Multiple primary cancers (MPCs) are defined as the occurrence of more than one synchronous or metachronous cancer in a patient. For reporting purposes, the identification of MPCs is variable across the regulatory agencies. The Surveillance, Epidemiology and End Results (SEER) program considers the histology, site, laterality, and time since initial diagnosis of the tumor, while for the International Agency for Research on Cancer (IARC) only one tumor is registered for an organ, irrespective of time. The definition of synchronous and metachronous tumors also varies, having a threshold of 2 months in SEER and 6 months in IARC (1,2).

The development of screening tests and new cancer therapies improved cancer patient survival. By 2005, almost 900,000 of the 11 million cancer survivors were diagnosed with more than one cancer. The risk of developing subsequent cancers varies with regards to the type of first primary site, age at diagnosis, environmental exposure and genetic factors (3).

The identification of genes associated with hereditary cancer risk started 25 years ago with the discovery of BRCA pathogenic variants in families with breast and ovarian cancers. Since then, more than 80 cancer predisposition syndromes and 100 cancer-related genes were identified (4). For each syndrome, the most commonly associated cancers are defined but the exact risk profile remains often unknown. One common characteristic of these conditions is the young age at cancer diagnosis in most patients, although a considerable heterogeneity is observed. Additionally, it is widely stated that patients with multiple primary cancers are more likely to carry germline pathogenic variants in cancer-related genes (5).

For patients with MPCs without a germline molecular genetic diagnosis after initial evaluation, single-nucleotide variants (SNVs), copy number variants (CNVs), or de novo chromosomal rearrangements could be the cause of the disease. Another possibility could be the additive effect of multiple genetic events influencing a single biochemical mechanism. The aim of this study was to identify novel genetic mechanisms associated with risk of tumor development in patients with MPCs.

In this study only two out of 10 patients with MPCs presented clearly pathogenic SNVs. Previous reports showed similar results. When performing a whole-genome sequencing (WGS) of a cohort of patients with MPCs who had undergone genetic assessment, undetected germline pathogenic variants were identified in 15.2% cases (16). In another study, 21% of patients with MPCs had at least one PV identified (17). The rest of our patients had one or more VUS that could potentially act together to cause the disease, supporting the hypothesis of an oligogenic effect, which has been described before (18).

The oligogenic effect of combinations of low penetrance SNVs in cancer-related genes has been suspected for a long time, including in young patients with breast and lung cancers. In a case-control study of 631 women with breast cancer diagnosed under the age of 53, it was proposed that SNVs in ten genes with known or predicted roles in breast cancer interact to affect a woman's cancer risk in a way unpredictable from single gene effects (19). On the other hand, sets of germline SNVs were identified in young non-smokers with lung adenocarcinoma that underwent WES (20).

The additive effect of multiple genetic events influencing a few biochemical mechanisms was observed in patient 1, where the identified variants were mapped into the cancer-related pathways. The SNVs affected genes of the DNA repair pathways (double-strand break repair and Fanconi anemia: CHEK2, ATRIP) and cell cycle checkpoints (EME2), while the genes involved in the microdeletion were mapped into the cell cycle checkpoints (BUB1). The pathogenic variant observed in CHEK2 could explain each of the cancers observed in the patient separately due to the reports that CHEK2 can be associated with risk of melanoma (21), seminoma (22), and other cancers (23). However, each of these reported patients showed a narrower spectrum of cancer, therefore the contribution of other pathogenic variants cannot be ruled out to explain multiple cancers in a single patient. The study of this single patient with a very significant number of precocious malignant tumors thus suggests an oligogenic effect. This patient, now healthy at the age of 46, benefits from a regular clinical and imaging surveillance. However, genetic counselling for his 5 children is very difficult; so far, family testing was not proposed as patient 1 had his first diagnosis at the age of 18 (a melanoma), a clinical follow-up from the age of 13 should probably be proposed.

Patient 2 in addition to two SNVs presented a deletion of the four last exons in MSR1 gene. Germline variants in these gene have been linked to prostate cancer (24) and esophageal carcinoma (25). The importance of this gene in breast cancer and melanoma is yet to be identified. The second CNV was a complete deletion of APOBEC3B gene in patient 7. Germline deletions of this gene have been associated with breast cancer risk (26). The CNVs identified in patient 2 and patient 7 were confirmed by nanopore sequencing.

Three of the patients presented microdeletions/microduplications that could alter micro-RNA expression. However, none of the micro-RNA included in the CNVs has been associated with germline risk of cancer.

The other patients from our cohort showed limited numbers of variants or not any identified variant (patient 4). Of course, we cannot exclude mutations in non-coding areas. Other explanations could include epigenetic or environmental factors. Interestingly, there were not any variants identified in patient 4, who presented with a medulloblastoma at an early age, and a bilateral mucinous ovarian borderline tumor later in life, while her monozygotic twin sister never had a tumor, which suggested a non-genetic cause or low penetrance factors that we could not detect.

In five patients, we could analyze the tumors to search for second hits and confirm the involvement of germline variants in tumor suppressor genes. We could not identify such somatic genetic events. However, for further studies, a simultaneous analysis of the germline and tumor DNA should be done, including epigenetic analysis of the tumor DNA.

On the tumors, we investigated the mutational signatures, as they could orient the investigation toward specific oncologic mechanisms, such as DNA repair defects. We could not identify a recurrent signature in the different tumors from a single patient, except for a defective DNA mismatch repair (MMR) signature observed in the three cancers from patient 6, in which we have not identified any mutation in the MMR genes. Interestingly, several tumors showed a signature indicating exposure to chemotherapy. Although these patients were not previously exposed to chemotherapy nor radiotherapy, exposure to therapeutic or environmental DNA-damaging agents could contribute significantly to cancer risk if there is any constitutional defect in DNA-repair pathways (27).

Our study thus suggests that the simultaneous presence of multiple germline variants can confer a significant cancer risk. We also observed three families with multiple early cancers in patients carrying BRCA2 and ATM mutations (28) and other reports showed similar data (29,30). Moreover, the polygenic risk scores can identify a small population with high cancer risk, as demonstrated for breast cancer (31). Finally, rare recessive conditions could also induce a significant cancer risk as indicated for MCM9 mutations associated with primary ovarian insufficiency and cancer risk (32). Taken together these studies indicate that: i) investigations should not be limited to single gene studies in patients/families with multiple and/or early cancers; ii) In order to identify all the genetic events that could be associated with cancer genes, SNVs, CNVs, and chromosomal rearrangements including gene fusions should be tested combining different techniques such as NGS and optical genome mapping; iii) further studies are needed to investigate the respective role of genetic and epigenetic events in these patients; iv) the exact role of the observed variants and their cumulative effect should be addressed by functional studies.

The limitations of the study include the small number of participants, the challenges of correlating VUS with the disease and the limited availability of somatic samples. The limited number of participants restricts the generalizability of the findings, and makes impossible a mutation frequency analysis to identify common mutation patterns or recurring PVs. At the moment, it will be not possible to reclassify the VUS into other categories, as the information on the variants is limited, these variants are not recurrent in several patients and consequently there is not enough evidence of pathogenicity to initiate functional validation of the variants. Further studies such as dynamic variant analysis, data integration and bioinformatic analysis can provide insights into the variant role in MPCs, and discover potential biomarkers, signaling pathways, and therapeutic targets. Furthermore, it will be necessary to evaluate the effect of gene-gene, gene-environment and protein-protein interactions, and the role of genetic modifiers and environmental factors in MPCs. Indeed, it will be a real challenge to study the role of oligogenic variants in cancer predisposition. Among possibilities, it could be envisioned to inactivate several genes in cell models and study specific biochemical pathways, controlling for instance cell proliferation, apoptosis or DNA repair mechanisms. In these cell models, and in human tumors studied in parallel, proteomic and transcriptomic studies will evaluate simultaneous loss of gene or protein expression as well as protein-protein interactions. In animal models, one could cross heterozygous or homozygous knock-out animals and evaluate spontaneous or induced cancer development.