TBIO-04. Comprehensive analysis of mutational signatures in pediatric cancers

Abstract

Abstract Analysis of mutational signatures can reveal the underlying molecular processes that have caused the somatic mutations found in genomes. We performed an extensive analysis encompassing mutational signatures of single base substitution and, for the first time, signatures of small insertions/deletions in ~800 whole-genome-sequenced tumor-normal pairs from 25 molecularly defined pediatric cancer types. More than half of this cohort consists of pediatric CNS-tumors (n=416), including high- and low-grade gliomas, ependymomas, and embryonal tumors. These were subsequently compared with COSMIC v.3 signatures to identify overlap with the latest set of known mutational signatures. We identified only a small number of mutational signatures active in pediatric cancers when compared to previously analyzed adult cancers. Amongst these, SBS1 and SBS5 were present in nearly all pediatric tumors analyzed, with a significant correlation of signature activity with age at diagnosis, as expected. Further, we found SBS21 activity in a fraction of high-grade gliomas, which is the result of defective DNA mismatch repair, SBS36 in fractions of ETMR and Group4-medulloblastoma, the result of defective base excision repair, and SBS44 in Group4-medulloblastomas, caused by defective DNA mismatch repair. For these signatures, no consistent genetic alteration was identified. We report a significantly smaller proportion of pediatric tumors which show homologous-recombination repair defect signature SBS3 compared to previously published analyses. Additionally, the previous mutational signature analysis of this cohort based on COSMIC v.2 reference signatures had identified a novel substitution signature (Signature.P1), active in the brain tumors ATRT and ependymoma. Our updated results suggest that Signature.P1 is not a pediatric specific mutational signature, but a treatment-associated signature identified in a small fraction of brain tumors that were annotated as treatment-naïve. This analysis provides a systematic overview of mutational signatures in pediatric cancers, which is relevant for understanding tumor biology and future research in defining biomarkers for treatment response.