Voxel-based morphometry of comorbid schizophrenia and learning disability: analyses in normalized and native spaces using parametric and nonparametric statistical methods T. William J. Moorhead, * Dominic E. Job, Heather C. Whalley, Tracy L. Sanderson, Eve C. Johnstone, and Stephen M. Lawrie Image Analysis Laboratory, Division of Psychiatry, Royal Edinburgh Hospital, University of Edinburgh, Edinburgh EH10 5HF, UK Received 22 July 2003; revised 15 October 2003; accepted 8 December 2003 We employed voxel-based morphometry (VBM) to compare the distributions of grey matter found in structural magnetic resonance imaging (MRI) brain scans of patients with comorbid learning disability with schizophrenia, schizophrenia alone, learning disability alone, and normal controls. Our primary aim was to replicate a previous region of interest (ROI) finding that comorbids and schizophrenics belong to the same population. Nonparametric analysis in normalized space showed no significant differences in grey matter distribution between the comorbid and schizophrenia groups. Fur- thermore, this analysis showed significant grey matter reductions in the comorbid and schizophrenia groups when compared to the learning- disabled or the normal controls. Parametric analysis localized the significant grey matter reductions between the normal controls and the comorbid and schizophrenia groups to the prefrontal and temporal lobes. It also identified an area of increased grey matter, on the inferior aspect of the postcentral gyrus, in the learning-disabled alone compared to the other groups. Native space parametric and nonparametric analyses, based on modulation of the normalized scans, confirmed the similarity in grey matter distribution of the comorbid and schizophrenia groups. Results confirm the ROI finding that in native space the learning-disabled group possesses the least and normal controls the most grey matter for the cohort. An increase in the basal ganglia of patients with schizophrenia vs. the learning-disabled, probably attributable to antipsychotic medication, was identified in the native space analysis. The native space results did not however register statistically significant temporal lobe reductions found under normalized analysis between schizophrenics and normal controls. This may be attributable to minor physical anomalies (MPA) in the schizophrenic cranium. Overall, these VBM results replicate previous ROI findings and are compatible with the view that comorbid learning disability with schizophrenia is a severe form of schizophrenia, rather than a consequence of learning disability. VBM has the facility to compare grey matter distributions in this structurally diverse cohort. D 2004 Elsevier Inc. All rights reserved. Keywords: Schizophrenia; Learning disability; Voxel-based morphometry Introduction Systematic reviews of the literature reveal consistent evidence that schizophrenia is associated with reductions in the volumes of the whole brain, temporal lobes, and amygdalo-hippocamal com- plexes, while the lateral ventricles are increased in size (Lawrie and Abukmeil, 1998; Wright et al., 2000). A neuropathology review by Harrison (1999) concluded that the pattern of brain abnormalities in schizophrenia is suggestive of disturbance in connectivity within and between affected regions that most likely occurs during brain development. Marenco and Weinberger (2000) reviewed literature relating to the neurodevelopmental hypothesis of schizophrenia and concluded that it is likely that a brain insult before or at birth plays a role in the expression of schizophrenia, but environmental variables after birth cannot be ruled out. While it is generally accepted that there is a biological basis for many cases of learning disability and those structural studies that have been conducted demonstrate abnormalities (Deb, 1997), the research base in this area is very much weaker than it is for schizophrenia. Sanderson et al. (1999, 2001) examined a group of comorbid schizophrenia with learning disability that had been identified from the records of inpatient care and discharged between 1970 and 1993. This was part of a wide-ranging study designed to examine the relationship between comorbid learning disability and schizo- phrenia and the separate conditions alone (Doody et al., 1998). The neuroanatomy of these patients was compared to that of age- and sex-matched schizophrenia, learning-disabled, and normal controls. The correlations of Sanderson et al. (2001) between clinical and historical variables, and cerebral structural variables were indicative of a strong neurodevelopmental component in the presentation of comorbid learning disability with schizophrenia. Our prime aim in the work described in this paper was to replicate, through voxel- based morphometry (VBM) grey matter analysis, the region of interest (ROI) finding by Sanderson et al. (1999) that the comorbid and schizophrenia groups belong to the same population. We also wished to assess the feasibility of applying automated VBM analysis to the disparate set of brains presented by the study cohort. VBM provides for the grouped analysis of tissue distribution (Ashburner and Friston, 2000; Friston et al., 1995a; Wright et al., 1995). This is an automated methodology that avoids the scrutiny 1053-8119/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.neuroimage.2003.12.012 * Corresponding author. Image Analysis Laboratory, Division of Psychiatry, Royal Edinburgh Hospital, The University of Edinburgh, Morningside Park, Edinburgh EH10 5HF, UK. Fax: +44-131-537-6531. E-mail address: tmoorhea@staffmail.ed.ac.uk (T.W.J. Moorhead). Available online on ScienceDirect (www.sciencedirect.com.) www.elsevier.com/locate/ynimg NeuroImage 22 (2004) 188 – 202