The role of mitochondrial energy metabolism in drug resistance and prognosis of lung adenocarcinoma: a multi-omics and machine learning strategy for predictive and personalized therapy.

Background: As the leading histological form of lung cancer, lung adenocarcinoma (LUAD) displays considerable intratumoral heterogeneity, frequent therapeutic resistance, and an unfavorable clinical outcome. Although rewiring of mitochondrial energy metabolism is known to drive tumor progression and treatment failure, a comprehensive understanding of its dual role in LUAD drug resistance and prognosis has yet to be established. Here, we built a robust predictive signature that integrates mitochondrial metabolism with drug resistance through multi-omics integration and machine learning frameworks. Methods: We explored single-cell RNA sequencing profiles together with TCGA-LUAD transcriptomic data. Weighted gene co-expression network analysis (WGCNA) was applied to extract gene modules linked to mitochondrial-related genes (MRGs) and drug resistance-related genes (DRGs). From these, a five-gene (KLF4, KLF10, CAT, ALDOA, HLA-DRA) prognostic classifier, designated MDrisk, was formulated using LASSO-Cox regression and 101 combinations of 10 machine learning algorithms. Results: The MDrisk model demonstrated reliable and precise prognostic capacity across training, internal test, and external GEO cohorts, serving as an independent risk factor. Elevated MDrisk scores correlated with an immunosuppressive microenvironment, higher tumor mutational burden, distinct copy-number alteration profiles, and decreased drug sensitivity in computational predictions. Conclusion: The MDrisk signature derived from mitochondrial energy metabolism and drug resistance may be useful for distinguishing prognosis, immune contexture, and computationally inferred drug susceptibility in LUAD. It may offer a tool for further exploration of individualized therapy and sheds light on the interplay between metabolic dysregulation and antitumor immunity.