This study of patients with breast cancer, ovarian cancer, or both compared with controls uses whole-exome sequencing to investigate whether non-BRCA genes are associated with an increased risk of those cancers.
Key Points
Question
Which non-BRCA genes are associated with breast or ovarian cancer and what are the magnitudes of those risks?
Findings
In this study assessing whole-exome sequencing results from 11 416 patients with breast cancer, ovarian cancer, or both and 3988 controls, an increased risk of breast cancer was associated with PALB2, ATM, CHEK2, and MSH6 genes, whereas MSH6, RAD51C, TP53, and ATM genes were associated with an increased risk of ovarian cancer.
Meaning
In addition to confirming several well-known breast or ovarian cancer gene associations, this study identified MSH6 and ATM as possible moderate-risk breast and ovarian cancer predisposition genes, respectively.
Abstract
Importance
Since the discovery of BRCA1 and BRCA2, multiple high- and moderate-penetrance genes have been reported as risk factors for hereditary breast cancer, ovarian cancer, or both; however, it is unclear whether these findings represent the complete genetic landscape of these cancers. Systematic investigation of the genetic contributions to breast and ovarian cancers is needed to confirm these findings and explore potentially new associations.
Objective
To confirm reported and identify additional predisposition genes for breast or ovarian cancer.
Design, Setting, and Participants
In this sample of 11 416 patients with clinical features of breast cancer, ovarian cancer, or both who were referred for genetic testing from 1200 hospitals and clinics across the United States and of 3988 controls who were referred for genetic testing for noncancer conditions between 2014 and 2015, whole-exome sequencing was conducted and gene-phenotype associations were examined. Case-control analyses using the Genome Aggregation Database as a set of reference controls were also conducted.
Main Outcomes and Measures
Breast cancer risk associated with pathogenic variants among 625 cancer predisposition genes; association of identified predisposition breast or ovarian cancer genes with the breast cancer subtypes invasive ductal, invasive lobular, hormone receptor–positive, hormone receptor–negative, and male, and with early-onset disease.
Results
Of 9639 patients with breast cancer, 3960 (41.1%) were early-onset cases (≤45 years at diagnosis) and 123 (1.3%) were male, with men having an older age at diagnosis than women (mean [SD] age, 61.8 [12.8] vs 48.6 [11.4] years). Of 2051 women with ovarian cancer, 445 (21.7%) received a diagnosis at 45 years or younger. Enrichment of pathogenic variants were identified in 4 non-BRCA genes associated with breast cancer risk: ATM (odds ratio [OR], 2.97; 95% CI, 1.67-5.68), CHEK2 (OR, 2.19; 95% CI, 1.40-3.56), PALB2 (OR, 5.53; 95% CI, 2.24-17.65), and MSH6 (OR, 2.59; 95% CI, 1.35-5.44). Increased risk for ovarian cancer was associated with 4 genes: MSH6 (OR, 4.16; 95% CI, 1.95-9.47), RAD51C (OR, not estimable; false-discovery rate–corrected P = .004), TP53 (OR, 18.50; 95% CI, 2.56-808.10), and ATM (OR, 2.85; 95% CI, 1.30-6.32). Neither the MRN complex genes nor CDKN2A was associated with increased breast or ovarian cancer risk. The findings also do not support previously reported breast cancer associations with the ovarian cancer susceptibility genes BRIP1, RAD51C, and RAD51D, or mismatch repair genes MSH2 and PMS2.
Conclusions and Relevance
The results of this large-scale exome sequencing of patients and controls shed light on both well-established and controversial non-BRCA predisposition gene associations with breast or ovarian cancer reported to date and may implicate additional breast or ovarian cancer susceptibility gene candidates involved in DNA repair and genomic maintenance.
Introduction
Understanding the genetic bases for clinical phenotypes is an important goal of both clinical diagnostics and biomedical research. At least 3% of all cancer cases are attributed to hereditary variation in cancer predisposition genes,1 and 5% to 20% of breast cancer (BC) or ovarian cancer (OV) cases are linked to genetic abnormalities.2,3 In the 1990s, family-based genetic linkage studies identified 2 major BC and OV genes, BRCA1 and BRCA2, which account for at least 25% of familial risk,4 implying a more complex “polygenic” genetic architecture of BC.5 In the past decade, several more BC and OV susceptibility genes were identified from large case-control association studies.6,7,8,9 Although next-generation sequencing has enabled rapid screening for new cancer susceptibility genes, there are currently only 11 well-characterized BC and OV susceptibility genes (ATM [OMIM 607585], BRCA1 [OMIM 113705], BRCA2 [OMIM 600185], BRIP1 [OMIM 605882], CDH1 [OMIM 192090], CHEK2 [OMIM 604373], PALB2 [OMIM 610355], PTEN [OMIM 601728], RAD51C [OMIM 602774], RAD51D [OMIM 602954], and TP53 [OMIM 191170])10,11,12,13 and 6 others (BARD1 [OMIM 601593], CDKN2A [OMIM 600160], NF1 [OMIM 162200], MRE11A [OMIM 600814], RAD50 [OMIM 604040], and NBN [OMIM 602667])11,14,15,16,17 reported to have associations with BC but for which large-scale robust evidence is lacking (eTable 1 in the Supplement). A recent study of 65 057 patients with BC referred for multigene panel testing supports the role of BARD1 but not CDKN2A, NF1, MRE11A, RAD50, or NBN in breast cancer.18 Moreover, although mismatch repair (MMR) genes MLH1 (OMIM 120436), MSH2 (OMIM 609309), MSH6 (OMIM 600678), and PMS2 (OMIM 600259) are widely accepted and treated clinically as OV susceptibility genes, data regarding their gene-specific contributions to BC or OV risk are limited.
Further investigation of cancer risks associated with less well-characterized genes is imperative, particularly as multigene panel testing that includes moderate-risk genes are increasingly used in oncology practice.19 Currently, only approximately 10% to 24% of patients referred for BC or OV risk assessment with genetic testing are found to harbor known pathogenic variants identified by multigene panel testing.11,14,20,21 To determine whether and to what extent additional cancer genes not currently included in most multigene panel tests contribute to these risks, we performed whole-exome sequencing for 11 416 patients with clinical features of BC or OV and for 3988 controls. The results have the potential to add to our understanding of the genomic landscape of BC and OV predisposition and to inform comprehensive genetic testing in the context of patient care.
Methods
Samples
The study participants comprised 11 416 unrelated patients with BC, OV, or both referred for genetic testing from 1200 hospitals and clinics across the United States between 2014 and 2015 (Table 1; eTable 2 in the Supplement). The in-laboratory controls consisted of 3988 unrelated patients referred for genetic testing for noncancer conditions: 445 patients (11.1%) with cystic fibrosis, 1766 patients (44.3%) with rare inherited cardiovascular conditions, and 1777 reportedly healthy parents of children with epilepsy, congenital anomalies, autism, or psychiatric disorders (44.6%) enrolled in our clinical exome sequencing program. Patients with self-reported personal history of cancer were not included in the in-laboratory control group. Demographic, clinical history, and family history information were collected from documents provided by the ordering clinicians. A consistent set of protocols pertaining to exome library construction, exome capture, sequencing platform, and bioinformatics pipeline (eAppendix in the Supplement) was applied to all case and control samples to eliminate potential biases associated with differences in sequencing platforms and to enable further filtration of artifacts in cancers when compared with a reference population, the Genome Aggregation Database (gnomAD) (eAppendix in the Supplement). This study was submitted to the Solutions Institutional Review Board (Little Rock, Arkansas) and was deemed exempt from review. In addition, this board waived the need for informed patient consent because all data used were from previously existing, deidentified samples gathered for the purpose of research.
Table 1. Clinical Charateristics of Patients With Breast Cancer (BC) or Ovarian Cancer (OV) and In-Laboratory Controlsa.
| Charateristic | Patients, No. (%) | ||
|---|---|---|---|
| With BC (n = 9639) | With OV (n = 2051) | Controls (n = 3988) | |
| Race/ethnicity | |||
| White | 7208 (74.8) | 1718 (83.8) | 2765 (69.3) |
| African American | 985 (10.2) | 80 (3.9) | 323 (8.1) |
| Hispanic | 874 (9.1) | 161 (7.8) | 561 (14.1) |
| Asian | 572 (5.9) | 92 (4.5) | 339 (8.5) |
| Age at diagnosis, y | |||
| Mean (SD) | 48.7 (11.5) | 55.7 (14.1) | 39.7 (14.7)b |
| ≤45 | 3960 (41.1) | 445 (21.7) | 2671 (67.0) |
| 46-60 | 3876 (40.2) | 768 (37.4) | 951 (23.8) |
| >60 | 1518 (15.7) | 751 (36.6) | 332 (8.3) |
| Not providedc | 285 (3.0) | 87 (4.2) | 34 (0.9) |
| Sex | |||
| Male | 123 (1.3) | 0 | 1315 (33.0) |
| Female | 9516 (98.7) | 2051 (100.0) | 2673 (67.0) |
| BC histology | |||
| IDC | 6829 (70.8) | NA | NA |
| ILC | 493 (5.1) | NA | NA |
| IDC and ILC | 141 (1.5) | NA | NA |
| Not providedc | 2176 (22.6) | NA | NA |
| Bilateral BC | |||
| Yes | 1321 (13.7) | NA | NA |
| Not providedc | 8318 (86.3) | NA | NA |
| BC subtypes (receptor status) | |||
| ER+, PR+, and HER2− (HR+) | 2476 (25.7) | NA | NA |
| ER−, PR−, and HER2+ (HR−) | 278 (2.9) | NA | NA |
| ER+, PR+, and HER2+ (TPBC) | 539 (5.6) | NA | NA |
| ER−, PR−, and HER2− (TNBC) | 1154 (12.0) | NA | NA |
| Additional cancer primary | |||
| Colorectal cancer | 166 (1.7) | 35 (1.7) | NA |
| Uterine cancer | 378 (3.9) | 145 (7.1) | NA |
| Others | 97 (1.0) | 28 (1.4) | NA |
| None | 8998 (93.3) | 1843 (89.8) | NA |
| 1st- or 2nd-Degree relative with any cancer | |||
| Yes | 8152 (84.6) | 1670 (81.4) | NA |
| No | 282 (2.9) | 72 (3.5) | NA |
| Not providedc | 1205 (12.5) | 309 (15.1) | NA |
| 1st- or 2nd-Degree relative with BC | |||
| Yes | 5404 (56.1) | 787 (38.4) | NA |
| No | 3030 (31.4) | 955 (46.6) | NA |
| Not providedc | 1205 (12.5) | 309 (15.1) | NA |
| 1st- or 2nd-Degree relative with OV | |||
| Yes | 981 (10.2) | 245 (11.9) | NA |
| No | 7453 (77.3) | 1497 (73.0) | NA |
| Not providedc | 1205 (12.5) | 309 (15.1) | NA |
Abbreviations: ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; HR, hormone receptor; IDC, invasive ductal carcinoma; ILC, invasive lobular carcinoma; PR, progesterone receptor; NA, not applicable; TNBC, triple-negative BC; TPBC, triple-positive BC; +, positive for the receptor; −, negative for the receptor.
Inferred ethnicity using principal component analysis to identify population substructure based on exome-sequencing data.
Age at testing for controls.
Incomplete or unavailable clinical information.
Statistical Analysis
After data cleaning and filtering (eAppendix in the Supplement), we collapsed pathogenic variants (loss of function and known pathogenic22) by gene and performed burden tests for 625 cancer genes (eAppendix, eTable 1, and eFigure 1 in the Supplement) among 9639 BC and 2051 OV cases vs 3988 in-laboratory controls. We excluded truncations located beyond the last 55 base pairs of the penultimate exon and not residing in functional domains (and thus may not be influenced by nonsense-mediated messenger RNA decay).18,23 We also examined the distribution of pathogenic variants in 11 characterized BC or OV genes (eAppendix and eFigure 2 in the Supplement). Although the coverage of our sequencing data was comparable to that for the gnomAD database, we excluded sites with 10× coverage (per-site sequencing depth) or less in gnomAD or a missing rate of 30% or more at 10× coverage, which avoided inflation of signals in either direction.
For both case vs in-laboratory control comparisons and case vs gnomAD comparisons (eTables 3 and 4 in the Supplement), we used the cohort allelic sums test,24 a burden test widely used for association analysis of rare variants.10,25,26,27 Odds ratios (ORs) and 2-sided P values were computed using the Fisher exact test, followed by the Benjamini-Hochberg false-discovery rate (FDR) multiple testing correction. An FDR-corrected P < .05 was considered statistically significant. For genes with variants observed in cases but absent in controls, ORs could not be estimated. Genes with 2 or fewer detectable pathogenic variants were excluded. Suspected pathogenic variants in BC- or OV-associated genes were reviewed and confirmed by Sanger sequencing.
To assess the contribution of candidate genes to specific clinical features of BC or OV, we tested their associations with breast cancer subtypes. Moreover, we performed 5 sensitivity analyses to evaluate the robustness of our findings and 2 case-case analyses to explore genetic associations with these clinical features (eTables 5-11 and eAppendix in the Supplement). Pathogenic variants carried by cases and in-laboratory controls are given in eTables 12 through 14 in the Supplement. All statistical analyses were conducted using R, version 3.3.3.
Results
Patient Characteristics
The overall racial/ethnic distribution for 11 416 patients with BC, OV, or both is shown in eFigure 3 in the Supplement. Of 9639 patients with BC (Table 1), 3960 (41.1%) were early-onset cases (diagnosis age ≤45 years) and 123 (1.3%) were male, with men having an older age at diagnosis than women (mean [SD] age, 61.8 [12.8] vs 48.6 [11.4] years). In addition, 1321 patients (13.7%) had bilateral or multiple BC, with age at first onset similar to the diagnosis age for patients with 1 BC (mean [SD] age, 49.0 [10.7] vs 48.7 [11.6] years), and 641 patients (6.6%) had additional cancer primaries. Of 2015 patients with OV, 445 (21.7%) received a diagnosis at 45 years or younger.
Most patients with BC (6829 [70.8%]) had invasive ductal breast cancer (IDC) (Table 1). Invasive lobular breast cancer (ILC) was less common (493 patients [5.1%]) and developed in patients at a slightly older age than those with IDC or other histologic types of BC. Of the 4447 patients with BC who also had information on tumor pathologic features (46.1%) (eTable 2 in the Supplement), 2476 (55.7%) were hormone (estrogen and progesterone) receptor–positive, 278 (6.3%) were human epidermal growth factor receptor–positive but hormone receptor–negative, 539 (12.1%) were triple-positive (TPBC), and 1154 (26.0%) were triple-negative (TNBC). The mean [SD] age of patients who were TPBC (45.3 [10.7] years) and hormone receptor–negative (45.4 [10.9] years) was less than that of patients who were TNBC (48.8 [11.1] years) and hormone receptor–positive (50.1 [11.6] years). Demographics and BC-associated gene frequencies for patients with BC without pathology information were similar to those described above.
In addition, 8152 patients with BC (84.6%) and 1670 patients with OV (81.4%) had a family history of cancer (Table 1). Also, 2060 patients with BC or OV (18.0%) provided information on BRCA1/2 testing prior to exome sequencing in our laboratory, of whom 2034 (98.7%) had received a negative result. Thus, the cases described in this report are enriched for noncarriers of the deleterious BRCA1/2 mutations, which resulted in underestimation of risks for BRCA1/2 genes.
The 3988 in-laboratory controls were younger than the patients with BC or OV at the time of testing (mean [SD] age, 39.7 [14.7] vs 50.9 [14.8] years). Although the racial/ethnic distribution among controls was somewhat similar to that of cases, the proportion of white individuals was slightly lower among controls and the proportion of Hispanic individuals was slightly higher among controls (eFigure 3 in the Supplement).
Breast Cancer Susceptibility Genes
We identified 4 genes that confer significantly increased BC risk (Table 2 and eTable 3 in the Supplement), including a newly identified BC association with the Lynch syndrome28 susceptibility gene MSH6. Our findings were supported by comparison with the gnomAD database as a reference population, in which all 4 genes were present and significantly associated with similar or higher estimated risks (eTable 3 in the Supplement). For both comparisons, PALB2 was associated with high BC risk, consistent with other studies,10,18 whereas ATM and CHEK2 conferred 2- to 3-fold risks, also consistent with prior reports.10,18 The CHEK2 founder mutation c.1100delC was enriched among mutation carriers (eTable 14 in the Supplement) and was significantly associated with BC risk (OR, 4.00; 95% CI, 2.04-8.73; uncorrected P = 2.66 × 10−6, FDR-corrected P = .003). Without this variant, the association between CHEK2 and BC was substantially attenuated and nonsignificant (all other variants combined: OR, 1.42; 95% CI, 0.76-2.81, uncorrected P = .32). Although TP53 is known to confer risk for Li-Fraumeni syndrome, our findings indicated that TP53 might confer moderate to high risk for BC at marginal significance (eTable 3 in the Supplement), in agreement with a recent report.29 The MMR gene MSH6 was associated with moderate BC risk, even after excluding 981 BC cases (Table 1) with first- and second-degree OV-affected relatives (eTable 7 in the Supplement). The MSH6 mutation c.2945delC, enriched in MSH6 mutation carriers (eTable 14 in the Supplement), primarily accounted for the observed association (c.2945delC variant-specific OR, 2.73; 95% CI, 1.22-7.18, uncorrected P = .009 vs all other variants combined OR, 2.21; 95% CI, 0.74-8.89, uncorrected P = .18). Increased risks were observed for 4 genes among white individuals (eTable 4 in the Supplement) and women (eTable 6 in the Supplement), although power was diminished for these stratified analyses.
Table 2. Estimated ORs for Associations With Breast Cancer and Its Subtypes for Casesa.
| Gene | Breast Cancer | Male Breast Cancer | ||||
|---|---|---|---|---|---|---|
| Mutated Alleles, No.
(Total No. of Alleles) |
OR (95% CI) | P Value | Mutated Alleles, No. (Total No. of Alleles) |
OR (95% CI) | P Value | |
| ATMb | 79 (11 512) | 2.97 (1.67-5.68) | 4.02 × 10−5 | 0 (137) | ND | ND |
| CHEK2b | 110 (13 553) | 2.19 (1.40-3.56) | 2.66 × 10−4 | 1 (152) | 1.66 (0.04-10.34) | .46 |
| MSH6 | 65 (17 362) | 2.59 (1.35-5.44) | 1.69 × 10−3 | 0 (246) | ND | >.99 |
| PALB2b | 61 (15 532) | 5.53 (2.24-17.65) | 6.50 × 10−6 | 3 (180) | 22.73 (3.50-118.22) | 8.86 × 10−4 |
| Invasive Ductal Carcinoma | Invasive Lobular Carcinoma | |||||
| ATMb | 41 (6659) | 2.67 (1.42-5.30) | 1.01 × 10−3 | 6 (738) | 3.50 (1.10-9.73) | .02 |
| CHEK2b | 60 (7190) | 2.11 (1.30-3.55) | 1.39 × 10−3 | 8 (960) | 2.49 (0.96-5.74) | .05 |
| MSH6 | 27 (10 670) | 1.79 (0.86-4.01) | .10 | 7 (1180) | 4.23 (1.39-11.97) | 5.63 × 10−3 |
| PALBb | 45 (9466) | 6.91 (2.75-22.33) | 4.09 × 10−7 | 1 (1131) | 1.34 (0.03-11.97) | .56 |
| ER+, PR+, and HER2− | ER−, PR−, and HER2+ | |||||
| ATM b | 23 (2959) | 3.39 (1.67-7.14) | 2.96 × 10−4 | 0 (316) | ND | ND |
| CHEK2b | 33 (3152) | 2.66 (1.52-4.71) | 3.56 × 10−4 | 2 (328) | 1.54 (0.18-6.25) | .39 |
| MSH6 | 13 (4724) | 1.95 (0.81-4.81) | .14 | 2 (534) | 2.67 (0.29-12.29) | .20 |
| PALB2b | 17 (4210) | 5.87 (2.08-20.36) | 1.68 × 10−4 | 1 (451) | 3.09 (0.07-27.68) | .31 |
| Triple-Positive Breast Cancer | Triple-Negative Breast Cancer | |||||
| ATMb | 4 (653) | 2.70 (0.65-8.64) | .09 | 3 (1321) | 0.99 (0.18-3.55) | >.99 |
| CHEK2b | 9 (783) | 3.14 (1.28-7.03) | 6.50 × 10−3 | 3 (1470) | 0.51 (0.10-1.70) | .34 |
| MSH6 | 0 (1032) | 0.00 (0.00-3.02) | .63 | 7 (2222) | 2.25 (0.74-6.36) | .09 |
| PALB2b | 4 (830) | 6.61 (1.31-30.77) | .01 | 11 (1936) | 8.27 (2.65-30.37) | 5.40 × 10−5 |
Abbreviations: ER, estrogen receptor; HER2, human epidermal growth factor receptor 2; PR, progesterone receptor; ND, not determined; OR, odds ratio; +, positive for the receptor; −, negative for the receptor.
Derived from Fisher exact test.
Known breast or ovarian cancer gene.
No significant BC associations were observed with BARD1, NF1, or PTEN after correction for multiple tests (all FDR-corrected P > .10). Moreover, the MMR genes MLH1, MSH2, and PMS2, the MRN complex genes MRE11A, RAD50, and NBN, the OV susceptibility genes BRIP1, RAD51C, and RAD51D, and the melanoma predisposition gene CDKN2A did not significantly increase BC risk (all FDR-corrected P > .20). No increased BC risk was observed for CDH1 (FDR-corrected P = .99) given its primary association with ILC30 and the enrichment of IDC cases in the present study (Table 1).
We also identified roles for high- and moderate-risk BC genes in disease pathology (Table 3). The PALB2 mutations were associated with greater than 5-fold risks of IDC, hormone receptor–positive BC, TPBC, TNBC, and male BC (MBC); association with TNBC has been previously noted.31 The ATM and CHEK2 genes conferred moderate IDC, ILC, and hormone receptor–positive risks; CHEK2 was also associated with moderate TPBC risk. The MSH6 gene was more closely associated with ILC than with IDC. Moreover ATM, CHEK2, MSH6, and PALB2 were associated with early-onset BC with similar or higher risks than those estimated for overall breast cancer.
Table 3. Candidate Genes Associated With Early-Onset Breast or Ovarian Cancera.
| Gene | Cases, Mutated Alleles, No. (Total No. of Alleles) |
Controls, Mutated Alleles, No.
(Total No. of Alleles) |
OR (95% CI)b | P Value |
|---|---|---|---|---|
| Breast cancer | ||||
| ATMc | 34 (4804) | 14 (6038) | 3.07 (1.60-6.19) | 2.26 × 10−4 |
| CHEK2c | 57 (5630) | 24 (6442) | 2.73 (1.67-4.61) | 1.69 × 10−5 |
| MSH6 | 29 (7601) | 11 (7806) | 2.71 (1.31-6.03) | 3.90 × 10−3 |
| PALB2c | 24 (6812) | 5 (7314) | 5.17 (1.93-17.35) | 2.40 × 10−4 |
| Ovarian cancer | ||||
| ATM | 1 (492) | 14 (6095) | 0.88 (0.02-5.84) | >.99 |
| MSH6c | 8 (852) | 11 (7826) | 6.73 (2.34-18.43) | 2.35 × 10−4 |
| RAD51Cc | 1 (344) | 0 (4804) | ND | .07 |
| TP53c | 3 (749) | 1 (7209) | 28.96 (2.32-1506.64) | 3.09 × 10−3 |
Abbreviation: ND, not determined.
Diagnosis age 45 years or younger.
Derived from Fisher exact test.
Known breast or ovarian cancer gene.
In the case-case analyses of BC clinical features, ATM and CHEK2 mutations were enriched among estrogen-progesterone–positive patients compared with estrogen-progesterone–negative patients (eTable 10 in the Supplement); CHEK2 may contribute to early onset of BC compared with an onset age of at least 45 years (eTable 11 in the Supplement).
Ovarian Cancer Susceptibility Genes
Four genes were significantly associated with increased OV risks in both case vs in-laboratory control and case vs gnomAD comparisons (eTable 3 in the Supplement). We confirmed associations with known OV susceptibility genes RAD51C (OR, not estimable; FDR-corrected P = .004) and TP53 (OR, 18.50; 95% CI, 2.56-808.10). An OR for RAD51C could not be estimated because of the absence of pathogenic variants among in-laboratory controls; however, RAD51C was associated with high OV risk in the comparison with gnomAD, in agreement with previous reports.32,33 Ovarian cancer risks associated with MSH6 or ATM or both have only recently been estimated.29,34,35 We observed moderately increased OV risk associated with MSH6 (OR, 4.16; 95% CI, 1.95-9.47) and ATM (OR, 2.85; 95% CI, 1.30-6.32). However, BRIP1, RAD51D, CDKN2A, and the MRN complex genes MRE11A, RAD50, and NBN were not significantly associated with elevated OV risk. All 4 genes increased OV risks among white individuals (eTable 4 in the Supplement) and among women (eTable 6 in the Supplement); 1 gene, MSH6, was significantly associated with early-onset OV (Table 3).
Discussion
We undertook large-scale whole-exome sequencing of 15 404 patients referred for genetic testing to provide new insights into predisposition genes for breast and ovarian cancers. Owing to the lack of significantly powered studies, the number and selection of risk genes for hereditary cancer testing have not been standardized,36 and robust evidence for association with BC or OV risk is only available for a modest set of characterized genes typically included in most panels.10,11 This is particularly the case for moderate-penetrance genes, for which there exists either conflicting evidence or insufficient data for reliable estimation.14
Many of our findings regarding the known BC or OV or both susceptibility genes are consistent with the literature reported to date.10,11,14,18 However, some anticipated associations were not observed and other associations were newly identified in this study of case patients and controls. Moreover, we were able to examine associations with clinical features of breast cancer and found that ATM and CHEK2 carriers were more likely to have estrogen-progesterone receptor–positive disease than estrogen-progesterone receptor–negative disease, a finding partially supported by a previous study examining the CHEK2 association with estrogen receptor–positive breast cancer.37 The role of PALB2 as a high penetrance BC gene has been established,10,18 but evidence for its association with OV is mixed.13,34 Consistent with Ramus et al,13 we did not observe elevated OV risk associated with PALB2 (OR, 2.1; 95% CI, 0.5-9.1; P = .31). We also observed increase risk for multiple breast cancer subtypes associated with PALB2, including MBC, for which prior evidence is scant.31 The RAD51C gene conferred OV but not BC risk, in agreement with previous reports indicating that it may be a high penetrance OV32,33 but not BC gene.2,10 Likewise, RAD51D was recently reported to increase BC risk18; however, we and others observed no such association.12 Moreover, we and others18,38 failed to find any association of BRIP1 with BC. The MMR genes MSH2 and PMS2 were also not associated with increased BC risk, in agreement with recent findings.18 Despite earlier reports15,16,17 of BC and OV associations with MRN complex genes and CDKN2A, we and others29 did not observe higher BC or OV risk associated with those genes.
We also identified BC and OV associations with additional genes involved in genomic stability pathways. Protein kinase produced by ATM, a gene known to be associated with BC, plays a crucial role in the early stages of homologous recombination and regulation of cell cycle checkpoints.39 However, prior evidence linking ATM and OV is limited. In both comparisons with in-laboratory and gnomAD controls, among all participants or a subset of white individuals, we observed a 2- to 3-fold increased risk of OV associated with ATM, with effect sizes similar to or slightly higher than those in recent observations.29,34,35 The well-characterized MMR complex also has a role in maintaining genomic stability, repairing small genomic alterations and ensuring error-free homologous recombination repair. Mutations in MMR genes have been shown to increase susceptibility to BC, OV, or both,39 although evidence to date regarding MSH6 is inconsistent. Whereas some studies have failed to observe significantly increased BC or OV risks associated with MSH6,28,39 others have reported it to be an OV susceptibility gene.40,41 Our data suggest that MSH6 has a strong significant association with OV. We also observed modestly increased risk for BC, early-onset BC, and the ILC subtype associated with MSH6, in agreement with a small prior study that reported BC association but failed to identify the MSH6 pathogenic variants.42 The MSH6 c.2945delC variant, which is reported herein with relatively high frequency, is a known pathogenic variant and therefore no upfront orthogonal confirmation was performed, per the research study design. We suspect that at least some of these are false-positive calls. Further analysis of the surrounding sequence suggests a mechanism for the false-positive calls due to polymerase slippage. Our findings support MSH6 as a weak BC susceptibility gene.18 These 2 associations were confirmed in multiple sensitivity analyses, including comparisons with gnomAD and analyses among white individuals and women.
Limitations and Strengths
Although many of our findings are supported by the literature, there are some limitations to the present study. First, we had incomplete information on the personal disease history of the in-laboratory controls and cannot rule out the possibility that some individuals may have been affected with BC or OV. The same issue underlies the use of Exome Aggregation Consortium and gnomAD data sets, particularly because approximately 6% of the gnomAD samples were from The Cancer Genome Atlas. However, analysis of affected controls would result in a bias toward null and implies that the true risks are greater than what we reported here. Second, we based our analysis on the aggregation of both protein truncating and known pathogenic variants, which could result in underestimation of risk for less-studied genes. Third, information on hormone receptor status was available for only approximately 46% of cases. Although power was diminished for these analyses because of the reduced sample size, inferences were unlikely to have selection bias because demographics and mutation frequencies of cases with or without pathology data were comparable (eTable 2 in the Supplement). Last, our study was composed of patients referred for genetic testing and may therefore be enriched for individuals with early-onset BC or OV, bilateral and TNBC, Lynch syndrome–associated cancers, and a family history of BC, OV, or other cancers. We showed that BC associations persisted despite exclusion of cases with Lynch syndrome and other cancers and exclusion of cases with close relatives affected with OV. Although the enrichment of personal and family histories for patients with BC may have afforded us an opportunity to identify new associations, particularly those of moderate risk, these characteristics should be considered when generalizing our study findings.
Despite these limitations, our study had several strengths. As a clinical diagnostic laboratory processing hundreds of thousands of samples annually, we were able to generate high-quality and high-sensitivity next-generation sequencing results (eFigures 3-5 in the Supplement) linked to detailed patient phenotype, personal and family history, and demographic information. The high resolution of sequencing data using a large sample of patients and controls enabled us to identify novel associations of BC and OV with susceptibility genes in addition to providing evidence for reported but unconfirmed associations. Although other large studies relying solely on the use of publicly available reference data sets13,31,34 are susceptible to bias when estimating genetic risks because of systematic differences in sequencing methods among cases and controls,43,44 we uniformly sequenced and analyzed cases and controls with the same technology and in the same laboratory to avoid this bias. We were then able to support our results by comparisons with the gnomAD database, while requiring 10× minimum site coverage to avoid inflation of estimates due to no or low coverage. Moreover, suspected pathogenic variants in the associated genes were subsequently confirmed by Sanger sequencing, eliminating the false-positives that are commonly present in next-generation sequencing results.45
Conclusions
Our findings in a large sample of patients referred for genetic testing confirmed several known or suspected associations with BC or OV and implicate new roles for genes involved in genomic maintenance. These results, therefore, have the potential to serve as the foundation for future epidemiologic, clinical, or functional studies of BC or OV and to inform comprehensive genetic testing and clinical practice.
eTable 1. Search Window of 625 Cancer-Associated Genes
eTable 2. Clinical Characteristics of BC and OV Patients and In-Lab Controls
eTable 2. Clinical Characteristics of BC Patients With or Without Pathologic Information
eTable 3. BC and OV Associations Identified From Comparisons of Cases With In-Lab Controls and Reference Population
eTable 4. BC and OV Associations Among Caucasians
eTable 5. BC and OV Associations Tested With the Combined Multivariate Collapsing Method
eTable 6. BC and OV Associations Tested Among Females
eTable 7. BC Associations After Excluding Patients With First- and Second-Degree Relatives With OV
eTable 8. BC and OV Associations After Excluding Patients With Other Cancers
eTable 9. BC and OV Associations Stratified by Loss of Function, Missense and Other Types of Variants
eTable 10. Case-Case Analysis (ER+PR+ vs ER-PR-b) Among BC Patients
eTable 11. Case-Case Analysis (Onset Age <45 vs ≥45 yr) Among BC and OV Patients
eTable 12. Pathogenic Variants in BC Patients
eTable 13. Pathogenic Variants in OV Patients
eTable 14. Pathogenic Variants in In-Lab Controls
eFigure 1. Power (1-β) Versus Significance Level (α) Given 10,000 BC, 2000 OV Cases and 4000 Controls
eFigure 2. Distribution
eFigure 3. Inferred Ethnicity of (a) 11,416 Cancer Patients and (b) 3,988 In-Lab Controls
eFigure 4. Quality Metrics of Samples
eFigure 5. Characteristics of Detected Variants
eFigure 6. Evaluation of Variant Calling Accuracy
eAppendix. Methods
References
- 1.Rahman N. Realizing the promise of cancer predisposition genes. Nature. 2014;505(7483):302-308. doi: 10.1038/nature12981 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Apostolou P, Fostira F. Hereditary breast cancer: the era of new susceptibility genes. Biomed Res Int. 2013;2013:747318. doi: 10.1155/2013/747318 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Toss A, Tomasello C, Razzaboni E, et al. Hereditary ovarian cancer: not only BRCA 1 and 2 genes. Biomed Res Int. 2015;2015:341723. doi: 10.1155/2015/341723 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Easton DF. How many more breast cancer predisposition genes are there? Breast Cancer Res. 1999;1(1):14-17. doi: 10.1186/bcr6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Pharoah PD, Antoniou A, Bobrow M, Zimmern RL, Easton DF, Ponder BA. Polygenic susceptibility to breast cancer and implications for prevention. Nat Genet. 2002;31(1):33-36. doi: 10.1038/ng853 [DOI] [PubMed] [Google Scholar]
- 6.Seal S, Thompson D, Renwick A, et al. ; Breast Cancer Susceptibility Collaboration (UK) . Truncating mutations in the Fanconi anemia J gene BRIP1 are low-penetrance breast cancer susceptibility alleles. Nat Genet. 2006;38(11):1239-1241. doi: 10.1038/ng1902 [DOI] [PubMed] [Google Scholar]
- 7.Rahman N, Seal S, Thompson D, et al. ; Breast Cancer Susceptibility Collaboration (UK) . PALB2, which encodes a BRCA2-interacting protein, is a breast cancer susceptibility gene. Nat Genet. 2007;39(2):165-167. doi: 10.1038/ng1959 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Renwick A, Thompson D, Seal S, et al. ; Breast Cancer Susceptibility Collaboration (UK) . ATM mutations that cause ataxia-telangiectasia are breast cancer susceptibility alleles. Nat Genet. 2006;38(8):873-875. doi: 10.1038/ng1837 [DOI] [PubMed] [Google Scholar]
- 9.CHEK2 Breast Cancer Case-Control Consortium CHEK2*1100delC and susceptibility to breast cancer: a collaborative analysis involving 10,860 breast cancer cases and 9,065 controls from 10 studies. Am J Hum Genet. 2004;74(6):1175-1182. doi: 10.1086/421251 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Easton DF, Pharoah PD, Antoniou AC, et al. Gene-panel sequencing and the prediction of breast-cancer risk. N Engl J Med. 2015;372(23):2243-2257. doi: 10.1056/NEJMsr1501341 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Tung N, Lin NU, Kidd J, et al. Frequency of germline mutations in 25 cancer susceptibility genes in a sequential series of patients with breast cancer. J Clin Oncol. 2016;34(13):1460-1468. doi: 10.1200/JCO.2015.65.0747 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Loveday C, Turnbull C, Ramsay E, et al. ; Breast Cancer Susceptibility Collaboration (UK) . Germline mutations in RAD51D confer susceptibility to ovarian cancer. Nat Genet. 2011;43(9):879-882. doi: 10.1038/ng.893 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ramus SJ, Song H, Dicks E, et al. ; AOCS Study Group; Ovarian Cancer Association Consortium . Germline mutations in the BRIP1, BARD1, PALB2, and NBN genes in women with ovarian cancer. J Natl Cancer Inst. 2015;107(11):djv214. doi: 10.1093/jnci/djv214 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Tung N, Domchek SM, Stadler Z, et al. Counselling framework for moderate-penetrance cancer-susceptibility mutations. Nat Rev Clin Oncol. 2016;13(9):581-588. doi: 10.1038/nrclinonc.2016.90 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Debniak T, Górski B, Huzarski T, et al. . A common variant of CDKN2A (p16) predisposes to breast cancer. J Med Genet. 2005;42(10):763-765. doi: 10.1136/jmg.2005.031476 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Borg A, Sandberg T, Nilsson K, et al. High frequency of multiple melanomas and breast and pancreas carcinomas in CDKN2A mutation-positive melanoma families. J Natl Cancer Inst. 2000;92(15):1260-1266. doi: 10.1093/jnci/92.15.1260 [DOI] [PubMed] [Google Scholar]
- 17.Damiola F, Pertesi M, Oliver J, et al. Rare key functional domain missense substitutions in MRE11A, RAD50, and NBN contribute to breast cancer susceptibility: results from a Breast Cancer Family Registry case-control mutation-screening study. Breast Cancer Res. 2014;16(3):R58. doi: 10.1186/bcr3669 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Couch FJ, Shimelis H, Hu C, et al. Associations between cancer predisposition testing panel genes and breast cancer. JAMA Oncol. 2017;3(9):1190-1196. doi: 10.1001/jamaoncol.2017.0424 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Katsanis SH, Katsanis N. Molecular genetic testing and the future of clinical genomics. Nat Rev Genet. 2013;14(6):415-426. doi: 10.1038/nrg3493 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Lincoln SE, Kobayashi Y, Anderson MJ, et al. . A systematic comparison of traditional and multigene panel testing for hereditary breast and ovarian cancer genes in more than 1000 patients. J Mol Diagn. 2015;17(5):533-544. doi: 10.1016/j.jmoldx.2015.04.009 [DOI] [PubMed] [Google Scholar]
- 21.Walsh T, Casadei S, Lee MK, et al. Mutations in 12 genes for inherited ovarian, fallopian tube, and peritoneal carcinoma identified by massively parallel sequencing. Proc Natl Acad Sci U S A. 2011;108(44):18032-18037. doi: 10.1073/pnas.1115052108 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Landrum MJ, Lee JM, Benson M, et al. ClinVar: public archive of interpretations of clinically relevant variants. Nucleic Acids Res. 2016;44(D1):D862-D868. doi: 10.1093/nar/gkv1222 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Richards S, Aziz N, Bale S, et al. ; ACMG Laboratory Quality Assurance Committee . Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med. 2015;17(5):405-424. doi: 10.1038/gim.2015.30 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Morgenthaler S, Thilly WG. A strategy to discover genes that carry multi-allelic or mono-allelic risk for common diseases: a cohort allelic sums test (CAST). Mutat Res. 2007;615(1-2):28-56. doi: 10.1016/j.mrfmmm.2006.09.003 [DOI] [PubMed] [Google Scholar]
- 25.Basu S, Pan W. Comparison of statistical tests for disease association with rare variants. Genet Epidemiol. 2011;35(7):606-619. doi: 10.1002/gepi.20609 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Lu C, Xie M, Wendl MC, et al. Patterns and functional implications of rare germline variants across 12 cancer types. Nat Commun. 2015;6:10086. doi: 10.1038/ncomms10086 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Le Scouarnec S, Karakachoff M, Gourraud JB, et al. Testing the burden of rare variation in arrhythmia-susceptibility genes provides new insights into molecular diagnosis for Brugada syndrome. Hum Mol Genet. 2015;24(10):2757-2763. doi: 10.1093/hmg/ddv036 [DOI] [PubMed] [Google Scholar]
- 28.Bonadona V, Bonaïti B, Olschwang S, et al. ; French Cancer Genetics Network . Cancer risks associated with germline mutations in MLH1, MSH2, and MSH6 genes in Lynch syndrome. JAMA. 2011;305(22):2304-2310. doi: 10.1001/jama.2011.743 [DOI] [PubMed] [Google Scholar]
- 29.Kurian AW, Hughes E, Handorf EA, et al. . Breast and ovarian cancer penetrance estimates derived from germline multiple-gene sequencing results in women. JCO Precision Oncol. 2017;(1):1-12. doi: 10.1200/PO.16.00066 [DOI] [PubMed] [Google Scholar]
- 30.Dossus L, Benusiglio PR. Lobular breast cancer: incidence and genetic and non-genetic risk factors. Breast Cancer Res. 2015;17:37. doi: 10.1186/s13058-015-0546-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Couch FJ, Hart SN, Sharma P, et al. Inherited mutations in 17 breast cancer susceptibility genes among a large triple-negative breast cancer cohort unselected for family history of breast cancer. J Clin Oncol. 2015;33(4):304-311. doi: 10.1200/JCO.2014.57.1414 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Loveday C, Turnbull C, Ruark E, et al. ; Breast Cancer Susceptibility Collaboration (UK) . Germline RAD51C mutations confer susceptibility to ovarian cancer. Nat Genet. 2012;44(5):475-476. doi: 10.1038/ng.2224 [DOI] [PubMed] [Google Scholar]
- 33.Song H, Dicks E, Ramus SJ, et al. Contribution of germline mutations in the RAD51B, RAD51C, and RAD51D genes to ovarian cancer in the population. J Clin Oncol. 2015;33(26):2901-2907. doi: 10.1200/JCO.2015.61.2408 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Norquist BM, Harrell MI, Brady MF, et al. Inherited mutations in women with ovarian carcinoma. JAMA Oncol. 2016;2(4):482-490. doi: 10.1001/jamaoncol.2015.5495 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Lilyquist J, LaDuca H, Polley E, et al. Frequency of mutations in a large series of clinically ascertained ovarian cancer cases tested on multi-gene panels compared to reference controls. Gynecol Oncol. 2017;147(2):375-380. doi: 10.1016/j.ygyno.2017.08.030 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Thompson ER, Rowley SM, Li N, et al. Panel testing for familial breast cancer: calibrating the tension between research and clinical care. J Clin Oncol. 2016;34(13):1455-1459. doi: 10.1200/JCO.2015.63.7454 [DOI] [PubMed] [Google Scholar]
- 37.Decker B, Allen J, Luccarini C, et al. Rare, protein-truncating variants in ATM, CHEK2 and PALB2, but not XRCC2, are associated with increased breast cancer risks. J Med Genet. 2017;54(11):732-741. doi: 10.1136/jmedgenet-2017-104588 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Easton DF, Lesueur F, Decker B, et al. ; Australian Ovarian Cancer Study Group; kConFab Investigators; Lifepool Investigators; NBCS Investigators . No evidence that protein truncating variants in BRIP1 are associated with breast cancer risk: implications for gene panel testing. J Med Genet. 2016;53(5):298-309. doi: 10.1136/jmedgenet-2015-103529 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Nielsen FC, van Overeem Hansen T, Sørensen CS. Hereditary breast and ovarian cancer: new genes in confined pathways. Nat Rev Cancer. 2016;16(9):599-612. doi: 10.1038/nrc.2016.72 [DOI] [PubMed] [Google Scholar]
- 40.Song H, Cicek MS, Dicks E, et al. . The contribution of deleterious germline mutations in BRCA1, BRCA2 and the mismatch repair genes to ovarian cancer in the population. Hum Mol Genet. 2014;23(17):4703-4709. doi: 10.1093/hmg/ddu172 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Baglietto L, Lindor NM, Dowty JG, et al. ; Dutch Lynch Syndrome Study Group . Risks of Lynch syndrome cancers for MSH6 mutation carriers. J Natl Cancer Inst. 2010;102(3):193-201. doi: 10.1093/jnci/djp473 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Wasielewski M, Riaz M, Vermeulen J, et al. Association of rare MSH6 variants with familial breast cancer. Breast Cancer Res Treat. 2010;123(2):315-320. doi: 10.1007/s10549-009-0634-4 [DOI] [PubMed] [Google Scholar]
- 43.Lee S, Kim S, Fuchsberger C. Improving power for rare-variant tests by integrating external controls. Genet Epidemiol. 2017;41(7):610-619. doi: 10.1002/gepi.22057 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Hendricks A, Billups S, Zeggini E, Barroso I, Santorico SA, Dupuis J. A New Method for Genetic Region Association Testing With Massively Different Sequencing Depths of Coverage. Vancouver, BC, Canada: The American Society of Human Genetics; 2016. [Google Scholar]
- 45.Mu W, Lu HM, Chen J, Li S, Elliott AM. Sanger confirmation is required to achieve optimal sensitivity and specificity in next-generation sequencing panel testing. J Mol Diagn. 2016;18(6):923-932. doi: 10.1016/j.jmoldx.2016.07.006 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Search Window of 625 Cancer-Associated Genes
eTable 2. Clinical Characteristics of BC and OV Patients and In-Lab Controls
eTable 2. Clinical Characteristics of BC Patients With or Without Pathologic Information
eTable 3. BC and OV Associations Identified From Comparisons of Cases With In-Lab Controls and Reference Population
eTable 4. BC and OV Associations Among Caucasians
eTable 5. BC and OV Associations Tested With the Combined Multivariate Collapsing Method
eTable 6. BC and OV Associations Tested Among Females
eTable 7. BC Associations After Excluding Patients With First- and Second-Degree Relatives With OV
eTable 8. BC and OV Associations After Excluding Patients With Other Cancers
eTable 9. BC and OV Associations Stratified by Loss of Function, Missense and Other Types of Variants
eTable 10. Case-Case Analysis (ER+PR+ vs ER-PR-b) Among BC Patients
eTable 11. Case-Case Analysis (Onset Age <45 vs ≥45 yr) Among BC and OV Patients
eTable 12. Pathogenic Variants in BC Patients
eTable 13. Pathogenic Variants in OV Patients
eTable 14. Pathogenic Variants in In-Lab Controls
eFigure 1. Power (1-β) Versus Significance Level (α) Given 10,000 BC, 2000 OV Cases and 4000 Controls
eFigure 2. Distribution
eFigure 3. Inferred Ethnicity of (a) 11,416 Cancer Patients and (b) 3,988 In-Lab Controls
eFigure 4. Quality Metrics of Samples
eFigure 5. Characteristics of Detected Variants
eFigure 6. Evaluation of Variant Calling Accuracy
eAppendix. Methods