Article Text

An aetiological study of 25 mentally retarded adults with autism
  1. C D M van Karnebeek1,
  2. I van Gelderen3,
  3. G J Nijhof3,
  4. N G Abeling4,
  5. P Vreken4,*,
  6. E J Redeker1,
  7. A M van Eeghen1,
  8. J M N Hoovers1,
  9. R C M Hennekam1,2
  1. 1Department of Clinical Genetics, Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands
  2. 2Department of Paediatrics, Emma Children's Hospital, Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands
  3. 3Institute for the Mentally Retarded “Sherpa” (Eemeroord), Baarn, The Netherlands
  4. 4Laboratory of Metabolic Diseases, Academic Medical Centre, University of Amsterdam, Amsterdam, The Netherlands
  1. Correspondence to:
 Dr R C M Hennekam, Department of Paediatrics, Emma Children's Hospital, Academic Medical Centre, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands;

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Autism is a chronic and severe neuropsychiatric disorder with an early onset, characterised by qualitative impairments of social interaction, deviant development of language and other communicative skills, delayed cognitive development, and restricted repetitive and stereotyped interests and behaviours.1,2 The prevalence in the general population was estimated at 5.5/10 000 but more recent investigations report higher rates.3 Males are affected more often than females, with a predominance of 3 to 1.4 Mental retardation is present in about 75 to 85% of patients with autism,1,5,6 and almost half of all autistic patients are functionally mute.5

The causes and mechanisms underlying autism are heterogeneous, varying from genetic causes, that is, chromosome abnormalities and conditions with Mendelian inheritance such as metabolic disturbances, to infectious causes and teratogenic influences.7–9 The prevailing view is that autism is caused by a pathophysiological process arising from the interaction of an early environmental insult and a genetic predisposition.10 However, the aetiology often remains obscure and earlier studies reported that causal medical conditions were detectable in a relatively small percentage of autistic patients.11,12 There is increasing evidence that genetic factors may well play a major role in the remaining idiopathic cases.13 In support of this hypothesis is the recent identification of deletions in the short arm of the X chromosome,14 duplications of the Prader-Willi/Angelman critical region on chromosome 15q,15,16 linkage to 7q31 and 2q,17,18 as well as the high monozygotic twin concordance rates,19,20 and high recurrence risk among sibs of patients with idiopathic autism.21

Autism poses an extremely heavy burden for affected subjects, their families, and society. Research focusing on biological causes and on guidelines for these studies in each specific person is important, both in diagnostics, management, and genetic counselling. Here we report on the results of a full diagnostic work up of 25 patients with autism, all residing in an institute for the mentally retarded.


Study subjects

In 1993, a behavioural study concerning stereotypic movements in autism was initiated in 25 adults with autism residing in an institute for the mentally retarded (“Eemeroord”, Baarn, The Netherlands). Inclusion criteria at that time were: confirmed diagnosis of autism using DSM-IV criteria,1 age >14 years, and residency in either of the two wards chosen for the study for practical reasons. The level of functioning was assessed in each patient using the SRZ scale22,23 and subsequently classified based on DSM-IV criteria.1 All subjects participated after written informed consent was obtained from parents or other legal care givers. The study was approved by the Medical Ethical Committee of the Academic Medical Centre in Amsterdam.

Data collection

Archives were searched to collect data retrospectively on each patient.

Physical examination

All patients were examined by one physician for the mentally retarded and one clinical geneticist.

Additional investigations


Chromosome preparations from peripheral blood cultures and cytogenetic analyses were performed using standard techniques.

Fluorescence in situ hybridisation (FISH)

FISH analysis of the minimal DiGeorge critical region was performed using the cosmid M51 probe. FISH analysis of the subtelomeric regions of all chromosomes was performed using the Cytocell Ltd MultiprobeT technique,24 and scored by two investigators following the protocol described in Appendix 1 (see The FISH probes used to detect the origin and size of the marker chromosome identified in one of the patients are described in Appendix 1.

Screening for 15q11-13 interstitial duplications

The Prader-Willi/Angelman syndrome critical region (15q11-13) was screened for duplications by applying three different methods: (1) FISH analysis using the D15S10 and SNRPN probes (Vysis Inc), (2) densitometry using the microsatellite markers described in Appendix 1, (3) quantitative Real time PCR of loci D15S122 and GABRA5 using probes and primers chosen with the assistance of the Primer Express software program (Applied Biosystems) and ordered from Applied Biosystems as well (for detailed description see Appendix 1) (fig 1).

Figure 1

Fluorescence in situ hybridisation with LSI SNRPN/PML/CEP 15 dual colour probe (Vysis, Inc) which hybridises to the PW/AS critical region 15q11-13 (SpectrumOrange SNRPN), and as a control also to 15p11.2 (SpectrumGreen CEP 15p11.2, D15Z1) and to 15q22 (SpectrumOrange PML). (A) Metaphase cell showing a single orange signal (normal) at locus 15q11-13 (arrow). (B) Metaphase cell showing double orange signals at locus 15q11-13 (arrow), indicating a possible interstitial duplication. (C) Metaphase cell showing a large merged orange signal at locus 15q11-13 (arrow), indicating a possible interstitial duplication.

Other molecular analysis

The presence of an FMR1 gene expansion was analysed using standard molecular PCR procedures.25MECP2 gene mutation screening is described in detail in Appendix 1.

Neuroimaging and EEG

Neuroimaging (CT and/or MRI scanning) of the brain was performed in all patients with either neurological signs at physical examination or in patients with microcephaly or macrocephaly (that is, an occipitofrontal head circumference below the 2nd centile or above the 98th centile, respectively) (n=10). Electroencephalography was performed on all patients using standard methods.

Metabolic investigations

These included a general urine screen as well as a search for peroxisomal, mitochondrial, glycosylation, and cholesterol metabolism disturbances (for detailed description see Appendix 1).

Ophthalmological and ear, nose, and throat (ENT) investigations

The investigations were performed according to international standards in all patients.

Other investigations

If clinical history, physical examination, or one of the other above mentioned studies produced clues for a specific diagnosis, further investigations in search of this diagnosis were initiated.


Patient characteristics (table 1)

Table 1

Patient characteristics

Detailed information on single patients is provided in Appendix 2 (see All patients are of Dutch extraction except for one Indonesian male and one Nigerian female. Age at physical examination varied from 22 to 45 years (mean 33.6 years). Intellectual abilities were limited in all 25 patients, the severity of mental retardation being mild (IQ 50/55-70) in two, moderate (IQ 35/40-50/55) in 11, severe (IQ 20/25-35/40) in 11, and profound (IQ <20/25) in the remaining two patients.1 Six patients had mothers whose age at conception was 36 years or older. Mental retardation was present in seven relatives; psychiatric disorders occurred in five relatives.

In two patients, teratogenic influences were present during pregnancy: excessive maternal alcohol abuse in one and anticonvulsant use in the other. The patient born to the mother with alcohol abuse had multiple features fitting fetal alcohol syndrome; in the other patient no stigmata of fetal anticonvulsant syndrome were present. In a third patient, phenotypic features were strongly suggestive of a teratogenic influence. However, the mother denied use of medication or abuse of alcohol or other drugs during pregnancy, although she had been known to have periods of alcohol abuse before.

On physical examination, microcephaly or macrocephaly was diagnosed in eight patients, and neurological examination showed abnormalities (signs of a confirmed HNP L5-S1, drowsiness, tremor, and cogwheel rigidity) in three patients, while neurological side effects of medication were present in two patients. Minor congenital anomalies were present in 17 patients, varying from only one to multiple dysmorphic features. In eight of them the features were suggestive of the presence of a multiple congenital anomalies syndrome. More expressed congenital anomalies (urethral stenosis, umbilical hernia) were found twice. Skin depigmentation disturbances including café au lait spots and erythema were found in three patients; none had symptoms of neurofibromatosis, tuberous sclerosis, or other neurocutaneous disorders.

Additional investigations

The results of all additional investigations performed are listed in table 2. In this section only more specific details are provided.

Table 2

Results of additional investigations and diagnosis


The supernumerary marker chromosome detected in one patient was positive for FISH analysis with the centromere 13/21 probe (pZ21A), as well as for the whole chromosome 13 paint. Further studies showed the marker to be dicentric, without proximally located long arm material (13q11-12). In another patient, a mosaic pattern of the sex chromosomes was found, indicating his karyotype to be 45,X (75%)/46,XY (25%). In a third patient, subtelomeric FISH analysis yielded an Xpter deletion using cosmid probe CY29. However, this finding was not reproduced by the BAC probe 98C4 nor by molecular analysis. We concluded that the subtelomeric Xp abnormality was a polymorphism.

Screening for 15q11-13 interstitial duplications

The scoring results of SNRPN and D15S10 probes in interphase and metaphase showed considerable intra- as well as interobserver variability (data not shown). Thus results were difficult to interpret. The number of scored interstitial duplications, applying the definitions described in the Methods section, for the SNRPN probe varied between 0 and seven (28%) patients and for the D15S10 probe between one (4.2%) and 11 (45.8%) patients. There was no patient in whom a duplication was scored in both interphase and metaphase by both technicians.

Densitometric studies did not show a true duplication in the patients in our cohort, but dosimetric abnormalities were found in six patients, indicating a mosaic pattern of duplication in the region of marker D15S122 (n=4) and in the GABRA5 region (n=2) (fig 2). These results did not correspond to those of FISH analysis. Real time quantitative PCR analysis of the D15S122 region in the former four patients did not show any abnormalities, nor did analysis of the GABRA5 region in the latter two. As a positive control, a patient known in our Department with a cytogenetically visible duplication of the region 15q11-13 was used, in whom the duplication was confirmed with Real time quantitative PCR. Thus, by defining the latter technique as the gold standard, no person in our cohort had an interstitial 15q duplication.

Figure 2

Electropherograms for marker D15S122 in four patients. (A) Positive control with visible duplication of the 15q11-13 region. The amplitude ratio of the 154.8 mobility unit (mu) peak (blue) v the 148.9 mu peak (orange) is 55.5:37.7 (=1.5). (B) Normal patient without a duplication. The amplitude of the 146.8 mu peak (blue) v the 144.9 mu peak (orange) is 42.4:64.6 (=0.7). (C) Patient with a possible mosaic duplication of this locus. The blue (150.9) v orange (146.8) peak ratio is 61.4:53.1 (=1.2), which is similar to, but smaller than, gold standard (A). (D) Another patient with a possible mosaic duplication of this locus. The blue (149) v orange (144.9) mu peak ratio is 43.6:36.7 (=1.2), which is also similar to but smaller than (A).

Molecular analysis

In one female, analysis of MECP2 showed a C to T transition at position 1125 (the C-terminal region) causing a substitution of serine for proline at amino acid position 376. In the subsequent screening of 200 X chromosomes of normal controls, a similar mutation was detected in one control, indicating that it was most probably a polymorphism.

Metabolic investigations

In one patient, a previously established diagnosis of phenylketonuria (PKU) was reconfirmed by detection of raised plasma concentration of phenylalanine. In another patient, in whom the clinical diagnosis of fetal alcohol syndrome (FAS) was established, an abnormality of the distal cholesterol biosynthesis was detected: plasma concentrations of 5;7-dehydrocholesterol as well as 5;8-dehydrocholesterol were raised and plasma bile acid concentrations were low. Molecular analysis of the complete gene gave normal results, however. It seems likely that the detected cholesterol abnormalities can be attributed to the patient's intake of haloperidol, as has been reported before.26 No abnormalities were detected in the cholesterol biosynthesis of five other patients in our cohort also using haloperidol.


In 10 patients neuroimaging studies were performed. Cerebellar atrophy was found in one and in another possibly diffuse cerebral atrophy was present; in all others neuroimaging showed no abnormality.


The individual case histories of the following nine patients are described in detail in Appendix 2. In five (20%) patients the following unequivocal aetiological diagnoses were established: a prenatal factor in one (FAS, case 1), a perinatal factor in one (kernicterus, case 2), a metabolic disturbance in one (PKU, case 3), and a chromosome abnormality in two (marker chromosome 13, case 4) (mosaic karyotype, 45,X (75%)/46,XY (25%), case 5). In four (16%) patients a diagnosis was probable but without firm proof: a private syndrome in one patient (case 6), a teratogenic factor in one (case 7), and Orstavik syndrome in two patients (cases 8 and 9). Features of the latter two are compared to published cases in table 3.27

Table 3

Orstavik syndrome: comparison of our patients to published reports27

Diagnostic yield of individual investigations

The diagnostic yield of each investigation differed widely (table 4). A detailed clinical history identified abnormalities in 13 patients and in five of them this investigation contributed significantly to the diagnosis. Family history showed abnormalities in 12 patients, which contributed significantly in one patient. Physical examination showed abnormalities in 20 patients and in seven of them these findings contributed to establishing a diagnosis. Karyotyping led to diagnosis in two patients, while the results of FISH screening for subtelomeric rearrangements as well as for 22q11 deletions were normal. Screening for duplications in the PW/AS region through densitometry indicated possible mosaic duplications in six patients, but these could not be confirmed by Real time PCR. Molecular analysis for FMR1 gene expansions did not contribute to the diagnosis, nor did MECP2 gene mutation screening. Finally, EEG abnormalities were detected in eight patients, contributing to a diagnosis in two. Based on neuroimaging findings a diagnosis could be established in two subjects.

Table 4

Yield of investigations performed in all 25 patients

Remaining patients

Of the remaining 16 (64%) patients without any diagnosis, one had a major congenital malformation and eight had minor anomalies on physical examination. Of the latter group, no person had four or more anomalies, which would have been suggestive of a MR/MCA syndrome.


Since Kanner's description of autism as a developmental disorder characterised by “extreme autistic loneliness” and “an obsessional desire for the maintenance of sameness”, autism has often been referred to as a diagnosis on its own.28 However, autism should be considered a symptom, caused by a variety of underlying disorders. Most investigators at present assume that the underlying pathophysiological process causing autism arises from the interaction between a genetic predisposition and an early environmental insult.10

Comparison to other studies

In the present study, a diagnosis was made in nine (36%) patients, in five unequivocally and in four probably. These results fall in the middle range when compared to other similar studies: an organic aetiologic factor for autism was found to range from 10-60%.2,5,7–9,12,19,20,29–32 The variation in the results of these studies may be explained by a difference in diagnostic criteria used to establish the diagnosis of autism, in patient selection, and in diagnostic work up.11

Present study: advantages and limitations

In the present study we analysed a group of patients assembled previously for another study (with a different scope), all residing in the same institute for the mentally retarded, fulfilling all the DSM-IV criteria for autism,1 and with a standardised work up in all patients that was (almost) complete for present standards. The work up is limited only in the absence of screening of family members for components of the broad autism phenotype, as suggested by Piven et al.33 However, this approach is especially important in linkage studies, but less so for finding organic causes, as in the present study.33

The fact that all patients were mentally retarded adults may have created a bias. However, as 75-85% of patients with autism are mentally retarded and as the distribution of the different degrees of delay within the present study group is similar to the distribution within persons with autism in general, the introduced bias is limited.7,8,29,32 For subjects with autism whose characteristics differ from those of our cohort, such as children and subjects without mental retardation, the results and conclusions of the present study are only partly applicable. Therefore the yield expected for a diagnostic work up should be adjusted to the age and developmental level of the autistic patient.

Value of each diagnostic investigation

Clinical history and physical examination

Clinical history and physical examination have proven to be the most rewarding parts of the diagnostic work up. A patient's history may provide clues to prenatal causes, for example, fetal teratogen exposure such as thalidomide34 and alcohol.35 The exact teratogenic mechanism of alcohol is unknown, although a clue may be the disturbance in the migration of neuronal and glia cells,36 supported among other things by the cerebellar anomalies present in both animal models of FAS37 and subjects with autism.38 Perinatal history may provide clues for factors such as neonatal jaundice. Before the advent of phototherapy and exchange blood transfusions, neonatal jaundice often caused kernicterus, characterised by sensorineural hearing loss, mental retardation, and evidence on MRI of damaged basal ganglia, especially the globus pallidus.39 The latter characteristic may well explain the autistic behaviour reported in patients with kernicterus,40 as there is increasing evidence that the volume and function of basal ganglia are different in autistic subjects compared to controls.41

Detailed physical examination, including anthropometric, neurological, and dysmorphology examinations, may provide clues to many possible aetiologies. In general, the incidence of minor physical anomalies in patients with autism is increased when compared to controls,42 indicating that in the former group structural development was disrupted during early embryogenesis as a result of underlying disorder. Abnormal cephalic measures occur among autistic persons in a significantly higher proportion than in the general population, and have been suggested to be the single most consistent physical characteristic of autistic subjects.12,43 In our cohort, the frequency of macrocephaly (20%) and microcephaly (12%) is similar to the frequencies among the general autistic population (20% and 15%, respectively).44

Many syndromes are known to have a distinct behavioural phenotype, indicating the potential for the causative genes to influence human cognitive development.45 Autism is considered to be such a behavioural phenotype associated with several well described syndromes, such as Williams syndrome and Moebius syndrome.46 In our cohort, a syndromic genesis was suspected for two subjects whose phenotype strongly resembled the autosomal recessive syndrome described by Orstavik et al,27 and which is characterised by epilepsy, mental retardation, facial dysmorphism, and macrocephaly (table 3). This combination of features may have been described before in 1995 by Andermann47 in four patients. As neuroimaging data are lacking in both our patients, an unequivocal diagnosis of Orstavik syndrome is currently not possible.


A broad spectrum of chromosome anomalies in autism has been reported, involving almost all chromosomes and many types of rearrangements.48 Most frequently documented are (de novo) structural and numerical abnormalities of sex chromosomes and anomalies of chromosome 15.13 In our cohort, two (8%) subjects were identified with numerical chromosome anomalies, a rate similar to the 5% reported by Bailey et al49 in 1996 and 6.3% by Konstantareas et al in 1999.50 Three patients similar to our case with 45,X/46,XY mosaic and autistic behaviour have been described.2,51 The present patient with mosaic Turner karyotype showed bilateral shortening of the fourth metatarsal bones, a common feature in Turner syndrome. This is additional evidence that the chromosome anomaly exerts its effect on the phenotype. Furthermore, the maternal origin of the remaining X in the present patient is in agreement with the hypothesis of a parent of origin effect in the X chromosome influencing social cognition.52 Although supernumerary marker chromosomes have been reported in subjects with autism,50,53 to our knowledge no other cases with a marker derived from chromosome 13 were reported to have autism.54 Interestingly, a chromosome 13q region was the most significant result in one collaborative linkage study (CLSA) genome scan.55 Other linkage studies also showed increased sharing of this locus, although weaker.56 It seems likely that the marker 13 chromosome has affected the phenotype, including the presence of autism.

Submicroscopic deletions

Patients with the velocardiofacial (VCF) syndrome have a distinct behavioural phenotype and often suffer psychiatric disorders, including autism.57 FISH analysis in all the present patients was performed despite the absence of the dysmorphic features of VCF syndrome, and no deletion was identified. This concurs with a recent paper, in which no 22q11 deletions were found in autistic patients without a phenotype suggesting VCF syndrome.58 Submicroscopic rearrangements involving (sub-)telomeric regions are emerging as an important cause of mental retardation.59 In our cohort, no subtelomeric rearrangements were found. Owing to the limited number of patients in our cohort, however, no firm conclusions can be drawn. To ascertain the presence and frequency of these cryptic rearrangements among autistic subjects, further similar, and if possible larger, studies are required.

15q11-13 abnormalities

Over the past decade, numerous reports have mentioned abnormalities of chromosome 15 associated with autism, with or without mental retardation. Frequently, they concerned supernumerary isodicentric chromosomes 15, less frequently interstitial duplications, and rarely triplications of the 15q11-13 region, almost all maternally derived.60–69 The region may harbour a potential susceptibility gene or genes for autism 13,70,71 In a large study performed by Schroer et al,15 a cohort of 100 consecutive patients with autism was screened and abnormalities were identified in four (4%): two with supernumerary bisatellited marker chromosomes 15, one with a 15q11-13 deletion, and one with an interstitial duplication of region 15q11-13. In most published cases, an interstitial duplication (or triplication) was already visible on G banded chromosomes. However, there is as yet no generally accepted reconfirmation method, nor a method to screen for submicroscopic duplications in this region. FISH analysis, sometimes followed by PCR microsatellite analysis, is the most widely applied method.15,16,62,67 There is only a single report of subtle 15q11-13 interstitial duplications, which according to the authors might have been missed in routine chromosome or FISH analysis: microsatellite densitometry analysis was used to detect this abnormality in two sibs with autism, which was also present in their unaffected mother.72 In our cohort, we found no patients with cytogenetically visible abnormalities of chromosome 15. We first performed FISH analysis, but this technique is not suitable for detecting cryptic 15q11-13 duplications because the company making the probes (Vysis, Inc) states in their “Quality Assurance Certificate” that “Occasionally, the LSI probes may appear as three or four signals, depending upon condensation of the DNA and the relative distances between chromatids. All probe signals may also appear diffuse or split”. No validated method has been published to quantify and qualify results of duplication screening of region 15q11-13 by FISH, and various subjective terms have been used, such as “a large merged signal”, “a double signal”, or “a split signal”16,73; interobserver and intraobserver variability studies are lacking. Consequently, we have performed densitometric studies. This technique yielded possible interstitial duplications (in mosaic form) in several patients; however, these did not coincide with those found by FISH. Furthermore, we doubted its validity, as the amplitude ratio of the electropherogram peaks in our six patients was less than the ratio in the positive control (fig 2). Therefore, quantitative Real time PCR was performed, but no abnormalities were found, indicating that no person in our cohort had such a duplication. FISH analysis and densitometry would have led to an overestimated frequency of interstitial 15q duplications in our cohort. Hence, this seems to be a less frequent cause of autism than suggested by previous clinical studies.15,72

Real time PCR has proven to be a sensitive, specific, and reproducible method for diagnosing changes in gene dosage, especially in diagnostic cancer research.74 A recent exemplary study was performed by Aarskog et al,75 who used Real time PCR to detect cryptic PMP22 duplications and deletions in patients with Charcot-Marie-Tooth type 1A.75 For our study, its technical validity was proven by confirmation of a duplication in a cytogenetically positive control.

Whether duplications in the PWAS region are really pathophysiologically significant in autism remains to be elucidated. Recently, duplications of the GABRA5 gene were detected in unaffected subjects, suggesting that (some) 15q11-13 duplications may be benign polymorphisms.76

Fragile X syndrome

No patients were identified with fra(X). Indeed, the relationship between these two conditions remains uncertain. At first, males and females with fra(X) were reported to display autistic behaviour frequently,77 but later on autism was proven to occur no more frequently in a fra(X) population than in the mentally retarded population in general.78 Also, no linkage with the FMR1 gene region was proven in a study of families multiplex for autism but without cytological evidence of fragile X expression.79

MECP2 gene screening

The recent study by Lam et al80 reporting a MECP2 gene mutation in one of the 21 screened (non-RTT) subjects with infantile autism and mental retardation motivated us to perform a similar screening in our cohort. No true mutations were found, as the single point mutation we found is probably a (hitherto unreported) polymorphism.

DNA repair disorder

One of the present patients (case report 6) showed many features compatible with a DNA repair disturbance, although firm proof at the cellular level is still lacking. To our knowledge, only one earlier report mentioned autism in a patient with a DNA repair disturbance.81 It remains uncertain how this relationship, if any, might be explained. Well known repair disturbances such as Bloom, Cockayne, Rothmund-Thomson, and Werner syndromes do not include autism.82

Metabolic screening

Phenylketonuria,83 hypersuccinylpurinaemia,84 changes of aromatic acids and monoamines,85 lactate acidosis,86 and cholesterol anomalies87 are metabolic disturbances reported in patients with autism. In our cohort, extensive screening of multiple metabolic pathways resulted in the (re)confirmation of metabolic disturbances in two patients: phenylketonuria (PKU) in one and an abnormality in the distal cholesterol biosynthesis in another patient. The mechanism through which PKU causes autism remains uncertain, but has been suggested to involve dopaminergic pathways.88 The so-called enzyme/brain-barrier has also been suggested to be involved.89 In our cohort, no proof of a mitochondrial dysfunction was present. A disturbance in brain energy metabolism owing to mitochondrial dysfunction has been proposed as a cause of autism, supported by reports of lactate acidosis and urinary excretion of Krebs cycle metabolites in autistic patients.86,90 It has been hypothesised that mitochondrial dysfunction may act through excessive nitric oxide production.91

Finally, abnormalities in the glycosylation process cause the CDG syndrome phenotypes.92 As these are still in the process of delineation,93 we investigated all patients, all with negative results.


The recognition of a high incidence of EEG abnormalities and of seizure disorders involving all areas of the cerebral cortex was among the earliest evidence of an organic basis of autism.94 In autistic subjects with a single EEG, abnormalities are present in about 40% versus 65% with multiple EEGs.95 In our cohort, single EEGs identified abnormalities suggestive of epilepsy in about one-third of patients (n=8). However, epilepsy should not be considered a separate aetiological entity but a symptom.

Neuroanatomical imaging

Neuroanatomical abnormalities are frequently found in autistic subjects, affecting the cerebral cortex, thalamus, brain stem, and most often the cerebellum.96 It has been difficult to integrate these findings in an explanatory model.97 Of the present 10 patients in whom neuroimaging was performed, findings concurring with the diagnosis of kernicterus were identified in one patient, and in the patient with a possible DNA repair disorder signs of cerebral atrophy were present. Otherwise all neuroradiological studies gave normal results. Thus, in our cohort, neuroimaging provided few clues for previously unsuspected diagnoses.

ENT and ophthalmological investigation

Investigation of both vision and hearing is important in any person with developmental delay for adequate support and care and to detect specific causes for the mental handicap.98 Although in our cohort the ENT and ophthalmological investigations did not provide new aetiological insights, a considerable number of subjects were identified with insufficient sensory function, with therapeutic consequences.


In autism, a specific aetiological diagnosis is of considerable value, both to establish prognosis and to provide adequate care, as well as for genetic counselling and for unravelling its pathogenetic mechanisms. The present study shows that an extensive, structured work up yields a diagnosis in at least 20%, and possibly up to 36% of adult autistic subjects with mental retardation. As these underlying medical conditions encompass teratogenic, metabolic, and syndromic influences, a prerequisite for such a yield is a multidisciplinary approach, of which clinical history taking and physical examination form the basis. In future similar studies, diagnostic data should be gathered on a larger number of patients, allowing more firm conclusions. Linkage studies, used to screen multiplex families in order to identify autism susceptible loci,56,99 will profit from an initial diagnostic work up of all subjects before inclusion, as all other causes of autism will be ruled out. The diagnostic yield of linkage studies will undoubtedly increase if the cohorts comprise only those remaining cases with truly idiopathic autism.18

  • Twenty-five consecutive mentally retarded adults (20 males) with autism (DSM-IV criteria) were investigated for somatic causes of their developmental disorder.

  • Each patient underwent complete physical examination, ophthalmological and ENT investigations, EEG studies, karyotyping, FISH analysis of 22q11 region and all subtelomeric regions, molecular analysis for FMR1 gene expansions, MECP2 gene mutations, and duplications of the 15q11-13 region, metabolic investigations including a general urine screen, search for peroxisomal, mitochondrial, glycosylation, and cholesterol metabolism disturbances, and neuroradiological studies (if neurological symptoms were present).

  • In five patients an unequivocal diagnosis was established: fetal alcohol syndrome, kernicterus, phenylketonuria, disturbed cholesterol metabolism, marker chromosome 13, and mosaicism for XY/XO. A probable diagnosis was made in four other patients: Orstavik syndrome in two, a private syndrome in one, and a teratogenic cause in one.

  • A complete work up of autistic mentally retarded adults yields a diagnosis responsible for autism in at least 20% and possibly up to 36%. If such studies are performed in cohorts of familial cases used for linkage analysis, such studies may well be more successful.


This project was financially supported by grants from the “Stichting tot Steun van het Emma KinderZiekenhuis” and the “Stichting Klinische Genetica Amsterdam”. We are grateful to the patients and their parents for their cooperation in this study project. We thank M Roelink-Reynhoudt for performing venepunctures, E de Boer and C Koevoets for performing cytogenetic analyses, R Rust, Drs R van den Boogaard, M Alders, and M Mannens for molecular analyses, Dr V Kalscheuer for providing YAC clones, Dr H Waterham for metabolic studies, N Briare, Drs W de Bruin, M Dudok van Heel, A Kraak, and J A P M de Laat for audiometric/ENT studies, and Dr W Dorsman, F Gunther, and G Kinds and for ophthalmological/optometric investigations.


Supplementary materials

    Web-only appendices

    Appendix 1: Detailed description of methods
    Appendix 2: Case reports

    View pdf


  • * *Deceased.