Effect of Oophorectomy on Survival After Breast Cancer in BRCA1 and BRCA2 Mutation Carriers

Importance  Women who carry a germline mutation in either the BRCA1 or BRCA2 gene face a lifetime risk of breast cancer of up to 70%, and once they receive a diagnosis of breast cancer, they face high risks of both second primary breast and ovarian cancers. Preventive bilateral salpingo-oophorectomy is recommended to women with a BRCA mutation at age 35 years or thereafter to prevent breast and ovarian cancer, but it is unclear whether oophorectomy has an impact on survival in women with BRCA-associated breast cancer.

Objective  To estimate the impact of oophorectomy on survival in women with breast cancer with a BRCA1 or BRCA2 mutation.

Design, Setting, and Participants  Retrospective analysis of patients selected by pedigree review of families who received counseling at 1 of 12 participating clinical genetics centers. Patients were 676 women with stage I or II breast cancer and a BRCA1 or BRCA2 mutation who were observed for up to 20 years after receiving a diagnosis between 1975 and 2008. Survival experience was compared for women who did and who did not undergo oophorectomy.

Main Outcomes and Measures  In all analyses, the primary end point was death due to breast cancer.

Results  Of the 676 women, 345 underwent oophorectomy after the diagnosis of breast cancer and 331 retained both ovaries. The 20-year survival for the entire patient cohort was 77.4%. The adjusted hazard ratio for death attributed to breast cancer in women who underwent oophorectomy was 0.38 (95% CI, 0.19-0.77; P = .007) for BRCA1 carriers and 0.57 (95% CI, 0.23-1.43; P = .23) for BRCA2 carriers. The hazard ratio for breast cancer–specific mortality was 0.76 (95% CI, 0.32-1.78; P = .53) for women with estrogen receptor–positive breast cancer and 0.07 (95% CI, 0.01-0.51; P = .009) for women with estrogen receptor–negative breast cancer.

Conclusions and Relevance  Oophorectomy is associated with a decrease in mortality in women with breast cancer and a BRCA1 mutation. Women with estrogen receptor–negative breast cancer and a BRCA1 mutation should undergo oophorectomy shortly after diagnosis.

Women who carry a germline mutation in either the BRCA1 or BRCA2 gene face a lifetime risk of breast cancer of up to 70%,¹ and once they receive a diagnosis of breast cancer, they face high risks of both second primary breast and ovarian cancers.^2,3 Preventive salpingo-oophorectomy is recommended to women with a BRCA mutation at age 35 years or thereafter to prevent breast and ovarian cancer,^4-6 but the benefit of oophorectomy may extend to reducing mortality from breast cancer as well.^7-9 In a large prospective study, Finch et al⁷ reported that the mortality rate from all causes combined before age 70 years was reduced by 77% in women with a BRCA1 or BRCA2 mutation who underwent an oophorectomy. The risk reduction with oophorectomy was present with women with a history of breast cancer (hazard ratio [HR], 0.31 [95% CI, 0.24-0.39]) and was particularly strong among those women with a BRCA1 mutation (HR, 0.21 [95% CI, 0.12-0.37]). In a similar study, Domchek et al⁸ reported an odds ratio of 0.35 (95% CI, 0.19-0.67) for oophorectomy and breast cancer–specific mortality among women with a history of breast cancer. These 2 studies imply that if an oophorectomy is performed after a diagnosis of breast cancer then mortality is reduced, but neither study adjusted for stage of cancer at diagnosis or for other treatments received. A study by Huzarski et al⁹ of women with early-onset breast cancer in Poland looked at predictors of mortality in women with and without a BRCA1 mutation. In the mutation-positive subgroup, oophorectomy was associated with a large and significant reduction in breast cancer mortality (adjusted HR, 0.30 [95% CI, 0.12-0.75]). The beneficial effect of oophorectomy was much higher for mutation carriers than for noncarriers.

Two other recent studies of BRCA mutation carriers have also reported reductions in mortality associated with oophorectomy for women with a history of breast cancer^10,11; for 1 of these, the difference was statistically significant. The majority of breast cancers in BRCA1 carriers are estrogen receptor (ER) negative, and the effect of oophorectomy on cancer mortality is perhaps unexpected. The large effect sizes observed in these observational studies support the recommendation that women with breast cancer and BRCA1 mutation should undergo oophorectomy as a therapeutic measure shortly after diagnosis, but in the absence of data from randomized trials, it is important that these findings be replicated in well-controlled observational studies. We sought to confirm these earlier observations in an independent cohort of BRCA1 and BRCA2 carriers with early-stage breast cancer. We have previously studied this cohort for the effect of contralateral mastectomy on mortality,¹² but in the present analysis, we focus on oophorectomy. To enhance the clinical relevance of our study, we restricted our attention to women who received a diagnosis of breast cancer with both ovaries intact and we sought to determine the impact of a subsequent oophorectomy on mortality from breast cancer. We review the 20-year mortality experience of 676 women with early-stage breast cancer who are known to be mutation carriers or who are likely to carry a deleterious mutation in the BRCA1 or BRCA2 gene and estimate the impact of oophorectomy on mortality.

Box Section Ref ID

To identify study participants, the pedigrees of BRCA families who received genetic counseling at the 10 participating cancer genetics clinics were reviewed. A family was considered to be eligible for the study when a BRCA1 or a BRCA2 mutation was documented in the family and at least 1 case of invasive breast cancer was recorded. Eligible study participants included all women from these families who received a diagnosis of stage I or II breast cancer at age 65 years or younger, between 1975 and 2008. Living and deceased women were eligible, but those with a prior diagnosis of cancer (including breast cancer) or those who resided outside North America were excluded. It was not necessary to be a proven carrier of the family mutation to be included in the study; however, affected women who were known to be noncarriers were excluded. All study procedures were approved by the institutional review boards at each of the participating centers. All participants (or their next of kin) gave written informed consent before participating in the study.

We identified 615 families, which, in aggregate, contained 1773 breast cancer cases. Of the 1773 cases of breast cancer, 417 women were excluded because the date of diagnosis indicated on the pedigree was before 1975, and 70 women were excluded because the age at diagnosis was older than 65 years. An additional 29 women were known not to carry the familial mutation and were therefore excluded. Nineteen women were excluded because they had a diagnosis of other cancer prior to breast cancer, and 26 women were excluded because they were treated outside North America.

Of the remaining 1212 women, we were able to obtain identifying information for 927 (76%). An attempt was made to contact each woman or her next of kin to obtain permission to review the medical records. Nineteen women (or their next of kin) declined to provide consent for the release of the medical records. The medical record was requested from the hospital where treatment was received for the remaining 908 women. In 76 cases, the hospital was not able to locate the record or did not forward the documents. The medical record was obtained for 796 of the 908 women (88%).

After review of the medical records, an additional 120 women were excluded. Of these, 32 women were excluded because tumor stage was greater than II, 18 women were excluded because the tumor was noninvasive (ductal or lobular carcinoma in situ), and 2 women were excluded because they declined treatment. We excluded 49 women who had undergone an oophorectomy prior to diagnosis. We excluded 19 women who had an initial diagnosis of bilateral breast cancer. The remaining 676 women are included in the analysis. Of the 676 women, 608 were proven to be gene carriers (90%) and 68 (10%) were not tested but were likely to be gene carriers. The 676 women were from 493 different families.

The treatment records and pathology documents were reviewed. We recorded tumor size (in centimeters), nodal status (positive/negative), and tumor grade (I to III). Estrogen receptor status was recorded as positive, negative, equivocal, or unknown. We recorded the use of chemotherapy (yes/no), tamoxifen (yes/no), and radiotherapy (yes/no). We established whether the patient had undergone a bilateral oophorectomy and, if so, the date of the operation. We considered as exposures those oophorectomies between the diagnosis of breast cancer and prior to distant recurrence. Women who had undergone oophorectomy prior to breast cancer diagnosis were excluded, and women who had an oophorectomy after a distant recurrence (n = 7) or after a diagnosis of ovarian cancer (n = 55) were categorized under no oophorectomy. We recorded whether the initial surgery was lumpectomy, unilateral mastectomy, or bilateral mastectomy. In some cases, the initial lumpectomy was followed later by an ipsilateral mastectomy and/or contralateral mastectomy. Dates and causes of death were abstracted from the medical record and were recorded as due to breast cancer, due to ovarian cancer, due to another cancer, due to another cause, or unknown.

Differences in proportions between subgroups of patients with and without oophorectomies were compared for statistical significance using the χ² test. A series of survival analyses was performed. The primary end point was death of breast cancer. A secondary end point was death of all causes. We considered the woman to be at risk for death from the date of the first surgical procedure until the last date of follow-up or until death of breast cancer, death of another cause, or a diagnosis of ovarian cancer. Hazard ratios were estimated using the Cox proportional hazards model, implemented in SAS. We evaluated the use of tamoxifen and chemotherapy as dichotomous variables. Oophorectomy was evaluated as a time-dependent variable. We estimated the effect of oophorectomy in the entire patient series and then in subgroups defined by mutation (BRCA1 vs BRCA2), age (<50 or ≥50 years), stage (I or II), and ER status. We did not perform formal tests of statistical interaction to evaluate the significance of differences in the HRs in the various subgroups. We also included ipsilateral mastectomy and contralateral mastectomy (compared with breast-conserving surgery) as exposures of interest. The HRs were adjusted for mutation status (BRCA1 vs BRCA2), age at diagnosis, ER status, tumor size, lymph node status, receipt of chemotherapy, and receipt of oophorectomy. P < .05 was considered to be statistically significant.

The characteristics of the patients with breast cancer who did and who did not undergo oophorectomy and the treatments that they received are presented in Table 1. The patients received a diagnosis between 1977 and 2009 and were observed after diagnosis for a mean (range) of 12.5 (0.7-20.0) years. Patients were observed until June 2012. A total of 128 participants (18.9%) died of breast cancer during the follow-up period. Of those who died, the mean (range) time to death was 7.8 (0.7-19.5) years after diagnosis. The 10-year breast cancer survival of women with a BRCA1 mutation was 84.0% and of women with a BRCA2 mutation was 87.5%. The 20-year survival for the entire patient cohort was 77.4%.

Women who had undergone oophorectomy prior to diagnosis were excluded. Of the 676 women who had intact ovaries at the time of diagnosis, 407 underwent oophorectomy after diagnosis and 269 did not. Of the 407 women with an oophorectomy, 7 women underwent oophorectomy after distant recurrence and 55 underwent oophorectomy after ovarian cancer; these 62 women were included in the no oophorectomy group. After these reclassifications there were 345 women in the oophorectomy group. In the entire cohort, there was a 56% reduction in breast cancer–specific mortality associated with oophorectomy in the adjusted analysis (HR, 0.46 [95% CI, 0.27-0.79]; P = .005) (Table 2). Oophorectomy was associated with a significant reduction in breast cancer–specific mortality in women with a BRCA1 mutation (HR, 0.38 [95% CI, 0.19-0.77]; P = .007) but not in women with a BRCA2 mutation (HR, 0.57 [95% CI, 0.23-1.43]; P = .23) (Table 3).

In addition, there were 9 deaths of ovarian cancer in the no oophorectomy group. There was a 65% reduction in all-cause mortality associated with oophorectomy in the adjusted analysis (HR, 0.35 [95% CI, 0.22-0.56]; P < .001).

Oophorectomy was performed a mean (range) of 6.1 (0.1-31.8) years after diagnosis. For the 70 BRCA1 carriers for whom the oophorectomy was performed within 2 years of breast cancer diagnosis, the HR was 0.27 (95% CI, 0.11-0.66; P = .004) compared with women with a BRCA1 mutation who never underwent oophorectomy. The protective effect of oophorectomy on deaths of breast cancer was apparent immediately after diagnosis and persisted for 15 years (Table 4). Among BRCA1 carriers, the impact of oophorectomy was similar for women with stage I and II breast cancers (Table 5) and for women who did and who did not receive chemotherapy. Among BRCA2 carriers, none of the patients with ER-negative breast cancer and an oophorectomy has died (Table 5), but the sample size was small (n = 41).

In the entire study group, the observed protective effect of oophorectomy was strong for women who had ER-negative breast cancer (HR, 0.07 [95% CI, 0.01-0.51]; P = .009). Fourteen women with ER-negative breast cancer diagnosed at older than 50 years did not undergo oophorectomy; of these, 3 (21%) died. In contrast, of the 15 women with ER-negative breast cancer older than 50 years who underwent oophorectomy, none died.

We also modeled the effects of breast surgery and chemotherapy on breast cancer–specific survival (Table 2). The 20-year survival of women who had a contralateral mastectomy was better than that of women who had unilateral mastectomy or lumpectomy (adjusted HR, 0.59 [95% CI, 0.34-1.04]; P = .07). Chemotherapy was associated with nonsignificant reductions in breast cancer mortality in BRCA1 carriers (adjusted HR, 0.61 [95% CI, 0.33-1.33]; P = .12) and in BRCA2 carriers (HR, 0.57 [95% CI, 0.23-0.47]; P = .23).

In this study, we confirm that women with breast cancer and a BRCA1 mutation improve their prognosis if they undergo oophorectomy and that the protective effect of the oophorectomy persists beyond age 50 years. In previous studies, we have shown that oophorectomy reduces the risk of both ovarian and breast cancers, and we now report that the benefits of oophorectomy also include a reduction in the risk of breast cancer mortality. We have recently reported that breast cancer patients with BRCA mutations who are treated with bilateral mastectomy have superior survival than women treated with unilateral mastectomy.¹²

There are several limitations to this study. This is a nonrandomized historical cohort study of patients enrolled starting in 1975, and the mortality experience reflects therapies in use at the time of diagnosis. Furthermore, these are predominantly clinically detected cancers and the survival of patients who received a diagnosis in the modern era through magnetic resonance imaging surveillance might be better. If so, then the current survival advantage associated with oophorectomy might be less than that reported here. Furthermore, this study included only women with stage I and II breast cancer and the results may not be applicable to women with advanced-stage breast cancer. The women were treated prior to the introduction of aromatase inhibitors, and we are not able to evaluate the use of the recent antihormonal drugs. Similarly, details on the human epidermal growth factor receptor 2 (HER2) status of the cancers were not available for most women.

These results extend the rationale for testing all women with early-stage breast cancer for BRCA1 mutations at the time of diagnosis.¹³ We did not observe a significant decline in mortality associated with oophorectomy in BRCA2 mutation carriers, but the number of BRCA2 carriers was much smaller and the 95% CI was wide (0.23-1.43). Future studies need to be done to extend the findings on BRCA2 to a larger number of patients. Surprising perhaps, the benefit associated with oophorectomy extends to women who received a diagnosis at older than 50 years. This cut point corresponds to the mean age of menopause in carriers.¹⁴ This observation complements our recent finding that the risk of first primary breast cancer also decreases when an oophorectomy is performed after age 50 years.¹⁵ We conclude from these 2 studies that the postmenopausal ovary remains active in terms of androgen production and that this affects cancer risk and progression, either directly or through aromatization to estrogen. If the latter is true, then we would expect maximum risk reduction to be achieved through the administration of an aromatase inhibitor after oophorectomy. This will be the topic of a future study.

Recently, 3 other cohort studies of BRCA1-positive patents with breast cancer have generated a similar result. Valentini et al¹⁰ followed 397 young women with a BRCA1 mutation (mean age at diagnosis, 33 years, 128 of whom had a pregnancy near the time of diagnosis) and found in a multivariate analysis that oophorectomy was associated with an 80% reduction in mortality (HR, 0.20 [95% CI, 0.06-0.62]; P = .006). In a historical cohort study of 234 BRCA1 mutation carriers with breast cancer from Poland, Huzarski et al⁹ found that oophorectomy was associated with a 70% reduction in mortality (HR, 0.30 [95% CI, 0.12-0.76]; P = .01). In a small study from Norway of 63 patients with invasive cancer with a BRCA1 mutation who were identified in the course of a screening program, oophorectomy was associated with a large reduction in mortality (HR, 0.27 [95% CI, 0.04-1.89]) but the sample size was small and this result was not statistically significant.¹¹ We confirm the result of Domchek et al,⁸ ie, that oophorectomy reduces breast cancer mortality in women with a BRCA1 mutation. However, in contrast to the study of Domchek et al,⁸ we found that the benefit was restricted to women with a BRCA1 mutation.

The strengths of our study include the large sample size and the confirmation of all treatments by review of the medical records. We distinguished deaths of breast cancer from deaths of other causes, and we identified and excluded oophorectomies that took place as treatment for distant recurrence or ovarian cancer. We included untested women and deceased women in the study to avoid the survivorship bias that would arise if genetic testing were a condition for inclusion, but most participants (90%) were confirmed mutation carriers. In our extended study, 720 women with breast cancer were tested for the family mutation; 691 (96%) were found to carry the mutation and 29 women (4%) were found not to carry the mutation (and were excluded). This suggests that, of the 68 women included in this study for whom genetic testing had not been performed, we would expect only 3 to be noncarriers (ie, <1% of the 676 participants). It is not practical to randomize women to oophorectomy or no oophorectomy; because of the high risk for ovarian cancer in this population, the operation is generally recommended for all carriers older than 35 years, including those with early-stage breast cancer.

In our study, the majority of oophorectomies were performed for prevention of ovarian cancer and not for the treatment of breast cancer. The mean time from diagnosis to oophorectomy was 6.4 years. In order to focus on the therapeutic benefit of performing an oophorectomy after diagnosis, we excluded women who had undergone oophorectomy prior to receiving a diagnosis of breast cancer. Furthermore, 7 women underwent oophorectomy after distant recurrence and 55 women underwent oophorectomy as treatment for ovarian cancer. We excluded these oophorectomies from the present study. It is important that follow-up studies be performed on women who undergo oophorectomy as part of their initial treatment, in particular, those women who undergo oophorectomy in the first year after diagnosis. It is also important that our observations be confirmed in other study populations. Further data are needed, in particular for BRCA2 carriers in order to confirm the benefit of oophorectomy in this population.

The data presented here suggest that oophorectomy should be discussed with the patient shortly after diagnosis. We recommend that the operation be performed in the first year of treatment to maximize the benefit. These data, coupled with emerging data on the unique sensitivities of BRCA-related breast cancers to certain classes of chemotherapy,¹⁶ and the mortality benefit derived from chemotherapy^16,17 and from contralateral mastectomy,¹² suggest that women with newly diagnosed breast cancer and a BRCA mutation might benefit from treatments that are tailored to the hereditary subgroup.

Accepted for Publication: March 4, 2015.

Corresponding Author: Steven A. Narod, MD, Women’s College Research Institute, 790 Bay St, Toronto, ON M5G 1N8, Canada (steven.narod@wchospital.ca).

Published Online: April 23, 2015. doi:10.1001/jamaoncol.2015.0658.

Author Contributions: Dr Metcalfe had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Metcalfe, Narod.

Acquisition, analysis, or interpretation of data: Metcalfe, Lynch, Tung, Kim-Sing, Olopade, Eisen, Rosen, Snyder, Gershman, Sun, Narod.

Drafting of the manuscript: Metcalfe, Olopade, Snyder, Gershman, Narod.

Critical revision of the manuscript for important intellectual content: Metcalfe, Lynch, Tung, Kim-Sing, Eisen, Rosen, Sun, Narod.

Statistical analysis: Sun, Narod.

Obtained funding: Metcalfe, Narod.

Administrative, technical, or material support: Lynch, Olopade, Snyder, Gershman.

Study supervision: Narod.

Conflict of Interest Disclosures: None reported.

Funding/Support: This research was funded by the Canadian Breast Cancer Foundation (Ontario Chapter). Dr Metcalfe is supported by the Canadian Institutes of Health Research and the Ontario Women’s Health Council.

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.