The primary finding of this study was a significant reduction in midsagital vermian area, vermal lobes I-V, and vermal lobes VI-VII when adjusted for increases in total cerebral or cerebellum volume in the children with ASD as compared to the children with typical development. Children with developmental delay exhibited cerebellar volumes that were persistently smaller compared to children with typical development and ASD, after controlling for age, gender, and total cerebral volume. These findings are similar to previous reports suggesting hypoplasia of vermis VI-VII in children with autism and children with PDD aged 1.9 to 5.2 years (Akshoomoff et al., 2004
) and an overlapping sample of children with autism compared to typical controls aged 2 to 16 years (Courchesne et al., 2001
). This pattern of decreased vermal midsagittal area may be a general anatomical finding associated with ASD, consistent with the results of a meta analysis which noted that between 84%- 92% of individuals with ASD exhibited vermal hypoplasia (Courchesne et al., 1994
As well, we examined whether there existed a bimodal distribution for vermal areas and whether the distributions for the volume of the vermal lobules VI-VII for both the ASD and TD groups deviated from normal. Thus, while we did find evidence of a non-normal distribution, this pattern does not appear to be specific to the ASD sample. As seen in , the ASD and TD groups have extreme overlap in absolute (raw) values () and some overlap in volumes that are corrected for age, gender and cerebrum volume (). Furthermore, while the mean value of the ASD group significantly differed from the TD group, only a small number of individual had volumes that were beyond 1 SD of the mean.
Our findings are in contrast with Akshoomoff's report of increased anterior vermis areas (Akshoomoff et al., 2004
). Specifically, we found smaller anterior vermis I-V areas in children with ASD compared to TD (). Examining the variability in vermis I-V areas, as compared to the TD mean, 34% of the ASD sample and 71% of the DD sample had values 1 SD below the mean of the TD group, while 6% of the ASD sample and 0% of the DD sample had values 1 SD above the mean of the TD group. There are several demographic differences, as well, between the Akshoomoff sample and ours. Akshoomoff included only male
children between 1.9 and 5.2 years. Our sample included both male and female children between 3.0 and 4.8 years. More importantly, our analyses used gender, age, and total cerebral volume or cerebellum volume as covariates, whereas Akshoomoff did not. Given that Akshoomoff also found enlarged total brain volume in children with low (p
< .05) and high (p
= .08) functioning autism versus controls; it is possible that their pattern of findings would change if scaled for total cerebral volume.
In this study, we found no relation between cerebellar measures and measures of symptom severity and verbal or nonverbal IQ in the ASD group. However, there was a relation within the DD group with larger vermal area and cerebellum volume being correlated with better daily living skills, better communication, and more typical motor skills. It is unclear why these skills would be correlated within the DD group but not the ASD group. (Scores were not available for the TD sample.) One possibility is that children in the DD group represented the lower end of the normal distribution and thus demonstrate the same relation found in neurotypicals. In neurologically typical individuals, larger cerebellum size has been shown to be correlated with better neuro-physiological functioning, including memory, fine motor dexterity, and IQ (Andreasen et al., 1993
; Pardiso et al., 1997
There have been some structural correlations with neurophysiolgoical findings in ASD. Of note, in a recent report with the same group of children with ASD, we found that larger right amygdala volume, but not hippocampal volume, was predictive of poorer social and communication abilities (Munson et al., 2006
). There are several possible explanations to account for the absence of a relation between cerebellar measures and measures of autism symptom severity. First, it is possible that measures that more directly relate to cerebellar function are needed to fully explore the functional significance of these anatomical findings (e.g., manual dexterity such as finger tapping, verbal memory). Second, the cerebellum may mediate general behaviors (such as attention, exploration) that are reflective of the autism phenotype but not directly assessed by the measures in this study.
In contrast to autism, individuals with Williams Syndrome (WMS) show conserved cerebellar volume and disproportionately enlargement of VI-VII and VIII-X despite reduced cerebral volumes (Jernigan & Bellugi, 1990
; Reiss et al., 2000
; Schmitt et al., 2001
). Individuals with WMS are socially motivated, although the disorders share characteristics of abnormalities in motor coordination and gait. Of note, Schaefer et al. (1996)
have observed that hypoplasia of cerebellar vermal lobules VI and VII is a diagnostically nonspecific finding that occurs in a number of other developmental disorders. For example, decreases in vermal size have been reported in association with ADHD, fragile X, Schizophrenia, velo-cardio-facial syndrome, and Joubert syndrome (e.g., Holroyd et al., 1991
; Mostofsky et al., 1998
; Nopoulos et al., 1999
; Eliez et al., 2001
; Campbell et al., 2006
). Thus, cerebellum structures may be genetically unstable, environmentally susceptible, or sensitive to downstream alterations by other affected regions.
What might cause relative decreases in vermal area? It has been suggested that there are two types of cerebellar pathology in autism. First, there may be structural hypoplasia of posterior vermal lobes VI-VII and cerebellar hemispheres related to severe Purkinje neuron loss. Second, hypoplasia may result from a failure to develop appropriate prefrontal – neocerebellar circuitry. The posterior vermis has a long postnatal development period and develops in parallel with frontal and prefrontal structures (Schmanhmann, 1991
; Ciesielski et al., 1997
). Deficits at any levels of the circuitry may result in eventual neuronal loss in the cerebellum.
Although the cerebellar system it has consistently been identified as a region of abnormality in histopathological postmortem studies of autism, the link between anatomical measurements and the autism phenotype is unclear. Future investigations of cerebellar dysmorphology that include targeted functional assessments, are appropriate for across species use, and allow for the assessment of developmental stages of the disorder will provide valuable evidence about the role cerebellar circuitry plays in ASD.