Abstract
BACKGROUND: We aimed to investigate if left ventricular electromechanical delay (LVEMD) prolongation predicts trastuzumab-induced cardiotoxicity (TIC) in breast cancer patients. HYPOTHESIS: LVEMD prolongation on serial echocardiograms could be an indicator of subclinical TIC. METHODS: We included 237 breast cancer patients receiving trastuzumab chemotherapy, who underwent echocardiography at baseline and at 6 and 12 months after trastuzumab initiation. LVEMD was defined as the time from electrical activation to myocardial contraction. TIC was defined as left ventricular ejection fraction (LVEF) worsening to <55%, either as symptomatic decrease of ≥5% or asymptomatic decrease of ≥10%. RESULTS: During a mean follow-up of 547 days, TIC occurred in 27 patients (11.4%). Changes in the time intervals from QRS onset on electrocardiography to the beginning and peak of transaortic flow on pulsed-wave Doppler echocardiography (ie, ΔLVEMDi and ΔLVEMDp, respectively) were independent predictors of TIC. On receiver operating characteristic curve analysis, the optimal cutoff value for TIC prediction was 23 milliseconds for ΔLVEMDi (sensitivity, 0.85; specificity, 0.78; area under the curve [AUC], 0.882) and 21 milliseconds for ΔLVEMDp (sensitivity, 0.96; specificity, 0.68; AUC, 0.860). The C-index for TIC prediction increased significantly after adding ΔLVEMDi and ΔLVEMDp to conventional models that included clinical variables, baseline LVEF, and changes in global longitudinal peak systolic strain. Similarly, adding ΔLVEMDi or ΔLVEMDp to conventional models provided significant improvement in discrimination capability for TIC prediction (integrated discrimination improvement and continuous net reclassification improvement index). CONCLUSION: ΔLVEMDi and ΔLVEMDp may serve as predictors of subclinical cardiac dysfunction in breast cancer patients receiving trastuzumab.