Complete loss of BRCA1 or BRCA2 function is associated with sensitivity to DNA damaging agents. However, not all BRCA1 and BRCA2 germline mutation-associated tumors respond. Herein we report analyses of 160 BRCA1 and BRCA2 germline mutation-associated breast and ovarian tumors. Retention of the normal BRCA1 or BRCA2 allele (absence of locus-specific loss of heterozygosity (LOH)) is observed in 7% of BRCA1 ovarian, 16% of BRCA2 ovarian, 10% of BRCA1 breast, and 46% of BRCA2 breast tumors. These tumors have equivalent homologous recombination deficiency scores to sporadic tumors, significantly lower than scores in tumors with locus-specific LOH (ovarian, P = 0.0004; breast P < 0.0001, two-tailed Student’s t-test). Absence of locus-specific LOH is associated with decreased overall survival in ovarian cancer patients treated with platinum chemotherapy (P = 0.01, log-rank test). Locus-specific LOH may be a clinically useful biomarker to predict primary resistance to DNA damaging agents in patients with germline BRCA1 and BRCA2 mutations.
Approximately 5% of breast and 20% of ovarian cancers arise in women carrying heterozygous germline mutations in the cancer susceptibility genes BRCA1 or BRCA2 1. Germline BRCA1 and BRCA2 mutations in women are associated with an increased lifetime risk of breast and ovarian cancers2, 3. BRCA1 and BRCA2 are critical in double strand break repair utilizing homologous recombination (HR)4, disruption of which leads to high levels of genomic instability in BRCA1 and BRCA2 germline mutation-associated ovary5 and breast6 tumors. These tumors require additional somatic mutations, as in TP53, to suppress induction of DNA damage cell-cycle checkpoints, as otherwise genomic instability leads to cell-cycle arrest or apoptosis7.
BRCA1 and BRCA2 are canonical tumor suppressor genes; loss of the non-mutated (wild-type) allele at the BRCA1 or BRCA2 locus, termed locus-specific loss of heterozygosity (LOH) is observed in tumors8, 9. Cells with complete loss of BRCA1 or BRCA2 function and resultant HR-based DNA repair deficiency (HRD) have exquisite sensitivity to DNA damaging agents, such as platinum-based chemotherapeutics10 and PARP inhibitors11, 12. Tumors in BRCA1 and BRCA2 mutation carriers show high sensitivity to these agents in clinical trials13,14,15,16,17. The sensitivity of BRCA-deficient cells to platinum agents is due to the inability of cells to repair damage-induced lesions such as interstrand crosslinks10. PARP inhibitor sensitivity relies, in part, on a synthetic lethal interaction resulting from the inability of BRCA-deficient cells to repair stalled replication forks generated by PARP trapping on DNA18, 19. Cells with heterozygous BRCA mutations are significantly less sensitive to platinum agents and PARP inhibitors than cells with homozygous mutations, both in vitro12, 20,21,22,23,24 and in mouse models24, suggesting that complete loss of BRCA1 or BRCA2 function is a requirement for efficacy of these therapeutics. When treated with platinum chemotherapy and PARP inhibitors, individuals with and without BRCA1 and BRCA2 germline mutations have the same rates of adverse effects related to cell death in rapidly proliferating tissues, such as the gastrointestinal tract and hematopoietic system, further demonstrating the lack of sensitivity of heterozygous cells15, 16. These data support a requirement for homozygous loss of BRCA1 or BRCA2 function for sensitivity to DNA damaging agents.
Although clinical trials report excellent response rates of tumors in patients with germline BRCA1 and BRCA2 mutations to platinum chemotherapy and PARP inhibitors13,14,15,16,17, primary resistance has been noted25,26,27. Genomic studies have suggested that a subset of germline BRCA1 and BRCA2 mutation-associated tumors may not have BRCA locus-specific LOH28,29,30. Reversion to the heterozygous state and presumed restoration of BRCA1 or BRCA2 function has been noted as a mechanism of secondary resistance20, 21, 31. However, the rates of primary resistance due to maintenance of the heterozygous state (absence of locus-specific LOH) and its relationship to genomic measures and clinical outcomes are currently unknown.
We have performed an in-depth examination of the genomic profiles of primary breast and ovarian tumors in patients with germline BRCA1 and BRCA2 mutations with the goal of identifying correlates of therapeutic response, using two data sets. The first data set was derived from The Cancer Genome Atlas (TGCA). The second independent data set was uniformly generated from patients seen at our institution, and a tissue microarray was available for correlative studies on a subset of tumors. We show that a proportion of BRCA1 and BRCA2 germline mutation-associated tumors do not have locus-specific LOH. Absence of locus-specific LOH is associated with a lack of genomic measures of BRCAness, and, in ovarian cancer, poorer overall survival when treated with platinum chemotherapy. We propose that locus-specific LOH may be an important clinical tool to predict primary resistance to DNA damaging agents in patients with BRCA1 and BRCA2 germline mutated-associated tumors.
Establishment of analysis pipeline
We established an analysis pipeline for identification of genomic markers for BRCA1 and BRCA2 functional deficiency (termed BRCAness19) and BRCA locus-specific LOH from whole exome sequencing (WES) data. Analysis of primary data from TCGA identified 100 breast and ovarian tumors with germline BRCA1 (n = 55) and BRCA2 (n = 45) mutations (Supplementary Fig. 1, Supplementary Data 1, Supplementary Table 1). We created a nonBRCA tumor set (n = 764 breast, n = 215 ovary) from the remaining TCGA tumors excluding tumors with somatic mutations in, homozygous copy loss of, or transcriptional repression of BRCA1 and/or BRCA2, and any breast tumors from patients who had received neoadjuvant chemotherapy.
We identified BRCA mutational signatures32 using deConstructSigs33 and the Somatic Signatures nonnegative matrix factorization (NMF) function34. When TCGA tumors were stratified by mutation and tumor type, both BRCA1 and BRCA2 germline mutation-associated breast and ovarian tumors had a significantly higher proportion of BRCA signature (Signature 3) compared to nonBRCA breast and ovarian tumors (Supplementary Fig. 2a, b). In both analyses, there were no significant differences between any groups of BRCA1 and BRCA2 germline mutation-associated tumors, although nonBRCA ovarian tumors had a significantly higher proportion of BRCA mutational signature than nonBRCA breast tumors (Supplementary Fig. 2a, b). Both BRCA1 and BRCA2 germline mutation-associated breast and ovarian tumors had a significantly lower proportion of the aging signature (Signatures 1 and 5) compared to nonBRCA breast and ovarian tumors (Supplementary Fig. 2c). Only a small proportion of BRCA1 and BRCA2 germline mutation-associated breast and ovarian tumors (n = 3, 3%) and nonBRCA ovarian tumors (n = 8, 4%) had over 20% of mutations attributed to any signatures other than Signatures 1,3, and 5, with Signatures 2 (APOBEC) and 6 (defective mismatch repair) observed. In contrast, 110 (16%) of nonBRCA breast tumors had over 20% of their mutations attributed to other signatures (Signatures 2, 6, 10, 13, 15, 18, or 20) (Supplementary Fig. 2d).
We developed a method to calculate genomic loss of heterozygosity (HRD-LOH)35, non-telomeric allelic imbalance (HRD-NtAI)36, and large state transitions (HRD-LST)37 scores from WES data using Sequenza38 derived allele-specific copy number (ASCN) data (Supplementary Fig. 3a–c). When tumors were stratified by mutation and tumor type, the means of these three scores (HRD-Mean) were significantly higher in BRCA1 and BRCA2 germline mutation-associated breast vs. nonBRCA breast tumors, and between BRCA1 germline mutation-associated and nonBRCA ovarian tumors (Supplementary Fig. 3d). HRD-Mean scores were not significantly different between BRCA1 and BRCA2 germline mutation-associated breast and ovarian tumors. However, nonBRCA ovarian tumors had significantly higher HRD-Mean scores than nonBRCA breast tumors (Supplementary Fig. 3d). Of the TCGA breast and ovarian tumors, 26 and 95% of samples, respectively, underwent whole genome amplification (WGA) of tumor and/or normal DNA prior to WES. WGA is known to affect mutational profiles, copy number calls, and LOH calls39, 40. The proportion of mutations due to the BRCA signature were significantly lower and HRD-Mean scores were significantly higher for breast tumors whose DNA was prepared using WGA (n = 180) compared to those for which WES was performed directly from nascent DNA (n = 635) (Supplementary Fig. 4). Only 12 ovarian cancer TCGA samples did not have WGA performed prior to being profiled.
BRCA locus-specific LOH in TCGA BRCA1 and BRCA2 tumors
We next determined BRCA locus-specific LOH in the TCGA data set using VarScan241 statistical analysis, allele frequency comparisons28, and Sequenza38 ASCN calls of the genomic region containing the BRCA1 or BRCA2 locus (Supplementary Data 2). Fifty-two of 55 BRCA1 germline mutation-associated breast and ovarian tumors had locus-specific LOH (Fig. 1, Supplementary Table 2), with two breast (11%) and one ovarian (3%) tumors lacking locus-specific LOH. In contrast, 10 of 19 (53%) breast and four of 26 (15%) ovarian BRCA2 germline mutation-associated tumors did not have locus-specific LOH (Fig. 1). The most common mechanism of locus-specific LOH was copy neutral LOH in both breast and ovarian tumors (Table 1). Tumors without BRCA locus-specific LOH may inactivate the wild-type allele via alternative mechanisms. One BRCA1 and one BRCA2 breast tumor each had somatic pathogenic mutations in the corresponding gene, at 21 and 35% allele frequency, respectively (Table 1). These two tumors were included in the group with locus-specific LOH for genomic analyses. Promoter methylation analysis of TCGA data showed that BRCA1 promoter methylation status in ovarian and breast tumors correlates with BRCA1 but not BRCA2 expression levels (Supplementary Fig. 5). BRCA1 promoter methylation as a mechanism of inactivation of the wildtype allele in the BRCA1 tumors without locus-specific LOH was not identified.
BRCA locus-specific LOH in Penn BRCA1 and BRCA2 tumors
We performed WES on nascent DNA from 60 primary untreated breast (n = 39) and ovarian (n = 21) tumors from individuals with germline BRCA1 and BRCA2 mutations seen at Penn Medicine. We wanted to further explore BRCA locus-specific LOH and its potential relationship with genomic measures of BRCAness and clinical outcome in a uniformly sequenced data set, not subjected to WGA, with well-annotated clinical characteristics. Three of 23 (13%) breast and three of 15 (20%) ovarian tumors associated with BRCA1 germline mutations, and seven of 16 (44%) breast and one of six (17%) ovarian tumors associated with BRCA2 germline mutations did not have BRCA locus-specific LOH (Fig. 1, Supplementary Data 3, Supplementary Table 2). With the exception of BRCA1 germline mutation-associated ovarian tumors, the proportion of tumors without locus-specific LOH was nearly identical in the Penn and TCGA data sets (Fig. 1). The most common mechanism of locus-specific LOH in this data set was copy neutral LOH in BRCA1 tumors and LOH due to loss of the wild-type allele in BRCA2 tumors (Table 1). No tumor had evidence of a second somatic mutation. Methylation specific PCR was used to analyze the BRCA1 promoter in BRCA1 tumors. Eight of 23 BRCA1 germline mutation-associated breast (one without and seven with locus-specific LOH) and three of 15 BRCA1 germline mutation-associated ovarian tumors (two without and one with locus-specific LOH) had somatic promoter methylation (Supplementary Fig. 6). Of note, the allelic distribution of promoter methylation (wildtype vs. mutant allele vs. both) is not possible using this technique.
BRCA1 immunohistochemistry (IHC) was performed for 30 of 38 tumors associated with germline BRCA1 and 13 of 22 tumors associated with germline BRCA2 mutations. Combining tumors with both diffusely positive and heterogeneously positive nuclear staining, 12 of 13 tumors with BRCA2 mutations had positive BRCA1 IHC; these tumors had similar BRCA1 staining in nuclei of normal stromal tissue (Table 2, Fig. 2a). In contrast, all BRCA1-mutated tumors with locus-specific LOH had decreased or absent BRCA1 nuclear staining compared to normal stromal cell nuclei within the tumor (Table 2, Fig. 2b, Supplementary Figs. 7 and 8). In six tumors with BRCA1 locus-specific LOH, at least 25% of tumor nuclei retained positive BRCA1 staining, indicating intratumoral heterogeneity; two had missense mutations and four had truncating mutations in BRCA1. Four of the five tumors without locus-specific LOH, all with truncating mutations and including both with BRCA1 promoter methylation, showed either diffusely or heterogeneously positive BRCA1 staining in tumor nuclei similar to normal stromal cells (Table 2, Fig. 2c, Supplementary Figs. 7 and 8). Therefore, the two tumors without BRCA1 locus-specific LOH, but with promoter methylation, were included in the group without locus-specific LOH for genomic analyses.
BRCA locus-specific LOH and genomic signatures of BRCAness
Tumors in the Penn data set, all sequenced from nascent DNA, had a similar distribution of mutational signatures (majority Signatures 1 and 3) as the TCGA BRCA1/2 germline mutation-associated tumors. Tumors in the Penn data set without locus-specific LOH had a significantly lower proportion of BRCA mutational signature (BRCA1, P = 0.007 and BRCA2, P = 0.02) compared to tumors with locus-specific LOH, when mutational signatures were analyzed by both NMF and deconstructSigs (Fig. 2d). No significant differences in proportion of mutational signature were observed in the TCGA data set; WGA may alter these profiles (Supplementary Fig. 9a). Tumors without locus-specific LOH in both data sets trended towards having higher proportions of mutations due to Signature 1 (aging) (Supplementary Fig. 9b, c). Five tumors without locus-specific LOH had evidence of either Signature 2 or 13 (APOBEC)-associated mutations (Supplementary Fig. 9d, e, Supplementary Table 2).
Tumors without locus-specific LOH in the Penn data set had significantly lower HRD-Mean scores compared to tumors with locus-specific LOH for BRCA1 (P = 0.001) and BRCA2 (P = 0.01) (Fig. 2e). Similar results were seen in the TCGA data set for both BRCA1 (P < 0.0001) and BRCA2 (P = 0.01) (Supplementary Fig. 10a). When BRCA1 and BRCA2 tumors in both data sets were combined, despite differences in sequencing platform, both ovarian and breast tumors associated with germline BRCA1 and BRCA2 mutations without locus-specific LOH had significantly lower HRD-Mean scores compared to ovarian and breast tumors with locus-specific LOH (P < 0.0005) (Fig. 2f). HRD-Mean scores in tumors without locus-specific LOH were similar to nonBRCA tumors (Fig. 2f). To exclude a possible effect of WGA samples on this result, the combined Penn and TCGA breast data were reanalyzed including only TCGA tumors derived from nascent DNA (all Penn tumors were derived from nascent DNA). In this analysis, HRD-Mean scores remained significantly higher in tumors with locus-specific LOH compared to both tumors without locus-specific LOH and nonBRCA tumors (P < 0.0001, Supplementary Fig. 10b).
Molecular correlates of BRCA locus-specific LOH
We evaluated whether somatic mutations differed in BRCA1 and BRCA2 germline mutation-associated tumors with and without locus-specific LOH. We identified single nucleotide and insertion/deletion somatic mutations in the 60 Penn and 100 TCGA tumors. Mutational burden was defined as the number of somatic nonsynonymous mutations per megabase42. In the Penn data set, BRCA1 germline mutation-associated tumors had an average of 0.96 ± 0.39 mutations/Mb and BRCA2 germline mutation-associated tumors an average of 1.08 ± 1.00 mutations/Mb, consistent with mutational burden in studies of TCGA BRCA1 and BRCA2 germline mutation-associated ovarian tumors43. BRCA1 germline mutation-associated tumors, with and without locus-specific LOH, had similar mutational burdens (0.78 ± 0.20 vs. 1.00 ± 0.07, comparison not significant). However, BRCA2 germline mutation-associated tumors without locus-specific LOH had a significantly lower mutational burden compared to those with LOH (0.37 ± 0.11 vs. 1.41 ± 0.27, P = 0.02). Mutational burden was similar between tumors with and without locus-specific LOH in the TCGA data set (0.92 ± 0.36 vs. 1.29 ± 0.06, comparison not significant for BRCA1 and 0.99 ± 0.13 vs. 1.34 ± 0.11, P = 0.09 for BRCA2).
MutSigCV42 analysis identified only TP53 (q = 0) and PTEN (q = 0.006) as significantly mutated genes in Penn BRCA1 and BRCA2 germline mutation-associated tumors (Supplementary Data 4). In the TCGA BRCA1 and BRCA2 germline mutation-associated tumors, TP53 (q = 3.1e −12) and RB1 (q = 2.2e −5) were significantly mutated; PTEN had q = 0.2 (Supplementary Data 5). Both BRCA1 and BRCA2 germline mutation-associated tumors without locus-specific LOH were significantly less likely to have a TP53 mutation than BRCA1 and BRCA2 tumors with locus-specific LOH (44 vs. 84%, P = 0.01 for BRCA1; 24% vs. 68%, P = 0.001 for BRCA2). Beyond TP53 and PTEN, 41% of BRCA1 and BRCA2 germline mutation-associated tumors had a likely pathogenic/pathogenic mutation in a cancer gene (defined as COSMIC cancer gene census gene and/or as reported44) (Fig. 3). Across the 160 tumors, six cancer genes had mutations in more than two tumors: NF1 (8), PIK3CA (6), RB1 (6), ARID1A (4), TDG (4), and ERCC6 (3). There were no differences in the spectrum of cancer gene mutations between tumors with and without locus-specific LOH (Fig. 3, Supplementary Table 3).
PTEN loss is a common feature of breast and ovarian tumors associated with germline mutations in BRCA1 and BRCA2 45. PTEN allele-specific copy number status by WES was compared to PTEN status by IHC on 44 Penn tumors so that PTEN status in tumors without IHC could be estimated (Supplementary Fig. 11, Supplementary Table 4). Using this method, the rate of PTEN loss did not differ significantly between tumors with and without locus-specific LOH for BRCA1 and BRCA2 (38 vs. 30%, and 32 vs. 21%, respectively).
Clinical correlates of BRCA locus-specific LOH
We next evaluated the association of BRCA locus-specific LOH with clinical characteristics. No significant association between BRCA locus-specific LOH status and age of cancer diagnosis, breast tumor size, node positivity, or hormone receptor status or with ovarian tumor grade was observed (Supplementary Table 5). In ovarian cancer patients treated with adjuvant platinum based chemotherapy, patients whose tumors lacked BRCA locus-specific LOH had a significantly worse overall survival compared to patients whose tumors had locus-specific LOH, similar to nonBRCA patients (Fig. 4a). Absence of locus-specific LOH remained significantly associated with overall survival using a Cox proportional hazard model to control for site (Penn vs. TCGA), stage at diagnosis, and gene (BRCA1 vs. BRCA2) (Supplementary Fig. 12a, b). In breast cancer patients, absence of locus-specific LOH was associated with better survival compared to nonBRCA patients but was similar to patients whose tumors had locus-specific LOH (Fig. 4b). However, this was likely due to an enrichment of Stage III and/or triple negative patients in the nonBRCA group (Supplementary Fig. 12c).
It has been previously assumed that all tumors associated with germline BRCA1 and BRCA2 mutations have locus-specific LOH, and therefore loss of BRCA1 or BRCA2 function. However, large-scale studies have suggested that not all tumors from individuals with germline BRCA1 or BRCA2 mutations have locus-specific LOH28,29,30, and we hypothesized the absence of locus-specific LOH may be a biomarker of primary resistance to platinums and PARP inhibitors.
In the TGCA and Penn data sets, most tumors with germline mutations associated with BRCA1 had locus-specific LOH. Only nine of 93 (10%) of BRCA1 tumors in the combined data set lacked locus-specific LOH. All of the BRCA1 germline mutation-associated tumors without locus-specific LOH had HRD-Mean and BRCA mutational signature scores well below the mean for tumors demonstrating locus-specific LOH. Thus, a small percentage of BRCA1 germline mutation-associated tumors do not display locus-specific LOH and lack a BRCAness phenotype. Of the nine tumors lacking BRCA1 locus-specific LOH, three came from patients with a history of receiving chemotherapy for a prior cancer, and three were tumors of types not typically associated with BRCA1 germline mutations (an ovarian carcinosarcoma, one ER+ breast ductal carcinoma, and one ER+ breast lobular carcinoma), suggesting that second primary and non-classic histological tumors may be enriched for absence of locus-specific LOH.
We evaluated whether the presence or absence of locus-specific LOH of BRCA1 was associated with retention of the protein as measured by BRCA1 IHC. All tumors demonstrating locus-specific LOH had decreased BRCA1 protein as compared to the surrounding stroma, with most tumors demonstrating complete absence of BRCA1 protein. However, of note, a subset of tumors with BRCA1 locus-specific LOH retained BRCA1 expression in at least 25% of nuclei. In contrast, in 80% (four of five) of tumors without BRCA1 locus-specific LOH, BRCA1 protein was retained; of these two also had BRCA1 promoter methylation. Although promoter methylation of the wildtype allele would be expected to result in absence of protein, these data are consistent with prior studies showing approximately 20% of tumors with BRCA1 promoter methylation have retained BRCA1 protein46,47,48, which could be due to promoter methylation of the mutant as opposed to the wildtype allele. Overall, these results suggest that BRCA1 locus-specific LOH as measured genetically is a marker for protein expression. However, we found that tumors associated with BRCA1 mutations demonstrate heterogeneity in protein expression, and thus bulk sequencing, promoter methylation and LOH analysis may mask sub-populations of tumor cells that vary in regards to BRCA1 locus-specific LOH.
Almost identical percentages of tumors associated with BRCA2 germline mutations in the TCGA and Penn set did not have locus-specific LOH. Absence of locus-specific LOH was found at a particularly high rate for BRCA2 germline mutation-associated breast tumors (46%). Over 75% of BRCA2 germline mutation-associated tumors without locus-specific LOH had HRD scores significantly below the mean for tumors demonstrating locus-specific LOH. Additionally, we identified molecular correlates associated with absence of locus-specific LOH, also consistent with a lack of BRCAness, including a lower mutational burden in tumors associated with BRCA2 germline mutations and a significantly decreased frequency of TP53 mutations in tumors associated with BRCA1 and BRCA2 germline mutations.
Tumors without locus-specific LOH in patients with BRCA1 or BRCA2 germline mutations may arise due to various mechanisms of carcinogenesis. Several tumors arose in patients previously treated with chemotherapy and so subjected to other types of DNA damage which may induce tumor formation irrespective of germline BRCA status, although breast and ovarian cancers are not typically considered chemotherapy-induced malignancies. BRCA1 and BRCA2 mutant cells may be more sensitive in general to chemotherapy49 which could thereby eliminate precancerous clones promoting growth of tumors that are not fully deficient in BRCA function. Tumors also may develop through mechanisms similar to sporadic breast and ovarian tumors, related to estrogen exposure or aging, for example. Supportive of this postulate is a similar rate of an absence of locus-specific LOH in BRCA2 carriers in the two independent tumor sets, similar HRD-Mean scores to nonBRCA tumors and higher percentage of aging signature in those tumors without BRCA locus-specific LOH in the Penn data set. However, age of cancer diagnosis was not associated with LOH status and seven of 30 (23%) of the tumors without BRCA locus-specific LOH developed in patients under age 40, and 50% under age 50. As breast and ovarian cancers diagnosed under age 50 are rare in the general population, these data suggest possible alternate mechanisms of tumoriogenesis related to the underlying inherited mutation. BRCA1 and BRCA2 heterozygous mutant states have been shown to contribute to tumor formation in pancreatic and ovarian cancer mouse models50, 51, consistent with studies that demonstrate that haploinsufficiency for BRCA1 and BRCA2 leads to multiple levels of cellular dysfunction29, 52,53,54. Radiation-induced tumors in heterozygous Brca1 mice do not demonstrate locus-specific LOH55. It is therefore possible that tumors without locus-specific LOH in BRCA1 and BRCA2 germline mutation carriers develop due to haploinsufficiency of BRCA1 or BRCA2 function56 in conjunction with other predisposing events, such as environmental exposures.
BRCA1 and BRCA2 germline mutation status is known to be associated with improved overall survival in ovarian cancer patients57. We found that both BRCA1 and BRCA2 ovarian tumors without locus-specific LOH treated with adjuvant platinum based chemotherapy have lower overall survival than tumors with locus-specific LOH, at rates similar to sporadic tumors. This finding is consistent with preclinical data demonstrating that both human cell lines and mouse models lacking BRCA1 or BRCA2, due to biallelic mutations or knock out, are responsive to platinum and PARP inhibitors, whereas those that retain some level of BRCA1 or BRCA2 function are not12, 20,21,22,23,24. Our data support that BRCA locus-specific LOH is necessary for tumors in BRCA1 and BRCA2 germline mutation carriers to respond to platinum therapy, and potentially by extension PARP inhibitors.
There are limitations to the current study. We are using locus-specific LOH status to extrapolate BRCA1 or BRCA2 function. As our data suggest the presence of intratumoral heterogeneity protein expression of tumors associated with germline BRCA1 and BRCA2 mutation carriers, it is possible that further information may be obtained by single cell analysis. Sub-populations of cells within tumors may vary in terms of BRCA locus-specific LOH, and thus responsiveness to DNA damaging agents.
In aggregate, our findings from the largest single study of breast and ovarian tumors associated with BRCA1 and BRCA2 germline mutations suggest that locus-specific LOH may be a biomarker for a BRCAness phenotype in whole tumor formalin-fixed paraffin embedded (FFPE) specimens. There are other recently published BRCAness scores58, 59 which may predict therapy response. However, these scores require additional tumor genomic testing using single nucleotide polymorphism (SNP) arrays59 or whole genome sequencing58. Locus-specific LOH, which can be determined from already existing DNA-based FFPE tumor testing pipelines remains a clinically promising and more cost-effective means of predicting therapy response in germline BRCA1 and BRCA2 carriers. Our findings need to be confirmed prospectively, particularly in breast cancer patients stratified for stage and therapy, but provide support for the use of a BRCA locus-specific LOH assay to predict primary response to platinums, and potentially PARP inhibitors, in patients with germline BRCA1 and BRCA2 mutations. Such an assay could have broad applicability as germline BRCA1 and BRCA2 mutations have been associated with 3% of metastatic tumors of multiple subtypes60, 61. These results also emphasize that tumors associated with germline BRCA1 and BRCA2 mutations, and likely tumors in other carriers of mutations in DNA damage response genes, should not be considered uniformly. Further study is needed to investigate the precise role of the individual inherited mutations in DNA damage response pathway genes in the development of cancer, and whether or not locus-specific LOH may be useful to predict response to DNA damaging agents in carriers of mutations in other DNA damage response genes.
Identification of TCGA tumors with germline BRCA1 and BRCA2 mutations
Primary (Level 1) WES data was obtained to create the TCGA dat sets. To access the Level 1 DNA sequencing data for breast and ovarian tumors from TCGA, a project request was submitted and approved by the National Center for Biotechnology Information Genotypes and Phenotypes Database (NCBI dbGaP) Data Access Request system, Protocol #5309 “BRCA1 and BRCA2 mutations in breast and ovarian cancer”. Breast and ovarian samples potentially with controlled-access Level 1 WES binary alignment (.bam) files were identified from the Genomic Data Commons (GDC) (https://gdc.cancer.gov/) (n = 1098 breast tumors and 590 ovarian tumors) (Supplementary Fig. 1). After filtering for samples with WES data not available on the GDC commons (n = 406) and samples not passing our sequencing pipeline quality control analysis (n = 104), 1178 tumor/normal pairs were subjected to downstream analysis. Germline BRCA1 and BRCA2 mutations were identified using VarScan241 (http://dkoboldt.github.io/varscan/); and from the normal variant calls of Mutect62 (https://github.com/broadinstitute/mutect) and subjected to variant classification63. Tumors were determined to be associated with germline BRCA1 and BRCA2 mutations (n = 100) if they met the following criteria: (1) known pathogenic BRCA1 or BRCA2 mutation as per ENIGMA classification in germline and tumor sample; (2) germline allelic fraction (AF) >0.30; and (3) total depth >30 in germline and tumor at the mutation locus. Germline mutations were confirmed by review of traces in the Integrated Genome Viewer IGV (http://software.broadinstitute.org/software/igv/). We thus curated a data set of 100 BRCA1 and BRCA2 germline mutation-associated breast and ovarian tumors from the TCGA consisting of 37 breast (18 BRCA1 and 19 BRCA2) and 63 ovarian tumors (37 BRCA1 and 26 BRCA2).
Identification of nonBRCA TCGA tumors
To create a nonBRCA tumor set from the remaining 1078 tumor/normal BAM pairs (Supplementary Fig. 1), we analyzed primary WES data using Mutect62, VarScan241, and Sequenza38 (https://cran.r-project.org/web/packages/sequenza/index.html). Tumors were excluded if they were found to have: (1) a pathogenic somatic BRCA1 or BRCA2 mutation (n = 23 breast and n = 18 ovarian) and/or (2) homozygous copy number deletion of BRCA1 or BRCA2 (n = 11 breast and n = 4 ovarian). Finally, Level 3 RNAseq z-scores, microarray Z-scores and HK27/HK450m methylation beta values were bulk downloaded from The Cancer Genomics Hub of the University of Santa Cruz (https://cghub.ucsc.edu/, project now completed) for the tumor/normal pairs. Samples were excluded with z-scores < −1.5 and HK450m>0.5 for BRCA1 and RNAseq z-score < −1.5 for BRCA2 expression (n = 12 breast and 20 ovarian). Breast tumors were also excluded from patients who had received neoadjuvant chemotherapy (n = 11 breast).
Sample acquisition and preparation of Penn germline BRCA1 and BRCA2 mutated tumor set
The Penn study population was ascertained from academic and community hospital sites within Penn Medicine and the Penn Cancer Network. Acquisition of the patient samples was approved by the Institutional Review Board of the University of Pennsylvania, and informed consent was obtained from each participant for use of their samples and clinical data in genetic studies. Eligibility criteria for the study were (1) diagnosis of breast or ovarian cancer; (2) positive BRCA1/2 sequencing in a Clinical Laboratory Improvement Amendments-approved laboratory; and (3) available archived blood DNA and FFPE tumor blocks. Two hundred and twenty-three patients with breast and ovarian tumors meeting criteria were identified and 151 blocks were available or received by a central Pathology core (Supplementary Fig. 1). FFPE tumor blocks were sectioned and stained with hematoxylin and eosin to ensure sections of over 70% invasive tumor were used for DNA extraction, 68 tumors failed pathology quality control. Selected tumor areas of slides or rolls were dissolved in Deparaffinization Solution (Qiagen) and purified using Gentra PureGene Reagents (Qiagen) following manufacturer's protocols. All DNA samples were quantitated with a Qubit and tumor DNA was subjected to a quantitative PCR based QC Kit (Kapa Biosystems) for analysis of DNA quality (sample QC), and 23 samples failed laboratory quality control. This process resulted in a set 60 tumors, 38 from BRCA1 (23 breast, 15 ovarian tumors) and 22 from BRCA2 (16 breast, six ovarian tumors) mutation carriers (Supplementary Fig. 1). Germline DNA from blood or saliva was extracted using standard protocols in the laboratory. Library preparation of tumor and matched germline DNA was as described64, 65. In all Penn cases, nascent DNA, not subjected to WGA, was used for WES. Details of the Penn population are found in Supplementary Table 1 and summarized in Supplementary Table 2. Libraries were subjected to WES using the Agilent All-Exon Kit v5 (Santa Clara, CA). Tumors were sequenced on an Illumina Hi-Seq 2000 to an average mean depth of 141× and matched blood DNA was sequenced to an average mean depth of 155×.
Bioinformatics analysis and identification of somatic mutations in the Penn and TCGA tumor set
Quality control measures were determined with Picard Tools (https://broadinstitute.github.io/picard/). Penn tumors and matched germline were aligned to the hg19 assembly of the human genome using Burrows–Wheeler Aligner for short-read alignment66 (http://bio-bwa.sourceforge.net/). Variants underwent initial quality control filtering according to Genome Analysis ToolKit (GATK)67 best practices. Downloaded TCGA BAM files had been aligned to the hg38 assembly of the human genome. All exonic single nucleotide and insertion/deletion variants were identified using a combination of GATK67 (https://software.broadinstitute.org/gatk/), Mutect62, and VarScan241. Variants identified by initial bioinformatic analysis were annotated using ANNOVAR68 (http://annovar.openbioinformatics.org/en/latest/). Exonic variants were kept as somatic variants if they met the following criteria: (1) alternate allele depth in germline less than five reads; (2) population frequency <1% in EVS6500 (http://evs.gs.washington.edu/EVS/) and 1000 genomes (http://www.1000genomes.org/) databases; (3) not found in genomic regions of high (>89%) segmental duplications; (4) not categorized as synonymous variants; and (5) alternate allele supported by greater than 10 reads in the tumor. Manual curation of variants using a custom derived pipeline64 was used to classify variants as deleterious (D), likely deleterious (LD), variants of uncertain significance (VUS), likely benign and benign (B). Somatic mutations were defined as D or LD variants and rare VUS (variants with population allele frequency <0.1%) in the tumors. MutSigCV (http://software.broadinstitute.org/cancer/software/)42 was used to identify significantly mutated genes. Cancer genes are those in the COSMIC cancer gene census (http://cancer.sanger.ac.uk/census/) and/or reported44.
Determination of BRCA locus-specific loss of heterozygosity in Penn and TCGA data sets
A combination of VarScan241, allele frequency comparisons, and allele-specific copy number calls were used to determine BRCA locus-specific LOH. Estimates of tumor purity (cellularity) were determined using Sequenza and inputted into VarScan2 variant calling. The sample was assigned a locus-specific LOH positive status if the VarScan2 somatic P-value was significant and a locus-specific LOH negative status if the VarScan2 germline P-value was significant. Allele-specific copy number calls of the genomic region containing the BRCA1 or BRCA2 mutation were determined by Sequenza. The copy number of the genomic region surrounding the germline BRCA1 or BRCA2 mutation (CN) and the number of mutant (m) alleles as per the output of the Sequenza program were used to assign two states of absent locus-specific LOH—heterozygous diploid (CN = 2;m = 1) or amplified with gain of non-mutant (wildtype) allele (CN>2;m = 1)—and three states of locus-specific LOH—loss (CN = 1;m = 1), copy neutral LOH (CN = 2;m = 2), and amplified with LOH (CN>2;m>2). The genomic regions surrounding the germline BRCA1 and BRCA2 mutation ranged from less than one to over 100 Mb in length. In cases where the VarScan2 and ASCN calls differed (six of 100 TCGA tumors and four of 60 Penn tumors), the difference between cellularity corrected tumor allele frequency and blood allele frequency (ΔAF) was determined; the sample was assigned a locus-specific LOH positive status if ΔAF>0.2028. Finally, five of 100 TCGA tumors had a germline BRCA1 or BRCA2 mutation identified only by normal variant calling by Mutect; for these a combination of ASCN and ΔAF was used to determine locus-specific LOH.
Determination of mutational signatures in the Penn and TCGA data sets
Somatic mutations for determination of mutational signatures in the local and TCGA data set were identified using the MuTect2 variant calls derived from BAM files obtained as above. Mutational signatures were then extracted with the SomaticSignatures program34 using r = 4 and NMF options simultaneously for n = 80 Penn breast and ovarian tumors and n = 1186 TCGA breast and ovarian tumors. In addition, the deconstructSigs program was additionally used to determine the proportion of known mutational signatures in the data33. Using deconstructSigs, 60 TCGA tumors had greater than 30% of their mutational signatures due to sequencing artifact (signature R1–R3, U1–U2)32, including one BRCA2 tumor without locus-specific LOH, three BRCA1 tumors with locus-specific LOH, seven nonBRCA ovarian, and 49 nonBRCA breast tumors. These tumors were therefore excluded from all subsequent genomic analyses. No Penn tumors had greater than 10% of their mutational signatures due to sequencing artifact. Mutational signatures were calculated blinded to locus-specific LOH status.
Determination of homologous recombination deficiency in the Penn and TCGA data sets
Allele-specific copy number states were determined using Sequenza and used to calculate the HRD scores NtAI36, LST37, and genomic LOH (HRD-LOH)35 using custom R-scripts, which are available upon request. Non-telomeric allelic imbalance (NtAI) scores were derived from the Sequenza data by summing the number of segments of allelic imbalance that were post-centromeric to the sub-telomeric regions and >11 Mb in length. Large state transition scores were derived from the Sequenza data by summing the number of breakpoints creating >3 Mb segments that were >10 Mb from one another. Raw LST scores were corrected for ploidy (LSTm) using the equation LSTm = LST–15.5 × ploidy. HRD-LOH scores were derived from Sequenza data by summing the number of segments of LOH >15 Mb in length excluding segments found on Chromosome 17. HRD scores were calculated blinded to locus-specific LOH status.
Methylation specific PCR analysis of the BRCA1 promoter
Methylation specific PCR analysis was performed using primers specific for unmethylated and methylated BRCA1 promoter as described69, 70. Primers were obtained from Integrated DNA Technologies (Coralville, IA). Bands were resolved on a 6% Tris-buffered EDTA (TBE) gel from Invitrogen (Carlsbad, CA). A representative full blot is found in Supplementary Fig. 6. Band intensity of samples was compared to a fully methylated and an unmethylated control and a band intensity ≥ two-fold of the band seen in the unmethylated control was considered positive for BRCA1 promoter methylation. Constitutional blood DNA was concurrently tested and demonstrated an absence of constitutional methylation of the BRCA1 promoter in tumors with positive BRCA1 promoter methylation. Methylation was quantified blinded to locus-specific LOH status.
PTEN and BRCA1 immunohistochemistry
Tissue microarrays containing 44 of the sequenced tumors were made. Sections were prepared of tumor and surrounding normal tissue in triplicate for each sample available. Immunohistochemistry was performed for BRCA1 using MS110 antibody from EMD Millipore (Billerica, Massachusetts), staining for one tumor failed. Immunohistochemistry was performed for PTEN using D4.3 antibody from Cell Signaling Technology (Danvers, Massachusetts). Scoring for immunohistochemical stains was performed in triplicate for each tissue sample. Stain intensity was graded on a four-point scale (0–3) by a board-certified pathologist (A.B., M.F.), with 0 = no staining, 1 = weak or partial staining, 2 = moderate staining, 3 = strong staining. BRCA1 and BRCA2 mutation and LOH status and PTEN allele-specific copy number status was blinded. Two cell populations (tumor and non-tumor cell types) were analyzed in each sample, and the intensity of cytoplasmic and nuclear staining in each population was assessed independently. In addition, the percentage of cells with the highest intensity of staining was also reported.
Use of allele-specific copy number calls at the genomic PTEN locus to extrapolate PTEN status
To determine if PTEN allele-specific copy number status by Sequenza38 based WES analysis accurately predicted protein loss, we analyzed PTEN status by IHC on 44 tumors from the Penn data set as above (Supplementary Fig. 11). All 10 tumors with PTEN copy number loss (n = 7) or a truncating PTEN mutation with copy neutral LOH (n = 3) did not stain positively for PTEN on IHC (Supplementary Table 4). Twenty-four of 25 tumors that had wildtype PTEN with a copy number state of two or more had staining of PTEN on IHC (concordance 97%). Of nine tumors without an identifiable mutation but copy neutral LOH of PTEN on IHC: four stained positively and five did not stain for PTEN. Given these results, for the 16 Penn tumors without available IHC data and the 100 TCGA tumors, the ASCN state of the PTEN genomic locus was used to extrapolate PTEN status. PTEN copy number loss or truncating mutation was used to extrapolate PTEN loss (n = 15), and PTEN wildtype status at copy number two or more to extrapolate retention of PTEN (n = 64). Eleven tumors with copy neutral LOH were excluded from analysis.
Statistical and clinical data analyses
Means of continuous variables were compared using a two-tailed Student’s t-test. Outliers were excluded based on Grubb’s test (extreme studentized deviate test). Comparisons of rates in different groups were determined using a two-tailed Fisher’s exact test of significance for two groups or a one-way ANOVA with correction for multiple comparisons for three or more groups. Clinical data were obtained for the patients in the Penn data set by IRB approved chart review. The overall survival time for all patients was determined from the date of diagnosis to the time of last follow-up or death by query of the medical record for each Penn patient or by bulk data download from The Cancer Genomics Hub of the University of California at Santa Cruz (https://cghub.ucsc.edu/, project now completed) for the TCGA data set. Patients alive at the end of follow-up were censored. Survival was compared using a log-rank (Mantel–Cox) test. In addition, a Cox proportional hazards model was used to examine the effect of locus-specific LOH status, site (TCGA vs. Penn) and stage at diagnosis for ovarian cancer patients. For breast cancer patients, a Cox proportional hazards model was used to examine the effect of locus-specific LOH status, site (TCGA vs. Penn), stage at diagnosis, and hormone receptor status. Kaplan–Meier curves were constructed for each analysis.
The WES data that supports this study have been deposited in the National Center for Biotechnology Information (NCBI)’s Sequence Read Archive (SRA, https://www.ncbi.nlm.nih.gov/sra) with BioProject ID PRJNA38804 and can be accessed at: http://www.ncbi.nlm.nih.gov/bioproject/388048. The TCGA data is available from the National Cancer Institute’s Genome Data Commons (https://gdc.cancer.gov/). The remaining data are available within the article and its Supplementary Information files or available from the authors upon request.
Daly, M. B. et al. Genetic/familial high-risk assessment: breast and ovarian. J. Natl Compr. Cancer Netw. 8, 562–594 (2010).
Antoniou, A. et al. Average risks of breast and ovarian cancer associated with BRCA1 or BRCA2 mutations detected in case series unselected for family history: a combined analysis of 22 studies. Am. J. Hum. Genet. 72, 1117–1130 (2003).
Cancer risks in BRCA2 mutation carriers. The Breast Cancer Linkage Consortium. J. Natl Cancer. Inst. 91, 1310–1316 (1999).
Turner, N., Tutt, A. & Ashworth, A. Hallmarks of 'BRCAness' in sporadic cancers. Nat. Rev. Cancer. 4, 814–819 (2004).
Kamieniak, M. M. et al. DNA copy number profiling reveals extensive genomic loss in hereditary BRCA1 and BRCA2 ovarian carcinomas. Br. J. Cancer 108, 1732–1742 (2013).
Stefansson, O. A. et al. Genomic profiling of breast tumours in relation to BRCA abnormalities and phenotypes. Breast Cancer Res. 11, R47 (2009).
Dasika, G. K. et al. DNA damage-induced cell cycle checkpoints and DNA strand break repair in development and tumorigenesis. Oncogene 18, 7883–7899 (1999).
Smith, S. A., Easton, D. F., Evans, D. G. & Ponder, B. A. Allele losses in the region 17q12-21 in familial breast and ovarian cancer involve the wild-type chromosome. Nat. Genet. 2, 128–131 (1992).
Gudmundsson, J. et al. Different tumor types from BRCA2 carriers show wild-type chromosome deletions on 13q12-q13. Cancer Res. 55, 4830–4832 (1995).
Bhattacharyya, A., Ear, U. S., Koller, B. H., Weichselbaum, R. R. & Bishop, D. K. The breast cancer susceptibility gene BRCA1 is required for subnuclear assembly of Rad51 and survival following treatment with the DNA cross-linking agent cisplatin. J. Biol. Chem. 275, 23899–23903 (2000).
Bryant, H. E. et al. Specific killing of BRCA2-deficient tumours with inhibitors of poly(ADP-ribose) polymerase. Nature 434, 913–917 (2005).
Farmer, H. et al. Targeting the DNA repair defect in BRCA mutant cells as a therapeutic strategy. Nature 434, 917–921 (2005).
Byrski, T. et al. Response to neoadjuvant therapy with cisplatin in BRCA1-positive breast cancer patients. Breast Cancer. Res. Treat. 115, 359–363 (2009).
Ledermann, J. et al. Olaparib maintenance therapy in platinum-sensitive relapsed ovarian cancer. N. Engl. J. Med. 366, 1382–1392 (2012).
Sandhu, S. K. et al. The poly(ADP-ribose) polymerase inhibitor niraparib (MK4827) in BRCA mutation carriers and patients with sporadic cancer: a phase 1 dose-escalation trial. Lancet Oncol. 14, 882–892 (2013).
Balmana, J. et al. Phase I trial of olaparib in combination with cisplatin for the treatment of patients with advanced breast, ovarian and other solid tumors. Ann. Oncol. 25, 1656–1663 (2014).
Fong, P. C. et al. Inhibition of poly(ADP-ribose) polymerase in tumors from BRCA mutation carriers. N. Engl. J. Med. 361, 123–134 (2009).
Helleday, T. The underlying mechanism for the PARP and BRCA synthetic lethality: clearing up the misunderstandings. Mol. Oncol. 5, 387–393 (2011).
Lord, C. J. & Ashworth, A. BRCAness revisited. Nat. Rev. Cancer 16, 110–120 (2016).
Sakai, W. et al. Functional restoration of BRCA2 protein by secondary BRCA2 mutations in BRCA2-mutated ovarian carcinoma. Cancer Res. 69, 6381–6386 (2009).
Sakai, W. et al. Secondary mutations as a mechanism of cisplatin resistance in BRCA2-mutated cancers. Nature 451, 1116–1120 (2008).
Sawyer, S. L. et al. Biallelic mutations in BRCA1 cause a new Fanconi anemia subtype. Cancer Discov. 5, 135–142 (2015).
Husain, A., He, G., Venkatraman, E. S. & Spriggs, D. R. BRCA1 up-regulation is associated with repair-mediated resistance to cis-diamminedichloroplatinum(II). Cancer Res. 58, 1120–1123 (1998).
Drew, Y. et al. Therapeutic potential of poly(ADP-ribose) polymerase inhibitor AG014699 in human cancers with mutated or methylated BRCA1 or BRCA2. J. Natl Cancer. Inst. 103, 334–346 (2011).
Audeh, M. W. et al. Oral poly(ADP-ribose) polymerase inhibitor olaparib in patients with BRCA1 or BRCA2 mutations and recurrent ovarian cancer: a proof-of-concept trial. Lancet 376, 245–251 (2010).
Kaufman, B. et al. Olaparib monotherapy in patients with advanced cancer and a germline BRCA1/2 mutation. J. Clin. Oncol. 33, 244–250 (2015).
Tutt, A. et al. Oral poly(ADP-ribose) polymerase inhibitor olaparib in patients with BRCA1 or BRCA2 mutations and advanced breast cancer: a proof-of-concept trial. Lancet 376, 235–244 (2010).
Kanchi, K. L. et al. Integrated analysis of germline and somatic variants in ovarian cancer. Nat. Commun. 5, 3156 (2014).
Martins, F. C. et al. Evolutionary pathways in BRCA1-associated breast tumors. Cancer Discov. 2, 503–511 (2012).
Nik-Zainal, S. et al. Landscape of somatic mutations in 560 breast cancer whole-genome sequences. Nature 534, 47–54 (2016).
Swisher, E. M. et al. Secondary BRCA1 mutations in BRCA1-mutated ovarian carcinomas with platinum resistance. Cancer Res. 68, 2581–2586 (2008).
Alexandrov, L. B. et al. Signatures of mutational processes in human cancer. Nature 500, 415–421 (2013).
Rosenthal, R., McGranahan, N., Herrero, J., Taylor, B. S. & Swanton, C. DeconstructSigs: delineating mutational processes in single tumors distinguishes DNA repair deficiencies and patterns of carcinoma evolution. Genome Biol. 17, 31 (2016).
Gehring, J. S., Fischer, B., Lawrence, M. & Huber, W. SomaticSignatures: inferring mutational signatures from single-nucleotide variants. Bioinformatics 31, 3673–3675 (2015).
Abkevich, V. et al. Patterns of genomic loss of heterozygosity predict homologous recombination repair defects in epithelial ovarian cancer. Br. J. Cancer 107, 1776–1782 (2012).
Birkbak, N. J. et al. Telomeric allelic imbalance indicates defective DNA repair and sensitivity to DNA-damaging agents. Cancer Discov. 2, 366–375 (2012).
Popova, T. et al. Ploidy and large-scale genomic instability consistently identify basal-like breast carcinomas with BRCA1/2 inactivation. Cancer Res. 72, 5454–5462 (2012).
Favero, F. et al. Sequenza: allele-specific copy number and mutation profiles from tumor sequencing data. Ann. Oncol. 26, 64–70 (2015).
Pinard, R. et al. Assessment of whole genome amplification-induced bias through high-throughput, massively parallel whole genome sequencing. BMC Genomics 7, 216 (2006).
Stokes, A. et al. Copy number and loss of heterozygosity detected by SNP array of formalin-fixed tissues using whole-genome amplification. PLoS ONE 6, e24503 (2011).
Koboldt, D. C. et al. VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing. Genome Res. 22, 568–576 (2012).
Lawrence, M. S. et al. Mutational heterogeneity in cancer and the search for new cancer-associated genes. Nature 499, 214–218 (2013).
Birkbak, N. J. et al. Tumor mutation burden forecasts outcome in ovarian cancer with BRCA1 or BRCA2 mutations. PLoS ONE 8, e80023 (2013).
Ye, K. et al. Systematic discovery of complex insertions and deletions in human cancers. Nat. Med. 22, 97–104 (2016).
Saal, L. H. et al. Recurrent gross mutations of the PTEN tumor suppressor gene in breast cancers with deficient DSB repair. Nat. Genet. 40, 102–107 (2008).
Garg, K. et al. BRCA1 immunohistochemistry in a molecularly characterized cohort of ovarian high-grade serous carcinomas. Am. J. Surg. Pathol. 37, 138–146 (2013).
Shilpa, V. et al. BRCA1 promoter hypermethylation and protein expression in ovarian carcinoma--an Indian study. Tumour Biol. 35, 4277–4284 (2014).
Wang, C. et al. Expression of BRCA1 protein in benign, borderline, and malignant epithelial ovarian neoplasms and its relationship to methylation and allelic loss of the BRCA1 gene. J. Pathol. 202, 215–223 (2004).
Kriege, M. et al. Sensitivity to first-line chemotherapy for metastatic breast cancer in BRCA1 and BRCA2 mutation carriers. J. Clin. Oncol. 27, 3764–3771 (2009).
Perets, R. et al. Transformation of the fallopian tube secretory epithelium leads to high-grade serous ovarian cancer in Brca;Tp53;Pten models. Cancer Cell 24, 751–765 (2013).
Skoulidis, F. et al. Germline Brca2 heterozygosity promotes Kras(G12D) -driven carcinogenesis in a murine model of familial pancreatic cancer. Cancer Cell 18, 499–509 (2010).
Konishi, H. et al. Mutation of a single allele of the cancer susceptibility gene BRCA1 leads to genomic instability in human breast epithelial cells. Proc. Natl Acad. Sci. USA 108, 17773–17778 (2011).
Pathania, S. et al. BRCA1 haploinsufficiency for replication stress suppression in primary cells. Nat. Commun. 5, 5496 (2014).
Sedic, M. et al. Haploinsufficiency for BRCA1 leads to cell-type-specific genomic instability and premature senescence. Nat. Commun. 6, 7505 (2015).
Jeng, Y. M. et al. Brca1 heterozygous mice have shortened life span and are prone to ovarian tumorigenesis with haploinsufficiency upon ionizing irradiation. Oncogene 26, 6160–6166 (2007).
Bartek, J., Lukas, J. & Bartkova, J. DNA damage response as an anti-cancer barrier: damage threshold and the concept of 'conditional haploinsufficiency'. Cell Cycle 6, 2344–2347 (2007).
Bolton, K. L. et al. Association between BRCA1 and BRCA2 mutations and survival in women with invasive epithelial ovarian cancer. JAMA 307, 382–390 (2012).
Davies, H. et al. HRDetect is a predictor of BRCA1 and BRCA2 deficiency based on mutational signatures. Nat. Med. 23, 517–525 (2017).
Telli, M. L. et al. Homologous Recombination Deficiency (HRD) score predicts response to platinum-containing neoadjuvant chemotherapy in patients with triple negative breast cancer. Clin. Cancer Res. 22, 3764–3773 (2016).
Meric-Bernstam, F. et al. Incidental germline variants in 1000 advanced cancers on a prospective somatic genomic profiling protocol. Ann. Oncol. 27, 795–800 (2016).
Schrader, K. A. et al. Germline variants in targeted tumor sequencing using matched normal DNA. JAMA Oncol. 2, 104–111 (2016).
Cibulskis, K. et al. Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples. Nat. Biotechnol. 31, 213–219 (2013).
Maxwell, K. N. et al. Evaluation of ACMG-guideline-based variant classification of cancer susceptibility and non-cancer-associated genes in families affected by breast cancer. Am. J. Hum. Genet. 98, 1–17 (2016).
Fishbein, L. et al. Whole-exome sequencing identifies somatic ATRX mutations in pheochromocytomas and paragangliomas. Nat. Commun. 6, 6140 (2015).
Maxwell, K. N. et al. Prevalence of mutations in a panel of breast cancer susceptibility genes in BRCA1/2-negative patients with early-onset breast cancer. Genet. Med. 17, 630–638 (2015).
Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25, 1754–1760 (2009).
McKenna, A. et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 20, 1297–1303 (2010).
Wang, K., Li, M. & Hakonarson, H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res. 38, e164 (2010).
Esteller, M. et al. Promoter hypermethylation and BRCA1 inactivation in sporadic breast and ovarian tumors. J. Natl Cancer. Inst. 92, 564–569 (2000).
Turner, N. C. et al. BRCA1 dysfunction in sporadic basal-like breast cancer. Oncogene 26, 2126–2132 (2007).
Financial support for this work was provided by the Department of Defense (W81XWH-13-1-0338, K.N.M.), National Institutes of Health (5T32GM008638-15, K.N.M.), Basser Center for BRCA at the University of Pennsylvania (K.L.N., S.M.D., R.D.), Konner Family Foundation (K.N.M.), Breast Cancer Research Foundation (K.L.N.), MacDonald Cancer Risk Evaluation Program (S.M.D.), Susan G Komen Foundation (S.M.D.) and Rooney Family Foundation (K.L.N., S.M.D.). Views and opinions of, and endorsements by the authors do not reflect those of the US Army or the Department of Defense. All work contained in this manuscript is original. We thank Louisa Pyle, M.D., Ph.D. and Kevin Nead, M.D. for helpful comments on the manuscript.
The authors declare no competing financial interests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Maxwell, K.N., Wubbenhorst, B., Wenz, B.M. et al. BRCA locus-specific loss of heterozygosity in germline BRCA1 and BRCA2 carriers. Nat Commun 8, 319 (2017). https://doi.org/10.1038/s41467-017-00388-9
This article is cited by
Validation of genomic and transcriptomic models of homologous recombination deficiency in a real-world pan-cancer cohort
BMC Cancer (2022)
Landscape of homologous recombination deficiencies in solid tumours: analyses of two independent genomic datasets
BMC Cancer (2022)
Identification of biomarkers of response to preoperative talazoparib monotherapy in treatment naïve gBRCA+ breast cancers
npj Breast Cancer (2022)
Somatic loss of the remaining allele occurs approximately in half of CHEK2-driven breast cancers and is accompanied by a border-line increase of chromosomal instability
Breast Cancer Research and Treatment (2022)
npj Breast Cancer (2022)