Abstract

Methods

Results

The studies that formed the basis of this meta-analysis included case–control studies as well as prospective and retrospective cohort studies (Table 1). As can be seen in this summary, limitations of the currently available data regarding RRSO in BRCA1/2 mutation carriers include variable study designs, small sample sizes for individual studies, many of which are retrospective in nature, and short post-RRSO follow-up times in prospective studies. Eight studies (8-10,12-16) estimated the risk of breast cancer in BRCA1/2 mutation carriers who were treated with RRSO relative to BRCA1/2 mutation carriers who did not receive this treatment (Table 2). As summarized in Table 3 and Figure 1, three nonoverlapping studies (14–16), which included 5703 participants, estimated the risk of breast cancer in BRCA1/2 mutation carriers who received RRSO relative to BRCA1/2 mutation carriers who did not receive the procedure, giving a summary HR estimate of 0.49 (95% CI = 0.37 to 0.65). Four nonoverlapping studies (12,14–16) estimated the risk reduction associated with RRSO for breast cancer in BRCA1 mutation carriers, giving a summary HR estimate of 0.47 (95% CI = 0.35 to 0.64). Finally, three nonoverlapping studies (14–16) estimated the relative risk for breast cancer in BRCA2 mutation carriers, giving a summary HR estimate of 0.47 (95% CI = 0.26 to 0.84) (Table 3, Figure 1).

Six studies (8,10,11,13,16,17) (Table 2) estimated the risk of gynecologic cancer in BRCA1/2 mutation carriers treated with RRSO relative to BRCA1/2 mutation carriers who did not receive this treatment. Based on data from the three nonoverlapping datasets (11,16,17), which included 2840 participants, the summary HR was 0.21 (95% CI = 0.12 to 0.39) (Table 3, Figure 1). Only one study (16) estimated the risk of gynecologic cancer in BRCA1 mutation carriers treated with RRSO relative to untreated BRCA1 carriers (HR = 0.15, 95% CI = 0.04 to 0.56) (Table 2). No study estimated the risk reduction associated with RRSO in BRCA2 mutation carriers. Kauff et al. (16) did investigate risk reduction in 294 women with BRCA2 mutations, but observed no post-RRSO gynecologic cancers in this sample.

We found no evidence of publication bias of any of our estimates based on the Begg and Majumder test statistics presented in Table 3. No evidence of study heterogeneity was found based on the x² test (Table 3).

Discussion

The clinical management of cancer risk in BRCA1 and BRCA2 mutation carriers is complex and should consider patient preferences; these preferences can be informed by accurate knowledge of the risks and benefits of the interventions considered (Table 4). The results of our meta-analysis suggest an 80% reduction in ovarian/fallopian tube cancer risk and a 50% reduction in breast cancer risk associated with RRSO in women who carry mutations in BRCA1 or BRCA2. The consistency of these findings across the included studies confirms the strong association of RRSO with reduced risks of breast and ovarian cancer in BRCA1 or BRCA2 mutation carriers. In addition, modeling studies have also demonstrated that salpingo-oophorectomy has a large effect on years of life added, particularly when adjusted for quality of life (31). Furthermore, in a prospective study with short-term follow-up, RRSO was associated with a 90% reduction in breast cancer–specific mortality, a 95% reduction in gynecologic cancer–specific mortality, and a 76% reduction in overall mortality (13). Therefore, all of the available data demonstrate the utility of salpingo-oophorectomy in this population of patients.

Despite the consistent evidence favoring RRSO in women with mutations in BRCA1 or BRCA2, the existing data remain somewhat limited in a number of ways. First, the influence of cohort effects on cancer risk over time remain unclear, despite evidence that differences in risk over time may reflect changing exposures, lifestyle, reproductive history, and use of screening or preventive surgeries (32). We lacked the data necessary to evaluate the effects of birth cohort, timing of surgery, or other factors that may influence the risk reduction estimates associated with RRSO. Therefore, at this time it is difficult to infer whether specific cohorts, exposure groups, or other strata may experience different risk reduction effects than others.

To limit the possibility that reporting bias influenced our findings, we included all published studies of RRSO in BRCA1/2 mutation carriers. However, we did not include any studies that reported the association of RRSO with cancer risk without providing estimates of risk reduction because these data would not contribute to pooled estimates of risk reduction. Because some studies included in the this analysis were limited in sample size and statistical power, their effect estimates for RRSO were large but not statistically significant, suggesting that a meta-analysis and presentation of summary statistics was appropriate. Two studies (15,17) were included in the summary estimates even though they used case–control designs, and therefore they yielded odds ratios rather than hazard ratios. Although odds ratios may slightly overestimate the risk reduction associated with RRSO, the annual incidence of ovarian and breast cancer in BRCA1/2 mutation carriers is no more than 2%–4%, with the result that odds ratios are likely to be similar to hazard ratios in this setting.

Some of the variability in the individual study estimates reported may reflect study design differences, including the use of retrospective vs prospective samples and poorly characterized selection biases. Despite these differences, we noted no statistically significant heterogeneity in the estimates of risk reduction after RRSO. In addition, cohort studies estimated a greater reduction in cancer risk associated with RRSO (particularly ovarian/fallopian tube cancers) compared with the case–control studies (Table 1). As a result, there is some variability in the estimates obtained using case–control and cohort studies; nonetheless, the estimates all consistently reflect risk reduction associated with RRSO.

We have included all of the large collaborative group studies that addressed the question of reduced risk conferred by RRSO and whose study populations come from and are representative of mutation carriers in North America and Europe. No studies of RRSO in nonwhite populations have been reported, and additional data may be needed to understand the role of RRSO in these groups. Finally, the samples of women with BRCA1/2 mutations reported here represent those who have generally been identified through high-risk clinics. Thus, these women may not be representative of the general population. However, they do represent the population of women who receive genetic testing and may be candidates for RRSO. Therefore, the populations summarized here represent the most relevant group in whom RRSO may be applied at this time.

Despite the strength and consistency of the data in the literature as reflected in our meta-analysis, a number of questions remain. There are only a few estimates of the association of RRSO with cancer risk in populations composed exclusively of BRCA1 mutation carriers or BRCA2 mutation carriers (12,14–16), and it is critical to understand how risk reduction may differ by gene. Using a prospective cohort approach and a large consortium dataset, we recently estimated gene-specific risks and found that hormonal modulation by RRSO may be associated with a greater reduction in breast cancer risk in BRCA2 mutation carriers than in BRCA1 mutation carriers (16). In contrast, the studies that used retrospective cohort (14) or case–control approaches (14,15) did not observe this difference, and therefore, there was no difference in the pooled estimates of breast cancer risk reduction reported in Table 1. Thus, differences in study design may influence the inferences we can make about the differences in risk reduction associated with RRSO in BRCA1 vs BRCA2 mutation carriers. The potentially larger risk reduction associated with RRSO in BRCA2 vs BRCA1 mutation carriers is of interest, given the high proportion of estrogen receptor (ER)–negative breast tumors in BRCA1 mutation carriers compared with BRCA2 mutation carriers (33). Our observation of a higher risk in BRCA2 mutation carriers should be followed up in larger studies that specifically evaluate tumor markers. In addition, attention needs to be given to the time interval between RRSO and breast cancer diagnosis. For example, it is possible that there is greater breast cancer risk reduction in BRCA2 mutation carriers, in whom the majority of tumors are ER positive, given that RRSO may treat some subclinical breast tumors. In contrast, in BRCA1 mutation carriers, who have predominantly ER-negative breast cancer, it is unclear whether a “treatment effect” may exist, and any primary prevention effect may require more time to emerge.

Finally, the effect of age at RRSO on cancer risk reduction remains unresolved. Eisen et al. (15) reported that the breast cancer risk reduction with RRSO was greater in BRCA1/2 mutation carriers who underwent surgery before age 50 than in women who underwent surgery after age 50. Among BRCA1 mutation carriers older than age 50, no risk reduction was evident with RRSO. No statistically significant association of RRSO at any age with risk reduction was observed in BRCA2 mutation carriers. Although these findings are consistent with effects of removal of hormone exposures in premenopausal women and not in postmenopausal women, the sample sizes in this analysis (15) were relatively small. Thus, additional studies are required to resolve the optimal age at surgery.

The importance of understanding the optimal age at which a woman should consider RRSO is underscored by a recent study (34) conducted in the general population that suggests that RRSO in women younger than age 45 is associated with an increased mortality, particularly if hormone replacement therapy (HRT) is not used. An initial report of HRT use after RRSO suggests that women can undergo RRSO and take HRT for a short time if needed after surgery because breast cancer risk is not substantially elevated in HRT users after RRSO (35). Although data on postmenopausal women do not demonstrate a cardiovascular benefit from HRT (36), an important limitation of this study (36) was the older age of the participants. More recent data have suggested that younger women going through natural menopause may indeed derive a cardiovascular benefit from HRT (36,37), and it is possible that BRCA1/2 mutation carriers undergoing abrupt surgical menopause to reduce ovarian cancer risk who receive HRT may in fact derive important cardiovascular, bone health, and quality-of-life benefits. Although the risk–benefit ratio of RRSO is very different in BRCA1/2 mutation carriers than in the general population, and RRSO in BRCA1/2 mutation carriers has been associated with improved overall survival in the short term, these studies pointing to the potentially complex relationship of RRSO and HRT exposure raise important and difficult questions. For example, it is not yet clear whether the long-term effects of long-term HRT in unaffected mutation BRCA1/2 carriers will ultimately be more beneficial in preventing noncancer mortality in these women or more harmful by increasing their risk of breast cancer (or potentially increasing cardiovascular events) compared with the general population. Given this possibility, studies that address the type, timing, and length of administration of HRT as well as its long-term effects on the association between RRSO and cancer risk and on other health factors in BRCA1/2 mutation carriers are urgently needed. In the interim, we provide a summary of clinical recommendations related to the detection and prevention of cancer in BRCA1/2 mutation carriers (Table 4).

Finally, although RRSO has become the standard of care for cancer risk reduction in women who have inherited BRCA1/2 mutations, other options for risk reduction also exist. Women with BRCA1/2 mutations who have been treated with risk-reducing mastectomy have a substantially reduced breast cancer risk (30). Furthermore, a study of breast cancer screening that added yearly magnetic resonance imaging to screening mammography suggested that combination of these modalities may also have benefit in the early detection of breast cancer in this group of women (28).

In conclusion, the summary risk reduction estimates presented here confirm that BRCA1/2 mutation carriers who have been treated with RRSO have a substantially reduced risk of both breast and ovarian cancer. However, residual cancer risk remains after surgery. Therefore, additional cancer risk reduction and screening strategies are required to maximally reduce cancer incidence and mortality in this high-risk population.

Funding

National Institutes of Health (R01-CA083855 and R01-CA102776 to TRR); the Department of Defense Breast Cancer Research Program (DAMD17-03-1-0375 to NDK); the Cancer Genetics Network (HHSN21620074400C to SMD); the Eileen Genet Fund and the Project Hope Fund for Ovarian Cancer Research and Education.

Footnotes

References