Identification and validation of significant gene mutations to predict clinical benefit of immune checkpoint inhibitors in lung adenocarcinoma

Objective: Immune checkpoint inhibitors (ICI) has achieved remarkable clinical benefit in advanced lung adenocarcinoma (LUAD). However, effective clinical use of ICI agents is encumbered by the high rate of innate resistance. The aim of our research is to identify significant gene mutations which can predict clinical benefit of immune checkpoint inhibitors in LUAD. Methods: The “mafComapre” function of “MafTools” package was used to screen the differentially mutated genes between durable clinical benefit (DCB) group and no durable clinical benefit (NDB) group based on the somatic mutation data from NSCLc_PD1_mSK_2018. Machine learning was performed to select significantly mutated genes to accurately classify patients into DCB group and NDB group. A nomogram model was constructed based on the significantly mutated genes to predict the susceptibility of patients to ICI. Finally, we explored the correlation between two classifications of immune cell infiltration, PD-1 and PD-L1 expression, tumor mutational burden (TMB) and prognosis. Results: Through utilize machine learning, 6 significantly mutated genes were obtained from 8 differentially mutated genes and used to accurately classify patients into DCB group and NDB group. The DCA curve and clinical impact curve revealed that the patients can benefit from the decisions made based on the nomogram model. Patients highly sensitive to ICI have elevated immune activity, higher expression of PD-1 and PD-L1, increased TMB, and well prognosis if they accept ICI treatment. Conclusions: Our research selected 6 significantly mutated genes that can predict clinical benefit of ICI in LUAD patients.