Normalized QRS Duration by LV Dimension: Eliminating Sex-Specific Differences and Predicting Cardiac
Enrui Zhang1, Jiaxin Zeng1, Fengwei Zou2
1Department of Cardiology, First Affiliated Hospital, Nanjing Medical University, Nanjing, China.
Background:
Increasing evidence shows sex-specific differences in the efficacy of cardiac resynchronization therapy (CRT). Without a clear cardiac rationale, relying solely on sex to evaluate resynchronization outcomes is insufficient.
Objectives:
This study sought to assess whether QRS duration (QRSd) normalized to cardiac size (QRSd/left ventricular end-diastolic dimension [QRSd/LVEDD]) could underlie the observed sex-specific differences in resynchronization efficacy and guide CRT implantation.
Methods:
This multicenter retrospective study screened 602 patients with heart failure who underwent biventricular or left bundle branch area pacing for resynchronization therapy. The primary composite outcome included all-cause mortality, heart failure hospitalizations, and ventricular tachyarrhythmic events. Interaction analysis and structural equation modeling were applied to evaluate whether QRSd/LVEDD could independently predict outcomes and elucidate sex-specific differences.
Results:
Compared with male patients, female patients (n = 216) had a higher QRSd/LVEDD ratio (2.66 vs 2.48; P < 0.001). After QRSd/LVEDD adjustment, female patients no longer exhibited greater left ventricular ejection fraction improvement. Sex-specific differences initially observed in clinical outcomes (HR: 1.71; 95% CI: 1.17-2.48; P = 0.005) were eliminated by QRSd/LVEDD (HR: 1.14; 95% CI: 0.76-1.69; P = 0.527), with a significant dose-response relationship (P < 0.001). QRSd/LVEDD also stratified prognosis within each sex group (P < 0.001). No sex-specific differences were found among patients with the same QRSd/LVEDD (QRSd/LVEDD ≥2.55, P = 0.744; QRSd/LVEDD <2.55, P = 0.123). Structural equation modeling revealed that sex did not directly influence clinical outcomes (P = 0.693) but exerted an indirect effect through QRSd/LVEDD (P < 0.001). QRSd/LVEDD exhibited superior ability to predict clinical outcomes compared with QRSd and was integrated into machine learning modeling for individualized CRT risk prediction.
Conclusions:
QRSd/LVEDD effectively eliminated sex-specific differences in CRT efficacy and outperformed QRSd in predicting clinical outcomes. Sex did not directly influence CRT outcomes but mediated its effect indirectly through QRSd/LVEDD.
