Comparative accuracy of ECG LV hypertrophy scores and human readers using CMR as reference standard

View at publisher
Authors: Jędrzej Sarnecki, Martin Segeroth, PB Badertscher, I Strebel, Jan Gerrit van der Stouwe, P Lopez Ayalo, J Reinhardt, S Frey, M J Zellweger, R Schmid, Felix Mahfoud, C Mueller
+1 more Philip Haaf
Cited by: 0

Abstract

Abstract Background Electrocardiographic (ECG) criteria for left ventricular hypertrophy (LVH), including the widely used Sokolow-Lyon index, are frequently employed in clinical practice. However, newer criteria such as the Peguero-Lo Presti and Groningen scores have been proposed as superior alternatives. This study aimed to compare the diagnostic accuracy of expert ECG readers and various ECG-based LVH criteria, using cardiovascular magnetic resonance imaging (CMR) as the reference standard, and to assess the impact of myocardial fibrosis and infarction on diagnostic performance. Methods In this retrospective cohort study, 1,034 patients referred for CMR between 2018 and 2020 were analysed. Digital 12-lead ECGs were independently reviewed by two blinded cardiologists and automatically assessed for LVH using 25 established ECG criteria as well as the novel Peguero-Lo Presti and Groningen scores. CMR provided the reference standard for LV mass index (LVMi) and identified late gadolinium enhancement (LGE) patterns indicative of myocardial fibrosis or infarction. Results CMR confirmed LVH in 25% of patients. Expert ECG interpretation demonstrated low sensitivity (12.2%) and high specificity (97.6%), comparable to the Sokolow-Lyon index. The Peguero-Lo Presti, Groningen, and Cornell duration product criteria significantly outperformed both expert readers and traditional criteria, achieving higher sensitivity (39.0–45.7%) while maintaining high specificity (83.3–89.7%) (Table 1). An ECG heatmap showed that the S wave amplitudes exhibit the strongest correlation with LVMi across multiple leads (Figure 1, left). The presence of non-ischaemic LGE was associated with higher diagnostic performance for select ECG criteria, whereas ischaemic LGE reduced diagnostic accuracy (e.g. Peguero-Lo Presti, Figure 1, right). Conclusions Commonly used ECG criteria and expert readings have limited sensitivity for LVH detection using CMR as reference for LVH. Recently published criteria (Peguero-Lo Presti, Groningen) offer improved performance, though fibrosis patterns, especially ischaemic LGE, impact accuracy.Table 1 Summary Results Figure 1 Heatmap, I vs. NI-LGE

Fields of Study & Subjects

ECG Monitoring and Analysis Cardiomyopathy and Myosin Studies Cardiac Imaging and Diagnostics

Publication details

Journal / source
European Heart Journal - Cardiovascular Imaging
Volume
27
Issue
Supplement_1
Publication date
1 January 2026
Language
EN