Lamia A Alamri, Laura K Kaizer, Jason Gien, S Christopher Derderian, Mary D Sammel, Mariana L Meyers, Michael V Zaretsky, Henry L Galan
In this cohort of left-sided CDH, all prenatal prognostic measures by US and MRI were highly predictive of ECMO use. O/E LHR by US and PPLV by MRI were the best predictors of ECMO in newborns, and these were most predictive in late gestational age.
INTRODUCTION: Prenatal ultrasound (US) and magnetic resonance imaging (MRI) prognostic measurements are routinely used to predict congenital diaphragmatic hernia (CDH) severity. Practice variability exists in the type and timing of measurements obtained. The primary objective of this study was to determine the best prenatal prognostic measures to predict the need for extracorporeal membrane oxygenation (ECMO). Similarly, the secondary objective was to determine which measures best predicted a variety of other neonatal outcomes.
METHODS: This is a retrospective study of left-sided CDH affected pregnancies at a single fetal care center from 2013 to 2023 who had at least one US and MRI. The cohort was divided into early (24 week) and late (34 week) groups, based on gestational age at the time of evaluation. Lung-to-head ratio (LHR), observed-to-expected LHR (O/E LHR), O/E total lung volume, and percent predicted lung volume (PPLV) were obtained at each visit. Clinical data were obtained from the electronic medical record. Pregnancies with other major anomalies, genetic syndromes, and <32 weeks gestation at the time of delivery were excluded. To identify the prenatal measures most predictive of ECMO, logistic regression models were estimated using standardized measures as covariates, and the standardized odds ratios were compared across measures. Following the identification of the strongest predictive measure for each imaging modality, each measure from both timepoints was included in one model to identify the best time (early vs. late) for outcome prediction.
RESULTS: Of the 124 patients affected with CDH, 90 met inclusion criteria. 11 out of the 90 neonates did not survive to discharge (12.2%). Patients on ECMO were more likely to be female (59% p = 0.034) and have lower birthweight (2810 [413] vs. 3080 [483], p = 0.014). All prenatal measures between MRI and US were significantly associated with ECMO use; however, MRI PPLV and US O/E LHR at the late evaluation had the strongest association.
CONCLUSION: In this cohort of left-sided CDH, all prenatal prognostic measures by US and MRI were highly predictive of ECMO use. O/E LHR by US and PPLV by MRI were the best predictors of ECMO in newborns, and these were most predictive in late gestational age.