LAUSR

Abstract Funding Acknowledgements Type of funding sources: Public Institution(s). Main funding source(s): This work was supported by the National Institute for Health Research Biomedical Research Centre at Guy’s and St. Thomas’ Trust and King’s College, the Centre of Excellence in Medical Engineering funded by the Wellcome Trust and Engineering and Physical Sciences Research Council (EPSRC; WT088641/Z/09/Z). M.J.B. is supported by a Medical Research Council New Investigator Grant (MR/N011007/1) and British Heart Foundation (Project grant PG/18/74/34077). This work was supported by EPSRC 2018/19 DTP - EP/R513064/1 grant. This work was supported by a National Heart and Lung Institute Foundation grant awarded to Professor Sanjay Prasad and Dr Richard Jones. Background Arrhythmic risk stratification of patients with stable coronary artery disease (CAD) for defibrillator implantation remains imprecise. Current clinical guidelines center on a basic assessment of adverse cardiac remodelling, an approach with well documented limitations. More advanced analysis of intricate and regional LV remodelling in these patients may shed new insight for enhanced risk prediction of life-threatening arrhythmia in stable CAD. Objective We explored a novel computational 3D LV shape analysis strategy to accurately represent, and quantify, arrhythmogenic LV remodelling patterns in late gadolinium enhancement cardiovascular magnetic resonance images (LGE-CMR) in patients with stable CAD. We assess the utility of our patient-specific 3D LV arrhythmic shape (LVAS) score predicting major arrhythmic event; comparing to the current clinical benchmark of severely reduced LVEF and NYHA functional class. Methods Patients with stable CAD were prospectively recruited into a CMR registry and followed up for a composite endpoint of major arrhythmic events (MAE) consisting of sudden cardiac death (SCD), aborted SCD or haemodynamically unstable VT. Time-to-event data was also recorded. 2D LGE-CMR images were contoured and used as input to an advanced computational shape-analysis approach (Figure 1A). Reconstructed 3D LV shape models were analysed using Principle Component Analysis and Cox-Lasso to characterise the average arrhythmogenic LV shape, which was then used to quantitatively define a patient-specific LV arrhythmic shape (LVAS) score. Results Of 397 patients (mean [SD] LVEF% 45.4 ± 16.0) followed for a median (IQR) of 6 (3) years, 55 (14%) experienced a major arrhythmic event (MAE). The mean (SD) LVEF for patients who experienced MAE was lower than for those that did not (37.1% ± 13.2 vs 46.7% ± 16.0), yet there was no difference in NYHA class. LVAS mean (95% confidence interval [CI]) score was different between event and event-free patients: 2.98 (0.48-6.44) vs 19.83 (10.66-29), P<0.001. On univariate Cox regression analysis both LVAS score (hazard ratio [HR] 2.1, 95% CI 1.6-2.7, P<0.001) and LVEF<35% (HR 2.9, 95% CI 1.6-5.0, P<0.01) were associated with the primary endpoint. NYHA Class >I was not associated with the primary endpoint (HR 1.5, 95% CI 0.8-2.7, P=0.221). In multivariate Cox regression analysis, LVAS remained independently associated with the primary endpoint after adjusting for LVEF<35% and NYHA>Class I: LVAS score HR 2.1, 95% CI 1.5-3.0, P<0.001; LVEF<35% HR 1.0, 95% CI 0.5-2.2, P=0.923; NYHA>Class I HR 1.3, 95% CI 0.7-2.5, P=0.443 (Figure 1 B, C). The addition of the LVAS score increased the C-statistic from 0.6 to 0.75 between models. Conclusion A Left Ventricle Arrhythmic Shape score improved arrhythmic risk stratification compared to guideline-based clinical parameters, highlighting a potentially novel approach to identifying candidates for implantable cardioverter defibrillators in stable CAD.