Dissecting Age-dependent Molecular Heterogeneity of Colorectal Cancer and Establishing a Probabilistic Model for Robust Prediction of High-risk Young-onset Colorectal Cancer Patients

Description

Young-onset colorectal cancer, defined as arising in people younger than age 50, is a growing threat of health worldwide, with estimated increases of incidence in a decade by about 90% and 140% in western and eastern world, respectively. Conventional colorectal cancer risk factors such as obesity, alcohol consumption and smoking have been extensively investigated, yet no compelling explanation has been identified for this epidemiological trend. More importantly, the etiology and underlying biology of young onset CRC are not understood. Although the pathogenesis of hereditary CRC syndromes has been well characterized, they only account for a small fraction of early-onset CRC, and therefore, cannot explain the rapidly increasing incidence, more aggressiveness and relatively poorer outcome. So far, no biomarker is available for effective prediction of high-risk young-onset CRC patients, and as a result, these patients were overtreated with significantly more adjuvant chemotherapies at all stages, but only received minor gain in survival. Elucidating the biological mechanisms in relation to clinical outcomes in young-onset CRC is critical for selection of patients for more optimized clinical management and individualized therapy. Colorectal cancer is a heterogeneous disease, which can be subdivided into four consensus molecular subtypes (CMSs): CMS1-MSI, CMS2-canonical, CMS3-metabolic and CMS4-mesenchymal. In our pilot study, we unveiled that young-onset CRC is also a biologically heterogeneous disease comprising the same four molecular subtypes, but the prevalence of CMSs is different from late-onset CRC. More specifically, we found that the observed poor clinical outcomes in young-onset CRC were not due to age per se, but higher prevalence of CMS4-mesenchymal subtype, known to be associated with poor relapse-free survival and higher risk of metastasis. Furthermore, we provided evidences to show that the higher prevalence of CMS1 in late-onset CRC may be potentially due to DNA hypermethylation and mutation burden progressively accumulated during aging. In this proposal, we aim to comprehensively dissect the age-dependent heterogeneity of CRC by integrative analysis of multi-omic and clinical data involving over 7000 patient samples across 9 public datasets and 2 independent in-house clinical cohorts. Our 1st objective is to perform comprehensive clinical characterizations of young-onset CRC based on a multi-center cohort study. Based on curated public datasets, our 2nd objective is to systematically delineate multi-omic molecular properties associated with each CMS in young-onset CRC. To robustly identify high-risk young-onset CRC patients, our 3rd objective is to develop a probabilistic model that integrates age-specific prevalence, molecular subtyping and subtype-specific recurrence risk prediction.