Autism Biomedical Information Network

Autistic regression, epilepsy, epileptiform EEG, and Landau-Kleffner syndrome


Abstract of article by Tuchman and Rapin
Comment on abstract
Reference to article on Landau-Kleffner syndrome by Dr. Max Wiznitzer in
... The Advocate
Reference to review article on autism by Dr. Isabelle Rapin in the
... New England Journal of Medicine

Many parents are confused about the relationships among autistic regression, epilepsy, an epileptiform EEG, Landau-Kleffner syndrome, and whether or not treatment should include anticonvulsants or corticosteroids in children without overt seizures. The abstract of an important article by Drs. Tuchman and Rapin follows. The full-text article can be ordered through Loansome Doc. The article identifier is UI 97247171.


Title: Regression in pervasive developmental disorders: Seizures and epileptiform electroencephalogram correlates

Authors: RF Tuchman, I Rapin

ABSTRACT. Background. Approximately one third of the parents of children with pervasive developmental disorders or autistic spectrum disorders reports an early regression of unknown cause in their children's language, sociability, and play. Seizures or an epileptiform electroencephalogram (EEG) are associated with language regression in acquired epileptic aphasia (Landau-Kleffner syndrome) and some other pediatric epileptic syndromes. The importance of epilepsy or epileptic EEGs as contributors to autistic regression is not known.

Method. Subjects were 482 boys and 103 girls on the autistic spectrum seen consecutively in consultation by one child neurologist. Data on autistic regression, seizures, sleep EEGs, and cognitive function were entered prospectively into a data base.

Results. Of the 585 children, 176 (30%) had a history of regression, and 66 children (11%) had a history of epilepsy, defined as two or more unprovoked seizures. Among 392 children with available sleep EEGs, the EEG was epileptiform in 59% of the 66 epileptic children and 8% of the 335 nonepileptic children. Regression had occurred equally among children without seizures and in those with epilepsy. Regression was associated with an epileptiform EEG in 14% of 155 nonepileptic children who had undergone a regression, as opposed to 6% of 364 children with neither regression nor epilepsy. Mean age at regression was 21 months. There was no difference in the proportion of children with epilepsy or epileptiform EEGs who had regressed before or after 2 years of age. Approximately half of the epileptiform discharges were centrotemporal, whether or not the child was epileptic or had regressed. Children with lower cognitive function were more likely to have undergone regression than those with better cognitive skills (34% vs 20%).

Conclusion. Epilepsy or epileptiform EEGs occur in a significant minority of autistic children with a history of regression and in a smaller minority without regression. Prompt recognition of regression and recording of prolonged sleep EEGs is recommended, even though information on the potential efficacy of antiepileptic treatment to improve language and behavior in autistic children with epilepsy or an epileptiform EEG is still lacking. Pediatrics 1997;99:560-566; pervasive developmental disorders, autism, regression, seizures, electroencephalogram.

Abstract Copyright ©1997 by the American Academy of Pediatrics.

[End of abstract]

Comment on abstract

This is an important paper because autistic regression superficially resembles Landau-Kleffner syndrome (LKS, acquired epileptic aphasia). The full text of this article has an in-depth discussion of the differences between autistic regression and LKS.

(Also see the article on LKS by Dr. Max Wiznitzer in the Austism Society of America newsletter, The Advocate, November-December 1996).
Return to Overview

Since the EEG changes seen in LKS often respond to corticosteroid therapy (although language regression in LKS appears less responsive to treatment), parents of children with autistic regression have sought treatment with corticosteroids. In a recent excellent review of autism, Dr. Isabelle Rapin indicates that the possible therapeutic role of steroids remains uncertain.

(See Rapin I, Autism, New England J Medicine 1997 (July 10);337(2):97-104. This journal is available in some public libraries. A copy of the article may be ordered through Loansome Doc, National Library of Medicine. The article identifier is UI 97337878).
Return to Overview

As mentioned in the abstract above, the incidence of epilepsy was low, 11%. The frequency was the same in the subgroup with a history of regression compared to those without regression (12% vs 11%). There was also no significant difference in the proportion with an eleptiform EEG in the epileptic children with regression compared to those without regression (52% vs 62%). However, of even greater interest in the subgroup without epilepsy is that 14% of those with regression had an epileptiform EEG vs 6% without regression (for the entire study group). If the analysis is restricted to the subgroup with EEG data, but without epilepsy, the proportion with an epileptiform EEG is 19% in those with regression compared to 10% in those without regression. Thus it appears that even in those with regression, it is a relatively small, though significant, minority that have an eleptiform EEG. Also, it appears that regression is a risk factor for subclinical epilepsy. Since the mean age at evaluation was 70 months, these children, with or without regression, may be marked to develop overt epilepsy with the passage of years, accounting for the ultimate incidence of epilepsy of about one-third of young adults with autism.

Although centrotemporal localization of paroxysmal EEG changes is characteristic of LKS, only about one-half of this study group had such localization.

It is not yet clear that autistic individuals with apparent subclinical epilepsy should be treated with anticonvulsants, let alone corticosteroids (see review by Rapin cited above). The authors recommend that all children with autistic regression have a prolonged sleep EEG as soon as possible.

-- Ronald J. Kallen, M.D.

Loansome Doc requires a prior agreement with a local medical library and registration with the National Library of Medicine, which will issue a user ID and a password. See How to order full text of an article through Loansome Doc for more information.
Return to Overview

|Return to list of commentaries| |Links to other autism resources|

Comment above ©1999, This page last updated on 3/1/99