Development and validation of an inflammatory response-related gene and clinical factor-based signature for...

Gastric cancer (GC) is an aggressive disease that requires prognostic tools to aid in clinical management. The prognostic power of clinical features is unsatisfactory, which might be improved by combining mRNA-based signatures. Inflammatory response is widely associated with cancer development and treatment response. It is worth exploring the prognostic performance of inflammatory-related genes plus clinical factors in GC.

An 11-gene signature was trained using the least absolute shrinkage and selection operator (LASSO) based on the messenger RNA (mRNA) and overall survival (OS) data of The Cancer Genome Atlas-stomach adenocarcinoma (TCGA-STAD) cohort. A nomogram was established using the signature and clinical factors with a significant linkage with OS and was validated in 3 independent cohorts (GSE15419, GSE13861, and GSE66229) via calculating the area under the receiver operator characteristic curve (AUC). The association between the signature and immunotherapy efficacy was explored in the ERP107734 cohort.

A high risk score was associated with shorter OS in both the training and the validation sets (the AUC for 1-, 3-, 5-year in TCGA-STAD cohort: 0.691, 0.644, and 0.707; GSE15459: 0.602, 0.602, and 0.650; GSE13861: 0.648, 0.611, and 0.647; GSE66229: 0.661, 0.630, and 0.610). Its prognostic power was improved by combining clinical factors including age, sex, and tumor stage (the AUC for 1-, 3-, 5-year in TCGA-STAD cohort: 0.759, 0.706, and 0.742; GSE15459: 0.773, 0.786, and 0.803; GSE13861: 0.749, 0.881, and 0.795; GSE66229: 0.773, 0.735, and 0.722). Moreover, a low-risk score was associated with a favorable response to pembrolizumab monotherapy in the advanced setting (AUC =0.755, P=0.010).

In GCs, the inflammatory response-related gene-based signature was related to immunotherapy efficacy, and its risk score plus clinical features yielded robust prognostic power. With prospective validation, this model may improve the management of GC by enabling risk stratification and the prediction of response to immunotherapy.