An Immune-Related Gene Signature for Predicting Neoadjuvant Chemoradiotherapy Efficacy in Rectal Carcinoma

BACKGROUND: Locally advanced rectal cancers (LARC) show a highly variable response to neoadjuvant chemoradiotherapy (nCRT), and the impact of the tumor immune response in this process is poorly understood. This study aimed to characterize the immune-related gene expression profiles (GEP), pathways, and cell types associated with response or resistance to neoadjuvant chemoradiotherapy. METHODS: The transcriptomic and clinical data of Rectal carcinoma from the Gene Expression Omnibus database and Immune-related genes (IRGs) from ImmPort were downloaded to identify the differentially expressed immune-related genes (DEIRGs) between responder and non-responder to neoadjuvant chemoradiotherapy. Gene set enrichment analyses were performed to uncover significantly enriched GO terms and KEGG pathways. Immune cell infiltration was estimated from RNA-sequencing data using ImmuCellAI. Afterward, we constructed an immune-related gene-based predictive model (IRGPM) by Support Vector Machine and validated it in an external cohort. RESULT: A 15-gene signature (HLA-DPB1, HLA-DQA1, CXCL9, CXCL10, TAP2, INHBB, BMP2, CD74, IL33, CCL11, CXCL11, DEFB1, HLA-DPA1, CCN3, STAT1) was identified as DEIRGs and found to be significantly associated with nCRT outcomes. Gene set enrichment analyses indicated that the 15 genes play active roles in inflammation-related biological processes. In addition, ImmuCellAI revealed that CD4 naive T cells, Tex, Th1 were significantly up-regulated (p=0.035, p=0.02, p=0.0086, respectively), while Tfh were significantly down-regulated (p=0.015) in responder subgroup. Finally, a novel predictive model was developed by SVM based on DEIRGs with an AUC of 80% (internal validation) and 73.5% (external validation). CONCLUSION: Our team conducted a genomic study of the relationship between gene expression profile and response to nCRT in LARC. Our data suggested that the DEIRGs signature could help predict the efficacy of nCRT. And a DEIRGs‐based SVM model was developed to monitor the outcomes of nCRT in LARC.