Frequency of mutations in individuals with breast cancer referred for BRCA1 and BRCA2 testing using next-generation sequencing with a 25-gene panel

Abstract

BACKGROUND

Next-generation sequencing (NGS) allows for simultaneous sequencing of multiple cancer susceptibility genes and, for an individual, may be more efficient and less expensive than sequential testing. The authors assessed the frequency of deleterious germline mutations among individuals with breast cancer who were referred for BRCA1 and BRCA2 (BRCA1/2) gene testing using a panel of 25 genes associated with inherited cancer predisposition.

METHODS

This was a cross-sectional study using NGS in 2158 individuals, including 1781 who were referred for commercial BRCA1/2 gene testing (cohort 1) and 377 who had detailed personal and family history and had previously tested negative for BRCA1/2 mutations (cohort 2).

RESULTS

Mutations were identified in 16 genes, most frequently in BRCA1, BRCA2, CHEK2, ATM, and PALB2. Among the participants in cohort 1, 9.3% carried a BRCA1/2 mutation, 3.9% carried a mutation in another breast/ovarian cancer susceptibility gene, and 0.3% carried an incidental mutation in another cancer susceptibility gene unrelated to breast or ovarian cancer. In cohort 2, the frequency of mutations in breast/ovarian-associated genes other than BRCA1/2 was 2.9%, and an additional 0.8% had an incidental mutation. In cohort 1, Lynch syndrome-related mutations were identified in 7 individuals. In contrast to BRCA1/2 mutations, neither age at breast cancer diagnosis nor family history of ovarian or young breast cancer predicted for other mutations. The frequency of mutations in genes other than BRCA1/2 was lower in Ashkenazi Jews compared with non-Ashkenazi individuals (P=.026).

CONCLUSIONS

Using an NGS 25-gene panel, the frequency of mutations in genes other than BRCA1/2 was 4.3%, and most mutations (3.9%) were identified in genes associated with breast/ovarian cancer. Cancer 2015;121:25–33. © 2014 American Cancer Society.

INTRODUCTION

Breast cancer is expected to cause 40,000 deaths in the United States in 2014.¹ This constitutes 7% of annual cancer deaths in the United States and underscores the importance of identifying at-risk individuals for prevention, early detection, and treatment. At least 10% of breast cancers occur in individuals who have germline mutations in high-penetrance or moderate-penetrance cancer susceptibility genes, which carry a relative cancer risk of >5-fold and 2-fold to 5-fold, respectively.2 Mutations in the early onset breast cancer genes BRCA1 and BRCA2 (BRCA1/2) are responsible for up to half of the heritable mutations in breast cancer. Although germline mutations in other recognized breast cancer susceptibility genes have been identified (eg, ataxia telangiectasia mutated [ATM]; E-cadherin [CDH1]; checkpoint kinase 2 [CHEK2]; partner and localizer of BRCA2 [PALB2]; phosphatase and tensin homolog [PTEN]; serine/threonine kinase 11 [STK11]; and tumor protein 53 [TP53]), mutations in any 1 of these genes are rare; thus, testing 1 gene at a time is both inefficient and expensive. Recently, germline mutations in other genes that function in the same DNA repair pathway as BRCA1/2 have been identified and are associated with hereditary breast cancer (eg, nibrin [NBN], RAD51 paralog C [RAD51C], PALB2, BRCA1 interacting protein C-terminal helicase 1 [BRIP1], and BRCA1 associated RING domain 1 [BARD1]).3-12 Guidelines for appropriate testing are not yet available for these genes.13

With the advent of next-generation sequencing (NGS), simultaneous sequencing of multiple cancer susceptibility genes is available through multiplex panels at a cost comparable to that for single gene testing.14, 15 However, testing multiple genes simultaneously creates the potential for unanticipated findings as well as identification of genetic changes for which clinical management is less clear. These panels are now commercially available, but no study has yet assessed the frequency of germline mutations in cancer susceptibility genes other than BRCA1/2 in a large cohort of individuals with breast cancer who were referred for genetic evaluation.

The primary objective of the current study was to determine the frequency of germline mutations in individuals with a history of breast cancer using a 25-gene NGS panel, including 1781 individuals who were referred for commercial BRCA1/2 testing (cohort 1) and 377 individuals who previously tested negative for BRCA1/2 mutations through an academic high-risk program (cohort 2). Cohort 1 was used to determine the relative frequency of mutations in BRCA1/2 and additional cancer susceptibility genes, and cohort 2 was used to evaluate the frequency of mutations in cancer susceptibility genes in BRCA1/2-negative individuals with known, detailed personal and cancer family histories. A secondary objective was to determine whether any clinical factors (eg, age at cancer diagnosis, extent of cancer family history, and tumor pathology) predict for mutations in additional cancer susceptibility genes, information that could guide the appropriate use of NGS panels for individuals who have tested negative for BRCA1/2.

The 25 genes were selected based on their putative role in the development of hereditary cancers. Most, but not all, are associated with hereditary breast and ovarian cancer. The selected genes are associated with an increased risk of melanoma or breast, colorectal, ovarian, endometrial, prostate, gastric, and pancreatic cancers, and almost all have evidence supporting a greater than 2-fold increase in risk. Although the inclusion of genes not associated with hereditary breast cancer introduces the possibility of incidental findings, it supports the goal of developing a comprehensive hereditary cancer susceptibility panel. And, as additional data emerge, it is reasonable to expect the list of genes appropriate for inclusion in such a panel will expand.

MATERIALS AND METHODS

Patient Characteristics

Cohort 1 was comprised of DNA extracted from 1781 anonymized blood samples consecutively submitted to Myriad Genetic Laboratories (Salt Lake City, Utah) for BRCA1/2 testing between November 2012 and April 2013 from individuals with breast cancer. Clinical information was derived from data submitted. Because 1 objective of this study was to determine the relative frequency of mutations in BRCA1/2 compared with mutations in the 23 other cancer susceptibility genes, and because it is known that BRCA1/2 mutations are more prevalent in Ashkenazi Jews, individuals of Ashkenazi Jewish heritage were excluded to determine the relative frequency of mutations in a generalizable population.

Cohort 2 was comprised of 377 DNA samples extracted from blood of patients with breast cancer who were referred to Beth Israel Deaconess Medical Center for genetic testing between 1998 and 2013. Initially, 384 individuals tested negative for mutations in BRCA1/2 (n=352) or declined clinical testing (n=32) and consented to an institutional review board-approved protocol to collect clinical information and blood for future research. Previous BRCA1/2 testing included 3-site Ashkenazi testing (n=63), full sequencing without large rearrangement testing (n=207), and full sequencing with large rearrangement analysis (n=82). Seven BRCA1/2 mutations were identified through NGS (5 sequence variants and 2 large rearrangements) and were excluded from this cohort. Clinical data included personal and family cancer histories, cancer histology and receptor status, and ancestry. Because all individuals in cohort 2 had already tested negative for BRCA1/2 mutations, Ashkenazi Jews were included in this cohort. This study was approved by the Institutional Review Board of the Dana Farber/Harvard Cancer Center.

NGS Assay

Sample preparation for NGS was performed using the RainDance Thunderstorm emulsion polymerase chain reaction (PCR) system (RainDance Technologies, Billerica, Mass). Modified sample preparation using long-range and nested PCR was used for portions of the CHEK2 and PMS2 postmeiotic segregation increased 2 (PMS2) genes to alleviate the challenge of pseudogenes. NGS was performed using the Illumina HiSeq2500 or MiSeq (Illumina Inc., San Diego, Calif) for the 25 genes (Table 1) (Supporting Methods; see online supporting information).29 Analysis of the NGS data identified both sequence variants and large rearrangements (deletions and duplications). Variants were classified using American College of Medical Genetics and Genomics recommendations (Supporting Methods; see online supporting information).30-32 Gene variants that were deemed deleterious or suspected deleterious were considered mutations. Analyzed genes were categorized into 3 groups (Table 1).

Table 1. Cancer Susceptibility Genes Other Than BRCA1/2

Gene	References	Breast Cancer RR or Inclusion Criterion
Group 1: Breast cancer susceptibility genes (other than BRCA1/2)
TP53	Gonzalez 200916	4.3-9.3
CDH1	Kaurah 2007,17 Pharoah 200118	5.9-7.3
PTEN	Pilarksy 2004,19 Tan 201220	2.0-5.0
ATM	Renwick 200621	1.5-3.8
CHEK2	Weischer 200822	2.7-4.8
STK11	Hearle 2006,23 Lim 200424	2.0-4.0
RAD51C	Meindl 201011	1.5-7.8
PALB2	Rahman 20075	1.4-3.9
BARD1	Ratajska 2012,3 De Brakeleer 201025	Breast cancer association reported: RR not yet determined
BRIP1	Seal 20066	1.2-3.2
NBN	Bogdanova 2008,8 Seemanova 2007,26 Zhang 201227	1.4-6.6
Group 2: Ovarian cancer susceptibility genes (do not confer an increased risk for breast cancer)a
MLH1	—	Lynch syndrome
MSH2	—	Lynch syndrome
MSH6	—	Lynch syndrome
PMS2	—	Lynch syndrome
EPCAM	—	Lynch syndrome
RAD51D	Loveday 201128	Ovarian cancer: RR, 6.3
Group 3: Cancer susceptibility genes unrelated to breast or ovarian cancera
APC	—	Familial adenomatous polyposis
MUTYHb	—	MUTYH-associated polyposis
CDKN2A	—	Melanoma and pancreas cancer syndrome
SMAD4	—	Juvenile polyposis syndrome
CDK4	—	Melanoma syndrome
BMPR1A	—	Juvenile polyposis syndrome

• Abbreviations: APC, adenomatous polyposis coli; ATM, ataxia telangiectasia mutated; BARD1, BRCA1 associated RING domain 1; BMPR1A, bone morphogenetic protein receptor, type 1A; BRCA1/2, early onset breast cancer genes BRCA1 and BRCA2; BRIP1, BRCA1 interacting protein C-terminal helicase 1; CDH1, E-cadherin; CDK4, cyclin-dependent kinase 4; CDKN2A, cyclin-dependent kinase inhibitor 2A; CHEK2, checkpoint kinase 2; EPCAM, epithelial cell adhesion molecule; MLH1, mutL homolog 1; MSH2, mutS homolog 2; MSH6, mutS homolog 6; MUTYH, biallelic mutY homolog; NBN, nibrin; PALB2, partner and localizer of BRCA2; PMS2, PMS2 postmeiotic segregation increased 2; PTEN, phosphatase and tensin homolog; RAD51C, RAD51 paralog C; RAD51D, RAD51 paralog D; RR, relative risk; SMAD4, SMAD family member 4; STK11, serine/threonine kinase 11;TP53, tumor protein 53.
• ^a Mutations in group 3 were deemed “incidental” for all patients as were mutations in group 2 for those without a history of ovarian cancer.
• ^b Only tumors with biallelic MUTYH mutations were considered for this analysis.

RESULTS

Frequency of Mutations

The clinical characteristics of patients in cohorts 1 and 2 are provided in Table 2. In cohort 1, 241 individuals (13.5%) carried a mutation in at least 1 of the genes tested: 162 (67%) had mutations in BRCA1/2, and 76 (32%) had mutations in at least 1 of the other 23 genes (Table 3). In addition, 3 individuals (1%) had a mutation in both BRCA2 and another gene (ATM, CHEK2, or NBN). Two individuals also had 2 different mutations in CHEK2. Thus, 81 mutations in additional cancer susceptibility genes were identified in 79 individuals, 76 of whom did not have a BRCA1/2 mutation. Of the 165 BRCA1/2 mutations identified, 159 were sequence changes, and 6 were large rearrangements. For mutations in other genes, 69 were sequence changes, and 12 were large rearrangements. The frequency of mutations was 9.3% with BRCA1/2 testing and 13.5% with NGS, for a difference of 4.3% (95% confidence interval [CI], 2.1%-6.4%). Considering only mutations in breast/ovarian cancer susceptibility genes relevant to the individual's cancer history (ie, excluding group 2 genes in individuals without ovarian cancer), the frequency was 13.2%, for a difference of 3.9% (95% CI, 1.8%-6.1%). When BRCA1/2 mutation carriers were excluded from cohort 1, a mutation was detected in 76 of 1616 individuals (4.7%; 95% CI, 3.8%-5.9%), and the frequency was 4.3% (95% CI, 3.4%-5.4%) if only relevant breast/ovarian cancer susceptibility genes were considered.

Table 2. Demography and Clinical Characteristics of Patients by Cohort

	No. of Patients (%)
Characteristic	Cohort 1, n=1781	Cohort 2, n=377
Mean age of first cancer diagnosis, y	47.2	45.6
BRCA1/2 mutation present	165 (9.3)	0 (0)
Mean BRCAPRO model scorea	NA	15.9%
Ashkenazi Jewish heritage	0 (0)	94 (24.9)
Personal cancer history
Breast cancer	1735 (97.4)	369 (97.9)
Breast and ovarian cancer	40 (2.2)	7 (1.8)
Breast and pancreatic cancer	6 (0.3)	1 (0.3)
Family cancer historyb
High risk	644 (36.2)	167 (44.3)
Low risk	1137 (63.8)	210 (55.7)
Ancestry
White/Caucasian	911 (51.2)	343 (91)
African	100 (5.6)	13 (3.4)
Latin American/Caribbean	75 (4.2)	10 (2.7)
Asian	43 (2.4)	6 (1.6)
Other/mixed	267 (15)	5 (1.3)
Not specified	385 (21.6)	0 (0)
Sex
Women	1750 (98.3)	374 (99.2)
Men	22 (1.2)	3 (0.8)
Unknown	9 (0.5)	—

• Abbreviations: BRCA1/2, early onset breast cancer genes BRCA1 and BRCA2; NA, not available.
• ^a BRCAPRO (CancerGene v5.1) is a statistical model with associated software for assessing the probability that an individual carries a germline deleterious mutation of the BRCA1 and BRCA2 genes.
• ^b High risk was defined as family history of ovarian cancer at any age or breast cancer before age 50 years in a first-degree or second-degree relative. Any other family history was defined as low risk.

Table 3. Frequency of Mutations by Personal Cancer History

	No. of Patients (%)
	Cohort 1						Cohort 2
			Genes Other Than BRCA1/2					Genes Other Than BRCA1/2
Personal Cancer History	Total No.	BRCA1/2	Group1a	Group 2b	Group 3c	Total	Total No.	Group 1a	Group 2b	Group 3c	Total
Total	1781	165 (9.3)d	71 (4)d	7 (0.4)	1	79 (4.4)	377	11 (2.9)e	0 (0)	3 (0.8)	14 (3.7)
Breast cancer	1735	157d	68d	5	1	74	369	10e	0	3	13
Women	1704	153	63	5	1	69	366	10	0	3	13
Men	22	4	3	0	0	3	3	0	0	0	0
Unspecified	9	0	2	0	0	2	—	—	—	—	—
Breast and ovarian cancer	40	8	2	2	0	4	7	0	0	0	0
Breast and pancreatic cancer	6	0	1	0	0	1	1	1	0	0	1

• Abbreviations: BRCA1/2, early onset breast cancer genes BRCA1 and BRCA2.
• ^a Group 1 includes breast cancer susceptibility genes other than BRCA1/2: TP53, CDH1, PTEN, ATM, CHEK2, STK11, RAD51C, PALB2, BARD1, BRIP1, and NBN (no mutations were identified in PTEN, STK11, or RAD51C).
• ^b Group 2 includes ovarian cancer susceptibility genes other than BRCA1/2 that do not confer an increased risk of breast cancer: Lynch syndrome genes (MLH1, MSH2, MSH6, PMS2, and EPCAM) and RAD51D (no RAD51D mutations were identified).
• ^c Group 3 includes cancer susceptibility genes that do not confer an increased risk of breast or ovarian cancer: APC, MUTYH, CDKN2A, SMAD4, CDK4, and BMPR1A.
• ^d Three women had both a BRCA2 mutation and an additional mutation (ATM, CHEK2, or NBN).
• ^e One woman had a mutation in both ATM and BARD1.

In cohort 2, 15 sequence mutations in additional cancer susceptibility genes were identified in 14 women for a frequency of 3.7%, and the frequency was 2.9% if only breast/ovarian cancer susceptibility genes relevant to the individual's history were considered (Table 3). In cohort 2, no mutations were identified among the 94 Ashkenazi individuals (P=.026).

Of the 22 men in cohort 1, 4 (18.2%) had BRCA1/2 mutations, and 3 (13.6%) had other mutations. The frequency of any mutation was significantly greater in men (31.8%) than in women (13.3%; P=.0210), whereas the frequency of mutations in genes other than BRCA1/2 was relatively higher, but the difference was not statistically significant (13.6% vs 4.1%; P=.0609).

Distribution of Mutations in Genes Other Than BRCA1/2

In genes other than BRCA1/2, CHEK2 mutations were most common, representing approximately 33% of the mutations identified in both cohorts (Table 4). No mutations in either cohort were identified in the following genes: bone morphogenetic protein receptor, type 1A (BMPR1A); cyclin-dependent kinase 4 (CDK4); epithelial cell adhesion molecule (EPCAM); mutL homolog 1 (MLH1); PTEN, RAD51C, RAD51D; SMAD family member 4 (SMAD4); and STK11.

Table 4. Distribution of Mutations in Genes Other Than BRCA1/2

	No. of Patients (%)
Gene	Cohort 1, N=79	Cohort 2, N=14
CHEK2	29 (36.7)a	5 (35.7)
ATM	12 (15.2)a	1 (7.1)
PALB2	12 (15.2)	1 (7.1)
BRIP1	7 (8.9)	—
BARD1	6 (7.6)	1 (7.1)b
NBN	3 (3.8)a	1 (7.1)
TP53	2 (2.5)	—
CDH1	—	2 (14.3)
PMS2	4 (5.1)c	—
MSH6	2 (2.5)	—
MSH2	1 (1.3)d	—
MUTYH	1 (1.3)	1 (7.1)
APC	—	1 (7.1)
CDK2NA	—	1 (7.1)

• Abbreviations: APC, adenomatous polyposis coli; ATM, ataxia telangiectasia mutated; BARD1, BRCA1 associated RING domain 1; BRCA1/2, early onset breast cancer genes BRCA1 and BRCA2; BRIP1, BRCA1 interacting protein C-terminal helicase 1; CDH1, E-cadherin; CDKN2A, cyclin-dependent kinase inhibitor 2A; CHEK2, checkpoint kinase 2; MSH2, mutS homolog 2; MSH6, mutS homolog 6; MUTYH, biallelic mutY homolog; NBN, nibrin; PALB2, partner and localizer of BRCA2; PMS2, PMS2 postmeiotic segregation increased 2; TP53, tumor protein 53.
• ^a One proband each had an additional mutation in BRCA2.
• ^b This proband had an additional mutation in ATM.
• ^c One of these cases was not incidental; the individual had ovarian cancer.
• ^d This was not an incidental case; the individual had ovarian cancer.

In cohort 1, the next most common mutations were in PALB2 (15%) and ATM (15%). Of the 1781 individuals tested, 6 had presumed incidental mutations identified, 5 had mutations in Lynch syndrome genes (ie, mutS homolog 6 [MSH6] and PMS2), and 1 had a biallelic mutY homolog (MUTYH) mutation. Two additional individuals had mutations in Lynch syndrome genes (MSH2 and PMS2), but these were not considered “incidental” because of a history of ovarian cancer.

The mutation distribution by gene in cohort 2 is provided in Table 4. The personal and family cancer history for these mutation carriers is presented in Supporting Table 1 (see online supporting information). Of the 377 individuals, 3 had “incidental” mutations identified in the CDK2NA, MUTYH (biallelic), and adenomatous polyposis coli (APC) genes. Monoallelic mutations in MUTYH were identified in 32 individuals in cohort 1 and in 10 individuals in cohort 2. A complete list of mutations identified in both cohorts is reported in Supporting Data (see online supporting information).

Factors That Predict for Mutations in Genes Other than BRCA1/2

For cohort 1, only aggregate clinical and family history was available because of anonymization restrictions, thus preventing multivariate modeling. Univariate statistical analysis did not identify clinical predictors, such as age at breast cancer diagnosis or strength of family history, for mutations in genes other than BRCA1/2 regardless of whether all genes were analyzed or only group 1 genes (breast cancer susceptibility) were analyzed (Table 5). It is noteworthy that the age at first breast cancer diagnosis was relatively younger for those with BRCA1/2 mutations (median, 42 years) than for those with mutations in other breast cancer susceptibility genes (ie, group 1 genes; median, 45 years), but this difference did not reach statistical significance (P=.075).

Table 5. Frequency of Mutations in Genes Other Than BRCA1/2 by Age at First Breast Cancer Diagnosis and Family Historya

		No. of Patients (%)
			Mutations in Genes Other than BRCA1/2
Age at First Breast Cancer Diagnosis, y	Family Historyb	Mutations in BRCA1/2 Genes	Breast Cancer Susceptibility Genesc	Any Gene	Total No. of Patients
<50	Low risk	61 (7.7)	30 (3.8)	33 (4.2)	789
	High risk	55 (18.4)d	14 (4.7)d	15 (5)d	299
≥50	Low risk	17 (5.4)e	9 (2.9)e	10 (3.2)e	312
	High risk	24 (7.2)	15 (4.5)	16 (4.8)	335

• Abbreviations: BRCA1/2, early onset breast cancer genes BRCA1 and BRCA2.
• ^a Data here are for cohort 1, which included patients with breast cancer only.
• ^b High risk is defined as a family history of ovarian cancer at any age or breast cancer before age 50 years in a first-degree or second-degree relative. Any other family history is defined as low risk.
• ^c The breast cancer susceptibility genes identified other than BRCA1/2 (group 1 genes) were TP53, ATM, CHEK2, PALB2, BARD1, BRIP1, and NBN.
• ^d Two patients had both a BRCA2 mutation and an additional mutation (ATM or NBN).
• ^e One patient had a mutation in both BRCA2 and CHEK2.

For cohort 2, multivariate modeling using both exact logistic regression and hierarchical regression was conducted to identify predictors for carrying mutations in genes other than BRCA1/2. These predictors included age at first cancer diagnosis, number of breast cancers, personal history of other cancers, BRCAPRO (CancerGene v5.1) score,51, 52 breast cancer histology, and the number of relatives with breast or other cancers. None of these variables were statistically significant.

Variants of Uncertain Significance

In cohort 1, a variant of uncertain significance (VUS) was identified in at least 1 of the 25 genes tested in 742 individuals (41.7%), for a total of 1052 variants. Excluding BRCA1/2, 700 individuals (39.3%) had a VUS, for a total of 966 variants (Supporting Data; see online supporting information). All 9 large VUS rearrangements were in additional cancer susceptibility genes, and all other VUS findings were sequence changes. In cohort 2, a VUS was identified in at least 1 of the 22 genes tested in 157 individuals (41.6%), for a total of 231 variants. Other than 1 large rearrangement, all others VUS findings were sequence changes.

DISCUSSION

In the largest investigation of NGS for hereditary breast cancer to date, we report the frequency of mutations in 25 cancer susceptibility genes among 2158 individuals with a history of breast cancer who were referred for BRCA1/2 testing. The frequency of germline mutations was 9.3% with BRCA1/2 testing alone and 13.5% with NGS, or 13.2% if only breast and ovarian cancer susceptibility genes relevant to the individual's cancer history were considered. The frequency of mutations in those who did not have a BRCA1/2 mutation was 4.7% in cohort 1 and 3.7% in cohort 2 (4.3% and 2.9%, respectively, when only breast/ovarian cancer susceptibility genes were considered). The finding that approximately 66% of the mutations identified in cohort 1 were in BRCA1/2 is similar to that reported in another study of women with ovarian cancer who were tested with an NGS assay.15 Our results are also consistent with the 4.3% frequency of germline mutations in additional cancer susceptibility genes detected through analyses of breast cancers in The Cancer Genome Atlas (TCGA) Project.2

Four mutations in high-penetrance breast cancer susceptibility genes other than BRCA1/2 were identified (TP53, CDH1). Mutations in PALB2, a gene that encodes for a BRCA2-interacting protein, were even more common, comprising 15% of the mutations in genes other than BRCA1/2. Although initial reports suggested that the risk of breast cancer with PALB2 mutations was moderate,5 recent reports suggest that some mutations confer a high risk.33-35 Mutations in high-penetrance genes confer a significant risk of subsequent breast cancers, and consideration of prophylactic mastectomies is reasonable for these patients. Individuals with TP53 mutations should avoid radiation if possible, because it may increase their risk of future malignancies,36-38 and those with CDH1 mutations, who are at risk for gastric cancer, should consider prophylactic gastrectomy. Equally important is the identification of family members without an inherited mutation who can avoid unnecessary screening and surgery.

CHEK2 and ATM are moderate-penetrance breast cancer susceptibility genes and confer a 2-fold to 3-fold increase in the risk of breast cancer,21, 22, 39, 40 and mutations in these genes accounted for approximately half of the mutations identified in genes other than BRCA1/2 in both cohorts. It is important that individuals and their physicians understand these risks when choosing prevention strategies and not generalize management recommendations from guidelines written for high-penetrance genes. In addition, CHEK2 mutations frequently do not track with the breast cancer in families,41, 42 prompting speculation that there may be several low-penetrance or moderate-penetrance breast cancer risk genes, including CHEK2, segregating independently within these families. Therefore, clinicians must be cautious about giving reassurance to relatives who test negative for familial mutations in moderate-penetrance genes, because these genes alone may not explain a strong family history of breast cancer.

For mutations in more recently identified genes (eg, NBN, BARD1, BRIP1), the penetrance for breast cancer and the clinical spectrum of other cancers are not well characterized. Larger numbers of mutation carriers will need to be followed to determine these risks. Many of these genes, such as NBN, RAD51C, PALB2, BRIP1, and BARD1, function in the same double-strand DNA break repair pathway as BRCA1 and BRCA2. Preclinical and early clinical data suggest that mutations in some of these genes may predict response to DNA-damaging drugs that are effective in treating BRCA1/2-deficient cancers (eg, platinum and Poly[ADP-ribose] polymerase [PARP]-1 inhibitors).43, 44 These findings and their clinical implications warrant further exploration.

Some identified mutations were not associated with a previously reported breast cancer phenotype or family cancer history, 2 standard triggers for genetic testing. For example, whereas CDH1 mutations are associated with lobular breast cancer and diffuse gastric cancer, 1 of the 2 CDH1 mutation carriers in cohort 2 had an invasive ductal cancer and no family history of gastric cancer in a large family (Supporting Table 1; see online supporting information).17 Similarly, the patient with breast cancer in cohort 2 who had a PALB2 mutation had no family history of pancreatic cancer. Germline mutations in PALB2 have been reported in 3% of families with hereditary pancreatic cancer,45 but data are inconsistent regarding the risk of pancreatic cancer in patients who have breast cancer with a germline mutation.4, 46

Of the 2158 breast cancer patients studied, 9 had mutations in genes that presumably were unrelated to their breast cancer, including 8 mutations in colon cancer susceptibility genes (5 mutations in Lynch syndrome-related genes, 2 biallelic mutations in MUTYH, and 1 APC mutation) and 1 CDK2NA mutation in a patient without a personal or family history of melanoma or pancreatic cancer. Although an increased risk of breast cancer in women with Lynch syndrome has been suggested, this association is still controversial; therefore, these mutations were considered incidental. Monoallelic MUTYH mutations were identified in 42 individuals. These findings may confer a small increase in the risk of colorectal cancer but, when inherited in an autosomal-recessive pattern, have more significant risk implications for offspring.47, 48 Although these mutations are incidental to breast cancer, they are clinically important, because their identification can lead to initiation of screening and prevention strategies for both colon and extracolonic cancers.49

We attempted to identify predictors of mutations in additional cancer susceptibility genes that could guide the selection of panel testing for BRCA1/2-negative individuals. No mutations were identified in Ashkenazi Jewish individuals. Men had a relatively higher frequency of mutations in additional cancer susceptibility genes than women. Both findings must be confirmed in additional studies. However, we were unable to identify any other predictors. Although detailed clinical and family history was available for cohort 2, the cohort may not have been large enough to enable the identification of significant predictive factors. In addition, the cancer spectrum and phenotype may be different for mutations in each gene, necessitating a larger number of mutation carriers before specific predictors for each gene can be elucidated.

We identified at least 1 genetic VUS in approximately 40% of individuals in both cohorts. Further testing of these genes will eventually reclassify these variants as deleterious mutations or benign polymorphisms. Experience with BRCA1/2 suggests that the majority will be reclassified as benign polymorphisms. Initially, approximately 20% of individuals tested for BRCA1/2 had a VUS; whereas, currently, only approximately 2% receive such a result through commercial testing.31, 32 A VUS finding should not be used to manage patients; family history should guide surveillance and prevention recommendations for these individuals.13, 49

Strengths of this study include its large sample size and the extensive personal and family histories for the individuals in cohort 2. These enabled an evaluation of the breast cancer phenotype and cancer clinical spectrum of mutations in genes other than BRCA1/2 (Supporting Table 1; see online supporting information). In addition, assessing the frequency of mutations in 2 different cohorts was valuable. Samples for cohort 1 were obtained from individuals who were undergoing genetic testing in a broad spectrum of clinical settings (ie, both community-based and academic), whereas samples for cohort 2 were accrued solely from a high-risk academic genetic testing program. The frequency of mutations in those who did not have a BRCA1/2 mutation was 4.7% in cohort 1 and 3.7% in cohort 2, suggesting that the occurrence of mutations in genes other than BRCA1/2 may be similar in different breast cancer populations that are suitable for genetic testing. Limitations of this study include the restricted family history for individuals in cohort 1 and the relatively small size of cohort 2, both of which may have precluded the identification of predictors for mutations in genes other than BRCA1/2. Another limitation is the absence of accepted guidelines for managing individuals with many of the mutations identified, which currently limits the clinical value of our findings.

In conclusion, the frequency of mutations in genes other than BRCA1/2 was 4.3% in the NGS 25-gene panel, and most mutations (3.9%) were in genes associated with breast/ovarian cancer. Multiple-gene sequencing may benefit appropriately selected patients, especially those with a personal or family history of more than 1 possible genetic syndrome. At least half of the mutations identified were in moderate-penetrance genes. Although breast cancer risk-management guidelines for these individuals are not yet available, identification of such mutations in combination with family history may justify more intensive surveillance (eg, breast magnetic resonance imaging), prevention (eg, tamoxifen), or both.39 Clinical guidance for such individuals is important to ensure that inappropriate interventions derived from mutations in high-penetrance genes are not pursued. Panel testing may also identify mutations, such as those in Lynch-related genes, that trigger surveillance or prevention strategies for cancers that would not otherwise be justified by family history alone. Identifying a PALB2 or CDK2NA mutation in a patient with a family history of pancreatic cancer may prompt pancreatic cancer surveillance.50 Multigene panel testing identifies several kinds of VUS, which clinicians should anticipate and be prepared to address with their patients. Future research should elucidate the clinical implications of cancer susceptibility genes and will guide evidence-based management recommendations.

FUNDING SUPPORT

This research was supported by grants from the Breast Cancer Research Foundation.

CONFLICT OF INTEREST DISCLOSURES

Dr. Allen, Mr. Kaldate, Dr. Bhatnagar, Dr. Bowles, Dr. Timms, Dr. Roa, Dr. Wenstrup, and Dr. Hartman are employees of Myriad Genetics, Inc. and Myriad Genetic Laboratories, Inc. and receive salaries and stock options. Dr. Tung reports research with Myriad Genetics. Dr. Garber reports research with Myriad, Astra-Zeneca, Novartis, and Pfizer. Dr. Ellisen reports being a paid consultant for Gene Dx.