Characteristics of BRCA1 mutations in a population-based case series of breast and ovarian cancer

Breast and ovarian cancers account for approximately 210 000 newly diagnosed cases per year. More than half a million American women are estimated to be carriers of a breast cancer susceptibility gene. The purpose of this study was to assess the association of characteristics such as, age at diagnosis, race/ethnicity and family history of cancer with inherited BRCA1 mutations in a population-based sample of breast and ovarian cancer cases. No selection was made by race, age at diagnosis or positive family history of breast or ovarian cancer. The population under study was all breast cancer cases diagnosed in Orange County, CA, during the 1-year period beginning 1 March 1994 and all ovarian cancer cases diagnosed in Orange County during the 2-year period beginning 1 March 1994. This report focuses on the ﬁrst consecutively ascertained 802 participating probands enrolled in the study, of which 9 were male breast cancer probands, 673 were female breast cancer probands and 120 were ovarian cancer probands. We observed 11 BRCA1 mutations or 1.6% (95% CI: 0.8–2.9) among the 673 female breast cancer probands and 4 BRCA1 mutations or 3.3% (95% CI: 0.8–8.3) among the 120 ovarian cancer probands. No BRCA1 mutations were identiﬁed among the 98 non-white breast and ovarian cancer probands. The prevalence of BRCA1 mutations in non-Hispanic-white breast cancer cases below the age of 50 years was 2%. Positive family history of breast or ovarian cancers was signiﬁcantly associated with BRCA1 mutation status among breast cancer probands. Similarly, positive family history of breast or ovarian cancer was signiﬁcantly associated with BRCA1 mutation status among the ovarian cancer probands. In summary, we present results on the prevalence of BRCA1 mutations in a signiﬁcantly larger sample of population-based breast and ovarian cancer cases than previously reported. The results indicate that, using a conservative approach to targeted genotyping of BRCA1 , the frequency of mutations was consistent with those reported using similar methods of population-based case ascertainment. # 2000 Published by Elsevier Science Ltd. All rights reserved.


Introduction
Breast cancer is the most common cancer and ovarian cancer is the ®fth most common cancer in women in the US. Breast and ovarian cancers account for approximately 210 000 newly diagnosed cases per year [1]. Of the 180 000 new breast cancer cases diagnosed yearly in the US, the estimated number of cases associated with a breast cancer susceptibility gene(s) is more than 10 000 [2,3]. More than half a million American women are estimated to be carriers of a breast cancer susceptibility gene. Some of these women may have a signi®cantly increased risk of developing breast cancer and perhaps additional cancers over their lifetime depending on the degree of family history of cancer [2,4]. For these women, a prevention and early detection strategy is an essential aspect of managing their risk.
Early genetic linkage studies using multiple case families have suggested that mutations in the BRCA1 gene account for 45% of hereditary breast cancer and 80±90% of breast and ovarian cancer [4±7]. Based on some of these studies, it was determined that BRCA1 mutation carriers had a lifetime risk of 85% for breast cancer and 63% for ovarian cancer. More recent studies have included breast cancer families that are not highly selected for their strong family history of breast and ovarian cancer. These studies suggest that risk estimates are far lower than previously reported [8±15]. These studies also suggest that mutations in BRCA1 are attributable to only 16% of women with a family history of breast or ovarian cancer [13]. The study by Whittemore and colleagues [14] indicated that the risk of ovarian cancer among carriers is 27.8% based on populationbased case±control studies of ovarian cancer.
In the present study, we describe the frequency of BRCA1 mutations in a population-based series of 682 consecutive breast cancer probands and 120 consecutive ovarian cancer probands. We further describe the characteristics of breast and ovarian cancer probands with BRCA1 mutations with respect to age at diagnosis, demographic characteristics including race and ethnicity, family history (presence of breast or ovarian cancer among ®rst-degree or ®rst-and second-degree relatives of the proband) and presence of multiple primary cancer. Thus, the purpose of the present study was to assess the characteristics of inherited BRCA1 mutations in a population-based sample that was not selected by race, age at diagnosis or family history of breast or ovarian cancer.

Patients
The population under study is all breast (female and male) cancer cases older than 18  Protocols for population-based ascertainment of breast and ovarian cancer have been established [21]. Breast cancer probands were identi®ed within 6 months of diagnosis. However, for ovarian cancer probands we have incorporated a rapid reporting system to identify ovarian cancer cases within 4 weeks of diagnosis. Physicians were noti®ed that their patient(s) would be contacted regarding study participation. This was followed by a letter of introduction sent to the patient, then by a telephone interview regarding the family history of cancer. Positive family history was determined as at least 1 ®rst-degree relative with breast or ovarian cancer or no ®rst-degree relative with breast or ovarian cancer but at least 2 second-degree relatives with breast or ovarian cancer on the same side of the family. Trained interviewers elicited information on dates of birth, types of cancer and diagnosis and death dates on all ®rst-and second-degree relatives and ®rst cousins of the proband, including both aected and unaected family members. After the subject signed a consent form, an 18-ml blood sample was drawn, and the proband completed an epidemiological risk factor questionnaire. All reported malignancies among family members were veri®ed whenever possible by obtaining pathology reports, tumour tissue, clinical records, interview of another family member besides the proband and, if applicable, death certi®cates.
The population ascertained consisted of 2030 probands, of whom 342 were ovarian cancer cases, 17 male breast cancer cases, and 1671 female breast cancer cases. All probands were invited to participate in this breast and ovarian cancer family registry study. The distribution of the probands by race/ethnicity shows that 86.6% were non-Hispanic white, 7.2% Hispanic, 0.5% African-American and 5.7% Asian. The age distribution of contacted cases was: 8% below the age of 40 years, approximately 21% in each 10-year interval from 40 to 80 years of age and 10% for 81 years of age or older. More than 78% of the patients agreed to complete the family history interview. For breast cancer, 362 patients declined to participate (361 females and 1 male), and for ovarian cancer, 70 patients declined to participate. Of the non-participants, the most common reasons given for non-participation were`not interested' (40%),`bad health' (20%) and`time commitment' (19%). Participants and non-participants did not dier signi®cantly with respect to age and race/ethnicity. Details about participation status of the above-referenced population were presented by Ziogas and colleagues [22].
Overall 1598 probands were enrolled in the study (16 male cancer, 1310 female breast cancer and 272 ovarian cancer cases). This report focuses on the ®rst consecutively ascertained 802 participating probands enrolled in the study, of which 9 were male breast cancer probands, 673 were female breast cancer probands and 120 were ovarian cancer probands. Approximately 4.5% of breast cancer probands were Ashkenazi Jewish and among the ovarian cancer probands 3.3% were Ashkenazi Jewish. 29% of the breast cancer probands were under the age of 50 years at diagnosis, and approximately 38% of the ovarian cancer probands were under the age of 50 years at diagnosis. Among the families with no family history of breast or ovarian cancer, there was an average of 7.6 female ®rst-or second-degree relatives compared with 8.5 among families with positive family history.
Reported malignancies for the study in probands and family members were veri®ed by obtaining pathology reports, clinical records, an interview of another family member in addition to the proband and review of death certi®cates. All probands' tumours were veri®ed by pathology. 76% of ®rst-degree relatives aected with breast or ovarian cancer were veri®ed, whereas approximately 65% of aected second-degree relatives were veri®ed. Details about veri®cation of cancer among ®rst-and second-degree relatives in the population under study were presented in Ziogas and colleagues [22].

Laboratory methods
The details of the laboratory procedures have been described previously [23,24]. Allele-speci®c oligonucleotide (ASO) assays for BRCA1 mutations 185delAG, 5382insC, R1443X, int5-11T-G, 4184delTCAA, R841W and 2594delC among breast cancer probands were performed using appropriate reverse transcriptase polymerase chain reaction (RT-PCR) or genomic PCR products prepared with nucleic acids puri®ed from buy coat samples as described elsewhere [23]. Of the 120 ovarian cancer cases, 13 were screened for mutations in the coding region of the BRCA1 gene using the same ASO methods as in the breast cancer cases and 107 were screened for mutations usin the Rnase Mismatch Cleavage Assay (NIRCA; Ambion, Austin, TX, USA). Laboratory methods for genotyping of ovarian cancer probands have been presented elsewhere [24].

Statistical analysis
Proportions and 95% con®dence intervals (CIs) for the two groups (breast and ovary) and for subgroups categorised by race, age-at-diagnosis and family history were computed by exact binomial probabilities. Tests for trends were computed by use of unconditional logistic regression.

Results
The characteristics of the 682 breast cancer probands and 120 ovarian cancer probands with respect to BRCA1 status are presented in Table 1. Using the laboratory methods outlined above, where targeted mutations were tested in the breast cancer cohort and a full screening was done on the BRCA1 gene in the ovarian cancer group, we observed 11 BRCA1 mutations or 1.6% (95% CI: 0.8±2.9) among the 673 female breast cancer probands and 4 BRCA1 mutations or 3.3% (95% CI: 0.8±8.3) among the 120 ovarian cancer probands. The mean age at onset in female breast cancer probands was 54.0AE14.5 years among the BRCA1 carriers versus 58.4AE13.0 years among non-carriers. Among the ovarian cancer probands the mean age at onset among BRCA1 carriers was 49.8AE5.7 years compared with 55.5AE15.7 among the non-carriers. Both breast and ovarian cancer probands who were BRCA1 carriers tended to be younger than non-carriers. However, the dierences observed were not statistically signi®cant.
Male breast cancer probands were genotyped for both BRCA1 and BRCA2 gene. No BRCA1 mutations were identi®ed among the nine male breast cancer probands. However, we detected two BRCA2 mutations among the male breast cancer probands. Of the two BRCA2 mutations among the male breast cancer probands, one occurred in a proband with no family history of breast or ovarian cancer among ®rst-or second-degree relatives and the other occurred in a proband with family history of breast cancer in a ®rst-degree relative. The mean age at onset of male breast cancer probands was 67.0AE11.3 years (Table 1), whereas the mean age at onset of the 2 BRCA2 mutation carriers was 61AE1.4 years. Since the number of male breast cancer probands was small, the following analyses are limited to female breast or ovarian cancer families.
No BRCA1 mutations were identi®ed among the 98 non-white breast and ovarian cancer probands. Thus, the proportion of BRCA1 carriers among the 592 non-Hispanic-white breast cancer probands was estimated to be 1.9% (95% CI: 0.9±3.3) and among the 103 non-Hispanic-white ovarian cancer probands was 3.9% (95% CI: 1.1±9.6) ( Table 2). In addition, among the 30 breast cancer Ashkenazi probands, 6.7% were carriers of a BRCA1 mutation. No statistically signi®cant differences were observed among breast or ovarian cancer probands with respect to BRCA1 status and tumour dierentiation, histology or presence of multiple primaries. Since no BRCA1 mutations were identi®ed among the non-white breast or ovarian cancer probands, subsequent analyses include only the non-Hispanicwhite probands. Of the 11 breast cancer probands with BRCA1 mutations, 4 had the 185delAG mutation, 4 had the R841W mutation, 1 had the int5-11T-G, 1 had the 2594delC, and 1 had the 5382insC mutation. Among the 4 ovarian cancer probands with BRCA1 mutations, 1 had the 5382insC mutation, 1 had the R841W, 1 had the 3600del-AAGATACTAGT, and 1 had the 962delCTCA mutation ( Table 3). The mean age at onset of individuals identi®ed with the R841W mutation was 66.0AE9.5 years whereas for individuals with the 185delA-G mutation, mean age at onset was 48.8AE9.2 years.
Positive family history of breast or ovarian cancer was signi®cantly associated with BRCA1 mutation status among breast cancer probands (P=0.018) (Table  4). Similarly, positive family history of breast or ovarian cancer was signi®cantly associated with BRCA1 mutation status among the ovarian cancer probands (P=0.0028). However the frequency of BRCA1 mutations was higher for ovarian cancer probands than for breast cancer probands. In particular, the association of ovarian cancer family history with the presence of BRCA1 mutations in both breast and ovarian cancer probands was statistically signi®cant (P=0.019 and P=0.029, respectively). BRCA1 mutations were responsible for 4.0% of breast cancer probands with positive family history of breast or ovarian cancer and 16% of ovarian cancer probands with positive family history of breast or ovarian cancer. In addition, BRCA1 mutations were responsible for 3 of 15 breast cancer probands (20%) with positive family history of ovarian cancer and 2 of 8 ovarian cancer probands (25%) with positive family history of ovarian cancer.
We also observed statistically signi®cant trends between BRCA1 status and the number of breast and ovarian cancers among ®rst-degree or ®rst-and seconddegree relatives in both breast and ovarian cancer probands ( Table 5). The proportion of BRCA1 positive breast cancer probands with 2 or more ®rst-degree relatives with breast or ovarian cancer was 2 out of 31 (6.5%) compared with 5 out of 437 (1.1%) for probands with negative family history (test for trend P=0.032). Signi®cant trends were also observed for ovarian cancer probands when we included in the family history second-degree relatives with breast and ovarian cancer ( Table 5). When we classi®ed the data considering the age of onset of breast cancer of the aected relatives (less than the age of 50 years and aged 50 years or more), we observed statistically signi®cant trends between   a Positive family history was considered for probands having either at least one ®rst-degree relative with the de®ned phenotype or no ®rst-degree relative and at least two maternal or paternal second-degree relatives. BRCA1 status of the breast and ovarian cancer probands and the presence of a ®rst-and/or second-degree relative with breast cancer below the age of 50 years ( Table 6). In particular, BRCA1 mutations were responsible for 7.8% of breast cancer probands with positive family history of breast cancer diagnosed below the age of 50 years (test for trend P=0.0068) and 28.6% of ovarian cancer probands with such a family history (test for trend P=0.0021).
Of the 11 breast cancer probands who carried BRCA1 mutations, 1 did not report a history of breast or ovarian cancer among ®rst-degree or ®rst-and second-degree relatives (Table 5). However, of the 4 ovarian cancer probands who carried a BRCA1 mutation, all 4 reported a history of breast or ovarian cancer among ®rst-or second-degree relatives.
Colon and stomach cancer among ®rst-or seconddegree relatives of the breast cancer probands was reported more frequently among carriers of BRCA1 mutation compared with families of non-carriers (Table  7). Similarly, in addition to family history of ovarian cancer among ®rst-and second-degree relatives of the ovarian cancer probands, positive family history of kidney cancer was more frequent among mutation carriers compared with non-carriers (Table 7).

Discussion
In this paper we report BRCA1 mutation frequency in 120 population-based ovarian cancer cases. For a majority of cases (107), the mutation analysis was FamHx, family history. a No ®rst-degree relative with breast cancer diagnosed below the age of 50 years. b At least one ®rst-degree relative with breast cancer diagnosed below the age of 50 years. Table 7 Percentages of carriers and non-carriers of BRCA1 mutations with a family history among ®rst-and second-degree relatives for selected cancers

Breast cancer
Ovarian cancer Total carried out through a full screening of mutations throughout the BRCA1 gene using RNase mismatch cleavage assay (NIRCA; Ambion, Austin, TX, USA) method. However, the BRCA1 mutation testing in the breast cancer group in this study was done through screening for targeted mutations using ASO methods. The frequencies of BRCA1 mutations in a populationbased series of non-Hispanic-white breast (592 cases) and ovarian (103 cases) cancer probands found in this study were 1.9% and 3.9% respectively. Several studies in the literature reported BRCA1 frequencies in series of breast cancer patients to evaluate the population mutation prevalence and penetrance as well as lifetime risk of developing breast cancer in mutation carriers and non-carriers. It is clear that there is a wide range of variability in these estimates, which may be attributed to dierences in the characteristics and the origin of study populations. Thus, it is important when making comparisons in mutation frequencies between studies to do so only if the study populations are comparable at least with respect to age distribution, ascertainment source (high-risk versus population-based series), and ethnic/racial origin. In this study, we found that the overall frequency of BRCA1 mutations in population-based, non-Hispanic-white breast cancer patients was 1.9%. The age range of our non-Hispanicwhite breast cancer series was older than 18 years. Newman and colleagues [9] reported a frequency of 3.3% in a series of breast cancer cases ranging in age from 20±74 years. When we compared the frequency for non-Hispanic-white women below the age of 40 years, in our study 6.1% (data not included in tables) were found to have BRCA1 mutations compared with 3.6% in the Australian women in the same age group cited in Newman [9]. Furthermore, in breast cancer probands below the age of 50 years, the frequency of BRCA1 mutations in our study was 2% compared with 1.4% reported by Newman [9]. Malone and colleagues [25] reported a 6.2% frequency of BRCA1 mutations in cases diagnosed below the age of 35 years, none of whom were selected on the basis of family history, and a 7.2% frequency in cases diagnosed below the age of 45 years who had a ®rst-degree relative with breast cancer.
Clearly, the age of the groups studied is important in estimating the frequency of BRCA1 mutations. Among breast cancer probands, we estimated the probability of developing breast cancer among carriers by the age of 40 years to be 0.17 and by the age of 70 years to be 0.48 [22]. It is important to point out that the mutation frequencies among the breast cancer group are probably underestimated since the mutation analysis was targeted to speci®c mutations using ASO methodology. Although this is a potential weakness in this study, a full sequencing of the BRCA1 gene in a large population-based series is prohibitively expensive and the yield is low, particularly in population-based series such as in this study.
Most of the current literature on BRCA1 mutations in ovarian cancer has been concerned with its occurrence in Ashkenazi Jews [26,27], and frequencies in other populations were only based on estimates from family studies [8,5]. In our study of 103 population-based, non-Hispanic-white women with ovarian cancer, the frequency of BRCA1 using full sequencing of the gene [24] was found to be 3.9% (95% CI: 1.2±9.6). No mutations were found in the age groups below 40 years or above 60 years of age. The mutation frequency in women of 40±49 years of age was 3/29 (10.3%) and for women 50± 59 years was 1/24 (4.2%).
In 81 non-white breast cancer probands and 17 nonwhite ovarian cancer probands there were no mutations found in BRCA1. This could possibly be due to the small number of non-whites in this study (4 African-American, 42 Hispanic and 35 Asian breast cancer patients and 12 Hispanic and 5 Asian ovarian cancer patients). Mutations in BRCA1 were previously identi-®ed in high-risk African-American families [28,29]. However, in a series of 76 African-American women with breast cancer, no mutations were detected [9]. In the study by Newman and colleagues [9], the cases were selected from a population-based registry and, therefore, it is likely that BRCA1 mutations in African-American women associated with breast cancer are extremely rare and probably limited to very high-risk families with early-onset disease similar to those included in the study by Gao and colleagues [29].
Cases that were positive for BRCA1 in our study were younger than those who were mutation-negative. The mean age at onset for female breast cancer probands who were BRCA1 positive was 54 years compared with 58.4 years in BRCA1 mutation-negative cases. For ovarian cancer, BRCA1 mutation-positive probands were younger than mutation-negative probands, 49.8 compared with 55.5 years. The dierence was not statistically signi®cant, and the dierence of 6 years in the age of onset among carriers and non-carriers might not be epidemiologically meaningful. Similar data were observed in a study of BRCA1 mutations among Jewish women with ovarian cancer, where mutation-positive patients had a mean age of 48.3 years while negative patients had a mean age of 55.5 years [30]. Thus, although the frequency of BRCA1 mutations in ovarian cancer in Ashkenazi Jews was higher than in our non-Hispanic-white probands, the ages of onset were quite similar. In this study the frequency of BRCA1 mutations in the Ashkenazi subgroup of breast cancer probands was 6.7%.
The age of onset may also vary between cases with speci®c mutations. In our study, the mean age at onset of individuals identi®ed with the R841W mutation was 66 years, whereas for cases with the 185delAG mutation the mean age at onset was 48.8 years. We did not ®nd dierences in the mutation frequency when we classi®ed our patients by histology or dierentiation. It may be of importance to report speci®c mutations and patient and tumour characteristics to further de®ne the BRCA1 mutation-positive phenotype.
Positive family history of breast or ovarian cancer was signi®cantly associated with BRCA1 mutation status among breast cancer probands (Table 4). Similarly, positive family history of breast or ovarian cancer was signi®cantly associated with BRCA1 mutation status among the ovarian cancer probands. However, the frequency of BRCA1 mutations was higher for ovarian cancer probands than for breast cancer probands.

Conclusions
In summary, the frequencies of BRCA1 mutations in this population-based study of breast cancer were consistent with those reported using a similar approach and similar study populations ascertainment. There were no BRCA1 mutations found in non-white breast and ovarian cancer probands. In this study we found that the frequency of BRCA1 mutations in ovarian cancer patients was signi®cantly higher than that for breast cancer for the same age groups however the number of mutation carriers was small. In addition, the age of onset of cases with positive BRCA1 mutations was younger than cases negative for BRCA1 mutation. Furthermore, the frequency of BRCA1 mutations in ovarian cancer in Ashkenazi Jewish cases in our study group was higher than in our non-Ashkenazi Jewish/non-Hispanic-white probands, although the age of onset was quite similar in the two groups. The frequency of BRCA1 mutations in the Ashkenazi subgroup of breast cancer probands was 6.7%. Finally, the proportion of BRCA1 positive breast cancer probands with 2 or more ®rst-degree relatives with breast or ovarian cancer was 6.5% compared with 1.1% for probands with negative family history. In particular, the BRCA1 mutations were responsible for 7.8% of breast cancer probands with positive family history of a ®rst-degree relative with breast cancer diagnosed at an age below 50 years and 28.6% of ovarian cancer probands with such a family history. In addition, BRCA1 mutations were responsible for 20% of breast cancer probands with positive family history of ovarian cancer and 25% of ovarian cancer probands with positive family history of ovarian cancer.