ABSTRACT
PURPOSE: Colorectal cancer (CRC) is the third most commonly diagnosed cancer worldwide. Biomarkers to aid in prognostication and treatment decisions are in high demand, and to facilitate their development, a better understanding of the underlying biology of the highly heterogeneous disease is needed. METHODS: A genome-scale alternative splicing (AS) analysis using RNA-sequencing data from primary microsatellite stable (MSS) CRCs from 127 patients was performed. Splice variant-specific expression levels of individual cancer samples were compared with the total set of samples, and a metric for a tumor sample's global amount of deviating AS was developed. This metric varied considerably across the cohort and ranged from 6 to 282 deviating AS events per tumor sample. A threshold of 45 or more deviating events was set to distinguish cancers with high (n = 44) and low (n = 83) levels of deviating AS. RESULTS: Patients with high amounts of AS deviations had significantly shorter time to relapse compared with patients with fewer deviations (P = .04). Furthermore, differential gene expression analysis revealed nine known cancer-critical genes that are significantly upregulated in samples with high amounts of deviating AS. Validation of the results in an independent cohort of MSS CRCs showed the same tendency toward shorter progression-free survival among the high-deviation group. In both cohorts, enrichment for growth factors was identified among upregulated genes associated with this phenotype. CONCLUSION: There is a large variation in the amount of deviating AS among MSS CRCs, and we provide evidence that those with high amounts of deviations represent different cancer biology.
