Patterns of Epileptiform Activity and Clinical Response to Steroid Therapy in Autism

Background: Magnetoencephalography [MEG] reveals that a high percentage of children with an autism spectrum disorder [ASD] demonstrate epileptiform activity within the peri-sylvian region. The peri-sylvian region has also been shown to be the primary epileptic focus in Landau-Kleffner syndrome [LKS], an acquired epileptiform aphasia characterized by continuous spike wave activity in slow wave sleep. Several studies have shown that high doses of steroids can suppress epileptiform activity in LKS and lead to improvements in language, so there is growing interest in the possibility that steroids can be of benefit in autism.

Objectives: The current study sought to determine if specific patterns of epileptiform activity are predictive of the quality of clinical response to high-dose steroid therapy in the ASDs.

Methods: MEG data were retrospectively reviewed for 36 children [ages 3-10] with autism spectrum disorders and epileptiform EEG abnormalities who had been treated with high-dose prednisone. In 22 cases the MEG data that were available for review had been obtained 2-6 months prior to initiation of steroid treatment. In the remaining 14 cases the electrophysiological data had been obtained 2-24 months following steroid withdrawal. Based on a blinded review of clinical notes, standardized tests [as available], and parental interview, 20 of the 36 children were classified as steroid responders.

Results: In the comparison of MEG data from responders versus non-responders, the most predictive information about treatment responsivity came from the actual pattern of epileptiform activity. Specifically, the highest predictive value was seen for the following rule -- 'Subjects with epileptiform activity restricted to peri-sylvian, inferior-frontal, and peri-Rolandic regions were likely to respond to steroids, whereas those with activity outside of these regions were unlikely to respond to steroids.  This simple rule led to correct classification of 17/20 responders (85%) and 15/16 non-responders (94%).

Conclusions:   A subset of children with ASDs and epileptiform activity show transient positive improvements in language when they are treated with steroids. Unfortunately, there is often a significant regression in skills following treatment withdrawal [as demanded by the development of medical side-effects], although some children do maintain improvements after medication taper. MEG patterns of epileptiform activity help to identify those children most likely to show transient positive benefits from steroid therapy. Best responses are seen for male subjects without clinical seizures for whom epileptiform activity is restricted to peri-sylvian, inferior-frontal, and/or peri-rolandic regions. The present data suggest that, for some children with autism, language compromise is associated with an active process that can be controlled by appropriate medical management, with MEG helping to identify those children most likely to benefit from aggressive intervention with steroids. Even though steroids are obviously not a good long-term treatment strategy for children with ASDs, a better understanding of how steroids can effect positive changes in some children may provide critical insight into the nature of autism and may lead to safer, more effective therapeutic interventions.