Development and validation of a CAF-related signature for prognosis and therapy response in colorectal cancer: new insights on HSPB1

Abstract Colorectal cancer (CRC) is a globally prevalent malignancy with high mortality rates. Cancer-associated fibroblasts (CAFs) are crucial in CRC progression and therapeutic response. This study systematically screened 22 CAF-related prognostic genes using single-cell and spatial transcriptomics analysis. By integrating 101 combinations of 10 machine learning algorithms, we developed and validated a comprehensive predictive model (CRPS) based on large-scale public and in-house datasets (1,541 patients in total), which exhibited superior prognostic predictability compared to 58 existing CRC prognostic models. CRPS score not only effectively evaluates biological functions, immune infiltration, and gene mutation levels, but also serves as a valuable tool for predicting immunotherapy efficacy in various cohorts (478 patients in total). In-house single-cell and spatial transcriptomics data, microarray cohort analysis, and experimental validation revealed that model key gene HSPB1 is closely associated with malignant transformation and subtype conversion of CAFs. In vitro and in vivo experiments further demonstrated that HSPB1 -overexpressing CAFs enhance tumor cell malignancy, underscoring the therapeutic promise of targeting the HSPB1 –CAF axis in CRC.