JLE

Epileptic Disorders

MENU

Epilepsia, epileptiform abnormalities, non-right-handedness, hypotonia and severe decreased IQ are associated with language impairment in autism Volume 9, supplement 5, Supplement n°1, December 2007

Auteur(s) : Hana Oslejskova1, Ladislav Dusek2, Zuzana Makovska1, Ivan Rektor3

1Department of Pediatric Neurology, Brno Epilepsy Centre, Children’s Medical Center, University Hospital, Brno
2Institute of Biostatistics and Analyses, Masaryk University, Brno
3Department of Neurology and Neuroimaging, Brno Epilepsy Centre, St. Anne’s University Hospital, Brno, Czech Republic

Pervasive developmental disorders (PDD) alias autistic spectrum disorders (ASD) are neurodevelopmental disturbances and encompass a heterogeneous group of individuals with multiple cognitive/behavioral impairment presenting with onset symptoms early in childhood. Diagnosis of ASD (besides Rett syndrome) is made at the syndromological level. Recently, a strong genetic background has been described (Rutter 1999; Muhle et al. 2004). However, etiology and pathogenesis of these disorders remains probably multifactorial. Nearly all patients suffer from some degree of verbal impairment since disturbances in verbal communication are one of the core symptoms in autistic disorders. Expressive language function across the autistic spectrum ranges from complete mutism to verbal fluency. Problems in the structure of language can also occur (phonology and syntax) similar to those in developmental language disorders (DLD) or specific language impairments (SLI) (Rapin and Dunn 2003; Allen and Rapin 1992; Folstein and Mankoski 2000; Bishop 2003). Verbal fluency in children with ASD is often accompanied by a number of semantic and verbal pragmatic errors. Single types often combine together. The aim of this study is to contribute to an explanation of complex factors leading to substantial language impairment in autistic patients. Although the level of functional impairment and severity of symptoms remain the strongest factor of language impairment, clinical experience indicates that these factors do not possess sufficient predictive power. One can observe children with relatively mild clinical autistic symptoms but significantly impaired speech abilities and vice versa. Furthermore, we were interested in the possible correlation of epilepsy and epileptic activity with language impairment in autistic disorders because their precipitating influence on the possible induction of autistic disorders has been postulated (Tuchman 1999; Gabis et al. 2005). Finally, an analogous clinical phenomenon is thought to be speech impairment in several syndromes such as Landau Kleffner, etc. (Ballaban-Gil and Tuchman 2000; Tomblin et al. 2003).

The aims of our study were: i) to define a position of speech-related impairment among the other markers of patient health status, ii) to define speech-related principal end-points that reasonably describe patient status and respect the already given typology of autistic disorders and the patient’s functionality, and iii) to examine factors that can influence the process of speech impairment and to classify their impact, especially the role of epilepsy.

Subjects and methods

Two hundred and five children have been diagnosed with ASD and followed up at the Department of Pediatric Neurology Univ. Hospital Brno from 1999 to April 2006. Their clinical and demographical characteristics are given in table 1. Children were diagnosed as being on the autistic spectrum on the basic of their history and by clinical observation by parents, teacher, psychiatrist, psychologist and child neurologist of qualitative impairments in social interactions, verbal and nonverbal communication, and a restricted, repetitive and stereotyped pattern of interest and behaviors applying the ICD-10 classification of World Health Organization (MKN-10 revision 1992, ICD-10 1993) and Childhood Autism Rating Scale (CARS) questionnaire (Schopler et al. 1980) and Childhood Asperger Syndrome Test (CAST) questionnaire (Scott et al. 2002). IQ was tested using the Gesell Developmental Scales for the youngest children and the Stanford-Binet Intelligence Scale, 4th Edition, for older children. Using a modified De Meyr’s classification we grouped our patients into high-functioning autism, medium functioning autism and low-functioning autism groups (De Myer et al. 1981). In this classification, the principal categorizing discriminant is the ability of a patient to cope with real life situations.

A comprehensive family and personal history of each patient, including perinatal risk factors, with a specific focus on epileptic disorders in the proband and his/her family was taken. Each patient was examined by a neurologist, a psychologist and a psychiatrist. Laboratory functional studies involved recoding of 24- or 32-channel EEG, EEG after sleep deprivation, and neuroimaging studies (MRI or CT or both). Where indicated on the basis of clinical suspicion, the respective patient was seen by a specialist in medical genetics and was screened for fragile X, inborn errors of metabolism, tuberous sclerosis and Rett syndrome.

Categorization of speech impairment was performed in collaboration with a speech therapist and a specialist in phoniatrics with respect to: i) difficulty with structural aspects of language (phonology and syntax), ii) delayed development of speech, iii) abnormal use of words and phrases (abnormal use of words and phrases with idiosyncratic meanings, iv) use of made-up words - neologisms, v) failure to use contextual information in comprehension, vi) dysphatic symptoms, vii) complete mutism (no development of speech, severe language impairment), viii) marked impairment in language comprehension, ix) verbal auditory agnosia-like symptoms, x) immediate or delayed echolalia or scripts, and xi) regression of speech.
Table 1 Patients’ characteristics (n = 205).

Parameters

Initial values*

Male sex (boys)

n = 145 (70.7%)

Female sex (girls)

n = 60 (29.3%)

Age (yrs)

10 (5; 15)

Characteristics of autism

Age of distinguishing first symptoms (months)

30 (4; 60)

Age at diagnosis (yrs)

7 (2; 12)

Childhood autism

n = 127 (62.0%)

Atypical autism

n = 57 (27.8%)

Asperger’s syndrome

n = 21 (10.2%)

Autism with regres

n = 71 (34.6%)

Age of regres (months)

24 (14; 48)

Autism: functionality

High

n = 75 (36.6%)

Medium

n = 79 (38.5%)

Low

n = 51 (24.9%)

CARS

38 (32; 48)

IQ

55 (15; 104)

Handedness

Right handed

n = 82 (40.0%)

Non-right handed

n = 123 (60.0%)

Follow-up of family history

Abnormalities in family history

n = 89 (43.4%)

Epilepsy

n = 19 (9.3%)

Psychiatric disorders

n = 47 (22.9%)

Diagnostics autism

n = 4 (1.9%)

Social patology

n = 9 (4.4%)

Neurologic and psychologic patology

n = 13 (6.3%)

Genetic

n = 12 (5.9%)

Migraine

n = 8 (3.9%)

Follow-up of personal history

Prenatal risk factors

n = 94 (45.9%)

Perinatal risk factors

n = 80 (39.0%)

Postnatal risk factors

n = 80 (39.0%)

Follow-up of epileptic disorders

Epileptic seizures

64 (31.2%)

Age of first seizure (months)

20 (4; 85)

Abnormal finding on examination

Abnormal neurologic examination

n = 153 (74,6%)

Hypotonia

n = 32 (15.6%)

Spasticity (cerebral palsy)

n = 45 (21.9%)

Gross motor deficits (clumsy and uncoordinated movements)

n = 132 (64.4%)

EEG abnormal background

n = 115 (56.0%)

EEG nonepileptiform abnormalities

n = 64 (31.2%)

EEG epileptiform abnormalities

n = 98 (47.8%)

EEG subclinical epileptiform discharges

n = 24 (15.8%)

CT

n = 48 (23.4%)

MRI

n = 74 (36.1%)

CT and MRI

n = 87 (42.4%)

Genetic

n = 24 (11.7%)

Screens for Inborn Errors of Metabolism

n = 5 (2.4%)

Hearing impairment

n = 12 (5.85%)

Optical impairment

n = 54 (26.3%)

*Continuous variables are summarized as median with 10-90% percentiles (in parenthesis).

Statistical analysis

All statistical tests were performed on the intention-to-treat principle, and no case was excluded prior to the analyses. A value of α < 0.05 was taken as a universal limit for statistical significance. Standard robust descriptive measures were used to express differences among subgroups of cases (median supplied with 10%-90% percentiles; relative frequency). The ML chi-square test was applied to study associations among binary or categorical outcomes. The Robust Mann-Whitney U-test was applied to estimate differences between subgroups of patients in continuous variables. Both univariate and multivariate approaches were adopted to examine the contribution of risk factors to specified speech-related endpoints. Logistic regression was used for analyzing relationships between one or more variables and the speech impairment measure, coded at two levels (0/1, 1 always being the risk event). The odds ratio supplied with 95% confidence limits was estimated on the basis of logistic regression models (Altman 1991; Zar 1984).

Results

The results of our study are presented in tables 2 to 7, which are organized in a consecutive and iterative manner reflecting processes of statistical computation and taking into consideration the stepwise character of results processing. Table 2 shows that the recruited sample evidently covered all important types of autism and degrees of functionality advantageous for the robustness of subsequent conclusions presented in tables 3 to 7. The examined field was extremely heterogeneous, necessitating definition of several principal endpoints related to various degrees of speech impairment. All descriptive parameters defined in table 1 were independent and therefore several categories could have been effectively combined together. We aggregated subtypes of autism and associated degree of functionality into four categories (table 2) based on significance of their correlation. In our cohort, Asperger’s syndrome (AS) was classified as “highly functional” (n = 21) while, on the other side, low functionality was detected only in the group Childhood Autism (CHA) (n = 51). Finally, two categories of medium (n = 79) and high (N = 54) functionality were defined in the Atypical Autism group (AA) and CHA. These four categories of autistic children provided a significant gradient of growing risk that was at least partially reflected in speech-related measures. Nevertheless, the functionality and subtype of autism could not completely explain the growing risk trends and therefore, we hypothesized about the influence of other factors as well. Subsequent distribution analyses specifically revealed the three following categories of speech impairment associated with increasing severity of autism, i.e. from high to low functionality. The selected three impairment categories included the whole spectrum of examined autistic children and were therefore employed as key endpoints in further analyses. First, “delayed development of speech” being a problem of relatively mild severity, occurring even in the group of highly functional patients but with incidence > 92% in autistic children with decreased functionality. Second, “complete mutism (no development of speech, typically severe impairment of speech)” being a severe problem with remarkable incidence in groups with medium and low functionality. It was defined as a measure of severe problems to be examined only within the group of patients with decreased functionality. Third, “regression of speech” was defined as clinically important but a specific problem that did not correlate with the other speech-related measures. Other speech-related measures (shown in table 2) were not included in the multivariate analyses because they were found redundant as related to the selected endpoints (correlation analysis, not shown). Selected endpoints formed a three-dimensional space representative for speech disorders of varying severity. “Delayed development of speech” and “regression of speech” were diagnosed in various patients with different functionality thus standing rather in the central position as measured by behavioral and IQ scoring (data not shown). On the other hand, “Severe impairment of speech” aggregated disorders clearly related to decreased functionality and IQ. Then, the key question for further analyses was the multivariate typology of patients bearing these three types of impairment. Next, we adopted multivariate stepwise procedure to select only risk factors that can independently contribute to the increasing risk of speech impairment presented in tables 3, 4, and 5. Comparing outcomes from tables 3 to 5 we observed that both “delayed development of speech” and “complete mutism” were associated with a number of different factors while only a very limited number of factors was found to correlate significantly with “regression of speech”. This observation confirmed a position of this problem already mentioned – it was also found to correlate neither with the type of autism nor with the degree of functionality. The profile of contributing risk factors in tables 3 and 4 is somewhat redundant because many of them are mutually correlated.

The outcomes of multivariate models are further presented in table 6 showing logistic models based on potential risk factors. The significant position of epilepsy among potential risk factors associated with speech impairment in children with autism emerged from computational models presented in tables 3 to 6. Although measures indicating that certain epileptic elements were included in all models, the type of a specific epileptic element was different in different speech-related endpoints. In table 6A and B, a profile of risk factors related to the “delayed development of speech” and “complete mutism” was relatively similar with the remarkable position of several neurological abnormalities and heavily decreased IQ < 35 points. None of these parameters were included in the final model built for a “regression of speech” (table 6C). All models included certain factors related to epileptic disorders, namely infantile spasms, EEG epileptiform abnormalities and non-right-handedness: these factors dominated in univariate models as well. Non-right-handedness and coincidence of epilepsy were found to be remarkably frequent in the whole sample of autistic children (table 1). The results shown in table 6 thus confirmed the specific contribution of these factors to the decreased speech ability of children. Finally, table 7 gives a detailed description of epilepsy and EEG findings as speech-related risk factors. Active epileptic seizures contributed to the increasing risk of “severe impairment of speech (complete mutism)” and “delayed speech development” while epileptiform abnormalities found in EEG were related to “regression of speech”. Comparing groups with and without specified speech problem within the cohort of epileptic patients revealed the following significant facts – “delayed development of speech” was proved to be associated namely with epileptic seizures – in table 7 we can see even more the increased occurrence of epileptic syndromes in children with this disorder – it was proved namely for generalized epileptic seizures (tonic and infantile spasms) and syndromes, background abnormalities and non-epileptiform abnormalities in EEG (table 7); “complete mutism” was selected as the endpoint relevant for severe subtypes of autistic disorders (table 2) and multivariate logistic models found its association with active epilepsy – i.e. with epileptic seizures. Table 7 proved this fact also within the cohort of children with epilepsy – “complete mutism” was specifically associated with several types of generalized seizure (myoclonic, tonic and infantile spasms), further with generalized epileptic syndromes and generalized epileptiform abnormalitis as well as background abnormalities and non-epileptiform abnormalities. “Regression of speech” was the only one from the selected key endpoints that was associated only with a very limited number of potential risk factors, one of them being proved focal EEG epileptiform abnormalities (tables 5 and 6). No specific association was found with active epileptic seizures (data not shown). Based on these results we conclude that the regression of speech is a core symptom in autism.
Table 2 Speech-related measures within diagnostic categories of autistic children.

1

2

3

4

5

6

7

8

9

Categories of patients according to type of autism and functionality

Difficulty with structural aspects of language

Dysphatic symptoms

Delayed development of speech

Abnormal use of words and phrases

Complete mutism (no development of speech)

Marked impairment in language comprehension

Symptoms of verbal auditory agnosia

Immediate or delayed echolalia or scripts

Regression of speech

AS/HFA (n = 21)

28.6%a

0%a

19.1%a

4.8%a

0%a

0% a

0%a

19.1%a

0%a

AA + CHA/HFA (n = 54)

64.8%b

11.1%a

79.6%b

53.7%b

0%a

35.2% b

11.3%a

38.9%b

27.9%b

AA + CHA/MFA (n = 79)

55.7%b

5.1% a

96.2%c

50.6%b

18.9%b

69.6% c

31.2%b

44.3%b

21.6%b

CHA/LFA (n = 51)

9.8%c

3.9%a

92.2%c

11.8%a

80.4%c

100.0% d

88.2%c

13.7%a

35.3%b


Table 3 Risk factors association with parameter I. “Delayed development of speech”*.
  • Risk factor and its
  • incidence (n)


Relative part of patients with “delay development of speech”

Odds ratio

(95% conf. limits)

p level

IQ

IQ < 35: n = 56

98.2%

16.32 (7.91; 33.31)

p < 0.001

IQ ≥ 35: n = 149

77.2%

Global developmental delay and mental retardation

No: n = 92

69.6%

6.60 (2.71; 16.14)

p < 0.001

Yes: n = 113

93.8%

Spasticity (cerebral palsy)

No: n = 160

80.0%

3.50 (1.01; 12.11)

p = 0.048

Yes: n = 45

93.3%

Abnormal neurologic examination

No: n = 103

76.7%

2.51 (1.15; 5.48)

p = 0.021

Yes: n = 102

89.2%

Optical impairment

No: n = 151

79.5%

3.22 (1.08; 9.69)

p = 0.037

Yes: n = 54

92.6%

Epileptic seizures

No: n = 141

79.4%

2.52 (1.19; 5.2)

p = 0.036

Yes: n = 64

90.6%

EEG epileptiform abnormalities

No: n = 107

77.6%

2.29 (1.04; 4.98)

p = 0.038

Yes: n = 98

88.8%

CT and MRI abnormal finding

No: n = 118

78.8%

2.07 (1.01; 4.24)

p = 0.043

Yes: n = 87

88.5%

Handedness

Right handed: n = 136

78.7%

2.84 (1.12; 7.27)

p = 0.018

Non-right handed: n = 69

91.3%

*End-point defined for the whole population of patients with autism (n = 205; see table 2).

Only factors with statistically significant association are included. Odds ratio calculated on the basis of univariate logistic regression with 95% confidence limits and level of statistical significance.


Table 4 Risk factors association with parameter II. ”Complete mutism (severe language impairment, no development of speech)”*.

  • Risk factor and its
  • incidence (n)


Relative part of patients with severe language impairment (complete mutism)

Odds ratio

(95% conf. limits)

p level

IQ

IQ < 35: n = 55

67.3%

6.06 (2.79; 13.21)

p < 0.001

IQ ≥ 35: n = 75

25.3%

Age at onset of autism

< 1 yr: n = 35

60.0%

2.57 (1.15; 5.73)

p = 0.021

≥ 1 yr: n = 95

36.8%

Global developmental delay and mental retardation

No: n = 41

21.9%

3.98 (1.69; 9.37)

p < 0.001

Yes: n = 89

52.8%

Spasticity (cerebral palsy)

No: n = 95

37.9%

2.19 (1.09; 4.39)

p = 0.045

Yes: n = 35

57.1%

Abnormal neurologic examination

No: n = 52

30.8%

2.37 (1.12; 4.99)

p = 0.024

Yes: n = 78

51.3%

Hypotonia

No: n = 108

38.9%

2.75 (1.05; 7.18)

p = 0.035

Yes: n = 22

63.6%

Optical impairment

No: n = 88

35.2%

2.70 (1.26; 5.79)

p = 0.011

Yes: n = 42

59.5%

Epileptic seizures

No: n = 78

33.3%

2.73 (1.31; 5.66)

p = 0.007

Yes: n = 52

57.7%

Age at onset epileptic seizures

< 1 yr: n = 25

72.0%

2.29 (1.05; 4.97)

p = 0.047

≥ 1 yr: n = 27

44.4%

EEG epileptiform abnormalities

No: n = 57

33.3%

2.07 (1.01; 4.21)

p = 0.046

Yes: n = 73

50.7%

Handedness

Rihgt handed: n = 74

31.1%

3.18 (1.53; 6.62)

p = 0.002

Non-right handed: n = 56

58.9%

*End-point defined only for patients with middle functioning autism and low functioning autism (n = 130).

Only factors with statistically significant association are included. Odds ratio calculated from uni-variate logistic regression models with 95% confidence limits and level of statistical significance.


Table 5 Risk factors association with parameter III. “Regression of speech”*.

  • Risk factor and its
  • incidence (n)


Relative part of patients with regres of speech

Odds ratio

(95% conf. limits)

p level

Handedness

Right handed: n = 136

19.9%

2.85 (1.35; 6.01)

p = 0.005

Non-right handed: n = 69

33.3%

EEG epileptiform abnormalities

No: n = 107

18.8%

1.49 (1.02; 2.21)

p = 0.040

Yes :n = 98

31.2%

Epileptic seizures before genesis (or identify) of autism

No: n = 177

22.9%

2.28 (1.01; 5.18)

p = 0.048

Yes: n = 28

39.3%

*End-point defined for the whole population of patients with autism (n = 205).

Only factors with statistically significant association are included. Odds ratio calculated from uni-variate logistic regression models with 95% confidence limits and level of statistical significance.


Table 6 Principal speech-related end-points in the multivariate stepwise logistic regression models*.

A. Delayed development of speech

Parameters included

Coefficient (SE)

Model Log-Likelihood

Log-Likelihood Ratio Test

Odds ratio (95% conf. limits)

All diagnostic categories of patients (n = 205)

Null model

0.251

(0.318)

-187.4

Step 1. IQ < 35

1.520

(0.485)

-170.1

0.014

4.57

(1.77; 11.83)

Step 2. Global developmental delay and mental retardation

1.327

(0.474)

-159.3

0.001

3.77

(1.48; 9.62)

Step 3. Epileptic seizures

0.856

(0.406)

-154.5

< 0.001

2.35

(1.06; 5.22)

Step 4. Handedness “non-right handed”

0.621

(0.309)

-149.7

< 0.001

1.86

(1.02; 3.41)

ARRAY(0x369ee4) 

B. Severe impairment of speech (complete mutism, no development of speech)

Parameters included

Coefficient (SE)

Model Log-Likelihood

Log-Likelihood Ratio Test

Odds ratio (95% conf. limits)

Patients with medium and low functionality (n = 130)

Null model

-2.975

(0.645)

-151.9

Step 1. Handedness “non-right handed”

1.643

(0.571)

-139.3

0.005

5.16

(1.67; 16.04)

Step 2. IQ < 35

1.528

(0.479)

-127.3

< 0.001

4.61

(1.79; 11.92)

Step 3. Hypotonia

1.252

(0.606)

-120.1

< 0.001

3.50

(1.05; 11.63)

Step 4. Epileptic seizures

1.109

(0.491)

-113.4

< 0.001

3.03

(1.15; 8.03)

ARRAY(0x370918) 

C. Regression of speech

Parameters included

Coefficient (SE)

Model Log-Likelihood

Log-Likelihood Ratio Test

Odds ratio (95% conf. limits)

All diagnostic categories of patients (n = 205)

Null model

-1.909

(0.342)

-205.5

Step 1. Handedness “non- right handed”

1.025

(0.372)

-198.1

0.007

2.78

(1.31; 5.89)

Step 2. EEG epileptiform abnormalities

0.365

(0.179)

-193.1

0.001

1.44

(1.01; 2.04)

*Multivariate stepwise procedure was driven only by statistical measures (log-likelihood function).

Odds ratio associated with variables entered in multivariate models as independent predictors

Base used for specified models was given by incidence of specified end-point within diagnostic categories of patients (see table 2).


Table 7 Detailed description of epilepsy and EEG findings as speech-related risk factors.

Epilepsy and EEG findings

Delayed development of speech

Complete mutism (no development of speech)

Regression of speech

Yes: n = 92/No: n = 11

Yes: n = 39/No: n = 64

Yes: n = 29/No: n = 74

Epileptic seizures

Partial seizures

Simple partial seizures

18.5%/27.3%

25.6%/15.6%

13.8%/21.6%

Complex partial seizures

28.3%/36.4%

23.1%/29.7%

27.6%/27.0%

Partial, evolving into secondary generalized seizures

28.3%/36.4%

25.6%/31.3%

31.0%/28.4%

Generalized seizures

Absence

6.5%/0%

10.3%/3.1%

3.5%/6.8%

Myoclonic

9.8% / 0%

15.4%/4.7%*

6.9%/9.5%

Clonic

3.3%/9.1%

7.7%/1.6%

7.0%/1.9%

Tonic

15.3%/0%*

23.1%/3.1%*

6.9%/12.1%

Atonic

11.9%/9.1%

12.8%/10.9%

17.2%/9.5%

Tonic-clonic

2.2%/9.1%

2.6%/3.1%

0%/4.0%

Infantile spasms

16.2%/0%*

25.6%/1.6%*

13.8%/9.5%

Unclassified seizures

15.1%/25.2%

17.9%/14.1%

10.3%/17.6%

Epileptic syndromes

Localization-related

30.4%/27.3%

23.1%/34.4%

27.6%/31.1%

Generalized

25.0%/9.1%*

46.2%/9.4%*

24.1%/22.9%

Epilepsies undetermined whether focal or generalized

4.3%/0%

2.6%/4.7%

0%/5.4%

Special syndrome

5.4%/18.1%

7.7%/6.3%

3.5%/8.1%

EEG findings

Background’s abnormalities

76.1%/54.6%*

84.6%/67.2%*

65.5%/77.0%

Non-epileptiform abnormalities

48.9%/18.2%*

56.5%/38.9%*

37.9%/47.3%

Epileptiform abnormalities

Focal epileptiform abnormalities

75.0%/72.7%

69.1%/76.3%

89.7%/68.9%*

Generalized epileptiform abnormalities

33.7%/27.3%

43.6%/26.5%*

31.0%/33.8%

Only patients with epilepsy were included in the analysis (overall sample size n = 103).

*Mark for statistically significant difference of two compared groups (YES/NO); ML Goodness of fit test, p < 0.05.

Discussion

Hypotonia, neurological abnormalities and severely decreased IQ scoring are significantly associated with autism. Rapin identified that hypotonia was present in 25% of autistic children and gross motor deficits (clumsy and uncoordinated movements) in 30% of children. The incidence of these disturbances is higher in autistic children compared to the normal population, but not compared to the control group with a mental handicap (Rapin 1996). In 1999, Sadock and Sadock (1999) published that up to 75% of autistic children had an IQ scoring below 70 IQ, and out of this subgroup about 45% of children suffered from deep mental retardation. In our cohort, abnormal neurological examinations were present in 74.6% of children and out of this subgroup, hypotonia was found in 15.6% with a median IQ 55 (range 15 - 104). Hypotonia proved to be a statistically significant endpoint in complete mutism. Another endpoint – decreased IQ score – was found to be significant in delayed development of speech and complete mutism as well.

Left hemisphere dominance represents the typical language lateralization profile for the majority of neurologically healthy right-handed individuals. Speech dominance and laterality are substantially genetically determined. The high number of non-right-handers in our group (60%) also indicates a less characteristic disposal of speech dominance, thereby indicating inclination to a greater vulnerability of speech development. The appearance of non-right-handedness as a risk factor at all observed end-points of speech impairment can imply a common genetic basis and, concurrently, vulnerability of development of speech in children with autism. Similar results have recently been published by Brazdil in patients with epilepsy. He showed that functional organization of the language-related neuronal network is significantly modified in patients suffering from left mesiotemporal epilepsy. Lateralization of the language functions in these subjects significantly decreased in connection with an earlier age of initial insult (Brázdil et al. 2003 and 2005).

We think that an important finding in our cohort was association of epilepsy and/or EEG epileptiform abnormalities with speech impairment in autism. Appearance of epileptic elements in autistic patients was higher than in the normal population of children and adolescents. Furthermore, epileptic seizures were found in 31.2% of children thus reaching the upper limit of occurrence of this phenomenon declared in the literature. However, this may be due to the fact that these patients tend to concentrate in tertiary care epilepsy centers such as ours. A number of patients who were primarily referred to our center with a diagnosis of epilepsy were subsequently rediagnosed as having autism. Children with epilepsy have a higher tendency to having ASD as followed from a publication by Clark et al. (2005).

In 47.8% of autistic children we identified epileptiform discharges both in awake EEG and EEG taken after sleep deprivation. Subclinical epileptiform activity was found in 15.8% of patients. Our findings are in good accordance with findings found by other authors (Kagan-Kushnir et al. 2005; Chez et al. 2006; Kim et al. 2006; Tuchman and Rapin 2002). Furthermore, our results support the hypothesis that epileptic seizures may be in a causal relationship with delayed development of speech and complete mutism, while EEG epileptiform abnormalities may play significant role in complete mutism and regression of speech.

In complete mutism, generalized epileptic abnormalities are present, whereas in regression of speech focal discharges in EEG are the characteristic findings. Interestingly, generalized epileptiform discharges were not found significant regarding their correlation with the delayed development of speech. This was probably due to the fact that seizures were being ascertained throughout the patient’s entire history and assigned EEG results were from the time of diagnosis of autism established in our Department. In some cases patients had been treated by antiepileptic drugs and were free from seizures. Therefore, the fact that at the time of diagnosis of autism no generalized epileptic abnormalities in EEG were present may, in our opinion, explain this paradoxical finding.

A possible causal relationship between epilepsy, epileptiform EEG abnormalities, behavioral, language and cognition functions was recognized by Tuchman and Rapin and others (Tuchman and Rapin 1997; Ballaban-Gil and Tuchman 2000, Galanopoulou et al. 2002; Ribeiro et al. 2002). Developmental and acquired disabilities such as autistic spectrum disorders, Landau-Kleffner Syndrome, electrical status epilepticus in sleep, and developmental dysphasias have been associated with epileptoform abnormalities and they have many common characteristic features. Tuchman established the term “epileptiform autistic regression”, (1997) which comprehends well the essential potential causality. He and others speculated about possible therapeutic implications of elimination of EEG epileptiform discharges and epileptic seizures in autism (Tharp 2004) but reliable data are still missing. The causality in these phenomena was supported by some studies (Tuchman and Rapin 1997; Mc Vicar et al. 2005; Canitano and Zapella 2006) but not by others (Canatino et al. 2005; Baird et al. 2006; Hrdlicka et al. 2004a,b).

Epilepsy and/or epileptiform abnormalities may enhance the risk of speech impairment in autistic children. Moreover, it may negatively alter the cognitive-behavioral profile in patients without autism (Aldenkamp 1997) as well.

Our results support the hypothesis that diverse pathophysiological elements may be responsible for the development of speech pathology in autistic children. A causal relationship was identified between generalized seizures and complete mutism and delayed development of speech, however, no significant association was found with the third determinant, regression of speech thus contrasting with the finding by Hrdlicka et al. (2004a,b).

Finally, in regression of speech, we identified only focal epileptic activity, particularly in those individuals with atypical distribution of speech dominance.