|Home | About | Journals | Submit | Contact Us | Français|
Women with BRCA1/2 mutations inherit high risks of breast and ovarian cancer; options to reduce cancer mortality include prophylactic surgery or breast screening, but their efficacy has never been empirically compared. We used decision analysis to simulate risk-reducing strategies in BRCA1/2 mutation carriers and to compare resulting survival probability and causes of death.
We developed a Monte Carlo model of breast screening with annual mammography plus magnetic resonance imaging (MRI) from ages 25 to 69 years, prophylactic mastectomy (PM) at various ages, and/or prophylactic oophorectomy (PO) at ages 40 or 50 years in 25-year-old BRCA1/2 mutation carriers.
With no intervention, survival probability by age 70 is 53% for BRCA1 and 71% for BRCA2 mutation carriers. The most effective single intervention for BRCA1 mutation carriers is PO at age 40, yielding a 15% absolute survival gain; for BRCA2 mutation carriers, the most effective single intervention is PM, yielding a 7% survival gain if performed at age 40 years. The combination of PM and PO at age 40 improves survival more than any single intervention, yielding 24% survival gain for BRCA1 and 11% for BRCA2 mutation carriers. PM at age 25 instead of age 40 offers minimal incremental benefit (1% to 2%); substituting screening for PM yields a similarly minimal decrement in survival (2% to 3%).
Although PM at age 25 plus PO at age 40 years maximizes survival probability, substituting mammography plus MRI screening for PM seems to offer comparable survival. These results may guide women with BRCA1/2 mutations in their choices between prophylactic surgery and breast screening.
More than 300,000 women in the United States are estimated to carry a mutation in the BRCA1 or BRCA2 genes1; with these mutations, women inherit 5- to 20-fold increased risks of developing breast and ovarian cancer.2 Cancer risk management strategies for BRCA1/2 mutation carriers incorporate earlier, more frequent, and more invasive intervention than those for the general population. Practice guidelines recommend prophylactic bilateral salpingo-oophorectomy (PO) for ovarian cancer risk reduction by age 40 years (all ages are given in years hereafter) and support various alternatives for managing breast cancer risk: either prevention with prophylactic mastectomy (PM) or early detection with screening mammography plus breast magnetic resonance imaging (MRI).3,4
Although studies in BRCA1/2 mutation carriers have reported efficacy of PO and PM for cancer prevention5–9 and of mammography plus MRI screening for early breast cancer detection,10–13 most were limited in size, and none directly compared survival after screening versus surgery. Moreover, all risk-reducing options have disadvantages. PM and premenopausal PO are permanent procedures that limit reproductive choices; PM can impair body image, and PO may impose health risks of early menopause.14–18 Screening MRI yields frequent false positives, thereby increasing anxiety and costs.19 To the best of our knowledge, no randomized trial is planned to compare survival benefit and other health outcomes from different strategies aimed at reducing cancer mortality in BRCA1/2 mutation carriers. Moreover, patients may not accept random assignment between prophylactic surgery and screening, and many clinicians question the ethics of asking them to do so. Therefore, women and their physicians must navigate among disparate and invasive alternatives with little empiric guidance when choosing how best to manage cancer risks. To optimally inform such decisions, we developed a simulation modeling approach to compare the survival probability for BRCA1/2 mutation carriers after several risk-reduction strategies, based on the best available evidence. This approach builds on our prior evaluation of the cost-effectiveness of MRI screening in BRCA1/2 mutation carriers19 and additionally considers PO and PM as alternatives or complements to MRI-based breast screening. We now report survival probability with various clinically relevant risk-reducing strategies for women with BRCA1/2 mutations, aiming to target survival estimates to individual patients and enable personalized cancer risk management.
We used a computer simulation model that integrates empiric data from the literature (Table 1) to estimate survival probability and causes of death at ages 70 and 80 years for 25-year-old women with a BRCA1 or BRCA2 mutation. Risk-reducing interventions were modeled alone and in combination, at ages specified by cancer care guidelines3,4: breast screening with mammography plus MRI started at age 25 and continued annually to age 69, PO was performed at age 40 or 50, and PM was evaluated for ages 25 to 50 years.
We previously developed a Monte Carlo simulation model to analyze the impact of screening and treatment on outcomes of individual breast cancer patients.42,43 We then modified the model to incorporate breast and ovarian cancer incidence, tumor characteristics, prognosis under standard treatments,2,24,37,39,44–47 and the performance of screening with mammography and MRI10–13 in BRCA1/2 mutation carriers.19 For this study, we added literature-based estimates of the effects of PM and PO on survival probability and causes of death (Table 1).
The model simulates the life histories of a 1980 birth cohort of 1,000,000 female BRCA1 or BRCA2 mutation carriers from age 25 until age 100 or death, whichever occurs first. We extrapolated cancer risks for BRCA1 and BRCA2 mutation carriers from meta-analyses.2,20 Since approximately 30% of BRCA1/2 mutation carriers undergo PO at a mean age of 45,7 we assumed that reported breast cancer incidence2 incorporates 30% use of PO. We adjusted our estimates of breast cancer risk without PO, given that PO in premenopausal BRCA1/2 mutation carriers reduces breast cancer risk by approximately 50%.6–8,25 We assumed that BRCA1/2 mutation carriers receive standard breast and ovarian cancer therapies and that treatment efficacy and cancer prognosis equal those of the general population39,45–48 (Table 1).
We assumed that PM reduces breast cancer risk by 90%.9,24 We assumed that PO reduces ovarian cancer risk by 85%7,8 and breast cancer risk by 50% when performed between ages 40 and 50, with no impact on breast cancer risk when performed at or after age 50 (Table 1).6,49 We assumed that the breast cancer risk reduction after premenopausal oophorectomy persists indefinitely.25
We assumed no use of menopausal hormone therapy, given uncertainty about its effect on breast cancer in BRCA1/2 mutation carriers.50–52 After PO before age 50,3 we assumed a two-fold increased risk of cardiovascular disease40 and a 50% increased risk of osteoporotic hip fracture and dementia16,17,41; as have previous studies,53,54 we assumed these increases would last until age 65. To compute other-cause mortality, we adjusted data from the Berkeley Mortality Database 1980 birth cohort table55 by removing breast and ovarian cancer deaths based on 2004 US rates reported by the Centers for Disease Control and Prevention (diagnostic codes C50 and C56)56 and adjusting mortality rates from cardiovascular disease, dementia, and hip fracture (codes I20-I25, F0, F3, G20-G21, and G30)54,56 according to the assumed relative risks (Table 1).
We varied model parameters about which significant uncertainty exists: risk of developing breast and ovarian cancer with a BRCA1 or BRCA2 mutation, breast tumor volume doubling time (TVDT), sensitivity of MRI for cancer detection, impact of oral contraceptive pills on breast and ovarian cancer risk, and impact of premenopausal PO on breast cancer and other-cause mortality (Table 1).2,6,8,17,20–23, 26–36,38,49 Sensitivity analyses considered four clinically relevant scenarios: no intervention, mammography plus MRI screening without surgery, screening plus PO at age 40, and screening plus PO and PM at age 40.
In the absence of breast screening, PM at age 40 reduces breast cancer risk to 27%; PO at age 40 reduces breast cancer risk to 49% and ovarian cancer risk to 9%. Performing PM and PO at age 40 reduces breast cancer risk to 25% and ovarian cancer risk to 9%. Screening has no impact on ovarian cancer risk and minimal impact on breast cancer risk (data not shown).
With no intervention, survival probability by age 70 years is 53% for BRCA1 mutation carriers versus 84% for the general US population (Table 2). The most effective single intervention is PO at age 40, yielding a survival probability of 68% by age 70, which represents a 15% absolute gain compared with no intervention (68% v 53%). Delaying PO to age 50 yields half the survival gain provided by PO at age 40 (8%: 61% v 53% with no intervention). In comparison, PM at age 25 yields a 13% gain relative to no intervention, whereas delaying PM to age 40 yields a small (2%) decrement in gain compared with PM at age 25. Breast screening alone from ages 25 to 69 yields the lowest gain (6%).
The most effective combination strategy is PM at age 25 plus PO at age 40, providing a 26% survival gain by age 70 compared with no intervention (79% v 53%). Postponing PM until age 40, in the presence of screening from ages 25 to 39 and PO at age 40, reduces survival gain by 2%. Eliminating PM and substituting breast screening from ages 25 to 69 while performing PO at age 40 reduces survival gain by an incremental 3%. When added to PO at age 40, breast screening offers 5% lower survival probability than does PM at age 25 (74% v 79%) and 3% lower survival probability than does PM at age 40 (74% v 77%). If PO is delayed until age 50, breast screening offers 5% lower survival probability than PM at age 40 (69% v 74%). Results by age 80 are similar (Table 2). Figure 1A presents survival probability, and Figure 2A presents distribution of health status by age 70 years in BRCA1 mutation carriers under various intervention scenarios.
Among BRCA1 mutation carriers who choose no intervention, the likelihood of death from breast versus ovarian cancer is similar (41% v 36%, conditional on death by age 70; Table 2). With PO at age 40 only, death from ovarian cancer decreases dramatically, making breast cancer deaths most frequent (45%, followed by other-cause [43%] and ovarian cancer deaths [12%]). When PO is delayed from age 40 to age 50, ovarian cancer deaths nearly double; breast cancer deaths increase less markedly. Among women who choose breast screening until age 40 and then PO plus PM at age 40, 23% will die by age 70; most (64%) die of non-cancer causes, followed by breast cancer (18%) and ovarian cancer (18%). Results are comparable for women who choose breast screening until age 69 and then PO at age 40, but not PM, and follow a similar pattern by age 80 (Table 2).
In the absence of breast screening, PM at age 40 reduces breast cancer risk to 14%; PO at age 40 reduces breast cancer risk to 31% and ovarian cancer risk to 2%. Performing PM and PO at age 40 reduces breast cancer risk to 13% and ovarian cancer risk to 2%. Screening has no impact on ovarian cancer risk and minimal impact on breast cancer risk (results not shown).
With no intervention, survival probability by age 70 is 71% for BRCA2 mutation carriers versus 84% for the general population (Table 2). The most effective single intervention is PM at age 25, yielding an 8% gain compared with no intervention (79% v 71%); postponing PM to age 40 reduces gain by 1%. In comparison, PO at age 40 yields a 6% gain relative to no intervention, and breast screening alone, with annual MRI plus mammography, provides a 4% gain. Delaying PO from age 40 to age 50 reduces gain by 2%.
The most effective combination strategy is PM at age 25 plus PO at age 40, providing a 12% survival gain by age 70 compared with no intervention (83% v 71%). Postponing PM until age 40 in the presence of breast screening from ages 25 to 39 and PO at age 40 reduces survival gain by 1%. Eliminating PM and substituting breast screening from ages 25 to 69 while performing PO at age 40 reduces survival gain by an incremental 2%. In the presence of PO at age 40, screening yields 3% lower survival than does PM at age 25 (80% v 83%), and 2% lower survival than does PM at age 40 (80% v 82%; Table 2). If PO is delayed to age 50, breast screening offers 4% lower survival probability than PM at age 40 (79% v 83%). Results by age 80 are similar (Table 2). Figure 1B presents survival probability and Figure 2B presents distribution of health status by age 70 years in BRCA2 mutation carriers under various intervention scenarios.
Among BRCA2 mutation carriers who choose no intervention, more die from breast than ovarian cancer (36% v 20%, conditional on death by age 70), but non-cancer deaths are more frequent (44%; Table 2). With PO at age 40 only, ovarian cancer deaths decrease substantially; other-cause deaths remain most frequent (66%), followed by death from breast cancer (30%) and ovarian cancer (4%). Delaying PO from age 40 until age 50 yields little increase in ovarian cancer deaths (2% to 3%) but a larger increase in breast cancer deaths (12%). Among women who choose breast screening until age 40 and then PO plus PM at age 40, 18% will die by age 70; most (85%) die of non-cancer causes, followed by breast cancer (9%) and ovarian cancer (6%). Results are comparable for women who choose breast screening until age 69 and then PO at age 40, but not PM, and follow a similar pattern by age 80 (Table 2).
We observed the largest differences in overall survival with variation in assumptions about breast cancer risk (9% to 10%), TVDT (4% to 6%), and hazard ratio for breast cancer after premenopausal oophorectomy (3% to 6%). Variation in these three parameters also yielded the greatest differences in breast cancer–specific death (13% to 22%). Ovarian cancer–specific deaths varied most with changes in breast and ovarian cancer risk and TVDT (4% to 20%). Other-cause deaths varied most with breast cancer risk, TVDT, and the hazard ratio for breast cancer after premenopausal oophorectomy in BRCA1 mutation carriers (10% to 16%) and with breast and ovarian cancer risk and TVDT in BRCA2 mutation carriers (9% to 15%). Variations in other model parameters yielded smaller differences in overall and cause-specific survival (Table 3).
We developed a Monte Carlo model to simulate and compare different strategies for reducing cancer mortality in BRCA1/2 mutation carriers. The most effective strategy is PO at age 40 plus PM at age 25; for BRCA1 mutation carriers, this approach substantially improves survival by age 70 (79% v 53%, with no intervention), while for BRCA2 mutation carriers, the absolute increase is smaller (83% v 71%) because of their lesser cancer risks. We evaluated a delay in PO until age 50, which is 10 years later than recommended by current practice guidelines3 but which may appeal to women because it approximates the age of natural menopause. In BRCA1 mutation carriers, delayed PO provides half the survival gain of PO at age 40 (8% at 50 v 15% at 40), whereas for BRCA2 mutation carriers, delaying PO makes less difference (4% v 6%). For both BRCA1 and BRCA2 mutation carriers, combining PO at age 40 with PM at age 25 provides survival approaching that of women without mutations (79% for BRCA1, 83% for BRCA2, and 84% for the general population); however, postponing PM until age 40, when it may prove more acceptable than at age 25, reduces survival gain by only 1% to 2%. Most notably, we found that replacing PM with MRI-based breast screening in the presence of PO at age 40 yields only a 3% to 5% decrement in survival. Approximately 36% of US BRCA1/2 mutation carriers now choose PM, whereas 24% undergo breast screening incorporating MRI.57 Our finding that mammography plus MRI screening offers survival probability comparable to that of PM may alter women's choices between these options. In the related field of breast cancer treatment, some women will accept adjuvant chemotherapy for an anticipated survival improvement of 5% or less, whereas others consider its side effects too morbid for such a small gain58,59; current research focuses on targeting chemotherapy to women with larger estimated benefits.60
Many prior model-based analyses have addressed cancer risk management in BRCA1/2 mutation carriers.19,53,61–65 Most concluded that prophylactic surgeries improve life expectancy62,64,65; however, few19,63 considered recent improvements in cancer detection with breast MRI.10–13,66 Our study represents an advance because we directly compared prophylactic surgery with screening, incorporating an updated understanding of the options available to high-risk women. We modeled the effect of MRI screening on cancer detection and prognosis, on the basis of the tumor grade, growth rate, and hormone receptor profiles in BRCA1/2 mutation carriers.10,37,38,44,67 We incorporated data on age-specific breast cancer risk reduction after premenopausal PO6,8 and considered the use of oral contraceptives,26–36 which most prior analyses did not. However, given controversy about chemoprevention with tamoxifen and raloxifene in BRCA1/2 mutation carriers,57,68–70 we did not model them. Although some practice guidelines recommend ovarian cancer screening with transvaginal ultrasound and CA125 in BRCA1/2 mutation carriers who do not undergo PO,3 we chose not to model this strategy, given the lack of compelling evidence that it impacts survival or other health outcomes.
As with all modeling studies, our results depend on our assumptions. We performed sensitivity analyses on all major parameters, including BRCA1 and BRCA2 mutation penetrance, the growth patterns of BRCA1/2-associated breast cancers and their detectability by screening, the impact of oral contraceptive use on breast and ovarian cancer risk, and the effect of premenopausal PO on breast cancer and other health outcomes. In sensitivity analyses, none of these factors dramatically affected the ranking of interventions but did alter our absolute estimates of overall and cause-specific survival by up to 22%. Our assumptions about breast cancer risk, breast TVDT, and the risk of breast cancer after PO at age 40 were most influential. If BRCA1/2 mutation carriers have higher breast cancer risks than reported by large meta-analyses,2,20 have more interval breast tumors than reported in MRI screening studies,10–13 or gain less benefit than reported from PO at age 40,6,49 then the survival difference between PM and MRI-based breast screening increases, although by a relatively small amount: from 3% up to 8% for BRCA1 and from 2% to 3% for BRCA2 mutation carriers.
Many uncertainties remain about the clinical management of BRCA1/2 mutation carriers. Important questions include the impact of menopausal hormone therapy after premenopausal PO on breast cancer and other health outcomes, the efficacy of nipple or skin-sparing techniques compared with simple prophylactic mastectomy, the potential risk of breast cancer due to mammography, and the prognosis of BRCA1/2-associated cancers as targeted therapies emerge.50,52,71–73 Answers to these questions may alter judgments about the relative efficacy and tolerability of different risk-reducing strategies and better inform future decision analyses.
Model-based analyses cannot replace empirical studies but can address clinically important questions that are poorly amenable to randomized trials. Given the complex, personal nature of decisions about prophylactic surgery, women will not likely accept random assignment between PM and breast screening; therefore, direct evidence about survival differences will remain elusive. Our analysis aims to enhance patient care by bridging the evidence gap: we provide a computer model that integrates the best available data, permitting recommendations calibrated to the variable effects of risk, age, intervention efficacy, and personal preferences. Individual women make widely disparate choices about how to manage their cancer risks, depending on their family history, health care access, reproductive concerns, and concurrent diagnoses.18,47,57,74–76 Our results can anchor such choices quantitatively, helping a woman weigh strategies that yield small differences in survival, yet potentially larger differences in physical and emotional effects, according to her preferences. Computer-based decision support tools are now widely used to assist patients' cancer treatment choices.58,77 Our model may similarly facilitate shared decision making, guiding women with BRCA1/2 mutations toward better-informed choices between prophylactic surgery and screening alternatives.
See accompanying editorial on page 189
Supported by Grants No. R01 CA829040, U01 CA088248, and R01 CA66785 from the National Institutes of Health; by Stanford University Cancer Center 2007 Developmental Research Award in Population Sciences; and by Robert Wood Johnson Foundation 2008 Physician Faculty Scholars Award (64317).
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
The author(s) indicated no potential conflicts of interest.
Conception and design: Allison W. Kurian, Sylvia K. Plevritis
Financial support: Allison W. Kurian, Sylvia K. Plevritis
Provision of study materials or patients: Sylvia K. Plevritis
Collection and assembly of data: Bronislava M. Sigal, Sylvia K. Plevritis
Data analysis and interpretation: Allison W. Kurian, Bronislava M. Sigal, Sylvia K. Plevritis
Manuscript writing: Allison W. Kurian, Bronislava M. Sigal, Sylvia K. Plevritis
Final approval of manuscript: Allison W. Kurian, Bronislava M. Sigal, Sylvia K. Plevritis