Survival after bilateral risk-reducing mastectomy in healthy BRCA1 and BRCA2 mutation carriers

Women with a germline BRCA1/2 gene mutation have high risks of developing breast cancer (BC), estimated to range from 45 to 88% for a first BC up to the age of 70 years [1‐4]. Moreover, BC is diagnosed at a younger age in BRCA1/2 mutation carriers than in the general population [4‐6], with an increased risk from the age of 25 years. For healthy BRCA1/2 mutation carriers, the options are to follow a BC surveillance program aimed at early BC detection, or to opt for bilateral risk-reducing mastectomy (BRRM) to reduce BC risk. In healthy BRCA1/2 mutation carriers, BRRM reduces the risk of BC with estimates even up to 100% [7‐12], and this method may have beneficial effects on quality of life by diminishing the strong anxiety of getting BC. However, despite the strong BC risk-reduction, no clear survival benefit of BRRM over BC surveillance has been reported so far.

Mathematical models with simulated cohorts suggested that surveillance with both mammography and magnetic resonance imaging (MRI) in combination with risk-reducing salpingo-oophorectomy might offer an almost comparable survival as BRRM with risk-reducing salpingo-oophorectomy, due to improved imaging techniques and better systemic treatment options in recent years [13‐15]. However, no convincing prospective data are available so far. Previously, we observed better 10-year overall survival in the BRRM group than in the surveillance group (99% vs. 96%) among 570 healthy BRCA1/2 mutation carriers, but this difference was not significant [10].

To investigate whether BRRM leads to survival benefit, we determined the overall and breast cancer-specific mortality rates among 2857 healthy BRCA1/2 mutation carriers opting for either BRRM or surveillance with follow-up until 2017. Since BRCA2-associated BCs have more favorable characteristics than BRCA1-associated BCs [10, 16, 17], and BRCA2 mutation carriers have shown lower recurrence rates than BRCA1 mutation carriers [10], we performed all analyses for BRCA1 and BRCA2 mutation carriers separately.

In this nationwide cohort study, we observed lower overall and breast cancer-specific mortality rates among BRCA1 mutation carriers opting for BRRM than among those under surveillance. For BRCA2 mutation carriers, BRRM was nonsignificantly associated with lower overall mortality when compared with surveillance. Not one BRCA2 mutation carrier died of BC after BRRM, while the surveillance group performed almost as good. In addition, BRCA2-associated BCs were diagnosed less frequently, and had more favorable characteristics than BRCA1-associated BCs.

All analyses were performed separately for BRCA1 and BRCA2 mutation carriers, which is more accurate because BRCA1-associated BCs and BRCA2-associated BCs represent different entities. The current results are in line with our previous observation of a small but nonsignificant better 10-year overall survival after BRRM than under surveillance (99% vs. 96%) for a smaller combined cohort of BRCA1/2 mutation carriers [10]. The observation that BRRM was associated with lower breast cancer-specific mortality for BRCA1 mutation carriers, and not for BRCA2 mutation carriers underscores that counseling for BRCA1 and BRCA2 mutation carriers regarding the choice between risk-reducing mastectomy and surveillance might be tailored, although confirmation in a larger cohort of especially BRCA2 mutation carriers is warranted.

To the best of our knowledge, this is the first cohort study comparing BRRM with surveillance with respect to survival in healthy BRCA1 and BRCA2 mutation carriers separately. Previous investigations have shown that BRRM effectively reduces BC risk [7‐12, 24, 25], but convincing data regarding survival after BRRM in BRCA1/2 mutation carriers are scarce and mainly derived from modeling studies. Using a simulated cohort and Markov modeling of outcomes, Grann et al. estimated that BRRM plus risk-reducing salpingo-oophorectomy at the age of 30 may extend survival by 4.9 years over surveillance alone [26]. Further, Sigal et al. yielded from their Monte Carlo simulation model gains in life expectancy after BRRM plus risk-reducing salpingo-oophorectomy varying from 6.8 to 10.3 for BRCA1 and 3.4 to 4.4 years for BRCA2 mutation carriers [15]. Recently, Giannakeas and Narod showed in a simulated cohort that for BRCA mutation carriers who underwent bilateral mastectomy at the age of 25, the probability of being alive at age 80 increased by 8.7% [27]. In addition, in an exploratory study in unaffected BRCA1/2 mutation carriers and untested female first-degree relatives, Ingham et al. showed overall survival benefit of ~ 10% after risk-reducing surgery [28]. However, this study is not directly comparable to the current study since the authors compared three groups of women undergoing risk-reducing surgery (i.e., BRRM only, risk-reducing salpingo-oophorectomy only, or both) with women without any surgery, while we currently incorporated undergoing risk-reducing salpingo-oophorectomy (yes/no) in the model. In our opinion, this better reflects daily practice: as a result of directive counseling due to ineffective screening protocols for early ovarian cancer detection, the uptake of risk-reducing salpingo-oophorectomy is high for both women undergoing BRRM (~ 75%) and women not (yet) opting for BRRM (~ 60%).

For BRCA1 mutation carriers under surveillance, BC and ovarian cancer were the main causes of death. The high percentage of ovarian cancer deaths in this group—which was similar to that of BC deaths—emphasizes the need for RRSO for BRCA mutation carriers. While in the surveillance group 20 out of 990 women (2.0%) died due to BC, only one out of 722 women (0.1%) died from BC after BRRM. The latter patient was identified with a BRCA1 mutation at the age of 38, and underwent BRRM 1 year later (in 2007). At the age of 42, she was diagnosed with a triple-negative BC with lung metastases, and died 1 year later. This emphasizes—in addition to the fact that eight BCs occurred 4.4 median years after BRRM in the current cohort—that BRRM does not fully protect against the occurrence of BC and BC-related death.

Of the 29 deceased BRCA2 mutation carriers in the surveillance group, 24% died of BC, 59% of another malignancy—including two deaths due to ovarian cancer and seven due to pancreatic cancer—and 17% died of nonmalignancy-related causes. The higher numbers of non BC-related deaths in the surveillance group seem to be coincidental, but may explain the higher overall mortality rate though comparable breast cancer-specific mortality rate among BRCA2 mutation carriers under surveillance.

In BRCA2 mutation carriers, we observed no BCs and no BC-related deaths after BRRM versus 144 BC cases and seven BC-related deaths in the surveillance group, suggesting a maximal risk-reduction of developing BC and dying due to BC after BRRM. However, the absolute breast cancer-specific survival benefit at the age of 65 was minimal (2%), partly due to the low BC-specific mortality in the surveillance group (i.e., 0.9 per 1000 person-years of observation). The latter can be explained by the observation that BRCA2-associated BCs were diagnosed with more favorable characteristics, i.e., diagnosed at older age, more often in situ, better differentiated, and less often showing a triple-negative phenotype—than BRCA1-associated BCs. This supports previous suggestions that BRCA2-associated BC patients face a better prognosis than BRCA1-associated BC patients [10, 16]. The current results suggest that regarding breast cancer-specific mortality, BC surveillance may be a reasonable and balanced alternative to BRRM for BRCA2 mutation carriers.

The main strengths of the current study are (1) the sufficient numbers of BRCA1 and BRCA2 mutation carriers allowing analyses for both groups separately, (2) with long enough follow-up, and (3) the availability of data on cause of death, enabling to specifically address the ultimate goal for BRRM, i.e., breast cancer-specific survival.

This study also has limitations. First, information regarding BC screening modality and frequency was derived from self-reported data, and unknown for ~ 50% of the women in the surveillance groups. However, we do know that all women had been counseled by clinical geneticists and were aware of an identified BRCA mutation at the start of the observation period. Therefore, we assume that the vast majority of the women did participate in a BC surveillance program for high-risk women according to Dutch guidelines. This assumption is supported by the experience from the Rotterdam Family Cancer Clinic that after being positively tested for a pathologic mutation in one of the BRCA genes, 97% of the mutation carriers is yearly screened; 79% of the mutation carriers are yearly screened by both MRI and mammography, 11% by MRI only (aged < 30 years), and 7% by mammography only (aged > 60 years). Only three percent of the proven mutation carriers seem not to attend the national screening program for BRCA1/2 mutation carriers, or are screened in another hospital (unpublished data). These numbers are in line with recently reported international trends in the uptake of cancer screening among BRCA1/2 mutation carriers [29].

Still, if the BC patients with unknown screening status were not under BC surveillance, BCs consequently would be diagnosed at a more advanced stage with worse prognosis. As a result, the observed number of BC-related deaths in the surveillance group could be an overestimation of the actual number of BC-related deaths under surveillance, and a potential breast cancer-specific survival benefit may be overestimated. However, BCs occurring among BRCA1/2 mutation carriers in the surveillance group with unknown screening status showed in fact slightly more favorable characteristics (i.e., more often in situ and smaller than two centimeters; see Supplementary Table S2) than the patients with known screening status. In addition, the absolute number of women dying from BC was lower among the women with unknown screening status: 8 out 864 (0.9%) versus 19 out of 865 (2.2%) among the women with known screening status (P value 0.033; Supplementary Table S2). Thus, it seems plausible that the majority of the women with unknown screening status were actually under BC surveillance, and an overestimation of the observed breast cancer-specific survival is unlikely.

Thirdly, there might be some bias toward BRRM being offered more often to healthier women. This could be supported by the fact that BRCA2 mutation carriers in the surveillance group show more other cancers (i.e., no BC or ovarian cancer) than those in the BRRM group (9% vs. 6%, P = 0.048; Table 1). However, we did not observe this difference for BRCA1 mutation carriers, where the incidence of other tumors was 7% for both groups. In addition, the median age at diagnosis of cancer other than BC or ovarian cancer is higher in the surveillance group than in the BRRM group (both for BRCA1 and BRCA2 mutation carriers), suggesting that with longer follow-up—and thus growing age—the numbers of patients with other tumors could increase. Unfortunately, data about health-related issues such as weight and past and current smoking habits are not available for the current cohort.

In conclusion, BRRM was associated with lower overall and breast cancer-specific mortality rates than surveillance for BRCA1 mutation carriers. For BRCA2 mutation carriers, BRRM may lead to similar breast cancer-specific survival as surveillance. The latter is most probably due to the more favorable characteristics of BRCA2-associated BCs. Therefore, for BRCA2 mutation carriers BC surveillance may be as effective as BRRM regarding breast cancer-specific survival. Although the number of events are small—especially for the analyses on breast cancer-specific mortality—our findings may support a more individualized counseling based on BRCA mutation type regarding the difficult choice between BRRM and BC surveillance.