Integrative bulk and single-cell transcriptomics identify a chemoresistance-related risk model for recurrence prediction in colorectal cancer

Chemoresistance remains a major obstacle to improving outcomes in colorectal cancer (CRC), particularly among patients receiving 5-fluorouracil (5-FU)–based therapy. Here, we developed and validated a drug resistance–related gene signature to predict recurrence and treatment response in CRC. In the GSE190826 cohort, patients without pathological complete response (non-pCR) exhibited significantly poorer tumor-free recurrence survival. Using differential expression analysis in TCGA-READ, we identified 4,277 differentially expressed genes and constructed a weighted gene co-expression network (WGCNA) to define clinically relevant modules. Genes from modules with strong clinical correlations were further screened using logistic regression and univariate Cox regression, yielding six prognostic chemoresistance-related genes. LASSO–Cox regression subsequently refined these to a four-gene signature (ENHO, BLACAT1, LEMD1 and ASNS), which was used to establish a READ drug resistance–related gene (RDRG) risk score. High-risk patients consistently showed poorer recurrence-free survival in the TCGA-READ training and test sets, as well as in external cohorts (TCGA-COAD and GSE17536), with robust predictive performance (3-year AUC, 0.766; 5-year AUC, 0.714 in TCGA-READ). Multivariate Cox analyses confirmed the risk score as an independent prognostic factor, and a nomogram incorporating disease stage further improved clinical utility. The RDRG score also demonstrated moderate ability to predict chemotherapy response across multiple datasets. Single-cell RNA-seq analysis revealed that malignant cells contributed the highest risk scores, with enrichment of EMT, KRAS signaling, coagulation and hypoxia-related programs in high-risk tumor cells. Notably, LEMD1 emerged as a key contributor and was broadly upregulated across multiple cancer types. Functional assays further showed that LEMD1 knockdown suppressed CRC cell migration and increased sensitivity to 5-FU. Collectively, the RDRG risk score provides a practical model for predicting CRC recurrence and chemotherapeutic response, and identifies LEMD1 as a potential therapeutic target.