Baseline cardiac biomarkers and inflammatory indices for new-onset ECG abnormalities and survival in patients receiving immune checkpoint inhibitors: a single-center exploratory study

Routine inflammatory and cardiac biomarkers may help risk-stratify patients receiving immune checkpoint inhibitors (ICIs). The ECG monitoring endpoint and oncologic survival outcomes are distinct. We examined whether baseline cardiac biomarkers were associated with adjudicated new-onset electrocardiographic (ECG) abnormalities after ICI initiation and whether inflammatory and cardiac biomarkers were associated with overall survival (OS) and progression-free survival (PFS). This single-center retrospective cohort included 162 patients with advanced gastrointestinal or lung cancers who received at least two cycles of ICI monotherapy or ICI-based combination therapy. The ECG monitoring endpoint was any adjudicated new-onset ECG abnormality. Baseline ECGs were used to identify pre-existing findings, and ICI initiation was the time origin for the 12-month Cox, Kaplan–Meier, and score analyses. Patients without an event were censored at the earliest of the last evaluable ECG date, death, or 365 days after ICI initiation. ECGs were independently reviewed by two cardiologists, and disagreements were resolved by consensus. Logistic regression and Cox models were used for ECG, OS, and PFS analyses; age-stratified Cox sensitivity analyses addressed non-proportionality. New-onset ECG abnormalities occurred in 86 patients (53.1%). Study-defined Grade ≥ 2 ECG abnormalities occurred in 25 patients (15.4%). Binary normal/abnormal classification was concordant in all 162 patients, and Cohen’s kappa for the specific primary ECG type was 0.975. In the adjusted 12-month Cox model, cTnI > 0.02 ng/mL (HR, 2.00; 95% CI, 1.26–3.18; P = 0.003) and CK-MB > 13 U/L (HR, 2.54; 95% CI, 1.60–4.01; P < 0.001) were associated with shorter time to the first documented adjudicated new-onset ECG abnormality. Age showed non-proportionality, but biomarker estimates were similar in age-stratified sensitivity models. The 3-point score had modest apparent, in-sample discrimination (C-index, 0.645). During follow-up, 53 deaths and 137 PFS events occurred. cTnI remained associated with both OS and PFS after forced adjustment. Baseline cTnI and CK-MB were associated with the first documented adjudicated new-onset ECG abnormality, although discrimination was modest and performance estimates were apparent and in-sample. Separately, cTnI, D-dimer, NLR, and ECOG performance status contributed to exploratory survival risk stratification. ECG abnormalities should not be interpreted as definite ICI-related cardiotoxicity. Prospective multicenter validation is required before these findings can inform clinical monitoring strategies.