Platelet-Related Signature in Colorectal Cancer: Prognosis and Therapy Guidance.

Platelets are key components of the tumor microenvironment that contribute to colorectal cancer (CRC) progression. However, effective platelet-related prognostic tools for CRC risk stratification and prognostic assessment are lacking. Therefore, we aimed to develop a platelet-related prognostic model to improve risk stratification and prognostic evaluation. A tumor-cell-expressed platelet-related transcriptional signature, termed platelet-related risk score (PLRS), was established using LASSO and Cox regression based on TCGA-COADREAD, GSE39582 and GSE183635. After strict screening for intact survival information, 573, 531 and 85 tumor samples were retained correspondingly. Patients were stratified into high- and low-PLRS groups according to the median cutoff value of PLRS derived from the training cohort. A single-cell dataset GSE178341 was utilized for single-cell transcriptomic validation. All analyses were performed in R. Wilcoxon test, Kaplan-Meier method, log-rank test, multivariate Cox regression and ROC curves were applied with two-sided p < 0.05. qRT-PCR was conducted for hub gene validation. The PLRS model involved six prognostic genes (TIMP1, TTYH3, GPX4, PLCB4, CA2, and LITAF). The high- and low-PLRS groups showed significant differences in disease stage and survival in TCGA datasets, which was confirmed in the GEO dataset (GSE39582). The multivariate analysis established the PLRS as an independent prognostic factor. Single-cell analyses revealed distinct transcriptional trajectories, immune microenvironments, and communication patterns between the PLRS groups. Stromal, immune, and ESTIMATE scores positively correlated with PLRS, whereas tumor purity was inversely correlated. Exploratory drug sensitivity analysis using GDSC and CCLE cell line datasets suggested potential associations between PLRS and multiple therapeutic agents, warranting validation in clinical cohorts. qRT-PCR in RKO cells compared with the CCD-18Co fibroblast control partially confirmed the expression trends of four PLRS genes (TIMP1, GPX4, LITAF, and PLCB4), whereas TTYH3 showed no significant difference and CA2 was not tested due to primer-design failure. Validation in normal colonic epithelial cells is warranted. The PLRS model demonstrated statistically significant prognostic value and may serve as a complementary tool for CRC risk stratification, providing a basis for further investigation into platelet-related transcriptional programs.