Abstracts

Rationale: Prophylactic (preceding first clinical seizure) antiseizure treatment in people at high-risk for epilepsy is an emerging area of study. Prophylactic vigabatrin, for example, can prevent infantile spasms in infants with Tuberous Sclerosis Complex (TSC).¹ Epilepsy often occurs in children with autism spectrum disorder (ASD). We hypothesized that there are factors in people with ASD predictive of comorbid seizures and that these factors may be used to devise predictive models for seizure risk in children with ASD. This may be used to identify high-risk groups for clinical trials, allow targeted counseling, and deepen understanding of the ASD-epilepsy link.


Methods: A retrospective cross-sectional study was conducted on 280 individuals in the Weill Cornell Medicine electronic health record for children with ASD born 1/1/07-6/30/07 (N=135) or 1/1/16-4/15/16 (N=145). Fisher’s Exact Tests were performed to determine statistically significant differences in distribution of individual factors between people with ASD with versus without a history of one or more seizures. Best subsets logistic regression was used to determine the best 4-variable model (based on deviance and AIC) from factors significant by Fisher’s Exact Test. Multivariable logistic regression was performed with these 4 variables. The 4-variable model was used to predict the probability of seizure given combinations of the 4 factors.

Results: Ten factors were significantly different between groups (Fig 1). Best subsets logistic regression of the 10 factors identified intellectual disability (ID), movement disorder, self-injurious behavior, and developmental delay in 2+ domains as the best 4-variable combination. A logistic regression model estimated the probability of seizure as 88% in people with ASD and all 4 factors as compared to 5% in those with ASD without any of the 4 factors (Fig 2).

Conclusions: This study (a) identifies 10 possible seizure predictors in patients with ASD: ID, developmental delay in 2+ domains, minimally verbal/nonverbal, self-injurious behavior, physical aggression, sleep disorder, visual impairment other than refractive errors, movement disorder, wheelchair-dependency, and potential metabolic etiology for ASD, epilepsy, and/or ID, and (b) provides a framework for using multivariable models incorporating such factors to predict patients with ASD at highest risk for seizures. In ongoing work, we are refining the prediction models and developing a scoring system to stratify the risk of developing epilepsy in children with ASD.

Funding: Funding: Not applicable.

References:
1. Kotulska K, Kwiatkowski DJ, Curatolo P, et al. Prevention of Epilepsy in Infants with Tuberous Sclerosis Complex in the EPISTOP Trial. Ann Neurol. 2021;89(2):304-314. doi:10.1002/ana.25956