Background

Breast cancer (BC) is the second cause of death by cancer in women [1]. Screening for inherited variants in patients with BC has significantly increased over the past 30 years. About 20% of all BC cases described worldwide have a genetic origin with a large heterogeneity in the percentage of pathogenic variants (PVs) [2,3,4].

Somatic variants represent the most common cause of cancer while germline variants accounts for approximately 5% of BC. Although two-thirds of these variants were found in BRCA1/2, other genes such as ATM, PALB2, CHEK2, PTEN and TP53 have been also reported in hereditary BC, ovarian cancer (OC), and pancreatic cancer (PC) [5, 6]. Using NGS-based multi-gene panel testing, many cases with strong personal and/or family history of cancer were indeed found to be BRCA1/2-wild-type. For instance, a significant proportion of 15.1% of BC, OC, or PC germline pathogenic variants was observed in susceptibility genes other than BRCA1/A2 among bilateral BC patients and therefore would have been over looked [7, 8].

The major risk of developing BC due to these variants and the large benefit of their early detection have made genetic screening for hereditary BC recommended to patients with family history in many countries even though the implication of these PV variants in defining the cancer clinical and phenotypic features remained unanswered [9]. Due to its comprehensive genomic coverage and higher sensitivity, next generation sequencing (NGS) has become an essential tool in tumor profiling [10]. However, this technology remained poorly used in African countries mainly due to its elevated running cost [11,12,13].

In this study, using NGS methodology, we reported the first BRCA1/2 mutational profile of a BC patients’ cohort in Mauritania and assessed the relevance of detected variants to the carriers’ demographic and clinical characteristics.

Patients and methods

Patients

In this study, 137 Mauritanian BC patients (132 women and 5 men) whose demographic and clinical data were complete and available to participate to the genetic screening were selected if they met one of the following criteria:

  • Age below 45 years when diagnosed with BC

  • Bilateral BC

  • BC detected in two or more relatives (first or second degree) in the family

  • BC in men

  • Triple negative breast cancer (TNBC) diagnosed before the age of 60 years.

Demographic and clinical characteristics of recruited subjects included age at diagnosis, family history of the disease, cancer staging and histological grading. Immunohistochemical staining (IHC), when accessible, was carried out on patient tissue samples embedded in paraffin blocks. Grading (from T0 to T4) was performed according to the American joint committee on cancer/ /Union for international cancer control (AJCC /UICC) systems. Evaluation of cancer stage (from 0 to 4) used TNM staging. TNBC patients were referred to subjects with IHC slides showing no antibody staining or a tumor cells fluorescence less than 1% for, concomitantly, receptors of estrogen (ER), progesterone (PR) and hormone epidermal growth factor receptor 2 (HER-2).

BRCA1 and BRACA2 molecular screening

Extraction of DNA from peripheral blood collected in EDTA tubes was carried out using QIAampDNA Blood Midi Kit (Qiagen, Hilden, Germany). DNA concentration and quality were assessed by NanoDrop (Jenway) spectrophotometer and agarose electrophoresis, respectively. Screening of the BRCA1/2 variants was performed using the ONCO/Reveal™ BRCA1/BRCA2 panel (Pillar Biosciences, Natick, MA). This technique consisted of a robust NGS assay to sequence the entire coding regions of the BRCA1 and BRCA2 genes plus 20 bp of flanking introns. It is based on the use of proprietary Stem-Loop Inhibition-Mediated amplification (SLIMamp®), a tiled amplicon-based library prep chemistry for efficient single-tube target enrichment. Sequencing was then performed on MiSeq platform as recommended by the manufacturer (Illumina, Inc., San Diego, CA) at Colorado sequencing center (https://thesequencingcenter.com/).

Bioinformatics

Alignment of the sequence reads, in FastQ format, was referred to the reference human genome 19 (hg19). Variant annotations were carried out using databases including HapMap project (http://hapmap.ncbi.nlm.nih.gov/), 1000 Genome Project (http://www.1000genomes.org/), Exome Variant Server (EVS, http://evs.gs.washington.edu/EVS/), and Exome Aggregation Consortium (EXAC, http://exac.broadinstitute.org/).

DNA variants were considered as pathogenic if cited by ClinVar record or reported by in silico analysis with PolyPhen2, (http://genetics.bwh.harvard.edu/pph2/), Sorting Intolerant From Tolerant (SIFT, http://sift.jcvi.org/) or Mutation Taster (http://www.mutationtaster.org/).

Statistics

Demographic and clinic-pathologic characteristics of patients were statistically assessed by SPSS data analysis package version 23.0 software (Chicago, Ill).

Results

Demographic and pathological features of the cohort

The demographic and clinico-pathological characteristics of 137 BC patients (132 women and 5 men) were shown in Table 1. Women were aged between 26 and 76 years with an mean age of 45 years at time of diagnosis of BC. Family history (patients with first- or second-degree relatives with BC) was reported in 41(30%) BC patients and consanguinity observed in 63 (46%) patients.

Table 1 Demographic and clinico-pathological characteristic of the study population
Full size table

Cells abnormality was diagnosed at poorly differentiated status (grade III) in 45 (32.8%) patients. BC was at stage III in 61 (44.5%) and metastasized (stage IV) in 32 (23.4%) patients.

Cancer stretching into breast surrounding tissues showed presence of invasive ductal carcinomas (IDC) and invasive lobular carcinomas (ILC) in 114 (83.2%) and 12 (8.8%) patients, respectively.

Among 96 patients with satisfactory immunohistochemical data, 45 (46.9%) were classified triple negative breast cancer (TNBC) as no antibody staining above cut off was detected simultaneously for ER, PR and HER-2 receptors.

The five men with BC recruited had an age between 42 and 95 years. They all had unilateral (left or right) advanced BC with IDC in 4 of them. Only one male was TNBC.

BRCA1 and BRCA2 sequence variants in the study population

BRCA1/2 NGS-based screening of the cohort identified sixteen pathogenic (PV) or likely pathogenic variants (PLPVs) (8 in BRCA1 and 8 in BRCA2) observed in 38 patients (28.8% of total BC patients) (Table 2). PV/LPV variants were carried by 11 different families among 41 enrolled patients with family history (1st or 2nd degree affected) of BC recruited. Pedigrees of the most prevalent pathogenic variants were presented in Fig. 1. Among the PV/LPV variants, we had seven missense, four insertions, four deletions and one splice site variants. The two predominant variants were frameshift (c.815_824dup) in exon 10 and splice site variant (c.4986 + 6 T > C) in intron 15, found in 13 and 7 BC patients, respectively. Both changes were located in BRCA1 gene. Frameshift variants (c.7234_7235insG and c.6280_6286del) of BRCA2 gene were identified in 4 and 2 patients, respectively.

Fig. 1
figure 1

Examples of pedigrees with patients carrying variants in BRCA1 and BRCA2 genes in the Mauritanian population

Full size image
Table 2 List of pathogenic and likely pathogenic BRCA1 and BRCA2 genes variants found in Mauritanian hereditary BC patients
Full size table

Interestingly, we identified four variants never reported before (Tables 2 and 3): One (c.813_814insTAGCCATGTG) in BRCA1 and three (c.256del, c.2892_2893insC and c.10247A > G) in BRCA2. Three of these novel variants were predicted pathogenic (Table 2).

Table 3 List of non-pathogenic BRCA1 and BRCA2 genes variants found in Mauritanian hereditary BC patients
Full size table

Twelve of the 42 TNBC patients carried pathogenic variants: 7 in BRCA1 and 5 in BRCA2. Nineteen non pathogenic variants (8 in BRCA1 and 11 in BRCA2) were detected in 96 BC patients (73.8%) (Table 3).

Carriers of all variants originated from different regions of the country (Fig. 2).

Fig. 2
figure 2

Geographic distribution of reported BRCA1/2 variants in Mauritania

Full size image

Discussion

Although poor socioeconomic conditions and reduced access to adequate healthcare may have been determinant in BC development and outcome among our cohort, genetic origin was also likely giving the high rate of consanguinity (46%), the early average age of women (45 years) and advanced phase of BC at diagnosis observed in our cohort. Indeed, numerous data showed a higher incidence of common adult diseases such as hypertension, diabetes mellitus or cancer in patients from consanguineous marriages [14]. Reports also indicated that close-kin marriage continued to be preferential in North and sub-Saharan Africa with a prevalence of marriages between couples related as second cousins or closer exceeding 40% in countries including Mauritania [15, 16].

Gene predisposition has been suggested in the younger median age at BC diagnosis among Eastern Mediterranean women compared to women in Western European countries [17, 18]. An equivalent early age at BC detection (47 years) was also observed in Sub-Saharan and British black women while most Caucasian women had a BC onset at 67 years [19,20,21,22].

In this first study providing molecular data on BC in our country, among the 137 BC patients recruited at the only state cancer referring center, 38 (27.7%) carried a BRCA1/2 pathogenic or likely pathogenic variant. In neighboring North African region, the frequency of BRCA1/2 pathogenic variants in Moroccan population varied from 16.7 to 31.6% [23]. In Tunisia, 25% of hereditary BC patients carried a BRCA1/2 gene pathogenic variant [24]. The BRCA1/2 variant ratio we found (1:1.18) was also consistent with the fraction reported by these different studies which confirmed the significant contribution of BRCA1/2 germline variant in BC risk among North African populations [25, 26]. Although we did not perform functional validation of the BRCA1/2 deleterious variants identified in our cohort, most common PV/LPVs we found were already reported as associated with BC. Indeed (c.815_824dup p.Thr276Alafs*14) (found here in 34.2% of patients), was diagnosed in 15 of 27 (55.5%) patients of hereditary BC cases in neighboring Senegal with a founder effect in Afro-American patients [27]. This BRCA1 duplication leading to a premature stop codon was revealed as pathogenic variant by ClinVar here and in other studies [27, 28]. It was also submitted by HGSC-CL as pathogenic germline variant using clinical testing [29, 30]. The relatively high and concordant frequency in our cohort and in Senegal further highlights the pathogenic character of this variant in BC assessment and prevention in West African families [27, 31,32,33].

The second pathogenic variant (c.4986 + 6 T > C) in BRCA1 carried by seven of our BC patients has also been detected in multiple African patients with an early-onset of BC [34, 35]. This splicing site variant is located in highly conserved human genome region and seems to activate a cryptic splice donor site which alters the reading frame resulting in absent or truncated protein [36, 37]. It was proposed by ClinVar as pathogenic germline allele by numerous submitters [35, 38].

We also found that among the 12 TNBC patients carrying a pathogenic or likely pathogenic variant, 7 had a BRCA1 variant which substantiates data suggesting that patients with BRCA1 variants were more likely to have TNBC than those with BRCA2 variants [39, 40].

Due to the limited number of BC patients here, we could not conclude on the association of the pathogenic variant frequency and the BC phenotypic features (molecular tumor subtypes) as investigated elsewhere [40]. For instance, BRCA1-633delC was detected with relatively higher prevalence in patients with TNBC, whereas BRCA2-1466delT was found mainly in Luminal B tumors, but not in TNBC patient [40].

Our study also identified three novel predicted pathogenic BRCA1/2 variants never reported before. These variants could be specific to our population. Territorial prevalence has for instance reported among families with hereditary breast and ovarian cancer in families from southern Italy with a higher prevalence of PVs in Sicilian population [41].

One limitation of this single gene testing was that only BRCA1/2 was explored in this cohort while a NGS-based multi-gene panel testing could have revealed more potentially BC pathogenic variants. It was thus showed that 19 out of 53 positively tested bilateral BC patients harbored a germline PVs in a known (no-BRCA) BC susceptibility gene which clearly support the inclusion of multi-gene panel with high or intermediate penetrant BC genetic predisposition [7, 8]. Although highly pertinent, we could not currently perform such a study giving the high cost of this profiling.

Conclusions

Our study gave the first data on BRCA1/2 alterations likely underlying hereditary BC in Mauritania using a powerful NGS based screening.

A multi-gene panel testing of all BC patients followed by Sanger sequencing confirmation should avoid underestimation of affected patients and pave way to more cancer associated PV/LPV variants identification in hereditary tumor surveillance and targeted therapy choice.

Most of PV/LPVs variants identified in our cohort were previously reported in other African populations geographically remote and culturally apparently not related. This BC heterogeneous heredity confirmed the complex genetic structure of African populations shaped by successive voluntary or forced migrations, integrations and assimilations over many centuries.