Multi-omics and machine learning reveal DPPC as a key contributor to colorectal cancer progression and tumor immune microenvironment remodeling

Although previous studies have linked colorectal cancer (CRC) with lipid metabolism and inflammatory signaling, the specific roles of phosphatidylcholine metabolites and their interactions with inflammatory cytokines in the tumor microenvironment remain poorly understood. This study utilizes an integrated multi-omics analysis approach. We performed a two-sample Mendelian randomization analysis using data from 1400 metabolites and 91 inflammatory cytokines to investigate their associations with CRC, followed by experimental validation of the findings. Single-cell transcriptomics revealed metabolic state differences, while machine learning constructed a predictive model. SHAP analysis interpreted the model, with spatial transcriptomics validating key findings. The phosphatidylcholine metabolite DPPC was identified as causally associated with CRC risk. Our results demonstrated DPPC promotes tumor progression by inhibiting TNFSF14 secretion. Our DPPC-based model effectively predicted CRC progression, with SHAP analysis identifying ARL8A, MTUS1, and TMEM184A as key contributors. These findings were validated spatially and translated into a clinical nomogram for prognosis and immunotherapy guidance. In summary, this study highlights the significance of DPPC-mediated regulation within the tumor microenvironment in predicting CRC progression and guiding potential therapeutic strategies.