Multi-center real-world validation of the HFA-ICOS cardio-oncology risk tool in HER2-positive breast cancer patients receiving anti-HER2 therapy.

e24015 Background: The HFA-ICOS baseline cardiovascular risk stratification framework is widely recommended to guide surveillance strategies in patients receiving potentially cardiotoxic cancer therapies. However, evidence supporting its real-world clinical utility and risk separation across diverse populations remains limited. Methods: We conducted a multicenter retrospective cohort study including adult patients with HER2-positive breast cancer treated with trastuzumab-based regimens between 2016 and 2023. Baseline cardiovascular risk was classified using the HFA-ICOS framework (low, moderate, high, and very high). The primary outcome was composite cardiotoxicity, defined by declines in left ventricular ejection fraction, deterioration in global longitudinal strain, ECG/arrhythmic events, or elevation of cardiac biomarkers. Observed event rates, clinical risk separation, and time-to-event patterns were compared across risk categories. Discrimination and calibration were explored descriptively. Associations were assessed using multivariable logistic regression, and time-to-event analyses were performed using Kaplan–Meier methods. Results: Among 687 patients, 273 (39.7%) experienced composite cardiotoxicity during follow-up. Baseline HFA-ICOS classification categorized 35.7% as low risk, 37.4% as moderate risk, 26.4% as high risk, and 0.6% as very high risk. Cardiotoxicity events occurred across all risk categories, with substantial overlap in observed event rates and time-to-event distributions. Exploratory Receiver Operating Characteristic (ROC) analyses demonstrated modest discrimination for EF-defined (AUC 0.51), GLS-defined (AUC 0.55), ECG-defined (AUC 0.51), and composite cardiotoxicity (AUC 0.52). Sensitivity was low, while specificity was moderate; negative predictive values were consistently higher than positive predictive values. Calibration analyses demonstrated acceptable agreement between predicted and observed risk patterns. In multivariable analysis, baseline HFA-ICOS risk category was not independently associated with composite cardiotoxicity (adjusted OR per category increase 0.88; 95% CI 0.56–1.37). Event-free survival curves showed marked overlap among low-, moderate-, and high-risk groups. Conclusions: In this large real-world cohort of HER2-positive breast cancer patients, baseline HFA-ICOS risk stratification demonstrated limited clinical risk separation for cardiotoxicity, with substantial overlap in outcomes across risk categories. These findings support integrating baseline HFA-ICOS stratification with dynamic, longitudinal cardio-oncology surveillance strategies that incorporate on-treatment imaging and biomarkers, rather than relying solely on baseline risk assessment.