Perioperative cardiac safety of neoadjuvant chemo-immunotherapy: how neutral is neutral?

To the Editor, Yan et al report a bidirectional cohort study examining whether neoadjuvant chemo-immunotherapy increases postoperative cardiac adverse events in patients undergoing resection for non-small cell lung cancer (NSCLC) and show no significant increase in overall postoperative cardiac adverse events (PCAEs) compared with neoadjuvant chemotherapy alone1. As neoadjuvant immunotherapy is increasingly adopted in surgical oncology, defining its perioperative cardiovascular safety has become a clinically relevant issue. In this context, several aspects of study design and interpretation merit further consideration to clarify the boundaries of the reported safety signal. The study also aligns with the TITAN 2025 framework for transparent reporting of AI-assisted research2. First, the primary reliance on a composite PCAE endpoint may obscure heterogeneity among its constituent events. Postoperative atrial fibrillation, myocardial injury after non-cardiac surgery, myocardial infarction, and acute heart failure differ substantially in pathophysiology, timing, and prognostic significance. Aggregating these outcomes into a single binary endpoint improves statistical efficiency but assumes comparable clinical weight and biological relevance. In particular, postoperative atrial fibrillation is largely driven by perioperative stress and autonomic imbalance, whereas myocardial injury reflects myocardial vulnerability to ischemic and inflammatory stress, and infarction or heart failure represents less frequent but more definitive ischemic events3. When analyzed together, higher-frequency rhythm disturbances may dominate the composite and reduce sensitivity to more specific myocardial injury signals. Although component outcomes are reported, interpretation remains centered on the composite endpoint, leaving uncertainty as to whether neoadjuvant chemo-immunotherapy (nCIT) is neutral across distinct cardiac phenotypes or whether apparent neutrality reflects endpoint averaging rather than a uniform biological effect. Second, the analysis is conditional on completion of neoadjuvant therapy and progression to surgery, which introduces an inherent selection structure. In contemporary neoadjuvant pathways, failure to proceed to resection is often related to treatment toxicity, physiological intolerance, or clinical reassessment rather than random attrition4. These processes may correlate with cardiovascular vulnerability and inflammatory stress responses. Restricting evaluation to patients who successfully reach surgery therefore preferentially captures individuals with greater treatment tolerance, potentially attenuating observable differences between treatment groups5. From a clinical perspective, perioperative risk estimates are most informative when they reflect the full preoperative trajectory, including the likelihood and determinants of not proceeding to surgery, rather than risk conditional on arrival in the operating room. Reporting attrition patterns and adopting pathway-based analytical frameworks may better inform multidisciplinary decision-making, particularly as nCIT indications expand to older patients and those with greater comorbidity burden. Third, the absence of immune-cardiac biological markers limits translation of a negative association into individualized perioperative risk assessment. Event-based analyses cannot distinguish true biological neutrality from subclinical myocardial perturbations that are buffered by perioperative care or fall below adjudication thresholds. In the context of immune checkpoint blockade, perioperative cardiac vulnerability may depend on baseline inflammatory state, endothelial stress, and myocardial reserve, factors not captured by clinical events alone6. This limitation is highlighted by the observed age-related effect modification, which suggests potential heterogeneity but cannot be mechanistically interpreted within the current dataset. Incorporation of longitudinal cardiac biomarkers and inflammatory profiles could help determine whether nCIT is uniformly benign or associated with latent, context-dependent risk relevant to perioperative monitoring strategies. In summary, Yan et al provide important real-world evidence supporting the perioperative cardiac safety of nCIT in selected surgical candidates with NSCLC. However, the interpretation of safety depends on endpoint definition, conditioning on surgical completion, and the absence of biological resolution. Addressing these issues in future studies may improve causal interpretation and enhance translation of safety data into surgical planning, perioperative surveillance, and guideline development as neoadjuvant immunotherapy continues to evolve. Ethical approval Ethical approval was not required for this study, as it did not involve human participants, patient data, or any procedures requiring ethical clearance. Consent This study did not involve human participants, patients, or volunteers; therefore, written informed consent was not required.