Association of AI derived right ventricular function with all-cause mortality in patients referred for routine cardiovascular magnetic resonance

Abstract

Abstract Funding Acknowledgements Type of funding sources: Public grant(s) – National budget only. Main funding source(s): Wellcome/EPSRC Centre for Medical Engineering at King’s College London (WT 203148/Z/16/Z), the National Institute for Health Research (NIHR) Cardiovascular MedTech Co-operative award to the Guy’s and St Thomas’ NHS Foundation Trust, and the London AI Center, funded through the UK Research and Innovation (UKRI) and the Department of Health and Social Care (DHSC), Office of Life Sciences, delivered through Innovate UK. Background Right ventricular (RV) function is an important prognostic marker in assessment of cardiovascular disease. Artificial Intelligence (AI) based quantification of RV function has been validated against ground truth metrics however their relationship to outcomes is not yet well defined. Purpose To evaluate the relationship between RV biomarkers obtained using our validated AI tool, AI-CMRQC with all-cause mortality in patients who underwent routine cardiovascular magnetic resonance (CMR). Methods 3,108 routine clinical CMR scans, performed between 2016 and 2020 were retrospectively analysed using AI-CMRQC (1) to obtain RV function parameters of end-diastolic volume (EDV), end-systolic volume (ESV), stroke volume (SV) and ejection fraction (EF). All-cause mortality and date of death were obtained from hospital electronic records and Personal Demographics Service from the Office for National Statistics (ONS). Cox proportional hazard risk models were used to derive hazard ratios for measuring association between RV function and all-cause mortality. All variables were stratified based on upper or lower limits of reference values (2), RVEF <50% or >50%, EDV >217ml males, >175ml females, ESV >91ml males, >75ml females, SV <71ml males, <56ml females. Results The distribution of mean values (±SD) of RV parameters were EDV 166.14 ± 0.83ml, ESV 77.83 ± 0.57ml, SV 88.3 ± 0.45ml and EF 55.9 ± 0.17%. The median follow up duration was 3.4 years. On univariate analysis, ejection fraction [Hazard Ratio (HR) 2.903 (95% Confidence Interval [CI] 2.197 to 3.834] and stroke volume [HR 3.385 (95% CI 2.534 to 4.522)] were significant predictors of mortality. After adjustment for covariates of left ventricular function (LVEF, LVEDV, LVESV) RVEF remained an independent predictor of mortality. Conclusions Automated RV ejection fraction is significantly associated with all-cause mortality in patients undergoing routine CMR.