Abstract
Purpose: To evaluate the prevalence of BRCA1 and BRCA2 mutations and associations with clinical correlates of disease in a population-based series of ovarian cancer cases from Denmark.
Methods: DNA sequencing and multiplex ligation-dependent probe amplification analysis were used to analyze the BRCA1 and BRCA2 genes for coding sequence mutations and large genomic rearrangements in 445 confirmed cases of ovarian cancer. We evaluated associations between mutation status and clinical characteristics, including cancer risks for first-degree relatives and clinicopathologic features of tumors.
Results: Deleterious BRCA1 or BRCA2 mutations were identified in 26 cases; thus, mutations in these genes are responsible for at least 5.8% of ovarian cancer cases in this population. Five different mutations were identified in more than one individual, suggesting that they may be founder mutations in Denmark. We identified several differences between mutation carriers and noncarriers: mutation carriers were diagnosed at a significantly early age (median, 49 and 61 years, respectively; P = 0.0001); the frequency of BRCA1 mutation carriers was 23% for women diagnosed <40 years, 15% for 40 to 49 years, 4% for 50 to 59 years, and 2% for ≥60 years (P = 0.00002); ovarian cancer in carriers was diagnosed at a later stage (P = 0.002) and tumors were of poorer grade (P = 0.0001); and first-degree relatives of mutation carriers had greater relative risks of both ovarian cancer [10.6 (95% confidence interval, 4.2-26.6); P < 0.0001] and breast cancer <60 years [8.7 (95% confidence interval, 3.0-25.0); P < 0.0001].
Conclusion: These data may have a significant effect on risk assessment and clinical management of individuals from Denmark who are predisposed to ovarian cancer because they carry a BRCA1 or BRCA2 mutation.
- BRCA1 and BRCA2 mutations
- clinical correlates of disease
- ovarian cancer
Epithelial ovarian cancer is the leading cause of death from gynecologic malignancy in the western world. The strongest known risk factor is a family history of the disease; an individual with a first-degree relative affected with ovarian cancer has a 3-fold increased risk of developing the disease (1). Two genes, BRCA1 and BRCA2 (2, 3), are responsible for the majority of families containing multiple cases of breast and ovarian cancer (4–6). The cumulative lifetime risks of ovarian cancer associated with these genes have been estimated as 40% to 50% for a BRCA1 mutation carrier and 20% to 30% in BRCA2 carriers (4, 7).
There have been several studies reporting the prevalence of BRCA1 and BRCA2 mutations in ovarian cancer cases unselected for family history (8–13). Together, these studies suggest that mutations in these genes may cause 3% to 15% of all ovarian cancers. The first published study of 374 ovarian cancer cases from southern England, which analyzed just BRCA1, identified truncating mutations in 3% of cases (8). Another larger study reported a higher prevalence (8%) in 977 patients from Canada (13). Less data are available for BRCA2, but the Canadian study reported deleterious mutations for 5.9% of cases (13).
There are several reports of BRCA1 and BRCA2 analysis in ovarian and breast cancer cases from some Scandinavian countries (Finland, Norway, Sweden, and Iceland; refs. 13–19), but the data from the Danish population are much more limited. The largest study of BRCA1 and BRCA2 mutations to date from Denmark is the analysis of 103 multifocal or bilateral early-onset breast cancers, which identified mutations in 20% of cases (20). There are no reports in the literature describing the analysis of BRCA1 and BRCA2 in population-based ovarian cancer cases from Denmark.
For all but one of the population-based studies described above, screening for mutations in BRCA1 and BRCA2 was limited to the coding sequence and splice sites of both genes, which will only identify single-base changes or small deletions and insertions of a few nucleotides. However, several studies have now reported the common occurrence of large, genomic alterations (rearrangements and deletions) in BRCA1 and BRCA2 (21, 22). For BRCA1, this type of mutation represents 8% to 40% of all BRCA1 mutations identified in families from the United Kingdom, United States, France, Germany, The Netherlands, and Italy, but may be rare in the Danish population (6, 23, 24). This suggests that most studies will have underestimated the prevalence of BRCA1 and BRCA2 mutations in ovarian and breast cancer populations.
Some studies have also reported on the associations between BRCA1 and BRCA2 mutation status and clinical features of disease. These studies suggest that tumors from BRCA1 carriers are more likely to be of serous histology and higher grade (25); BRCA2 tumors appear to be similar by pathology to BRCA1 tumors (25). Patients with BRCA1 mutations are also diagnosed at a younger age (12, 13). Reports describing associations between survival and BRCA1 and BRCA2 mutation status are conflicting (26–29).
The purpose of the current study was to establish the prevalence of BRCA1 and BRCA2 mutations, including coding sequence alterations and large genomic deletions/rearrangements in a population-based series of 445 invasive ovarian cancer cases unselected for family history of cancer from Denmark. A secondary aim was to correlate BRCA1 and BRCA2 mutation status with both a family history of ovarian and breast cancer and clinical features of disease including age at diagnosis and histopathologic characteristics of ovarian tumors.
Materials and Methods
Patient samples. The MALOVA study is an epidemiologic, population-based study of ovarian cancer cases from Denmark. Details of this population have been published previously (30). Briefly, eligible cases were women ages 35 to 79 years with suspected ovarian cancer, who were diagnosed with an ovarian tumor from December 1994 to May 1999. Study participants were recruited from 16 different hospitals throughout Denmark. To ensure that all eligible cases in the study area were included, we also identified cases by searching the Danish Cancer Registry, which is a nationwide register, every 2 months. Participants provided blood and tissue samples and completed an epidemiologic, questionnaire. Ethics committee approval was obtained for the collection and genetic analysis of all samples, and an informed written consent was obtained from all participants. Likewise, approval from the Danish Data Protection Agency was obtained. In total, 681 patients with invasive ovarian cancer, 235 cases of borderline ovarian cancer, and 450 cases with a benign ovarian tumor were enrolled in the study. Lymphocyte DNA samples of sufficient quantity and suitable quality were available from 445 invasive epithelial ovarian cancer cases.
We attempted to establish the history of cancer in first-degree relatives of all ovarian cancer cases in this study using the Danish Civil Registration System, which assigns a unique 10-digit personal identification number to every Danish person alive on April 1, 1968 and born thereafter. Index cases were asked about the name and date of birth of their mother and biological sister(s) and daughter(s). The Civil Registration System was then used to confirm this information. When information on the relatives could not be found, we contacted the National Register (Folkeregisteret) in the municipality where the woman was born. The National Register includes information on every inhabitant in 271 different municipalities from 1924 up to the time the Civil Registration System was introduced. If information was unavailable from these systems, we manually went through parish registers. In total, we obtained information on 427 mothers (96% complete), 389 fathers (87% complete), 494 sisters, and 310 brothers. We used the Danish Cancer Registry, established in 1942 to obtain information about cancer incidence in this cohort. Since 1987, it has become compulsory to notify the registry of all cases of cancer. Even before this date, the register was 96% to 98% complete. Cancers diagnosed up to December 31, 2003 were included. The Civil Registration System contains information on dates of death and emigration and was used to perform active clinical follow-up of cases up until 2004.
Genetic analysis. All cases were screened for mutations in the coding region and splice site (intron/exon) boundaries by sequencing. DNA samples were amplified using Variant SeqR primer sets for BRCA1 (v1) and BRCA2 (v2; Applied Biosystems). Custom primers were designed for regions not covered by these kits (primer sequences available on request). Sequencing reactions were done using the Big Dye terminator v1.1 kit (Applied Biosystems). M13 forward and reverse primers were used to sequence each fragment for both sense and antisense strands. Sequencing reactions were size fractionated on a 3730xl, 96 capillary DNA analyzer (Applied Biosystems). SeqScape software version 2.5 (Applied Biosystems) was used to analyze sequence traces. Pass rates for the number of samples aligned by SeqScape were determined for each PCR fragment and those with <70% pass were repeated. All detectable mutations were confirmed in an independent PCR amplified product.
Successful sequencing was achieved for 91% of all PCR fragments for both genes (93% for BRCA1 and 89% for BRCA2). It was not possible to sequence across all of the coding region for each gene. We achieved 98% coverage of BRCA1 and 95% coverage of BRCA2. Thus, by combining successful sequencing rates and gene coverage, we estimate that we were able to achieve complete gene sequencing of at least 91% of BRCA1 and 85% of BRCA2 for the study population. These figures are based on pass rates per fragment and are likely to be an underestimate because many of the fragments are overlapping.
Cases were screened for large genomic alterations of BRCA1 and BRCA2 by multiplex ligation-dependent probe amplification (MLPA) using the SALSA P002-BRCA1 or P045-BRCA2 Exon Copy Number Test Kits (MRC-Holland). Details of MLPA analysis have been described previously (6). MLPA was successfully done in 93% of cases for BRCA1 and 85% of cases for BRCA2. We attempted to characterize the breakpoints of the one genomic rearrangement we found by long-range PCR using TaKaRa LA enzyme (TaKaRa). However, it was not possible to amplify the genomic fragment containing the MLPA alteration. The naming of mutations was based on the nucleotide number of the cDNA sequences in GenBank U14680.1 (BRCA1) and GenBank U43746.1 (BRCA2). For deletions and insertions in short tandem repeats, the most 3′ nucleotide was arbitrarily assigned as recommended by Antonarakis and the Nomenclature Working Group (31).
Statistical analysis.t tests were used to analyze differences in the mean ages of diagnosis. χ2 tests were used to analyze differences in family history and the proportions for age at diagnosis. Fisher's exact tests were used to analyze differences in histology, stage, and grade, by mutation status. Unconditional logistic regression analysis was used to calculate the relative risks and 95% confidence intervals of ovarian cancer, breast cancer, prostate cancer, or any cancer in first-degree relatives of mutation carriers.
Results
BRCA1 and BRCA2 mutations in ovarian cancer cases from Denmark. We evaluated the prevalence of germ-line BRCA1 and BRCA2 mutations in a population-based series of 445 invasive epithelial ovarian cancer cases from Denmark. Both genes were screened using a combination of DNA sequencing to identify coding sequence and splice-site alterations and MLPA to identify large genomic rearrangement mutations.
Deleterious BRCA1 mutations were found in 22 cases (4.9%) and BRCA2 mutations in 4 cases (0.9% Table 1 ). Thus, BRCA1 and BRCA2 mutations are responsible for at least 5.8% of ovarian cancer cases in this population. Twenty-five mutations were in the coding region and were either frame-shift deletions or insertions of 1 to 5 bp or nonsense substitutions. We identified one genomic alteration by MLPA, which resulted in an in-frame deletion of exons 17 to 19. We have included this as a deleterious mutation, although it is not predicted to lead to protein truncation. The deletion would remove 207 bp of the BRCA1 coding sequence and 69 amino acids (3.7% of the predicted protein). Five different mutations were found in more than one individual from the study, suggesting that they may be founder mutations in the Danish/Scandinavian population. Combined, these mutations accounted for nearly half of all the mutations identified. The most common mutations were 2594delC and 3829delT, both in BRCA1, which were found in four and three cases, respectively.
Mutation and family history information and histologic classification in patients with a BRCA1 or BRCA2 mutation
No alterations that are predicted to affect splicing were identified. Neither did we find any of a small number of previously reported missense mutations that are known to be deleterious. However, we did identify several rare nonsynonymous base changes, some of which have been reported before in the Breast Cancer Information Core (BIC) database (32); others are previously unreported variants (see Table 2 ). Thirteen variants in BRCA1 and 27 variants in BRCA2 had an allele frequency of <0.01 and were nonsynonymous sequence changes for which there may potentially be disease associated function (Table 2). We attempted to model the predicted functional effect for these alterations using the programs PMut, MuPro, and SIFT (33–35). Eight of these variants were predicted to be pathogenic using at least two of the measures. These were L1198W and R1347G in BRCA1 and A75P, S1832P, D2665G, T2766I, N2781I, and K2860T in BRCA2. However, two of these variants (L1198W and T2766I) were present in two patients that also had different termination mutations in BRCA1, which suggests that they are unlikely to be pathogenic. The variants R1347G, A75P, and D2665G have been reported on the BIC database 154, 49, and 26 times, respectively, where they are described as of “unknown” or of “no” clinical significance. Finally, several known common single nucleotides polymorphisms in the coding region were also found (data not shown).
Rare, nonsynonymous sequence variants identified in patients from the MALOVA study
Associations between BRCA1 and BRCA2 mutation status, family history, and clinical characteristics of disease. We examined the relationship between mutation status and a family history of breast, ovarian, colorectal, prostate, and “other” cancers in first-degree relatives of cases from this population (Table 3 ). Nine BRCA1 and BRCA2 mutation carriers (35%) had one or more first-degree relative(s) with confirmed epithelial ovarian cancer. This contrasts with cases in which no mutations were found; only 5% of these cases had a first-degree relative with ovarian cancer (P = 0.00001). Similarly, six mutation carriers (23%) had a family history of breast cancer <60 years in first-degree relatives compared with 3% of noncarriers (P = 0.0007). The relative risks (95% confidence intervals) of ovarian and breast cancer <60 years in first-degree relatives of mutation carriers were 10.6 (4.2-26.6) and 8.7 (3.0-25.0), respectively (P < 0.0001 for both). We found no statistically significant differences in the risks for prostate or any other cancers in carriers compared with noncarriers, although the numbers were small (data not shown).
Clinical and family history information for all cases from the MALOVA study screened for BRCA1 and BRCA2 mutations and correlation with mutation status
Age at diagnosis and family history of cancer were strong predictors of BRCA1/BRCA2 mutation status (Table 3). Approximately half of all cases in this population were diagnosed with ovarian cancer over 60 years. The median age of ovarian cancer diagnosis in mutation carriers was substantially less than for individuals without a mutation (49 versus 61 years; P < 0.0001). Fifty-four percent of BRCA1 and BRCA2 carriers were diagnosed before age 50 years compared with 19% of noncarriers. Mutations were identified in 23% of cases aged <40 years, 13% of cases aged 40 to 49 years, 4% of cases aged 50 to 59 years, and 2% of cases aged ≥60 years; this trend was significantly different compared with noncarriers (P = 0.00002). Of 445 cases in this study, 45 (10%) had a family history of ovarian cancer or breast cancer <60 years in first-degree relatives; 12 of these cases harbored a BRCA1 or BRCA2 mutation (46% of all mutation carriers). In total, 20 of 26 mutation carriers (77%) were either diagnosed <50 years and/or had a family history of ovarian/breast cancer, although only 27% of cases from this cohort fulfilled one or both of these criteria.
Previous studies have suggested that the presence of a germ-line BRCA1 or BRCA2 mutation may correlate with the histologic subtype of ovarian cancers. In this series, 62% of tumors from nonmutation carriers were serous histology compared with 64% of mutation carriers. The only notable difference between the two groups was in the proportion of mucinous cases, although the numbers were small; no tumors from carriers were mucinous compared with 10% of non-BRCA1/BRCA2 tumors that were. We found significant differences between carriers and noncarriers for stage at diagnosis and tumor grade. Ninety-five percent of carriers were diagnosed at stages III/IV compared with 65% of noncarriers (P = 0.002); 86% of BRCA1/BRCA2 tumors were poorly differentiated compared with only 35% of non-BRCA1/BRCA2 tumors (P = 0.0001; Table 3). Ovarian cancer cases with no identifiable mutation, but with a family history of ovarian/breast cancer, more closely resembled the noncarriers without a family history rather than the mutation carriers (Table 3). We were unable to evaluate associations between mutation status and survival due to the small numbers of mutation carriers identified in the study.
Discussion
We have established the prevalence of BRCA1 and BRCA2 mutations in a population-based study of 445 ovarian cancer cases from Denmark. Twenty-six deleterious mutations were identified: 22 mutations in BRCA1 (4.9%) and 4 mutations in BRCA2 (0.9%). Thus, the combined prevalence of BRCA1 and BRCA2 mutations in this series was 5.8%. To our knowledge, this is the first such investigation in an ovarian cancer case series from Denmark, so we are unable to draw direct comparisons with other Danish studies. However, there have been a few studies that have reported the contribution of BRCA1 and/or BRCA2 to ovarian cancer in other populations. These studies provide different and wide-ranging estimates of mutation prevalence for these genes, suggesting a degree of genetic heterogeneity between different populations. Other possible explanations for the variation seen between studies include ascertainment bias in population collections, the presence of founder mutations and ethnic bias between different populations, and the completeness and accuracy of mutation screening.
The BRCA1 and BRCA2 prevalence estimates from this study are consistent with those reported in other populations, but they are substantially lower than for two other studies (11–13). Risch et al. initially reported on 515 invasive ovarian cancer cases and later on an additional 462 cases (977 cases in total) as part of the same cohort from Ontario, Canada (13). BRCA1 and BRCA2 mutations were found in 13.2% of cases. Pal et al. found BRCA1 and BRCA2 mutations in 15.2% of 208 ovarian cancer cases from west central Florida (12). A main reason for the differences between these studies and ours is the high frequency of BRCA2 mutations identified (5.5% and 5.8%, respectively). The ratios of BRCA1 to BRCA2 mutations were 1.4:1 and 1.7:1, respectively, compared with 5.5:1 for our study. In this respect, the current study is more similar to a recent analysis of BRCA1 and BRCA2 in 283 ovarian cancer families, in which the ratio of BRCA1 to BRCA2 mutations was 5.9:1 in the UK population (6). Another contributing factor to the high frequency of mutations identified by Risch et al. and Pal et al. is the presence of Ashkenazi Jewish founder mutations, which represent 10.1% and 21.9%, respectively, of all mutations identified in each study. Finally, the studies of Risch et al. and Pal et al. both contain an unusually high proportion of cases with a family history of ovarian and/or breast cancer, which are much more likely to have a BRCA1 and BRCA2 mutation. Twenty-eight percent of cases in the study by Pal et al. and 25% of cases in the study by Risch et al. had at least one first-degree relative diagnosed with breast or ovarian cancer; this compares with only 10% of cases in our study.
Other BRCA1 and BRCA2 mutation screening studies from Scandinavia provide insights into mutation prevalence in northern Europe and of founder mutations that may also be common in Denmark. Most reports of mutation screening in Scandinavian ovarian cancer cases are based on the analysis of selected founder mutations in population studies. In a study from Finland, 20 different BRCA1 and BRCA2 mutations that had been identified previously in the Finish population were present in 6% of 233 unselected ovarian cancer cases (14). In another study, two common Norwegian BRCA1 founder mutations were identified in 3% of 615 ovarian cancer cases (16). More recently, a Swedish study identified deleterious BRCA1 and BRCA2 mutations in 8% of 161 ovarian cancer cases (8).
Four of the five BRCA1 mutations found more than once in this study may be founder mutations in Denmark or Sweden. The 5386insC mutation (also known as 5382insC) is common throughout Europe and is also a founder mutation in the Ashkenazi Jewish population. The 2594delC mutation constituted 18% of BRCA1 mutations in this study and was found at a similar frequency (17%) in the study of breast cancer cases from Denmark (20). The 3829delT and Q563X mutations have also been reported in studies from Sweden, suggesting that they may be Swedish/Danish founder mutations. However, we did not identify other north European founder mutations including the BRCA2 999del5 mutation common to Iceland, the BRCA1 1675delA and 1135insC mutations common to Norway, and the BRCA1 3172ins5 and 1201del11 mutation common to Sweden. Finally, the 2644insG mutation, which we found in two cases, has never been reported before in Scandinavia or elsewhere.
It is unlikely that any of the population-based studies described for ovarian cancer will have detected all of the mutations present in the populations studied, so the prevalence of BRCA1 and BRCA2 mutations will have been underestimated. Only one of the studies published thus far analyzed BRCA1 and BRCA2 for large genomic rearrangement mutations, although this type of alteration can be common, particularly in the BRCA1 gene (23). In the current study, one of the 22 BRCA1 mutations (∼5%) was a large genomic deletion. Mutations may also have been missed due to the sensitivity of the mutation detection techniques. We used sequencing, which is widely considered the most accurate approach to mutation screening, but mutations may have been missed because we were unable to completely sequence the entire coding region in all samples. Finally, some missense alterations, which are often not recorded as deleterious alterations, may be pathogenic.
We identified several correlations between mutation status and either a family history of cancer or clinical features of disease. In the main, these analyses were restricted to BRCA1 carriers due to the small number of BRCA2 mutation carriers identified in this study. Not surprisingly, mutation carriers were more likely than noncarriers to have a history of ovarian and breast cancer. However, it is perhaps surprising that more than half of all mutation carriers had no affected first-degree relatives, suggesting that breast/ovarian cancer risk estimates are not as high as suggested from familial studies. Antoniou et al. (36) estimated the risks of breast and ovarian cancer at age 70 years to be 72% and 53%, respectively, for BRCA1 carriers in a study of ovarian cancer families. In the same study, the risks associated with BRCA2 were 71% for breast cancer and 31% for ovarian cancer. However, breast and ovarian cancer risks calculated from a meta-analysis of population-based breast/ovarian cancer studies suggests much lower risks; the average cumulative risks of breast and ovarian cancer by age 70 years in BRCA1 carriers were 65% and 39% and in BRCA2 mutation carriers 45% and 11%, respectively (37). In this study, the estimated relative risk (95% confidence interval) for ovarian cancer [10.6 (4.2-26.6)] is comparable with those of the study of 977 ovarian cancer cases from Canada [10.3 (6.01-17.6) for BRCA1 mutation carriers and 3.46 (1.55-7.72) for BRCA2 mutation carriers; ref. 13].
There are now several studies that have investigated links between germ-line genetic variation and clinicopathologic features of ovarian cancer. We found no evidence that BRCA1 tumors are more likely to be serous compared with non-BRCA1 tumors as suggested by a previous study (25). Lakhani et al. (25) examined 178 invasive ovarian cancers from BRCA1 mutation carriers and found that 44% were serous compared with 31% of non-BRCA1 tumors. As in the current study, Lakhani et al. also found a significant difference in grade between BRCA1 and non-BRCA1 tumors; in both studies, BRCA1 tumors were more poorly differentiated. We additionally found that patients with BRCA1 tumors were more likely to be diagnosed at a later stage.
Patients in which a mutation was identified were diagnosed at a younger age than noncarriers; this was a highly significant result (P = 0.00002). Given the strength of this finding, it is perhaps surprising that the Canadian and Florida studies described above did not report something similar (5.7% and 3.1% of BRCA1 carriers, respectively, were <40 years compared with 23% in the current study; refs. 12, 13). However, this trend with decreasing age appears to be validated by data from the Canadian study, which shows a similar effect; 11% of cases were diagnosed ≤50 years, 5% were 51 to 60 years, and 3% were >60 years (13). If these results are true, then they could have implications for clinical genetic testing. In the current study, by stratifying the ovarian cancer case population by age at diagnosis and family history of ovarian/breast cancer, we would have identified 77% of BRCA1/BRCA2 mutation carriers by screening only 27% of the population.
In conclusion, we have evaluated the contribution of BRCA1 and BRCA2 mutations to ovarian cancer in a population-based series of cases from Denmark. The data were used to identify correlates between mutation status and both a family history of ovarian and breast cancer and clinical features of disease. Because this study is the first of its kind for ovarian cancer cases from Denmark, the findings are likely to be of clinical benefit in the future for individuals undergoing genetic testing and counseling for BRCA1 and BRCA2 because they have a family history of ovarian and/or breast cancer.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Footnotes
-
Grant support: Mermaid/Eve Appeal. This work was partly undertaken at University College London Hospital/University College London, which received a proportion of funding from the Department of Health's National Institute of Health Research Biomedical Research Centres funding scheme.
-
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.
- Accepted March 6, 2008.
- Received November 2, 2007.
- Revision received March 6, 2008.