Background: The incidence of early-onset colorectal cancer (EOCRC; diagnosed <50 years of age) is increasing in Australia and globally, the cause of which is unknown. The aim was to characterise the patterns of somatic mutations, measured as tumour mutational signatures, to understand the processes driving EOCRC development.
Methods: Non-hereditary, mismatch repair proficient EOCRCs (n=236) from the ANGELS study (n=149) and the Colon Cancer Family Registry (n=97) underwent tumour and matched germline whole exome sequencing. Single base substitution (SBS), indel (ID) and doublet (DBS) signatures were calculated with COSMIC v3.4 definitions. Dimensionality reduction with UMAP was applied using all SBS, ID and DBS calculated signatures as input features to identify clusters of EOCRC with similar mutational processes.
Results: Of the 236 EOCRCs (mean age at diagnosis 38.5±7.3 years, 56.4% females), 36.4% were diagnosed between 18-35yrs, 44.5% between 36-45yrs and 19.1% between 46-55yrs with 26.7% located in the proximal colon, 37.3% distal and 33.5% in the rectum. Dimensionality reduction revealed distinct clusters including EOCRCs with Homologous Recombination Deficiency (HRD)-associated signatures SBS3 and ID6, and EOCRCs with SBS10a and SBS10b corresponding to somatic mutations in the exonuclease domain of POLE gene. A unique cluster representing 14% of EOCRCs had both SBS89 and DBS8 signatures which do not currently have a known aetiology. EOCRCs with both SBS89 and DBS8 were more common in the proximal colon compared with distal and rectal EOCRCs (23% vs 11% p=0.027), were more prevalent in the youngest age at diagnosis group (18-35years 26% vs 7%, p=6x10-5) and were more likely to be born after 1980 (p=9.5x10-5).
Discussion: Tumour mutational signature profiling identified clusters of EOCRCs with distinct aetiologies providing novel insights into the causes of EOCRC. EOCRCs with mutational processes related to SBS89 and DBS8 had the youngest age at diagnosis and displayed evidence of a birth cohort effect.