Cumulative doxorubicin-equivalent anthracycline exposure and incident cardiac abnormalities in children with cancer: a retrospective cohort study

Anthracycline chemotherapy improves survival in children with cancer but may cause dose-dependent cardiotoxicity. A broad composite endpoint dominated by frequent, nonspecific ECG findings may obscure an exposure-response association with more specific echocardiographic abnormalities. Cumulative doxorubicin-equivalent exposure was therefore evaluated using prespecified composite and component endpoints. A retrospective cohort study was conducted among 644 pediatric cancer patients with baseline-normal cardiac evaluation and cumulative anthracycline exposure ≥100 mg/m 2 . Exposure was standardized as doxorubicin-equivalent dose and categorized into quartiles. The prespecified primary endpoint combined electrocardiographic, echocardiographic, or cardiac enzyme abnormalities. Associations were evaluated using Kaplan-Meier and Cox models, with spline and sensitivity analyses. During follow-up, 584 patients (90.7%) developed the prespecified composite endpoint, mainly driven by ECG abnormalities. The static final-dose analysis showed no significant association with the composite endpoint; however, a time-varying cumulative-dose analysis showed a modest association (adjusted HR per 50 mg/m², 1.17; 95% CI, 1.03–1.33; P = 0.018). Thirty-seven patients (5.7%) developed incident echocardiographic abnormalities, and Q4 had higher risk than Q1 (adjusted HR, 3.38; 95% CI, 1.11–10.28; P = 0.032). In a post hoc analysis requiring a LVEF decline ≥ 10% points from the available pretreatment value together with follow-up LVEF < 60%, 13 of 279 evaluable patients met the endpoint and the continuous-dose association was no longer significant (adjusted HR per 50 mg/m², 0.84; 95% CI, 0.51–1.38; P = 0.489). Restricted cubic spline analysis was inconclusive regarding the precise exposure–response shape or a specific threshold. Associations between anthracycline exposure and cardiac outcomes were sensitive to endpoint definition and temporal exposure assignment. Higher cumulative exposure was associated with the prespecified echocardiographic abnormality endpoint, but this association was not reproduced using a more specific within-patient LVEF-decline criterion, indicating that the observed association is sensitive to endpoint definition.These findings support cautious interpretation of retrospective cardiac endpoints and do not establish a new surveillance strategy or dose threshold.