To the Editor:

Section:

ChooseTop of pageTo the Editor: <<REFERENCES

Knowledge of BRCA1/2 mutation status plays an important role in multiple aspects of oncologic care.¹ Inherited mutations in BRCA1/2 are highly penetrant for hereditary breast and ovarian cancer syndrome, with lifetime risks—dependent upon the gene and population—of 50% to 70% for breast and 15% to 55% for ovarian cancer.² BRCA1/2 mutations also lead to increased risks of other malignancies, such as prostate, pancreatic, and male breast cancers, although penetrance is lower.

Germline BRCA1/2 testing has traditionally been guided by personal and family cancer history¹; however, two developments are changing the landscape of BRCA1/2 testing. First, the increased use of tumor genomic sequencing is leading to identification of germline BRCA1/2 mutations in patients without typical personal or family histories.³ Tumor genomic sequencing panels that contain BRCA1/2 that do not subtract matched germline DNA are fundamentally opportunistic screenings of patients with cancer. Second, for populations with a high frequency of BRCA1/2 mutation carriers,⁴ screening of all individuals has been proposed.⁵ For consideration of population-based BRCA1/2 testing, the expected frequency of a positive result is a required piece of critical information.

It has been reported that the population frequency of pathogenic BRCA1/2 mutations is 1:400, with the exception of populations with high frequency founder mutations, such as the Ashkenazi Jewish population.⁶ We therefore noted with interest that the heterozygote frequency of pathogenic BRCA1/2 mutations in the control population of the study by Thompson et al⁷ was 0.65% (1:153); BRCA1 0.20% (1:500) and BRCA2 0.45% (1:222). It is possible that the prevalence of BRCA1/2 mutations in the control population—Australian women in the Lifepool study—is greater than that of a true population-based ascertainment; however, these are cancer-free women who were ascertained via a population-based mammographic screening program, which suggests that the prevalence could be similar to that of the general population. We therefore examined publicly available data from the Exome Variant Server (EVS)⁸ and the Exome Aggregation Consortium (ExAC) database⁹ to calculate the germline BRCA1/2 heterozygote frequency in these large populations. BRCA1/2 known pathogenic missense mutations¹⁰ and all protein-truncating variants were identified, and variants were excluded if they were distal to the last known protein-truncating variants in the corresponding gene, led to predicted protein truncation in a noncanonical transcript, or were inserted or deleted nucleotide(s) in a repetitive region of the same nucleotide(s), as such regions are subject to sequencing error by using massively parallel sequencing technology.

BRCA1 and BRCA2 population frequencies in ExAC were 0.26% (1:381) and 0.36% (1:277; combined 0.62% [1:161]; Fig 1 and Table 1). When samples from The Cancer Genome Atlas (TCGA) were removed, frequencies were 0.15% (1:646) and 0.26% (1:390; combined 0.41% [1:243]). However, there was substantial variability between populations; BRCA1 and BRCA2 frequencies excluding TCGA samples were 0.21% (1:480) and 0.31% (1:327; combined 0.51% [1:195]) in individuals of European non-Finnish descent, but 0.08% (1:1251) and 0.24% (1:417; combined 0.32% [1:313]) in the Finnish population. This difference, in part, is the result of contributions of Ashkenazi Jewish founder mutations, which are present in ExAC only in the non-Finnish population. Although nonindependent from ExAC, EVS also does not include TCGA data, and in this noncancer cohort, BRCA1 and BRCA2 frequencies were 0.11% (1:924) and 0.49% (1:203; combined 0.60% [1:166]). There are only two Ashkenazi Jewish alleles in EVS; however, 14 of 31 pathogenic BRCA2 alleles are one of three founder mutations (Southern European, Scandinavian, or Pakistani). BRCA1/2 population frequencies greatly varied among African, Latino, East Asian, and South Asian populations in ExAC, both with TCGA samples included and excluded (Fig 1 and Table 1).

Fig 1. Graphical representation of BRCA1/2 heterozygous population frequencies in Exome Variant Server (EVS) and Exome Aggregation Consortium (ExAC) databases. All known pathogenic missense and all protein-truncating variants (PTVs) were identified from EVS and ExAC databases for BRCA1/2 and filtered as per the main text. PTVs were excluded from analysis for the following reasons: eight indels in EVS were insertion or deletion of one or more nucleotide(s) in a repeat of the same nucleotide(s); three PTVs in EVS and 10 in ExAC were distal to the last known pathogenic mutation in the gene; and three variants in ExAC were predicted to be PTVs only in a noncanonical transcript. The minor allele frequencies of these PTVs were summed to obtain the value of q. Heterozygote prevalence was calculated by using the equation 2q(1−q). Heterozygous prevalence is plotted as a percentage. AA, African American; AFR, African; EA, European American; EAS, East Asian; EUR, European; FIN, European Finnish; LAT, Latino; SAS, South Asian.

Table 1. BRCA1/2 Heterozygote Population Frequencies

View larger version (581K)

These calculations have a number of limitations. They do not incorporate large genomic rearrangements or uncharacterized, but potentially pathogenic, missense mutations and, therefore, could be underestimates of true population frequencies of BRCA1/2 heterozygotes. Furthermore, inclusion of individuals with founder mutations in control groups significantly impacts population frequencies. On the basis of the data in Thompson et al⁷ and this analysis, however, the combined frequency of germline BRCA1/2 mutation carriers may be as high as 1:200 in select populations, and closer to 1:400 in others. In addition, BRCA2 mutations seem to be more frequent than BRCA1 mutations in most populations, and in some populations BRCA1 mutations may be quite rare. Given that the penetrance of BRCA2 mutations is lower² and the associated tumor spectrum potentially wider,³ counseling individuals with mutations that were detected on population screening could be correspondingly more complex. Overall, these data show that in the discussion of population screening for BRCA1/2 mutations, we must determine population frequency estimates in a diverse group of population-based cohorts. Furthermore, population frequencies and penetrance estimates obtained from patients with high-risk familial hereditary breast and ovarian cancer may not be applicable in the discussion of population-based BRCA1/2 testing.