Abstract
BACKGROUND: Cancer-associated fibroblasts (CAFs) play a crucial role in the progression of colorectal cancer (CRC). However, the impact of CAF subpopulation trajectory differentiation on CRC remains unclear. METHODS: In this study, we first explored the trajectory differences of CAFs subpopulations using bulk and integrated single-cell sequencing data, and then performed consensus clustering of CRC samples based on the trajectory differential genes of CAFs subpopulations. Subsequently, we analyzed the heterogeneity of CRC subtypes using bioinformatics. Finally, we constructed relevant prognostic signature using machine learning and validated them using spatial transcriptomic data. RESULTS: Based on the differential genes of CAFs subpopulation trajectory differentiation, we identified two CRC subtypes (C1 and C2) in this study. Compared to C1, C2 exhibited worse prognosis, higher immune evasion microenvironment and high CAF characteristics. C1 was primarily associated with metabolism, while C2 was primarily associated with cell metastasis and immune regulation. By combining 101 combinations of 10 machine learning algorithms, we developed a High-CAF risk signatures (HCAFRS) based on the C2 characteristic gene. HCAFRS was an independent prognostic factor for CRC and, when combined with clinical parameters, significantly predicted the overall survival of CRC patients. HCAFRS was closely associated with epithelial-mesenchymal transition, angiogenesis, and hypoxia. Furthermore, the risk score of HCAFRS was mainly derived from CAFs and was validated in the spatial transcriptomic data. CONCLUSION: In conclusion, HCAFRS has the potential to serve as a promising prognostic indicator for CRC, improving the quality of life for CRC patients.