Risk Prediction Model of Cardiotoxicity in Breast Cancer Patients; the Multicenter Prospective CHECK HEART-BC (comprehensive Heart Imaging to Evaluate Cardiac Damage Linked with Chemotherapy in Breast

Abstract Background Cardiotoxicity due to cancer therapy has been emerging as an important issue along with the improvement in prognosis of breast cancer patients. Although the usefulness of global longitudinal strain and biomarker analysis was previously reported separately, there are few studies based on systematic evaluation of cardiotoxicity. Furthermore, large prospective cohort studies regarding cardiotoxicity and its prediction in breast cancer patients were also limited. Purpose Multicenter prospective study was performed to reveal the current characteristics of cardiotoxicity and develop a robust prediction model for cardiotoxicity using systematic evaluation of cardiac function. Methods and results Breast cancer patients who were scheduled for neoadjuvant and/or adjuvant chemotherapy were prospectively screened at the 25 participating institutions from August 2017 and March 2020. As a study protocol, follow-up visits were planned every 3 months from pre-treatment to 12 months after chemotherapy (median follow-up of 366 days). To evaluate their cardiac function, echocardiography, high-sensitive cardiac troponin, natriuretic peptide, and 12-lead ECG were acquired at every follow-up visit. cardiotoxicity was defined as a reduction in LVEF >10% point from baseline and to a value <53% in LVEF. Results 542 chemotherapy-naïve patients were included for analysis from 25 institutions. cardiotoxicity developed in 46 (8.5%) during follow-up period. The HEART2 score based on pre-treatment data (model A) was composed of 6 variables including heart rate >64 bpm, LV end-systolic volume index >36.0 mL/m2, treatment with anthracycline, with radiation, with trastuzumab, and TAPSE <26 mm. The receiver-operating characteristic (ROC) curve showed acceptable accuracy (an area under curve (AUC) 0.74 [95%confidence interval 0.66-0.81], P<0.01). Furthermore, the ROC curve of the HEART2 score adding variables of 3 months after chemotherapy (model B) (difference in heart rate from pre-treatment, LV end-systolic volume index >42.0 mL/m2) showed a greater AUC than the model based on only pre-treatment data (AUC 0.81 [95%confidence interval 0.73-0.90], P<0.01). No patient with low-risk group in the HEART2 score based on variables of pre-treatment and 3 months after chemotherapy developed cardiotoxicity, while 28.1% developed cardiotoxicity in patients with high-risk group in both models. Conclusions CHECK HEART-BC study revealed current clinical characteristics of cardiotoxicity and developed the HEART2 score which showed good accuracy and consistency. The HEART2 score can provide an efficient and effective strategy for early detection of cardiotoxicity.The HEART2 scoreThe combination of model A and B