Abstracts

Rationale: Hypoxic-ischemic encephalopathy (HIE) is a major cause of neurological morbidity in neonates, including cerebral palsy, cognitive impairment, and epilepsy. Randomized clinical trials have shown that therapeutic hypothermia improves motor outcomes in term infants with HIE, but post-TH epilepsy outcomes are not well described. We aimed to identify clinical, MRI, and EEG features from the neonatal period that predict the subsequent development of seizures before age 2, in this population. Methods: Retrospective chart review was performed on term neonates who underwent selective head cooling at a tertiary care center neonatal ICU. Subjects were monitored on cEEG after the 72-hour cooling period and thereafter if clinically indicated. At 7-10 days of life each patient had an MRI performed. After discharge, subjects were longitudinally followed in a neonatal follow-up clinic by a multidisciplinary team including neonatologists, child neurologists, occupational/physical therapists and a developmental pediatrician. Subjects were excluded if they lacked 2 year follow up (lost to follow up or died before age 2). A multivariable logistic regression model was built to identify significant independent risk factors for the development of epilepsy by the age of 2, using clinical features from the first day of life, EEG data (based on chart review) at 72 hours, and details from brain MRI report. We performed variable selection with the best subsets method to find the model that minimized the AIC (Aikake Information Criteria). The probability of developing epilepsy was calculated based on this model. Results: 49 infants had 2-year follow up data. Of these, 9 developed epilepsy outside of the neonatal period, 3 of which were infantile spasms. In bivariate analysis, initial pH 6.8 (p = 0.03, OR [95% CI] 10.8 [1.2,95]), and MRI injury to thalamus (p = 0.006, OR 15.2 [2.2, 105]), and basal ganglia (p = 0.001, OR 20 [3.3,120]) were each significantly associated with any seizures before age 2. The multivariable analysis found a three-variable logistic regression model best fit the data, with pH <6.8 (p = 0.03, adjusted OR 31 [1.4,660]), cEEG at 72-96h with burst suppression pattern (p = 0.04, adjusted OR 20.2 [1.2,333]), and MRI with evidence of basal ganglia injury (p = 0.004, adjusted OR 57 [3.7,863]) as predictive factors of significance. Using this analysis, the predictive model found a low-risk (0-1 risk factors) and high-risk group (2-3 risk factors) for the development of epilepsy. In the low-risk group, 1/37 neonates developed seizures by age 2 vs 8/12 neonates in the high-risk group. Conclusions: Our model predicts that neonates with HIE who undergo TH are at greatest risk of epilepsy if they fall into a high-risk group with 2-3 identified risk factors based on EEG, clinical and MRI criteria. We anticipate that this classification of risk will help direct close follow-up and surveillance of the development of epilepsy in this population.