Introduction

Male breast cancer (MBC) is a rare cancer entity, representing just 1% of all breast cancer cases [1, 2] and with the tendency to be diagnosed at an advanced stage, in part due to lack of awareness. Dedicated research studies on male breast cancer are limited, mainly due to the rarity of the disease, and therefore, MBC is currently being treated based on the knowledge extrapolated from female breast cancer (FMB). Quite recently, a large set of MBC cases was retrospectively centrally analyzed, showing that breast cancers arising in men are usually ER-, PR-, and AR-positive, Luminal B-like/HER2-negative [3].

MBC risk increases with hormonal abnormalities and advanced age, as well as with the number of first-degree relatives diagnosed with breast cancer [4]. More importantly, cancer-predisposing mutations, specifically in BRCA1 and BRCA2 genes, have been long-standing the key genetic risk factor for MBC diagnosis. Loss-of-function (LoF) mutations in BRCA1 and BRCA2 genes confer cumulative lifetime MBC risks of 1–5 and 5–0%, respectively, with BRCA2 mutations occurring more frequently [5, 6].

The prevalence of mutations in MBC in the post-BRCA era has been mainly assessed by individual efforts, with LoF mutations in PALB2 and CHEK2 mainly been reported. Specifically, through population-specific studies, the CHEK2 c.1100delC mutation was linked to tenfold increased risk for MBC [7, 8], while PALB2 mutations are reported to increase MBC risk by eight times [9]. In addition to that, males carrying LoF mutations in the syndromic genes NF1 and PTEN have been reported to be at increased risk for MBC diagnosis, but these studies are small and have ascertainment bias [10, 11].

The evolution of sequencing technologies, along with the implementation of panel testing in clinical practice, has enabled the identification of multiple breast cancer-predisposing mutations in additional genes, the majority of which involved female breast cancer (FBC) cases. Depending on the stringency of the selection criteria, mutation in genes other than BRCA1 and BRCA2 can account for 4–11% of FBC cases [12,13,14,15].

On the contrary, data deriving from comprehensive testing among MBC cases are limited, with the recent report by Pritzlaff et al., being the largest published, so far [16]. Through utilization of various gene panels, pathogenic or likely pathogenic variants were identified in sixteen genes in 18.1% of MBC patients, of various ethnicities, tested. The great majority of variants (12.3%) lied within BRCA1 and BRCA2, while CHEK2 was the second more frequent gene, detected in 4.1% of the cases.

To better understand the contribution and association of mutations in cancer genes to MBC, we interrogated a cohort of 102 Greek MBC patients, unselected for family history or age at diagnosis, by the largest tested so far, comprehensive gene panel that includes 94 cancer genes. The aim of this study was to evaluate the clinical utility of implementation of panel testing, at least for some genes, among MBC cases through investigation of patients deriving from a specific population with known founder effects.

Patients and methods

Patient cohort

One hundred and two male individuals diagnosed with breast cancer and treated mainly at Papageorgiou Hospital of Thessaloniki and the University Hospital of Heraklion, both in Greece, were invited to participate and to donate their biological material for future research purposes. Patients were unselected for family history and age at diagnosis. The study was approved by the Bioethics committee of NCSR “Demokritos,” as well as both hospitals’ ethic committees, in agreement with the 1975 Helsinki statement, revised in 1983. Genetic counseling was mandatory prior to genetic testing, and written informed consent was obtained from all individuals before performing genetic analysis.

Analysis of 94 cancer genes through next-generation sequencing

Genomic DNA, extracted from whole blood using the salt-extraction procedure [17], was used to prepare indexed libraries to target the sequence of 94 cancer predisposing genes using the Illumina Trusight Cancer Panel and was sequenced on a MiSeq analyzer (Illumina, San Diego, USA). FASTQ, BAM, and VCF files were produced through Basespace (Illumina, San Diego, USA). The minimum base and amplicon coverage was 50× and 100× respectively. All called variants of interest were confirmed by Sanger sequencing.

Variant annotation and classification

Called variants were annotated by VariantStudio version 3 (Illumina, San Diego, USA) against the human reference genome GRCh38. Variants were filtered and classified based on the recommendations published by American College of Medical Genetics and Genomics (ACMG) [18]. More specifically, all variants considered of unknown significance (VUS), with minor allele frequency (MAF) lower than 1%, have been assessed for their pathogenicity with the use of five in silico software (Align-GVGD, SIFT, PolyPhen, Mutation Taster and PhastCons). Therefore, the evolutionary amino acid conservation, biochemical and transactivation consequence of the amino acid change, as well as the effect on canonical splicing have been interrogated, while testing by functional assays has been also considered. Based on the data collected from all the aforementioned methods, each VUS has been categorized with a scale 1–5, with 1 and 5 can be considered polymorphic and pathogenic, respectively.

Multiplex ligation-dependent probe amplification (MLPA)

All samples were subsequently tested by MLPA for the detection of large genomic rearrangements. Specifically, SALSA MLPA KIT P002 and SALSA MLPA KIT P090 (MRC-Holland, Netherlands) were used for the BRCA1 and BRCA2 genes, respectively, according to the manufacturer’s instructions.

Results

Patient characteristics

This study cohort included 102 MBC cases (mean age at diagnosis: 62.8 ± 12 years). Among them, a small proportion (16/102; 15.6%) was diagnosed at a young age (< 50 years). Histology reports were available for eighty-two individuals; information on histology type, hormone receptors, and HER2 status was extrapolated. The majority of tumors (73/82; 89%) were estrogen receptor-positive; three cases were triple negative (3/82; 3.65%), while 14.6% (12/83) showed HER2 amplification. The most common histology type was ductal (80/82; 97.5%, of which two in situ diagnoses), with other types being rare (papillary and lobular, seen once and twice, respectively).

In total, 14.7% (15/102) of the patients were diagnosed with a second primary cancer, of which gastrointestinal malignancies (colorectal and duodenal cancer), represented one-third of them. Other cancer types involved are prostate, thyroid, pancreatic, bladder, laryngeal, and non-Hodgkin’s lymphoma. Interestingly, two of these patients were diagnosed with a metachronous breast cancer.

Mutation prevalence

Definitely pathogenic mutations were identified in 12.7% (13/102) of the patients tested, distributed in six cancer susceptibility genes. BRCA2 was the most commonly mutated gene, accounting for more than half of the mutations identified (7/102; 6.9%), followed by ATM (2/102; 1.96%). LoF mutations in all other genes (BRCA1, CHEK2, PMS2, and FANCL) were single events. The mutation distribution is illustrated in Fig. 1.

Fig. 1
figure 1

Proportions of male breast cancer patients with germline loss-of-function mutations in BRCA2, ATM, BRCA1, CHEK2, PMS2, and FANCL genes

Full size image

Two individuals (2/102; 1.96%) were found to carry a MUTYH monoallelic variant (p.Arg245His) and an APC allele p.Ile1307Lys, respectively, which are both associated with slight increase in lifetime colorectal cancer risk. Pathogenic variants in PALB2, NF1, and PTEN genes, which have been previously associated with MBC, have not been detected during this study. All definitely pathogenic mutations identified during this study, along with age at diagnosis, available clinicopathological data, and family history, are summarized in Table 1.

Table 1 Deleterious mutations identified in male breast cancer cases, along with tumor characteristics and details on family history
Full size table

Tumor pathology and characteristics of mutation carriers

Among the thirteen MBC carriers, the histology subtypes were ductal (10/12 available; 83.3%), lobular and papillary (1/12 available; 8.3%, each). The clinical subtypes included ten (90.9%) luminal and one (9.1%) HER2-positive breast cancers.

Additional characteristics of mutation carriers

The median age of breast cancer diagnosis in mutation carriers was 61.3 ± 11.3 years, which was not statistical significantly different to that of non-carriers (63 ± 12.1 years; p = 0.64). Second primary cancer diagnosis was reported in two cases, pancreatic cancer and multiple myeloma, both in BRCA2 carriers. Strong family history was reported in three cases, again all of which carried BRCA2 mutations, while four cases reported at least one family relative to breast cancer diagnosis. Selected pedigrees are illustrated in Fig. 2.

Fig. 2
figure 2

Selected pedigrees of male breast cancer families, carrying LoF mutations. Probands are represented by the arrow, while breast cancer patients are colored in black. BrCa breast cancer, Ca cancer, CRC colorectal cancer, PrCa prostate cancer, WT/MT allele carriers of mutations, WT/WT wild type for the familial mutation

Full size image

Change of clinical care based on panel test result

In total, an additional 2% of the patients tested will be offered amended surveillance protocols based on their test result. More specifically, PMS2 and CHEK2 carriers are potential candidates for gastrointestinal and colorectal cancer screening, respectively. This proportion can be doubled if specific guidelines are proposed for male ATM mutation carriers, specifically for prostate and pancreas surveillance for which at the moment there is insufficient evidence.

Variants of unknown clinical significance—variant classification

Assessment and classification have been performed for all detected variants located in gene coding regions, having MAF below 1%. Nine variants were identified in eight cancer susceptibility genes, have been considered suspicious, and/or have a potential to be associated with pathogenicity. Of these, BRCA1 p.Gly1727Ala has been classified as likely pathogenic, due to its rarity and the aggravating predictions of all tools used. The remaining missense changes were considered as variants of unknown significance, due to the conflicting evidence that is currently available, and were seen as follows: ATM (2), BRIP1 (1), CHEK2 (1), MLH1 (1), MSH2 (1), NBN (1), and PALB2 (1). All the above information is summarized in Table 2.

Table 2 Variants of unclassified significance along with the results of the in silico tools used and the available data on functional assays
Full size table

Discussion

In this study of unselected males with breast cancer, of Greek descent, 12.7% carried germline LoF alleles in six genes, previously associated with cancer predisposition. Traditionally, early age at cancer diagnosis, multiple primary cancers, and breast cancer family history are red flags for genetic testing referral. Interestingly, this does not seem to be the case for MBC. In line with previously published studies [16, 19], age at MBC diagnosis cannot be used as indicator of an underlying genetic defect. The study was designed to collect unselected MBC cases and, therefore, patient follow-up for secondary diagnosis, as well as detailed family history were not collected prior to analysis. Therefore, a limitation of our study is that statistical associations cannot be accurately performed and appropriately addressed for these cofactors.

BRCA2 mutations dominate the mutation spectrum, with a prevalence of 6.9%. Previous studies have reported BRCA2 mutations in 4–20% of MBC patients [13, 20, 21], with the upper end most likely involving ascertainment bias due to selection based on family history and population-specific alleles. The prevalence reported as 11 and 12%, in a recent large series of MBC [16] and in an Italian multicenter study [22], respectively, can be considered representative for mixed populations due to a large number of patients included and is higher than ours. This can be attributed to the genetic peculiarity of Greeks, characterized by BRCA1 strong founder effects, as shown through the extensive studies performed on female breast and ovarian cancer patients [23, 24]. Similarly, the BRCA2 mutation rate in Finnish MBC patients, another population with strong founder effects, was reported as 7.8% [25], which is comparable to ours.

The most frequent mutations in genes other than BRCA1 and BRCA2 were seen in ATM and CHEK2, which have been previously associated with both MBC and FBC. In our study, ATM mutations were the second most prevalent, accounting for 2% of the cases. Heterozygous ATM mutations have been associated with two- to fourfold increased risk for FBC [26, 27], while they also seem to increase pancreatic [28] and, possibly, ovarian cancer risk [29]. While there is no evidence for elevated risk for other cancers in ATM heterozygotes, this is the third published report of ATM mutation in MBC patients. Through multigene panel testing, ATM mutations were detected in six MBC patients (1.5%), of which one was an ataxia telangiectasia patient and two carried a pathogenic BRCA2 mutation [16]. Therefore, it seems that ATM mutations can be rare predisposing events for MBC, indicating that increased breast cancer surveillance can be proposed for male carriers, as for female carriers, based on National Comprehensive Cancer Network (NCCN) guidelines (https://www.nccn.org/).

Whether CHEK2 mutations confer increased risk for breast cancer is in the spotlight for many years. Clear associations and risks mostly derive from data on c.1100delC mutation, which is associated with a fourfold increased breast cancer risk in women and can be found in significant proportion among high-risk breast cancer patients in populations with founder effects. More specifically, among Finnish FBC and MBC patients, the prevalence is approximately 5 [30] and 5.9%, respectively [31]. In our series, a single CHEK2 LoF, but not truncating, allele (p.Gly167Arg) was identified. The low prevalence of CHEK2 mutations can be due to the rarity of c.1100delC allele in Mediterranean populations and especially, in individuals of Greek descent [32]. Quite recently, protein-truncating CHEK2 variants were associated with a 3.8-fold increased MBC risk and identified in ~ 2% of the MBC patients studied, indicating a clear association among CHEK2 truncating variants and MBC [16]. On the contrary, association with other LoF CHEK2 alleles is possible but questionable, mainly due to small number of carriers, and requires further investigation.

Interestingly, LoF mutations in two genes, which encode for proteins that are involved in DNA repair, and more specifically FANCL (Fanconi anemia pathway) and PMS2 (mismatch repair (MMR) pathway), that are not established predisposing breast cancer genes, have been identified. To the best of our knowledge, this is the first report of inactivating mutations among MBC patients. Specifically, the change in the initiator codon of FANCL has been detected in a MBC patient diagnosed at a relatively young age, who also had family history of breast cancer. A borderline association with increased FBC has been reported among Czech high-risk breast cancer patients carrying a four-base pair duplication in FANCL (c.1096_1099dupATTA) [33], located in the last exon of FANCL. This association was further doubted in a subsequent study on German breast cancer patients, where it was reported as a possible low-risk allele [34], mostly due to the likely hypomorphic nature of the specific allele. On the contrary, the mutation reported herein is pathogenic based on ACMG guidelines [35] and is predicted to alter normal protein FANCL production, therefore can be associated with both FBC and MBC susceptibilities.

Moreover, breast cancer predisposition conferred by mutations in MMR genes, including PMS2, has been questioned for long time. Quite recently, through a modified segregation analysis in a significant number of PMS2 carriers, the standardized incidence ratio of 3.8 for FBC risk showed to be significant. Due to the low number of carriers, this cannot be a definite association, but can be taken into account, especially in families that show breast cancer clustering [36]. Consistently, through a retrospective evaluation of patients who undergone panel testing, 11.9% carried MMR mutations alone and had breast cancer diagnosis only, with MSH6 and PMS2 mutations being statistically significant more frequent than MLH1 and MSH2 mutations, suggesting a possible breast cancer predisposition [37].

Germline LoF mutations among MLH1, MSH2, MSH6, and PMS2 genes are now known to confer variable lifetime cancer risks, with PMS2 mutations proposed to be associated with an attenuated Lynch syndrome phenotype, characterized by lower penetrance, later age at diagnosis, and lower risks for extracolonic Lynch syndrome-associated cancers [37,38,39]. Therefore, the cancer phenotypic spectrum associated with PMS2 mutations may be different, involving predisposition to breast cancer, as well. In the absence of a clear association of breast cancer to PMS2 mutations, the surveillance of this PMS2 carrier and his family relatives who will be tested positive for this mutation possibly need to be altered. Following the NCCN guidelines, gastrointestinal and gynecological surveillance is proposed for carriers, while chemoprevention by aspirin intake can be also proposed (https://www.nccn.org/). As a whole, clinical actionability will be proposed for ~ 2% of the MBC patients included in this study, based on the available evidence and current guidelines.

Interestingly, pathogenic variants in the genes PALB2, NF1, and PTEN, those have been previously reported among MBC cases [10, 11, 40], have not been identified through our study. This can be attributed to the relatively small size of our cohort, but can be indicative of the rarity or non-association of these genes with MBC pathogenesis.

In conclusion, this is an extensive evaluation of a comprehensive multicancer gene panel in a cohort of Greek MBC cases, unselected for family history or age at diagnosis. Our findings indicate that genetic testing for BRCA2 and BRCA1 can be sufficient for these patients, if taking into account prompt for change of care, which will be offered to only 2% of patients in our cohort beyond BRCA carriers. Although mutations in new genes that can be possibly associated with breast cancer predisposition emerged from this study, genetic events in genes, other than BRCA2 and BRCA1, involve rare cases. Until the production of sufficient evidence and clinical guidelines, from larger studies, it seems that gene panel testing may be of limited benefit for MBC patients of specific descent.