Echocardiography versus Cardiac MRI for Measurement of Left Ventricular Ejection Fraction in Individuals with Cancer and Suspected Cardiotoxicity

Two-dimensional echocardiography showed large variations compared with reference standard cardiac MRI for left ventricular ejection fraction measurement, which may lead to downstream impact on clinical decision-making for cardio-oncology patients.Purpose:To compare left ventricular ejection fraction (LVEF) measured with echocardiography and cardiac MRI in individuals with cancer and suspected cardiotoxicity and assess the potential effect on downstream clinical decision-making.Materials and Methods:In this prospective, single-center observational cohort study, participants underwent same-day two-dimensional (2D) echocardiography and cardiac MRI between 2011 and 2021. Participants with suboptimal image quality were excluded. A subset of 74 participants also underwent three-dimensional (3D) echocardiography. The agreement of LVEF derived from each modality was assessed using Bland-Altman analysis and at relevant thresholds for cardiotoxicity.Results:A total of 745 participants (mean age, 60 years ± 5 [SD]; 460 [61.7%] female participants) underwent same-day echocardiography and cardiac MRI. According to Bland-Altman analysis, the mean bias was –3.7% ± 7.6 (95% limits of agreement [LOA]: –18.5% to 11.1%) for 2D echocardiography versus cardiac MRI. In 74 participants who underwent cardiac MRI, 3D echocardiography, and 2D echocardiography, the mean LVEFs were 60.0% ± 10.4, 58.4% ± 9.4, and 57.2% ± 8.9, respectively (P < .001). At the 50% LVEF threshold for detection of cardiotoxicity, there was disagreement for 9.3% of participants with 2D echocardiography and cardiac MRI. Agreement was better with 3D echocardiography and cardiac MRI (mean bias, –1.6% ± 6.3 [95% LOA: –13.9% to 10.7%]) compared with 2D echocardiography and cardiac MRI (mean bias, –2.8% ± 6.3 [95% LOA: –15.2% to 9.6%]; P = .016).Conclusion: Two-dimensional echocardiography had variations of ±15% for LVEF measurement compared with cardiac MRI in participants with cancer and led to misclassification of approximately 10% of participants for cardiotoxicity detection. Three-dimensional echocardiography had better agreement with cardiac MRI and should be used as first-line imaging