Predictability of cardiotoxicity at a Belgian cardio-oncology clinic without mandatory physician referral

Abstract

Abstract Background and purpose Increased morbidity and mortality caused by side-effects of cancer treatment on cardiovascular function mandates careful monitoring and shared decision-making in cardio-oncology clinics. We report the experience of the cardio-oncology clinic at a large Belgian tertiary care center, without mandatory physician referral, and investigated the predictability of cardiotoxicity based on pre-existing cardiovascular risk factors, specific cancer treatment and existing risk scores of the American Society of Clinical Oncology (ASCO) and Mayo Clinic. Methods We included all patients seen at the outpatient cardio-oncology clinic between May 2018 and September 2020. We defined cardiotoxicity as a decline in ejection fraction (EF) of 10% in asymptomatic patients and 5% in symptomatic patients. Approval of the ethical committee was obtained (S65084). Results The majority were women (68%), with almost half (44%) having metastatic disease. Mean age was 63.4±16.0 years. The most frequent oncological diagnoses were breast cancer (33%) and haematological diseases (24%). Patients most frequently received radiation therapy (42%), anthracyclines (39%) and antimetabolites (35%). Mean follow-up was 443±245 days. Receiver operating characteristic (ROC) analysis of predictors of cardiotoxicity showed an area under the curve (AUC) of 0.580 (CI 95% bootstrap: 0.525–0.642) for cardiovascular risk factors alone, and an AUC of 0.613 (CI 95% bootstrap: 0.550–0.676) when treatment was added to the model. The ASCO risk score poorly predicted cardiotoxicity (sensitivity 64%, specificity 52%). The Mayo Clinic cardiotoxicity risk score was the best predictor of cardiotoxicity with an AUC of 0.685 (CI 95% bootstrap: 0.625–0.743). Discussion Classic cardiovascular risk factors alone, or in combination with the proposed cancer treatment cannot adequately predict cardiotoxicity risk. The Mayo Clinic Cardiotoxicity Risk score outperformed the ASCO risk score but requires further refinement to enhance adequate cardiovascular risk prediction. For future model building, we hypothesize that an all-comer population of oncological patients should be used as a derivation cohort, regardless of previous or current cardiac issues, risk factors or type of therapy. Admittedly, individual susceptibility due to genetic, epigenetic and environmental predisposition cannot be adequately incorporated into a risk score and emphasizes the need for individual cardiotoxicity risk evaluation using bed to bench tools such as the use of induced pluripotent stem cell derived cardiac or vascular cells to assess the susceptibility of individual patients to cancer drug-induced cardiovascular toxicities. In the absence of adequate risk prediction tools, we advocate standardized screening of all patients before oncological treatment starts. Funding Acknowledgement Type of funding sources: None. Figure 1. Cardiotoxicity