Impact of germline BRCA1 mutations and overexpression of p53 on prognosis and response to treatment following breast carcinoma
10-Year follow-up data
Fax: (514) 934-8273
W.D.F. is a Chercheur Clinicien Boursier (J2) of the Fonds de la Recherche en Santé du Québec.
Abstract
BACKGROUND
Overexpression of p53 has been associated with poor survival following breast carcinoma. BRCA1 interacts biochemically with p53 and may also contribute to poor outcome when constitutionally mutated. The joint effect of both abnormalities has not been studied. The primary objective of this study was to assess the impact of germline BRCA1 mutations and p53 overexpression on survival after 10 years of follow-up.
METHODS
A historical cohort of Ashkenazi Jewish women 65 years or younger with invasive breast carcinoma was tested for BRCA1 founder mutations. p53 overexpression was assessed by immunohistochemistry. Clinicopathologic information was obtained by chart review.
RESULTS
In total, 278 women were analyzed. On univariate analysis, p53 overexpression (n = 63) was prognostic for worse overall survival (relative risk [RR] 2.6, P = 0.001) whereas BRCA1 germline mutations (n = 30) were of borderline significance (RR 1.9, P = 0.052). In the lymph node-negative subpopulation, BRCA1 mutation status conferred a higher mortality on univariate (RR 5.6, P < 0.001) and multivariate (RR 3.5, P = 0.03) analyses. There was a trend in favor of a worse prognosis for women who carried a germline BRCA1 mutation and whose tumor overexpressed p53. When compared with noncarriers, BRCA1 mutation carriers had a worse overall survival if they did not receive adjuvant chemotherapy (RR 3.3, P= 0.01) or adjuvant hormonal therapy (RR 2.3, P = 0.02).
CONCLUSIONS
Germline BRCA1 mutations and p53 overexpression carry a negative prognosis that is not additive to known prognostic factors. Given the experimental sensitivity of BRCA1-mutated cells to chemotherapy, the worse survival among BRCA1 mutation-carrying lymph node-negative breast carcinoma patients may be partly explained by the significantly lower proportion of lymph node-negative patients who received adjuvant chemotherapy (P < 0.001). Cancer 2003;97:527–36. © 2003 American Cancer Society.
DOI 10.1002/cncr.11080
Breast carcinoma accounts for about 30% of new cancer cases and 18% of cancer deaths in Canada.1 Women in some areas of the Northeast United States have a significantly higher breast carcinoma mortality than women in other U.S. regions. Reproductive and socioeconomic characteristics explain much of this difference.2 Conversely, hereditary cancer syndromes account for 3–5% of breast carcinomas, with BRCA1 and BRCA2 mutations accounting for the greatest percentage of these syndromes.3, 4 In sporadic breast carcinoma, BRCA1 is rarely mutated,5 although expression is limited by DNA methylation-induced gene suppression.6 The BRCA genes are involved in DNA repair and function as tumor suppressors. Some pathologic characteristics of BRCA1-related breast carcinoma have been associated with a worse prognosis, such as higher grade,7-12 and with estrogen receptor (ER)9-13 and progesterone receptor (PR) negativity.9, 10, 13, 14 There are far fewer data for BRCA2-related cancer.15, 16 Prognosis for BRCA1 mutation carriers remains controversial. However, there is no mutation-based study showing improved survival in BRCA1 mutation carriers and some studies have found worse survival on univariate17, 18 and multivariate19 analyses.
The TP53 tumor suppressor gene may be the most commonly implicated gene in human malignancy.20 It acts as a checkpoint between the G1 and S phases of the cell cycle21 and plays a key role in directing a cell toward apoptosis when genomic integrity is threatened.20 The TP53 gene has been implicated in both sporadic22 and hereditary breast carcinoma, the latter as a part of the Li-Fraumeni syndrome.23
Somatic TP53 mutations are over-represented in germ-line BRCA1/2 mutation carriers.24, 25 As in the case of BRCA1 mutations, the presence of stabilized p53 or mutated TP53 is associated with higher-grade,26 ER-negative,26-29 and PR negative tumors.26, 30 It also increases the likelihood of local disease recurrence31 and suggests a worse prognosis,32-36 although sometimes only in specific patient subsets.30, 37, 38
In mice harboring mammary epithelial cells with conditionally excised exon 11 of the BRCA1 gene, mammary gland cancer occurred after a long latency period and at a relatively low rate of 25%. When mice with a Trp53-null allele were crossed with these mice, progeny bearing both the BRCA1 and Trp53 defects developed mammary gland cancers sooner and at a rate of almost 100%.39, 40 The same additive negative effect may occur in human breast carcinoma. Although somatic TP53 mutations are more frequent in the BRCA1 and BRCA2 mutation carrier background,24, 41 previous data lacked sufficient follow-up to assess the clinical impact of the interaction between TP53 and BRCA1 mutations.41
Some data suggest that breast carcinoma patients with TP53 mutations may respond differently to therapy. Some investigators have found that TP53 mutations or p53 overexpression confer resistance to tamoxifen,42-44 whereas others have found no such resistance.32, 45, 46 A similar discrepancy exists in terms of p53 status and the response to chemotherapy. Berns et al.,43 in their own negative study, noted that several studies gave conflicting reports in both the adjuvant and advanced settings. Other studies supported the influence of TP53 status on outcomes in patients treated with chemotherapy.47-49
More limited data exist with respect to BRCA1 mutation status and treatment with chemotherapy. Chappuis et al.50 found that patients with BRCA1 or BRCA2 mutations had a superior clinical and pathologic response to neoadjuvant anthracycline-containing chemotherapy when compared with noncarriers. In vitro work suggests that cells without functional BRCA1 transcription are more sensitive to treatment by doxorubicin,51, 52 cisplatin,53, 54 and radiation.51 However, cells with BRCA1 mutations are less sensitive to paclitaxel, possibly because of down-regulation of bcl-2 expression51, 55 and the importance of BRCA1 in facilitating apoptosis.51, 55, 56 Similar data on tamoxifen use in BRCA1 mutation-carrying cell lines is not available. Based on nonrandomized57 and retrospective analyses58, 59 of patients with BRCA1 mutations, oophorectomy is effective as primary prophylaxis and tamoxifen may prevent new contralateral tumors after primary breast carcinoma,60 suggesting a role for hormonal therapy in limiting BRCA1 mutation-related carcinogenesis.
The current retrospective cohort study evaluates a population of Ashkenazi Jewish women with breast carcinoma for interactions between BRCA1 germline mutations and p53 overexpression and prognosis. Three founder mutations, with an overall incidence of about 2.5% in this population, account for the overwhelming majority of BRCA mutations in the Ashkenazim.61 These three mutations are 185delAG and 5382insC in BRCA1 and 6174delT in BRCA2.62 Ashkenazi women who are BRCA1 mutation carriers have a greater incidence of p53 overexpression than noncarriers, making this a good population to evaluate such an interaction.16
Given the worse prognostic features seen in patients carrying TP53 mutations, we hypothesized that p53 overexpression would carry an adverse prognosis in our cohort of patients. Similarly, we hypothesized that patients with both a BRCA1 mutation and p53 overexpression would have an even worse prognosis, based on mouse data suggesting that the combination promoted greater tumorigenesis and reduced apoptosis. Finally, we investigated our population to determine whether BRCA1 mutations or p53 overexpression would affect the prognosis of patients receiving adjuvant treatment.
MATERIALS AND METHODS
Patients
Subjects included all patients who reported themselves as being Ashkenazi Jewish by birth and were diagnosed with primary invasive breast carcinoma between January 1, 1980 and November 1, 1995 at the Sir Mortimer B. Davis-Jewish General Hospital (SMBD-JGH; Montreal, Quebec). Women were 65 years old or younger and had undergone lymph node dissection. Routine evaluation included laboratory analyses, chest X-ray, mammography, liver ultrasonography, and bone scanning. Patients were excluded if there was no follow-up or if they had no information on disease or vital status at the time of last follow-up. The study was approved by the research ethics committee of the SMBD-JGH. Pathology specimens were identified from each of these women and clinicopathologic and follow-up information was obtained by chart review. In general, patients were seen every 6 months for 5 years, then yearly. Follow-up evaluations consisted of a clinical examination, laboratory testing, and radiologic imaging as needed.
Specimens were evaluated by one pathologist (L.R.B.) for histologic type, nuclear grade, lymph node status, and immunohistochemical assessment. They were coded and DNA was extracted from the paraffin wax-embedded blocks using standard techniques. Clinical, pathologic, and molecular data were collected in a mutually blinded fashion.
Expression of ER was detected as described previously,10 using a standard streptavidin-biotin-peroxidase complex immunohistochemical technique on paraffin-embedded tissue. Positivity suggests that greater than 10% of tumor cell nuclei show immunoreactivity.
BRCA1 and BRCA2 Mutation Status
Mutation analysis was carried out as previously described, looking specifically for the dominant mutations in the Ashkenazi population (BRCA1: 185delAG, 5382insC; BRCA2: 6174delT).18 Haplotype analysis was also used to confirm 5382insC mutations. The BRCA2 mutation, 6174delT, was sought using single-strand conformation analysis, mutation-specific polymerase chain reaction-restriction fragment length polymorphism endonuclease digestion analysis, and direct sequencing. We also used a size assay as a second assay for all three mutations.63 BRCA2 mutation carriers (n = 10) were excluded from further analysis because they constituted a group too small to be analyzed alone and it is has not been established that they can be combined with BRCA1 mutation carriers in survival analyses.
p53 Immunohistochemistry (IHC)
p53 protein accumulation was detected as previously described.64 A standard streptavidin-biotin-peroxidase complex immunohistochemical technique was used on paraffin-embedded tissue with an anti-p53 monoclonal antibody (DO-7; Dako, Carpinteria, CA). Microwave antigen retrieval with amplification was performed. A positive p53 score suggests that greater than 10% of tumor cell nuclei show immunoreactivity.
Statistical Analysis
For the descriptive analysis of continuous patient characteristics, significance was assessed using the t test and the Wilcoxon test. For the analysis of categorical variables, significance was assessed using the Fisher exact test. To assess significant trends in increasing odds ratios (OR), Cochran-Armitage's test was used.
Ten-year survival curves of overall and distant disease-free survival were calculated using the Kaplan–Meier method and significance was assessed using the log rank test. Overall survival was defined as the interval between surgical resection of the primary breast carcinoma and death from any cause. Distant disease-free survival was defined as the interval between surgical resection of the primary breast carcinoma and the first recurrence of breast carcinoma, excluding any ipsilateral or contralateral breast carcinoma disease appearance. Patients were censored at death or at the end of the follow-up period.
To assess the relative risk (RR) of death as well as the risk of distant disease recurrence, the Cox proportional hazards model was used. The following prognostic factors were evaluated: tumor size, axillary lymph node status, nuclear grade, p53 IHC status, and BRCA1 mutation status. All results were censored at 10 years.
To investigate a possible additive effect with BRCA1 germline mutations and p53 IHC status, survival curves were constructed for patients with and without each of these abnormalities and comparisons between curves were made for overall survival.
Kaplan–Meier curves and Cox multivariate models were constructed to compare overall survival by BRCA1 mutation status and p53 IHC status stratified by adjuvant chemotherapy and by adjuvant hormonal therapy.
RESULTS
Patient Characteristics
Of 292 women tested, 10 who carried a BRCA2 mutation and 4 who lacked information on vital status at last follow-up were excluded from analysis, including one BRCA1 mutation carrier. For the 278 women analyzed, the median follow-up was 8.0 years. Patient and disease characteristics are shown in Table 1. Thirty patients had BRCA1 mutations (20 with 185delAG and 10 with 5382insC) (10.8%, 95% confidence interval [CI] 7.4-15) and 63 patients had positive IHC for p53 (22.7%, 95% CI 18–28). Patients with BRCA1 mutation-related tumors were more likely to be diagnosed at a younger age (46.7 vs. 53.8 years, P = 0.004), to have higher grade tumors (P < 0.0001 for trend), to have ER-negative tumors (76.7%vs. 31.8%, P < 0.001), and to have positive p53 IHC (48.3% vs. 20.5%, P = 0.002) than patients without BRCA1 mutations. No difference was seen between the group with BRCA1 mutation-related tumors and the group without mutations with respect to the percentage of patients who had positive lymph nodes (34.5% vs. 45.8%, P = 0.3) or to the mean percentage of positive lymph nodes (10.0% vs. 14.8%, P= 0.3).
| Characteristics | All subjects (n = 278) (%) | BRCA1/2 noncarriers (n = 248) (%) | BRCA1 carriers (n = 30) (%) | P valuea |
|---|---|---|---|---|
| Age at diagnosis (n = 278) (yrs) | ||||
| Median | 53.4 | 53.8 | 46.7 | 0.004 |
| Range | 26.5–64.9 | 26.5–64.9 | 31.6–62.0 | |
| Tumor size (n = 266) (cm) | ||||
| Median | 1.6 | 1.6 | 2.0 | 0.3 |
| Range | 0.10–14 | 0.10–14 | 0.15–5.0 | |
| Nuclear grade (n = 276) | ||||
| 1 | 73 (26) | 72 (29) | 1 (3) | |
| 2 | 112 (41) | 104 (42) | 8 (27) | 0.09 |
| 3 | 91 (33) | 70 (28) | 21 (70) | < 0.001 |
| Estrogen receptor (n = 272) | ||||
| Negative | 100 (37) | 77 (32) | 23 (77) | |
| Positive | 172 (63) | 165 (68) | 7 (23) | < 0.001 |
| Axillary lymph nodes (n = 254) | ||||
| Negative | 141 (56) | 122 (54) | 19 (66) | |
| Positive | 113 (44) | 103 (46) | 10 (34) | 0.3 |
| p53 IHC (n = 268) | ||||
| Negative | 205 (76) | 190 (79) | 15 (52) | |
| Positive | 63 (24) | 49 (21) | 14 (48) | 0.002 |
| Surgery (n = 278) | ||||
| Lumpectomy | 218 (78) | 193 (78) | 25 (83) | |
| Mastectomy | 60 (22) | 55 (22) | 5 (17) | 0.6 |
| Adjuvant chemotherapy (n = 272) | ||||
| No | 143 (53) | 130 (54) | 13 (43) | |
| Yes | 129 (47) | 112 (46) | 17 (57) | 0.3 |
| Type of chemotherapy (n = 128) | ||||
| Anthracycline | 65 (51) | 57 (51) | 8 (47) | |
| Nonanthracycline | 63 (49) | 54 (49) | 9 (53) | 0.8 |
| Adjuvant hormonal therapy (n = 260) | ||||
| No | 131 (50) | 112 (48) | 19 (68) | |
| Yes | 129 (50) | 120 (52) | 9 (32) | 0.07 |
- a P value for the trend was calculated assuming a continuous definition of the variable grade.
- IHC: immunohistochemistry.
Patients with positive p53 IHC had tumors that were larger (median 2.0 cm vs. 1.5 cm, P = 0.003), had a higher nuclear grade (P < 0.001 for trend), were more commonly ER negative (67% vs. 27%, P < 0.001), and were more likely to be associated with BRCA1 mutation-carrying status (OR 3.6, 95% CI 1.6–8.0, P = 0.002). These patients also more commonly received adjuvant chemotherapy (65% vs. 42%, P = 0.002).
The administration of adjuvant systemic therapy did not vary with BRCA1 mutation status (Table 1). For the overall cohort, subjects with positive lymph nodes were more likely to receive adjuvant chemotherapy than those with negative lymph nodes (79% vs. 24%, P < 0.001), whereas lymph node-positive subjects were no more likely than lymph node-negative subjects to receive hormonal therapy (50% vs. 47%, P = 0.6).
Overall Survival
For the overall cohort, 5- and 10-year overall survival rates were 83% and 72%, respectively. Median overall survival has not been reached.
Univariate analysis (Table 2) shows that greater tumor size (RR 3.2), higher nuclear grade (Grade 3 vs. Grade 1, RR 8.8; Grade 2 vs. Grade 1, RR 3.5), positive lymph node status (RR 2.8), and a positive p53 IHC status (RR 2.6; Fig. 1) were significant prognostic factors for a poorer overall survival. A positive BRCA1 mutation status was of borderline significance on univariate analysis (RR 1.9, P= 0.052) and by the log rank test (P = 0.048; Fig. 2).
| Variable | Univariate analysis | Multivariate analysis (n = 236) | ||
|---|---|---|---|---|
| RR (95% CI) | P value | RR (95% CI) | P value | |
| Tumor size (cm) | ||||
| < 2 | 1.0 | 1.0 | ||
| ≥ 2 | 3.2 (1.9–5.5) | 0.0001 | 1.9 (1.01–3.4) | 0.04 |
| Nuclear grade | ||||
| I | 1.0 | 1.0 | ||
| II | 3.5 (1.4–9.3) | 0.01 | 2.4 (0.9–6.5) | 0.09 |
| III | 8.8 (3.5–22) | 0.0001 | 4.0 (1.5–11) | 0.007 |
| Lymph nodes | ||||
| Negative | 1.0 | 1.0 | ||
| Positive | 2.8 (1.7–4.7) | 0.0001 | 2.2 (1.3–3.9) | 0.006 |
| p53 IHC | ||||
| Negative | 1.0 | 1.0 | ||
| Positive | 2.6 (1.6–4.3) | 0.0001 | 1.5 (0.9–2.7) | 0.1 |
| BRCA1 mutation | ||||
| Noncarriers | 1.0 | 1.0 | ||
| Carriers | 1.9 (0.99–3.6) | 0.052 | 1.4 (0.7–2.9) | 0.3 |
- RR: relative risk; CI: confidence interval; IHC: immunohistochemistry.
Overall survival among patients with and without tumors with p53 overexpression (P = 0.0001).
Overall survival among patients with and without BRCA1 germline mutations (P = 0.05).
The Cox multivariate model shows that greater tumor size (RR 1.9), nuclear Grade 3 (RR 4.0), and positive lymph node status (RR 2.2) remained statistically significant, but p53 IHC status did not (Table 2).
When overall survival was evaluated in the lymph node-negative and positive subpopulations, general trends agreed with the overall population (Tables 3, 4). Univariate analysis with the 141 lymph node-negative patients showed that high nuclear grade (RR 4.3), p53 overexpression (RR 3.0), and positive BRCA1 mutation status (RR 5.6) were significantly associated with higher mortality. In the multivariate model, only positive BRCA1 mutation status (RR 3.5) retained significance.
| Variable | Univariate analysis | Multivariate analysis (n = 134) | ||
|---|---|---|---|---|
| RR (95% CI) | P value | RR (95% CI) | P value | |
| Tumor size (cm) | ||||
| < 2 | 1.0 | 1.0 | ||
| ≥ 2 | 1.9 (0.8–4.4) | 0.2 | 0.8 (0.3–2.2) | 0.7 |
| Nuclear grade | ||||
| I | 1.0 | 1.0 | ||
| II | 1.3 (0.4–4.6) | 0.7 | 0.9 (0.2–3.5) | 0.9 |
| III | 4.3 (1.4–14) | 0.01 | 2.2 (0.6–8.6) | 0.3 |
| p53 IHC | ||||
| Negative | 1.0 | 1.0 | ||
| Positive | 3.0 (1.2–7.3) | 0.02 | 1.5 (0.6–4.2) | 0.4 |
| BRCA1 mutation | ||||
| Noncarriers | 1.0 | 1.0 | ||
| Carriers | 5.6 (2.3–14) | 0.0001 | 3.5 (1.1–11) | 0.03 |
- RR: relative risk; CI: confidence interval; IHC: immunohistochemistry.
| Variable | Univariate analysis | Multivariate analysis (n = 102) | ||
|---|---|---|---|---|
| RR (95% CI) | P value | RR (95% CI) | P value | |
| Tumor size (cm) | ||||
| < 2 | 1.0 | 1.0 | ||
| ≥ 2 | 3.7 (1.5–8.8) | 0.003 | 2.6 (1.1–6.5) | 0.03 |
| Nuclear grade | ||||
| I | 1.0 | 1.0 | ||
| II | 8.9 (1.2–67) | 0.03 | 6.3 (0.8–48) | 0.08 |
| III | 17 (2.3–130) | 0.006 | 9.1 (1.2–71) | 0.04 |
| p53 IHC | ||||
| Negative | 1.0 | 1.0 | ||
| Positive | 2.6 (1.2–4.3) | 0.01 | 1.4 (0.7–2.8) | 0.3 |
| BRCAI mutation | ||||
| Noncarriers | 1.0 | 1.0 | ||
| Carriers | 0.8 (0.2–2.3) | 0.6 | 0.8 (0.2–2.5) | 0.6 |
- RR: relative risk; CI: confidence interval; IHC: immunohistochemistry.
On univariate analysis in the 113 lymph node-positive patients, size greater than or equal to 2 cm (RR 3.7), nuclear Grade 2 or 3 (RR 8.9 and RR 17.0, respectively), and p53 overexpression (RR 2.3) were significantly associated with worse survival, but BRCA1 mutation status was not. On multivariate analysis, only size greater than or equal to 2 cm (RR 2.7) and nuclear Grade 3 (RR 9.1) retained significance.
Distant Disease-Free Survival
For the overall cohort, 5- and 10-year distant disease-free survival rates were 76% and 67%, respectively. Median distant disease-free survival has not been reached.
The results of analyses on distant disease-free survival paralleled the results of overall survival (Table 5). On univariate analysis, greater tumor size (RR 3.9), higher nuclear grade (Grade 3 vs. Grade 1, RR 6.6; Grade 2 vs. Grade 1, RR 3.0), positive lymph node status (RR 2.8), and a positive p53 IHC status (RR 2.1), but not BRCA1 mutation status, were associated with a significantly worse distant disease-free survival.
| Variable | Univariate analysis | Multivariate analysis (n = 236) | ||
|---|---|---|---|---|
| RR (95% CI) | P value | RR (95% CI) | P value | |
| Tumor size (cm) | ||||
| < 2 | 1.0 | 1.0 | ||
| ≥ 2 | 3.9 (2.4–6.4) | 0.0001 | 2.4 (1.4–4.2) | 0.002 |
| Nuclear grade | ||||
| I | 1.0 | 1.0 | ||
| II | 3.0 (1.4–6.6) | 0.005 | 2.4 (1.04–5.6) | 0.04 |
| III | 6.6 (3.1–14) | 0.0001 | 3.8 (1.6–9.1) | 0.002 |
| Lymph nodes | ||||
| Negative | 1.0 | 1.0 | ||
| Positive | 2.8 (1.8–4.4) | 0.0001 | 2.0 (1.2–3.4) | 0.006 |
| p53 IHC | ||||
| Negative | 1.0 | 1.0 | ||
| Positive | 2.1 (1.3–3.2) | 0.002 | 1.2 (0.7–2.0) | 0.5 |
| BRCAI mutation | ||||
| Noncarriers | 1.0 | 1.0 | ||
| Carriers | 1.6 (0.9–2.9) | 0.1 | 1.2 (0.7–2.4) | 0.5 |
- RR: relative risk; CI: confidence interval; IHC: immunohistochemistry.
In the Cox multivariate model, greater tumor size (RR 2.4), higher nuclear grade (Grade 3 vs. Grade 1, RR 3.8; Grade 2 vs. Grade 1, RR 2.4), and positive lymph node status (RR 2.0) remained statistically significant but p53 IHC status did not (Table 5).
Combined Effect of BRCA1 Mutation Status and p53 IHC
We compared the interactive effects of BRCA1 mutations and p53 IHC on overall survival (Fig. 3). An overall difference was found among the four curves (P < 0.001), between the BRCA1/2 mutation noncarriers/p53 IHC-negative patients and the BRCA1 mutation noncarriers/p53 IHC-positive patients (P < 0.001), and between the BRCA1/2 mutation noncarriers/p53 IHC-negative patients and the BRCA1/2 mutation carrier/p53 IHC-positive patients (P = 0.001), but not between the other groups. For distant disease-free survival, the same pattern was found (data not shown). When an interaction term was evaluated in the multivariate models for overall survival and distant disease-free survival, no significant interaction was found between BRCA1 mutation status and p53 IHC status.
Overall survival according to combination of abnormalities. A, BRCA1/2 mutation noncarriers/p53 IHC-negative patients; B, BRCA1 mutation carriers/p53 IHC-negative patients; C, BRCA1 mutation noncarriers/p53 IHC-positive patients; D, BRCA1/2 mutation carriers/p53 IHC-positive patients. Overall, P < 0.001; A vs. C, P < 0.001; A vs. D, P = 0.001.
Treatment Effects
Treatment in this study was not assigned randomly. Therefore, a comparison between treatment groups is problematic. However, within groups, it may be possible to study the effects of certain variables on outcome. To this end, we evaluated the impact of BRCA1 mutation status and p53 IHC status on overall survival in the setting of adjuvant hormonal therapy or chemotherapy. Patients with BRCA1 mutations had a significantly worse overall survival compared with patients without BRCA1 mutations in the subset of patients (n = 131) who were not treated with hormonal therapy (RR 2.3, 95% CI 1.2–4.7, P= 0.02) and in the subset of patients (n = 143) who were not treated with adjuvant chemotherapy (RR 3.3, 95% CI 1.2–8.8, P= 0.01). p53 IHC-positive patients fared worse in terms of overall survival than p53 IHC-negative patients in the subset of patients being given adjuvant chemotherapy (RR 2.5, 95% CI 1.4–4.6, P = 0.003). However, the trend in the group not receiving chemotherapy was also in the same direction, if not significant (RR 1.8, 95% CI 0.7–4.7, P = 0.2). By contrast, p53 IHC-positive patients had a worse overall survival than p53 IHC-negative patients when they were not given hormonal therapy (RR 2.4, 95% CI 1.3–4.4, P = 0.004), with a trend toward worse overall survival also present in patients with p53 overexpression in the subset who did receive hormonal therapy (RR 2.5, 95% CI 0.9–6.7, P = 0.07). None of these results was significant on multivariate analysis and we stress the preliminary nature of these observations.
DISCUSSION
To date, despite convincing evidence that tumors harboring a germline BRCA1 mutation exhibit worse pathologic features, there is still controversy over whether BRCA1 mutations independently confer a worse prognosis. The results of the current study are consistent with an earlier assessment of this database18 and suggest that a BRCA1 mutation is a negative prognostic factor. Overall, women carrying a mutation had tumors with features that were prognostic for a poorer outcome and had a worse overall survival that was of borderline significance on univariate analysis (RR 1.9, P = 0.052). Patients in the lymph node-negative subset had a statistically significant poorer survival on multivariate analysis (RR 3.5, P= 0.03). The reasons for this finding bear consideration.
This study suggests that adjuvant treatment may be particularly important in patients carrying a BRCA1 mutation. Carriers had a worse survival than noncarriers in the subset of patients who did not receive adjuvant chemotherapy (RR 3.3, P = 0.01). We do not know whether the use of chemotherapy in this group of women would have lessened the negative impact of mutation status on outcome. However, there is evidence that BRCA1 mutations confer sensitivity to treatment with doxorubicin and cisplatin.51-54 Although cell lines with BRCA1 mutations are perhaps less sensitive to paclitaxel,51, 55, 56 the overwhelming majority of our data lies in the pretaxane era.
Further and not surprisingly, patients were less likely to receive chemotherapy if they had no involved lymph nodes (24% treated) than if they had involved lymph nodes (79% treated; P < 0.001 by the Fisher exact test). If BRCA1 mutation-bearing tumors are chemosensitive, then the poorer prognosis that might otherwise be seen in the lymph node-positive group may be improved by treatment and the worse prognosis observed in the lymph node-negative group may be related to the lack of treatment.
It is less clear why patients not treated with hormonal therapy fared worse if they harbored BRCA1 mutations (RR 2.3, P = 0.02). Noting that tumors bearing these mutations have features that portend a worse prognosis, it is possible that these patients benefit more from treatment than patients without BRCA1 mutations. Both tamoxifen60 and oophorectomy57-59 appear to prevent breast carcinoma among women carrying BRCA1 mutations. Although there is clearly insufficient evidence to assert that BRCA1 mutation-bearing tumors are especially sensitive to hormonal therapy, the results do not suggest that treatment of BRCA1-related breast carcinoma with hormonal therapy is ineffective. This is a surprising result because the majority of BRCA1-related breast carcinomas are ER negative and adjuvant hormonal therapy is usually ineffective in the absence of ERs.65 A portion of our patients were treated in the era before ER status guided tamoxifen administration. Five of nine BRCA1 mutation carriers who received hormonal therapy had ER-negative breast carcinomas and six are still alive at the time of median follow-up. Cell line work suggests that BRCA1 suppresses the transcriptional activity of the ER directly66 and by coactivator down-regulation.67 If so, tumors without functional BRCA1 may derive particular benefit from maneuvers that limit ER stimulation.
The presence of stabilized p53 or mutated TP53 is associated with worse pathologic features. The current study confirmed these findings. In accordance with several other studies32-34, 36, 38 and our own assessment of women with negative lymph nodes,30 the presence of p53 by IHC increased the risk of both distant disease recurrence and death on univariate analyses. Overexpression of p53 carried a worse survival on univariate analysis in both the lymph node-positive and negative subsets.
In contrast to the situation with BRCA1 mutations, patients with tumors positive for p53 IHC overexpression had a poorer survival in the setting of adjuvant chemotherapy (RR 2.5, 95% CI 1.4–4.6, P = 0.003). This is consistent with a role for p53 in chemotherapy resistance and could be related to the fact that p53 is partly responsible for causing apoptosis in chemotherapy-treated cells.20 Among women not treated with adjuvant chemotherapy, there was also a nonsignificant trend toward worse survival among those with p53 IHC positivity. It is likely that patients who did not receive adjuvant therapy had a lower risk of developing distant disease and differences would be more difficult to detect. This information must be viewed in the context of conflicting studies in the literature on the role of p53 as a predictive factor of response to treatment.43, 47-49
Patients with p53 overexpression fared worse than those who did not have p53 overexpression among patients not treated with hormonal therapy. There was only a trend toward a worse outcome if patients were treated with hormonal therapy and had p53 overexpression. The clinical data are contradictory for the effect of hormonal therapy on tumors with p53 overexpression.32, 42-46 Tamoxifen functions both by antiproliferative effects and by inducing apoptosis.68 However, multiple pathways are involved in tamoxifen-induced apoptosis69 and there is evidence to suggest that p53 is not involved in this process.70 There is reason to expect that p53 expression may have limited effect on the impact of tamoxifen.
We were interested in assessing whether a combination of the BRCA1 mutation and p53 IHC positivity might induce a worse survival. There is reason to suspect that a combination of defects would be detrimental. Together, they increase the risk of murine mammary tumors over BRCA1 mutations alone.39, 40 In mice, the elimination of functional p53 can mitigate the embryogenic demise and increased apoptosis induced by BRCA1 mutations,71 although there is some controversy over these data.72 It is conceivable that tumor cells with BRCA1 mutations may be less prone to apoptosis in the permissive setting of a coexisting TP53 mutation, potentially worsening prognosis. The curves showing the survival of women with each combination of these genes are compatible with such a hypothesis (Fig. 3). Although p53 IHC-positive tumors with or without a BRCA1 mutation caused a worse overall survival than cells without either defect, no statistical difference was seen between patient groups having just one abnormality and those having a combination of the two. This may be a result of insufficient power or the absence of such a combined effect. Analysis in larger data sets may be informative.
Our study has limitations. Despite a reasonably long median follow-up, the size of the database is limited, potentially undermining our ability to detect significant multivariate and additive effects between p53 overexpression and BRCA1 mutations. In addition, although assessment of p53 accumulation is less costly than sequence analysis, there is evidence to suggest that it is less accurate, with modest specificity and a relative insensitivity to nonsense mutations and deletions.48, 73, 74 Although a good correlation between p53 accumulation and sequencing has been found in some studies,36, 48 others have found a greater rate of false-positive and false-negative results.33, 75 However, given the known degree of correlation, it is unlikely that our results are spurious. Treatment was not assigned randomly in this study. As a result, our findings with respect to treatment must be interpreted cautiously.
To summarize, BRCA1 mutations and p53 overexpression carry both prognostic and predictive value in breast carcinoma. Not surprisingly, their prognostic value is limited by their association with other pathologic factors that are also linked with poor outcome, particularly high nuclear grade. A worse prognosis that was independent of other variables was seen in women carrying BRCA1 mutations who did not have involved lymph nodes. The prognostic value in this group may be due to less common use of adjuvant treatment in the setting of increased sensitivity to chemotherapy of tumors harboring BRCA1 mutations. If our findings are confirmed by other studies, this would have implications for the treatment of these patients. Despite our inability to come to definitive conclusions regarding the additive effects of BRCA1 mutations and p53 overexpression, our data are compatible with the results obtained using a mouse model.71 Larger, prospective studies will be required to further define prognostic groups and to clarify the optimal selection of treatment for women with hereditary breast carcinoma.
Acknowledgements
The authors thank Graciela Kuperstein and Maral Ouzounian for excellent technical assistance.
