5 Visual assessment of systemic right ventricles: does imaging modality and experience matter?

Introduction Reproducible and accurate ventricular assessment for systemic right ventricle (RV) is challenging but is essential for serial monitoring and optimising lifelong management for complex congenital patients. We investigated visual grading and formal quantification using 2D transthoracic echo and cardiac MRI to assess reproducibility and the impact of clinical experience and formal societal accreditation on these data. Methods Data from 10 patients with systemic RV were assessed. Apical ‘4-chamber’ transthoracic (Epic, Philips) and ‘4-chamber/HLA’ cardiac MRI (1.5T Avanto, Siemens) images were anonymised and duplicated twice. Each echo image was viewed for 30 s by 20 echocardiographers/physicians and CMR data by 12 physicians. The level of experience and accreditation varied amongst raters. Raters were requested to grade function by i) visually assessment (hyperdynamic, normal, mildly, moderately or severely impaired), ii) provide an ’eyeball’ estimate of ventricular function (EF) ±5% and iii) judge image quality (0%–100%). Results Images from 10 patients were assessed; female n=6, mean age 27±9 years (22–48 years), single ventricles (HLHS) n=8, atrial switch (Senning) n=1 and ccTGA n=1. In total, 16 echo and 6 CMR raters viewers were accredited (BSE, EACVI, SCMR or BSCMR). Formal CMR quantification of the systemic right ventricle was 61%±8% (50%–77%). Intra-observer The 95% limits of agreement for EF were ±13% for echo and ±17% for CMR. Observers gave identical measures of EF in 36% echo and 28% CMR and were within ±5% in 77% and 66% respectively (table 1). There was no significant difference between repeated measures in experienced echo (p=0.783) or those with regular ACHD lists (p=0.341) or CMR (p=0.095), although there was smaller difference in experienced CMR participants. Intraobserver ‘eyeball’ estimation was in the same grade in 63% (echo) and 60% (CMR) (table 2). There was no significant difference between experience in echo but intra-observer consistency in CMR was significantly greater in experienced observers, with same classification in 72% vs 47% in inexperienced observers (p=0.009) figure 1. Interobserver reproducibility is shown in figure 2, with Kappa statistic 0.236 using linear weighting, and 0.331 using quadratic weighting for echo and 0.085 and 0.107 for CMR respectively. Pairs of observers classified in the same category of ventricular function in 36% (echo) and 32% on CMR. Conclusion Visual grading and quantification of systemic RV function is difficult with limited reproducibility with both echo and CMR, even in experienced hands. Experience in congenital CMR improves reproducibility but for accurate systemic RV function, a full data-set with full quantitative analysis is recommended in both echo and CMR.