Better Life Expectancy in Women with BRCA2 Compared with BRCA1 Mutations Is Attributable to Lower Frequency and Later Onset of Ovarian Cancer

Purpose: No formal assessment of life expectancy in women with BRCA1 and BRCA2 mutations in these genes has been reported previously. We have evaluated life expectancy using actuarial analysis and assessed the effect of breast and ovarian cancers on premature death in >1,000 BRCA1/2 carriers.

Methods: Families with pathogenic mutations in BRCA1 and BRCA2 have been ascertained in a 10-million population region of United Kingdom since 1996. Mutation carriers and their first-degree relatives were used in an analysis of breast and ovarian cancer incidence and mortality as well as to derive and compare an actuarial assessment of life expectancy.

Results: Six hundred twelve BRCA1 and 482 BRCA2 female mutation carriers were identified from 482 families. Life expectancy was significantly reduced for BRCA1 carriers compared with BRCA2 (P = 0.0002). This effect was attributable to an increased death rate from ovarian cancer (P = 0.04). Kaplan-Meier analysis revealed a better long-term survival from early-stage ovarian cancer in BRCA2 carriers but no significant differences in deaths from breast cancer or from women presenting with late-stage ovarian cancer. There was no other major contributing cause to death other than breast/ovarian cancer in BRCA1/2 female carriers.

Conclusion: Interventions to reduce ovarian cancer incidence are likely to have a greater effect on life expectancy in BRCA1 compared with BRCA2 carriers. (Cancer Epidemiol Biomarkers Prev 2008;17(6):1535–42)

Although it has been assumed that life expectancy is reduced in female BRCA1 and BRCA2 carriers, no direct actuarial analysis has been undertaken to date. Decision analyses have been widely published, which have assessed the potential life years gained by interventions such as risk-reducing mastectomy and chemoprevention with tamoxifen (1, 2). These analyses were, however, based on assumptions about the incidence and survival rates from breast and ovarian cancer rather than a direct analysis of mutation carriers.

Mutations in the BRCA1/BRCA2 genes account for about 2% to 3% of breast cancer and 8% of ovarian cancer in women (3). About 1 in 500 individuals in most Western countries carries pathogenic mutations in one or other gene (3, 4), although this increases to approximately 1 in 40 in the Ashkenazi Jewish population (5). The cumulative risks of developing either breast or ovarian cancer in carriers depend on how an individual and family are ascertained. Breast cancer risks in high-risk familial breast cancer kindreds with BRCA1/BRCA2 mutations are substantially higher than the risks derived from population-based studies (3-11). In high-risk clusters (Breast Cancer Linkage Consortium data), BRCA1 and BRCA2 mutations were estimated to cause a cumulative lifetime risk of breast cancer by age 70 years of 85% to 87% and 77% to 84%, respectively (3, 8). However, estimates of breast cancer risks to age 70 years from data from population-based studies are much lower at 28% to 60% (5, 9-11) for BRCA1 and even lower for BRCA2. The risk of ovarian cancer is lower and occurs later in BRCA2 carriers. The risk of ovarian cancer in BRCA1 carriers is 25% to 60% by age 70 years (3, 7, 9), whereas risks in BRCA2 carriers are 10% to 27% (3, 9).

We have assessed and compared the actuarial life expectancy of women from 482 high-risk BRCA1/2 families and the attributable early death rate from breast/ovarian cancer. We have also evaluated ovarian cancer–related survival in 292 ovarian cancer patients from these families since 1975.

Mutation screening for BRCA1/2 has been undertaken since 1996 in the overlapping regions of Manchester and Birmingham (mid-north United Kingdom) covering a population of 10 million. Women who attend specialist genetic clinics in these two regions with a family history of breast and/or ovarian cancer have detailed three-generation family trees elicited. All cases of breast or abdominal cancers are confirmed by means of hospital/pathology records, from Regional Cancer Registries (data available from 1960), or death certification. Mutation analysis has been initiated with as little as one patient with breast cancer <35 y, two <50 y, or with a family with at least a single patient with ovarian cancer and an additional one with breast or ovarian cancer at any age. Once a family-specific pathogenic BRCA1/2 mutation is identified, predictive testing is offered to all blood relatives. Where possible, within these families, all affected women with breast or ovarian cancer are tested to establish the true extent of BRCA1/2 involvement in the family. This may mean identifying intervening deceased relatives who would have to have carried the mutation (obligate carriers). Mutation carriers and their close relatives are offered regular follow-up through the regional genetic service and dates of last follow-up were obtained from the clinical genetics notes. Dates of death were obtained from the cancer registry or from death certification. Kaplan-Meier curves were derived to assess overall life expectancy as well as breast and ovarian cancer incidence in proven mutation carriers. Incidence curves were censored at date diagnosis, death, or last follow-up and at risk-reducing mastectomy for breast cancer and at risk-reducing oophorectomy for ovarian cancer. These curves were also generated from date of diagnosis and censored at date of death or date of last follow-up for both breast and ovarian cancers. For ovarian cancer, only cases with a confirmed diagnosis from medical records or cancer registration after January 1, 1975 when platinum-based therapies became available were included in the survival analysis. The analysis was also repeated censoring for a prior or subsequent breast cancer. For survival from diagnosis, we included all first-degree relatives (FDR) with breast/ovarian cancer to correct for the survival bias of someone needing to be alive to be tested. All Kaplan-Meier statistics were based on log-rank scores with one to three degrees of freedom.

A total of 482 families were identified with BRCA1/2 mutations (263 BRCA1 and 219 BRCA2) from over 2,200 families screened for mutations. The 482 of 2,200 (25% detection rate) mutations were mainly in breast cancer families, as only 500 (23%) contained ovarian cancer, the remainder being breast cancer only. There were 1,094 proven female mutation carriers: 612 BRCA1 carriers and 472 BRCA2 carriers (Table 1). One hundred four of 612 (16%) BRCA1 carriers and 60 of 482 (12%) for BRCA2 were “obligate” deceased carriers. The birth date ranged from 1885 to 1985 (median, 1949) for BRCA1, 1890 to 1985 (median, 1950) for BRCA2, and 1907 to 1986 (median, 1956) for noncarriers. Dates of death were from 1937 to 2007 (median, 1993) for BRCA1 and 1938 to 2007 (median, 1995) for BRCA2. There was no difference in the number of female carriers per family (2.3 for BRCA1 and 2.2 for BRCA2). At the time of analysis, 64 unaffected mutation carriers had undergone risk-reducing mastectomy [37 of 145 (26%) BRCA1; 27 of 131 (21%) BRCA2] and 78 had undergone risk-reducing oophorectomy [53 of 145 (37%) BRCA1; 25 of 131 (21%) BRCA2]. Only 3 of 152 (2%) of risk-reducing surgeries had been carried out before 1990 and 114 of 152 (75%) had been undertaken within the last 10 years. It is too early therefore for these to have appreciably affected the results. Information was also available on a further 1,267 female FDRs of unknown mutation status. We estimate for the Manchester region that we have ascertained 70% of female carriers in the region through the identified families using a combined birth incidence rate of 1 per 1,000 (3).

Two hundred forty-eight of 612 (41%) of the identified BRCA1 carriers have died, of which 223 (90%) died before 70 years. Of those who died before 70 years, 58% were due to ovarian cancer, 38% to breast cancer, and 3% to other causes. In contrast, only 156 of 482 (32%) of BRCA2 carriers had died, with 131 (84%) dying before age 70. Of these deaths, 21% were due to ovarian cancer, 74% to breast cancer, and 4% to other causes (Table 2). As such, there was no other major contributing cause to death other than breast/ovarian cancer in BRCA1/2 female carriers. In analyzing actuarial survival from birth, there was a significantly greater chance of death in BRCA1 carriers before 70 years of age (P = 0.0002; Fig. 1).

We have not included an adjustment for untested relatives. Among FDRs of a mutation carrier, only 1 of 111 (0.9%) of ovarian cancer tested negative for the family mutation. It could therefore be assumed that all FDRs with ovarian cancer were carriers. However, 28 of 283 (10%) of breast cancer FDRs tested negative, so no such assumption could be made for breast cancer cases. Among unaffected female FDRs tested >55 years of age, only 5 of 48 (10%) BRCA1 and 14 of 55 (25%) of BRCA2 carriers tested positive (P = 0.05). Among untested FDRs, 219 BRCA1 females and 132 BRCA2 females had died (P = 0.002). This amounted to 26.5% of untested BRCA1 women and 18% of BRCA2 women dying by 50 years of age (P < 0.001), similar to the 22% and 17.5% in proven carriers (Fig. 2). In contrast to proven carriers, 17.5% of deaths [18 (3.5%) below 50 years] were not cancer related and 13% were related to other cancers besides breast/ovarian cancer. Nonetheless, most of these breast/ovarian cancer deaths are likely to have occurred in noncarriers.

Although the proportion of breast cancer deaths in BRCA2 mutation carriers is higher, there was no statistical difference in breast cancer survival from diagnosis between BRCA1- and BRCA2-related breast cancer (Fig. 2). The curves were almost identical when censoring for ovarian cancer diagnosis. Both the incidence and cumulative risk of breast cancer were very similar within the two groups, with a predicted risk of 78% to 83% by age 70 years. The mean age at diagnosis of breast cancer was 42.8 years (median, 41.4 years; range, 20.7-78.2) for 345 proven BRCA1 carriers and 44.75 years (median, 44 years; range, 21.3-89.2) for 310 BRCA2 carriers with breast cancer (P = 0.26; Fig. 3).

The ovarian cancer incidence was lower and at a later age for BRCA2 with a mean age of diagnosis of 50.3 years (median, 50.2 years; range, 30.1-76.2) in 178 proven BRCA1 carriers and 55.1 years (median, 54 years; range, 32.4-78.4) in 60 BRCA2 carriers (P < 0.0001). The cumulative risk of ovarian cancer, to 70 years, was 34% and 57% in BRCA2 and BRCA1 carriers, respectively (Fig. 4). In total, 482 known mutation families contained a total of 434 cases of ovarian cancer, with 314 (78%) occurring in BRCA1 families and 120 (28%) in BRCA2 families. There were 238 proven mutation carriers with ovarian cancer, of which 178 had BRCA1 mutations and 60 had BRCA2 mutations. If we assume that the remaining unknown FDRs with ovarian cancer were mutation carriers (only 1 of 111 tested negative; Table 3), a further 97 BRCA1 and 40 BRCA2 carriers with ovarian cancer would be available for analysis. Thus, a total of 275 BRCA1 ovarian cancers and 100 BRCA2 ovarian cancers could be analyzed for survival from diagnosis. At the time of last follow-up, 46 of 275 (17%) of BRCA1 ovarian cancers and 29 of 100 (29%) of BRCA2 ovarian cancer patients were alive (P < 0.05).

Survival data from diagnosis are shown in Fig. 5 for BRCA1/2 carriers diagnosed since January 1, 1975, including FDRs with unknown mutation status. A survival advantage seems to be shown for those patients carrying (or assumed to be carrying) a BRCA2 mutation, although this was not significant (5-year/10-year survival: BRCA1, 29%/19%; BRCA2, 49%/30%; P = 0.06). When analyzed only on proven mutation carriers (excluding FDRs of unknown status), this became significant (P < 0.05), although the curves are not presented as these would represent a substantial survival bias of those being alive to test.

When ovarian cancers were analyzed by stage, there was better long-term survival for early-stage disease (5-year/10-year survival stage I and II: BRCA1, 65%/47%; BRCA2, 100%/82%; P = 0.04) for BRCA2 versus BRCA1 but no difference in survival for late-stage disease (Table 4; Fig. 6), with very poor 10-year survival for women with either mutation. No substantial differences to these data were found when censoring for breast cancer diagnosis. Finally, we assessed the effect of birth cohort on life expectancy (Table 5). There was no improvement in survival from birth in more recent birth cohorts when excluding the index case (P = 0.24). When including the index case, there was a worsening trend for the more recent birth cohorts with 40% of the 1960+ cohort dead by 45 years of age compared with only 8% to 9% of the cohorts between 1920 and 1949 (log-rank = 27.60; degree of freedom, 6; P = 0.0001).

We have undertaken an actuarial analysis of BRCA1 and BRCA2 mutation-positive families to assess their life expectancy and the effect of breast and ovarian cancer diagnosis on this. The analysis shows a highly significant reduction in life expectancy in BRCA1 compared with BRCA2 carriers, which seems largely attributable to the later onset, lower frequency, and better long-term remission rates of early-stage ovarian cancer seen in the latter group.

There are several biases inherent in such an analysis. The index case tested within a family needs to be alive and affected with breast or ovarian cancer. We have, however, identified a large number of unaffected mutation carriers through predictive presymptomatic genetic testing and obligate carriers through connecting mutation-positive lineages. Although many women are now undertaking risk-reducing surgery, this was rare 10 years ago and is therefore unlikely to have biased the overall life expectancy figures within this study. Because there was a higher level of obligate carriers in BRCA1 families, this could reflect greater testing in families that had been identified earlier historically. However, there were actually more male obligate carriers in BRCA2 families (30:26) and the greater number probably reflects the earlier death rates particularly from ovarian cancer in BRCA1 mutation carriers. Although the inclusion of obligate carriers has partly corrected for the left truncation inherent in such analysis, we have also shown a significantly worse life expectancy among BRCA1 FDRs. The actuarial survival to 50 years is similar to proven carriers (there is a higher death rate for BRCA1). However, the inclusion of >50% of FDRs who would be noncarriers means that a higher proportion of these early deaths would be expected to have occurred among the carriers. Furthermore, our results of predictive testing would suggest that a higher proportion of the unaffected women aged >55 years would be BRCA2 carriers. We therefore believe that, regardless of the potential biases, this study highlights a clear difference in life expectancy between BRCA1 and BRCA2 female mutation carriers. We have also tested a large number of breast cancer only families, which avoids the bias of testing families only containing ovarian cancer. Cumulative risks for breast cancer were very similar in our high-risk families, consistent with other studies of such families (3, 6-8) and at least one population-based study (4). These should not be taken as a penetrance analysis for unaffected relatives as they did not include risks derived from untested relatives. The Breast Cancer Linkage Consortium cohort of high-risk families had lifetime risks of breast cancer of close to 85% for both genes (3, 6, 7). The estimate of ovarian cancer was also very similar with risks to 70 years of 60% for BRCA1 carriers and 27% (3) for BRCA2 carriers. Even population-based data sets give ovarian cancer risks of 39% (18-54%) for BRCA1 and 11% (2.4-19%) for BRCA2 (12). The differences in incidence rates are consistent, but as most women attending genetics clinics are from high-risk families, the penetrance estimates derived from these families are more likely to apply. Nonetheless, the results from our study would not be appropriate when advising women identified from lower-risk families or population screening as penetrance in these individuals is likely to be substantially lower particularly for breast cancer (12).

The similar survival curves after breast cancer diagnosis do not fit in with the higher proportion of breast cancer deaths in BRCA2 families. Indeed, most studies show either a worse or similar survival for BRCA1 compared with population controls or BRCA2 (13, 14). Therefore, the higher BRCA2 breast cancer deaths are likely to represent competing mortality from ovarian cancer, as BRCA1 mutation-positive women may well have gone on to die from breast cancer if they had not developed ovarian cancer. The individuals concerned would not contribute to the cumulative onset of breast cancer after their death from ovarian cancer.

Although the long-term outlook for stage III or IV disease is very similar across both groups, our data indicated that there is better survival in BRCA2-related early-stage ovarian cancer compared with that in BRCA1, which may therefore ratify the cell biological evidence of an increased response to DNA cross-linking agents such as carboplatin (15-21). Our data also suggest that although platinum-based therapy may result in extra years of life in late-stage disease, it is very unlikely to equate to a cure even in BRCA2. The results would support an effect of platinum agents in long-term remission from ovarian cancer, but the data cannot be compared with a population set due to the lack of similarly ascertained data (22). Nonetheless, the recent analysis from Israel with 213 mutation carriers does show significantly better survival in carriers (15). Median survival for carriers was significantly longer than for noncarriers (54 versus 38 months, respectively; P = 0.002). This was more pronounced among stages III and IV (5-year survival rates of 38.1% and 24.5% for carriers and noncarriers, respectively; P < 0.001). The number of BRCA2 carriers was probably not large enough to discriminate between the genes. Given the paucity of clinical data on BRCA2 mutation-related ovarian cancer, our study provides important information to oncologists and families. Given the lower and later incidence of ovarian cancer and potentially better cure rates, it is likely that ovarian cancer will have a much less significant effect on life expectancy in BRCA2 compared with BRCA1.

The main influence on long-term survival in ovarian cancer is stage at presentation (22). The majority of patients with ovarian cancer in both BRCA1 and BRCA2 presented with advanced (stage III/IV) disease, with no significant difference between BRCA1 and BRCA2 carriers. Any survival advantage for BRCA2 does not therefore seem to be related to any substantial effect of diagnosis of early-stage disease. Nonetheless, only ∼20% of BRCA1/2 mutation carriers are alive more than 10 years following diagnosis and only downstaging of disease at presentation is likely to improve long-term cure, with currently available therapies.

Reliance on ovarian screening with annual CA125 measurements and/or transvaginal ultrasound is likely to be misplaced, with the majority of ovarian cancers in BRCA1/2 carriers still presenting with stage III/IV disease (23-26).⁵

Indeed, our recent analysis of 49 mutation carriers in an annual screening program showed only 35% survival at 10 years.⁶ Nonetheless, if improvements in screening algorithms, such as more frequent CA125 measurement with analysis of rises within the reference range rather than a threshold, can reliably downstage ovarian cancer to at least stage II, then this may be sufficient for BRCA2 (24).

By use of decision analysis, several investigators estimated that women who had BRCA1/2 mutations would live longer if they had prophylactic surgery than if they had intensive screening alone (1, 2). This analysis on life expectancy adds weight to these decision analyses, although life expectancy gains from risk-reducing oophorectomy in BRCA1 are likely to considerably outweigh those for BRCA2. Nonetheless, we have shown not only a real incidence of ovarian cancer before 50 years of age in BRCA2 mutation carriers (Fig. 4) but also the death of six women from ovarian cancer under this age. It may therefore not be advisable to delay risk-reducing oophorectomy, in BRCA2 carriers who wish to undertake it, to 50 years, as has previously been suggested (23), unless there are improvements in screening. Although an argument can be made that much of the data in the present study are historical, the ovarian survival reflects modern treatments and we have evidence to show that life expectancy is not improving in BRCA1/2 carriers despite modern screening and treatment.⁶ Carriers in the most recent birth cohorts (1960+) do not have improved survival over cohorts before 1940.⁶ Therefore, the results of our analysis still have real meaning for BRCA carriers facing choices today.

We present historical data that female BRCA2 mutation carriers survived longer than those carrying BRCA1 mutations. This difference seems to be largely related to increased ovarian deaths in BRCA1 mutation carriers. It is likely that interventions aimed at reduction in ovarian incidence and death will have a larger effect for BRCA1 than BRCA2 carriers, but BRCA2 carriers may still benefit from risk-reducing oophorectomy before 50 years of age.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.