Background
Angelman syndrome (AS) is a neurodevelopmental disorder caused by an absent or non-functioning maternal allele of chromosome 15q11-q13 [
1]. The typical AS phenotype is characterized by intellectual disability (ID), lack of speech, hyperactivity, ataxic gait, microcephaly, sleep disturbances, frequent laughter/smiling and an apparently happy demeanour [
1‐
4]. ID ranges from moderate to profound, with most individuals functioning in the severe to profound range [
5,
6]. Epilepsy occurs in 80% or more of cases [
2,
7], usually involving multiple seizure types and starting in early childhood [
7,
8]. High rates of autistic symptoms are also reported [
9‐
11], with prevalence estimates of autism spectrum disorder (ASD) ranging from 24 to 81% [
6,
10]. AS can be due to
UBE3A mutations, uniparental disomy and imprinting defects [
1,
12], but deletions are the predominant cause and are found in 68–75% of patients. Deletions are also associated with more severe AS-phenotype, and co-deletion of GABA
A-receptor genes (
GABRB3,
GABRA5 and
GABRG3) located adjacent to
UBE3A gene is suggested as a possible explanation for this [
1]. Dysfunction of
GABRB3 is highly associated with both epilepsy and autism symptoms [
13,
14].
A strong association between autism symptoms, epilepsy and ID has been found in a number of other genetic syndromes, such as fragile X and tuberous sclerosis complex (TSC), as well as in AS [
6,
10]. It is evident, too, that the negative effect of seizures is particularly strong during infancy and early childhood [
15‐
18]. Thus, onset of seizures during the first year of life is associated with increased prevalence and severity of ID and ASD and increased prevalence of brain abnormalities [
19,
20]. However, there is a continuing debate [
21‐
24] as to whether autism symptoms, epilepsy and ID are independent comorbidities [
15,
16,
21,
25‐
27], whether they are all outcomes of the same underlying pathophysiological/genetic mechanisms [
17,
21,
25,
28], or whether the epilepsy itself contributes to more severe cognitive and behavioural impairments than might be expected from the underlying pathology alone [
15,
17,
29,
30], i.e. a so-called encephalopathic effect [
30].
There are several reasons why AS offers a suitable disease model to investigate the association between epilepsy, ID and autism symptoms. Firstly, the rate of epilepsy in AS (> 80%) is as high as or higher than other genetic disorders in which epilepsy and autism commonly co-occur (e.g. TSC [80–90%]; fragile X syndrome [10–20%]) [
29,
31,
32]. Secondly, epilepsy in AS tends to start in very early childhood. Seizures are also often treatment-resistant and refractory epilepsy has been shown to be an important predictor of autism symptoms [
33]. Thirdly, unlike genetic conditions such as TSC, in which the numbers and location of tubers are associated with autism symptoms [
17,
34], there are no specific structural brain abnormalities in AS that are known to affect the phenotype. Fourthly, knowledge of the specific genetic defects that cause AS makes it possible to evaluate the degree to which the association between epilepsy and autism symptoms is a result of the underlying genetic abnormality and to assess the independent contribution of seizures on level of autism symptoms.
The aims of the current study were to describe epilepsy characteristics and then investigate the relationship between epilepsy, autism symptoms, communication level and genetic cause in individuals with AS. Based on previous research on other populations with childhood epilepsy including TSC [
18,
33,
35‐
37], we hypothesized that age of onset of epilepsy would be related to the number of autism symptoms in AS independent of the effect of the specific genetic abnormality.
Discussion
This study explored the relationship between age of epilepsy onset, autism symptomatology, type of genetic aberration and nonverbal communication level in a Norwegian sample of individuals with AS. Among the 56 individuals with AS identified from the available databases, 48 (86%) had genetically verified AS. This is in line with other reports noting that no genetic abnormality can be identified in 10–15% of individuals with AS [
4]. Other clinical findings were similar to those of previous studies of AS. Thus, deletions were the most common genetic cause identified [
1,
4]. With regard to epilepsy, the prevalence in this study was 77%, somewhat lower than the rates of ≥ 80% commonly reported [
4,
7,
8,
41]. However, our sample included several very young participants who may not yet have had their first seizure. We also excluded individuals in whom the cause of AS was unknown and there is some indication that individuals with AS of unknown cause may have the highest prevalence of seizures [
7]. Epilepsy characteristics with early-onset epilepsy, multiple seizure types, a tendency to have seizures during febrile episodes and commonly treatment-resistant seizures, particularly in early childhood, are also in line with the findings reported by others [
2,
7,
8,
41,
42], and the use of anti-epileptic drugs is comparable to other studies [
7,
8,
41].
The main focus of the study was the association between age of epilepsy onset and extent of autism symptomatology when type of genetic abnormality was controlled for. Our findings from this study of individuals with AS provide support for the notion that seizures themselves contribute more to autism symptoms than might be expected from the underlying pathology alone [
15‐
17,
21]. As anticipated, individuals with a deletion of 15q11-q13 had substantially more autism symptoms than individuals with other genetic aberrations (
g = 1.48). However, when entered into a regression model with epilepsy onset, genetic aberration made no significant contribution to the number of autism symptoms reported. Although the lack of an independent effect of type of genetic aberration is likely due to the low number of causes other than deletion, it should be noted that the slope of the regression lines is similar for both genetic subgroups, thus supporting the importance of age at seizure onset across the sample. These findings from AS parallel evidence from studies in other rare disorders such as TSC; although both early seizures and encephalopathy are highly associated with type of genetic abnormality, early seizures may contribute to a worsening of developmental outcome [
17,
43]. Similarly, from fragile X syndrome, research indicates that males with the FMR1 premutation are more likely to have ASD and ID if seizures occur in childhood [
29,
44].
Although individuals with epilepsy had more autism symptoms than those without epilepsy, and despite a moderate to large effect size, this difference was not significant [
15]. This may be due to the rarity of non-epilepsy cases among individuals with AS and hence the very small size of the no-epilepsy group. However, the findings also point towards the importance of viewing epilepsy as a spectrum disorder rather than a dichotomy [
15]. Hence, the comorbidity between autism symptoms and epilepsy may be related both to the underlying pathology and to the effect of seizures. The high risk of ASD in populations with early-onset epilepsy has been used to support the encephalopathy hypothesis, i.e. that seizures may cause ASD [
16,
25]. Others have argued against this because the relationship is bi-directional and individuals with ASD are at increased risk of future epilepsy and seizures may occur in adolescence or adulthood [
21,
22,
45,
46]. This study highlights the importance of considering the additive effects of the underlying genetic aetiology and seizures contributing to autism symptoms in AS, which may be relevant also for other conditions [
15,
29]. The encephalopathic effect may be greater when seizures start early. Early-life seizures may result in molecular changes which impact neural network structure, and the hippocampal region may be of particular importance. Molecular changes may also influence the expression of genes involved in autism symptoms and genetic syndromes such as
GABRB3,
FMR1,
TSC1 and
TSC2 [
16,
29]. Moreover, research suggests that effects of seizures on GABA
A-receptor expression are age-dependent, a finding that further supports the notion that early seizures are particularly harmful [
16].
There was no difference in the level of nonverbal communication between the epilepsy group and no–epilepsy group. Age of first seizure however, was associated with nonverbal communication (
g = 0.56) and individuals with the lowest level of nonverbal communication had earlier seizure onset than those who used more signs to communicate. A number of other studies has found that earlier age of seizure onset is associated with poorer cognitive outcome [
18,
33,
35‐
37,
47,
48]. Our study did not include a measure of development, only a measure of nonverbal communication. However, supplemental analysis showed that age of epilepsy remained significant also when nonverbal communication was entered as a covariate. This suggests that the number of autism symptoms was not explained only by the level of nonverbal communication.
Although the findings of this exploratory study have potentially important implications for understanding the complex links between autism symptoms and epilepsy, there are a number of limitations that must be taken into account in the interpretation of the data. Firstly, the sample size was small and the age of participants was very wide, ranging from infancy to adulthood. In addition, we did not have data on the level of ID, only an estimate of nonverbal communication was available. There were also few individuals with a genetic cause other than the 15q11 deletion, and we lacked data on size of deletions. Furthermore, information from medical records was often incomplete and formal seizure classification, except for tonic-clonic seizures, was rarely performed. Hence, some individuals may have had more types and higher frequency of seizures than reported (particularly those of short duration or less severe such as absences and myoclonic seizures). Finally, there was no clinical assessment of autism, and rather than a categorical distinction between ASD/non-ASD, we focused on the frequency of autism symptoms as measured by the SCQ. While this avoided the problems of misdiagnosing ASD in a population with severe developmental delay, it is well established that the number of autism symptoms is highly related to severity of ID [
11]. Thus, high rates of autism symptoms were to be expected in this sample of individuals with AS [
9,
10]. The severity of ID in AS is the main limitation when using this disorder as a disease model for studying the relation between autism symptoms and epilepsy.
It is clear that information from a larger sample of individuals with AS, with a larger range of genetic causes other than deletions, and detailed information on developmental level is needed to increase confidence in the current findings. More details of the genetic aberration, such as size and exact break points of the deletions, are also needed. Finally, further studies in this area should investigate which autism symptoms are particularly vulnerable to early seizures and which are less affected. Such knowledge may be of relevance for better understanding of the biology of ASD.