The Core Hypothesis Mitochondria control cellular energy production, reactive oxygen species (ROS) generation, and apoptosis - all of which critically influence both tumor growth and anti-tumor immune responses. This study hypothesized that a mitochondrial pathway signature (MitoPS) derived from systematically cataloged mitochondrial proteins could predict immunotherapy outcomes in lung adenocarcinoma (LUAD).
MitoCarta3.0 Foundation The MitoCarta3.0 database comprehensively catalogs 1,136 human mitochondrial proteins organized into 149 functional pathways. This provided an unbiased, biology-driven starting point for the analysis, covering oxidative phosphorylation complexes, fatty acid oxidation, TCA cycle enzymes, apoptosis regulators, and mitochondrial dynamics proteins.
Multi-Cohort Validation Published in the Journal for ImmunoTherapy of Cancer in 2025, this study analyzed 14 cohorts comprising 1,682 LUAD samples total. MitoPS was benchmarked against 129 existing prognostic signatures, and NDUFB10 - the top-ranked gene - was validated in both patient cohorts and experimental mouse models.
Signature Construction Starting from MitoCarta3.0's 149 mitochondrial pathways, the researchers applied machine learning-based feature selection across 14 LUAD cohorts to identify mitochondrial pathways most consistently associated with survival and immunotherapy response. The final MitoPS was trained to distinguish immune-inflamed from immune-desert tumor phenotypes.
Benchmarking Against 129 Signatures The study rigorously compared MitoPS performance against 129 previously published LUAD prognostic signatures using concordance index, AUC for OS prediction, and association with immunotherapy response. MitoPS ranked first or in the top tier across all comparison metrics, demonstrating superior cross-cohort generalizability.
Tumor Immune Phenotyping MitoPS scores were correlated with immune cell infiltration patterns derived from CIBERSORT deconvolution and single-cell RNA sequencing data. High MitoPS scores associated with immune-desert phenotypes characterized by low CD8+ T-cell infiltration, high regulatory T-cell proportions, and suppressed interferon-gamma signaling.
NDUFB10 Identification Among all mitochondrial proteins, NDUFB10 emerged as the single most informative predictor of immune phenotype and immunotherapy response. NDUFB10 encodes a subunit of mitochondrial respiratory chain Complex I (NADH:ubiquinone oxidoreductase), the largest complex in the electron transport chain with 45 subunits.
High NDUFB10 Expression and Immune Desert LUAD tumors with high NDUFB10 expression showed markedly reduced CD8+ cytotoxic T-cell infiltration, lower PD-L1 expression, reduced tumor mutational burden, and suppressed interferon signaling - characteristics of an immune-excluded or immune-desert microenvironment. These patients had significantly shorter overall survival and poor responses to anti-PD-1/PD-L1 therapy.
Survival and Immunotherapy Response Data Across multiple independent cohorts including IMvigor210 and other immunotherapy trial datasets, high NDUFB10 expression consistently predicted non-response to immune checkpoint inhibitors. Kaplan-Meier analysis confirmed significantly worse OS in high-NDUFB10 patients treated with ICIs compared to low-NDUFB10 patients.
In Vivo Experimental Validation To test whether NDUFB10 is functionally causal rather than merely correlative, the researchers performed NDUFB10 knockdown in murine LUAD cell lines and implanted these into immunocompetent mice. NDUFB10 knockdown alone modestly reduced tumor growth, but the combination of NDUFB10 knockdown plus anti-PD-1 antibody treatment showed synergistic tumor volume reduction significantly exceeding either treatment alone.
Immune Cell Changes After NDUFB10 Knockdown Flow cytometry and immunohistochemistry of tumors from NDUFB10-knockdown plus anti-PD-1 treated mice showed markedly increased GZMB+ (granzyme B positive) CD8+ cytotoxic T cells within the tumor. This indicates that NDUFB10 knockdown reverses the immune exclusion phenotype, allowing anti-PD-1 therapy to reinvigorate anti-tumor cytotoxic T-cell responses.
Mechanistic Pathway NDUFB10 knockdown impaired Complex I-dependent NADH oxidation, altering mitochondrial membrane potential and ROS production in tumor cells. This metabolic disruption upregulated danger-associated molecular patterns (DAMPs) and increased MHC-I antigen presentation, making tumor cells more visible to cytotoxic T lymphocytes and thus more vulnerable to PD-1 blockade.
Metabolic-Immune Crosstalk Mechanism Mitochondrial electron transport chain activity in cancer cells generates a specific metabolic microenvironment rich in lactate and depleted of oxygen and nutrients. This environment directly impairs T-cell function: high lactate suppresses T-cell glycolysis and cytokine production, while hypoxia induces regulatory T-cell differentiation and M2 macrophage polarization.
NDUFB10 and ROS Signaling Complex I generates superoxide as a byproduct of electron transfer, which is normally balanced by antioxidant systems. NDUFB10 overexpression may alter ROS homeostasis in ways that activate NF-kB and HIF-1alpha transcription factors, which in turn upregulate immune checkpoint ligands and immunosuppressive cytokines like TGF-beta and IL-10.
Metabolic Reprogramming as an ICI Resistance Mechanism This study frames NDUFB10-mediated immune exclusion as a metabolic resistance mechanism to immunotherapy - distinct from classical acquired resistance through T-cell exhaustion or loss of neoantigen expression. Targeting mitochondrial metabolism may therefore represent a strategy to convert immunotherapy-resistant LUAD into treatment-responsive disease.
NDUFB10 as a Predictive Biomarker NDUFB10 expression by immunohistochemistry or RNA sequencing could be incorporated into clinical biomarker panels for immunotherapy selection. Patients with high NDUFB10 LUAD might be offered combination strategies upfront (NDUFB10 inhibitor plus ICI) rather than ICI monotherapy, which appears unlikely to provide meaningful benefit.
Therapeutic Targeting of Complex I Several Complex I inhibitors already exist, including IACS-010759 (in clinical trials for AML and solid tumors) and metformin (biguanide that weakly inhibits Complex I). Testing whether IACS-010759 or next-generation Complex I inhibitors can sensitize NDUFB10-high LUAD to anti-PD-1 therapy in clinical trials is a high-priority next step.
Broader Mitochondrial Targeting Strategy Beyond NDUFB10 and Complex I, the MitoPS framework identifies multiple mitochondrial pathways with immunotherapy predictive potential. A comprehensive mitochondria-targeted therapy approach - combining metabolic inhibitors with immune checkpoint blockade - could address the 60-70% of NSCLC patients who currently do not respond to ICI monotherapy.