Prediction of postnatal circulation in pulmonary atresia/critical stenosis with intact ventricular septum: systematic review and external validation of models.

Objective: A favorable postnatal prognosis in cases of pulmonary atresia/critical stenosis with intact ventricular septum (PA/CS-IVS) is generally equated with the possibility of achieving biventricular (BV) repair. Identification of fetuses that will have postnatal univentricular (UV) circulation is key for prenatal counseling, optimization of perinatal care and decision-making regarding fetal therapy. We aimed to evaluate the accuracy of published models for predicting postnatal circulation in PA/CS-IVS using a large internationally derived validation cohort.

Methods: This was a systematic review of published uni- and multiparametric models for the prediction of postnatal circulation based on echocardiographic findings at between 20 and 28 weeks of gestation. Models were externally validated using data from the International Fetal Cardiac Intervention Registry. Sensitivity, specificity, predictive values, area under the receiver-operating-characteristics curves (AUCs) and proportion of cases with true vs predicted outcome were calculated.

Results: Eleven published studies that reported prognostic parameters of postnatal circulation were identified. Models varied widely in terms of the main outcome (UV (n = 3), non-BV (n = 3), BV (n = 3), right-ventricle-dependent coronary circulation (n = 1) or tricuspid valve size at birth (n = 1)) and in terms of the included predictors (single parameters only (n = 6), multiparametric score (n = 4) or both (n = 1)), and were developed on small sample sizes (range, 15-38). Nine models were validated externally given the availability of the required parameters in the validation cohort. Tricuspid valve diameter Z-score, tricuspid regurgitation, ratios between right and left cardiac structures and the presence of ventriculocoronary connections (VCC) were the most commonly evaluated parameters. Multiparametric models including up to four variables (ratios between right and left structures, right ventricular inflow duration, presence of VCC and tricuspid regurgitation) had the best performance (AUC, 0.80-0.89). Overall, the risk of UV outcome was underestimated and that of BV outcome was overestimated by most models.

Conclusions: Current prenatal models for the prediction of postnatal outcome in PA/CS-IVS are heterogeneous. Multiparametric models for predicting UV and non-BV circulation perform well in identifying BV patients but have low sensitivity, underestimating the rate of fetuses that will ultimately have UV circulation. Until better discrimination can be achieved, fetal interventions may need to be limited to only those cases in which non-BV postnatal circulation is certain. © 2023 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.