Recurrence associations of a colorectal stemness gene set adapted to bulk expression in three cohorts

Abstract

This study assessed whether a fixed bulk adaptation of a published colorectal cancer stemness gene set shows consistent recurrence-related associations in stage II–III colon cancer. This exploratory analysis included 701 patients and 185 events from GSE39582, GSE17536 and The Cancer Genome Atlas colon adenocarcinoma cohort (TCGA-COAD). A fixed 45-gene stemness score (SCS45) and ninemarker comparator used within-sample expression ranks. Separate Cox models adjusted for age, sex and stage; six primary comparisons underwent Benjamini-Hochberg correction. SCS45 hazard ratios per one within-cohort standard deviation were 1.023 (95% confidence interval 0.853–1.228), 0.678 (0.484–0.949) and 1.169 (0.650–2.103), respectively. Associations were not consistent across cohorts, and no primary comparison met the adjusted 0.05 threshold. The nominal GSE17536 association (P = 0.024) had an adjusted P value of 0.142. Alternative scoring changed some estimates. These cohorts provide no consistent evidence of recurrence prognostic value for this bulk adaptation. Endpoint differences and limited events constrain interpretation; the findings neither establish absence of an association nor invalidate the original single-cell assay.