What have we learned from risk-reducing salpingo-oophorectomy?

In this issue of the Journal, Rebbeck et al. (1) have provided us with a meticulously executed, up-to-date meta-analysis that quantifies the reductions in the risks of breast and ovarian/fallopian tube cancer that follow risk-reducing salpingo-oophorectomy (RRSO) in BRCA1 and BRCA2 (BRCA1/2) mutation carriers. Their conclusion — " The summary risk reduction estimates presented here indicate that RRSO is strongly associated with reductions in the risk of breast and ovarian/fallopian cancers, and should provide guidance to women in planning cancer risk reduction strategies " — is fully supported by their analysis and will not surprise those who have followed the rapidly evolving management strategies for BRCA1/2 mutation carriers. From our perspective, four features commend this study. First, the meta-analysis that forms the basis of these conclusions was carefully designed and thoughtfully implemented. The authors identifi ed and included all relevant publications, used state-of-the-art statistical methods, excluded reports with overlapping study populations, evaluated potential interstudy heterogeneity, and tested for publication bias. Importantly, Rebbeck et al. attempted to disentangle potential differences between BRCA1 and BRCA2 mutation carriers who, despite having superfi cial similarities with regard to phenotype, have important biological differences that render a combined analysis suboptimal. Consequently, one can have confi dence in their fi ndings, particularly because they detected no evidence of heterogeneity in effect size among the 10 pooled studies, despite substantial heterogeneity in study designs. The lack of statistical heterogeneity provides considerable reassurance that the pooled estimates are accurate. The second valuable report feature, therefore, is its provision of improved breast and ovarian/fallopian tube cancer risk reduction estimates, to better inform the patient and provider about RRSO-related decision making. The meta-analysis gave point estimates of risk that are more precise and 95% confi dence intervals surrounding those estimates that are substantially narrower than those based on any of the individual studies. For example, when viewed separately, the published studies are consistent with as much as an 85% (the lowest 95% confi dence interval boundary) reduction in breast cancer risk in BRCA1 mutation carriers and up to a 22% (the highest 95% confi dence interval boundary) increase in risk [table 2 in Rebbeck et al. (1)]. Although there is no biological basis for anticipating an increase in breast cancer risk after RRSO, studies in which the 95% confi dence interval includes 1.0 are statistically compatible with no risk reduction. When the studies were pooled, the range became much tighter …