RESUMEN
Background/Purpose: This study aims to quantify the utility of monitoring LVEF, hs-cTnT, and NT-proBNP for dynamic cardiotoxicity risk assessment in women with HER2+ early breast cancer undergoing neoadjuvant/adjuvant trastuzumab-based therapy. Materials and methods: We used joint models of longitudinal and time-to-event data to analyze 1,136 echocardiography reports and 326 hs-cTnT and NT-proBNP measurements from 185 women. Cardiotoxicity was defined as a 10% decline in LVEF below 50% and/or clinically overt heart failure. Results: Median pre-treatment LVEF was 64%, and 19 patients (10%) experienced cardiotoxicity (asymptomatic n = 12, during treatment n = 19). The pre-treatment LVEF strongly predicted for cardiotoxicity (subdistribution hazard ratio per 5% increase in pre-treatment LVEF = 0.68, 95%CI: 0.48-0.95, p = 0.026). In contrast, pre-treatment hs-cTnT and NT-proBNP were not consistently associated with cardiotoxicity. During treatment, the longitudinal LVEF trajectory dynamically identified women at high risk of developing cardiotoxicity (hazard ratio per 5% LVEF increase at any time of follow-up = 0.36, 95% CI: 0.2-0.65, p = 0.005). Thirty-four patients (18%) developed an LVEF decline ≥ 5% from pre-treatment to first follow-up ("early LVEF decline"). One-year cardiotoxicity risk was 6.8% in those without early LVEF decline and pre-treatment LVEF ≥ 60% (n = 117), 15.9% in those with early LVEF decline or pre-treatment LVEF < 60% (n = 65), and 66.7% in those with early LVEF decline and pre-treatment LVEF < 60% (n = 3), (Gray's test p < 0.0001). Conclusion: Cardiotoxicity risk is low in two thirds of women with HER2+ early breast cancer who have pre-treatment LVEF ≥ 60% and no early LVEF decline > 5% during trastuzumab-based therapy. The longitudinal LVEF trajectory but not hs-cTnT or NT-proBNP allows for a dynamic assessment of cardiotoxicity risk in this setting.