Performance of cardiotoxicity risk scores in a tertiary center cardio-oncology clinic

Introduction

Cardiotoxicity worsens prognosis in cancer patients, making early detection and prevention crucial. Several predictive scores exist, but their applicability across therapies and populations is limited, highlighting the need for more reliable risk stratification.

Aim

To identify cardiotoxicity predictors and assess risk scores in cancer patients.

Metoths

Retrospective single-centre study of consecutive patients seen in a cardio-oncology clinic (2022–2023). Cardiotoxicity was defined as heart failure decompensation, >10% LVEF reduction, clinically significant arrhythmias, myocardial infarction, myocarditis, pericarditis, or refractory hypertension. Baseline HFA-ICOS, CardTox, CRS, SCORE2 and SCORE2-OP scores were calculated. A composite endpoint (CV hospitalization or CV death) was analysed using Kaplan–Meier curves and Cox regression.

Results

A total of 209 patients were included (mean age 60 ± 2 years, 57% female). Most received HER2-targeted therapy (18%), anthracyclines (17%) or hormone therapy (17%). Over a median follow-up of 16 months (IQR 11–23), 85 patients (41%) developed cardiotoxicity, most commonly HF (69%), LVEF decline (44%), AF/flutter (20%), and pericarditis (7%). Cardiotoxicity was significantly associated with baseline LVEF, GLS, TAPSE and troponin levels. In multivariate analysis, GLS ≤ −18% independently predicted cardiotoxicity (OR 5.5, p=0.002) and LVEF decline (OR 10.0, p=0.002), adjusted for age and baseline LVEF. HER2-targeted therapy was linked to higher risk of new-onset HF (OR 4.4, p=0.01), and HT to new-onset AF/flutter (OR 4.1, p=0.03), both adjusted for age, atrial size and concomitant cancer therapies.

A composite endpoint (CV hospitalization or CV death) occurred in 18 patients (21%). Independent predictors were LVEF decline (OR 6.3, p=0.04) and HF decompensation (OR 22.0, p=0.01), adjusted for age, cardiovascular risk factors and baseline LVEF. While risk scores did not predict cardiotoxicity directly, high/very high HFA-ICOS, CardTox, CRS scores and SCORE2 were associated with the composite endpoint (HR = 11.76; IC95%: 1.51–91.90; p = 0.019; HR = 4; IC95%: 1.5–9.8; p = 0.003; HR = 3.8; IC95%: 1.4-10.8; p = 0.009; HR = 4.4; IC95%: 1.1-19.9; p = 0.05, respectively). Subgroup analysis showed that HFA-ICOS best predicted cardiotoxicity in patients receiving multiple myeloma therapy (p=0.02), whereas SCORE2 and SCORE OP performed better in those on HT (p=0.01). All-cause mortality was not related to cardiotoxicity or cardiovascular hospitalization.

Conclusions

GLS and cardio-oncology risk scores such as HFA-ICOS, CardTox and CRS independently predict adverse CV outcomes. Routine integration of these tools could enhance individualized risk stratification and guide preventive strategies, but further multicenter studies are required to refine risk identification and validation.For image description, please refer to the figure legend and surrounding text.