The prediction of cancer therapeutics-related cardiac dysfunction (CTRCD) is an essential aspect of care for individuals who receive potentially cardiotoxic oncologic treatments. Certain clinical risk factors have been described for incident CTRCD, and measurement of left ventricular (LV) longitudinal strain by speckle tracking 2-dimensional echocardiography (2DE) is the best-validated myocardial mechanical imaging assessment to detect subtle changes in LV function during cancer treatment. However, the direct integration of clinical and imaging risk factors to predict CTRCD has not yet been extensively examined. This was a retrospective study of 183 women with breast cancer aged 50.9 ± 10.8 years who received treatment with anthracyclines (doxorubicin dose of 422 ± 69 mg/m2, with 41.2% of subjects also receiving trastuzumab) and underwent 2DE at clinically determined intervals. CTRCD was diagnosed when LV ejection fraction dropped ≥10% to a subnormal (<53%) value by 2DE. Left ventricular global longitudinal strain (LV-GLS) was assessed offline. The risk prediction tool based only on clinical factors previously described by Ezaz et al was applied to our cohort and accurately stratified these subjects into low-, intermediate-, and high-risk groups, with incident CTRCD in 7.4%, 26.9%, and 54.6%, respectively (chi-square = 20.7, p <0.0001). We developed novel multivariate models to predict CTRCD using (1) demographic variables only (c = 0.8674), (2) echocardiographic (peak LV-GLS) variables only (c = 0.8440), or (3) a combination of demographic and echocardiographic variables, with the combined model exhibiting superior receiver-operating characteristics (c = 0.9629). In conclusion, estimation of CTRCD risk should integrate all available data, including both clinical variables and an imaging assessment.
Copyright © 2018 Elsevier Inc. All rights reserved.