Abstract Background Operational feasibility and reproducibility of echocardiographic (Echo) parameters are essential for planning interventional trials with surrogate imaging endpoints. Existing data from large cohort studies performed at institutions with high expertise in Echo may not be representative of the multicentre clinical trial settings where substantial variability in local standards and equipment exists. Purpose To assess feasibility of standardized Echo data acquisition, measurability, and reproducibility of key Echo endpoints in a large multicentre clinical trial. Methods ENDEAVOR was a phase 2b/3 multicentre, international, placebo-controlled trial investigating the effects of mitiperstat on symptoms and function in participants with heart failure (HF) and left ventricular ejection fraction (LVEF) >40%. Echo was used for efficacy assessment at 3 time points (baseline, 16 and 24 weeks). Echo data were acquired according to the study protocol at sites with various equipment and expertise and analysed centrally by an academic core laboratory. Key Echo metrics were LVEF, LV global longitudinal strain (GLS), LV mass, E/E’ and left atrial volume (LAV). LVGLS was measured from three apical views (4, 2, and 3-chamber) by 2DSTRAIN TomTec software. All images were evaluated for image quality using the scale from 1 (Very poor) to 5 (Excellent) and measured by a single reader. Results 711 participants at 142 sites in 18 countries were randomised in ENDEAVOR trial (mean age 72.4y, mean BMI 30.5kg/m2). A total of 2052 Echo exams were analysed: 735 at baseline, 661 at 16 and 656 at 24 weeks. Majority were of excellent to good (1282/62%) or adequate (732/36%) image quality. Doppler images and apical 4ch view were the most feasible to acquire, while apical 2ch view was the most challenging (Fig, A). LVEF, LV mass, E/E’ and LAV were measurable in most studies (88-96%), however fewer studies were interpretable for LVGLS (61% at baseline to 70% at week 24). When matching echo exams at 2 or 3 visits were analysed, the number of interpretable matches dropped more significantly for LVGLS (to 47% for 2 and 41% for 3 time points), while the proportion of interpretable matches for standard echo endpoints remained acceptable (>83% and >75%, respectively). A total of 82 scans were re-read for reproducibility. Most of the echo endpoints demonstrated good or excellent reproducibility, with LVGLS showing better reproducibility than LVEF (Fig, B). Conclusions We report feasibility of Echo data acquisition, measurability, and reproducibility of key parameters of LV remodelling and function in a large multicentre, international, placebo-controlled clinical trial in HF and EF>40%. These data will help to inform future trial design to drive drug development decision making.
Read full abstract