Exploring immune-related signatures for predicting immunotherapeutic responsiveness, prognosis, and diagnosis of patients with colon cancer.

The present study focused on identifying the immune-related signatures and exploring their performance in predicting the prognosis, immunotherapeutic responsiveness, and diagnosis of patients with colon cancer. Firstly, the immunotherapeutic response-related differential expressed genes (DEGs) were identified by comparing responders and non-responders from an anti-PD-L1 cohort using the edgeR R package. Then, the immunotherapeutic response related DEGs was intersected with immune-related genes (IRGs) to obtain the immunotherapeutic response and immune-related genes (IRIGs). Then, an immunotherapeutic response and immune-related risk score (IRIRScore) model consisting of 6 IRIGs was constructed using the univariable Cox regression analysis and multivariate Cox regression analysis based on the COAD cohort from the cancer genome atlas (TCGA) database, which was further validated in two independent gene expression omnibus database (GEO) datasets (GSE39582 and GSE17536) and anti-PD-L1 cohort. A nomogram with good accuracy was established based on the immune-related signatures and clinical factors (C-index = 0.75). In the training dataset and GSE39582, higher IRIRScore was significantly associated with higher TMN and advanced pathological stages. Based on the anti-PD-L1 cohort, patients who were sensitive to immunotherapy had significantly lower risk score than non-responders. Furthermore, we explored the immunotherapy-related signatures based on the training dataset. Kaplan-Meier curve revealed a high level of T cells regulatory (Tregs) was significantly related to poor overall survival (OS), while a high level of T cells CD4 memory resting was significantly related to better OS. Besides, the TMB value of patients in the high-risk group was significantly higher than those in a low-risk group. Moreover, patients in the high-risk group had significantly higher expression levels of immune checkpoint inhibitors. In addition, the immune-related signatures were applied to establish prediction models using the random forest algorithm. Among them, and revealed excellent diagnostic predictive performance (AUC >0.8). In conclusion, the current findings provide new insights into immune-related immunotherapeutic responsiveness, prognosis, and diagnosis of colon cancer.