Genetic testing results in Slovenian male breast cancer cohort indicate the BRCA2 7806-2A > G founder variant could be associated with higher male breast cancer risk

Purpose To analyze the prevalence of pathogenic/likely pathogenic variants (P/LPVs) in BRCA1 and BRCA2 genes in the largest cohort of Slovenian male breast cancer (MBC) patients to date and to explore a possible correlation between the Slovenian founder variant BRCA2:c.7806-2A > G and predisposition to MBC. Methods We performed a retrospective analysis of 81 MBC cases who underwent genetic counseling and/or testing between January 1999 and May 2020. To explore a possible genotype–phenotype correlation, we performed additional analyses of 203 unrelated families with P/LPVs in BRCA2 and 177 cases of female breast cancer (FBC) in carriers of P/LPVs in BRCA2. Results Detection rate of P/LPVs in the BRCA1 and BRCA2 genes was 24.7% (20/81) with 95% of them in BRCA2 gene. The only two recurrent P/LPVs were BRCA2:c.7806-2A > G and BRCA2:c.3975_3978dupTGCT (9 and 5 MBC cases, respectively). In families with BRCA2:c.7806-2A > G, the incidence of MBC cases was higher compared to families with other P/LPVs in BRCA2; however, the difference did not reach statistical significance (17.8% vs. 8.9%, p = 0.105). BRCA2:c.7806-2A > G was detected in both families with multiple cases of MBC. This splice-site variant represented a significantly higher proportion of all BRCA2 P/LPVs detected in MBC carriers compared to FBC carriers (47.4% vs. 26%, p = 0.049). Conclusion We observed a high mutation detection rate and conclude this may be due to the prevalent BRCA2:c.7806-2A > G variant in Slovenia. Our results indicate a possible association between this variant and higher risk of breast cancer in males compared to other identified P/LPVs in BRCA2.


Introduction
Male breast cancer (MBC) is a rare disease, comprising up to 1% of all breast cancers and up to 1% of all cancers in males [1][2][3]. Age-adjusted incidence rate in Slovenia is reported to be 1.1 per 100,000 and is in line with incidence rates reported in other Caucasian populations [2,4,5]. Various demographic, environmental, hormonal, and genetic factors have been associated with MBC [2,6]. Among genetic factors, mutations in BRCA1 and, especially, BRCA2 genes are the most clearly established predispositions [2]. Genetic testing is recommended for all MBC patients in order to guide screening recommendations for other types of cancer and to identify other family members at risk [1,7], as well as for treatment with PARP (poly-ADP ribose polymerase) inhibitors in metastatic settings according to the American Society of Clinical Oncology 2020 guidelines [8].
In 2008, Besic together with members of our group published first results of genetic testing in a small cohort of 25 Slovenian MBC patients and reported a high prevalence of 16% of pathogenic/likely pathogenic variants (P/LPVs) in BRCA2 [9]. In three out of four MBC carriers of P/LPVs in BRCA2, the highly recurrent Slovenian founder splice-site variant BRCA2:c.7806-2A > G (formerly known as IVS16-2A > G) was detected [10,11]. A group from Aviano, Italy, also reported this variant as recurrent in northeast part of Italy and suggested an association between this splice-site BRCA2 variant and risk of breast cancer in males [12][13][14].
The aim of our study was to analyze the prevalence of P/ LPVs in BRCA1 and BRCA2 genes in the largest cohort of Slovenian MBC patients to date. We performed a retrospective analysis of all MBC cases who were referred to our Cancer Genetic Clinic at the Institute of Oncology Ljubljana, Slovenia, since its inception in 1999. Our primary objective was to re-evaluate the high detection rate of P/LPVs in BRCA1 and BRCA2 genes in a small cohort of Slovenian MBC patients, reported previously by Besic et al. in 2008 [9]. In addition, we wanted to document the spectrum of all P/LPVs as well as possible difference between carriers and non-carriers of P/LPV in BRCA1 and BRCA2 genes. Our secondary objective was to explore a possible correlation between the BRCA2:c.7806-2A > G variant and predisposition to MBC compared to other BRCA2 P/LPVs identified in our cohort. We also evaluated certain clinical characteristics of MBC in carriers of BRCA2:c.7806-2A > G compared to other P/LPVs in BRCA2.

MBC cohort and data collection
We performed a retrospective analysis of MBC cases identified in the Cancer Genetics Clinic's database at the Institute of Oncology Ljubljana, the only comprehensive cancer center in Slovenia. Our cohort is unselected for family history of cancer, since MBC has always been one of the independent inclusion criteria for genetic testing at the Institute [11]. All MBC cases referred to our Clinic since the introduction of genetic counseling and testing in 1999 up to May 2020 who opted for genetic testing were included. Only P/ LPVs in BRCA1 and BRCA2 genes were analyzed.
For the purpose of this study, positive family history for hereditary breast and ovarian cancer syndrome (HBOCrelated) cancers was defined as a family history of female breast, ovarian, prostate, and/or pancreatic cancer in first and/or second-degree relatives. As part of our routine clinical protocol, all probands are required to report their family history of cancer, which is then verified in the Cancer Registry of Republic of Slovenia, a national system of mandatory reported cancer cases, established in 1950. Also, we used this database to check if the MBC patients in our cohort had been diagnosed with any other cancers.
Information on relevant MBC clinical and pathological characteristics was retrieved from the Institute's electronic database and patients' medical records. We applied surrogate definitions of intrinsic subtypes of breast cancer as defined by the St. Galen International Expert Consensus 2013 [15]. For the purpose of exploring possible genotype-phenotype correlation between the BRCA2:c.7806-2A > G variant and predisposition to MBC compared to other BRCA2 P/ LPVs identified at our Institute, we performed an additional analysis of all unrelated families with P/LPVs in BRCA2 detected at our Institute since 1999. We recorded all cases of MBC in these families. Also, all cases of breast cancer in female carriers of P/LPVs in BRCA2 were identified from the Cancer Genetics Clinic's database.
We obtained the approval for our analysis from the Institute's Committee for Medical Ethics (#ERIDEK-0069/2020).

Genetic testing methods
Genetic testing was performed at the Department of Molecular Diagnostics at the Institute as well as other laboratories [11]. Together, we identified 81 MBC cases. Different genetic testing methods were used. In 55 MBC cases, genetic testing with next-generation sequencing (NGS) was performed on blood samples. Of these, 54 were tested using NGS of targeted panels Illumina's TruSight Cancer Panel or TruSight Hereditary Cancer Panel [16]. Large intragenic deletions in BRCA1 and BRCA2 genes were detected from NGS data with copy number analysis using SeqNext v4.4.0 (JSI medical systems) or with multiplex ligation-dependent probe amplification (MLPA). Identified P/LPVs in BRCA1 and BRCA2 were confirmed using Sanger sequencing or MLPA analysis from separate blood samples. In one patient, clinical exome sequencing was performed at another laboratory as previously described by Bergant et al. [17]. In 15 MBC cases, who were tested prior to the introduction of NGS in 2014, only a limited mutational screen for the six highly recurrent P/LPV in BRCA1 and BRCA2 in Slovenian population (c.7806-2A > G in BRCA2; c.5266dupC, c.1687C > T. c.191G > A, c.181 T > G and c.181 T > A in BRCA1) was performed. In these, denaturing gradient gel electrophoresis (DDGE) for exon 4, part of exon 10 and exon 19 of BRCA1, DDGE for exon 17 of BRCA2, and in two cases additionally protein truncation test (PTT) for exon 10 of BRCA1 and exon 11 of BRCA2 were used as detailed elsewhere by our group [11]. In four MBC cases, complete screening of all BRCA1 and BRCA2 exons with combination of high-resolution melting (HRM), DDGE, and Sanger sequencing methods as well as MLPA for detecting large genomic deletions was performed as previously described [18,19]. Out of 55 patients who underwent NGS testing, 42 were previously untested and 13 had already tested negative with previously performed genetic screening using DGGE, HRM, PPT, or Sanger sequencing; all living MBC patients who tested negative prior to the introduction NGS at our Institute in 2014 were re-contacted, and 13 of them responded and opted for re-testing with NGS. In three MBC cases, only Sanger sequencing was performed to determine known familial pathogenic variants in BRCA2 gene. In four MBC cases, NGS testing (TruSight Tumor 170) was performed on archived formalin-fixed paraffin-embedded (FFPE) tissue samples (two samples were obtained from tumor tissue and two from non-tumor) as previously described by Klancar et al. [20]. Flow chart of genetic testing methods performed as well as detection of P/LPVs in BRCA1 and BRCA2 with different testing methods used is depicted in the Fig. 1.

Statistical analysis
We used descriptive statistics to describe relevant MBC patients' clinical, pathological, and genetic as well as their families' characteristics. Association between categorical variables was evaluated using Pearson's chisquared or Fisher's exact test, as appropriate. To compare means, we used the independent-samples t-test. All significance tests were two-sided using an alpha level of 0.05. The statistical analysis was performed with the licensed IBM SPSS Statistics software, version 25.

MBC cohort characteristics
We identified 81 MBC cases from 80 unrelated families who underwent genetic counseling and/or testing at our Institute between January 1999 and May 2020. Pathological and clinical characteristics of MBC cases are summarized in Table 1. Patients were diagnosed with breast cancer between 1970 and 2019 with half of them diagnosed prior to 2011. Excluding non-melanoma skin cancer and contralateral breast cancer, 19/81 (23.5%) MBC patients had a personal history of additional malignant primary tumors. Melanoma was reported in four MBC patients, colorectal cancer in three, prostate cancer in three, renal cell carcinoma in two, hematological malignancy in two, lung cancer in one, conjunctival squamous cell carcinoma in one, unknown primary in one, while two patients had multiple additional primary tumors (prostate and gastric cancer; prostate and lung cancer). In 31/80 (38.7%) of unrelated families with MBC cases, there was a family history of female breast, ovarian, prostate, and/or pancreatic cancer, and in 2/80 (2.5%), there was a family history of male breast cancer (a brother in one and a father in the other).

Discussion
In our study, we aimed to analyze the prevalence of P/ LPVs in BRCA1 and BRCA2 genes in the largest cohort of Slovenian MBC patients to date. Our primary objective was to re-evaluate the high detection rate of P/LPVs in BRCA1 and BRCA2 as previously reported [9]. We hypothesized the detection rate could be even higher, as more accurate genetic screening methods including nextgeneration sequencing have been implemented in clinical practice since then. Here, we report a high detection rate (24.7%) of P/LPVs in BRCA2 (23.5%) and BRCA1 (1.2%) in the largest cohort of Slovenian MBC patients (n = 81). Our cohort is very similar to other published cohorts regarding the age and stage at diagnosis [21][22][23][24], histology, and intrinsic subtypes of breast cancer [2,23,[25][26][27].
As we predicted, the detection rate is higher than the first reported detection rate of 16% (4/16) in a small cohort of Slovenian MBC patients in 2008 [9]. Then, only limited genetic screening was performed for highly recurrent P/ LPVs in BRCA1 and BRCA2 detected in Slovenian HBOC families [11] and extended genetic analysis was limited only to those with a substantial family history of breast and/or ovarian cancer [9]. Recently published detection rates of P/LPVs in BRCA1 and/or BRCA2 in unselected MBC patients range between 7.8% and 22% in smaller cohorts [28][29][30][31][32]. In two of the largest recently published cohorts, consisting of 382 MBC patients in the collaborative Italian Multicenter Study and 708 mixed population MBC patients tested by Ambry Genetics, California, reported detection rates were 13% and 9%, respectively [33,34]. Not surprisingly, most deleterious variants were attributed to BRCA2 and only 1% or less to BRCA1, as also seen in our cohort [28,[30][31][32][33][34].
Reported family history of other HBOC-related cancers in 80 unrelated families with MBC cases in our cohort was similar to others (38.7% in our cohort vs. 32% to 39% reported in other unselected MBC cohorts) [30,32,33]. We observed a particularly high (43.7%) detection rate in those with a family history of other HBOC-related cancer in first-and/or second-degree relatives. Observed detection rates in our study are in line with high detection rates observed in MBC patients from populations with founder effects [35][36][37].
Interestingly, almost one third of MBC patients, who were carriers of P/LPVs in BRCA1 or BRCA2, had no family history of other HBOC-related cancers in their firstand/or second-degree relatives. The detection rate in these was high (12.2%). However, these results are similar to those published by other groups who reported between 3.6% and 13% detection rates in MBC patients without any family history of breast cancer [23,32,37,38]. Our findings confirm that genetic testing should be offered to all MBC patients regardless of their family history of other cancers.
In our cohort of MBC patients, the splice-site c.7806-2A > G in BRCA2 was the most common deleterious variant detected and represented almost half of all variants in BRCA2 gene. Our group previously reported this variant as a founder Slovenian mutation [10,11]. It is the most common pathogenic variant in the BRCA2 gene, detected in Slovenian HBOC families [18], most commonly detected in the central Osrednjeslovenska and eastern Savinjska region of Slovenia [39]. Different published reports on functional analysis of the BRCA2:c.7806-2A > G variant showed at least three abnormal transcripts: skipping of entire exon 17, out-offrame skipping of 20 nucleotides at 5′-end exon 17 and inframe skipping of 69 nucleotides at 5′-end exon 17 (13 out of 23 amino acids lost being strictly conserved) [14,40,41].
This variant was also reported as recurrent in the Italian region Friuli-Venezia-Giulia (FVG) [12][13][14], which neighbors the western part of Slovenia. Their analysis in 13 Italian carrier families from the FVG region demonstrated that all carriers shared a common haplotype and a common ancestor estimated around 94 generations or 2350 years ago [12]. It is not clear if the haplotype from Italian and Slovenian families is the same as such studies have never been performed [12].
Currently, there is no evidence that would clearly indicate genotype-phenotype correlation between the location of a particular variant within the BRCA2 gene and predisposition to MBC [42] and no MBC-specific cluster region has been identified [43,44]. However, it has been reported that some P/LPVs in BRCA2 were more frequent in MBC cases compared to FBC [36,37], that male carriers of a specific variant had significantly higher lifetime risks of breast cancer compared to other variants in BRCA2 [45] and that large genomic rearrangement in the BRCA2 gene tends to be more frequent in families with MBC cases [46][47][48].
Possible association between the BRCA2:c.7806-2A > G variant and predisposition to MBC was first suggested by the group of researchers from Aviano. Cini et al. found that 39% (7/18) of BRCA2:c.7806-2A > G families had MBC cases compared to 16% (13/80) of families with other BRCA2 P/ LPVs (p = 0.049) [12]. They also observed a family with this variant and three cases of MBC [12]. To explore possible genotype-phenotype correlation between the splicesite variant and predisposition to MBC in our cohort, we performed additional analysis of all unrelated BRCA2 families documented at our Institute. This variant was the most frequent among all deleterious variants in BRCA2 detected in MBC and FBC patients from these families; however, the frequency of the BRCA2:c.7806-2A > G variant was significantly higher among MBC compared to FBC carriers (p = 0.049). Also, MBC cases were reported twice as often in the BRCA2:c.7806-2A > G-positive families compared to families with other P/LPV in BRCA2 identified, but the difference did not reach statistical significance. However, both MBC patients in our cohort with family history of MBC were carriers of this splice-site variant, as was one out of two MBC patients with bilateral breast cancer. As reported by the group from Aviano [12], our analysis also suggest possible genotype-phenotype correlation between the BRCA2:c.7806-2A > G variant and predisposition to male breast cancer compared to other deleterious BRCA2 variants identified at our Institute. However, both studies are small and further research on larger cohorts will be needed. Breast and prostate screening for male carriers of BRCA1/2 P/LPVs has been offered at our institution for more than 10 years. If carriers of BRCA2:c.7806-2A > G variant are in fact at greater risk for developing MBC than other BRCA2 carriers, their screening recommendations could be modified in view of that risk.
There are several limitations to our study. Although our cohort represents the largest Slovenian set of MBC patients to date, the absolute number of included cases is still small. However, this is a rare disease and Slovenian population is less than 2.1 million people. As this is a retrospective analysis, our cohort is very heterogenous in view of genetic tests performed. We did not identify any P/LPV in BRCA1 and BRCA2 in 13 MBC cases who were re-tested using NGS after previously testing negative using other genetic testing methods. Still, some P/LPV could have been missed since in 14 negative MBC cases only limited genetic screening for the six highly recurrent P/LPV in BRCA1 and BRCA2 was performed [49]. Since the limited genetic screening in 14 negative MBC cases covered BRCA2:c.7806-2A > G variant but did not include BRCA2:c.3975_3978dupTGCT, which was the second most common recurrent variant identified in our cohort, this might led to an underestimated proportion of this variant compared to the c.7806-2A > G among MBC patients in our cohort. Also, variants in other non-BRCA1/BRCA2 genes have been linked to MBC or observed in MBC patients [34,50], but are not reported in our study. Finally, our study was conducted in a small cohort of patients from the same clinic serving an ethnically homogenous population. Bias due to population stratification could therefore have an important effect on the results of statistical testing.
In conclusion, we observed a high mutation detection rate in this largest cohort of Slovenian MBC patients to date. As predicted, detection rate was even higher than previously reported in a much smaller cohort [9]. We conclude the observed high detection rate may be due to the prevalent Slovenian founder variant BRCA2:c.7806-2A > G. As previously suggested by the group of researchers from Aviano [12], our results also indicate a possible association between this splice-site BRCA2 variant and higher risk of breast cancer in males compared to other identified P/LPVs in BRCA2 gene. Further research on larger cohorts is needed to explore this. Once genotype-phenotype correlations are better defined, personalized risk assessment and follow-up could be offered to carriers of genetic cancer predispositions.