Comparing MRI-based scoring systems for neuroprognostication in neonatal encephalopathy
Related Experiment Videos
Michal Yaar1, Vann Chau1, Jessy Parokaran Varghese2
1Department of Pediatrics, Division of Neurology, The Hospital for Sick Children, Toronto, ON, Canada.
Background:
Neonatal encephalopathy (NE) remains a significant cause of neurodevelopmental impairment. However, which MRI-based scoring system best predicts 36-month outcomes remains unclear.
Methods:
A prospective cohort of 102 neonates ≥36 weeks gestation with NE underwent early postnatal brain MRI. Images were scored using five established systems: Barkovich, Weeke, Trivedi, Rutherford, and NICHD-NRN. Neurodevelopmental outcomes were evaluated at 36 months. Prognostic performance was analyzed using logistic regression and receiver operating characteristic curves. The best predictive model was determined using pairwise correlation to reduce highly correlated variables, and then backward selection to optimize the linear regression model.
Results:
The Weeke total score showed the highest predictive accuracy for Bayley Scales of Infant Development motor outcomes (AUC 0.7647). The Weeke gray matter subscore was most predictive of Child Behavior Checklist internalizing problems (AUC 0.7735). No scoring system reliably predicted cognitive, language, or externalizing problems (AUC < 0.7). The optimal motor outcomes prediction model included 5 Weeke variables: basal ganglia, posterior limb of internal capsule, perirolandic cortex, cerebellar hemorrhage location, and basal ganglia lactate.
Conclusions:
MRI-based scoring systems have limited prognostic value beyond motor outcomes after NE. The Weeke score offers the most reliable motor prediction, achievable using only five parameters.
Impact:
Existing MRI scoring systems have limited prognostic value beyond motor outcomes after neonatal encephalopathy. The Weeke score offers the most reliable motor prediction. Only 5 Weeke scoring system variables are needed to optimize motor prediction. Deep gray matter injury is associated with internalizing behavioral outcomes. A simplified, user-friendly prognostic model can support accurate outcome prediction by a broader range of clinicians.