Detection of left ventricular hypertrophy through ECG: diagnostic performance and comparison of ten ECG scores in different hypertrophic conditions.

Abstract

Different electrocardiographic (ECG) criteria were proposed to improve ECG-based detection of left ventricular hypertrophy (LVH), but external validation data are limited and their diagnostic performances in different conditions leading to LVH are not extensively assessed. to assess the robustness and limitations of ECG scores across distinct LVH phenotypes. The LVH-MORE registry included 408 patients across four cohorts: HTN (n = 103), HCM (n = 104), sAS (n = 101), and controls (n = 100). LVH was defined by echocardiographic left ventricular mass index (LVMi) >115 g/m2 in male subjects and >95 g/m2 in female subjects. Ten ECG criteria were calculated, including Peguero-Lo Presti, Sokolow-Lyon, Cornell, Perugia, Romhilt-Estes, and others. Diagnostic performance was assessed using ROC curves (AUC), sensitivity (Sn), specificity (Sp). In the HTN cohort, Peguero-Lo Presti achieved the highest AUC for LVH detection (0.752 [95% CI: 0.616-0.888]), with Sn = 31% and Sp = 94%. Other scores showed similar or lower AUCs, with Perugia score showing higher sensitivity (54%). The highest specificity were reached by Sokolow-Lyon and Gubner-Ungerleider (100%). In the HCM cohort, all ECG scores showed modest diagnostic performance (AUC 0.529-0.628). The Perugia and Romhilt-Estes scores achieved the highest sensitivities (94% and 71%, respectively), whereas Sokolow-Lyon showed the highest specificity (90%). In the sAS cohort, AUCs ranged from 0.519 to 0.671. Sensitivity was higher for Perugia (66%) and Romhilt-Estes ≥4 (57%). Highest specificity was shown by Gubner-Ungerleider (95%). No single ECG criterion provided consistent diagnostic accuracy across different causes of LVH. Voltage-only ECG scores performed better in HTN patients while multi-parametric scores showed relatively better performance in HCM and sAS.