Racial and ethnic variation in BRCA1 and BRCA2 genetic test results among individuals referred for genetic counseling at a large urban comprehensive cancer center

The prevalence of pathogenic variants in BRCA1 and BRCA2 in populations other than Ashkenazi Jewish (AJ) is not well defined. We describe the racial and ethnic-specific prevalence of BRCA1/2 pathogenic variants and variants of uncertain significance (VUS) among individuals referred for genetic testing in a large urban comprehensive cancer center over a 20-year period. The population included 3,537 unrelated individuals who underwent genetic testing from January 1999 to October 2019 at the Karmanos Cancer Institute. We estimated the prevalence of pathogenic variants and VUS and evaluated associations with race and ethnicity for African American (AA), Arab, AJ and Hispanic individuals compared to Non-Hispanic Whites (NHW). We used multivariable models to adjust for other predictors of pathogenic variants. We also reported the most common pathogenic variants by racial and ethnic group. The racial and ethnic breakdown of our population was: NHW (68.9%), AA (20.3%), AJ (2.5%), Arab (2.2%), Hispanic (1.0%), Asian Pacific Islander, Native American/Alaskan Native (4.7%), and < 1% unknown. The overall prevalence of pathogenic variants in BRCA1/2 was 8.9% and the prevalence of VUS was 5.6%. Compared to NHW, there were no racial or ethnic differences in the rate of pathogenic variants. However, AA individuals were more likely to have VUS in BRCA1 (adjusted OR 2.43, 95% CI 1.38–4.28) and AJ were more likely to have VUS in BRCA2 (adjusted OR 3.50, 95% CI 1.61–6.58). These results suggest the continued need for genetic testing and variant reclassification for individuals of all racial and ethnic groups.


Introduction
Breast cancer is the most commonly diagnosed female malignancy in the United States and the second leading cause of cancer-related deaths in women after lung cancer, accounting for 15% of all cancer deaths [1]. The majority of breast cancer is sporadic, influenced by an interplay of several modifiable and non-modifiable risk factors. However, about 5-10% of individuals diagnosed with breast cancer have an inherited pathogenic variant in one or more cancer susceptibility genes [2]. Highly penetrant pathogenic germline variants in BRCA1 and BRCA2, inherited in an autosomal dominant pattern, account for the majority of hereditary breast cancer risk [3,4]. Since the landmark supreme court ruling invalidated the patenting of BRCA 1 and BRCA2 Nadine Abdallah and Kristen S. Purrington have contributed equally to the manuscript and should be considered co-first authors. genetic tests in 2013, genetic testing using next generation sequencing (NGS) has been widely available through several commercial labs for individuals at high risk for hereditary breast and ovarian cancer (HBOC) [5]. Testing options include targeted testing for site-specific BRCA1/2 pathogenic variants in the presence of known familial or founder mutations, or comprehensive sequencing and deletion/duplication testing of the BRCA1/2 genes.
Several studies have previously demonstrated ethnic variation in the prevalence of specific BRCA1/2 mutations [6][7][8][9]. Several mutations have been identified to occur at higher frequencies among different ethnic groups, termed "founder" mutations [10]. In the Ashkenazi Jewish (AJ) population, three founder mutations account for 90% of all BRCA1 and BRCA2 mutations [11]: 185delAG and 5382insC mutations in BRCA1 and the 6174delT mutation in BRCA 2 [12]. These mutations have also been identified at lower frequencies in other populations and ethnic groups [13]. Similarly, other recurring variants have been reported in different populations including Europeans and Hispanics [10,[14][15][16][17]. However, unlike the AJ population, specific mutation prevalence is less clearly defined in other racial and ethnic groups. As with pathogenic variants, ethnicity-specific differences have also been observed in the prevalence of variants of uncertain significance (VUS) [8].
Studies of genetic test results inclusive of a wide range of ethnic and racial groups are important for understanding the prevalence of specific variants in different racial and/or ethnic groups and in identifying recurring or novel pathogenic variants. These studies also have the potential to allow for analyses of novel variants in larger populations to better characterize penetrance and reclassify VUS. In this study, we describe the racial and ethnic-specific prevalence of BRCA1 and BRCA2 pathogenic variants and VUS among individuals referred for genetic testing in a large urban NCI designated comprehensive cancer center over a 20-year period and report common pathogenic variants by racial and ethnic group.

Study population
This is a retrospective analysis utilizing the Karmanos Cancer Institute (KCI) Cancer Genetic Counseling Service (CGCS) database, which includes data for individuals referred for evaluation and consideration of genetic testing for HBOC. The KCI is a National Cancer Institute designated Comprehensive Cancer Center located in Detroit, Michigan. Individuals were evaluated through KCI's CGCS at the downtown and Farmington Hills sites since 1 January 1999, and at 7 additional clinical sites across the State of Michigan by 2020. Our dataset includes both men and women, aged 18 years and older, with or without a personal history of cancer, who were referred for cancer genetic counseling for the possibility of HBOC from 1 January 1999 to 31 October 2019. All clinical information was stored in a secure electronic database. The study was approved by the Institutional Review Board.

Genetic test results
All genetic testing was ordered by KCI's CGCS according to testing guidelines utilized at the time of the genetic consultation. Testing was initially performed through Myriad Genetics Inc® (Salt Lake City, UT) until 13 June 2013, and thereafter in addition, ordered through other commercial genetic test laboratories including Ambry Genetics® and Invitae®. Initially, decisions on genetic testing were made utilizing BRCAPRO [18,19] and the Myriad Tables, [20,21]; by 2012, decisions on genetic testing were made based on NCCN guidelines [22].
Initially, genetic testing ordered by the CGCS included sequence analysis of the BRCA1 and BRCA2 genes as available through Myriad; in 2002 Myriad added the 5-site rearrangement test panel for selected cases, followed by large rearrangement testing (BART™) in 2006. Since May of 2013, all genetic testing has been completed utilizing NGS panels. Genetic test results reported here are only for BRCA1 and BRCA2 and were reported as negative, positive for a pathogenic variant, or as a VUS.

Socio-demographic and clinical variables
We extracted demographic data and data on personal cancer history including the type of cancer and date of diagnosis, family history of breast, ovarian or other cancers, and self-reported racial and ethnic group including: Non-Hispanic White (NHW), African American (AA), Arab, Ashkenazi Jewish (AJ), Hispanic, and others (Asian/Pacific Islander, Native American/Alaskan Native, Mixed Race and Unknown). For patients with mixed ancestry, the ethnicity category used in our analysis was based when applicable on the side of the family from which a pathogenic variant was inherited. Family cancer history was collected using a threegeneration pedigree and was stored in Progeny [23], which was electronically linked to the database.

Statistical analysis
Among individuals who underwent genetic testing, we estimated the prevalence of BRCA1 and BRCA2 pathogenic variants and VUS by race and ethnicity using frequency tables and percentages. When multiple members of the same family had genetic testing and were found to be carriers of a pathogenic BRCA1/2 variant, we only included the first individual identified in the family in our analysis. We conducted multinomial logistic regression to estimate odds ratios (ORs) and 95% confidence intervals (CIs) for associations between race and ethnicity and variant status. We also performed multivariable analyses to adjust for potential predictors of genetic test results including age at testing, gender, personal and family history of cancer associated with HBOC (yes vs. no), and the type of genetic test performed (BRCA1/2 only vs. panel testing). In addition, we reported the most common specific BRCA1/2 pathogenic variants in each racial and ethnic group. All statistical tests were two-sided, with a p-value of < 0.05 considered to be statistically significant. All data were analyzed using R statistical software [24].

Results
The KCI Genetics Cohort included 7,703 individuals evaluated for high risk of hereditary disease between 1 January 1999 and 31 October 2019. Among those, 3,537 (46%) unrelated individuals who underwent BRCA1/2 genetic testing for suspected HBOC were included in the analysis (Fig. 1). The racial and ethnic distribution of the genetic testing cohort was 68.9% NHW, 20.3% AA, 2.5% AJ, 2.2% Arab, 1% Hispanic, 4.7% Asian/Pacific Islander and American Indian/Alaskan Native, and < 1% (n = 6) unknown. Table 1 shows genetic test results and clinical characteristics of the KCI genetic testing population stratified by race and ethnicity. The rate of pathogenic variants in BRCA1/2 was 8.9% (4.1% in BRCA 1 and 4.8% in BRCA2), and the overall rate of VUS was 5.6% (1.8% in BRCA1 and 3.8% in BRCA2). The mean age at genetic testing was 52.5 (SD 13.4) years; 53.8% of patients were at least 50 years of age, including 28.3% ≥ 60 years and 9.2% ≥ 70 years. The cohort consisted of unaffected probands (36.9%), and others with a personal history of breast (44.5%), ovarian (5.5%), colorectal (3.2%), endometrial (2.8%), pancreatic (1.6%), skin (1.2%), prostate (1.1%), and thyroid cancer (1.0%). There were 2,269 individuals who had a family history of cancers related to HBOC including 81.9% with a family history of breast cancer, 40.5% with a family history of early onset breast cancer (diagnosed at < age 50), 23.8% with a family history of ovarian cancer, 16.4% with a family history of breast and ovarian cancer, 34% with prostate and 19.5% with pancreatic cancer. Only two individuals reported a family history of male breast cancer.
Compared to NHW, AA, Arab and Hispanic individuals who underwent genetic testing were more likely to be younger than 50 years of age, and to have a personal history of breast cancer and/or other cancers associated with HBOC. AAs and Arabs who underwent testing were also more likely to have a family history of breast cancer at < 50 years compared to NHWs. AJ individuals who underwent testing were more likely to be older, have no personal cancer history, and more likely to have a family history of breast cancer and other cancers associated with HBOC compared to other groups.
We evaluated the relationship between racial and ethnic group and genetic test results for BRCA1 and BRCA2 pathogenic variants and VUS. These results are presented in Table 2. In the unadjusted model, AJ individuals were 2.5-fold more likely to have a pathogenic BRCA2 variant compared to NHWs (95% CI 1.21-5.09, p = 0.013). This effect was attenuated after adjustment for confounders and was not statistically significant (OR = 1.59, 95% CI 0.62-4.06, p = 0.33). Otherwise, there were no significant differences in the prevalence of pathogenic variants in BRCA1 or BRCA2 by race or ethnic group in both the unadjusted and adjusted models.
To better understand the relationship between BRCA variants and the different cancer types, we evaluated the prevalence of BRCA1 and BRCA2 pathogenic variants and VUS by cancer site. These results are illustrated in Fig. 2. BRCA1 pathogenic variants were observed among individuals with breast, ovarian, colon, renal, and pancreatic cancers. Pathogenic variants for BRCA2 were observed among individuals with a wider range of cancer types, including breast, colon, thyroid, renal, ovarian, endometrial, skin, pituitary, and pancreatic cancers. Among individuals with breast cancer (n = 1,574), pathogenic BRCA1 variants were found in 3.9% and pathogenic BRCA2 variants in 4.8% of those tested; VUS were seen in 2.0% and 4.1% in BRCA1 and BRCA2, respectively. Among women with ovarian cancer (n = 195), 5.6% and 6.2% had pathogenic mutations in BRCA1 and BRCA2, respectively; the VUS rate was 1.5% for BRCA1 and 3.1% for BRCA2. The rates of pathogenic variants in BRCA1/2 among individuals with other cancers were 1.8%, 11.4% and 6.9% for colon (n = 112), skin (n = 44) and pancreatic cancers (n = 58), respectively. The VUS rates were 4.5%, 6.8%, and 3.4%, respectively. Among individuals with no personal history of cancer (n = 1,304), 4.6% had a pathogenic variant in BRCA1, 5.1% had a pathogenic variant in BRCA2, and 4.8% had a VUS in BRCA1 or BRCA2.

Discussion
Ethnic differences in the frequency of BRCA1 and BRCA2 mutations have been reported in several previous studies, with variability in prevalence estimates attributed at least in part to the heterogeneity of populations under study in terms of personal and family cancer history, and to variability in the definitions and classifications of ethnic groups [6,7,9,25,26]. In this study, we report an overall rate of pathogenic variants in BRCA1 and BRCA2 combined of 8.9%, which varied by race and ethnicity, ranging from 1.2% in AJ individuals to 6 [9]. In a more recent analysis including 1,090 individuals seen for cancer genetics at a tertiary community hospital in Michigan from 2008 to 2018, AAs had an increased prevalence of BRCA1/2 mutations compared to non-AJ Whites (8.1% vs 3.6%, p = 0.020) [26]. In our study, no differences were seen in the odds of pathogenic variants in BRCA1 or BRCA2 in all racial/ethnic groups. For VUS, the overall rate observed in our population was 5.6% (1.8% in BRCA1 and 3.8% in BRCA2), which is similar to that reported by Hall et. al (6.6%). We found a higher likelihood of VUS among AA individuals for BRCA1 and in AJ individuals for BRCA2 compared to NHW. This is consistent with results from previous studies where prevalence of VUS was higher among minority groups; in the Hall et al. study, AA women had the highest prevalence of VUS, compared to women of Western European descent (16.5% vs 5.7%) [9]. Similarly, a study by Eggington et (2) c.1813dupA (2041insA) (p.Ile605Asnfs) (7) c.1929del (2157delG) p.Arg645fs (2) c.2808_2811del (3036del4) p.Ala938Profs (2) c.6591_6592del (6819delTG) p. Glu2198fs (2) c.5656C > T (5884C > T) p. Gln1886Ter (2) c.3545_3546del (3773delTT) p.Gln1181_ Phe1182insTer (2) c.5857G > T (6085G > T) p.Glu1953Ter (3) c.9371A > T (9599A > T) p.Asn3124Ile (2) African American c.5173_5176GAAA (5296del4) p.Arg1726fs (2) c.4986 + 6 T > C (IVS16 + 6 T > C) ( among populations of African origin [8]. While our sample size of individuals of Arab background was small, it was relatively larger than in other studies, and we showed a higher, although non-significant, rate of pathogenic variants in individuals of Arab background and a higher, although non-significant, rate of BRCA1 VUS. Ethnic differences in the prevalence of VUS are likely due to limited availability of data in populations with lower rates of genetic testing. In this study, we observed that AA, Arab and Hispanic individuals who underwent genetic testing had more risk factors for HBOC compared to NHW and AJ individuals who underwent testing; this may reflect a disparity in referral patterns, lower uptake of genetic testing among the referred population for ethnic minorities compared to NHWs, and/or higher rates of cancers associated with HBOC and younger age of cancer diagnosis in these groups. We also looked at the most common pathogenic variants in the different racial and ethnic groups. All pathogenic variants seen in our AJ population were founder mutations. Among non-Hispanic whites, 5385insC was the most frequently detected pathogenic variant in BRCA1, which was also the most common pathogenic variant seen among Western and Central European populations in the study by Hall et al., and is one of the founder AJ variants [9]. Similarly, the most common BRCA1 pathogenic variants seen among AAs in our study (5296del4, IVS16 + 6 T > C, and 943ins10) are consistent with those reported in previous studies in this group [9,18]. The only BRCA1 pathogenic variant identified in Hispanics in our dataset was R71G which is a Spanish founder mutation [27], and among the common pathogenic variants reported by others in this group. [6,9,28]. Among Arabs, the most common BRCA1 mutation observed in our study was c.34C > T (153C > T) which was reported in one of seven patients with pathogenic variants in a study by Saghir et al. including 250 Lebanese women with breast cancer considered at risk for harboring a BRCA mutation [29]. Similarly, the most common BRCA2 mutation found in our Arab cohort (c.5576_5579del) in our study was also reported in one of seven patients with BRCA2 pathogenic variants in the same study by Saghir et al. [29]. The most common BRCA2 mutation seen among NHW in our study [c.1813dupA (2041insA)] is a recurring Fig. 2 Prevalence of BRCA1 and BRCA2 variants by cancer site. The number of individuals with personal cancer history who have no mutations (green bar), pathogenic variants (orange bar), or variance of uncertain significance (purple bar) in a BRCA1 and b BRCA2 by cancer site mutation reported in European populations with a possible founder effect [30]. This variant has also been reported in a large study including unselected women with a diagnosis of ovarian cancer [31]. Similarly, the other BRCA2 pathogenic variants we observed have previously been reported in European populations, some of which have a founder effect [c.1929del (2157delG), c.2808_2811del (3036del4), c.5857G > T (6085G > T)] [30,[32][33][34][35][36][37]; The c.5656C > T (p.Gln1886Ter) mutation has been reported in the Korean population [38]. The BRCA2 mutations seen in the AA population in our study are also consistent with those seen in AA populations in other studies [39][40][41]. The only BRCA2 pathogenic variant seen in Hispanics in our population [c.3922G > T (p.Glu1308Ter)] is a Puerto Rican founder mutation [33,42].
Limitations of our analysis include the fact that our results are based on clinical referrals rather than a population-based cohort, and are limited by the small number of tested individuals among race and ethnic groups other than NHW and AA. This may may be attributed to differences in referral rates by race and ethnicity and/or the racial and ethnic differences in the personal and family risk characteristics of referred patients. In addition, racial and ethnic group classification was based on self-report. However, the data presented from our cohort contributes to existing literature by providing results of clinic-based testing in a diverse population over a long period of time, using prospectively collected data which includes comprehensive data on self-identified race and ethnicity and personal and family cancer history in different racial and ethnic groups. There is a continued need for additional studies to enhance our understanding of the prevalence and genetic spectrum of pathogenic variants in ethnic minority groups, and to uncover the barriers to genetic testing. Such understanding has future implications on customization of genetic testing, classification of new founder mutations, and guiding treatment and surveillance strategies.