IMPROVED PREDICTION OF LEFT-VENTRICULAR MASS BY REGRESSION-ANALYSIS OF BODY-SURFACE POTENTIAL MAPS

Electrocardiographic left ventricular (LV) hypertrophy involving ST-T abnormalities, in addition to high QRS voltages, is associated with increased risk of cardiovascular disease mortality. Unfortunately, conventional electrocardiographic criteria have limited utility in the quantitative assessment of LV hypertrophy. Body surface potential maps, which contain diagnostic information not present in commonly used lead systems, were recorded from 117 thoracic sites and 3 limb electrodes in 72 normal subjects and 84 patients with LV hypertrophy. Multiple regression analysis was performed separately for 54 women and 102 men on 120-lead data, using as features instantaneous voltages on time-normalized P, PR, QRS and ST-T waveforms. Leads and features for optimal prediction of echocardiographically etermined LV mass were selected. A total of 6 features from 3 torso sites in men, and from the same 3 sites plus 2 others in women, yielded correlations between echocardiographic and electrocardiographic estimates of LV mass of 0.89 and 0.88, respectively. The standard errors of the estimate (SEE), or average errors in predicting LV mass from the regression equations, were 31 and 22 g, respectively. The single most potent predictor in both sexes was a mid-QRS voltage measured on a lead positioned 10 cm below V1; QRS duration, late QRS and early-to-mid T-wave amplitudes recorded in the lower left flank contributed significantly to the performance of both regression models. The optimal electrode sites for electrocardiographic prediction of LV mass were outside the conventional lead locations. In comparison, regression analysis on the standard 12-lead electrocardiogram in the same population produced a correlation of 0.76 in men and 0.75 in women; the SEE were 48 and 35 g, respectively. The Sokolow-Lyon criterion and the Romhilt-Estes point score yielded correlations of 0.57 and 0.69, respectively, in men and 0.51 and 0.67 in women. Substantial improvement in predicting LV mass from electrocardiographic measurements can be achieved by appropriate selection of a limited, practical subset of electrode positions from body surface potential maps. © 1990.