Screening of germline mutations in young Rwandan patients with breast cancers

Abstract Background In Sub‐Saharan Africa breast cancer is commonly detected at younger age and the profile is more aggressive with a high mortality rate compared to the European countries. It is suggested that African‐specific genetic background plays a key role in this matter. The present study aimed at understanding the role of genetic factors in breast cancer development in young Rwandan. Methods We performed a massive parallel sequencing on Illumina MiSeq NGS system for the screening of 26 genes associated with hereditary breast cancer from 40 patients under 35 years old from two University Teaching Hospitals in Kigali, Rwanda. Sanger sequencing was used to confirm pathogenic and likely pathogenic mutations. Results Five patients out of 40 (12.5%) presented with pathogenic mutations including four patients (10%) carrying BRCA1 or BRCA2 pathogenic variants. One patient showed a missense likely pathogenic TP53 variant. We have also detected additional missense, intronic, and 3’UTR variants of unknown significance in all study participants. Conclusion This preliminary study suggests that the frequency of germline mutations in young Rwandan patients with breast cancer is similar to the observations made in Caucasians. However, further large studies including patients and controls are needed to better understand the impact of genetic factors as well as the environmental risk factors in the development of breast cancer in young Rwandans.


| INTRODUCTION
Breast cancer (BC) is a common cause of mortality among women worldwide. In individuals under 40 years old, it is however considered to be rare as it affects less than 7% of patients (Brinton, Sherman, Carreon, & Anderson, 2008).
Although its incidence in African ancestry individuals is still lower compared to other ethnic groups, the mortality is rather higher. Indeed, breast cancer in sub-Saharan Africa is characterized by a younger age at diagnosis (Adebamowo et al., 2003;Fregene et al., 2005;Adesunkanmi, Lawal, Adelusola, & Durosimi, 2006). Compared to other age groups, BC in young people (YBC) has a worse prognosis due its advanced stage at diagnosis and a high proportion of hormone negativity subtype (Anders et al., 2008;Bharat, Aft, Gao, & Margenthaler, 2009;Colzani et al., 2011).
About 5%-10% of BC are caused by germline mutations. To date, inherited mutations associated with breast cancer risk have been identified in several genes. Those genes have been associated with different levels of risk of breast cancer ranging from high, moderate to low risk. High-risk genes include BRCA1 (OMIM: 113705), BRCA2 (OMIM: 600185), and TP53 (OMIM: 191170) and confer a lifetime relative risk of more than five. BRCA1 and BRCA2 are two major genes associated with a lifetime risk of 50%-80% of breast cancer. Other genes have been associated with a two to fivefold increase risk of breast cancer. These include genes that are involved in DNA breaks repair by homologous recombination such as PALB2 (OMIM: 610355), ATM (OMIM: 607585), and CHEK2 (OMIM: 604373; Wittersheim, Büttner, & Markiefka, 2015).
The cause of breast cancer associated with a high mortality rate in young African are still not well understood and remain understudied. A common hypothesis is that the YBC may be linked to African-specific genetic characteristics (Haffty et al., 2009;Rummel, Lovejoy, Shriver, & Ellsworth, 2017).
Few studies have been conducted in Africa to determine the role of genetic factors in development of BC in general, and in young patients in particular. The majority of those studies restricted their investigations to the screening of mutations in BRCA1 and BRCA2 genes (Abbad et al., 2018).
In Rwanda, genetic risk factors, incidence, and mortality rate of BC are not known. A recent study conducted at Butaro cancer center of excellence, it was reported that the median age at diagnosis of breast cancer was 49 and 32/144(22%) patients were below the age of 40 (Pace et al., 2015).
The determination of genetic variations associated with the occurrence of BC as well as genetic modifiers leading to the disease variability are necessary for accurate detection, prevention, and treatment.
We undertook this study to determine the germline mutations associated with BC disease in young Rwandan patients.

| Study participants
Patients were recruited between April 2016 and March 2018 from the two main public Hospitals in Kigali CHUK and RMH. These two hospitals receive patients from all parts of the country. To be eligible for this study, patients had to be diagnosed with BC before the age of 35. Forty patients consented to participate in the study. Clinical information as well as data on family history of cancer were collected from patient's medical records. Venous blood samples were collected during their routine hospital visits.

| DNA extraction
Whole blood samples were collected in EDTA tubes and stored at -20°C until use. Genomic DNA was extracted from 200 μL of whole blood samples using the QIAamp ® Blood DNA Mini Kit (Qiagen) according to the manufacturer's manual. The quality of the isolated DNA was assessed using NanoDrop Spectrophotometry ensuring the ratio of A 260 nm / A 280 nm ≈ 2.

| Next-generation sequencing and variant call
Libraries for next-generation DNA sequencing were constructed from 75 ng of the isolated DNA for each sample using BRCA Hereditary Cancer MASTR Plus kit (BRCA HC MASTR Plus; MR-0320.024, Agilent) following the manufacturer's instructions. This kit is designed for the identification of single nucleotide variants (SNVs), insertions and deletions (indels), and copy number variations (CNVs) within the 26 following genes: BRCA1, BRCA2, Briefly, targeted DNA regions made up by all coding exons and flanking intronic regions of the 26 genes, were first amplified in multiplex polymerase chain reaction (PCR)--five plexes per patient--using BRCA HC MASTR Plus (Agilent), then barcoded with specific molecular identifiers (MIDs) and ligated with adaptors using MID Dx (MID ML-2208.240) for Illumina Miseq ® sequencer following the manufacturer's instructions. Amplicons were pooled together and purified using AMPure XP Agencourt beads (Beckman Coulter Inc, USA) prior to sequencing. Sequencing was performed on Illumina Miseq NGS system using MiSeq Reagent Kit v3 (paired-end; 600-cycles; 2x 300 +++pb; MS-102-3003-Illumina) according to the manufacturer's instructions. Data were analyzed using MASTR reporter software v 1.0.2 (Agilent) and verified using SeqPilot (Module SeqNext) software v.4.3.1. Both softwares align to hg19 reference genome: At least 98% of all amplicons were covered with a minimum depth of coverage of 40x. For each retained variant, the MASTR reporter software provided annotations including its genomic position, the nucleotide change and the predicted protein change. The impact and consequence on the gene product were predicted by the Ensembl Variant Effect Predictor (VEP) tool v.83.
Sanger sequencing on ABI 3130 using standard dideoxy termination procedure was performed for the validation of plausible pathogenic variants.

| Variants classification and report
As there is no specific database for mutations in African population, the pathogenicity of variants was evaluated using two mutation databases: ClinVar (www.clini var.com) and dbSNPs (www.ncbi.nlm.nih.gov/snp). The pathogenicity effect of the VUS were evaluated using different in silico mutation interpretation softwares such as SIFT (Sorting intolerant form tolerant); http://sift-dna.org; which predicts where an amino acid substitution is deleterious to protein function; PROVEAN (Protein variation effect analyzer) www.//provean.jcvi.org which predicts the functional effect on protein sequence variations; and Mutation Taster www.mutti ontas ter.org which evaluate the disease-causing potential of DNA variants sequences.
Variants were classified into pathogenic, likely pathogenic, VUS, or likely benign/benign following the recom-

| Clinical characteristics of patients
Forty Rwandan breast cancer patients were enrolled in this study. They were selected based on the early age onset independently of family history of cancer. The mean age at diagnosis was 31.2 ± 3.6 years; ranging from 17 to 34 years. Histologically, 33 patients (82.5%) presented with invasive ductal carcinoma (IDC) tumors. Invasive carcinoma of both ductal and lobular and a sarcoma of the breast were present in one and two patients, respectively. Twenty-one patients (52.5%) were at stage three of tumor development. Twentythree patients in our cohort (58.9%) had ER negative tumors and 14 (38.9%) had Her2 positive tumors. Seven patients (17.5%) had a triple negative subtype. Nine patients (22.5%) had a family history with the first or second-degree relatives (FDR or SDR) with ovarian or breast cancer (HBOC). The characteristics of patients, family history, and others risk factors are summarized in Table 1

| Variants with unknown significance (VUS)
In total, 33 VUS were identified. Each patient carried at least one VUS and five patients had more than two VUS in the same or different genes. Twenty-nine VUS among them (87.8%) were unique in this cohort. Eight of those VUS were predicted by different prediction tools (SIFT, Provean, and MutationTaster) to have a damaging effect on protein and 13 variants were very rare and not previously observed in African population while five variants were simply novel (Table 3).

| DISCUSSION
In the present study, we have sequenced genomic DNA from 40 Rwandan patients aged below 35 at the time of breast cancer diagnostic. Young age at diagnosis and African origin are both well known to be associated with an advanced stage of the disease at diagnostic, a high proportion of hormone receptor negative tumors, and a worse prognosis. The reasons of this different severity when compared to Caucasian populations are still poorly understood but are thought to be related to African-specific genetic characteristics (Adesunkanmi et al., 2006), and/or environmental factors (Fregene et al., 2005). We conducted this study to gain insights into relevant genetic variations in young Rwandan with breast cancer. We observed an overall frequency of 5/40 (12.5%) pathogenic germline variants. Among them, 10% variants were detected in BRCA1 and BRCA2. This frequency was comparable with other high frequencies of pathogenic BRCA1/2 variants reported in other studies conducted in young Africans (Awadelkarim et al., 2007;Cherbal et al., 2010), Caucasians (Copson et al., 2018;De Sanjosé et al., 2003;Tonin et al., 2001), or African American women with breast cancers (Haffty et al., 2009;Malone et al., 2006; Table 4). The four pathogenic variants in BRCA1 and BRCA2 were well known in other populations (Heramb et al., 2018), or reported by ENIGMA.
We identified less BRCA1 pathogenic variants compared to BRCA2 (25%vs.75%) in this cohort. This is in consistence with results from other studies on breast cancer in patients of African ancestry where plausible causal variants in BRCA2 T A B L E 2 Five pathogenic variants identified in 40 young Rwandan patients.  gene were predominant compared to BRCA1 (Gao et al., 2000;Panguluri et al., 1999). Contrarily, studies in Caucasians (Krainer et al., 1997) report more BRCA1 mutations in early onset breast cancers than BRCA2. Unfortunately, we cannot make a final conclusion because of a small sample size of our cohort. In our cohort, only one patient (2.5%) harbored a likely pathogenic TP53: c.726C>G variant. This result is similar to previous studies (Bougeard et al., 2015;Hauke et al., 2018), where germline pathogenic TP53 mutations were found in up to 5% of young breast cancer patients.
Surprisingly, in the youngest patient in our cohort, who had familial history of breast cancer and presented with a sarcoma of the breast cancer and a TN tumor, did not have a plausible TP53 or BRCA1/2 variants, as one would expect. She did not harbor a plausible variant in the other genes of the tested panel neither.
A high number of VUS (n = 33) was observed in our study; which is in consistent with other studies in black women of African ancestry where NGS panel were evaluated (Awadelkarim et al., 2007;Fackenthal et al., 2012). These variants may have no functional implication in hereditary breast cancer, but their clinical significance remains to be elucidated. Our analysis indicated that eight among these VUS were predicted by SIFT, Provean, and Mutation Taster to have a damaging effect on protein (Table 3). Those predicted damaging VUS include one variant in CDH1 and BRIP1 genes, respectively, and two variants in each of the following three genes: ATM, CHEK2, and PMS2. The four genes namely CDHI, BRIP1, ATM, and CHEK2 are known to be associated with a high or moderate risk of breast cancer. However, the association of PMS2 gene with breast cancer is still unclear. In Caucasians, contradictory reports were published on the association of PMS2 gene mutations in breast cancer (Bernstein et al., 2019), (Roberts et al., 2018). The impact of this gene in the development of breast cancer in Africa needs further investigation.
Some variants identified in this study were classified as VUS because they are not, at the time of the redaction of this manuscript, found in mutation databases or clinical reports. Thus, the final frequency of germline mutations in our cohort is pending upon further evidences and reports from the literature and databases.
We found 13 variants that had never been observed before in African population (Ensembl: GMAF=0) and five variants that had never been observed before in any population. The lack of African reference and diseases databases are still preventing the full interpretation of NGS data from African individuals. This causes possible underestimation of the role of germline mutations in development of genetic diseases such as breast cancer in African population in general and in young Rwandan in particular. The development of such databases will allow more reliable determination of genetic contribution to breast cancer development in young Africans.
This study is among very few cohort-based studies in Sub-Saharan Africa investigating the contribution of germline T A B L E 4 BRCA1/2 mutations frequencies in young women of Caucasians, African American, and Africans.

Mutation frequencies
Sample size ( Malone et al., 2006) mutations to breast cancer within a large panel of breast cancer susceptibility genes. The majority of previous studies conducted in Africa were limited to the assessment of mutations in BRCA1 and BRCA2 only. However, we did not detect any clear relevant variant in 23 genes out of 26, indicating that BRCA1 and BRCA2 are probably the most commonly mutated genes associated with breast cancer predisposition in African women.
Our study had limitations related to the small size of our cohort, as well as the lack of reference mutation database for African population. These may lead to a false estimation of the frequency of genetic mutations. Additionally, we have only sequenced the coding sequences and their flanking intronic regions, and interrogated SNVs and small Indels. We may have missed the deep intronic variants or CNVs that would be associated with a risk of breast cancer.

| CONCLUSION
Our preliminary results showed that in young Rwandan patients with breast cancers, BRCA genes were the most mutated with a predominance of BRCA2 variants. The frequency of overall mutations was similar to the results observed in Caucasians. Further large studies including both large families and controls and interrogating more types and locations of variants would be interesting to better understand the impact of germline mutations and environmental risk factors in the development of breast cancer in young Rwandans.