医学
肺静脉
窦性心律
心房颤动
内科学
心脏病学
队列
作者
Zhi Rui Zhang,Don Ragot,Sophia Z. Massin,Adrian Suszko,Andrew C.T. Ha,Sheldon M. Singh,Vijay S. Chauhan
标识
DOI:10.1016/j.cjca.2023.04.014
摘要
Background Atrial low-voltage areas (LVAs) in patients with atrial fibrillation increase the risk of atrial arrhythmia (AA) recurrence after pulmonary vein isolation (PVI). Contemporary LVA prediction scores (DR-FLASH, APPLE) do not include P-wave metrics. We aimed to evaluate the utility of P-wave duration/amplitude ratio (PWR) in quantifying LVA and predicting AA recurrence after PVI. Methods In 65 patients undergoing first-time PVI, 12-lead ECGs were recorded during sinus rhythm. PWR was calculated as the ratio between the longest P-wave duration and P-wave amplitude in lead I. High-resolution biatrial voltage maps were collected and LVAs included bipolar electrogram amplitudes < 0.5 mV or < 1.0 mV. An LVA quantification model was created with the use of clinical variables and PWR, and then validated in a separate cohort of 24 patients. Seventy-eight patients were followed for 12 months to evaluate AA recurrence. Results PWR strongly correlated with left atrial (LA) (< 0.5 mV: r = 0.60; < 1.0 mV: r = 0.68; P < 0.001) and biatrial LVA (< 0.5 mV: r = 0.63; < 1.0 mV: r = 0.70; P < 0.001). Addition of PWR to clinical variables improved model quantification of LA LVA at the < 0.5 mV (adjusted R2 = 0.59 to 0.68) and < 1.0 mV (adjusted R2 = 0.59 to 0.74) cutoffs. In the validation cohort, PWR model–predicted LVA correlated strongly with measured LVA (< 0.5 mV: r = 0.78; < 1.0 mV: r = 0.81; P < 0.001). PWR model was superior to DR-FLASH (area under the receiver operating characteristic curve [AUC] 0.90 vs 0.78; P = 0.030) and APPLE (AUC 0.90 vs 0.67; P = 0.003) at detecting LA LVA and similar at predicting AA recurrence after PVI (AUC 0.67 vs 0.65 and 0.60). Conclusion Our novel PWR model accurately quantifies LVA and predicts AA recurrence after PVI. PWR model–predicted LVA may help guide patient selection for PVI.
科研通智能强力驱动
Strongly Powered by AbleSci AI