Variants of cancer susceptibility genes in Korean BRCA1/2 mutation-negative patients with high risk for hereditary breast cancer

Background We evaluated the incidence and spectrum of pathogenic and likely pathogenic variants of cancer susceptibility genes in BRCA1/2 mutation-negative Korean patients with a high risk for hereditary breast cancer using a comprehensive multigene panel that included 35 cancer susceptibility genes. Methods Samples from 120 patients who were negative for BRCA1/2 mutations, but had been diagnosed with breast cancer that was likely hereditary, were prospectively evaluated for the prevalence of high-penetrance and moderate-penetrance germline mutations. Results Nine patients (7.5%) had at least one pathogenic or likely pathogenic variant. Ten variants were identified in these patients: TP53 in two patients, PALB2 in three patients, BARD1 in two patients, BRIP1 in two patients, and MRE11A in one patient. We also identified 30 types of 139 variants of unknown significance (VUS). High-penetrance germline mutations, including TP53 and PALB2, tended to occur with high frequency in young (< 35 years) breast cancer patients (4/19, 21.1%) than in those diagnosed with breast cancer at ≥35 years of age (1/101, 1.0%; p = 0.003). Conclusions These combined results demonstrate that multigene panels offer an alternative strategy for identifying veiled pathogenic and likely pathogenic mutations in breast cancer susceptibility genes. Electronic supplementary material The online version of this article (10.1186/s12885-017-3940-y) contains supplementary material, which is available to authorized users.


Background
The identification of BRCA1 and BRCA2 germline mutations as predictors of cancer susceptibility significantly improved the diagnosis and prevention of hereditary breast and ovarian cancers (HBOC). Recent advances in genetic testing have enabled the discovery of novel genes that increase the risk of cancer in patients with familial predisposition. Multiple research laboratories have evaluated these cancer-associated mutations in patients who are negative for BRCA1/2 mutations, but still have a high risk of HBOC. These efforts have identified mutations in moderate-risk genes, such as ATM, BRIP1, CHEK2, BARD1, MRE11A, NBN, RAD50, RAD51, and XRCC2, as well as those in high-penetrance genes, including TP53, PTEN, STK11, CDH1, and PALB2, have been reported across diverse ethnic populations [1].
Next generation sequencing (NGS) can provide detailed genetic information via multi-gene panel assays [2]. However, the application of NGS multigene panel test in a clinical setting represents a considerable challenge. It is necessary to not only validate this novel technique, but also to select candidate susceptibility genes. Furthermore, mutations indicative of cancer susceptibility vary across ethnicities; therefore, it is important to understand the clinical and genetic characteristics of multiple susceptibility genes identified by NGS multigene panels in each ethnic population.
In this study, we used comprehensive multigene panels that included 35 known or suspected cancer susceptibility genes to examine BRCA1/2 mutation-negative Korean patients who had clinical features indicative of hereditary breast cancer. We also investigated the feasibility of multigene panel testing for Korean patients, and evaluated potential clinicopathological risk factors related to germline mutations other than BRCA1/2.

Study population
The study population included 182 Korean BRCA1/2 mutation-negative breast cancer patients with a familial predisposition who were referred to the Cancer Prevention Center, Yonsei Cancer Center, Seoul, Korea between March 1, 2015 and November 11, 2016. Sixty-two patients opted to not participate. Finally, a total of 120 patients were enrolled in the study. Suspected clinical features of hereditary breast cancer were defined as follows: (1) at least one case of breast or ovarian cancer in first-or second-degree relatives; (2) a first diagnosis of breast cancer before age 40; (3) bilateral breast cancer; and (4) co-diagnosis of breast and ovarian cancers in the same patient.
We identified all single base pair substitutions, insertion-deletions, and copy number variants (CNVs) in each gene. Split-read-based detection of large insertions and deletions was conducted using the Pindel and Manta algorithms. CNVs detected by ExomeDepth software [3] were further crosschecked with our custom pipelines, which retrieved base-level depth of coverage for each binary alignment map (BAM) file using SAMtools software (http://samtools.sourceforge.net) and normalized the depths in the same batch (Additional file 1: Figure S1). All likely deleterious mutations were validated by Sanger sequencing, and all possible large rearrangements were confirmed by the multiplex ligation-dependent probe amplification (MLPA) method (Additional file 1: Figure S2). Genetic variants were classified using a five-tier system following guidelines from the American College of Medical Genetics and Genomics (ACMG) as follows: pathogenic, likely pathogenic, variants of unknown significance (VUS), likely benign, or benign/polymorphism [4]. We used the Sorting Intolerant From Tolerant (SIFT, http://sift. bii.a-star.edu.sg/) and Polymorphism Phenotyping-2 (PolyPhen-2, http://genetics.bwh.harvard.edu/pph2) to generate in silico predictions of several of the identified nonsynonymous variants. Using large rearrangements of exons, pathogenic and likely pathogenic variants were considered as mutations, for consistency with previous studies [5].

Results
Baseline characteristics of the patients are presented in Additional file 2: Table S1. A total of 7.5% (9/120) of patients were found to carry at least one pathogenic or likely pathogenic variant. A total of ten gene variants ( Fig. 1a) were identified in nine patients: TP53 in two patients, PALB2 in three patients, BARD1 in two patients, BRIP1 in two patients, and MRE11A in one patient. We detected a large deletion from exon 2−9 in the TP53 gene, and the other pathogenic variants identified were as follows: PALB2 (c.3267_3268delGT, p.Phe1090SerfsTer6, rs587781890; c.2257C > T, p.Arg753Ter, rs180177110; and c.695delC, p.Gly232ValfsTer6); BARD1 (c.1345C > T, p.Gln449Ter); BRIP1 (c.1066C > T, p.Arg356Ter, rs730881633; and exon 5-6 deletion); and MRE11A (c.1773_1774delAA, p.Gly593LysfsTer4). Likely pathogenic variants were found in TP53 (c.733G > A, p.Gly245Ser, rs28934575). Pathogenic variants in PALB2 and MRE11A were identified in a 34-year-old patient who was codiagnosed with breast and gastric cancer ( Table 1). Three of the pathogenic variants identified in this study were not reported previously.
First diagnosis of breast cancer at a relatively young age (<35 years) was correlated with pathogenic or likelypathogenic variants in high-penetrance cancer susceptibility genes. Pathogenic variants in high-penetrance genes were detected in 21.1% (4/19) of these patients, which was significantly higher than that for patients who were first diagnosed with breast cancer at age ≥ 35 years (1/101, 1.0%, p = 0.003) ( Table 2).

Discussion
Previous studies using multigene panel tests identified cancer susceptibility genes in 2.1−16.8% of BRCA1/2 mutation-negative patients [5][6][7][8][9][10][11]. Our tests of highpenetrance genes identified a large exon deletion in TP53, and pathogenic and likely pathogenic variants in TP53 and PALB2 (Table 1). We also identified a frameshift mutation of MRE11A c.1773_1774delAA (p.Gly593LysfsTer4) in a patient with a PALB2 mutation. The MRE11 protein functions in non-homologous end-joining and homologous recombination, which occur during the repair of double-stranded DNA breaks [12]. Therefore, the risk for patients with concurrent dysfunction in PALB2 and MRE11A is unclear and should be assessed in future studies. Because the two frameshift variants in PALB2 (c.3267_3268delGT, p.Phe1090SerfsTer6, rs587781890; and c.695delG, p.Gly232ValfsTer6) were not found in the control group, the variants met the criteria to be likely pathogenic according to the ACMG guideline (PVS1 and PM2) (  Table S2) [4].
Pathogenic or likely pathogenic variants also were detected in BRCA1-associated RING domain 1 (BARD1) and BRCA1-interacting protein C-terminal helicase 1 (BRIP1). BARD1 and BRIP1 encode proteins that interact with the BRCA1 protein during the repair of DNA doublestranded break, and pathogenic variants of these genes have been investigated [13]. However, there is a controversy as to whether these rare variants are clinically associated with a risk of breast cancer [11,14]. In a previous study that screened for BRIP1 mutations among 235 Korean patients with BRCA1/2 mutation-negative high-risk breast cancers using fluorescent-conformation sensitive gel electrophoresis (F-CSGE), there was no case of a proteintruncating BRIP1 mutation, which suggests that the prevalence of BRIP1 mutations is likely to be low in the Korean population [15].
Cell cycle checkpoint kinase 2 (CHEK2) is a wellestablished moderate-penetrance breast cancer gene. Several studies have shown that essentially no case of CHEK2 (c.1100delC) was observed in Asian populations, in contrast to the observed prevalence in European populations [16][17][18][19]. Liu and colleagues reported that the CHEK2 c.1111C > T (p.His371Tyr, rs531398630) variant was observed in 4.24% (5/118) of Chinese familial breast cancer cases without BRCA1/2 mutations, and was associated with dysfunctional phosphorylation of T68 in the SQ/TQ rich domain, which is an activation point following DNA damage [18]. We also identified CHEK2 c.1111C > T variants in 2.5% (3/120) of Korean breast cancer patients without BRCA1/2 mutations (Additional file 2: Table S2). Population-based investigations are required to establish the prevalence of this variant, especially in Asian patients. We identified the CHEK2 c.908 + 2delT variant in one a b     Likely pathogenic (Table S2) [23] patient, and it was classified as likely pathogenic according to the ACMG guideline (Additional file 2: Table S2). However, we did not classify this variant as a positive result because the experimental study was not sufficient.
In the current study, clinically important likely pathogenic or pathogenic variants of high-penetrance genes were identified in only five (4.2%) patients (TP53 in two patients, and PALB2 in three patients). These variants were identified in 4 of 19 patients (21.1%) with earlyonset breast cancer (< 35 years old at onset) ( Table 2). A previous study identified cancer susceptibility mutations in 11% of BRCA1/2-negative patients with early-onset breast cancer (diagnosed at <40 years of age) [20]. Considering the frequency of pathogenic variants of high-penetrance genes in patients with early-onset cancer, clinicians should be encouraged to consider performing multigene panel tests for these patients if their conventional BRCA1/2 tests are negative.
This study has several limitations. The primary limitation is the small number of patients (n = 120), which provides only limited data for cancer susceptibility genes in Korean patients with breast cancer. A large-scale cohort study will be required to establish the accurate prevalence and spectrum of pathogenic variants in these patients. The majority of patients (87 of the 120, 72.5%) had VUS. A functional and population-based study will be necessary to clarify the clinical meaning of these VUS. Despite these limitations, to the best of our knowledge, this is the first prospective study to apply customized multigene panels to BRCA1/2 mutationnegative Korean patients with a high risk for HBOC. A recent study conducted by Couch et al. assessed the commercial multigene panel test results of 65,057 patients with breast cancer; however, the frequency, phenotypic association, and cancer risks related to each variant were analyzed among Caucasian women only [11]. Regarding diversity of prevalence of the genetic variants, more prospective studies will be required among diverse ethnic populations.

Conclusions
Wider application of multigene panel tests that include high-penetrance cancer susceptibility genes, so-called "beyond BRCA1/2 genes", will likely provide clinically relevant information for some patients with high risk for hereditary cancer [1,13,21]. However, these panels can produce abundant and conflicting results in clinical practice. To efficiently utilize these data, clinical databases should be established with respect to ethnic backgrounds, and genetic results should be carefully applied for high-risk patients.

Additional files
Additional file 1: Figures S1 and S2. This file includes the methods detecting pathogenic variants and lage deletion in this study; depth of coverage and method for detection of large insertion-deletion of exon using next-generation sequencing, and confirmation of