Risk Prediction Models of Cardiotoxicity in Patients With Breast Cancer

BACKGROUND: Early detection and treatment of cardiotoxicity are essential for reducing cardiac events. However, a reliable predictive model for cardiotoxicity in patients with breast cancer receiving chemotherapy is lacking. OBJECTIVES: In this study, we aimed to develop a risk prediction model and establish effective surveillance for cardiotoxicity in patients with breast cancer undergoing chemotherapy. METHODS: Patients with breast cancer scheduled for neoadjuvant and/or adjuvant chemotherapy were prospectively screened at 25 participating institutions between August 2017 and March 2020. Cardiotoxicity was defined as a reduction in left ventricular ejection fraction of >10% from baseline to a value <53%. RESULTS: The study included 559 chemotherapy-naïve female patients. Cardiotoxicity was observed in 46 of 559 patients (8.2%) during a median follow-up period of 366 days (Q1-Q3: 365-367 days). The CHECK HEART (Comprehensive Heart Imaging to Evaluate Cardiac Damage Linked With Chemotherapy in Breast Cancer Patients) score consisted of 6 variables: heart rate, left ventricular global longitudinal strain, left ventricular end-systolic and end-diastolic diameters, right ventricular fractional area change, and treatment with anthracycline and trastuzumab. The time-dependent area under the receiver operating characteristic curve (AUC) at 12 months based on pretreatment data showed acceptable accuracy (AUC: 0.82; 95% CI: 0.76-0.89). CONCLUSIONS: The developed multivariable risk prediction models can accurately predict cardiotoxicity and support effective surveillance in patients with breast cancer receiving chemotherapy.