There is controversy regarding the best echocardiographic diagnostic criteria for left ventricular noncompaction (LVNC). We assessed the diagnostic utility and reproducibility of the previously proposed echocardiographic diagnostic criteria in a pediatric population using a segmental approach.
Echocardiograms were matched for patients with and without a clinical diagnosis of LVNC. Blinded reviews of echocardiograms measured (1) depths of intertrabecular recesses (X/Y), (2) noncompaction-to-compaction ratio (NC/C), and (3) number of trabeculations, using a segmental approach. Measurements were analyzed for area under the receiver operating characteristic curves (AUC), sensitivity, and specificity.
There were 30 echocardiograms in the initial cohort (15 LVNC cases, 15 controls). Median age was 1.7 years (IQR 0.2-6.9 years) and systolic function was decreased in 40%. Comparison of diagnostic criteria demonstrated the best interrater agreement and AUC with an X/Y ratio measured in end-diastole in the parasternal short axis in the apical anterolateral segment (κ 0.72, CI 0.43-1.00, p value <0.001), yielding 100% sensitivity and 70-86% specificity, among readers. The least predictive and reproducible method was the NC/C ratio. A validation cohort confirmed the superiority of the X/Y ratio, although the interrater agreement and AUC decreased.
Measurements according to existing LVNC diagnostic criteria vary by echocardiographic view and segment. Modification of the Chin et al. criteria (Circulation 82:507-513, 1990) using an X/Y ratio <0.5 had the greatest interrater reliability and predictive validity when measured in end-diastole in the parasternal short axis in the apical anterolateral segment. The NC/C ratio had the lowest reliability and predictive validity.