Congenital Heart Disease
Development and Validation of an Echocardiographic Model for Predicting Progression of Discrete Subaortic Stenosis in Children

https://doi.org/10.1016/S0002-9149(97)00911-9Get rights and content

Abstract

The clinical course of discrete subaortic stenosis (DSS) varies considerably between patients. This study was performed to identify echocardiographic characteristics of DSS that distinguish progressive from nonprogressive disease. The study included 100 patients from 2 institutions and was performed in 2 stages. In phase I, a prediction model was developed based on multivariate analysis of morphometric and Doppler variables obtained from the initial echocardiogram in 52 children with DSS from Texas Children’s Hospital. In phase II, the performance characteristics of the prediction model were tested in 48 patients with DSS followed at Children’s Hospital in Boston. Patients were divided into 3 outcome groups: nonprogressive, progressive, and intermediate progression. In phase I, multivariate analysis identified 3 independent predictors of progressive disease: indexed aortic valve to subaortic membrane distance, anterior mitral leaflet involvement, and initial Doppler gradient. The logistic regression equation—Probability = [1 + e−(−3.22+0.334X1+4.06X2−0.708X3)]−1, where X = initial gradient in mm Hg; X2 = absence (0) or presence (1) of mitral leaflet involvement; and X3 = indexed distance between aortic valve and subaortic membrane in mm/body surface area0.5 were used to predict progression. When the prediction model was applied to phase II study patients, none of the patients with nonprogressive DSS had a prediction value >0.29 and none of the patients with progressive DSS had a prediction value <0.58. Thus, a prediction value >0.55 yielded a 100% sensitivity and 100% specificity for distinguishing progressive from nonprogressive DSS. Patients with intermediate progression were indistinguishable from progressive DSS but were clearly separable from nonprogressing patients. We conclude that progressive subaortic obstruction in children with DSS can be predicted from morphologic, morphometric, and Doppler echocardiographic analysis of left ventricular outflow.

Section snippets

Study Design:

The study included 100 patients from 2 institutions and was performed in 2 stages. In phase I, a multiple logistic regression prediction model was developed based on multiple clinical and echocardiographic variables of 52 patients with DSS from Texas Children’s Hospital. In phase II, the prediction model was applied to 48 patients with DSS from Children’s Hospital in Boston to examine its diagnostic accuracy in an independent study sample.

Phase I Study:

All patients diagnosed by echocardiography as having DSS

Patients:

One hundred patients with DSS who met inclusion criteria were included in the study: 52 patients from Texas Children’s Hospital participated in phase I and 48 patients from Children’s Hospital, Boston, were included in phase II. Based on the defined criteria, 33 patients were classified as having nonprogressive DSS, 50 as having progressive stenosis, and the remaining 17 patients formed the intermediate progression group. The demographic and clinical characteristics of the groups are summarized

Discussion

This study identified 3 echocardiographic features that were independently predictive of progressive DSS in children: (1) the distance of subaortic obstruction from the aortic valve, (2) involvement of the anterior mitral leaflet by the subaortic membrane, and (3) initial Doppler gradient. These 3 variables can be applied using a prediction equation to discriminate between nonprogressive and progressive subaortic stenosis.

Early studies based largely on cardiac catheterization data from patients

References (30)

Cited by (42)

  • Aortic Stenosis Morphology

    2015, ASE’s Comprehensive Echocardiography
  • Valve Disease in Children

    2009, Valvular Heart Disease: A Companion to Braunwalds Heart Disease Expert Consult - Online and Print
View all citing articles on Scopus
View full text