Autism spectrum disorders (ASD) are neurological and developmental disorders characterized by impairment of social relatedness, communication skills, restriction of interests and stereotyped behaviors¹⁾. ASD is believed to be related with central nervous system (CNS) dysfunction, and epilepsy is strongly associated. Epilepsy is reported to occur in one third of ASD patients but the exact prevalence is unknown with the literature reporting a wide range of estimates from 5²⁾ to 46%³⁾. In general population, epilepsy is reported in only 0.5-1%. The fact that the epilepsy rate is much higher in ASD patients suggests that ASD and epilepsy might share a common pathophysiological basis. This could also be supported by the reports of high rates of epilepsy or epileptiform electrocencephalography (EEG) in children with autism, and vice versa, as well as much higher prevalence of autism in children with epilepsy who also have learning problems. It is unclear if this suspicion is just an epiphenomenon of an underlying neural dysfunction in ASD or there is any causal relationship between ASD and epilepsy. Many authors have not reached a definite conclusion and there is no consensus on the neuropathophysiology of ASD.

In approximately 1/3 of ASD patients, the regression of language is observed between 18-24 months of age with the appearance of autistic behavior. In childhood disintegrative disorder (CDD), autistic regression occurs after 3 years of age, and mainly includes motor regression and a loss of bowel and bladder use. According to Kurita et al.⁴⁾, EEG abnormalities are significantly more common in CDD patients than infantile autism patients. However, according to Canitano et al.⁵⁾, an abnormal EEG is found usually after regression but there is no evidence of a causal relationship between them. Therefore, there is still controversy regarding the role of epilepsy or epileptiform EEGs in the regression of ASD patients. Currently, there has not been any controlled study examining the efficacy of treating epileptiform EEG discharges in ASD patients. The aim of this study was to examine the clinical and electroencephlography (EEG) characteristics of autistic spectrum disorder patients who later showed clinical seizures. An attempt was made to clarify the relationship between regression or any type of problematic clinical findings and severity of EEG abnormalities in ASD patients with late-onset epilepsy, by reviewing the serial EEG findings and underlying abnormalities of the background rhythms and paroxysmal discharges.

According to this study, 1/3 of ASD patients with late-onset epilepsy showed definite epileptiform paroxysmal discharges or a slow and disorganized background rhythm that might affect the cortical dysfunction. Those patients largely showed associated clinical problems, such as mood disorders, sleep disturbances, behavioral problems and a regression of language. On the follow-up, 80% of them (8 out of 10) showed EEG improvement after an average of 26.7±13.9 months (range 16-53 months), and 70% (7 out of 10) showed clinical seizure improvement. The result suggests that although the EEG abnormality was severe, it was treated well by AEDs in ASD.

Autism spectrum disorders (ASD) are neurological and developmental disorders characterized by an impairment of social relatedness, communication skills, restriction of interests and stereotyped behaviors¹⁾. ASD is believed to be related with central nervous system (CNS) dysfunction, and epilepsy is strongly associated, with a wide range of estimates from 5²⁾ to 46%³⁾. This high rate of epilepsy suggests that ASD and epilepsy might share a common pathophysiological basis. However, there is no consensus regarding the neuropathophysiology of ASD⁷⁾.

According to DSM-IV⁶⁾, ASD consists of five different subgroups. Asperger's syndrome is autism without mental retardation or delayed language development, with a reported epilepsy rate of 5-10% of early childhood⁸⁾. PDD-NOS includes patients with generally milder autism, that does not fit criteria for any other subtypes. Autistic disorder is classic autism with all three behavioral domains showing early unexplained regression before three years of age, maximally between eighteen to twenty-four months of age⁹⁾. Clinical epilepsy occurs in more than one third of children until adolescence. Childhood disintegrative disorder indicates more severe autism, and epilepsy has been reported in almost 70% of patients¹⁰⁾. Rett's syndrome is characterized by a specific X-linked genetic deficit, mutations of the MeCP₂ gene. In this group, the epilepsy rate is reported to be more than 90%¹¹⁾, which is the highest of all subgroups.

Many factors affect the variable rate of reported epilepsy rate in ASD. According to Tuchman and Rapin¹²⁾, there is a bimodal age distribution of seizure in ASD. The first peak is in infancy before 5 years of age, and the second peak is in adolescence, after 10 years of age¹¹⁾. In our study, the mean age for the onset of epilepsy was 7.1±4.9 years and there was no significant difference between the ASD subtypes. This is not precisely concordant with the reported studies but the patients selected in this study could be considered as a late-onset epilepsy group.

Epileptiform discharges are rare (1-4%) in healthy children13, ¹⁴⁾. Until recently, the reported rates of epileptiform discharges varied from 6-30% of ASD patients12, 13, 16, ¹⁷⁾. In this study, although all patients suffered clinical seizure events, the EEG results were variable and 23 out of 30 patients (76.6%) showed epileptiform EEG abnormalities. These EEG changes might be considered a biomarker of a cortical dysfunction in this population. There are no specific epileptiform EEG patterns known in ASD patients yet. These EEG changes can show the following: slowing, asymmetry, spikes or sharps, sharp and slow waves, generalized sharp and slow waves or generalized polyspikes in diffuse or generalized, multifocal or focal discharges, unilateral or bilateral and localized to many different brain areas3, 5, 18-²¹⁾. All seizure types can be associated with autism, which is consistent with our results. There is no report on the prevalence of CSWS pattern in ASD, and in this study 7 out of 30 patients (23.3%) showed CSWS.

Approximately, one third of ASD patients show a regression of language between 18-24 months of age, with the appearance of autistic behavior¹²⁾. Many studies focused on the potential association between epilepsy and autistic regression. Some reported higher rates of epilepsy in children with autism and regression but others showed no association between autism, epilepsy and regression. According to Tuchman and Rapin¹²⁾, about 20% of children with autistic regression, even without epilepsy, have an abnormal EEG, and Canitano et al.⁵⁾ reported that this abnormal EEG occurs after regression but still there is no evidence of a causal relationship between the epileptiform abnormality and regression. In this study, we could not confirm that each patient clearly showed autistic regression but many of them had associated neuropsychological problems, which was considered to indicate a tendency to regress. Although there was no statistical significance, there was slightly higher association with neuropsychological problems in group II, which included CSWS. These results suggest that if CSWS is accompanied in late-onset epilepsy in ASD, it might cause epileptic encephalopathy, as in Landau-Kleffner syndrome, and may be a reason for the neuropsychological problems including cognitive impairment. There is no report that clarify the relationship between CSWS and autistic regression. According to Nickels et al.²²⁾, CSWS is considered distinct epileptic encephalopathy with common clinical features including seizures, regression, and epileptiform abnormalities that are activated by sleep. It is thought that CSWS is related with regression of global skills and can show behavioral phenotype that overlaps with autism²³⁾. In opposite direction, this can suggest that somehow autism and regression have possible relationship with CSWS. As most ASD patients already have impaired communication skills, this study could not confirm whether there was language regression. However, in Group II, even Rett syndrome patients showed unusual regression as well as emotional and behavioral disturbances. Overall, the relationship between the occurrence of regression and epileptiform EEG abnormalities is unclear.

Although epileptiform EEG abnormalities are much more common in children with ASD, whether we should treat these abnormalities even if the patient does not show clinical seizure is a controversial problem. Generally, interictal discharges are thought to interfere with normal neural processing. In addition, attentions are paid to the frequent cognitive and behavioral disturbances in many epilepsy syndromes²⁴⁾. For example, in Benign Epilepsy with Centro-Temporal Spikes (BECTS), the academic problems related to the frequency of epileptiform discharges and attention problems largely improved after EEG had normalized25, ²⁶⁾. In many cases of Landau-Kleffner Syndrome (LKS), both the seizures and language impairment improve with normalization of the EEG¹⁾. According to these reports, the treatment of EEG abnormalities might have a positive effect on the symptomatic improvement of children with ASD and epileptiform EEGs. Many studies reported specific improvements in ASD symptoms after therapy to suppress the discharges with AEDs27-²⁹⁾.

Even with these promising reports, it is difficult to make a decision as to whether to treat the epileptiform EEG abnormalities if an autistic child with no clinical seizure events has a history of regression.

In this study, the patients experienced clinical seizures some time after the diagnosis of ASD. Some of them showed nothing remarkable in the EEG but others showed definitely severe epileptiform paroxysmal discharges, such as CSWS or a slow and disorganized background rhythm that might affect the cortical dysfunction. In addition, those patients largely showed associated neuropsychological problems, which were considered to be a marker of regression or disintegration of the patient's cortical function. As every patient showed a definite clinical seizure, they were treated with anti-epileptic drugs. The follow-up revealed, 18 out of 30 patients (60%) to be seizure free with only 1 patient showing seizure aggravation. Therefore, late-onset epilepsy in ASD patients appears to be treated well by AEDs. Even in the severe EEG group, the EEG results and clinical progress improved. However, as formal cognitive function tests were not performed in all patients, and as matched untreated children were not included, it cannot be concluded that the treatment of EEG abnormality was related to the improved neuropsychological symptoms. In addition, these improvements might be due to the natural history of the disorder itself. However, these results suggest that there is some support for such an assumption.

As this study was a retrospective study and included a small number of patients at one institute, these patients might not reflect all autistic children. Therefore, a prospective, placebo-controlled, double-blinded study will be needed to further test this hypothesis.

Despite the limitations of this study, these results suggest an early intervention for seizure control when an autistic patient shows clinical seizure or neuropsychological problems. In this case, EEG should be performed to confirm if there are any epileptiform EEG abnormalities. It is natural to treat EEG abnormalities with clinical seizures. However, although there is still some controversy regarding the treatment of epileptiform EEGs without clinical seizures, as it is believed that a large portion of these ASD patients already have intellectual disability, as they show clinically problematic findings, such as seizures and maladjusted behavior without their own expressions, it appears that more should be done to improve their quality of life by screening for any type of epileptiform EEGs that can affect their brain function and treating them if possible.