Incremental Value of Myocardial Deformation in Predicting Postnatal Coarctation of the Aorta: Establishment of a Novel Diagnostic Model

Abstract

Prenatal detection of coarctation of the aorta (CoA) still suffers from high false-positive and false-negative rates. The objective of this study was to develop a novel model to improve the diagnostic accuracy of fetal CoA. A retrospective study was conducted in 122 fetuses with suspected CoA who also had postnatal follow-up. Fetuses with confirmed diagnoses of CoA after birth were defined as the CoA group, and the non-CoA group comprised those with false positives. Conventional fetal echocardiographic measurements, including great arterial dimensions and flow characteristics, were obtained. Left ventricular (LV) functional parameters were determined using two-dimensional speckle-tracking echocardiography. A novel multiparametric diagnostic model, including gestational age at diagnosis, aortic isthmus (AOI) Z score, and LV longitudinal strain (LVLS), was developed in univariate and multivariate logistic regression analyses. The model was validated prospectively by a validation cohort of 48 fetuses. CoA was confirmed in 62 of 122 patients (50.8%) after birth. Fetuses with postnatal CoA were diagnosed significantly earlier than those with false-positives (median, 24.5weeks [interquartile range, 23.3-26.4weeks] vs 27.8weeks [interquartile range, 24.5-30.4weeks]; P<.001). The Z scores of aortic dimensions (aortic valve annulus, ascending aorta, transverse aortic arch, and AOI) were significantly smaller (P<.001 for all), whereas the Z scores of pulmonary dimensions (pulmonary valve annulus and main pulmonary artery) were significantly greater (P<.05 for all), in cases of confirmed CoA than false positives. Compared with the non-CoA group, the CoA group displayed lower LV ejection fraction (P=.005), LV fractional area change (P<.001), and LVLS (P<.001). A multivariate logistic regression model incorporating gestational age (odds ratio, 0.74; 95% CI, 0.60-0.88; P=.001), AOI Z score (odds ratio, 0.20; 95% CI, 0.08-0.41; P<.001), and LVLS (odds ratio, 1.79; 95% CI, 1.41-2.42; P<.001) was shown to diagnose CoA more accurately (Akaike information criterion=81.77, C statistic=0.945). The performance of this model was confirmed prospectively in the validation cohort. In fetuses with suspected CoA, speckle-tracking analysis of LVLS may have incremental value in predicting postnatal CoA. The diagnostic model presented here, including gestational age, AOI Z score and LVLS, may provide a good tool for stratification of risk in fetal CoA and contribute to patient-specific perinatal management.