Investigation of germline variants in Bahraini women with breast cancer using next-generation sequencing based-multigene panel

Germline variants in BRCA1 and BRCA2 (BRCA1/2) genes are the most common cause of hereditary breast cancer. However, a significant number of cases are not linked to these two genes and additional high-, moderate- and low-penetrance genes have been identified in breast cancer. The advent of next-generation sequencing (NGS) allowed simultaneous sequencing of multiple cancer-susceptibility genes and prompted research in this field. So far, cancer-predisposition genes other than BRCA1/2 have not been studied in the population of Bahrain. We performed a targeted NGS using a multi-panel covering 180 genes associated with cancer predisposition to investigate the spectrum and frequency of germline variants in 54 women with a positive personal and/or family history of breast cancer. Sequencing analysis revealed germline variants in 29 (53.7%) patients. Five pathogenic/likely pathogenic variants in four DNA repair pathway-related genes were identified in five unrelated patients (9.3%). Two BRCA1 variants, namely the missense variant c.287A>G (p.Asp96Gly) and the truncating variant c.1066C>T (p.Gln356Ter), were detected in two patients (3.7%). Three variants in non-BRCA1/2 genes were detected in three patients (1.85% each) with a strong family history of breast cancer. These included a monoallelic missense variant c.1187G>A (p.Gly396Asp) in MUTYH gene, and two truncating variants namely c.3343C>T (p.Arg1115Ter) in MLH3 gene and c.1826G>A (p.Trp609Ter) in PMS1 gene. Other variants of uncertain significance (VUS) were also detected, and some of them were found together with the deleterious variants. In this first application of NGS-based multigene testing in Bahraini women with breast cancer, we show that multigene testing can yield additional genomic information on low-penetrance genes, although the clinical significance of these genes has not been fully appreciated yet. Our findings also provide valuable epidemiological information for future studies and highlight the importance of genetic testing, and an NGS-based multigene analysis may be applied supplementary to traditional genetic counseling.


Introduction
Breast cancer is the most common malignancy and the leading cause of cancer death among women globally [1].The incidence of breast cancer is significantly high in the Western world with an estimated lifetime risk of one in nine [2].The prevalence of breast cancer is also increased in Arab and Middle-Eastern women, with an earlier diagnosis age compared to the Western countries [3].The Gulf Cooperation Council (GCC) is comprised of Bahrain, Saudi Arabia, United Arab Emirates (UAE), Qatar and Oman.Bahrain (officially the Kingdom of Bahrain) is an island nation located in a bay on the southwestern coast of the Arabian Gulf.It is the smallest among the GCC states and rich in its culture and history.According to 2021 statistics, the total number of population in Bahrain is 1,504,365, of whom 719,333 (47.8%) are Bahraini and 785,032 (52.2%) are non-Bahraini.The majority of non-Bahraini population are form the major part of the working force [4].The Arabs represent a mix of many ethnicities with a distinctive genetic profile.Genome sequencing analysis of 104 unrelated natives of the Arabian Peninsula placed indigenous Arabs as direct descendants of the first Eurasian populations established by the out-of-Africa migrations [5].A study by Bahri et al., [6] showed that Bahrainis ancestors were mainly emigrants from Arabia and Iran according to Alu insertion polymorphism analysis.Another study by Garcia-Bertrand et al., [7] showed that UAE and Bahrain share 23.7% and 22.9%, respectively, of their DNA with Southwest Asian populations.Some of the factors that contribute to the unique genetic makeup in the GCC countries include consanguinity, which is very high among the Arabs [8], and waves of migration of different ancestral populations into the region.Among the GCC states, Bahrain has a high incidence of breast cancer [9,10], with an overall mean age at diagnosis of 50.9 years [9].Although, there is a lack of studies related to genetic epidemiology of breast cancer in Bahrain, the recent foundation of Bahrain genome project [11] will help in the identification of specific genetic variants associated with susceptibility to diseases including breast cancer, and will contribute to the development of diagnostic methods and personalized approaches for disease management.
While the majority of breast cancer cases are sporadic, about 5-10% of cases are hereditary and 15-20% show familial aggregation [12,13].Germline variants in two high penetrance genes, BRCA1 and BRCA2 (BRCA1/2), are the most common cause of hereditary breast cancer [14][15][16][17].The cumulative risk of breast cancer in carriers of variants in BRCA1 and BRCA2 genes was reported to be 72% and 69%, respectively [18], and women with BRCA1/2 variants are advised to undergo prophylactic risk-reducing surgery to decrease cancer-related mortality [19].Remarkably, BRCA1/2 germline variants have been regarded for some target therapies such as Poly (ADP-ribose) polymerase (PARP) inhibitors to improve survival, especially in patients with early or metastatic breast cancer [20][21][22][23].Nevertheless, about 30% of the breast cancer patients who have a family history of inherited breast cancer do not carry BRCA1/2 variants [24][25][26][27][28].Moreover, approximately a four-fold risk of breast cancer was reported in women with a significant family history of breast cancer but who tested negative for BRCA1 or BRCA2 variants [26].Other high-penetrance genes such as CDH1, PALB2, PTEN and TP53 and additional moderate-and low-penetrance genes may also increase breast cancer risk [27,28].Many of these genes are involved in DNA damage response, homologous recombination (HR) repair, or mismatch repair pathways [27,28].The National Comprehensive Cancer Network (NCCN) guidelines for Genetic/Familial High-Risk Assessment: Breast, Ovarian, and Pancreatic (version 2.2022, accessed on March 2022) recommend genetic testing for additional genes beside BRCA1/2 genes [29].Therefore, screening for breast cancer susceptibility genes provides an essential insight into the contribution of clinically relevant variants, which can impact breast cancer management for patients and at-risk family members.
The advent of next-generation sequencing (NGS) has greatly expanded the use of multigene panel testing.The application of NGS-based-multigene testing, which enables sequencing of a large number of genes simultaneously, can provide more information than a single-gene test and has a great implication in breast cancer risk prediction and selection of precise treatment.
More than 200 multigene panels in which BRCA1/2 genes are included have been proposed by academic or commercial laboratories [30].In this context, several recent studies have used multigene panels and provided valuable information of the possible role of several genes in hereditary breast cancer risk.For instance, Yang et al., [31] investigated germline variants in a panel containing 152 genes associated with cancer in a cohort of Chinese breast cancer patients, and found variants in genes other than BRCA1/2 such as TP53 and CDH1, as well as DNA mismatch repair genes and Fanconi anemia genes.Shin et al., [32] analysed germline variants in a panel containing 64 cancer predisposition genes in Korean breast cancer patients with clinical features of hereditary cancer syndrome, and detected BRCA1/2 variants in 63.2% of patients were carriers for and 40.0% were carriers for non-BRCA1/2 genes such as CHEK2, PALB2, MRE11, and RAD50.Using a panel containing 93 cancer predisposition genes in Egyptian patients with familial breast cancer, Nassar et al., [33] identified 27 deleterious germline variants in 11 cancer susceptibility genes including ATM, BRCA1, BRCA2, VHL, MSH6, APC, CHEK2, MSH2, MEN1, PALB2, and MUTYH.These studies showed differences in the spectrum and prevalence of germline variants in breast cancer patients among ethnicities, signifying the need for respective studies in ethnically specific populations.The identification of population-specific variants is crucial to incorporate accurate genetic testing into clinical practice to meet the need for more specific precision therapy.To date, only one study in Bahrain was conducted to investigate the frequency of germline variants in familial breast cancer women, and it was solely focused on BRCA1/2 gene profiling [34].However, cancer-predisposition genes other than BRCA1/2 have not been well studied in the population of Bahrain.The aim of this study was to investigate the spectrum and frequency of germline variants in Bahraini women with a positive personal and/or family history of breast cancer using a targeted NGS-multi-panel covering 180 genes associated with cancer predisposition.

Patients
Women diagnosed with breast cancer in the period between 1 st January 2019 and 31 st December 2020, and reported to the Bahrain National Tumor Board at the Bahrain Oncology Center (BOC)-King Hamad University Hospital (KHUH) in Bahrain were recruited in this study.The period was chosen because initial search revealed that recording and notification of breast cancer cases were at maximum during that period.Patients included in this study met one of the following criteria based on the latest National Comprehensive Cancer Network (NCCN) guidelines for Genetic/Familial High-Risk Assessment: Breast, Ovarian, and Pancreatic (version 2.2022, accessed on March 2022) https://www.nccn.org/guidelines/guidelines-detail? category=2&id=1503 for genetic testing: 1) Personal history of breast cancer at age � 50 years; 2) personal history of breast cancer at any age with treatment indications (e.g.PARP inhibitors for metastatic breast cancer) or triple-negative breast cancer; 3) personal history of breast cancer at any age with � 1 close relative with breast cancer at age � 50 years, or with other cancers (ovarian, pancreatic and prostate cancers); 4) personal history of breast cancer at any age with � 2 close blood relative with either breast cancer or prostate cancer (any grade).Genetic counseling was offered to the patients in the Medical Genetics clinic at KHUH to assess their eligibility for inclusion and to explain the purpose of genetic testing.At this time, the personal and family history data and pedigrees of patients were documented.All patients were not related by means of three-generation pedigree cross-comparison to each other, and only one patient per family was included.Electronic medical records of patients were reviewed for clinical information including age at diagnosis, laterality of breast cancer, histology type and grade.The expression status of estrogen receptor (ER), progesterone receptor (PR), and Her2 was also obtained from the oncology reports of patients.The molecular subtypes of breast cancer were divided into the following categories: Luminal A (ER+, PR+, HER2-), Luminal B (ER+, PR-, HER2+), HER2+ markers and triple-negative (TNBC).All data were collected and accessed for research purposes between 15 th May 2022 and 26 th June 2022.

Ethical consideration
The study was conducted in accordance with the Helsinki declaration.Informed written consents were obtained from all patients.Ethical approval was obtained from the Institutional Review Board at King Hamad University Hospital (KHUH) (Reference # 22-509).

Genomic DNA extraction
Genomic DNA was extracted from EDTA peripheral blood samples using the MagNa Pure LC DNA Isolation Kit (Roche Diagnostics GmbH, Mannheim), according to the manufacturer's recommended protocol.DNA concentration was evaluated by the Nano Drop™ 2000 spectrophotometer (Thermos Fisher, Yokohama, Japan) and DNA quality was assessed using 0.8% agarose gel electrophoresis.

Gene selection
A targeted gene-panel covering 180 cancer susceptibility genes was chosen in this study.The genes were selected based on the NCCN guidelines and published studies in PubMed with a keyword search of "Neoplasms" AND "germline mutation", with the possibility to identify novel candidate genes associated to hereditary breast cancer risk.

Next generation sequencing and data analysis
The targeted region included all coding exons and exon-intron boundaries of each gene included in the panel.Library preparation was performed using the Ion AmpliSeqTM Exome RDY Kit (Thermo Fisher Scientific, Inc.) according to the manufacturer's instruction.This kit enables high-efficiency enrichment with target base coverage of 94.47% at depth � 20X and quality threshold of 97.71%.The libraries were sequenced on the Ion Pro-tonTM following the manufacturer's instructions (Thermo Fisher Scientific, Inc.).After filtering, reads were aligned against the human reference genome hg19/GRCh37 using the Torrent Mapping Alignment Program v.5.0.13 included in the Torrent Suite software for Sequencing Data Analysis V.5.0.4 (Thermo Fisher Scientific, Inc.).Variant calling was performed using a computational pipeline built on the variant caller Platypus v0.81.For final variant calling, filtering was carried out to eliminate erroneous base calling by visually examining variants using Integrative Genomics Viewer software (http://www.broadinstitute.org/igv).Candidate variants were obtained after filtering by gene function, focusing on those genes with a potential role in cancer.Nonsynonymous (missense and nonsense); insertions and deletions (InDels); and splicing variants, and variants occur in the 1000 Genome database with a minor allele frequency of < 0.01 were included.Synonymous variants and variants occur in the 1000 Genome database with a minor allele frequency of > 0.01 were excluded [35].Summary of the data analysis pipeline in the present study is shown in Fig 1 .For interpretation of variants, we followed the American College of Medical Genetics and Genomics (ACMG) 2015 guidelines [36], in which variants were classified into five categories: Pathogenic, likely pathogenic (> 90% certainty of a variant being disease-causing), variant of uncertain significance (VUS), likely benign (> 90% certainty of a variant being benign) and benign.Accordingly, variants were classified as pathogenic with very strong evidence of pathogenicity (PVS1) if they produced premature termination codons associated with non-functional or truncated proteins including nonsense, frameshift and splice-site mutations.Variants that do not fulfil the criteria for pathogenic/likely pathogenic, benign/ likely benign or with conflicting interpretation of pathogenicity were classified as variants of uncertain significance (VUS).

Bioinformatics analysis
Unclassified variants were analyzed using in silico prediction tools that use sequence homology, evolutionary conservation, and protein structural information [37].These tools include: 1) Combined Annotation Dependent Depletion (CADD) integrates multiple information sources including conservation, structure-based features and functional information, and categorizes variants as benign or deleterious using a machine learning approach.CADD predicts a continuous PHRED and a cut-off score above 15 is considered deleterious.2) REVEL is an ensemble method for predicting the pathogenicity of variants based on a combination of scores from 13 tools including MutPred, FATHMM, VEST, PolyPhen, SIFT, PROVEAN, MutationAssessor, MutationTaster, LRT, GERP, SiPhy, phyloP, and phastCons.REVEL cutoff score above 0.5 is considered deleterious.

Characteristics of patients
The characteristics of 54 women with breast cancer included in this study are summarized in Table 1.The mean age at diagnosis of breast cancer was 48.7 years (± 9.5), ranging from 27 to 65 years.Three patients (5.6%) were diagnosed before the age of 30 years, 18 patients (33.3%) were diagnosed at age � 40 years, 16 patients (29.6%) were diagnosed at age � 50 years, 15 patients (27.7%) were diagnosed at age � 60, and 2 patients (3.7%) were diagnosed at age > 60 years.The majority of patients (53, 98.1%) were diagnosed with unilateral breast cancer and only one patient was diagnosed with primary bilateral breast cancer.Most breast cancers were invasive ductal carcinoma (45, 83.3%), and ductal carcinoma in situ was reported in 7 patients (13%).Regarding the molecular subtype, 30 (55.6%) were Luminal A, five (9.3%)Luminal B, five (9.3%)HER2+, five (9.3%)HER2/ER+, and five (9.3%)TNBC.One patient was diagnosed with thyroid cancer alongside breast cancer.Of all breast cancer patients, 44 (81.5%) had at least one first-and/or second-degree blood relative with breast cancer, and two (3.7%) had a family history of ovarian cancer.Furthermore, 40 patients (74.1%) had a family history of other cancers (such as colon, colorectal, endometrial, testicular, gastric, pancreatic, prostate, leukemia).

Gene-panel findings
Sequencing data of 54 breast cancer patients in multi-gene panel testing showed that 25 (46.3%)patients were not carriers for germline variants and 29 (53.7%)patients were carriers for germline variants.The results revealed 173,000 distinctive variants before filtering.After variant filtering a total of 47 germline variants in 33 cancer susceptibility genes were identified, including 43 missense variants, three nonsense variants, and one frameshift deletion.Five patients (9.3%) were pathogenic or likely pathogenic variant carriers.Two patients (3.7%) were identified as harboring BRCA1 gene variants.Three other patients were carriers for non-BRCA1/2 variants including one in MUTYH gene (1.85%), one in MLH3 gene (1.85%) and one in PMS1 gene variant (1.85%).Twenty five patients (46.3%) were found to carry VUS, since some of them also carried pathogenic or likely pathogenic variants.The identified genes with germline variants and their association with various cancer types and syndromes are listed in Table 2. Summary of gene-panel findings is provided in Fig 2.

Pathogenic/likely pathogenic variants
We found five pathogenic/likely pathogenic variants according to the American College of Medical Genetics and Genomics (ACMG) 2015 guidelines [36] in five unrelated patients.They included one missense variant and one nonsense variant in BRCA1 gene, identified in two patients, one missense monoallelic variant in MUTYH gene, one nonsense variant in MLH3 gene and one nonsense variant in PMS1 gene, in three other patients (Table 3).The BRCA1 missense variant c.287A>G (p.Asp96Gly) lies in exon 5 at a mutational hotspot and a critical well-established functional domain, and results in loss of normal protein function and cellular response to DNA damage.The patient harboring this variant (P-7) was diagnosed with unilateral, TNBC at age 35 years.The relevant family history included her maternal

Variants of uncertain significance
We also found several variants of uncertain significance (VUS) in different cancer predisposition-genes among twenty five patients (46.3%) (Table 4).Of them, two missense variants namely (p.Ile1307Lys) in APC gene and c.1441G>T (p.Asp481Tyr) in CHEK2 gene were classified as likely benign according to the ACMG criteria, and as variants with conflicting interpretation of pathogenicity in the ClinVar database.Bioinformatics analysis using CADD and REVE tools predicted both variants to be benign.Two other interesting variants were identified.The missense variant c.5671C>A in FAT1 gene, was reported as likely pathogenic in Clin-Var database and as VUS according to ACMG/AMP criteria.However, this variant was predicted to be benign by CADD and REVEL tools.The missense variant c.2074A>G in  POLH gene was reported as pathogenic in ClinVar database and as VUS according to ACMG/ AMP criteria.This variants was predicted to be deleterious by REVEL (score 0.62).Furthermore, the variant c.661G>A (p.Ala221Thr) in PRSS1 gene was not previously reported in any database and may be considered as novel.This variant was predicted to be deleterious by  REVEL (score 0.86).Several other variants with uncertain significance were predicted to be deleterious by REVEL (bold), (Table 4).In addition, fourteen patients (25.9%) were found to carry variants in two or more cancer susceptibility genes.Particularly, four patients carried pathogenic/likely pathogenic or potential pathogenic variants together with distinct other VUS.Patient P-7 harboring the likely pathogenic variant c.287A>G (p.Asp96Gly) in BRCA1 gene was found to carry a VUS in ATM gene.Patient P-24 harboring the pathogenic variant c.1187G>A (p.Gly396Asp) in MUTYH gene was found to carry two VUS in APC and MSH2 genes.Patient P-52 harboring the potentially pathogenic variant in POLH gene was found to carry a VUS in SAMD9 gene.Patient P-52 harboring the pathogenic variant c.1826G>A (p.Trp609Ter) in PMS1 gene was found to carry a VUS in SOC2 gene.

Discussion
Traditionally, genetic testing for breast cancer has been restricted to high-risk predisposition genes, such as BRCA1 and BRCA2.(BRCA1/2).Although germline variants in BRCA1/2 confer a higher risk of hereditary breast cancer [14][15][16][17][18], many breast cancer patients test negative for variants in these two genes [24][25][26][27][28] and additional non-BRCA genes, particularly those participating in the DNA repair mechanisms, have been identified as predisposing genes for breast cancer [27].Deficiency in DNA repair is the underlying cause of genomic instability that contribute to tumorigenesis and has implications for therapy in a variety of cancers [38][39][40].In the current era of NGS, the usage of multigene panels provides more information on a large number of cancer susceptibility genes, allowing for more accurate risk stratification and tailored cancer care.A number of recent studies have shown a higher rate of variants in non-BRCA1/2 genes, and also found differences in the spectrum and prevalence of germline variants in cancer susceptibility gene in breast cancer patients among ethnicities [31][32][33].Since multigene panel-testing data are still missing in the Bahraini population, the current study investigated germline variants in a cohort of 54 Bahraini women with a positive personal and/ or family history of breast cancer using an NGS-based gene-panel covering 180 cancer susceptibility genes.Abbreviations: VUS, variant of uncertain significance; conflicting, conflicting interpretation of pathogenicity; P, pathogenic; LP, likely pathogenic, LB, likely benign; NR, not reported; ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/);ACMG, American College of Medical Genetics.

*Variants
did not fulfil 2015 ACMG guidelines as pathogenic or likely pathogenic [36] but were classified as VUS or pathogenic/likely pathogenic in the ClinVar database.CADD, Combined Annotation Dependent Depletion; REVEL combines scores from 13 different individual tools for the prediction of pathogenicity; variants predicted to be deleterious by REVEL (bold).https://doi.org/10.1371/journal.pone.0291015.t004 Sequencing data revealed 47 germline variants in 29 patients (Fig 1 and Table 2).Of all patients, five were pathogenic or likely pathogenic variant carriers (Fig 2 and Table 3) including two patients with variants in BRCA1 gene and three patients with variants in non-BRCA1/ 2 genes namely MUTYH, MLH3 and PMS1 genes.
BRCA1 gene variants included a loss-of-function missense variant (c.287A>G; p.Asp96Gly) (Fig 3) [41], and a truncating nonsense variant (c.1066C>T; p.Gln356Ter) (Fig 4), detected in two unrelated patients.BRCA1 is a key DNA repair gene that plays an important role in maintaining genome stability by repairing double-strand DNA breaks through a homologous recombination repair (HRR) pathway [42].BRCA1 also acts as a tumor suppressor gene, which is frequently mutated in familial breast and ovarian cancers.Approximately 11.2% of patients with TNBC carry germline variants in BRCA1 or BRCA2 genes, and they are usually diagnosed at a younger age and have a positive family history of breast cancer [43].TNBC is characterized by the lack of expression of estrogen receptor (ER), progesterone receptors (PR) and human epidermal growth factor receptor 2 (HER2/neu), and is associated with poor prognosis [43].Detection of variants in BRCA1/2 genes is clinically important in breast cancer patients.Cancer cells with harmful variants in BRCA1/2 are unable to repair DNA doublestrand breaks and they are heavily rely on single-strand break repair pathways [44].The PARP enzyme is involved in repairing single-strand DNA breaks, and PARP inhibitors result in the accumulation of unrepaired single-strand breaks, leading to cell death.Therefore, patients with BRCA1/2 variants were predicted to benefit from PARP inhibitors such as Talazoparib and Olaparib [20][21][22].In recent clinical trials, PARP inhibitors were also approved for TNBC patients, which showed progression-free survival benefit when compared to chemotherapy [45].In addition, PARP inhibitors were found to be effective options for treatment of patients with advanced or metastatic TNBC [45,46].
The prevalence of BRCA1/2 variants differs according to ethnic, geographical and other factors.In a systematic review by Armstrong et al,.[47], the prevalence of BRCA1/2 variants was found to vary widely within key clinical and demographic subgroups across countries (Australia, Canada, France, Germany, Israel/Palestine, Italy, Japan, Russia, South Korea, Spain, United Kingdom, and United States).In other ethnic groups, a relatively high rate of BRCA1/2 germline variants was reported in familial breast cancer in Turkey (9.1%) [48], and in Chinese patients with hereditary breast/ovarian cancer (9.4%) [49].Whereas, a greater rate of BRCA1/2 germline variants was reported in familial breast cancer patients from Lebanon (15.5%) [50], and Egypt (19.8% for BRCA1 and 30.6% for BRCA2) [33].Data from the Gulf Cooperation Council (GCC) states have shown higher rates of BRCA1/2 variants in breast cancer patients.For instance, the prevalence of BRCA1 and BRCA2 variants was 89.3% and 14.3% respectively in Saudi Arabia [51], and the overall prevalence of BRCA1/2 variants was 39.6% in Qatar [52].A previous study in Bahrain by Al Hannan et al., [34], which investigated the frequency of BRCA1/2 germline variants in familial breast cancer women, showed that only one patient (1/ 25, 4%) was a carrier of BRCA1 gene variant and one patient was a BRCA2 variant carrier.In our study, the prevalence of BRCA1 variants was 3.7%, which is almost similar to Al Hannan et al., study [34].
Three other germline variants were found in non-BRCA genes: MUTYH, MLH3 and PMS1 genes.In MUTYH gene, a deleterious missense variant (c.1187G>A; p.Gly396Asp) was detected in a breast cancer patients with a strong family history of breast cancer (Fig 5).MUTYH (human MutY homolog) is a DNA repair gene, which encodes DNA glycosylase involved in base excision repair during DNA replication and DNA damage repair [53].Due to the essential role of MUTYH gene in DNA repair, it has been implicated in many other cancers [54].Specifically, biallelic pathogenic germline variants in MUTYH gene have been associated with MUTYH-associated polyposis, an autosomal recessive condition which increases the risk of colorectal cancer [55].It is generally believed that recessive genes are not pathogenic in heterozygotes.The association between monoallelic variants in MUTYH gene and risk of breast cancer is controversial, with some studies showing no link [56,57] and others reporting an association [58,59].In recent studies, MUTYH-monoallelic germline variants have been reported in patients with early-onset or familial breast cancer [60] and in BRCA1/2 negative breast cancer patients [61][62][63].In particular, the identified MUTYH c.1187G>A (p.Gly396Asp) variant in our study was one of the prevalent variants in Dutch patients with adenomatous polyposis, accounting for about 75% of all MUTYH pathogenic variants [64].Interestingly, this specific variant was previously reported in Egyptian patients with familial breast cancer [33], and in Dutch families with breast cancer and colorectal cancer [59].Monoallelic variants in MUTYH gene may act as low-penetrance breast cancer risk and could contribute to breast cancer development in synergy with additional risk factors such as age, ethnicity, or even environmental and lifestyle factors.The current guidelines of the NCCN for Genetic/ Familial High-Risk Assessment: Breast, Ovarian, and Pancreatic (version 2.2022, accessed on March 2022) recommended including MUTYH-monoallelic variants as one of the low-penetrance genes for multigene testing.Nevertheless, further studies are required to confirm the association between MUTYH-monoallelic variants and risk of breast cancer, which can provide additional knowledge for assessing the patient's risk and the potential of developing a new therapeutic target [39].
The other two germline variants in non BRCA1/2 genes in our study were truncating variants, found in two mismatch repair (MMR)-related genes namely c.3343C>T (p.Arg1115Ter) in MLH3 gene and c.1826G>A (p.Trp609Ter) in PMS1 gene (Figs 6 and 7).The MMR system is involved in the maintenance of genomic integrity during DNA replication and after meiotic recombination.Mutated MMR genes cause microsatellite instability (MSI) and replication of error-positive phenotype [65,66].Studies have shown that germline variants in MMR genes increase the risk of colorectal cancer and other cancers [67].
MLH3 gene encodes a protein that functions as a heterodimer with other MMR genes, and PMS1 gene encodes a protein that forms a heterodimer with MLH1 and PMS2, products of MMR genes involved in Lynch syndrome (nonpolyposis colorectal cancer, HNPCC) [68].Both MLH3 and PMS1 genes are suspected to play a role in Lynch syndrome, but the clinical significance of variants in these genes is less clear [69].Recently, germline variants in MLH3 gene have been reported in Finnish families with adenomatous polyposis who also exhibited breast cancer [70], and in Chinese patients with breast/ovarian cancer [32].In addition, unique germline variants in PMS1 gene have been found in patients with breast cancer and in patients with breast/ovarian syndrome [71,72].
Although germline variants in MMR genes are rare in breast cancer and the values of these genes are unknown with respect to breast cancer risk, it has been shown that breast cancer carriers of variants in these genes might have worse survival and some of them might benefit from immunotherapy [73].Thus, the advantage of MMR genes in breast cancer genetic testing cannot be completely ruled out.
It has been suggested that some pathogenic variants linked with cancer which are common in the European population show lower association with cancer in other ethnic groups such as people of Arab descent [74].In a previous study in Saudi Arabian breast cancer patients, 37 potential variants in 25 breast cancer risk associated genes other than BRCA1/2 were identified including variants in MLH1, MLH3 genes [75].Other studies in Saudi Arabian breast cancer patients showed that rare pathogenic variants in MUTYH gene and other MMR genes such as MLH1, MSH2, MSH6 may potentially increase the risk of breast cancer [76].https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10259259/In addition, pathogenic variants in MUTYH gene and MMR genes were also reported in breast cancer patients from the UAE [77].
Clinical intervention and management for breast cancer patients with BRCA1/2 variants are well established and widely applied in clinical practice.However, low-risk genes such as MLH3 and PMS1 are not available in the management guidelines, where variants in low-penetrance genes are very rare and correlate with < 2-fold risk of developing breast cancer [78].In this sense, medical decisions should be based on combined personal and family history of breast cancer patients in pre-and post-genetic counseling.
We also found a number of variants of uncertain significance (VUS) in different cancer predisposing-genes (Table 4).Of them, two missense variants namely c.3920T>A (p.Ile1307Lys) in APC gene and c.1441G>T (p.Asp481Tyr) in CHEK2 gene were classified as likely benign according to the ACMG criteria and as variants with conflicting interpretation of pathogenicity in the ClinVar database.Notably, in our study both variants were predicted to be benign by CADD and REVEL tools.Moreover, the missense variants c.5671C>A (p.Pro1891Thr) in FAT1 gene and c.2074A>G (p.Thr692Ala) in POLH gene were reported as VUS according to the ACMG criteria, but as likely pathogenic and pathogenic respectively in the ClinVar database.In our bioinformatic analysis, the POLH c.2074A>G variant was predicted to be deleterious by REVEL (score 0.62).Particularly this specific variant was previously detected in a Lebanese breast cancer family [50].Therefore, it deserves further investigation in a larger cohort with additional functional studies to assess its impact on breast cancer.We also found a novel missense variant (c.661G>A; p.Ala221Thr) in PRSS1 gene, which was not previously reported in any database.It was predicted to be deleterious by REVEL (score 0.86).However, there are no established functional studies to support the impact of these variants on protein function and further studies are needed to assess their consequences.
Interpretation of VUS can cause a great deal of uncertainty with inconclusive and frustrating results for patients, especially in case of family history of cancers.Currently there is a lack of consensus management guidelines for clinicians on VUS [29].However, reporting these variants may contribute to the knowledge of unclassified variants in breast cancer and gives us the opportunity to conduct further research to clarify their clinical significance.
Fourteen breast cancer patients (25.9%) in our study were found to carry variants in two or more cancer-predisposing genes.In particular, three patients were found to harbor pathogenic or likely pathogenic variants with other VUS.It was not surprising to detect more than one variant through analysis with high-throughput sequencing technology, and the co-presence of these variants can be normal polymorphism in the population with no biological significance.
Strengths of this study include the innovative use of NGS analysis in an underrepresented population, alongside multiple bioinformatics tools that assisted us with the in silico characterization of detected variants.However, our study was limited by the small number of breast cancer patients, and further studies in larger sample sizes are warranted.Additionally, the individuals included in our study represent unrelated patients and there is no prior knowledge about the status of germline variant in their family members.Consequently, further studies are required to assess their possible risk and their potential benefit of preventative management.Mainly, exploration of the impact of MLH3 and PMS1 germline variants in breast cancer would require further segregation analysis and screening tests such as MSI test in larger breast cancer cohorts.

Conclusions
Five pathogenic/likely pathogenic variants were reported here for the first time in Bahraini women with breast cancer.These variants were found in genes that play critical roles in DNA repair, including two variants in BRCA1 gene and three variants in non-BRCA1/2 genes namely MUTYH, MLH3 and PMS1.Several other variants of uncertain significance were detected, and some of them were found together with the pathogenic/likely pathogenic variants.To the best of our knowledge, this is the first application of NGS using a comprehensive cancer-predisposition gene-panel in Bahrain women with breast cancer.Our findings show that multigene testing can yield additional genomic information on low-penetrance genes and provide valuable epidemiological information for future studies.Our findings also highlight the importance of genetic testing, and an NGS-based multigene analysis may be applied supplementary to traditional genetic counseling.

Fig 1 .
Fig 1. Summary of the data analysis pipeline in this study.https://doi.org/10.1371/journal.pone.0291015.g001 aunt affected by breast cancer and thyroid cancer at a young age (28 years), and five other family members affected by breast cancer (Fig 3).The BRCA1 nonsense variant c.1066C>T (p.Gln356Ter) lies in exon 10 and results in truncated non-functional protein in the domain of the BRCA repeats, interfering with the cellular response to DNA damage.This variant was detected in a patient (P-44) who descended from consanguineous parents, and was diagnosed with bilateral breast cancer at age 46 years without a family history of breast cancer, but her maternal cousin was affected by bone cancer (Fig 4).The MUTYH missense variant c.1187G>A (p.Gly396Asp) lies in exon 13 at a mutational hotspot and a critical functional domain, and results in loss of normal protein function.The patient harboring this variant (P-24) was diagnosed with unilateral breast cancer at age 52 years.The relevant family history included her mother and other family members died due to breast cancer and a family member died due to prostate cancer (Fig 5).The MLH3 nonsense variant c.3343C>T (p.Arg1115Ter) lies in exon 3 and results in truncated non-functional protein.The patient harboring this variant (P-13) was diagnosed with unilateral breast cancer, and showed two paternal aunts affected by breast cancer, and other family members affected by breast cancer and/or leukemia (Fig 6).The PMS1 nonsense variant c.1826G>A (p.Trp609Ter) lies in exon 9 and results in truncated non-functional protein.Patient P-52 who carried this variant was diagnosed with unilateral breast cancer, and showed two sisters affected by breast cancer and a family member died due to leukemia (Fig 7).

Fig 3 .
Fig 3. Pedigree of patient P-7 carrying the missense variant c.287A>G (p.Asp96Gly) in BRCA1 gene.The patient was diagnosed with unilateral breast cancer at age 35 years, with her maternal aunt affected by breast cancer and thyroid cancer at age 28 years old, as well as other family members affected with breast cancer.The symbols used are proband (arrow), male (box), female (circle), affected female (black fill).https://doi.org/10.1371/journal.pone.0291015.g003

Fig 7 .
Fig 7. Pedigree of patient P-52 carrying the nonsense variant c.1826G>A (p.Trp609Ter) in PMS1 gene.The patient was diagnosed with unilateral breast cancer at age 43 years with a strong family history of her two sisters affected by breast cancer, as well as a family member who died due to leukaemia.The symbols used are proband (arrow), male (box), female (circle), affected male and female (black fill).https://doi.org/10.1371/journal.pone.0291015.g007

Table 1 . Characteristics of patients. Characteristics of patients (n = 54) no (%)
*Family history of breast cancer in the first-and/or second-degree blood relatives of the patients.https://doi.org/10.1371/journal.pone.0291015.t001