A Prognostic Nomogram Combining Immune-Related Gene Signature and Clinical Factors Predicts Survival in Patients With Lung Adenocarcinoma

• Published in: Frontiers in oncology Vol. 10; p. 1300
• Main Authors: Song, Congkuan, Guo, Zixin, Yu, Donghu, Wang, Yujin, Wang, Qingwen, Dong, Zhe, Hu, Weidong
• Format: Journal Article
• Language: English
• Published: Frontiers Media S.A 06.08.2020
• Subjects: immune related gene; lung adenocarcinoma; nomogram; Oncology; prognosis; signature
• ISSN: 2234-943X; 2234-943X
• DOI: 10.3389/fonc.2020.01300

The existence of tumor heterogeneity and complex carcinogenic mechanisms in lung adenocarcinoma (LUAD) make the most commonly used TNM staging system unable to well-interpret the prognosis of patients. Using transcriptome profiling and clinical data from The Cancer Genome Atlas (TCGA) database, we constructed an immune signature based on a multivariate Cox analysis (stepwise model). We estimated the half-maximal inhibitory concentration (IC50) of chemotherapeutic drugs in patients according to the pRRophetic algorithm. Gene-set variation analysis (GSVA) was used to reveal pathway enrichment between groups. Moreover, immune microenvironment landscape was described by single-sample gene-set enrichment analysis (ssGSEA) and CIBERSORT and systematically correlated with genomic of these patients. A prognostic nomogram combining the immune signature and TNM stage to predict the prognosis was developed by multivariate Cox regression. The novel signature with four immune-related genes (MAL, MS4A1, OAS1, and WFDC2) had good robustness, which can accurately distinguish between high- and low-risk patients. Compared with low-risk patients, high-risk patients with a worse prognosis (5-year OS: 46.5 vs. 59.4%, p = 0.002) could benefit more from immunotherapy and the application of common chemotherapeutic agents such as cisplatin and paclitaxel (Wilcoxon test, all p < 0.05). There were significant differences in tumor immune microenvironment and metabolic pathways between the two groups. Additionally, the constructed nomogram had reliable predictive performance with the C-index of 0.725 (95% CI = 0.668–0.781) in the development set ( n = 500), 0.793 (95% CI = 0.728–0.858) in the internal validation set ( n = 250) and 0.679 (95% CI = 0.644–0.714) in the external validation set ( n = 442). The corresponding calibration curves also showed good consistency. To sum up, we developed an immune-related gene signature and comprehensively evaluated LUAD immune landscape and metabolic pathways. Effective differentiation of high- and low-risk patients and accurate construction of nomogram would be helpful to the development of individualized treatment strategies.