Abstract 18300: Do Current Cardiovascular Risk Models Predict the Development of Cardiotoxicity in Lymphoma Patients Treated With Anthracycline?

Abstract

Introduction: Assessing baseline cardiovascular (CV) risk factors is crucial for identifying cancer patients at higher risk of cardiotoxicity (CT). While both CT specific scores, including Heart Failure Association-International Cardio Oncology Society (HFA-ICOS) and our institutional score, as well as general CV scores like Atherosclerotic cardiovascular disease (ASCVD), have been used to assess CT risk; in lymphoma patients undergoing anthracycline therapy, their utility remains unknown. Aim: Evaluate the performance of each risk score to predict CT in a cohort of newly diagnosed lymphoma patients undergoing anthracycline treatment, followed for a median of 5.4 years. Methods: 326 Hodgkin and non-Hodgkin lymphoma patients treated and followed from 2013 to 2021 were prospectively assessed. Baseline measurements of the three risk scores were obtained. ASCVD score was not assessed in patients >80 or <40 years old or without baseline cholesterol. CT was defined during or after one year of chemotherapy as severe, moderate, or mild according to 2022 ESC guidelines. Logistic regression and AUC were used to assess the association of each score and compare the models. Results: Among 326 patients, 78 developed CT (8 severe, 21 moderate, 49 mild). A high-risk score from HFA-ICOS was significantly associated with severe and moderate CT(OR: 6.1, p=0.001), sensitivity 37%, and specificity 88%. A very high-risk classification of the institutional score was associated with severe and moderate CT (OR: 8.4, p=0.006), sensitivity 44%, and specificity 81%. No association was found between the ASCVD score stratification and CT. When mild CT was included in the analysis, the AUC values were lower (see Table 1). Conclusions: Cardio-oncology risk scores were associated with severe and moderate CT, outperforming ASCVD, but their predictive capacity was weak. Incorporating other serum or imaging biomarkers into the models may improve the predictive accuracy of CT risk assessment.