Prediction of Neonatal Intensive Care Unit Mortality in Fetuses and Neonates with Vein of Galen Malformation

Abstract

Background and purpose: Vein of Galen malformation (VOGM) is a rare fetal arteriovenous shunt with presentations ranging from asymptomatic in infancy to high-output cardiac failure and death. Prenatal percutaneous embolization is being explored in fetuses predicted to be at elevated risk of death in the Neonatal Intensive Care Unit (NICU). The purpose of this study was to evaluate the reliability of previously reported measurements and identify any novel imaging markers predictive of NICU mortality in a cohort of patients with VOGM managed at our institution.

Materials and methods: This was a single-center retrospective cohort study of fetuses and neonates with isolated VOGM evaluated at our hospital from 2016 to 2024. Patients with fetal and/or neonatal MRI had assessment of the narrowest mediolateral diameter of the straight or falcine sinus at its narrowest dimension (SS-MD), ventriculomegaly, pseudofeeders, cerebral ischemia, and hydrops. Indexed combined cardiac output and tricuspid regurgitation severity were evaluated echocardiographically. Advanced postprocessing MRI techniques were used to calculate the VOGM varix volume, brain parenchymal volume, and varix-to-brain volume ratio (VBR). Receiver operating characteristic curves were used to determine cutoff values for predicting NICU mortality. All variables predicting NICU mortality were adjusted for age at MRI and delivery using regression analysis.

Results: Fourteen cases with pre-embolization MRIs were identified (6 fetal and 8 neonatal). NICU mortality was 29% (4/14). Survivors and nonsurvivors had significantly different SS-MDs (6.6 versus 11.0 mm, P = .007) and VOGM varix volume (4145 versus 12,758 mm³, P = .014), respectively; however, after adjusting for age at MRI and delivery, these differences were no longer statistically significant. However, VBR significantly differed between survivors and nonsurvivors (1.4% versus 7.7%, P = .008, respectively), even after adjusting for age at MRI and delivery (P = .038). A VBR >4.18% was 100% sensitive and 90% specific in predicting NICU mortality (area under the curve = 0.95, P = .011) with a positive likelihood ratio of 10.

Conclusions: Previously reported imaging findings did not predict NICU mortality from VOGM in this cohort after adjusting for age at MRI and delivery. However, a VBR >4.18% strongly predicted NICU mortality in our cohort, suggesting that this novel parameter may help identify patients with VOGM who could benefit from fetal intervention.