Tumor histone H3 lysine 18 lactylation for the prediction of immunotherapy efficacy in patients with advanced non–small cell lung cancer.

e20676 Background: Clinical benefit from immune checkpoint inhibitors (ICIs) in advanced non–small cell lung cancer (NSCLC) remains heterogeneous, and current biomarkers such as PD-L1 and immune infiltration incompletely capture metabolism-driven immune suppression. Lactate metabolism may promote an immunosuppressive tumor microenvironment, partly through histone lactylation. We evaluated the clinical relevance of a proposed metabolism–epigenetics axis linking baseline serum lactate dehydrogenase (LDH) to tumor histone H3 lysine 18 lactylation (H3K18la) in patients receiving first-line ICI-based therapy. Methods: We retrospectively enrolled 52 patients with stage IIIB–IV NSCLC treated with first-line ICI monotherapy or ICI-based combination therapy (January 2019–December 2024). Baseline serum LDH was collected prior to treatment initiation. Tumor H3K18la and CD8⁺ tumor-infiltrating lymphocytes (TILs) were assessed by immunohistochemistry; H3K18la was scored using a semiquantitative immunoreactive scoring system (0–12). X-tile identified an OS-driven LDH cutoff of 319.0 U/L. Clinical endpoints included progression-free survival (PFS), overall survival (OS), and milestone outcomes (1-year PFS/OS). Survival analyses used Kaplan–Meier and Cox models; predictive performance was assessed by ROC/AUC for milestone endpoints. Results: Baseline LDH positively correlated with tumor H3K18la expression (p < 0.01). LDH-high (≥319.0 U/L) patients had inferior PFS (p = 0.007) and OS (p = 0.019); LDH predicted shorter PFS in univariable analysis (HR 2.91, 95% CI 1.29–6.54) with attenuated significance after multivariable adjustment. Low tumor H3K18la was associated with improved PFS (p < 0.0001) and OS (p = 0.039) and remained independently associated with PFS in multivariable models (HR 0.14, 95% CI 0.04–0.47; p = 0.001). Joint stratification by H3K18la and CD8⁺ TIL density defined distinct immunologic phenotypes with markedly different PFS (log-rank p < 0.0001); among CD8-high tumors, high H3K18la still conferred worse PFS. For predicting 1-year PFS, H3K18la showed higher AUC than PD-L1 TPS and CD8⁺ TIL density (0.747 vs 0.703 vs 0.658); the difference between H3K18la and PD-L1 was not statistically significant. Conclusions: Tumor H3K18la is a clinically relevant, metabolism-linked biomarker associated with ICI outcomes in advanced NSCLC and may complement conventional immunophenotypic markers. These findings support a model in which systemic lactate burden (LDH) relates to epigenetic remodeling (H3K18la) and immune resistance. Prospective, multicenter validation and harmonized assessment of H3K18la are warranted.