may harbor premalignant changes and might contain proteins that are involved in the early steps of carcinogenesis and that would not be present in control endometriosis (without associated malignancy). We selected the bcl-2 and p53 proteins because they are involved in apoptosis8-10 and also are possible candidates for involvement in carcinogenesis.
Molecular genetic alteration of p53 such as overexpression or mutation has been reported to occur in up to 92% of cases of advanced epithelial ovarian carcinoma. 11 The p53 gene, which is found on the short arm of chromosome 17, encodes a nuclear protein that normally acts to restrain inappropriate cellular proliferation.12 It functions as a tumor suppressor gene and negatively regulates cell division.13 Mutated p53 accumulates in cells and regularly is detected by immunohistochemical staining because of the increased stability of the mutated form of the protein.19
bcl-2, a protooncogene located on chromosome 18, codes for the bcl-2 protein, which is important in determining whether a cell will be irreversibly committed to apoptosis. 14 This protein localizes to the mitochondrial membranes and interferes with programmed cell death independently of its ability to promote cell division.15 bcl-2 belongs to a growing family that often is divided into two different categories: inhibitors of apoptosis such as bcl-2, bcl-xl, and MCL-1, and accelerators or promoters such as bax, bcl-xS, and bad.16, 17
MATERIALS AND METHODS
One hundred thirteen patient samples were studied, and were comprised of the following: 30 benign ovarian endometriotic cysts, 24 endometriotic cysts containing endometrioid carcinomas, 19 endometriotic cysts harboring clear cell carcinomas, and 40 ovarian papillary serous cystadenocarcinomas. The cases had been diagnosed in a clinical setting by gynecologic pathology specialists, including one of the authors (T.K.), who also reviewed the complete histopathology of each case.
The slides were dewaxed in xylene, rehydrated through a graded ethanol series to distilled water, and air-dried. Blocking of endogenous peroxidase activity and protein was performed using blocking reagent (Chemicon secondary detection kit; Chemicon, Temecula, CA) for 10 minutes followed by rinsing. Antigen retrieval was performed by incubating the slides in 10mM citrate buffer (pH 3.0) for 15 minutes in a 95–96 °C waterbath, then allowing the hot solution to cool to room temperature for 20 minutes. Slides were rinsed in buffer (Chemicon).
Appropriate antibody dilutions were determined using controls: lymph nodes for the bcl-2 (1:50) (Chemicon) antibody and breast carcinoma tissue for the p53 antibody (1:80) (Calbiochem). Antibody solution (in phosphate-buffered saline [pH 7.3]) was applied to cover the slide tissue surface, and incubated as follows: 2 hours at 37 °C for bcl-2 and 2 hours at room temperature for p53. Slides then were rinsed (Chemicon secondary detection kit), secondary antibodies were applied (Chemicon kit), and the slides rinsed again. Color development employed streptavidin horseradish peroxidase and diaminobenzidene (Chemicon kit). After rinsing, slides were counterstained with hematoxylin, dehydrated through a graded ethanol series to xylene, and then coverslipped.
The immunohistochemical slides were reviewed on four separate occasions, by different combinations of authors, without knowledge of the diagnoses. For the groups of endometrioid and clear cell carcinomas arising in endometriotic cysts, both cancerous areas and benign-appearing areas were evaluated. Endometriotic cysts were found to exceed 1 cm in greatest dimension and were lined by endometriotic mucosa, which displayed the standard features for pathologic diagnosis: benign-appearing, müllerian-type epithelial lining with associated loose, capillarized, endometrial-like stroma, with associated hemorrhage. The p53 antibody was identified when golden brown intranuclear immunohistochemical staining occurred. bcl-2 stain was identified in the plasma membrane and membranes of intracellular organelles, and staining occurred as a cytoplasmic granular pattern. For the bcl-2 antibody, lymphocytes served as internal positive controls. The extent and intensity of staining were scored qualitatively. Extent assessed approximately how many of the pertinent cells in the tissue were stained, and was scored as a percentage. Intensity assessed the strength of staining, and was scored with 1+ indicating weak staining, 2+ indicating moderate staining, and 3+ indicating strong or dark brown staining. We defined a sample as positive when ≥ 10% of the cells demonstrated staining of at least 1+ to 2+ (weak to moderate) intensity.
Statistical analysis utilized the Fisher exact test using the Stat View computerized program (SAS Institute, Inc., Cary, NC). The statistical comparison was performed between benign ovarian endometriomas and endometriosis adjacent to the endometrioid and clear cell carcinomas. A P value < 0.05 was considered statistically significant.
RESULTS
Immunohistochemical staining results are summarized in Table 1.
Group | Benign endometriotic cyst | Endometrioid carcinoma | Clear cell carcinoma | Serous papillary carcinoma | ||
---|---|---|---|---|---|---|
Benign | Malignant | Benign | Malignant | |||
bcl-2 | 7/30 (23%) | 5/12a (42%) | 16/24 (67%) | 8/11b (73%) | 14/19 (74%) | 20/40 (50%) |
p53 | 0/35 (0%) | 3/12b (25%) | 10/24 (42%) | 1/11a (9%) | 7/19 (37%) | 21/38 (55%) |
- The numerator in each case was the number of cases per group demonstrating positive staining. Positive staining was defined as ≥ 10% of pertinent cells in a given sample demonstrating staining of weak to moderate intensity.
- The denominator in each case was the total number of different patient samples per group.
- a P value was not significant.
- b P value was significant at < 0.05.
Endometriotic Cysts
Only 23% of the endometriotic cysts were found to demonstrate positive staining for bcl-2; the majority of the cysts were unstained (Fig. 1). None of the cysts stained positively for p53 (Fig. 2).
Endometrioid Carcinoma in an Endometriotic Cyst
Of the 24 endometrioid carcinomas, 67% stained positively for bcl-2 (16 of 24 cysts) (Fig. 3) and 42% stained positively for p53 (10 of 24 cysts) (Fig. 4). Twelve of the 24 malignant cysts contained areas of recognizable, benign-appearing, endometriotic cyst lining. Of these areas, 42% were found to stained positively for bcl-2 (5 of 12 cysts; P < 0.274 [not significant]) and 25% were found to stain positively for p53 (3 of 12 cysts; P = 0.014) (Figs. 5 and 6).
Clear Cell Carcinoma in an Endometriotic Cyst
Of 19 clear cell carcinoma cases, 74% stained positively for bcl-2 (14 of 19 cysts) (Fig. 7) and 37% stained positively for p53 (7 of 19 cysts) (Fig. 8). Eleven of the 19 carcinoma cases in the current study demonstrated adjacent benign endometriotic cyst lining and, of these 11 cases, 73% were found to stain positively for bcl-2 (8 of 11 cysts; P = 0.008) (Fig. 9) and 9% were found to stain positively for p53 (1 of 11 cysts; P = 0.239).
DISCUSSION
Apoptosis has been implicated in the pathogenesis of several benign and malignant diseases. Many gene products, including the bcl-2 family and p53, have been reported to be associated with its process. 8-10 bcl-2 prevents cells from becoming irreversibly committed to apoptosis.
Approximately 23% of benign endometriotic cysts, 67% of endometrioid carcinomas, and 74% of clear cell carcinomas were found to stain positively for bcl-2. The values reported in the current study are higher than those reported by Skirnisdottir et al., 17 a finding that possibly may be attributable to differences with regard to the definition of positive staining, which their article did not specify. Our results are similar to those reported by Diebold et al.14 bcl-2 staining was found to be underexpressed in benign endometriotic cysts compared with malignant tumors. Underexpression of bcl-2 would be expected to diminish the protective effect of this protein against apoptosis, and possibly enhance sloughing. Indeed, endometriotic cysts frequently demonstrate some areas that are devoid of lining. The results of the current study also demonstrate an altered, increased expression of bcl-2 in both benign-appearing and associated malignant areas in the malignant endometriotic cysts, with the highest values in malignant areas compared with benign controls. This alteration was found to be statistically significant for the endometriotic lesions adjacent to the clear cell carcinoma compared with the benign endometriotic cysts. Although this comparison was not statistically significant for the endometrioid carcinoma, there was a trend toward this direction.
p53 is involved in many of the biologic phenomena that are important in the regulation of cell proliferation. Immunohistochemical staining and polymerase chain reaction studies have shown an absence of p53 mutations in benign endometriotic lesions 7, 18 whereas its overexpression has been detected in a variety of human tumors, including ovarian carcinoma.11, 12, 13, 19
Staining for mutated p53 essentially was negative in benign endometriotic cysts and positive in the malignancies. These findings correlate with those of Chan et al., 21 who reported positive staining in malignant ovarian epithelial tumors but negative staining in benign and borderline tumors and normal surface epithelium. Of interest in the current study are the three cases of endometrioid carcinoma arising in endometriotic cysts, which demonstrated positive staining in both the tumor and the adjacent benign-appearing lining of the cyst (Figs. 5 and 6). This finding is in keeping with that of Sato et al.21 who, using a different methodology and marker, found loss of heterozygosity of the PTEN tumor suppressor gene in tumors and associated endometriosis, but not in benign endometriosis. They also suggest that the tumor/endometriosis model may be useful for determining early changes of carcinogenesis.
The absence of detectable bcl-2 and p53 staining in the other malignant endometriotic cysts may be interpreted in several ways: 1) bcl-2 and p53 may be involved in malignant transformation in only a subset of patients or 2) insufficient sampling may account for the lack of detectable bcl-2 and p53 in the other cases. Although a number of sections of each endometriotic cyst were sampled, as per routine procedure for pathologic diagnosis, it is possible that more exhaustive sampling may have revealed positive cells.
The findings of the current study, indicating that bcl-2 and p53 proteins were altered in benign-appearing areas of malignant endometriotic cysts when compared with benign endometriotic cysts, suggest that alterations in the expression of these two proteins may be involved in the malignant transformation of endometriotic cysts. Alternatively, it is possible that these changes are consequences of, and not causal in, tumorigenesis. Further larger series studies of these and other markers might identify those women with endometriosis who are at risk for developing ovarian carcinoma, and might serve as a model for ovarian carcinogenesis.
Acknowledgements
The authors thank John Mandeli, Ph.D., Department of Biomathematics, The Mount Sinai School of Medicine, for statistical analysis.