RESUMEN
Stage II colon cancer (CC), although diagnosed early, accounts for 16% of CC deaths. Predictors of recurrence risk could mitigate this but are currently lacking. By using a DNA methylation-based clinical screening in real-world (n = 383) and in TCGA-derived cohorts of stage II CC (n = 134), we have devised a novel 40 CpG site-based classifier that can segregate stage II CC into four previously undescribed disease sub-classes that are characterised by distinct molecular features, including activation of MYC/E2F-dependant proliferation signatures. By multivariate analyses, hypermethylation of 2 CpG sites at genes CDH17 and LRP2, respectively, was found to independently confer either significantly increased (CDH17; p-value, 0.0203) or reduced (LRP2; p-value, 0.0047) risk of CC recurrence. Functional enrichment and immune cell infiltration analyses, on RNAseq data from the TCGA cohort, revealed cases with hypermethylation at CDH17 to be enriched for KRAS, epithelial-mesenchymal transition and inflammatory functions (via IL2/STAT5), associated with infiltration by 'exhausted' T cells. By contrast, LRP2 hypermethylated cases showed enrichment for mTORC1, DNA repair pathways and activated B cell signatures. These findings will be of value for improving personalised care paths and treatment in stage II CC patients.
RESUMEN
The heterogeneity of colon cancers and their reactions presents both a challenge and promise for personalized medicine. The challenge is to develop effective biologically personalized therapeutics guided by predictive and prognostic biomarkers. Presently, there are several classes of candidate biomarkers, including genomic probes, inhibitory RNAs, assays for immunity dysfunction and, not to be forgotten, specific histopathologic and histochemical features. To develop effective therapeutics, candidate biomarkers must be qualified and validated in comparable independent cohorts, no small undertaking. This process and subsequent deployment in clinical practice involves not only the strong association of the biomarker with the treatment but also careful attention to the prosaic aspects of representative tumor site selection, obtaining a fully adequate sample which is preserved and prepared to optimize high quality analysis. In the future, the clinical utility of biomarker analytical results will benefit from associated clinical and basic science data with the assistance of artificial intelligence techniques. By application of an individualized, selected suite of biomarkers, comprehensively interpreted, individualized, more effective and less toxic therapy for colon cancer will be enabled, thereby fulfilling the promise of personalized medicine.