Numerous ECG indices have been developed to diagnose left ventricular (LV) hypertrophy but ECG indexes assessing LV enlargement are lacking. The aim of the present study was to address the capacity of different ECG parameters to predict LV enlargement in different conditions, including myocardial infarction (MI), using cardiac MRI (CMR).

In a cohort of 501 patients with various clinical conditions, CMR and ECG were performed within a median period of 5 days. LV enlargement was defined by a LV end-diastolic volume indexed to body surface area (LVEDVI) >92ml/m².

In the whole cohort, amplitudes of RaVL, SV3, SV1 and QRS duration correlated in univariate analysis with LVEDVI (R=0.108, 0.304, 0161 and 0.256 respectively, p<0.01 for all). In multivariable analysis (adjusted for age, BMI, LV mass index, systolic blood pressure and gender), only QRS duration and SV3 remained associated with LVEDVI (β=0.130, p<0.001; β=0.057, p<0.001, respectively). Areas under ROC curves (AUC) demonstrate that SV3 had the best performance to predict LV enlargement (AUC 0.701, specificity 96.9%, sensitivity 19.1%, optimal threshold 2.2mV, see figure). In patients without MI (N=300), SV3 and QRS duration were independently correlated with LEVDVI (β =0.051, p<0.001; β=0.273, p<0.001, respectively), while in patient with MI only SV3 remained independently correlated with LEVDVI (β =0.051, p<0.001). AUC for SV3 were lower in patients with MI than those without (0.681 vs. 0.708), in women than in men (0.673 vs. 0.706) and in patients with QRS duration <120ms than those without (0.685 vs. 0.787). The optimal threshold of SV3 varies from 1.7 to 2.2mV according to different subgroups of patients.

Using the gold standard to assess LV volume, our results demonstrated that SV3 had a good specificity to detect LV enlargement.