Abstract
BACKGROUND: Early echocardiographic indicators of cardiac remodeling may enhance cardiomyopathy risk prediction in childhood cancer survivors (CCS). OBJECTIVES: The objective of the study was to assess whether influential echocardiographic measures can be combined to develop a robust cardiomyopathy risk prediction model in CCS. METHODS: Multicenter retrospective study of ≥1-year CCS with digitally archived surveillance echocardiograms, enrolled cardiomyopathy cases (left ventricular [LV] fractional shortening ≤28% or LV ejection fraction ≤50% on ≥2 occasions) and noncases (≥5-year CCS who maintained fractional shortening ≥ 30% and ejection fraction ≥55% without initiation of cardiac medications). Echocardiograms were centrally quantitated in a blinded fashion. Least absolute shrinkage and selection operator regression identified the most influential 2-year predictors of cardiomyopathy among 27 echocardiographic parameters. Logistic regression was used to generate ORs with 95% CIs. Estimates were applied to the training and test data sets to generate area under the receiver operating characteristic curves (AUC). RESULTS: Data from 146 CCS (52 cases; 94 noncases) with a median follow-up of 9.3 years post-cancer diagnosis and a total of 281 echocardiograms were included. A set of 7 echocardiographic measures were identified as the most influential predictors, with AUC of 0.82 (95% CI: 0.74-0.89) and 0.85 (95% CI: 0.74-0.95) in the training and test data sets, respectively. LV end-systolic dimension (OR(mm): 1.2; 95% CI: 1.1-1.4), apical 4-chamber longitudinal strain (OR%: 1.2; 95% CI: 1.0-1.3), and septal A' velocity (OR(cm/s): 1.3; 95% CI: 1.1-1.6) were strongly predictive of cardiomyopathy. AUCs were similar if cancer treatment exposures were included. CONCLUSIONS: Early abnormalities in echocardiographic parameters of structure and function predict subsequent cardiomyopathy in CCS and can identify high-risk survivors who warrant early intervention.