Electrocardiographic Versus Echocardiographic Left Ventricular Hypertrophy in Severe Aortic Stenosis

Abstract

Although ECG used to be a traditional method to detect left ventricular hypertrophy (LVH), its importance has decreased over the years and echocardiography has emerged as a routine technique to diagnose LVH. Intriguingly, an independent negative prognostic effect of the “electrical” LVH (i.e., by ECG voltage criteria) beyond echocardiographic LVH was demonstrated both in hypertension and aortic stenosis (AS), the most prevalent heart valve disorder. Our aim was to estimate associations of the ECG-LVH voltage criteria with echocardiographic LVH and indices of AS severity. We retrospectively manually analyzed ECG tracings of 50 patients hospitalized in our center for severe isolated aortic stenosis, including 32 subjects with echocardiographic LVH. The sensitivity of single traditional ECG-LVH criteria in detecting echocardiographic LVH was 9–34% and their respective specificity averaged 78–100%. The ability to predict echocardiographic LVH was higher for S-waves than R-waves (mean area under the receiver operating curve (AUC): 0.62–0.70 vs. 0.58–0.65). Among combinations of R- and S-waves, the discriminating ability was highest for the Cornell voltage (AUC: 0.71) compared to the Sokolow–Lyon, Romhilt and Gubner–Ungerleider voltage (AUC: 0.62–0.68). By multiple regression, peak aortic pressure gradient was positively related to the Sokolow–Lyon (β = 1.7 ± 0.5, p = 0.002) and Romhilt voltage (β = 1.3 ± 0.5, p = 0.01), but not Cornell (0.5 ± 0.3, p = 0.2) or Gubner-Ungerleider voltage (β = 0.0 ± 0.5, p > 0.9), regardless of LV mass index. In conclusion, echocardiographic LVH and stenosis severity appear to have distinct associations with traditional ECG-LVH criteria in AS. A moderate diagnostic superiority of the Cornell voltage criterion with regard to anatomic LVH might result from its unique ability to include depolarization vectors in both the frontal and horizontal plane with consequent lesser sensitivity to the confounding effect of obesity.

Diagnostic Ability of Peguero-Lo Presti Electrocardiographic Left Ventricular Hypertrophy Criterion in Severe Aortic Stenosis

“…Out of 83 pre-screened subjects previously hospitalized in our center with the final diagnosis of severe aortic stenosis by means of echocardiography [25], we selected 50 previously described subjects (30 women and 20 men; average age 77 ± 10 years) with isolated severe aortic stenosis without relevant coexistent diseases [26]. As reported [26], we excluded subjects with QRS duration >120 ms, His bundle branch or left anterior fascicular block, more than mild aortic regurgitation or disease of another valve, a history of myocardial infarction, and left ventricular (LV) ejection fraction below 40%.…”

“…We compared the agreement of the novel Peguero-Lo Presti criterion [11] and traditional ECG-LVH criteria on routine in-hospital 12-lead ECG tracing [5,26] with echocardiographic LVH, defined in accordance with current recommendations as a LV mass (by the Devereux equation) over 95 g/m 2 in women and over 115 g/m 2 in men [27].…”

“…Additionally, since by means of the receiver operating characteristics (ROC) curve analysis we previously observed the highest ability of the Cornell voltage to predict echocardiographic LVH among 13 classical LVH-related QRS voltages in our study group [26], the diagnostic performance of the Peguero-Lo Presti voltage against the Cornell voltage was also compared as a continuous variable by the ROC curve analysis.…”

Automated Estimation of Computed Tomography-Derived Left Ventricular Mass Using Sex-specific 12-Lead ECG-Based Temporal Convolutional Network

“…Respectively integrating both precordial-and limb-lead features by individual encoders may further increase the diagnostic accuracy. 25,26 Second, previous studies indicated that sex difference exists in QRS duration and voltage regardless of baseline body size or left ventricular mass. 16,27 Even with similar comorbidities or disease severity, there are significant differences in terms of left ventricular mass and extent of myocardial fibrosis between sexes.…”

“…Demographic data played a crucial role in predicting LVM using our model. The inclusion of demographic data significantly improved the prediction performance compared to using only raw ECG signals by 25.1% (P<0.01). Furthermore, the addition of the heart axis and QRS duration information provided an insignificant performance improvement (by an absolute difference of 0.7%, P=0.82).…”