|Home | About | Journals | Submit | Contact Us | Français|
The risks for cancers other than breast (BC) or ovarian (OC) cancer in breast cancer gene 1 and 2 (BRCA1/2) mutation carriers were elevated in studies of carrier families. However, case–control studies did not confirm this observation.
To compare the risks for other cancers in BRCA1/2 mutation carriers and non‐carriers, all affected with BC and/or OC. Both groups share risk modifiers of BC/OC, which enabled assessment of the role of BRCA1/2 mutations.
1098 Ashkenazi Jewish women affected with BC and/or OC were ascertained during 1995–2003; molecular testing revealed 229 BRCA1 and 100 BRCA2 carriers and 769 non‐carriers. COX proportional hazard models were used to evaluate the risk of other cancers. Analyses were conducted including all other cancers or only those diagnosed after BC/OC diagnosis.
The HRs for any other cancer were 2.6 (95% CI 1.7 to 4.2, p<0.001) and 1.8 (95% CI 0.95 to 3.6, p=0.07) in BRCA1 and BRCA2 carriers, respectively. The corresponding colon cancer HRs were 3.9 (95% CI 1.3 to 12.1, p=0.02) and 2.3 (95% CI 0.5 to 11.3, p=0.3) in BRCA1 and BRCA2 carriers. The HR for lymphoma was 11.9 (95% CI 3.1 to 46.2, p=0.001) in BRCA2 carriers. Risk estimates for other cancers after the onset of BC/OC were similar.
A 2.5‐fold increase in any other cancer and a fourfold risk of colon cancer were found among BRCA1 carriers. The corresponding HRs in BRCA2 carriers were non‐significant, except for the markedly elevated risk of lymphoma. These results suggest a role for BRCA1/2 mutations in colorectal cancer risk in a subgroup of BC/OC‐affected carriers.
Carriers of mutations in the breast cancer 1 (BRCA1) and breast cancer 2 (BRCA2) genes have a substantially elevated lifetime risk of breast cancer (BC) and ovarian cancer (OC). However, the risk of cancers other than breast and ovarian cancers is controversial. Different approaches have been used to elucidate this issue. Studies of associated cancer risks among mutation carriers were conducted in high‐risk families, and have suggested an elevated risk of several types of cancers in carrier families.1,2,3,4,5,6,7 Two large studies reported elevated risk for pancreas and colon cancer (CC) in BRCA1 carriers; however, these studies were in conflict regarding the risk of other tumour types.1,2 Using a similar approach, a study by the Breast Cancer Linkage Consortium (BCLC) reported increased incidence of pancreas, melanoma and prostate cancer among BRCA2 mutation carriers.4 These studies were criticised for selection bias that may overestimate the cancer risks in carriers.8,9
Another approach to address this question is the use of case–control studies estimating the cancer risk by comparing mutation carrier frequency in a series of individuals affected with a specific type of cancer with that of unaffected controls.10,11,12 These studies usually failed to confirm risk elevation for other cancers in BRCA1/2 carriers.
In this study, we evaluate the cancer risk in BRCA1/2 carriers identified through oncology or cancer‐genetic clinics. The frequencies of other cancers were compared between Ashkenazi BRCA1/2 carriers of any of the three Ashkenazi founder mutations (185delAG and 5382insC in the BRCA1 gene and 6174delT in the BRCA2) and non‐carriers of Ashkenazi. We limited the sample to individuals affected with breast cancer (BC) and/or ovarian cancer (OC), both in the carrier and in the control group. We hypothesised that both groups share common environmental, life style and genetic risk modifiers associated with BC and OC, and this comparison allows estimation of the additional role of BRCA1/2 carrier status in the risk of other cancers in this subset group.
A total of 1616 females, all of Ashkenazi origin, were tested for the BRCA1/2 Ashkenazi founder mutations through oncology and cancer‐genetic clinics in Hadassah Medical Center (Jerusalem, Israel) during 1995–2003. All participants signed an informed consent approved by the institutional ethics committee. We estimated the contribution of BRCA1/2 mutations to the risk of additional cancer development by comparing only those affected with breast or ovarian cancer. As the Ashkenazi founder mutations are frequent and easy to test, a large fraction of the Ashkenazi patients with BC and OC were referred for testing; therefore, selection due to diagnosis of any other cancer is probably low and equal among carriers and non‐carriers. Of 1616 women tested, we concentrate on 1098 affected with BC (n=947), OC (n=109), or both BC and OC (n=42). In this group, 329 were BRCA1/2 carriers (185delAG in 172, 5383insC in 36 and 6174delT in 100 individuals, three carried both 185delAG and 6174delT mutations and were omitted from analyses) and 769 did not carry either of these mutations. The frequency of non‐founder BRCA mutations in the Ashkenazi population is low,13 therefore the control group includes only a negligible number of carriers. Table 11 shows the comparison of primary diagnosis (BC and/or OC), age at onset of BC, OC and other cancer, and the average patient years of observation between BRCA1, BRCA2 carriers and non‐carriers.
As all individuals were tested while alive and the majority of them were treated in our institution, the accuracy of cancer diagnoses is very high. In all, 65 of 118 (55%) of all other cancers and 15 of 20 (75%) of CC were operated in Hadassah Medical Center; the rest were confirmed through medical files.
Clinical parameters were compared between BRCA1, BRCA2 carriers and non‐carriers. We used χ2 test for primary diagnosis (BC and/or OC), rate of BC diagnosis at age <50 years and timing of diagnosis (before/after BC/OC). Exact p values are reported. Analysis of variance was used for the series of age comparisons (age at onset of BC, OC, CC, all other cancers and patient years of observation from year of birth to last follow‐up date or death).
The hazard ratio (HR) for any other cancer was compared between BRCA1/2 mutation carriers and non‐carriers using cox proportional hazard models, with time of onset of other cancers or last follow‐up as end point. Average lengths of follow‐up were 52.8, 57.0 and 59.6 years in BRCA1, BRCA2 carriers and non‐carriers, respectively (p<0.001). To minimise the possible effect of ascertainment bias, treatment effect or surveillance bias HRs were also analysed, including only cancers which were diagnosed after the diagnosis of BC/OC. Separate analyses were conducted for BRCA1 and BRCA2 carriers. Specific tumour types with sufficiently high frequency for separate evaluation, or those reported to be associated with BRCA1/2 carrier status in previous studies, were also analysed (colorectal, lymphoma and uterine tumours).
Table 11 presents the clinical characteristics of carriers and non‐carriers that are potential confounders of other cancer risks. Higher rates of OC diagnosis (with or without BC) were found among BRCA1 (74/229, 32.3%) and BRCA2 carriers (28/100, 28%) compared with non‐carriers (49/769, 6.3%, p<0.001). Significant differences were observed in the mean age at onset of BC (p<0.001) and OC (p=0.006) and patient years of observation (p<0.001) between the various groups. However, the age at onset of other cancers was similar (55.1, 49.9 and 57.6 in BRCA1, BRCA2 carriers and non‐carriers, respectively, p=0.2).
Table 22 presents the rates and related HRs for other cancers in BRCA1/2 mutation carriers compared with non‐carriers. Of 1098 individuals, 118 were diagnosed with additional cancer (11 of them had more than one other cancer, and one individual, a BRCA2 mutation carrier, had three other tumours). Frequencies of individuals with other cancer diagnoses were 12.2% (28/229), 12.0% (12/100) and 10% (78/769) in BRCA1, BRCA2 carriers and non‐carriers, respectively. Although seven of 28 (25%) BRCA1 carriers were diagnosed with other cancers before or at the same time as BC/OC diagnosis, the corresponding rates were six of 13 (46.1%) and 31 of 76 (40.8%) in BRCA2 carriers and non‐carriers, respectively (p=0.3, table 11).
A significantly elevated HR of 2.4 (95% CI 1.6 to 3.6, p<0.001) for any other cancer was found in BRCA1/2 mutation carriers. The corresponding HRs were 2.6 (95% CI 1.7 to 4.2, p<0.001) and 1.8 (95% CI 0.95 to 3.6, p=0.07) in BRCA1 and BRCA2 carriers, respectively (table 22).). When only the cancers diagnosed after BC/OC were included, HRs were 4.1 (95% CI 2.3 to 7.1, p<0.001) and 2.0 (95% CI 0.83 to 4.9, p=0.1) in BRCA1 and BRCA2 respectively (table 3).
In separate analyses according to specific tumour type, estimated HRs were significantly elevated for CC among BRCA1 carriers (3.9, 95% CI 1.3 to 12.1, p=0.02), whereas in BRCA2 carriers a non‐significant HR of 2.3 (95% CI 0.5 to 11.3, p=0.3) was found (table 22).). The HR for developing CC after the diagnosis of BC/OC was 4.0 (95% CI 1.1 to 15.2, p=0.04) and 3.3 (95% CI 0.61 to 17.4, p=0.1) in BRCA1 and BRCA2 carriers, respectively. Among patients with CC, three of six BRCA1 carriers and one of two BRCA2 carriers were affected with OC. Interestingly, in BRCA1 carriers affected with CC, ages at onset were approximately 8 years more than the non‐carriers, for both BC/OC (61.8 compared with 55.8 years in non‐carriers, p=0.26) and CC (71.0 compared with 62.0 years in non‐carriers, p=0.61). BRCA2 carriers affected with CC included only two individuals affected approximately 9 years earlier than non‐carriers, both with BC/OC and with CC. The risk of lymphoma was markedly increased among BRCA2 carriers (HR, 11.9, 95% CI 3.1 to 46.2, p=0.001), but not among BRCA1 carriers. The estimated HRs for the rest of the tumour types were non‐significant.
In this study, we evaluate the risk of cancers other than BC and OC, in a subset of BRCA1/2 mutation carrier females affected with BC and OC. We have found an approximately 2.5‐fold increase in risk of any other cancer in BRCA1/2 mutation carriers, and a threefold increase in CC, which was significant in BRCA1 but not in BRCA2 carriers. Additional analyses including only cancers diagnosed after BC/OC diagnosis were conducted to correct for possible treatment, surveillance and ascertainment biases. HRs of 4 were found for the risk of developing any other cancer and CC after BC/OC diagnosis among BRCA1 carriers; among BRCA2 carriers, HRs were not significant. In addition, lymphoma risk was highly elevated among BRCA2 carriers.
Two large studies estimated the risk of other cancers in BRCA1 mutation carrier families.1,2 In these studies, a fraction of the carriers were tested directly, and the probability of being a carrier was calculated for family members who were not tested. Risks were compared with lifetime risk or incidence in the general population. Both studies found elevated risk of cancers other than BC and OC. In agreement with our results, Thompson et al2 reported an OR of 2.3 (95% CI 1.93 to 2.75, p=0.01) for any other cancer in female carriers but not in males. Brose et al,1 on the other hand, found a higher cumulative age‐adjusted lifetime risk of diagnosis of any cancer in male carriers of BRCA1 mutations (26.1%, 95% CI 17.5% to 34.6%) than in carrier women (10.3%, 95% CI 7.2% to 13.3%); however, their analyses were based on 381 females and only 102 males. In BRCA2 carriers, a relative risk (RR) of 2.45 (95% CI 2.15 to 2.78, p<0.001) for all cancers except BC, OC and non‐melanoma skin cancer was found.4 It was suggested that the elevated cancer risk found in carrier families, which includes a large fraction of affected individuals, may be overestimated, because of genetic or environmental exposures that are risk factors for other types of tumours in addition to BC and OC, which cluster in these families.8,9 Our unique approach of comparison of BC/OC‐affected carriers with non‐carriers enabled us to evaluate the effect of BRCA mutations controlling for these factors.
The strength of our study is the direct testing of all carriers and the high quality of verification of diagnosis of other cancer. In addition, we ascertained all participants in the same way, mainly due to BC or OC diagnosis and the Ashkenazi origin. The high frequency and ease of testing of Ashkenazi founder mutations yielded a series that is less selected on the basis of family history as opposed to previous studies.1,2,4,5,7 The ascertainment due to diagnosis of other cancer was probably low and equal in the two groups we compared, in addition, survival bias may have resulted in reduced incidence of other cancers in both groups and underestimation of the risks. The carrier group, especially BRCA1 mutation carriers, are younger, regarding both age of BC/OC diagnoses and average length of follow‐up; however, age at diagnosis of other cancer is similar to that for non‐carriers. Therefore, all estimated HRs were analysed from birth to age at onset of other cancers or last follow‐up.
CC was the major contributor to the elevated cancer risk we found in our carrier group. In BRCA1 carriers it was significantly fourfold higher, whereas the HR of 2.3 that was found in BRCA2 carriers was non‐significant. Similar HRs were found when only CCs diagnosed after BC/OC diagnosis were included. Several studies in high‐risk carrier families evaluated CC risk. Brose et al1 found a twofold increase in the risk of CC in BRCA1 carriers; separate analyses by gender were not reported. An older study by the BCLC5 reported a fourfold increase in the risk, whereas in a subsequent, much larger study,2 a twofold increased risk was found in female, but not in male, carriers. As the incidence of rectal cancer was markedly low, the authors suggested that the elevated risk could be attributable to OC cases misdiagnosed as CC. In BRCA2 carriers, risk of CC was not elevated.4 Recently, two large case–control studies in Ashkenazi patients with CC were reported.10,11 Neill et al10 found 2.4% (24/1002) of patients with CC to carry one of the three Ashkenazi founder mutations, compared with 1.9% (19/1038) in the controls, which translated to a non‐significant odds ratio of 1.24 (95% CI 0.68 to 2.26). A similar study11 among 586 Ashkenazi patients with CC reported an RR of 0.50 (95% CI 0.22 to 1.14) associated with these mutations.
Common risk modifiers for BC and colorectal cancer, such as diet14,15,16 and exercise,17,18 as well as genetic modifiers clustering in high‐risk families, may result in elevated CC risk in carrier families, regardless of the BRCA1/2 carrier status. However, by comparing BC/OC‐affected carriers with affected non‐carriers who share those factors, we suggest a role, either additive or synergistic, of BRCA1/2 mutations in occurrence of CC in this subset group. It is important to note that individuals affected with CC, especially BRCA1 carriers in our series, were older than the whole study group. Therefore, probably, longer survival is required for CC development in BRCA1 carriers. In the case–control studies mentioned above,10,11 carriers affected with BC/OC are probably under‐represented owing to a high mortality rate at younger age, which may have resulted in a lower carrier rate.
The risk of lymphoma was markedly elevated among BRCA2 carriers, with an HR of 11.9 (95% CI 3.1 to 46.2, p=0.001), whereas HR among BRCA1 carriers was non‐significantly reduced. In agreement with our results, the BCLC study reported an RR of 1.91 (95% CI 0.81 to 4.49) for non‐Hodgkin's lymphoma and 1.48 (95% CI 0.4 to 5.48) for Hodgkin's disease in BRCA2 carriers.4 The corresponding RRs in BRCA1 carriers2 were 0.23 (95% CI 0.09 to 0.60, p<0.001) and 0 (p=0.15), respectively. However, an RR of 2.6 was found for leukaemia and lymphomas, based on 649 patients with ovarian cancer (all BRCA1 carriers).6 Again, in a case–control study of 286 Ashkenazi Jewish patients with lymphoma, only two BRCA carriers were found.12 Whether the elevated lymphoma incidence we and others4 observed in BRCA2 carriers is related to treatment, owing to the role of BRCA2 in double‐strand breaks repair, should be further investigated. Of the four BRCA2 carrier patients with lymphoma in our series, two were diagnosed with lymphoma and treated with irradiation before BC diagnosis, the other two were diagnosed 5 and 34 years after the diagnosis of BC, and the causative role of the treatment for BC in the development of lymphoma is less clear.
In conclusion, our findings show an elevated risk of other cancers, especially that of CC, in a subset BC/OC‐affected carriers. Our results partly explain the controversy between studies of high‐risk carrier families and the case–control studies conducted in the general population. We speculate that the affected carriers, which comprise the majority of the carrier population in studies of high‐risk families, are the main contributors to the elevated cancer risks found in these studies.1,2,3,4,5,6,7 In the case–control studies10,11 conducted in the general population, the BC/OC‐affected carriers with early onset of disease are under‐represented, owing to reduced survival. In the future, better treatment and surveillance in carriers may result in longer life expectancy and therefore higher risk for other cancers, especially CC, among carriers.
BC - breast cancer
BCLC - Breast Cancer Linkage Consortium
BRCA1/2 - breast cancer genes 1 and 2
CC - colon cancer
HR - hazard ratio
OC - ovarian cancer
RR - relative risk
Competing interests: None declared.