1. Introduction
Hereditary breast, ovarian, and pancreatic cancers are associated with the presence of germline pathogenic (P) or likely pathogenic (LP) variants in the
BRCA1 and
BRCA2 genes. However, damaging mutations in these two genes justify no more than 20% of familial forms of these malignancies [
1]. As the remaining 80% is still waiting for genetic diagnosis, the discovery of new genes involved in the susceptibility of hereditary cancers is under continuous investigation. The identification of pathogenic variants in other genes at both a germline and somatic is, therefore, crucial for the future of primary prevention strategies (prophylactic surgery and drug-prevention), surveillance programs, and targeted therapy. In this scenario, the research of hereditary breast/ovarian cancer susceptibility genes is crucial.
BRCA1/2 have been primarily investigated for diagnostic purposes because their mutations show high penetrance, conferring the 5-fold higher risk of breast cancer in P/LP variant carriers compared to the general population [
2].
BRCA1 and BRCA2 play a crucial role in the DNA double-strand break repair (DSBR) machinery by homologous recombination (HR). In this highly conserved mechanism, they interact with different proteins, including ATM, a master kinase acting upstream in the genome surveillance pathway, mainly activated by double-strand breaks (DSBs) [
3]; MRN complex (MRE11, NBN, RAD50), able to detect DSBs [
4]; CHEK2 that allows DNA repair by arresting the cell cycle at the G1/S checkpoint [
5]; BARD1 and BRIP1, which interact with BRCA1 at N and C-terminal regions, respectively [
6,
7]; PALB2 and the paralog RAD51C and RAD51D, all involved in the BRCA complexes required for HR [
8,
9,
10], and LKB1, encoded by
STK11, which co-localizes with ATM and BRCA1 at the sites of the DNA damage [
11]. All of these DBSR genes have been already associated with hereditary breast/ovarian and pancreatic cancers, as well as colon and gastric cancers (CC and GC) [
12]. In addition, other genes involved in DNA damage repair (DDR) pathways different from HR and cell cycle control, such as
APC,
CDH1,
CDK4,
CDKN2A,
PTEN,
SMAD4,
TP53 [
13,
14,
15,
16,
17,
18], and the DNA helicase RECQL [
19], have been associated with high or moderate susceptibility to familial breast cancer (BC) and other types of malignancies.
Next-generation sequencing (NGS) studies have recently demonstrated that some genes causing hereditary gastrointestinal cancer syndromes are risk factors for breast, ovarian, and pancreatic cancers. Indeed, P/LP variants in the mismatch repair (MMR) genes, such as
MSH2,
MLH1,
MSH6,
PMS2, and
EPCAM, classically associated to hereditary colon and endometrial cancers (Lynch Syndrome), have also been identified in the breast, ovarian, biliary, and gastric tumors [
20,
21].
MUTYH, a gene involved in the DDR by base excision repair (BER) and responsible for the autosomal recessive form of familial colorectal cancer polyposis, has been recently proposed as a risk factor for breast cancer in males [
22].
The identification of P/LP variants in DDR and cycle cell genes is becoming one of the main goals of the oncology clinical research. Alterations in these genes are emerging as novel targets for treatment in different cancers and, particularly, for personalized therapies. PARP (Poly(ADP-ribose) Polymerase)-inhibitors, for instance, have been introduced in the treatment of
BRCA and, more recently, of other HR deficiency-related malignancies with encouraging results [
23].
To date, however, the prevalence of germline mutations in non-BRCA DDR genes is partially investigated in BC, ovarian cancer (OC), and pancreatic cancer (PC), and available data about these genetic risk factors in cancer disease are still poor.
By applying NGS technologies, we analyzed 25 genes involved in DDR and in the cell cycle control in a cohort of 113 non-BRCA patients with personal and/or family history of BC, OC, and/or PC. This study aimed at (1) broadening the mutational spectrum and better defining the prevalence of P/LP variants in non-BRCA cancer-related genes, (2) evaluating the clinical utility of the multigene panel, (3) identifying novel actionable variants, and (4) improving the efficiency of clinical diagnostic tests.
4. Discussion
The present study aimed at detecting pathogenic variants related to hereditary cancers by multigene panel testing. In total, 113 consecutive individuals with personal or family history of breast, ovarian, or pancreatic cancer and without P/LP variants in BRCA1 and BRCA2 genes were analyzed. By applying NGS technologies in our cohort, we investigated the frequency of germline deleterious variants in APC, ATM, BARD1, BRIP1, CDH1, CDK4, CDKN2A, CHEK2, EPCAM, MLH1, MRE11, MSH2, MSH6, MUTYH, NBN, PALB2, PMS2, PTEN, RAD50, RAD51C, RAD51D, RECQL1, SMAD4, STK11, and TP53 genes, all involved in DDR system. The choice of including these genes in our panel was based on several considerations.
The DNA damage response plays a critical role in maintaining genomic stability, and hereditary mutations in DDR genes often confer cancer susceptibility. Bi-allelic mutations in DSBR genes and in other DDR genes are indeed the basis of cancer-prone recessive hereditary syndromes, such as Ataxia–Telangiectasia, Nijmegen Breakage Syndrome, and Fanconi Anemia. Moreover, alterations in the MMR genes result in Lynch Syndrome, leading to an increased incidence of gastrointestinal, endometrial, and ovarian cancers.
Most chemotherapy agents currently used in cancer therapy cause DNA damage. DDR pathway inhibitors are now used to make cancer cells more sensitive to chemotherapy—an approach called synthetic lethality. In the era of PARP inhibitors, employed in BRCA carriers with OC and metastatic BC, the identification of further potential targets in other HRR (Homologous Recombination Repair) genes could provide new therapeutic opportunities for this and other cancers related to defects in DDR genes.
Genetic testing for hereditary cancer predisposition has evolved rapidly. Many genes included in multigene panels have inaccurate estimations about the degree of associated cancer risk, and there is no consensus on when to test a particular gene or how to manage an identified P/LP variant. A survey conducted in 61 centers from 20 countries by the clinical group of ENIGMA (Evidence-Based Network for the Interpretation of Germline Mutant Alleles) in 2018 [
66] showed that, beyond
BRCA1/2, only a small number of genes are currently analyzed worldwide, and management guidelines are limited. On the other hand, the clinical utility of detecting pathogenic germline variants in high and moderate penetrance genes is strong, as recommended by NCCN guidelines for cancer prevention, surveillance, and management.
In total, 14 different P/LP variants in 6/25 DDR pathway genes were identified in 16 probands. Despite the rarity of each damaging mutation, the overall pathogenic variants rate in these 25 DDR genes was 14% (16/113 unrelated
BRCA1/2-negative cases). Moreover, 23 probands carried VUS (20%), whereas 74 (66%) were negative (
Figure 1a). In several NGS studies of patients with BC and/or OC, beyond
BRCA1 and
BRCA2,
CHEK2 has been one of the most frequently mutated genes [
67,
68,
69]. This data was confirmed in our sample, in which 7.1% (8/113) of analyzed patients were found carriers of deleterious mutations in the
CHEK2 gene.
RAD51C, with a frequency of 2.6% (3/113), was the other most mutated gene in our patients. These data supported the rationale of including this gene in the NGS panels for the assessment of BC/OC risk.
Germline
CHEK2 P/LP variants have been associated with Li-Fraumeni like syndrome, BC, and other cancers, including prostatic, gastrointestinal, and, although still debated, OC [
70]. In our study, P/LP variants in
CHEK2 were detected in women with monolateral and bilateral BC, who referred at least one or more different types of malignancies, including BC, OC, and gastrointestinal cancers, in their relatives. Moreover, the c.793-1G>A variant was identified in a woman with a family history of osteosarcoma and young-onset BC that suggested a Li-Fraumeni like syndrome. Intriguingly, the proband had papillary thyroid cancer, recently described in
CHEK2 patients [
59]. Unfortunately, segregation studies could not be performed in the deceased parents, and we could not determine if the variant, detected also in her asymptomatic brother, was inherited from the maternal branch, in which breast, gastrointestinal, and brain cancers occurred, or from the paternal side, where three paternal uncles had PrC.
CHEK2 LP variants, c.1136C>G, p.(Ser379Cys) and c.1169A>C, p.(Tyr390Ser), were detected in a woman with melanoma (
P12) and in a Mullerian sarcoma case (
P04), respectively, both referring BC and gastrointestinal cancers in relatives. Previous studies have demonstrated that
CHEK2 c.1100del heterozygotes have a two-fold risk of malignant melanoma compared to non-carriers, while no clear associations have emerged between Mullerian sarcomas or, more generally, isolated uterine tumors and
CHEK2 [
71,
72]. Nevertheless, the recurrence of BC and GC and CC in many relatives of these two families was in line with the typical phenotype related to this gene.
Recently
CHEK2 pathogenic variants have been considered a proven genetic risk factor for male BC. Although many authors have confirmed this association worldwide [
70], the detection of P/LP
CHEK2 variants in Italian male BC cases was poor [
73]. Despite this data, among our nine mBC cases, one of the three mutated patients, who referred BC and OC among paternal and paternal relatives, carried the c.1367C>T LP variant in
CHEK2.
Finally, the c.470T>C, p.(Ile157Thr)
CHEK2 variant, described in BC and many other types of malignancies, was found in a proband with four primary different tumors (BC, OC, CC, and GC), who referred multiple cancers also in her father (RC and PrC).
CHEK2 damaging variants have been found in families with multiple primary cancers [
72,
74,
75]. However, to our knowledge, a high number of primary tumors has never been described in patients carrying the
CHEK2 c.470T>C mutation. Unfortunately, parental segregation studies could not be performed, preventing the investigation of the variant in her affected father.
Mutations in the
RAD51C gene, encoding a protein involved in HR, were found in 3/113 cases analyzed—two patients with OC and a woman with bilateral BC. Deleterious variants in this gene have been associated with a higher risk of epithelial ovarian carcinoma, especially with early onset [
76]. Although a well-defined genetic risk factor for OC, the role of
RAD51C in BC, is still debated [
38,
77]. The woman carrying the
RAD51C missense gene alteration had a bilateral metachronous BC, a lobular carcinoma on the left, and a high-grade triple-negative invasive DC occurred 11 years later on the right. Intriguingly, in patients with BC and damaging mutations in
RAD51C, triple-negative cancers recurred many times [
78,
79].
Bi-allelic P/LP variants in the
ATM gene cause Ataxia–Telangiectasia (AT), a neurodegenerative progressive disease complicated by immunodeficiency and cancer predisposition. Germline
ATM heterozygous carriers are about 0.75–1% of the population. Mono-allelic variants of this gene are proven as moderate risk factors for malignancies, including breast, pancreatic, prostatic, and other solid cancers. In our cohort,
ATM pathogenic variants were identified in two probands with multiple cancers. The
P29 patient, affected by BC and melanoma, had a paternal family history of different types of malignancies previously described in
ATM heterozygous patients; the
P89 patient had a metastatic PC and a previous diagnosis of GC. In his family, nine relatives deceased for gastric cancer. Although not frequently, pathogenic variants in
ATM have been previously described in patients with GC [
80,
81,
82]. PC remains one of the most lethal solid malignancies. The identification of damaging mutations in DDR system genes, including
ATM, in 17–25% of this type of cancer and the recent suggestion that PARP inhibitors could have therapeutic potential in cancers with loss or mutation of
ATM are opening up the possibility of new therapies, such as platinum and more recently PARP inhibitors, also in
ATM-mutated patients with PC [
83,
84,
85].
LP variants in the MMR genes
MLH1 and
MSH2 were found in two women with BC. Although BC is not included in the spectrum of Lynch Syndrome (LS)-related malignancies, an association between LS germline mutations and this cancer has been recently suggested. Many studies indeed have reported a higher risk of BC in patients with LS and a higher frequency of MMR gene variants in BC cases [
21,
86]. However, women with pathogenic variants in MMR genes are not usually advised to increase breast cancer screening. Furthermore, immunohistochemistry for MMR proteins is not performed for the identification of deficient breast cancers. Studies able to clarify this association are needed also for therapeutic implications. Checkpoint inhibitors, such as pembrolizumab, are becoming available treatments for all microsatellite instable-high and/or MMR deficient solid tumors (including breast cancer) [
87].
In this study, we described a new frameshift deletion c.363_364del in the
RECQL gene in a male with infiltrating ductal breast and prostatic cancer at 59 and 70 years of age, respectively, with a positive family history for breast and lung cancer (
Figure 2d).
RECQL encodes a protein that is part of a family of five RECQ helicases, including at least three implicated in cancer-prone syndromes, such as Bloom Syndrome, Werner Syndrome, and Rothmund-Thomson Syndrome. These diseases are indeed caused by bi-allelic mutations in the
BLM,
WRN, and
RECQ4 genes, respectively [
88].
RECQL is a helicase involved not only in the unwinding of the DNA but also in the promotion of complementary single-strand DNA annealing. Its role in DDR has been clearly proven by many studies that have demonstrated chromosomal instability, stalled and collapsed replication forks, oxidative damage, higher DSBs in cells with a deficit of RECQL [
29]. In 2015, two independent research groups associated the damaging variants in
RECQL with a higher risk of BC in studies conducted in Polish and Canadian populations [
65,
89]. However, this correlation was not confirmed in subsequent studies [
90]. More recently, mutations in
RECQL emerged as a moderate risk factor for BC in a cohort of African American women, indicating that mutations in the
RECQL gene confer a moderate risk of BC [
91]. These conflicting correlations could be explained by penetrance variability due to the consequences of the different types of the identified variants on the protein function and to the investigated cohorts belonging to different ethnicities [
92].
More than 20% of patients have carried VUS. This type of inconclusive result is a hard challenge to face. With the advent of panel analysis, the number of VUS has increased exponentially, and still, too often, their interpretation remains tangled and blurred. Unfortunately, in most cases, VUS causes difficulty in risk assessment, sometimes overtreatment, and usually anxiety in carriers. Bioinformatic analysis, functional studies, and periodical updates performed by international consortia [
93] are the currently available strategies to shed light on this issue [
94].
Laboratories and clinicians should collaborate in order to guarantee periodical re-evaluations and updates on their variants to VUS carriers [
27].
The comparison of clinical and familial features between patients with and without P/LP variants brought out interesting suggestions. In our cohort, P/LP variants were more frequent in women with bilateral BC, as reported before [
95]. There was not a prevalent histological phenotype in mutated BC women compared to BC cases without damaging mutation nor a strong recurrence of specific types of cancers among relatives. Intriguingly, but not surprisingly, a trend of higher frequency of prostatic cancer among relatives of BC women with P/LP variants emerged. This association was stronger in relatives of mutated OC patients. Pathogenic variants in DDR and MMR genes, usually analyzed in suspected hereditary BC and OC, have been recently associated also with hereditary prostatic cancer [
96].
A wider number of probands and the availability of segregation among relatives would have given the opportunity to furtherly strengthen the association between the identified variants and the predisposition to cancer. Future studies investigating P/LP variants in DDR genes in many BC, OC, and PC cases, familial or sporadic, and in their relatives would be necessary to better define the role and the weight of these genes in determining malignancies.
In conclusion, in this study, we described that 14% of non-BRCA patients with BC/OC tumors are carriers of pathogenic variants in other genes, particularly CHEK2, RAD51C, ATM, MLH1, MSH2, RECQL, all related with the BRCA1/2 DNA repair pathway. This result shows that the DDR genes panel significantly increases the diagnostic power in patients with personal and/or family history of breast/ovarian and pancreatic cancers.
The identification of mutations in genes involved in DNA damage response, other than BRCA, explains the strong tumor recurrence in some families and may contribute to the development of new and more specific clinical management programs and pave the way to new therapeutic opportunities.