Integrated single-cell and bulk transcriptomic analysis leverages liver metastasis-related genes to develop a prognostic model for colorectal cancer patients
Journal:
bioRxiv
Published Date:
Mar 30, 2026
Abstract
Based on single-cell RNA sequencing data, differentially expressed genes (LMR DEGs) between colorectal cancer liver metastasis epithelium and primary colorectal cancer epithelium show potential as novel biomarkers for colorectal cancer prognosis. This study first utilized single-cell RNA sequencing data to characterize the cellular landscape of primary colorectal cancer and liver metastasis, identifying LMR DEGs. Prognostic LMR DEGs were then screened through bulk RNA sequencing data. By comparing the C-index of various machine learning algorithm combinations, a randomized survival forest model combined with ridge regression was ultimately selected to construct a 15-gene scoring system (LMR score). In external validation, the 1-year and 5-year AUC values of the LMR score outperformed both AJCC staging and other scoring systems developed from similar datasets. Furthermore, the LMR score demonstrated close associations with key factors influencing colorectal cancer outcomes, such as immune infiltration. The findings suggest that the LMR score may serve as a reliable novel biomarker for predicting prognosis in patients with colorectal cancer.
