Introduction
Autism is a well-defined and validated child psychiatric disorder. It is characterized by impairments in social interaction and communication, restricted repetitive and stereotyped patterns of behaviour, interests and activities, and an onset prior to age 3 years (APA, 1994). Individuals diagnosed with pervasive developmental disorder – not otherwise specified (PDD-NOS) are characterized by the same set of impairments, but fail to meet the full criteria for autism and are therefore, often considered to have a milder variant of the disorder. However, the boundaries between PDD-NOS and other developmental disorders are poorly defined and unarticulated. Although PDD-NOS is supposedly a residual category of autism, epidemiological studies have shown that the prevalence of PDD-NOS outnumbers autism by up to three times (Wing & Gould, Reference Wing and Gould1979; Fombonne, Reference Fombonne2003, Reference Fombonne2005) and many children with PDD-NOS occupy beds in child psychiatric wards and have treatment needs similar to those of children with autism. The lack of explicit and positive diagnostic criteria for PDD-NOS hampers neurobiological and epidemiological research of its aetiology. Although there is discussion in the literature on different subtypes, there seems to be consensus on the necessity of the issue on differentiating within the PDD group for different reasons such as obtaining more insight in the neurobiology, developmental trajectories and predictive issues. To address this issue, Cohen and colleagues have attempted to formulate criteria that would define subgroups of PDD-NOS (Cohen et al. Reference Cohen, Paul and Volkmar1986, Reference Cohen, Paul, Volkmar, Cohen, Donnellan and Pauls1987). Perhaps the best defined and validated PDD-NOS subgroup to date is multiple complex developmental disorder (MCDD) (Towbin et al. Reference Towbin, Duykens, Pearson and Cohen1993; Van der Gaag et al. Reference Van der Gaag, Buitelaar, van den Ban, Bezemer, Njio and Van Engeland1995; Buitelaar & Van der Gaag, Reference Buitelaar and Van der Gaag1998; Ad-Dab'bagh & Greenfield, Reference Ad-Dab'bagh and Greenfield2001). MCDD is characterized by (1) impaired regulation of affective state and anxiety, (2) impaired social behaviour and sensitivity, and (3) impaired cognitive processing (thought disorder) (Cohen et al. Reference Cohen, Paul and Volkmar1986; Towbin et al. Reference Towbin, Duykens, Pearson and Cohen1993; Van der Gaag et al. Reference Van der Gaag, Buitelaar, van den Ban, Bezemer, Njio and Van Engeland1995; Buitelaar & Van der Gaag et al. Reference Buitelaar and Van der Gaag1998). Similar to children with autism, children with MCDD are disturbed in their social interactions, communication, and display stereotyped and rigid behaviour (Van der Gaag et al. Reference Van der Gaag, Buitelaar, van den Ban, Bezemer, Njio and Van Engeland1995). However, these behaviours are typically less marked in MCDD than in autism. Children with MCDD are further similar to children with autism in that they show some developmental delays in a number of domains, including language (late speech development), motor development (late walking), as well as physicial development (young skeletal age and more neurological soft signs and clumsiness than in children with other psychiatric diagnoses) (Van der Gaag, Reference Van der Gaag1993). In contrast, children with MCDD are found to be more impaired than children with autism on measures of (pre-) psychotic thinking (e.g. overengagement with fantasy figures; magical thinking; irrationality; marked loosening of association; marked fantasy activity; bizarre delusions), aggression (e.g. hurting other people; aggressive behaviour ‘outside’ and within the family; lack of appreciation of danger to others; impulsive behaviour and pervasive hyperactivity), and a ‘suspicious’ approach (e.g. suspicious or paranoid; odd, one-sided inappropriate approach; feeling of loneliness) (Van der Gaag, Reference Van der Gaag1993; Van der Gaag et al. Reference Van der Gaag, Buitelaar, van den Ban, Bezemer, Njio and Van Engeland1995; Buitelaar & Van der Gaag, Reference Buitelaar and Van der Gaag1998). Furthermore, children with MCDD typically have later onset of symptoms than their counterparts with autism (95% of children with autism has onset of symptoms before 30 months of age, versus 43% of children with MCDD), as well as higher verbal IQ scores, more family adversity and a higher incidence of psychiatric problems, a higher incidence of abnormal EEGs, and a higher incidence of schizophrenia spectrum disorders later in life (Van der Gaag, Reference Van der Gaag1993; Van Engeland & Van der Gaag, Reference Van Engeland and Van der Gaag1994). Two studies investigating the neurobiological basis of MCDD have suggested that there may be differences between children with MCDD and autism in event-related potentials to visual odd-balls, as well as in cortisol response to psychosocial stress (Kemner et al. Reference Kemner, Van der Gaag, Verbaten and Van Engeland1999; Jansen et al. Reference Jansen, Gispen-de Wied, Van der Gaag and Van Engeland2003). Taken together, these reports suggest that children with MCDD may represent a subtype of PDD-NOS that is neurobiologically and clinically distinct from autism (Buitelaar & Van der Gaag, Reference Buitelaar and Van der Gaag1998).
Structural neuroimaging studies in children with autism have demonstrated that this disorder is associated with changes in brain volume compared to typically developing children. The most reliable finding to date is probably that total brain and intracranial volumes are increased in children with autism (Courchesne et al. Reference Courchesne, Karns, Davis, Ziccardi, Carper, Tique, Chisum, Moses, Pierce, Lord, Lincoln, Pizzo, Schreibman, Haas, Akshoomhoff and Courchesne2001; Cody et al. Reference Cody, Pelphrey and Piven2002; Sparks et al. Reference Sparks, Friedman, Shaw, Aylward, Echelard, Artru, Maravilla, Giedd, Munson, Dawson and Dager2002; Palmen & Van Engeland, Reference Palmen and Van Engeland2004; Palmen et al. Reference Palmen, Hulshoff Pol, Kemner, Schnack, Durston, Lahuis, Kahn and Van Engeland2005; Lainhart, Reference Lainhart, Bigler, Bocian, Coon, Dinh, Dawson, Deutsch, Dunn, Estes, Tager-Flusberg, Folstein, Hepburn, Hyman, McMahon, Minshew, Munson, Osann, Ozonoff, Rodier, Rogers, Sigman, Spence, Stodgell and Volkmar2006), although this enlargement may be limited to early childhood (Courchesne et al. Reference Courchesne, Karns, Davis, Ziccardi, Carper, Tique, Chisum, Moses, Pierce, Lord, Lincoln, Pizzo, Schreibman, Haas, Akshoomhoff and Courchesne2001; Aylward et al. Reference Aylward, Minshew, Field, Sparks and Singh2002), or to individuals with high-functioning autism (Akshoomoff et al. Reference Akshoomoff, Pierce and Courchesne2002; Palmen & Van Engeland, Reference Palmen and Van Engeland2004; Palmen et al. Reference Palmen, Hulshoff Pol, Kemner, Schnack, Durston, Lahuis, Kahn and Van Engeland2005).
We set out to investigate to what extent children with MCDD show changes in brain volume similar to those found in children with autism and whether there are changes that differentiate between these two forms of PDD. To limit the heterogeneity in the samples investigated and as previous studies of MCDD all investigated high-functioning individuals (Van Engeland & Van der Gaag, Reference Van Engeland and Van der Gaag1994; Van der Gaag et al. Reference Van der Gaag, Buitelaar, van den Ban, Bezemer, Njio and Van Engeland1995; Kemner et al. Reference Kemner, Van der Gaag, Verbaten and Van Engeland1999; Jansen et al. Reference Jansen, Gispen-de Wied, Van der Gaag and Van Engeland2003), we included only high-functioning individuals. We collected MRI scans from a group of high-functioning children with MCDD and compared brain volumetric measures to those from (1) typically developing children and (2) high-functioning children with autism, matched for age, IQ, hand preference, socio-economic status and gender. We hypothesized that children with MCDD would show changes in brain volume similar to those found in the children with autism, but to a lesser extent. Specifically, we hypothesized that they would have larger total brain volume and larger ventricles. However, we reasoned that if MCDD does truly represent a neurobiologically distinct subtype of PDD, with an increased risk of psychosis, this may be reflected in neurobiological measures and therefore we might also expect differences between brain changes in autism and MCDD. Therefore, we formulated the alternative hypothesis that children with MCDD might display the opposite of children with autism and have a decrease in the total brain volume. An enlargement of ventricle volumes would be expected in both cases. However, given that none of the children in our sample had a history of psychosis, we focused on the first hypothesis of brain enlargement.
Method
Subjects
Twenty-two male subjects meeting the criteria for PDD-NOS (APA, 1994) and MCDD as defined by Cohen et al. (Reference Cohen, Paul and Volkmar1986, Reference Cohen, Paul, Volkmar, Cohen, Donnellan and Pauls1987), and 21 subjects meeting the criteria for autism (APA, 1994) were included. All subjects with autism and MCDD were recruited through the (out)-patient clinic at the Department of Child Psychiatry Unit at the University Medical Centre, Utrecht or the National Autism Society in The Netherlands. The diagnosis was established by expert clinical opinion (H.vE., B.L.) and confirmed using the Autism Diagnostic Interview – Revised version (ADI-R; LeCouteur et al. Reference LeCouteur, Rutter, Lord, Rios, Robertson, Holdgrafer and McLennan1989). Subjects were required to meet ADI-R criteria for autism as they are typically employed in research studies (a score within 2 points of full criteria) (Cox et al. Reference Cox, Klein, Charman, Baird, Baron-Cohen, Swettenham, Drew and Wheelwright1999; Sparks et al. Reference Sparks, Friedman, Shaw, Aylward, Echelard, Artru, Maravilla, Giedd, Munson, Dawson and Dager2002; Palmen et al. Reference Palmen, Hulshoff Pol, Kemner, Schnack, Durston, Lahuis, Kahn and Van Engeland2005). One subject in the autism group did not meet this criterion, but was included in the study, as a panel of experts agreed that the low score on ADI interview was due to underreporting by the parent. Eighty-one percent (17 patients) of subjects with autism fulfilled the traditional ADI-R threshold in all three domains (Lord et al. Reference Lord, Rutter and Le Couteur1994); 95% (20 subjects) fulfilled the modified criteria. Fifty-nine percent (13 patients) of subjects with MCDD met traditional ADI-R threshold for autism, whereas 95% (21 subjects) of MCDD subjects reached criteria when modified research criteria were applied. All subjects with MCDD met Cohen's criteria for MCDD (a minimum of six of 14 criteria, including at least two criteria from each of the three domains, whereas none of the subjects with autism did (see Table 1; Cohen et al. Reference Cohen, Paul, Volkmar, Cohen, Donnellan and Pauls1987). Mean score on the MCDD criteria for the subjects with autism was 3.5, compared to 8.6 for subjects with MCDD (t=8.6, df=38, p<0.001). Subjects with MCDD were matched to 21 subjects with autism and 21 controls for gender, age, height, weight, IQ, hand preference and parental socio-economic status, expressed as the highest completed level of education by either parent (see Table 2). Results from the sample of subjects with autism and controls have been previously reported (Palmen et al. Reference Palmen, Hulshoff Pol, Kemner, Schnack, Durston, Lahuis, Kahn and Van Engeland2005). Subjects with major physical or neurological illness (e.g. epilepsy) or full-scale IQ <70 were excluded. All subjects were male and of Caucasian ethnicity. None of the subjects met criteria for any DSM-IV-TR diagnosis other than autism (autism group) or PDD-NOS (MCDD group). Ten of 22 subjects with MCDD were on medication. All ten were taking risperidone for typical symptoms of MCDD, such as problems with affective regulation, combined with episodes of aggressive behaviour, anxieties, disorganized behaviour and problems separating fantasy from reality. One patient was also taking a SSRI for affective symptoms and one patient was taking methylphenidate for attentional problems. All subjects that met inclusion criteria were asked to participate in a 35-minute MR scan and neuropsychological assessment in order to estimate full-scale IQ [Wechsler Intelligence Scale for Children – Revised (WISC-R)] (Vandersteene et al. Reference Vandersteene, Van Haassen, De Bruyn, Coetsier, Pijl, Poortinga, Lutje Spelberg, Spoelders-Claes and Stinissen1986). The procedure was approved by the institutional review board of the University Medical Centre Utrecht. MRI scans were evaluated by independent clinical neuroradiologists. No clinically relevant abnormalities were present in any of the subjects included in the study.
Table 1. Diagnostic criteria for multiple complex developmental disorder (MCDD; Cohen et al. Reference Cohen, Paul, Volkmar, Cohen, Donnellan and Pauls1987)
Table 2. Demographic data
MCDD, Multiple complex developmental disorder; ADI-R, Autism Diagnostic Interview – Revised version.
Missing data: a one; b two.
MRI scan acquisition and analysis
The same scanner and procedures as in the Palmen et al. (Reference Palmen, Hulshoff Pol, Kemner, Schnack, Durston, Lahuis, Kahn and Van Engeland2005) study were used. Magnetic resonance images were acquired on a Philips Gyroscan (Philips Medical Systems, Best, The Netherlands) at 1.5 T. For volumetric measurements T1-weighted 3D fast-field echo (FFE) scans with 130–150 1.5-mm contiguous coronal slices of the whole head (TE 4.6 ms, TR 30 ms, flip angle 30°, FOV 256 mm, in plane voxel size 1 mm×1 mm) and T2-weighted dual-echo turbo spin-echo scans with 65–75 3.0-mm contiguous coronal slices (TE1 14 ms, TE2 80 ms, TR 6350 ms, flip angle 90°, FOV 256 mm, in plane voxel size 1 mm×1 mm) were acquired. In addition, T2-weighted dual-echo turbo spin-echo scans with 17 axial 5-mm slices and a 1.2-mm gap (TE1 9 ms, TE2 100 ms, flip angle 90°, FOV 250 mm, in plane voxel size 0.98 mm×0.98 mm) were acquired for clinical neurodiagnostic evaluation. The processing pipeline has been described previously and included semi-automated assessment of intracranial volume, total brain volume, lateral ventricles, third ventricle and cerebellum, as well as fully automated assessment of grey- and white-matter volumes (Durston et al. Reference Durston, Hulshoff Pol, Schnack, Buitelaar, Steenhuis, Minderaa, Kahn and Van Engeland2004; Palmen et al. Reference Palmen, Hulshoff Pol, Kemner, Schnack, Durston, Lahuis, Kahn and Van Engeland2005). Grey–white matter differentiation could not be obtained for two subjects with MCDD. Analyses were performed by two independent raters, blind to subject identity and diagnosis. Half of the scans were randomly flipped over the y-axis to ensure raters were also blind to laterality. Intra-class correlation coefficients were calculated to estimate reliability and were >0.85 for all measures.
Statistical analysis
All statistical analyses were conducted using the SPSS statistical package, version 11.5 (SPSS Inc., Chicago, IL, USA). ANOVA was used to assess differences between the three groups (subjects with MCDD, subjects with autism, and controls), and significant differences were further investigated using post-hoc independent-sample t tests (two-tailed). In order to control for global effects of differences in intracranial volume, analyses that yielded significant results were repeated with intracranial volume as a covariate. All analyses that yielded significant results were repeated excluding subjects on medication.
Results
The overall ANOVA showed significant differences between groups for intracranial, total brain, cerebellum and ventricular volume (F>3.15, df=2, 61, p<0.05), as well as a trend for grey-matter volume (F=2.71, df=2, 59, p=0.08). Differences between the autism and control groups have been previously reported (Palmen et al. Reference Palmen, Hulshoff Pol, Kemner, Schnack, Durston, Lahuis, Kahn and Van Engeland2005) and are summarized in Table 3. Subjects with MCDD had a significantly larger cerebellum than controls (t=2.93, df=41, p=0.005). After correction for intracranial volume, the enlargement in cerebellum remained significant (F=8.24, df=41, p=0.007) and an increase in total grey-matter volume also reached significance (F=7.84, df=39, p=0.008). When subjects on medication were excluded from the analyses, differences remained significant for the cerebellum (t=2.72, df=31, p=0.011) and at trend level for grey matter (t=1.69, df=30, p=0.10). When comparing subjects with MCDD to subjects with autism, intracranial and ventricular volumes were significantly larger in subjects with autism than in those with MCDD (t>2.15, df=41, p<0.05) and an increase in total brain volume reached trend level (t=1.73, df=41, p=0.09). After correction for intracranial volume, differences were no longer significant. When subjects on medication were excluded from the analyses, differences remained significant for the ventricular volume (F=7.231, df=31, p=0.048), and differences for intracranial and total brain volume were still in the same direction but no longer significant, probably related to reduced power.
Table 3. Brain volumes in subjects with multiple complex developmental disorder (MCDD), autism and typically developing controls
Values are mean±s.d.
* Indicates a significant difference from the control group at the p<0.05 level.
Discussion
We report differences in structural imaging measures of brain volume in both children with MCDD and children with autism compared to typically developing controls. Both groups show increases in grey-matter and cerebellar volume, but we find no evidence of intracranial enlargement in MCDD, contrary to our findings in autism.
Differences between children with autism and control subjects have been previously reported for this sample and are not further discussed here (Palmen et al. Reference Palmen, Hulshoff Pol, Kemner, Schnack, Durston, Lahuis, Kahn and Van Engeland2005). They are reported solely to enable comparisons with differences in the MCDD group. Children with MCDD showed increases in cerebellar and overall grey-matter volume compared to controls, similar to our previous results in children with autism (Palmen & Van Engeland, Reference Palmen and Van Engeland2004; Palmen et al. Reference Palmen, Hulshoff Pol, Kemner, Schnack, Durston, Lahuis, Kahn and Van Engeland2005) as well as reports by others (Cody et al. Reference Cody, Pelphrey and Piven2002). As such, our first hypothesis that children with MCDD would display changes in brain volume similar to those found in autism was confirmed. This suggests that neurobiological changes in MCDD are similar to those seen in autism.
However, we also found differences between children with MCDD and those with autism: children with autism had larger intracranial and ventricular volumes than children with MCDD and showed a trend towards an increase in total brain volume. These differences were related to an overall difference in head size between subjects with autism and MCDD, as they were no longer significant when corrected for intracranial volume. This finding of an overall difference in head size confirms our second hypothesis that there are also differences between brain changes in these two forms of autism spectrum disorders. These changes are unlikely to be secondary to differences in clinical severity, as ADI scores were similar in both groups (see Table 2).
Our main finding is that children with MCDD show no enlargement of intracranial volume, whereas head size is significantly enlarged in children with autism (see Fig. 1). Intracranial volume increases during the first years of life, under the influence of the growing brain. It continues to grow up until approximately age 5 years and then stabilizes (Durston et al. Reference Durston, Hulshoff Pol, Casey, Giedd, Buitelaar and Van Engeland2001). The finding of differences in head size between these two forms of autism spectrum disorder is provocative as it suggests that neurobiological differences may map onto clinical differences. If the lack of intracranial enlargement can be confirmed in other studies, it suggests that the onset of MCDD may be later than that of autism. This is in line with clinical evidence, as parents often report later onset of symptoms in children with this form of autism spectrum disorder (Van der Gaag, Reference Van der Gaag1993). As such, these findings suggest that MCDD may represent a form of PDD that is neurobiologically distinct from autism, at least to the extent that it may have a different developmental trajectory with later onset.
Fig. 1. Mean (a) intracranial volume and (b) cerebellum volume [ml (±s.e.)] in controls, subjects with multiple complex developmental disorder (MCDD), and subjects with autism.
In sum, we report a pattern of volumetric changes in the brains of subjects with MCDD, similar to that seen in autism. However, contrary to our results with autism, we find no evidence of intracranial enlargement in MCDD. This suggests that although neurobiological changes associated with MCDD are similar to those in autism, there may be differences in the developmental trajectories associated with these two subtypes of autism spectrum disorders.
These types of observations, although preliminary, can contribute to our understanding of the neurobiology of autism spectrum disorders and eventually facilitate the development of more effective tools for diagnosis and possibly even treatment of these disorders. To date, the number of studies available on the clinical concept and neurobiological background of MCDD are limited. Although the findings of this study should be considered preliminary, it does potentially add to our biological understanding of this subtype of autism spectrum disorders. As such, if these results can be replicated, they may have implications for considering differences in developmental trajectories within the PDD spectrum disorders. However, at this time MRI remains purely a research tool in the evaluation of autism spectrum disorders. While it is valuable in furthering our understanding of the neurobiological substrates of this disorder, MRI scanning is not appropriate for diagnostic purposes.
Limitations
There are a number of limitations to this study. First, although not atypical of paediatric neuroimaging studies, the sample size of the groups is relatively small. As such, we cannot exclude the possibility that some of our results that did not reach significance are false-negative findings. However, the difference in intracranial volume between subjects with MCDD and controls was negligible (0.23 ml; see Fig. 1 and Table 3), making it most unlikely that this particular finding was secondary to limited power. Second, we included only high-functioning individuals in all groups, with average or above average IQ scores. Therefore it is unclear how well our findings generalize to lower-functioning individuals with diagnoses in the PDD spectrum. Third, all children with autism were medication-naive. As this is unusual in daily practice, it is raises the question of how representative this sample is of children with high-functioning autism. In contrast, half of the subjects with MCDD were using medication at the time of scan. However, all analyses that yielded significant results were repeated without these subjects, and the findings were comparable, although not always significant due to diminished power.
Acknowledgments
The authors acknowledge Rutger Jan van der Gaag, Margreet Scherpenisse, and Jacqueline Jansen, for their contribution to subject recruitment and screening. Research support was provided by the Korczak Foundation.
Declaration of Interest
None.
