Key Predictors and Effectors of Immune Checkpoint Inhibitor-Associated Cardiotoxicity in Lung Cancer: A Large Retrospective Cohort Study

Background Immune checkpoint inhibitor (ICI)-associated cardiotoxicity is a rare but life-threatening complication. Its risk factors and underlying mechanisms in real-world lung cancer populations remain insufficiently characterized. Methods In this single-center, retrospective cohort study, we enrolled 1,633 lung cancer patients treated with ICIs between March 2013 and March 2023. Cardiotoxicity was diagnosed per international consensus criteria. We analyzed baseline/dynamic laboratory parameters, treatment regimens, and outcomes using multivariable logistic regression, Cox models, and causal mediation analysis. Robustness was validated via Firth and penalized regression. Results Among 1,633 patients, 93 (5.7%) developed cardiotoxicity (myocarditis: 35; pericarditis: 5; arrhythmias: 53). Elevated baseline platelet count (OR = 1.818, p = 0.025) and hemoglobin level (OR = 1.735, p = 0.023) were independent predictors. Anti-angiogenic therapy significantly increased cardiotoxicity risk (OR = 1.719, p = 0.019; HR = 1.668, p = 0.019). A transient rise in platelets/hemoglobin preceding cardiotoxicity onset offered dynamic warning signals. Causal mediation analysis excluded neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) as mediators, indicating direct toxicity of anti-angiogenic agents. Data-driven cutoffs (platelets: 363×10 ⁹ /L; hemoglobin: 128 g/L) refined risk stratification. Conclusion We propose an integrative risk-stratification model incorporating baseline platelets/hemoglobin, anti-angiogenic therapy, and dynamic parameter changes. This clinically feasible framework may enable early identification of high-risk patients, optimizing cardiotoxicity surveillance in ICI-treated lung cancer populations.