Comprehensive profiling of pathogenic germline large genomic rearrangements in a pan-cancer analysis.

The presence of large genomic rearrangements (LGRs) has been heavily investigated in breast and ovarian cancer. However, correlations between LGRs and cancer types beyond these two have not been extensively profiled, likely due to the highly inefficient methods of detecting these types of alterations. This study utilized next-generation sequencing (NGS) to analyze and classify the germline LGR profile in 17,025 cancer patients across 22 cancer types. We characterized newly identified LGRs based on predicted pathogenicity and took a closer look at genes that acquire both germline and somatic mutations within our samples. The detection method for LGRs was validated using droplet digital polymerase chain reaction (ddPCR) assay of commonly investigated LGR genes. In total, 15,659 samples from across 22 cancer types were retained for analysis after filtering. We observed that, in our cohort, the cancer types with the highest proportion of germline LGRs were ovarian cancer (4.7%), renal cell carcinoma (2.5%), breast cancer (2%), glioma (1.8%) and thyroid carcinoma (1.8%). Annotation of detected germline variants revealed several genes - MSH2, FANCA and PMS2 - that contain novel LGRs. We observed co-occurrences between germline LGRs in MSH2 and somatic single nucleotide variants/insertion and deletions (SNVs/InDels) in BRCA2, KTM2B, KDM5A, CHD8 and HNF1A. Furthermore, our analysis showed that samples with pathogenic and likely pathogenic germline LGRs tended to also have higher mutational burden, chromosomal instability, and microsatellite instability ratio compared to samples with pathogenic germline SNVs/InDels. In this study, we demonstrated the prevalence of pathogenic germline LGRs beyond breast and ovarian cancer. The profiles of these pathogenic or likely-pathogenic alterations will fuel further investigations and highlight new understanding of LGRs across multiple cancer types.