356Variability of left ventricular ejection fraction measurement by imaging modality for cardiotoxicity screening: Comparison between Radionuclide Ventriculography, 2D and 3D Echocardiography and CMR

Abstract

Background: The accuracy of left ventricular ejection fraction (LVEF) measurement depends on test-retest variability. Small changes gatekeeper major treatment decisions, particularly in cancer patients at risk of cardiotoxicity where guidelines recommend withholding treatment if LVEF falls by 10%. There are different imaging modalities used to assess this. Previous studies have compared technique inter-observer reproducibility. We compared precision in LVEF measurement in oncology patients using same day test-retest assessment by three imaging modalities. Method(s): Oncology patients being screened for chemotherapy-related cardiotoxicity by any imaging modality: equilibrium radionuclide ventriculography (ERNV), cardiovascular magnetic resonance (CMR) or echocardiography (2D and 3D echo) were recruited. All patients underwent same day repeat testing on their requested modality together with repeated imaging using a second modality. Image quality was categorised as 'good' or 'poor' using standard criteria, then measurement was made of LVEF and global longitudinal strain (GLS) for CMR and Echo. Coefficient of variation (CoV) and minimum detectable difference (MDD) were calculated. Result(s): 76 patients (mean age 51.6 +/- 11.7 years, 13% male) were recruited. The CoVs for ERNV, 2D echo, 3D echo and CMR were 6.9%, 8.8%, 7.4% and 5,4% respectively p = 0.076 (ERNV vs 2D echo) and p = 0.456 (ERNV vs 3D echo). The MDD was better for ERNV and CMR when compared to 2D echo and 3D echo (10.8% and 8.9% vs 11.8%, 11.5%, respectively) figure 1. CoV for GLS measured using CMR was non-inferior to 2D echo GLS (10.5 vs 9.2%, p = 0.385). When analysis was restricted to studies deemed good quality, >90% of CMR and ERNV studies fulfilled criteria versus only 75% and 62% of studies for 2D and 3D Echo. Just limiting the analysis to this group, only 2DE remained inferior to CMR for CoV (8.8 vs 5.4%, p = 0.002) Table 1 Conclusion(s): Serial screening for chemotherapy-related cardiotoxicity depends on precision. This study raises concerns that 2D echo may lack this precision to assess cardiac function and therefore may not be suitable for this. When image quality is good, 3D echo, ERNV and CMR are equivalent. (Table Presented).