Background

Breast Cancer (BC) is the most prevalent cancer worldwide and the second leading cause of death by cancer in women [1, 2]. In Tunisia, the incidence of BC is 27.1/100.000 per year affecting more often young women (< 35 years), and with more aggressive clinical behavior [3, 4]. Approximately 5 to 10% of BC patients harbor germline mutations that predispose to this malignancy at earlier onset compared to the general population [5, 6].

Evidently, the BC susceptibility genes BRCA1 (MIM# 113705) and BRCA2 (MIM# 600185) are tumor suppressor genes that play a key role in DNA repair through the homologous recombination pathway [7, 8]. Therefore, loss of the BRCA function results to an inefficient DNA repair process that increases the mutation rates and thus contributes to tumor development [9, 10]. Indeed, women carrying pathogenic germline mutations in BRCA1/BRCA2 genes have an increased risk of BOC compared to those with wild type BRCA genes [11,12,13]. Screening of BRCA1/BRCA2 genes for the identification of pathogenic mutations is essential to provide genetic counseling to the members of affected families, for medical follow-up, and for the targeted therapy selection which is based on the inhibitor of poly-ADP-ribose polymerase inhibitor namely the Olaparib [14]. Despite the importance of the mutational spectrum of BRCA1/BRCA2 genes, only few studies using Sanger DNA sequencing, investigated Tunisian patients with HBOC. In a recent study, Laitamn et al., reported all the pathogenic variants in both BRCA1 and BRCA2 genes in the Middle East, North Africa, and Southern Europe [15]. The authors identified 232 and 239 pathogenic sequence variants in BRCA1 and BRCA2, respectively that include only few variants that were found in Tunisian patients [15].

Besides, Riahi et al., reported twelve pathogenic mutations (25%); among them, three were found in BRCA1 (c.211dupA, c.5266dupC and c.1504_1508delTTAAA) and two were novel mutations detected in BRCA2 (c.1313dupT and c.7654dupT). The study was carried out on 92 families and was performed by direct sequencing [16]. Furthermore, two other studies reported deleterious mutations of the BRCA1/BRCA2 genes in 19% and 18% of familial patients from the north region of Tunisia respectively [17, 18]. Moreover, Fourati et al. by targeting the exons 5, 11, and 20 of the BRCA1 gene and exons 10 and 11 of the BRCA2 gene, showed that among 66 patients, only 12 patients (18%) had deleterious mutations in the BRCA1 or BRCA2 genes [19].

The study of Mahfoudh et al., included 16 Tunisian high-risk BC families and they were screened for only BRCA1 gene. The authors showed that the prevalence of BRCA1 carriers was 37.5% and identified 3 truncating mutations (c.916 delTT, c.3450 delCAAG, c.5382 insC) and one splice site mutation c.212 + 2insG [20].

Technological advances in DNA sequencing have improved and facilitated the screening procedure of BRCA1/BRCA2 genes since the large size of these genes and the absence of hotspots regions make this procedure expensive and time-consuming. In fact, NGS has resolved these problems and allows, henceforth the detection of mutations with higher sensibility [21,22,23]. Our study is the first using NGS to screen the whole exons of BRCA1/BRCA2 genes in 134 selected patients with high risk of BOC in the south region of Tunisia.

Methods

Patients

A total of 134 patients were selected (between 2016 and 2019) among them 113 patients with HBOC (98 cases) and/or OC (15 cases) that meet one of the following criteria: (1) Presence of at least three related first or second-degree BC cases; (2) BC in young patients aged less than 35 years, (3) Presence of male BC among the first or second-degree relatives, (4) Presence of at least two cases of BC or OC, regardless of age, and at least one case of prostate cancer in a related first or second degree patient.

In addition, 21 cases without evidence of family history were included in this study, 12 were diagnosed with BC at the age ≤ 40, and 9 patients diagnosed with OC at the age ≤ 60 years. All patients were recruited from the department of Medical Oncology of the CHU Habib Bourguiba of Sfax (Tunisia), and had provided written informed consent for participation in the research study and for the genetic testing.

Genetic testing

Genomic DNA was isolated from 0.2 mL of peripheral blood of the proband from each selected family using the “QIAamp DNA Blood Mini kit” (Qiagen), following the manufacturer’s instructions. Isolated DNA was quantified by Qubit 3.0 Fluorometric quantitation (Thermo Fisher Scientific). The BRCA/BRCA2 germline mutations were detected using the next-generation sequencing approach (NGS). Briefly, 50 ng of genomic DNA from each sample was used to prepare library using the “AmpliSeq™BRCA Panel”, “AmpliSeq™Library Plus” and “AmpliSeq™CD Indexes” according to the protocol provided by Illumina. The adaptors and sample unique DNA barcodes were incorporated into the amplified amplicons with a second PCR. Libraries were quantified with the Qubit® dsDNA HS Assay Kit (Life Technologies).The DNA library was pooled and prepared for sequencing using the Illumina MiSeq sequencer with Miseq Reagent Kit v2 (300-cycles) (Illumina, San Diego, CA) according to the manufacturer’s instructions to generate paired-end reads with a 151-bp read length. Reads were trimmed to remove low-quality sequences and then aligned to the human reference genome (GRCh37/hg19) using the Burrows-Wheeler alignment (BWA) package. The total PF Reads is 16,755,084 and the Q30 is 93.56%.

The BRCA1 (NM_007300.3) and BRCA2 (NM_000059.3) sequences from the National Center for Biotechnology Information (NCBI) database (http://www.ncbi.nlm.nih.gov) were used as reference. The NGS data was analyzed using the BaseSpace Variant Interpreter (https://basespace.illumina.com).

Sanger sequencing was performed to confirm the BRCA1/BRCA2 pathogenic mutation identified by NGS (see Additional file 1: Fig.1 in Supplementary data). Sequencing primers were designed using Primer 5.0 software and provided on request. PCR products were purified using the QIAquick PCR Purification Kit (Qiagen) and labeled using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems), and the sequence analysis was performed by BioEdit.

Variant annotation and classification

We annotated the mutations using in silico prediction tools to evaluate the functional effects of the candidate variants: PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/), which predicts the effects of an amino acid substitution on the structure and function of a protein. The PolyPhen score represents the probability that a substitution is damaging, so values between 0.9 to 1 are more confidently predicted to be deleterious. The program sorting intolerant from tolerant (SIFT) predicts whether an amino acid substitution is likely to affect protein function based on sequence homology and the physicochemical similarity between the alternate amino acid. The score is the normalized probability that the amino acid change is deleterious if the score is < 0.05 and tolerated if the score is > 0.05. Mutation Taster (http://www.Mutationtaster.org/) was used to assess the impact of mutations on protein function.

Statistical analysis

Associations between patient characteristics and presence of pathogenic BRCA1/BRCA2 mutations were tested by Chi-square test. A significance level of p ≤ 0.05 was used in the analysis. All analyses were performed using SPSS (version 20.0).

Results

Patients characteristics

A total of 134 breast and/or ovarian cancer subjects were included in this study. The characteristics of the study group are presented in Table 1. Unilateral BC was diagnosed in 104 (77.6%) patients, 6 (4.5%) women had bilateral BC, 22 (16.4%) had OC, and 2 (1.5%) had both BC and OC. Age at diagnosis of BC ranged from 27 to 65 years (median: 42.78 years), and of OC ranged from 38 to 64 years (median: 50.37 years). For patients with BC only 9 (6.7%) were diagnosed before the age of 30 years and 55/134 (41%) before 40 years of age. In the group of patients with OC only 4 patients were diagnosed at an age ≤ 40 years. Family history of BC and/or OC was found in 113 (84.3%) patients. Among the 21 patients without family history of BC/OC, 12 were diagnosed with TNBC and 9 with OC (Table 1).

Table 1 Clinicopathological parameters and association with BRCA1/BRCA2 carriers
Full size table

BRCA1/BRCA2 germline pathogenic mutations

Pathogenic BRCA1/BRCA2 germline mutations were identified in 14.17% (19/134) of the patients (Fig. 1, Table 2). Among 113 patients with strong family history for breast/ovarian cancer (HBOC), 18 (15.9%) were positive for heterozygous BRCA mutations (9 in BRCA1 and 9 in BRCA2). In the group of patients without evidence of HBOC, including 12 patients with TNBC and 9 patients with OC, only one patient 4.8% (1/21) with OC carried the c.2338C > T (Gln780X) pathogenic BRCA1 mutation and 2 cases (1 TNBC and 1 OC) carried the VUS c.9976A > T (Lys3326X) in BRCA2 exon 27 (Fig. 1, Tables 2,4). Most of the BRCA mutations were pathogenic (class 2 to 6) and 76.9% of the mutation (10/13) were frameshift deletion, only 1 was frameshift duplication, 2 nonsens mutations and one splice site mutation (Table 2, Fig. 1). Among the 13 mutations detected, 5 were identified at least twice in unrelated patients. Notably, the frameshift mutation c.1310_1313delAAGA of the BRCA2 gene was detected in 4 unrelated BC patients, and the c.17-20 delAAGA in 2 unrelated young patients with strong family history of BC. In BRCA1, the c.5030_5033delCTAA was found in 3 unrelated young patients (≤ 40 years) diagnosed with OC (2 without and 1 with BC), and the c.2338C > T mutation was detected in 2 unrelated patients with OC only. Interestingly, our data indicated that the c.1310_1313delAAGA of the BRCA2 gene, and the c.5030_5033delCTAA of the BRCA1 gene were found in 4% (4/98) and in 20% (3/15) of familial BC and of early onset familial OC respectively. We also detected the c.632-1G > A mutation in the splice acceptor site in 2 patients with BC (Table 2, Fig. 2). It is interesting to note that the 2 unrelated patients with the c.632-1G > A variant, carried also the pathogenic mutation c.1310_1313delAAGA in BRCA2. Our study outlined 6 out of 13 (46%) novel pathogenic BRCA mutations, according to the BIC and ClinVar databases.

Fig. 1
figure 1

Schematic diagram of BRCA1 and BRCA2 proteins with the position of all identified pathogenic variants. Novel variants are marked with arrow

Full size image
Table 2 Pathogenic BRCA1/BRCA2 mutations identified in breast/ovarian cancer patients
Full size table
Fig. 2
figure 2

BRCA mutation rate (%) in all BC (a), in all OC (b), and in TNBC (c) patients divided according to age at diagnosis (≤ 30, 31–40, 41–50, and ≥ 51 years old)

Full size image

Association of BRCA carriers and clinico-pathological parameters

In BC, there is a significant association between Triple Negative status and BRCA mutations. Among the 26 TNBC cases, 5 (19.23%) carried BRCA mutations and 4 (80%) were BRCA1 mutation carriers (p = 0.008, Table 1). Additionally, patients with OC (with or without BC) carried only mutations in BRCA1 gene as opposed to BRCA2 gene (5/24 vs 0/24), whereas BRCA2 mutations were more frequently detected in patients with only BC (9/110 vs 5/110, p = 0.029, Table 1). No other associations were observed with clinico-pathological parameters (Table 1).

On the other hand, for patients diagnosed with BC at an age ≤ 30 years, the BRCA mutation rate was 22.2% (2/9), and none of the patients was diagnosed with OC at an age ≤ 30 years in our series. In the group of patients between 31 and 40 years, BRCA mutation was detected in 13.72% (7/51) and in 75% (3/4) of women diagnosed with BC and OC respectively (p = 0.001, Fig. 2a, b; Table 3). Concerning women aged from 41 to ≤ 50 years old diagnosed with BC or OC, 8.33% (2/24) and 22.2% (2/9) carried a BRCA mutation respectively (p = 0.044, Fig. 2a, b; Table 3). None of OC patients over 51 years carried BRCA mutation while 11.5% (3/26) of BC patients were BRCA2 carriers (Fig. 2a, b; Table 3). The mutation rate for the BRCA gene in patients with only OC was 18.2% (4/22), all were BRCA1 carriers and only one OC patient carried the variant K3326X in the BRCA2 gene (Tables 1, 3). Furthermore, in young patients (≤ 30 years), 40% (2/5) of TNBC carried BRCA mutation while in the group of TNBC patients aged from 31 to 40 years old, BRCA mutations occurred more frequently in BRCA1 than in BRCA2 gene (3/13, (23%) vs 0/13 (0%), (p = 0.041, Fig. 2c, Table 3).

Table 3 Association of BRCA1/BRCA2 carriers with tumor site and triple negative status in patients divided according to the age at diagnosis
Full size table

Variants of uncertain significance in BRCA1/BRCA2 genes

In addition to deleterious BRCA1/BRCA2 mutations, 10 VUS (6 in BRCA1 and 4 in BRCA2) were predicted to be pathogenic by PolyPhen, SIFT or Mutation Taster (Table 4). Each VUS was identified in more than one patient and the c.9976A > T in exon 27 of the BRCA2 gene was detected in 3 unrelated patients, 2 with BC diagnosed at an age ≤ 40 years and 1 patient with OC (Table 4). All VUS have been reported in the BIC database except the variant N550H in BRCA1 exon 11.

Table 4 List of the uncertain significant variants (VUS)
Full size table

Discussion

BRCA1 and BRCA2 have key roles in the development of breast/ovarian cancer [9, 10]. The prevalence of BRCA1/BRCA2 mutations varies in different populations due to founder mutation effect [15, 24,25,26,27,28]. Genetic testing of patients with family history for breast/ovarian cancer have become standard clinical management in Western countries, however, in Tunisia studies of BRCA-associated breast/ovarian cancer remain less investigated. Clearly, the majority of studies that reported the spectrum of BRCA mutations in Tunisia included a small number of patients and used Sanger method for DNA sequencing [16,17,18,19,20]. Riahi et al. reported that the rate of BRCA mutations in patients from the north of Tunisia was 25% (12/48) [16]. Furthermore, in another study, it was showed that deleterious BRCA1 mutations were detected in 37% (6/16) of patients selected from high-risk breast cancer families [20]. In this study, we used NGS to determine the BRCA mutation rates in high risk breast and/or ovarian cancer in patients from the south of Tunisia. Thirteen truncating mutations and one splice acceptor mutation were identified in 19 among 134 patients (14.17%). The rate of BRCA1 mutations found in this study (7.46%, 10/134) was lower than those reported in previous studies. In fact, in a review of the literature, Cherbal et al., reported that in breast/ovarian cancer families from Algeria, Morocco and Tunisia, BRCA1 was mutated in 17.43% of cases (34/195) and that the c.798_799delTT was a recurrent mutation [29]. In our study none of the previously reported mutations in North African patients was identified except for the c.1310_1313delAAGA in BRCA2. Indeed, this mutation was detected in 4% (4/98) of patients from unrelated families with high risk of BC and among 9 BRCA2 carriers, 4 (44.4%) had the c.1310_1313delAAGA mutation suggesting that it might be recurrent in our population. Our finding is in line with a previous study showing that the c.1310_1313delAAGA mutation was detected in 6% (4/66) of patients from the north of Tunisia [19]. Furthermore, it was reported that the c.1310_1313delAAGA mutation was detected in 11.4% (14/122) of patients from the North-East region of Morocco, and in one Algerian family among 10 carrying BRCA mutation [30, 31]. According to the Breast Cancer Information Core database (BIC; http://research.nhgri.nih.gov/bic/), this mutation was found in different European patients and was recorded several times in the French UMD-BRCA2 database and classified as founder mutation [32]. Furthermore, we found that 2 BC patients are double heterozygous carrying both the c.1310_1313delAAGA frameshift mutation and the splice site acceptor variant c.-632G < A. This splicing mutation has been reported two times in the ClinVar database therefore, and in the best of our knowledge this is the first report describing the association of the splice site and a frameshift mutation in the BRCA2 gene in 2 unrelated patients with strong family history of BC.

The pathogenic mutation c.17-20 delAAGA (Lys6Xfs) in BRCA2 exon 2 was detected in 2 unrelated young patients (39 year-old) with strong family history of BC. This mutation has not been reported previously in the ClinVar and BIC databases.

We also detected in the BRCA2 exon 27, the c.9976A > T (K3326X) in 3 unrelated patients, 2 were diagnosed with family BC at 28 and 36 years old and one patient with OC at 56 years old and without family history. The BRCA2 variant (K3326X), was firstly interpreted as pathogenic, but its identification in control populations led to its classification as a benign polymorphism [33]. However, recent studies showed the association of K3326X variant with the risk of developing melanoma, pancreatic, breast and ovarian cancers [34,35,36]. Moreover, the large study of Meeks et al., provided evidence that the K3326X variant is associated with the risk of developing BC and OC independently of other pathogenic variants in BRCA2 [37]. Altogether, it was suggested that the K3326X variant is not neutral and that it may be included in SNP panels for evaluating BC risk.

In BRCA1gene, we identified the c.5030_5033delCTAA mutation in 3 out of 15 (20%) unrelated patients with early onset  HBOC. This mutation has not been reported in North Africa but was found in patients from Greece, Italy, Jordan, Lebanon, Kuwait, and Saudia Arabia [15]. In addition, the nonsense mutation c.2338C > T was shared by 2 young patients (< 50 years-old) diagnosed with OC. The c. 2338C > T has been reported in Caucasian patients according to the BIC and ClinVar databases. Altogether, these findings may suggest ethnic and genetic associations between unrelated populations, or that these mutations occur in mutational hotspots region.

On the other hand, it is interesting to note that the BRCA mutation rate in young patients diagnosed with BC (age ≤ 30 years) was higher (22.2%) compared to older patients (13.7% (31–40), 8.3% (41–50), and 11.5% (≥ 51 years). The recent study of Bakkach et al.,  reported that the frequency of mutations among young patients (≤ 40 years) with family history of BC was 16.7% (5 out of 33 patients) suggesting that the young age for BC diagnosis seems to be strongly predictive of BRCA mutation status in Moroccan patients [38].

Of patients with OC (with and without BC), 20.83% (5/24) carried pathogenic variants. Mutation detection rates were 75% (3/4) for patients diagnosed between 31 and 40 years compared to only 22.2% (2/9) for women between 41 and 50 years. All patients over 51 years, carried no mutations in BRCA1/BRCA2 genes.

It is well established that TNBC exhibit an aggressive behavior and has a worse prognosis [39]. In our cohort, among 110 patients with only BC, 26 (23.6%) were TN including 14 patients with family history of BC and 12 without evidence of family history of BOC. Five patients (5/26, 19.23%) carried BRCA mutations and 4 among them (80%) were BRCA1 carriers. Consistent with our findings, in a previous study, the BRCA1 mutation carriers had a higher incidence in TNBC than BRCA2 mutation carriers [40,41,42,43]. Moreover, in a recent work, Toss et al., showed that in TNBC Italian patients, BRCA mutation prevalence was 22.6% (21.4% BRCA1) and that 64.2% are ≤ 30 years old, which is in line with our findings [44]. Furthermore, Mahfoudh et al., reported that the 5382insC mutation in BRCA1 was detected in 25% (2/8) of Tunisian patients with TNBC [45]. In our study, none of the patients carried the BRCA1 5382insC mutation, however, in TNBC we detected 3 mutations in BRCA1 (c.3254delG, c.3364_3370dupACAGATT, c.4067_4071delAAGAA), and no BRCA2 pathogenic mutations were detected in TNBC cases except the VUS c.9976A > T identified in 2 unrelated patients (Additional file 2: Table 1 in Supplementary data).

It is important to note that previous studies used Sanger sequencing to identify BRCA mutations in Tunisian families with a history of BOC unlike the present study where the whole BRCA genes were screened by the NSG which allowed us to identify new mutations in Tunisian patients. However, some recent studies have performed the whole exome sequencing to identify candidate genes in few BRCA negative families. Hamdi et al., performed whole exome sequencing on seven Tunisian families with HBOC and identified four novel BC candidate genes (MMS19, DNAH3, POLK and KATB6) [46]. Furthermore, other studies identified by exome sequencing, RCC1 and RAD50 as BC candidate susceptibility genes in Tunisian BRCA negative families [47, 48].

Conclusions

This study assessed the prevalence of germline mutations and identified novel and recurrent mutations for BRCA1/BRCA2 genes in patients with high risk of BOC from the south region of Tunisia. Giving the high prevalence of pathogenic mutations in BRCA1/BRCA2 genes (14.17%), our data will contribute to the establishment of a service dedicated to the genetic screening and to the counseling of the families with high risk of HBOC in Tunisia.