Electrocardiographic predictors of haemodynamic benefit from biventricular pacing: the role of atrioventricular timing measures

Abstract

Abstract Background Patients with heart failure (HF) and left bundle branch block (LBBB) are selected for biventricular pacing (BVP) cardiac resynchronisation therapy (CRT) based on their 12-lead ECG QRS complex duration (QRSd). This is due to the long-held belief that BVP delivers its beneficial effects almost exclusively by reducing ventricular dyssynchrony. Recent work (1) suggests that atrioventricular delay (AVD) shortening is another key mechanism of benefit and, therefore, that atrioventricular timing may be an important CRT substrate. The ideal 12-lead ECG measure to facilitate patient selection that yields benefits from this mechanism is yet to be elucidated. Purpose We set out to establish whether incorporating a measure of AV delay provides additional benefit over and above using QRSd when using the 12-lead ECG to select patients who are likely to obtain haemodynamic benefit from BVP. Methods We identified patients in sinus rhythm with HF and LBBB meeting clinical criteria for CRT who had undergone implantation of a BVP device and had also undergone high-precision haemodynamic assessment of BVP. This is where pacing is alternated between atrium-only pacing (AAI) and DDD pacing using BVP multiple times for each of a full range of AVDs during continuous beat-by-beat blood pressure recording to determine the peak haemodynamic benefit of BVP in each patient. We extracted potentially predictive, pre-specified intrinsic (unpaced) ECG features that incorporated left or right AV timing to varying extents: PR interval (PR), end-of-p-wave-to-R-wave-peak-time-V6 (Pend-RWPTV6), start-of-p-wave-to-end-of-QRS (P-QRSend) as well as conventional QRSd. Pearson's test assessed the unadjusted correlation between intrinsic ECG intervals and BVP haemodynamic benefit(∆SBP). A multiparametric regression analysis model was constructed to account for confounders and interactions between different ECG features. Results 53 eligible patients were identified. 39 were male (73.6%), 31 had non-ischaemic aetiology (58.5%), mean LVEF was 27.6% ± SD 4.79 and mean LVEDD was 57.9± SD 9.4mm. Unadjusted correlations between intrinsic ECG features were not strong but favoured P-QRSend as a potentially predictive feature; R2 values: PR 0.28 (p=0.04), Pend-RWPTV6 0.19 (p=0.18), P-QRSend 0.27 (p=0.054), QRSd 0.045 (p=0.75). However, the multiparametric model revealed that utilising both PR interval and Pend-RWPTV6 in a predictive model showed better-adjusted correlation with BVP haemodynamic benefit (R2 0.54, p=0.04). Conclusion ECG features that incorporate either right (PR interval) or left (Pend-RWPTV6) AVD or both (P-QRSend) are superior to QRSd for predicting haemodynamic benefit from BVP in patients with HF and LBBB. This further reinforces the role of AVD shortening as a mechanism of benefit of BVP and suggests that selecting patients on the basis of AV timing may yield more overall benefit from CRT delivered by BVP or potentially conduction system pacing.ECG interval measurementCorrelation Graphs