DRD2 and PPP1R1B (DARPP-32) polymorphisms independently confer increased risk for autism spectrum disorders and additively predict affected status in male-only affected sib-pair families.

BACKGROUND: The neurotransmitter dopamine (DA) modulates executive functions, learning, and emotional processing, all of which are impaired in individuals with autism spectrum disorders (ASDs). Our previous findings suggest a role for dopamine-related genes in families with only affected males. METHODS: We examined two additional genes which affect DA function, the DRD2 and PPP1R1B (DARPP-32) genes, in a cohort of 112 male-only affected sib-pair families. Selected polymorphisms spanning these genes were genotyped and both family-based and population-based tests were carried out for association analysis. General discriminant analysis was used to examine the gene-gene interactions in predicting autism susceptibility. RESULTS: There was a significantly increased frequency of the DRD2 rs1800498TT genotype (P = 0.007) in affected males compared to the comparison group, apparently due to over-transmission of the T allele (P = 0.0003). The frequency of the PPP1R1B rs1495099CC genotype in affected males was also higher than that in the comparison group (P = 0.002) due to preferential transmission of the C allele from parents to affected children (P = 0.0009). Alleles rs1800498T and rs1495099C were associated with more severe problems in social interaction (P = 0.0002 and P = 0.0016, respectively) and communication (P = 0.0004 and P = 0.0046), and increased stereotypic behaviours (P = 0.0021 and P = 0.00072). General discriminant analysis found that the DRD2 and PPP1R1B genes additively predicted ASDs (P = 0.00011; Canonical R = 0.26) and explain ~7% of the variance in our families. All findings remained significant following corrections for multiple testing. CONCLUSION: Our findings support a role for the DRD2 and PPP1R1B genes in conferring risk for autism in families with only affected males and show an additive effect of these genes towards prediction of affected status in our families.

Developmental trajectories in syndromes with intellectual disability, with a focus on Wolf-Hirschhorn and its cognitive-behavioral profile.

Few studies exist of developmental trajectories in children with intellectual disability, and none for those with subtelomeric deletions. We compared developmental trajectories of children with Wolf-Hirschhorn syndrome to other genetic disorders. We recruited 106 children diagnosed with fragile X, Williams-Beuren syndrome, or Wolf-Hirschhorn syndrome, assessing their intellectual and adaptive behavior abilities. We retested 61 children 2 years later. We compared Time 1 and Time 2 difference scores related to genetic disorder, age, initial IQ, or adaptive behavior composite. Results show genetic disorder and initial IQ score were significant factors for IQ differences, but only genetic disorder affected adaptive behavior differences. Results suggest different gene-brain-behavior pathways likely exist for these genetic disorders. Different developmental trajectories will influence the type and intensity of intervention implemented by caregivers.

Dopamine receptor D4 gene variation predicts preschoolers' developing theory of mind.

Individual differences in preschoolers' understanding that human action is caused by internal mental states, or representational theory of mind (RTM), are heritable, as are developmental disorders such as autism in which RTM is particularly impaired. We investigated whether polymorphisms of genes affecting dopamine (DA) utilization and metabolism constitute part of the molecular basis of this heritability. Seventy-three 42- to 54-month-olds were given a battery of RTM tasks along with other task batteries that measured executive functioning and representational understanding more generally. Polymorphisms of the dopamine D4 receptor gene (DRD4) were associated with RTM performance such that preschoolers with shorter alleles outperformed those with one or more longer alleles. However, polymorphisms of the catechol-O-methyl transferase gene (COMT) and the dopamine transporter gene (DAT1) genes were not associated with children's RTM performance. Further tests showed that the association between DRD4 allele length and RTM performance was not attributable to a common association with executive functioning or representational understanding more generally. We conclude that DRD4 receptors, likely via their effects on frontal lobe development and functioning, may represent a neuromaturational constraint governing the stereotypical and universal trajectory of RTM development.

Development and initial validation of a parent report measure of the behavioral development of infants at risk for autism spectrum disorders.

We developed and evaluated a new parent report instrument--Parent Observation of Early Markers Scale (POEMS)--to monitor the behavioral development of infants at risk for autism spectrum disorder (ASD) because they have older affected siblings. Parents of 108 at-risk infants (74 males, 34 females) completed the POEMS from child age 1-24 months. The POEMS had acceptable psychometric properties and promising predictive validity. Most concerning items were social and communication deficits, and intolerance to waiting. Results provide preliminary evidence that prospective parent report measures can help to detect early ASD symptoms in infants at biological risk. We invite researchers to join us in multi-center studies of the POEMS.

Association of GTF2i in the Williams-Beuren syndrome critical region with autism spectrum disorders.

Duplications of 7q11.23, deleted in Williams-Beuren Syndrome, have been implicated in autism spectrum disorders (ASDs). A 1.5 Mb duplication was identified in one girl with severe expressive language deficits and anxiety among 1,142 ASD individuals screened for this duplication. Family-based association studies of Tag-SNPs in three genes (STX1A , CYLN2 and GTF2i) in two multiplex autism family cohorts revealed strong association of two GTF2i SNPs and their haplotype in Cohort 1 and the combined families. The risk alleles and haplotype were associated with severe problems in social interaction and excessive repetitive behaviors. Our findings suggest the GTF2i gene is important in the etiology of autism in individuals with this duplication and in non-duplication cases with severe social interaction problems and repetitive behaviors.

Understanding the impact of 1q21.1 copy number variant.

BACKGROUND: 1q21.1 Copy Number Variant (CNV) is associated with a highly variable phenotype ranging from congenital anomalies, learning deficits/intellectual disability (ID), to a normal phenotype. Hence, the clinical significance of this CNV can be difficult to evaluate. Here we described the consequences of the 1q21.1 CNV on genome-wide gene expression and function of selected candidate genes within 1q21.1 using cell lines from clinically well described subjects. METHODS AND RESULTS: Eight subjects from 3 families were included in the study: six with a 1q21.1 deletion and two with a 1q21.1 duplication. High resolution Affymetrix 2.7M array was used to refine the 1q21.1 CNV breakpoints and exclude the presence of secondary CNVs of pathogenic relevance. Whole genome expression profiling, studied in lymphoblast cell lines (LBCs) from 5 subjects, showed enrichment of genes from 1q21.1 in the top 100 genes ranked based on correlation of expression with 1q21.1 copy number. The function of two top genes from 1q21.1, CHD1L/ALC1 and PRKAB2, was studied in detail in LBCs from a deletion and a duplication carrier. CHD1L/ALC1 is an enzyme with a role in chromatin modification and DNA damage response while PRKAB2 is a member of the AMP kinase complex, which senses and maintains systemic and cellular energy balance. The protein levels for CHD1L/ALC1 and PRKAB2 were changed in concordance with their copy number in both LBCs. A defect in chromatin remodeling was documented based on impaired decatenation (chromatid untangling) checkpoint (DCC) in both LBCs. This defect, reproduced by CHD1L/ALC1 siRNA, identifies a new role of CHD1L/ALC1 in DCC. Both LBCs also showed elevated levels of micronuclei following treatment with a Topoisomerase II inhibitor suggesting increased DNA breaks. AMP kinase function, specifically in the deletion containing LBCs, was attenuated. CONCLUSION: Our studies are unique as they show for the first time that the 1q21.1 CNV not only causes changes in the expression of its key integral genes, associated with changes at the protein level, but also results in changes in their known function, in the case of AMPK, and newly identified function such as DCC activation in the case of CHD1L/ALC1. Our results support the use of patient lymphoblasts for dissecting the functional sequelae of genes integral to CNVs in carrier cell lines, ultimately enhancing understanding of biological processes which may contribute to the clinical phenotype.

2p15-p16.1 microdeletion syndrome: molecular characterization and association of the OTX1 and XPO1 genes with autism spectrum disorders.

Reports of unrelated individuals with autism spectrum disorder (ASD) and similar clinical features having overlapping de novo interstitial deletions at 2p15-p16.1 suggest that this region harbors a gene(s) important to the development of autism. We molecularly characterized two such deletions, selecting two genes in this region, exportin 1 (XPO1) and orthodenticle homolog 1 (OTX1) for association studies in three North American cohorts (Autism Spectrum Disorder - Canadian American Research Consortium (ASD-CARC), New York, and Autism Genetic Resource Exchange (AGRE)) and one Italian cohort (Società Italiana per la Ricerca e la Formazione sull'Autismo (SIRFA)) of families with ASD. In XPO1, rs6735330 was associated with autism in all four cohorts (P<0.05), being significant in ASD-CARC cohorts (P-value following false discovery rate correction for multiple testing (P(FDR))=1.29 × 10(-5)), the AGRE cohort (P(FDR)=0.0011) and the combined families (P(FDR)=2.34 × 10(-9)). Similarly, in OTX1, rs2018650 and rs13000344 were associated with autism in ASD-CARC cohorts (P(FDR)=8.65 × 10(-7) and 6.07 × 10(5), respectively), AGRE cohort (P(FDR)=0.0034 and 0.015, respectively) and the combined families (P(FDR)=2.34 × 10(-9) and 0.00017, respectively); associations were marginal or insignificant in the New York and SIRFA cohorts. A significant association (P(FDR)=2.63 × 10(-11)) was found for the rs2018650G-rs13000344C haplotype. The above three SNPs were associated with severity of social interaction and verbal communication deficits and repetitive behaviors (P-values <0.01). No additional deletions were identified following screening of 798 ASD individuals. Our results indicate that deletion 2p15-p16.1 is not commonly associated with idiopathic ASD, but represents a novel contiguous gene syndrome associated with a constellation of phenotypic features (autism, intellectual disability, craniofacial/CNS dysmorphology), and that XPO1 and OXT1 may contribute to ASD in 2p15-p16.1 deletion cases and non-deletion cases of ASD mapping to this chromosome region.

Population- and family-based studies associate the MTHFR gene with idiopathic autism in simplex families.

Two methylenetetrahydrofolate reductase gene (MTHFR) functional polymorphisms were studied in 205 North American simplex (SPX) and 307 multiplex (MPX) families having one or more children with an autism spectrum disorder. Case-control comparisons revealed a significantly higher frequency of the low-activity 677T allele, higher prevalence of the 677TT genotype and higher frequencies of the 677T-1298A haplotype and double homozygous 677TT/1298AA genotype in affected individuals relative to controls. Family-based association testing demonstrated significant preferential transmission of the 677T and 1298A alleles and the 677T-1298A haplotype to affected offspring. The results were not replicated in MPX families. The results associate the MTHFR gene with autism in SPX families only, suggesting that reduced MTHFR activity is a risk factor for autism in these families.

Disruption at the PTCHD1 Locus on Xp22.11 in Autism spectrum disorder and intellectual disability.

Autism is a common neurodevelopmental disorder with a complex mode of inheritance. It is one of the most highly heritable of the complex disorders, although the underlying genetic factors remain largely unknown. Here, we report mutations in the X-chromosome PTCHD1 (patched-related) gene in seven families with autism spectrum disorder (ASD) and in three families with intellectual disability. A 167-kilobase microdeletion spanning exon 1 was found in two brothers, one with ASD and the other with a learning disability and ASD features; a 90-kilobase microdeletion spanning the entire gene was found in three males with intellectual disability in a second family. In 900 probands with ASD and 208 male probands with intellectual disability, we identified seven different missense changes (in eight male probands) that were inherited from unaffected mothers and not found in controls. Two of the ASD individuals with missense changes also carried a de novo deletion at another ASD susceptibility locus (DPYD and DPP6), suggesting complex genetic contributions. In additional males with ASD, we identified deletions in the 5' flanking region of PTCHD1 that disrupted a complex noncoding RNA and potential regulatory elements; equivalent changes were not found in male control individuals. Thus, our systematic screen of PTCHD1 and its 5' flanking regions suggests that this locus is involved in ~1% of individuals with ASD and intellectual disability.

The course of cognitive-behavioral development in children with the FMR1 mutation, Williams-Beuren syndrome, and neurofibromatosis type 1: The effect of gender.

The course of cognitive-behavioral development in children with intellectual disabilities produced by genetic disorders has only recently begun to be examined systematically. Unfortunately, these studies are few in number. Previously, we examined cognitive-behavioral development in children with the fragile X (FMR1) mutation and found longitudinal decreases in both IQ and adaptive behavior (DQ) scores in most males and females with the full mutation. In this study, we examine longitudinal changes in IQ and DQ in children with neurofibromatosis type 1 (NF1) and Williams-Beuren Syndrome (WBS) by examining differences in composite IQ and DQ scores between the first test (T1) and retest (T2), and compare their developmental trajectory to children with the FMR1 mutation. Sixty-five children with the FMR1 mutation, or NF1, or WBS, ages 4-16 years, were retested two years after initial testing with the Stanford-Binet 4th Edition (SBFE) and the Vineland Adaptive Behavior Scale (VABS). In addition to significant longitudinal declines in IQ and DQ noted previously in children with the FMR1 mutation, we found significant decreases in IQ in males compared to females in the remainder of our sample. We also observed statistically significant decreases in DQ scores among children the FMR1 mutation, as noted previously, but not among children with NF1 or WBS. Moreover, significant declines were found only among males with the FMR1 mutation. Unlike declines in IQ scores, decreases in DQ were not significantly different between males and females.

Brief report: telephone administration of the autism diagnostic interview--revised: reliability and suitability for use in research.

The Autism Diagnostic Interview--revised is one of the "gold standard" diagnostic tools for autism spectrum disorders. It is traditionally administered face-to-face. Cost and geographical concerns constrain the employment of the ADI-R for large-scale research projects. The telephone interview is a reasonable alternative, but has not yet been examined for reliability with face-to-face administration. In this study, participants were interviewed both face-to-face and on the telephone using the complete ADI-R interview. Results indicate that there was no significant difference between the algorithm scores or the diagnoses arrived at for face-to-face and telephone administrations. Reliability statistics across the two modalities were very good and indicate that telephone interviews using the ADI-R are a viable option for researchers.

Association of MAOA, 5-HTT, and NET promoter polymorphisms with gene expression and protein activity in human placentas.

Monoamine oxidase A (MAOA) and the transporters for serotonin (5-HTT) and norepinephrine (NET) may play important roles in regulating maternal monoamine neurotransmitters transferred across the placenta to the fetus. We investigated whether promoter polymorphisms in MAOA (uVNTR), 5-HTT (5-HTTLPR), and NET (NETpPR AAGG(4)) could influence gene expression and protein activity in human placentas. Normal term human placentas (n = 73) were collected, and placental MAOA, 5-HTT, and NET mRNA levels and protein activity were determined. The mRNA levels or protein activities were compared between different genotype groups. Placentas hemizygous (male fetus) or homozygous (female fetus) for MAOA uVNTR 4-repeat allele had significantly higher MAOA mRNA levels than those hemizygous or homozygous for the 3-repeat allele (P = 0.001). However, no significant difference in MAOA enzyme activity was found for these two groups of genotypes (P = 0.161). Placentas with the 5-HTTLPR short (S)-allele (S/S+S/L) had significantly lower 5-HTT mRNA levels and serotonin uptake rate than those homozygous for the long (L)-allele (L/L) (mRNA: P < 0.001; serotonin transporting activity: P < 0.001). Placentas homozygous for the NET AAGG(4) L(4) allele had significantly higher NET mRNA levels, as well as dopamine and norepinephrine uptake rates, than those with the S(4)/L(4) genotype (mRNA: P < 0.001; dopamine transporting activity: P = 0.012; norepinephrine transporting activity: P = 0.011). These findings suggest that the three promoter polymorphisms of MAOA, 5-HTT, and NET influence gene expression levels and protein activity of these genes in human placentas, potentially leading to different fetal levels of maternal monoamine neurotransmitters, which may have an impact on fetal neurodevelopment.

Age at diagnosis of autism spectrum disorders in four regions of Canada.

OBJECTIVES: Early diagnosis of autism spectrum disorders ("autism") may lead to better treatment outcomes, reduces the stress parents experience when they do not understand the reasons for their child's behaviour, and empowers parents to make choices such as seeking genetic counseling. We examined the age at which Canadian children are diagnosed with autism, and analyzed whether there are geographic or temporal variations or differences by sex or diagnostic subtype. METHODS: As part of an autism surveillance program, in 2002/2003 we began collecting information on children with autism in Manitoba, Southeastern Ontario, Prince Edward Island, and Newfoundland and Labrador. For the analysis presented in this paper, we included children identified for our surveillance program who were diagnosed between 1997 and 2005 (n = 769). RESULTS: We found significant inter-regional differences in age at diagnosis, with Newfoundland and Labrador having the lowest median age at diagnosis (39.0 months) and Southeastern Ontario the highest (55.0 months). Diagnostic subtype was significantly associated with age at diagnosis in all regions. Southeastern Ontario was the only region where the overall age at diagnosis increased over time (p = 0.004), although in Manitoba the age at which children were diagnosed with PDD-NOS also increased significantly over the study period (p = 0.021). CONCLUSIONS: Our findings demonstrate that there are geographic differences and other sources of variation in the age at which Canadian children are diagnosed with autism. Further study is warranted to understand the factors contributing to these differences. Such research would inform best practices for early detection and timely access to treatment.

The DLX1and DLX2 genes and susceptibility to autism spectrum disorders.

An imbalance between excitation and inhibition in the cerebral cortex has been suggested as a possible etiology of autism. The DLX genes encode homeodomain-containing transcription factors controlling the generation of GABAergic cortical interneurons. The DLX1 and DLX2 genes lie head-to-head in 2q32, a region associated with autism susceptibility. We investigated 6 Tag SNPs within the DLX1/2 genes in two cohorts of multiplex (MPX) and one of simplex (SPX) families for association with autism. Family-based association tests showed strong association with five of the SNPs. The common alleles of rs743605 and rs4519482 were significantly associated with autism (P<0.012) in the first sample of 138 MPX families, with the latter remaining significant after correction for multiple testing (P(cor)=0.0046). Findings in a second sample of 169 MPX families not only confirmed the association at rs4519482 (P=0.034) but also showed strong allelic association of the common alleles at rs788172, rs788173 and rs813720 (P(cor)=0.0003-0.04). In the combined MPX families, the common alleles were all significantly associated with autism (P(cor)=0.0005-0.016). The GGGTG haplotype was over transmitted in the two MPX cohorts and the combined samples [P(cor)<0.05: P(cor)=0.00007 for the combined MPX families, Odds Ratio: 1.75 (95% CI: 1.33-2.30)]. Further testing in 306 SPX families replicated the association at rs4519482 (P=0.033) and the over transmission of the haplotype GGGTG (P=0.012) although P-values were not significant after correction for multiple testing. The findings support the presence of two functional polymorphisms, one in or near each of the DLX genes that increase susceptibility to, or cause, autism in MPX families where there is a greater genetic component for these conditions.

A DRD1 haplotype is associated with risk for autism spectrum disorders in male-only affected sib-pair families.

Individuals with autism spectrum disorders (ASDs) have impairments in executive function and social cognition, with males generally being more severely affected in these areas than females. Because the dopamine D1 receptor (encoded by DRD1) is integral to the neural circuitry mediating these processes, we examined the DRD1 gene for its role in susceptibility to ASDs by performing single marker and haplotype case-control comparisons, family-based association tests, and genotype-phenotype assessments (quantitative transmission disequilibrium tests: QTDT) using three DRD1 polymorphisms, rs265981C/T, rs4532A/G, and rs686T/C. Our previous findings suggested that the dopaminergic system may be more integrally involved in families with affected males only than in other families. We therefore restricted our study to families with two or more affected males (N = 112). There was over-transmission of rs265981-C and rs4532-A in these families (P = 0.040, P = 0.038), with haplotype TDT analysis showing over-transmission of the C-A-T haplotype (P = 0.022) from mothers to affected sons (P = 0.013). In addition, haplotype case-control comparisons revealed an increase of this putative risk haplotype in affected individuals relative to a comparison group (P = 0.004). QTDT analyses showed associations of the rs265981-C, rs4532-A, rs686-T alleles, and the C-A-T haplotype with more severe problems in social interaction, greater difficulties with nonverbal communication and increased stereotypies compared to individuals with other haplotypes. Preferential haplotype transmission of markers at the DRD1 locus and an increased frequency of a specific haplotype support the DRD1 gene as a risk gene for core symptoms of ASD in families having only affected males.

Trends in autism prevalence: diagnostic substitution revisited.

There has been little evidence to support the hypothesis that diagnostic substitution may contribute to increases in the administrative prevalence of autism. We examined trends in assignment of special education codes to British Columbia (BC) school children who had an autism code in at least 1 year between 1996 and 2004, inclusive. The proportion of children with an autism code increased from 12.3/10,000 in 1996 to 43.1/10,000 in 2004; 51.9% of this increase was attributable to children switching from another special education classification to autism (16.0/10,000). Taking into account the reverse situation (children with an autism code switching to another special education category (5.9/10.000)), diagnostic substitution accounted for at least one-third of the increase in autism prevalence over the study period.

Studies of age-correlated features of cognitive-behavioral development in children and adolescents with genetic disorders.

Studies of age-related features of cognitive-behavioral deficits produced by genetic mutations permit us to draw inferences about how brain development may be related cognitive ability as the child ages. Except for Down syndrome (DS) and the fragile X mutation (FRAXA), little is known about the longitudinal changes in cognitive-behavioral development in individuals with genetic abnormalities producing learning disabilities (LD) or mental retardation (MR). The purpose of this prospective study was to compare and contrast age related to cognitive abilities, adaptive and maladaptive behaviors in children and adolescents in the same age range, diagnosed with one of three genetic disorders: the FRAXA mutation, Neurofibromatosis type 1 (NF1) or Williams-Beuren syndrome (WBS). We also sought to examine whether cognitive-behavioral abilities associated with these three genetic disorders were related systematically to age. We examined 108 children, ages 4-15 years, with FRAXA, WBS, or NF1. Results show that there is a significant negative correlation between age and IQ, and between age and adaptive behavior (DQ) scores, in children with FRAXA and WBS, but not in children with NF1. All three groups of children have unusually high proportions of maladaptive behavior, ranging from 1/6 children with NF1 to 2/3 children with FRAXA. Cognitive and adaptive behavior profiles of children with FRAXA and WBS were also surprisingly similar. Our findings suggest the need for examining longitudinal developmental cognitive-behavioral changes in children and adolescents with all genetic disorders that produce LD or MR.

Large-scale copy number variants (CNVs): distribution in normal subjects and FISH/real-time qPCR analysis.

BACKGROUND: Genomic copy number variants (CNVs) involving >1 kb of DNA have recently been found to be widely distributed throughout the human genome. They represent a newly recognized form of DNA variation in normal populations, discovered through screening of the human genome using high-throughput and high resolution methods such as array comparative genomic hybridization (array-CGH). In order to understand their potential significance and to facilitate interpretation of array-CGH findings in constitutional disorders and cancers, we studied 27 normal individuals (9 Caucasian; 9 African American; 9 Hispanic) using commercially available 1 Mb resolution BAC array (Spectral Genomics). A selection of CNVs was further analyzed by FISH and real-time quantitative PCR (RT-qPCR). RESULTS: A total of 42 different CNVs were detected in 27 normal subjects. Sixteen (38%) were not previously reported. Thirteen of the 42 CNVs (31%) contained 28 genes listed in OMIM. FISH analysis of 6 CNVs (4 previously reported and 2 novel CNVs) in normal subjects resulted in the confirmation of copy number changes for 1 of 2 novel CNVs and 2 of 4 known CNVs. Three CNVs tested by FISH were further validated by RT-qPCR and comparable data were obtained. This included the lack of copy number change by both RT-qPCR and FISH for clone RP11-100C24, one of the most common known copy number variants, as well as confirmation of deletions for clones RP11-89M16 and RP5-1011O17. CONCLUSION: We have described 16 novel CNVs in 27 individuals. Further study of a small selection of CNVs indicated concordant and discordant array vs. FISH/RT-qPCR results. Although a large number of CNVs has been reported to date, quantification using independent methods and detailed cellular and/or molecular assessment has been performed on a very small number of CNVs. This information is, however, very much needed as it is currently common practice to consider CNVs reported in normal subjects as benign changes when detected in individuals affected with a variety of developmental disorders.

Trends in special education code assignment for autism: implications for prevalence estimates.

There is considerable controversy over reasons for observed increases in the prevalence of autism spectrum disorders. We examined trends in British Columbia education database coding of children with autism from 1996 to 2004. There was a significant linear increase in autism prevalence. The proportion of children identified by age 6 increased significantly from 1996 to 1999. When we calculated prevalence assuming onset prior to age 3, previously unidentified cases, particularly among girls in 1996 and 1997, accounted for substantial increases in estimated prevalence. The magnitude of under-identification decreased from 1996 to 2000, and rose slightly in 2001. Analyses of prevalence trends must take into account effects of earlier age at identification and inclusion of previously undetected cases on prevalence estimates.

Using cluster ensemble and validation to identify subtypes of pervasive developmental disorders.

Pervasive Developmental Disorders (PDD) are neurodevelopmental disorders characterized by impairments in social interaction, communication and behavior. Given the diversity and varying severity of PDD, diagnostic tools attempt to identify homogeneous subtypes within PDD. Identifying subtypes can lead to targeted etiology studies and to effective type-specific intervention. Cluster analysis can suggest coherent subsets in data; however, different methods and assumptions lead to different results. Several previous studies applied clustering to PDD data, varying in number and characteristics of the produced subtypes. Most studies used a relatively small dataset (fewer than 150 subjects), and all applied only a single clustering method. Here we study a relatively large dataset (358 PDD patients), using an ensemble of three clustering methods. The results are evaluated using several validation methods, and consolidated through an integration step. Four clusters are identified, analyzed and compared to subtypes previously defined by the widely used diagnostic tool DSM-IV.