Immune-related gene set prediction model contributes to the evaluation of prognosis in lung adenocarcinoma patients.

Abstract

e20550 Background: In recent years, rapid progress of immunotherapy (I/O therapy) development brings new hope to LUAD patients. The subtypes and/or quantality of TIL-IC cells were closely related to the prognosis and to the I/O therapy’s response in patients with a variety of solid tumors. In order to stratify high-risk and low-risk LUAD patients, we herein established a prognostic prediction model based on the expression of large-scale immune infiltration genes in LUAD patients. Methods: RNA sequencing data and relevant clinical information of LUAD patients ( n = 501) were obtained from the TCGA cohort. Stromal and immune scores were calculated using ESTIMATE algorithm. The differentially expressed immune-related genes (IRGs) were identified using the “DESeq2” package in R platform. Patients were randomly divided into training set and validation set with a ratio of 8:2. Univariate and Lasso Cox regression analyses were performed to identify overall survival (OS)-related IRGs. Next, stepwise multivariate Cox regression analysis was carried out to establish an optimal risk model of immune-related prognostic gene sets in the training cohort. Kaplan-Meier curves and receiver operating characteristic (ROC) curves were used to examine the predictive performance of the model in the validation cohort. Results: A total of 643 IRGs were identified in these LUAD patients, of which 140 IRGs were significantly associated with OS by univariate Cox regression analysis (P < 0.05). Lasso Cox and stepwise multivariate Cox regression analysis identified IRGs as independent prognostic factors, including C20orf197, SPRR1B, CHRNA5, SLC14A2, RP11-284F21.9, IGLV9-49, IGF2BP1, and C5orf64. The final immune-related gene set model was as follow: risk score = −0.13 * expression level of C20orf197 + 0.074 * expression level of SPRR1B + 0.23 * expression level of CHRNA5 − 0.11 * expression level of SLC14A2 + 0.11 * expression level of RP11-284F21.9 - 0.076 * expression level of IGLV9-49 + 0.048 * expression level of IGF2BP1 - 0.42 * expression level of C5orf64. Patients were divided into high-risk group and low-risk group with the median risk score. Kaplan–Meier survival analysis showed that the high-risk group had a significantly poor prognosis, when compared to the low-risk group in the training set ( HR 2.77, 95%CI; [1.96 - 3.93], P < 0.0001). The ROC curve analysis equally revealed that the area under curves for OS in the first and fifth year were 0.82 and 0.81, respectively, in the validation cohort. Taken together, the eight IRGs risk prediction model could effectively stratify LUAD patients according to prognosis. Conclusions: We constructed a prognostic prediction model for LUAD patients, which consisted of a robust signature of eight IRGs. This model could effectively stratify high-risk and low-risk LUAD patients and help to customize personalized treatment plans.