The potential role of vectorcardiographic QRS-area in determining capture type in left bundle branch area pacing

Abstract Background Confirming conduction system capture during left bundle branch area pacing (LBBAP) primarily relies on ECG criteria obtained at the time of implantation. Post-implant ECG interpretation, however, is often subjective to clinical variability. In this context, objective and more quantitative parameters could aid implanting cardiologists in identifying LBBAP capture. Purpose To investigate the potential role of vectorcardiographic QRS-area in identifying capture type during LBBAP, compared to clinical judgment. Methods Unipolar paced ECGs during LBBAP implantation from 48 patients with preserved ejection fraction and baseline narrow QRS were collected. Conduction system capture was attempted in all patients and assessed using QRS morphology transitions during threshold testing following the EHRA 2023 consensus statement (gold standard). Eight experienced LBBAP-implanting cardiologists from six tertiary referral centers were asked to classify the ECGs into capture types. The vectorcardiographic QRS area was calculated from the synthesized vectorcardiogram using the Kors transformation matrix, with custom-developed code programmed in MATLAB. The accuracy of QRS area in differentiating left ventricular septal pacing (LVSP) from non-selective (nsLBBP), compared to clinical judgment of each individual observer, was evaluated using sensitivity, specificity, and accuracy analyses. Differences between QRS-area and QRS duration between LVSP and nsLBBP were assessed with Mann-Whitney U tests. Results QRS area was significantly higher for LVSP than for nsLBBP (LVSP median: 38, range 28–49; nsLBBP median: 22, range 15–28; p <0.001). QRS duration (determined from onset stimulus) was not significantly different between LVSP and nsLBBP (LVSP median: 133, range 125–143; nsLBBP median: 131, range 110–141; p <0.108). A cutoff point of 26 mV.ms (determined from ROC curves) for QRS-area yielded a sensitivity of 81%, specificity of 69%, and accuracy of 77% for distinguishing LVSP from nsLBBP. Clinical judgement of the 8 observers yielded an average sensitivity of 82% (range 71-100%), specificity of 64% (range 52-85%), and accuracy of 72% (range 62-78%). An overview of the accompanying sensitivity and specificity analyses is provided in the figure. Conclusion Vectorcardiographic QRS-area is a promising parameter for distinguishing LVSP from nsLBBP. Its ability to differentiate capture types slightly outperforms clinical judgment of experienced cardiologists and may serve as a valuable quantitative adjunct to clinical assessment, enhancing the accuracy of conduction system capture determination during LBBAP.Accuracy analyses