Introduction

The p53 protein is the product of a tumor suppressor gene located on the human chromosome 17, thought to regulate the proliferation of normal cells. Mutations of p53 gene have been reported in human breast carcinoma,^1-4 especially in more advanced and/or more aggressive tumors.^5-7 Molina et al⁸ recently observed p53 positivity in 37.3% of 655 human breast carcinomas. These authors reported no significant correlation between p53 expression and tumor size, nodal involvement, or histologic type. However, in another study, Falette et al⁹ found that tumors carrying p53 alterations had a highly aggressive behavior and that the presence of altered p53 was an independent prognostic marker of early relapse and death. The p53 immunoreactivity correlated with tumor type in only a few instances, with 40% to 50% of medullary breast carcinomas expressing p53 oncoprotein.^10,11

The expression of Bcl-2 protein has also been reported in breast cancer^12-14 without correlation to tumor type. This protein is capable of preventing apoptosis and promoting tumor development. Leek et al¹⁵ found a direct correlation between Bcl-2 expression and the presence of estrogen receptors in breast carcinoma. These authors also noticed no correlation between Bcl-2 protein expression and nodal status, tumor size, or differentiation. Others have reported strong Bcl-2 expression in small, estrogen-receptor-positive, slowly proliferating, and p53-negative tumors.¹⁶ Interestingly, it has been demonstrated that in human breast carcinoma cell lines, a mutated and/or wild-type p53 downregulates Bcl-2 expression.¹⁷

To further investigate the role of these biomarkers in the progression of breast ductal neoplasia, we analyzed the expression levels of p53 and Bcl-2 by immunohistochemistry in a group of 26 invasive human breast carcinomas. We also evaluated the relationship between these markers and the expression of estrogen and progesterone receptors.

Materials and Methods

Twenty-six human breast carcinoma specimens from surgical resections performed at H. Lee Moffitt Cancer Center & Research Institute (Tampa, Fla) and Cooper University Hospital (Camden, NJ) were obtained. All the tumors were infiltrating ductal carcinomas. One case had focal lobular features, and seven exhibited medullary features. Clinical information including age, sex, tumor type, grade, stage, and therapy were also obtained from the medical records.

Immunohistochemical detection of p53 and Bcl-2 proteins was performed on formalin-fixed, paraffin-embedded sections using the avidin biotin peroxidase complex technique (Vectastatin Elite ABC Kit, Vector Laboratories, Inc, Burlingame, Calif) following the manufacturer’s instructions. For the detection of p53, a murine monoclonal antibody, clone DO1

(Santa Cruz Laboratories, Calif), directed against a denaturation-resistant epitope of human p53 located between amino acid 37 and 45 was used at a 1:500 dilution in all cases. The primary antibody used for the detection of Bcl-2 product was Bcl-2 DAKO M887, 124 (Dako Corp, Carpenteria, Calif) (dilution 1:500 in phosphate-buffered saline with 1% bovine serum albumin). This antibody was applied to sections after microwave antigen retrieval, as previously described.

Positive and negative controls were performed at the same time for each section. Controls for specificity included incubation of the tissue sections with unrelated primary mouse monoclonal antibodies, with unrelated secondary antimouse monoclonal antibodies, and with phosphate-buffered saline. Reactions were observed and evaluated by two of us (D.C. and E.C.) according to the intensity and percentage of cell staining independently. Nuclear or cytoplasmic staining was assigned a numerical value of 3 for strong staining, 2 for moderate staining, or 1 for weak staining. The percentage of positively stained cells were segregated into the following groups: 0 to 33% into group 1, 34% to 66% into group 2, and 67% to 100% into group 3.

As the variables analyzed were ordinal, associations were assessed by Spearman’s Rho, a nonparametric analog to the Pearson correlation coefficient. All P values reported are two-sided.

Results

The Table indicates the clinicopathologic features and results of the immunohistochemical stains. The age of the patients ranged between 38 to 88 years (mean = 57 years; median = 58 years). The size of their tumors ranged from 0.7 cm to 4.0 cm (mean = 1.9 cm; median = 1.7 cm). Following the Bloom and Richardson grading system, 6 invasive carcinomas were low grade, 13 were intermediate grade, and 7 were high grade. Disease stage was as follows: 12 patients had stage I, 7 patients had stage IIA, 2 patients had stage IIB, and 5 patients had stage IV. Three patients received chemotherapy after surgery; in 9 patients, both radiation and chemotherapy were given after resection. Fifteen tumors expressed both estrogen and progesterone receptors, 5 tumors expressed estrogen receptors only, and 6 tumors were negative for both. At follow-up, only one patient died of disease 36 months after surgery. Of the others, 4 were alive with disease (mean survival = 28 months; median = 31 months), and 19 were alive with no evidence of disease (mean survival = 38 months; median = 82 months). Two patients were lost to follow-up.

Moderate to strong (2 to 3) nuclear p53 immunoreactivity was present in 13 (50%) of 26 infiltrating ductal carcinomas (Fig 1), the other 13 tumors being p53 negative. Similarly, Bcl-2 cytoplasmic protein expression was moderate to strong (2 to 3) in 14 (54%) of 26 tumors. Seven carcinomas were only weakly Bcl-2 positive, and this stain was negative in 5 cases (Fig 2). An inverse correlation between p53 and Bcl-2 proteins expression was noted in 12 cases but was not statistically significant (rho = –0.35; P = .08).

In 7 cases (27%), negative or weak (1) Bcl-2 stain was detected in the presence of strong p53 positivity, but in 7 other cases (27%), the reverse was true. Furthermore, moderate to high expression of both oncoproteins was present in 6 (23%) of the 26 cases. The normal mammary duct epithelium and the hyperplastic ducts, when present adjacent to the tumor, were intensely decorated with Bcl-2 protein antibody (Fig 3) but were p53 protein negative (Fig 4). When present, the areas of apocrine metaplasia were p53 and Bcl-2 negative. In 33% of cases, ductal carcinoma in situ (DCIS) was present adjacent to the invasive carcinoma. This component always revealed a strong Bcl-2 positivity, whereas p53 immunoreactivity was present in 78% of such DCIS areas. Negative or decreased Bcl-2 expression was seen in 64% of the low-stage tumors (I and II). This was in contrast to the intense staining observed in the high-stage carcinomas (III and IV). At follow-up, of the 7 patients with strong (3) Bcl-2 staining, one died of disease after 36 months and 4 were alive with disease (survival mean = 27.7 months; median = 31 months). All of the patients with low-stage tumors that had negative or weak (1) Bcl-2 expression were alive without disease (mean survival = 53 months; median = 89 months). This difference approached statistical significance (P=0.08) in this small study. There was no significant association between p53 or Bcl-2 protein expression and estrogen/progesterone receptor status. However, strong (3) Bcl-2 cytoplasmic positivity significantly correlated with tumor stage (rho = 0.506; P=0.008).

Discussion

In this study, we analyzed the expression of p53 and bcl-2 gene products in a group of 26 infiltrating carcinomas of the breast to determine their distribution during the progression of ductal carcinoma of the breast. The relationships of these markers with clinicopathologic features and estrogen/progesterone receptor status were also evaluated.

The tumor suppressor gene p53 is expressed at low levels in normal cells when DNA damage occurs. The wild-type p53 molecule mediates the arrest of the cell cycle in G₁.¹⁸ The p53 protein recognizes a sequence-specific DNA binding site, resulting in activation of growth-inhibitory genes.^19-21 Abnormalities of the p53 gene and protein, which are usually associated with allelic loss on chromosome 17, have been widely reported in breast carcinoma.^22,23 Andersen et al²⁴ used constant denaturant gel electrophoresis to identify p53 mutations in exons 5 through 8 and/or protein accumulation in 42 of 163 primary tumors and in 5 of 16 metastases. These authors,²⁴ and others who reproduced similar results,²⁵ observed a statistically significant association between p53 alterations and tumor size, histologic and nuclear grade, DNA ploidy, mitotic rate, proliferation index, positive node status, distant metastases, and lack of estrogen receptors. They concluded that abnormal p53 can be used as an independent prognostic indicator of shortened survival and recurrence. Interestingly, it has been postulated that p53 mutations in codons that directly contact DNA are those able to predict poor outcome.²⁶ However, in a retrospective study of 164 cases of breast carcinomas without node metastases, Bosari et al²⁷ did not confirm the independent prognostic role of p53 expression. In our study, we found moderate to strong p53 immunoreactivity in 50% of the tumors examined, but no statistically significant correlation was found between p53 and tumor grade, stage, or estrogen/progesterone receptor status. This may reflect the low number of cases studied; however, other large studies have produced similar results.⁸ In 12 (46%) of the total cases studied, DCIS was present adjacent to the invasive tumors and showed strong p53 nuclear positivity. This finding is in agreement with previous studies and supports the implication that abnormal p53 protein is involved in the progression of early-stage mammary carcinoma.²⁸

The role of p53 in apoptosis has also been demonstrated; p53 gene mutations, inducing loss of p53 function, confers resistance to apoptosis.^29,30 In this scenario, the increase in the number of tumor cells would result from both unregulated proliferation and resistance to cell death. The latter mechanism is reminiscent of the one operating in follicular B-cell lymphomas in which overexpression of Bcl-2 perpetuates the increase in the number of malignant B lymphocytes.³¹ Normal Bcl-2 protein is usually synthesized in cells undergoing proliferation or in well-differentiated, noncycling cells (permanent cells). Furthermore, normal proliferating cells (ie, intestinal or breast ductal cells) lose their capability of synthesizing Bcl-2 prior to undergoing apoptosis.^32,33 In this study, we noticed that, when present, normal breast ductal cells, hyperplastic ducts, and/or foci of DCIS adjacent to the infiltrating tumor were intensely Bcl-2 positive. This protein was also present in 54% of the tumors examined. Specifically, we observed a negative or decreased Bcl-2 immunoreactivity in 64% of the low-stage tumors (I and II) when compared to the adjacent normal ducts. The Bcl-2 antibody did, however, always intensely decorate the tumor cells in the high-stage carcinomas (III and IV). Consequently, Bcl-2 overexpression showed an association with worse prognosis. This correlation was not statistically significant, perhaps because the small number of tumors studied, and deserves additional investigation on a larger number of cases. This result is not in agreement with the finding of Leek et al¹⁵ who reported the correlation of Bcl-2-negative breast tumors with poor prognosis. This incongruence may reflect a difference in stage between the tumors in our study and those examined by Leek and associates. Our finding, however, is similar to that of Silvestrini et al¹⁶ who, in their study of 283 breast carcinomas, found an unfavorable predictive value of Bcl-2 expression that was mainly dependent on p53 expression. Interestingly, we found an inverse relationship between p53 and Bcl-2 immunoreactivity independent of tumor type in 77% of the total cases. This finding may reflect a bcl-2 gene downregulation of a mutated and/or wild-type p53 protein, a phenomenon also observed in vitro, in human breast carcinoma cell lines.¹⁷ It is possible that a dual mechanism may be active at different times during the carcinogenic process, in which both Bcl-2 and p53 interfere with each other’s regulatory processes. This hypothesis is supported by the recent finding that overexpressed Bcl-2 may suppress p21 (Waf1) expression independently of p53 and may alter cell cycle regulation.³⁴

Reports have shown a strong negative relationship between estrogen receptors and p53 expression^35,36 and between progesterone receptors and Bcl-2 expression,³⁷ as well as a positive association between Bcl-2 expression and estrogen-receptor-positive tumors.¹⁶ In this small study, we did not detect any correlation between the expression of either estrogen or progesterone receptors and Bcl-2 expression.

In summary, Bcl-2 overexpression in some tumors may extend survival of the tumor cells. In other tumors, an altered p53 oncoprotein may be unable to suppress growth and proliferation of neoplastic cells but may be capable of downregulating Bcl-2 expression. Our study indicates that p53 and Bcl-2 protein expression may be important in the progression of breast carcinomas. We also noted a significant association between Bcl-2 expression and high-stage breast tumors.