Abstracts

Rationale: Although acute symptomatic seizures are self-limited in the neonatal period, ≥25% of surviving children develop post-neonatal epilepsy (chronic unprovoked seizures), including up to 10% who develop infantile spasms. Strategies to prevent infantile spasms will require reliable, clinically available, early risk factors that identify infants at the highest risk. Methods: This was a prospective, multicenter cohort of surviving infants with acute symptomatic neonatal seizures born between 7/2015 and 3/2018 and enrolled at a Neonatal Seizure Registry site. The onset of infantile spasms before age 12 months was determined by parent interview and corroborated by medical record review. To create a clinically-relevant, parsimonious, and multivariable model, we collected neonatal variables (clinical, EEG, and centrally-reviewed MRI) associated with infantile spasms with a significance level of p<0.05 and evaluated all possible logistic regression models with 1, 2, or 3 variables (i.e., the best subsets method). We refined the prediction tool through a consensus process that balanced the results of multivariable logistic regression models with clinical relevance and ease of implementation. Results: Among 209 infants with complete neonatal clinical, EEG, and MRI data and ≥12 months follow-up, 11 developed infantile spasms (5%). There were several possible predictors among birth, clinical exam, EEG, and MRI variables (Table), all of which combined to yield similarly predictive models. We selected a parsimonious prediction model using three factors: (1) injury to the deep grey nuclei or brainstem on the initial neonatal admission MRI (DWI or T2); (2) a severely abnormal EEG background (e.g. severe discontinuity, burst suppression, or flat trace); and (3) abnormal neonatal discharge exam (abnormal tone, reflexes or level of consciousness). The risk for infantile spasms if none of these factors were present was 0% [95% confidence interval 0-4%] (0/99). Among children with one or two neonatal risk factors, the risk for infantile spasms was 5% [95% CI 2-11%] (5/100). If all three neonatal factors were present, the risk for infantile spasms was 55% [95% CI 21-79%] (6/11). The performance of this tool to predict infantile spasms by age 12 months was fair, with sensitivity 55% and positive predictive value 60%. Conclusions: These results suggest that risk for infantile spasms after acute symptomatic neonatal seizures can be stratified using commonly available clinical data. Infantile spasms developed in more than half of infants with acute symptomatic neonatal seizures who had all three of: (1) MRI with deep grey or brainstem injury; (2) severely abnormal neonatal EEG; and (3) abnormal discharge neurological examination. Identifying children at high risk for infantile spasms is important for counseling, clinical monitoring, and clinical trials designed to test novel agents to prevent this post-neonatal epilepsy syndrome. Funding: This work was supported by PCORI and the Pediatric Epilepsy Research Foundation.