Prediction of prognosis and immunotherapy efficacy based on metabolic landscape in lung adenocarcinoma by bulk, single-cell RNA sequencing and Mendelian randomization analyses

Abstract

Immunotherapy has been a remarkable clinical advancement in cancer treatment, but only a few patients benefit from it. Metabolic reprogramming is tightly associated with immunotherapy efficacy and clinical outcomes. However, comprehensively analyzing their relationship is still lacking in lung adenocarcinoma (LUAD). Herein, we evaluated 84 metabolic pathways in TCGA-LUAD by ssGSEA. A matrix of metabolic pathway pairs was generated and a metabolic pathway-pair score (MPPS) model was established by univariable, LASSO, multivariable Cox regression analyses. The differences of metabolic reprogramming, tumor microenvironment (TME), tumor mutation burden and drug sensitivity in different MPPS groups were further explored. WGCNA and 117 machine learning algorithms were performed to identify MPPS-related genes. Single-cell RNA sequencing and in vitro experiments were used to explore the role of C1QTNF6 on TME. The results showed MPPS model accurately predicted prognosis and immunotherapy efficacy of LUAD patients regardless of sequencing platforms. High-MPPS group had worse prognosis, immunotherapy efficacy and lower immune cells infiltration, immune-related genes expression and cancer-immunity cycle scores than low-MPPS group. Seven MPPS-related genes were identified, of which C1QTNF6 was mainly expressed in fibroblasts. High C1QTNF6 expression in fibroblasts was associated with more infiltration of M2 macrophage, Treg cells and less infiltration of NK cells, memory CD8+ T cells. In vitro experiments validated silencing C1QTNF6 in fibroblasts could inhibit M2 macrophage polarization and migration. The study depicted the metabolic landscape of LUAD and constructed a MPPS model to accurately predict prognosis and immunotherapy efficacy. C1QTNF6 was a promising target to regulate M2 macrophage polarization and migration.