An antigen processing and presentation signature for prognostic evaluation and immunotherapy selection in advanced gastric cancer

• Published in: Frontiers in immunology Vol. 13; p. 992060
• Main Authors: Wang, Ke-wei, Wang, Mei-dan, Li, Zi-xi, Hu, Ben-shun, Wu, Jun-jie, Yuan, Zheng-dong, Wu, Xiao-long, Yuan, Qin-fang, Yuan, Feng-lai
• Format: Journal Article
• Language: English
• Published: Frontiers Media S.A 14.10.2022
• Subjects: antigen processing and presentation; immune checkpoint inhibitor; Immunology; immunotherapy; survival; tumor microenvironment
• ISSN: 1664-3224; 1664-3224
• DOI: 10.3389/fimmu.2022.992060

The aim of the study was to propose a signature based on genes associated with antigen processing and presentation (APscore) to predict prognosis and response to immune checkpoint inhibitors (ICIs) in advanced gastric cancer (aGC).ObjectiveThe aim of the study was to propose a signature based on genes associated with antigen processing and presentation (APscore) to predict prognosis and response to immune checkpoint inhibitors (ICIs) in advanced gastric cancer (aGC).How antigen presentation-related genes affected the immunotherapy response and whether they could predict the clinical outcomes of the immune checkpoint inhibitor (ICI) in aGC remain largely unknown.BackgroundHow antigen presentation-related genes affected the immunotherapy response and whether they could predict the clinical outcomes of the immune checkpoint inhibitor (ICI) in aGC remain largely unknown.In this study, an aGC cohort (Kim cohort, RNAseq, N=45) treated by ICIs, and 467 aGC patients from seven cohorts were conducted to investigate the value of the APscore predicting the prognosis and response to ICIs. Subsequently, the associations of the APscore with the tumor microenvironment (TME), molecular characteristics, clinical features, and somatic mutation variants in aGC were assessed. The area under the receiver operating characteristic curve (AUROC) of the APscore was analyzed to estimate response to ICIs. Cox regression or Log-rank test was used to estimate the prognosis of aGC patients.MethodsIn this study, an aGC cohort (Kim cohort, RNAseq, N=45) treated by ICIs, and 467 aGC patients from seven cohorts were conducted to investigate the value of the APscore predicting the prognosis and response to ICIs. Subsequently, the associations of the APscore with the tumor microenvironment (TME), molecular characteristics, clinical features, and somatic mutation variants in aGC were assessed. The area under the receiver operating characteristic curve (AUROC) of the APscore was analyzed to estimate response to ICIs. Cox regression or Log-rank test was used to estimate the prognosis of aGC patients.The APscore constructed by principal component analysis algorithms was an effective predictive biomarker of the response to ICIs in the Kim cohort and 467 aGC patients (Kim: AUC =0.85, 95% CI: 0.69-1.00; 467 aGC: AUC =0.69, 95% CI: 0.63-0.74). The APscore also was a prognostic biomarker in 467 aGC patients (HR=1.73, 95% CI: 1.21-2.46). Inhibitory immunity, decreased TMB and low stromal scores were observed in the high APscore group, while activation of immunity, increased TMB, and high stromal scores were observed in the low APscore group. Next, we evaluated the value of several central genes in predicting the prognosis and response to ICIs in aGC patients, and verified them using immunogenic, transcriptomic, genomic, and multi-omics methods. Lastly, a predictive model built successfully discriminated patients with vs. without immunotherapy response and predicted the survival of aGC patients.ResultsThe APscore constructed by principal component analysis algorithms was an effective predictive biomarker of the response to ICIs in the Kim cohort and 467 aGC patients (Kim: AUC =0.85, 95% CI: 0.69-1.00; 467 aGC: AUC =0.69, 95% CI: 0.63-0.74). The APscore also was a prognostic biomarker in 467 aGC patients (HR=1.73, 95% CI: 1.21-2.46). Inhibitory immunity, decreased TMB and low stromal scores were observed in the high APscore group, while activation of immunity, increased TMB, and high stromal scores were observed in the low APscore group. Next, we evaluated the value of several central genes in predicting the prognosis and response to ICIs in aGC patients, and verified them using immunogenic, transcriptomic, genomic, and multi-omics methods. Lastly, a predictive model built successfully discriminated patients with vs. without immunotherapy response and predicted the survival of aGC patients.The APscore was a new biomarker for identifying high-risk aGC patients and patients with responses to ICIs. Exploration of the APscore and hub genes in multi-omics GC data may guide treatment decisions.ConclusionsThe APscore was a new biomarker for identifying high-risk aGC patients and patients with responses to ICIs. Exploration of the APscore and hub genes in multi-omics GC data may guide treatment decisions.