Brain Anatomy in Adults With Velocardiofacial Syndrome With and WithoutSchizophrenia: Title and subTitle BreakPreliminary Results of a Structural Magnetic Resonance Imaging Study FREE

Velocardiofacial syndrome (VCFS), also known as 22q11 deletion syndromeor DiGeorge syndrome, is a genetic disorder that occurs in approximately 1in 4000 to 5000 live births.^{1 - 2} In85% of people with VCFS, an approximately 3-megabase deletion of 22q11 isdetected with fluorescence in situ hybridization.³ Thepresence of VCFS is associated with a characteristic physical phenotype (includingcongenital cardiovascular anomalies and velopharyngeal insufficiencies), mildto borderline learning disability, and specific cognitive deficits (eg, inobject perception, planning, and abstract reasoning).^{4 - 7} Inaddition, psychiatric problems are frequently reported in children with VCFS.These include social withdrawal, phobia, depression, attention-deficit/hyperactivitydisorder, and autistic spectrum disorder.^{8 - 10} Inadult life, people with VCFS are at increased risk of developing psychosis,particularly schizophrenia.¹¹ The reportedprevalence rate for psychosis (including bipolar disorder, and schizophreniaand related disorders) is approximately 30%, but methodologic differencescomplicate the interpretation of these studies, and some involve adults andothers include adolescents.^{11 - 14} Ina large study, Murphy et al¹¹ found psychosisin 30% (schizophrenia in 24%) of adults with VCFS.

Schizophrenia is a heterogeneous disorder that is likely to be causedby interaction of several susceptibility genes and environmental risk factors.The morbid risk of schizophrenia for a patient with VCFS is approximately25 times that of the general population, and, thus, possession of chromosome22q11 deletion, apart from being the offspring of 2 schizophrenic parentsor having a schizophrenic monozygotic co-twin, is the highest known risk factorfor the development of schizophrenia.¹⁵ Thissuggests that deletion of 1 or more gene(s) mapping to chromosome 22q11 underliessusceptibility to psychosis in VCFS.¹⁵ Thestudy of VCFS therefore provides a unique opportunity to increase our understandingof the neurobiology of schizophrenia in the general population.

There is growing consensus from a large body of in vivo neuroimagingstudies that people with schizophrenia have several structural brain abnormalities.Increased volume of cerebral ventricles is one of the most consistently reportedfindings, together with reduction in the volume of total brain and gray matter(for a review, see Shenton et al¹⁶ ). Also,localized volume and gray-matter reductions have frequently been describedin several brain regions, mostly implicating temporolimbic and frontal neocorticalregions, whereas there is less support for cerebellar abnormalities.^{16 - 18} In contrast, therehave been fewer studies on white-matter volumes, but recent studies usingnewer techniques including diffusion tensor imaging,^{19 - 20} magnetictransfer imaging,²¹ and microarray²² suggest that white-matter integrity, including oligodendrogliaand myelination, may be compromised as well in schizophrenia.²³

There are relatively few neuroimaging studies of people with VCFS. Casereports and qualitative studies have reported that people with VCFS have ahigh prevalence of white-matter hyperintensities (WMHIs) (which may reflectabnormalities in myelination and high water content²⁴ ),cavum septum pellucidum–cavum vergae, and a small cerebellar vermis.^{25 - 29} Quantitativestudies reported that learning-disabled children with VCFS, when comparedwith healthy children of normal intelligence, have a generalized reductionin volume of both cerebral hemispheres (mostly affecting white matter), combinedwith increased volume of frontal lobe and decreased volume of left parietalgray matter.^{30 - 31} There are, toour knowledge, only 2 quantitative neuroimaging studies in adults with VCFS.Our group reported that, when compared with IQ-matched controls, adults withVCFS have a smaller cerebellar volume, widespread deficits in white matter,and localized gray-matter deficits in temporal and cerebellar regions.²⁹ That study, however, did not differentiate betweenindividuals with VCFS with and without schizophrenia. Recently, Chow et al³² compared adults with VCFS and schizophrenia and healthycontrols without VCFS, and reported that adults with VCFS and schizophreniahad a smaller volume of total gray matter; regional differences in gray andwhite matter in frontal, temporal, parietal, and occipital lobes; and an increasedvolume of ventricular and sulcal cerebrospinal fluid (CSF) bilaterally. Althoughthis study was an important first step, their control group was not matchedfor the presence of VCFS or IQ, and so it is unclear whether the differencesthey reported are due to the presence of VCFS, schizophrenia, learning disability,or all 3.

To our knowledge, there have been no studies comparing the brain anatomyof adults with VCFS with and without schizophrenia. We therefore extendedour previous work and examined the brain structure of schizophrenic (S-VCFS)and nonschizophrenic (NS-VCFS) adults with VCFS and chromosome 22q11 deletion,and a healthy IQ-matched control group by means of structural magnetic resonanceimaging (MRI). We tested the hypotheses that (1) adults with S-VCFS have asignificant reduction of brain volume in frontal and temporolimbic brain regionsreflecting brain abnormalities associated with schizophrenia in the generalpopulation; and (2) both patients with S-VCFS and NS-VCFS would have a significantreduction of cerebellar volume reflecting a brain abnormality commonly associatedwith VCFS.

Approval for the study was granted by the local ethics committee, andall subjects (or their guardians, in case subjects were unable to consent)gave written informed consent after the procedure was fully explained. Allpatients with VCFS and control subjects were screened for medical conditionsaffecting brain function by means of a semistructured clinical interview androutine blood tests. Full-scale intelligence was measured by means of theCanavan et al shortened version of the Wechsler Adult Intelligence Scale–Revised,consisting of 5 subtests: vocabulary, comprehension, similarities, block design,and object assembly.³³ We included 25 subjectswith clinical features of VCFS and a 22q11 deletion detected by fluorescencein situ hybridization (Oncor Inc, Gaithersburg, Md). They volunteered andwere eligible to undergo MRI. All individuals with VCFS were interviewed byone of us (K.C.M.) by means of a semistructured psychiatric interview (Schedulesfor Clinical Assessment in Neuropsychiatry)³⁴ toestablish a DSM-IV diagnosis as described previously.¹¹ Fifteen individuals with VCFS (8 with schizophreniaand 7 without) were part of the original sample reported by Murphy and colleagues,¹¹ while 10 individuals with VCFS (5 with schizophreniaand 5 without) were recruited from the Behavioral Genetics Clinic, MaudsleyHospital, London, England. The VCFS group was subdivided into 2: those whomet DSM-IV criteria for schizophrenia (n = 13;7 women and 6 men; mean ± SD age, 34 ± 11years; IQ, 69 ± 8; all taking antipsychotic medication andhaving a duration of illness >1 year, 2 hospitalized at time of scanning)and those who had no history of psychosis (n = 12; 8 women and 4men; age, 31 ± 10 years; IQ, 74 ± 9). Weincluded a healthy control group, recruited from local community centers forpeople with mild or borderline learning disabilities and/or by local advertising(n = 14; 8 women and 6 men; age, 36 ± 10 years;IQ, 75 ± 16). Controls were included after screening forpsychiatric disorders and medical conditions affecting brain function andafter a deletion at 22q11 was excluded by fluorescence in situ hybridizationtesting.

Magnetic resonance imaging of the brain was performed on a 1.5-T MRIsystem (Signa; General Electric Co, Milwaukee, Wis) at the Maudsley Hospital.A coronal volumetric spoiled gradient acquisition in the steady state dataset covering the whole head was acquired (repetition time,13.8 milliseconds;echo time, 2.8 milliseconds; 124 sections; 1.5-mm section thickness). Thisdata set was used to perform manual tracing of brain volumes.³⁵ Inaddition, we acquired a whole-brain near axial dual-echo fast spin-echo dataset aligned with the anterior commissure–posterior commissure plane(repetition times, 4000 milliseconds; effective echo times, 20 and 85 milliseconds;3-mm section thickness). This data set was used to determine between-groupdifferences in gray- and white-matter volume by using a previously publishedmethod.^{29 ,36 - 37} Threetypes of analysis were performed, one qualitative and two quantitative; allwere blind to subject group status.

Both MRI data sets were assessed qualitatively by a neuroradiologist.The presence and extent of ventricular white-matter hyperintensities (WMHIs)were assessed by means of a 4-point rating scale adopted from Kozachuk etal²⁴ as follows: grade 0, ventricular WMHIsabsent; grade 1, frontal or occipital caps or pencil-thin lining of the lateralventricle; grade 2, smooth halo surrounding the lateral ventricles; and grade3, irregular ventricular WMHIs extending into the deep white matter. DeepWMHIs were graded as follows: grade 0, absent; grade 1, punctuate foci eitherfocal or symmetric; grade 2, mild confluence of foci; and grade 3, large confluenceof foci. Peripheral WMHIs were graded similarly to deep WMHIs. Congenitalabnormalities in cerebellum and cerebrum including cavum septum pellucidum–septumvergae were noted as being present or absent.

As described previously, volumetric analysis of total and regional brainareas was performed on a reformatted spoiled gradient acquisition in the steadystate data set by means of Measure software.³⁵ Total,right, and left caudate, putamen, hippocampus, amygdala, and frontal, occipitoparietal,and temporal lobes; cerebral hemispheres; and cerebellum, brainstem, and ventricularCSF volumes were traced by means of region of interest boundaries as previouslydescribed.^{29 ,36 ,38 - 40} Rightand left hemispheres, cerebral ventricles, and subcortical gray regions weremeasured on images aligned along the anterior commissure–posterior commissureline. The frontal lobe was defined as all supratemporal structures anteriorto the sylvian aqueduct. Temporal lobe was defined, from the anterior poleof the temporal lobe to the sylvian aqueduct. The medial temporal lobe boundarywas defined as a straight line from the angle of the medial temporal lobe,where it attaches to the temporal stem, to the midpoint of the operculum;the dura of the middle cranial fossa was then traced around each temporallobe to complete the region. The parietal lobe was defined as brain matterposterior to the sylvian aqueduct, extending to the medial transverse fissureof striate cortex. Remaining caudal portions of the cerebral hemispheres weredefined as parieto-occipital lobe. Regions of cerebellar brain and brainstemwere measured in the posterior fossa.

Images were realigned parallel to the sylvian fissure for hippocampaland amygdalar measures. The hippocampus was measured starting at the sectiondisplaying the sylvian aqueduct. Continuing anteriorly, the superior borderof the hippocampus merges with the inferior border of the amygdala. The hippocampus-amygdaladelineation is marked by white matter and the temporal horn of the lateralventricle. If that delineation is unclear, the inferior border of the posterioramygdala is marked arbitrarily by a horizontal line drawn medially from thehead of the temporal stem to the medial border of the amygdala; the amygdalais taken to be gray matter superior to that line.

The volume of each region was calculated by multiplying the summed pixelcross-sectional areas by section thickness. Intrarater and interrater reliabilities(range, 0.90-0.99) were determined by intraclass correlation computation forall brain regions traced by the operators and were highly significant (F>4.0and P<.002).⁴¹

Voxels representing extracerebral tissue were automatically set to zero,⁴² and the probability of each intracerebral voxel belongingto gray matter, white matter, CSF, or dura-vasculature tissue classes wasthen estimated by a modified fuzzy clustering algorithm.⁴³ Onthe basis of previous results, we equated these probabilities to the proportionalvolumes of each tissue class in the often heterogeneous volume of tissue representedby each voxel.⁴⁴ Thus, for example, if theprobability of gray-matter class membership was 0.8 for a given voxel, itwas assumed that 80% of the tissue represented by that voxel was gray matter.Because the voxel size was predetermined (2.2 mm³), we then estimatedthe volume in milliliters of gray matter, white matter, and CSF in each voxel.Summing these voxel tissue class volumes over all intracerebral voxels yieldedglobal tissue class volumes.

To allow estimation of between-group differences at each intracerebralvoxel (spatial extent statistics), the short echo (proton density–weighted)fast spin-echo images were coregistered by means of an affine transformation^{45 - 46} with a template image in the coordinatesystem of standard space as defined by Talairach and Tournoux.⁴⁷ Thisindividually estimated transformation was then applied to each of that subject'sgray- and white-tissue probability maps.

Between-group differences in age and IQ were assessed by means of a1-way analysis of variance and χ² test for sex distribution(P<.05, 2-tailed).

Group differences in frequencies of structural abnormalities were assessedwith the χ² test, whereas between-group differences in extentof WMHIs were assessed with a 1-way analysis of variance, with level of significancefor both tests at P<.05, 2-tailed.

Manually traced volumes (Measure) were analyzed with SPSS 10.0 for Windows(SPSS Inc, Chicago, Ill). Data were first examined for normality to conformto the assumptions of the parametric statistics used. Between-group differencesin uncorrected total regional brain volumes were calculated by means of aunivariate general linear model (GLM) with group (schizophrenic [S-VCFS] ornon-schizophrenic VCFS [NS-VCFS], controls) and sex (male, female) as thebetween-subject variables, and age and total intracranial volume (ICV) ascovariates and, where appropriate, Bonferroni adjustments for multiple comparisons.The significance level was defined as P<.05, 2-tailed.

Fast spin-echo data were unavailable for 2 of the patients with S-VCFSand 3 of the controls. Total gray- and white-matter and CSF volumes in theS-VCFS, NS-VCFS, and control groups were compared by univariate GLM usingsex, age, and ICV as covariates (SPSS 10.0) and, where appropriate, Bonferroniadjustments for multiple comparisons. Between-group differences in gray andwhite matter were localized by fitting an appropriate GLM at each intracerebralvoxel. Inference was via a permutation distribution of spatial extent statisticswith significance levels set to control for multiple comparisons by havingless than one estimated false-positive region (cluster) across the image (P<.001). In brief, the processing proceeded as follows.Maps of the standardized GLM model coefficient of interest (group) at eachvoxel were thresholded such that only voxels with a probability less than.05were retained. The sum of voxelwise statistics for each 3-dimensional suprathresholdcluster was the test statistic, the sign indicating a relative excess or deficitin local tissue density. Significance testing of the clusters was performedwith a null distribution of this test statistic similarly obtained after repeatedlyrandomly permuting the relevant factor in the GLM and refitting of the model.⁴⁸

There were no significant between-group differences in age (F = 0.71, P = .50), IQ (F = 1.1, P = .33), or sex distribution (χ² = 0.45, P = .80).

The prevalence of cavum septum pellucidum–septum vergae was notequally distributed over the 3 groups: 31% (4 patients) in the S-VCFS group,50% (6 patients) in the NS-VCFS group, and 0% in the control group (χ² = 7.8, P = .02). However,there were no significant between-group differences in any other qualitativevariable we measured or in severity of WMHIs (Table 1). However, when both VCFS groups were combined, the presenceof WMHIs was more common than in controls (χ² = 5.1, P = .02).

There was a significant group (F_2,32 = 3.4, P = .04) and sex (F_1,32 = 10.8, P = .002) effect on ICV, but no group × sexinteraction. Pairwise comparisons with Bonferroni adjustments for multiplecomparisons showed a significantly decreased ICV in the S-VCFS group comparedwith the control group and in the female compared with the male group. Inaddition, after ICV was added as a covariate to the model, there was a significantgroup effect on volume of total (F_2,31 = 6.08, P = .006), left (F_2,31 = 7.99, P = .002), and right (F_2,31 = 5.1, P = .01) cerebral hemispheres; total (F_2,31 = 5.4, P = .01), andright (F_2,31 = 5.5, P = .009)frontal lobes; total (F_2,31 = 5.1, P = .01), left (F_2,31 = 6.02, P = .006), and right (F_2,31 = 3.35, P = .048) temporal lobes; cerebellum (F_2,31 = 14.1, P<.001); brainstem(F_2,31 = 4.02, P = .03);and total sulcal CSF (F_2,31 = 9.3, P = .001) (Table 2).Pairwise comparisons with Bonferroni adjustments for multiple comparisonsshowed that cerebellar volume was significantly smaller in both S-VCFS andNS-VCFS groups than in controls. Furthermore, total and left cerebral hemispherevolumes were significantly smaller and sulcal CSF volume significantly largerin the S-VCFS group compared with both the NS-VCFS and the control groups.Decreases in volume of right hemisphere; total and right frontal lobe; total,left, and right temporal lobe; and brainstem volume were observed in the S-VCFSgroup compared with the control group only. There was a significant effectof age on total (F_1,31 = 9.14, P = .005),left (F_1,31 = 6.36, P = .02),and right (F_1,31 = 12.02, P = .002)frontal lobes, but no age × group interactions. There wereno significant effects of age and sex on any of the other brain structures.

When the level of significance was adjusted according to the numberof tests performed (P = .001), the onlyfindings that remained significant were those of group effects on cerebellumand sulcal CSF.

Total tissue class volumes are shown in Table 3. There was no significant effect of group on total gray-mattervolume. However, there was a significant effect of group on volume of totalwhite matter (F_2,26 = 8.73, P = .001)and total CSF (F_2,26 = 8.57, P = .001).Pairwise comparisons with Bonferroni adjustments for multiple comparisonsshowed that total white-matter volume was significantly decreased and totalCSF volume significantly increased in the S-VCFS group compared with boththe NS-VCFS and the control groups. Also, there was a significant effect onage for total gray matter (F_1,26 = 12.12, P = .002) and white matter (F_1,26 = 10.56, P = .003), and a sex × groupinteraction for white matter (F_2,26 = 3.88, P = .03) (with men in the S-VCFS group showing a greaterreduction in white-matter volume than women). After correction for the numberof tests performed, we adjusted the level of significance to P = .001, and this still yielded significant group effectson white matter and total CSF volume.

The central coordinates and volumes of the 3-dimensional clusters ofbrain tissues that were significantly different (P = .001)are shown in Table 4. The S-VCFS groupcompared with the control group had 6 significant gray-matter deficit regions,2 in (left and right) cerebellum, 1 in right superior temporal gyrus, and3 in right frontal (midfrontal, inferior frontal, and anterior cingulate gyrus)regions. Also, 1 gray-matter excess region was identified in the S-VCFS group;this was centered in the right anterior cingulate gyrus (Figure 1). The NS-VCFS compared with the control group had 1 gray-matterdeficit region that was located in right cerebellum extending to left cerebellum(Figure 2). In addition, there were3 gray-matter excess regions all centered in the precentral regions, bothleft and right. Comparisons between the 2 VCFS groups showed 2 gray-matterexcess regions in the S-VCFS compared with the NS-VCFS group, located in leftprecentral regions. There were no gray-matter deficit regions in the S-VCFSgroup compared with the NS-VCFS group (Figure3).

White-matter deficits in the S-VCFS group compared with the controlgroup were concentrated in 4 spatially extensive regions all covering frontallobe regions: 2 involving left and right precentral gyrus, 1 extending toleft anterior cingulate, and 1 involving right medial frontal region (Figure 4). In contrast, 1 area of excess white-mattervolume was observed centered in the brainstem of the S-VCFS group. In theNS-VCFS group, compared with controls, one cluster of white-matter deficitwas identified, and this was centered in right fasciculus longitudinalis superiorextending into right inferior frontal lobe (Figure5). Also, in the NS-VCFS group, significant excess white matterwas localized in 2 clusters, left and right fasciculus occipitofrontalis.Within-VCFS-group comparisons showed 1 cluster of white-matter excess in theS-VCFS group compared with the NS-VCFS group, located in the posterior cingulate;there were no white-matter deficit regions (Figure6).

We believe that this is the first (preliminary) study to investigatebrain anatomy in individuals with VCFS with and without schizophrenia withan IQ-matched healthy control group used as a reference. We found that bothS-VCFS and NS-VCFS groups had a smaller volume of cerebellum and increasedfrequency of cavum septum pellucidum–septum vergae. In addition, theS-VCFS group (compared with both control and NS-VCFS groups) had a significantreduction in volume of both total (white + gray) brain matter andtotal white matter. The S-VCFS group also had a significant increase in volumeof total and sulcal CSF. It is unlikely that our findings can be explainedby differences in IQ, age, sex, or nonspecific factors associated with beinglearning disabled, as the groups did not differ in IQ and we controlled forage and sex distribution.

Qualitative analysis of our data showed abnormalities that have beenpreviously reported in people with VCFS^{27 - 28} andin non–VCFS-related schizophrenia.^{49 - 50} Cavumseptum pellucidum was observed in both VCFS groups equally frequently, irrespectiveof whether schizophrenia was present. Septum pellucidum abnormalities areof unknown clinical significance and are also seen in people with and withoutlearning disabilities without mental illness^{29 ,51} (althoughthey were not present in our control group). The high prevalence of WMHIsin the majority of subjects in both VCFS groups at a relatively young agecould indicate white-matter tract disruption possibly of a vascular origin,although we did not use a method that is sensitive to fiber tract visualization,like diffusion tensor imaging.⁵² However, inour study these abnormalities were equally common in patients with VCFS withand without schizophrenia; there were no significant between-group differencesin frequency or severity of WMHIs, and WMHIs were also present in the controlgroup. However, when the 2 VCFS groups were combined, the frequency of WMHIwas significantly higher than in the control group. Thus, WMHIs and septumpellucidum abnormalities are frequently present in individuals with VCFS withand without schizophrenia and are most likely not related to the presenceor absence of schizophrenia.

Our quantitative analysis showed that the S-VCFS group, compared withboth the NS-VCFS and control groups, had a generalized reductionin total brain (gray + white matter) volume and total white-mattervolume and an increase in total and sulcal volume of CSF. However, we alsofound that the S-VCFS group (compared with the control group only) had a reductionin volume of the frontal and temporal lobes, although these findings disappearedafter correction for multiple comparisons. These findings are in agreementwith results from other MRI studies of schizophrenia in both the general population¹⁸ and children⁵³ andadults with VCFS.³² That white matter may beparticularly affected in VCFS has been supported by findings in children withVCFS,³⁰ and a recent diffusion tensor imagingstudy of children with VCFS has shown generalized white-matter tract disruption.⁵⁴

In contrast to a previous report in people with VCFS,³² wedid not find a significant diffuse loss of gray matter in adults with S-VCFS.We suggest that our results differ from those of Chow et al³² becausethey included non-VCFS controls with an above-average level of intellectualfunctioning (mean IQ, 116). Thus, the differences they observed in total gray-mattervolume could have been confounded by differences in subject intelligence andhealth status; eg, there is increasing evidence that IQ is positively correlatedwith total brain matter and cortical gray-matter volume.^{55 - 58} Itis not generally agreed as to which is the “best” control groupto use when studying people with genetically determined neurodevelopmentaldisorders such as VCFS. Our borderline–learning disabled control groupis not representative of a healthy population, and we cannot exclude an effectof genetically and environmentally determined causes of cognitive impairmentthat we did not detect by our screening techniques. However, ability to matchon IQ is important also in people with VCFS, as is supported by Kates et al,³⁰ who could not exclude an effect of IQ on their results.

Both the S-VCFS and NS-VCFS groups had a smaller total cerebellar volumeand cerebellar gray-matter volume than the control group. This finding furthersupports the results from our previous work³² andother qualitative²⁵ and quantitative⁵⁹ studies in children with VCFS. Others have suggestedthat in the general population schizophrenia is associated with cerebellarabnormalities.⁶⁰ Our results suggest that cerebellarabnormalities are more common in people with VCFS than IQ-matched controls;however, they are not related to the presence or absence of schizophrenia.

In contrast, the S-VCFS group demonstrated generalized loss of totalbrain (gray + white) and total white-matter volume, together withan increase in total and sulcal CSF volume, compared with both NS-VCFS andcontrol groups. In addition, our quantitative analysis showed regional differences in gray and white matter. These were most pronouncedas gray-matter deficits in cerebellar areas in both VCFS groups and differencesin anterior-posterior distribution between deficit and excess clusters, particularlyof white-matter frontal lobe regions in adults with S-VCFS and NS-VCFS comparedwith controls. Previous MRI studies in children and adults with VCFS reporteddifferences in development of frontal lobe.^{30 - 33} Moreover,abnormalities in the function, structure, and metabolism of frontal regionshave been frequently reported in the non-VCFS schizophrenic population. Forexample, both gray-matter⁶¹ and white-matter^{37 ,62} deficits have been reported in frontalbrain regions, and findings from longitudinal MRI studies^{63 - 64} suggesta progressive volume reduction in frontal regions, including frontal whitematter.⁶⁵ Normal brain maturation takes placelast in the frontal regions during adolescence and is accompanied by bothincreased synaptic pruning and myelination, leading to a reduction of volumein gray matter and an increase in white matter.⁶⁶ Wefound that people with S-VCFS had a reduction in whole-brain volume, but withregional differences in gray and white matter with an anterior-posterior distributionbetween deficits and excess areas particularly affecting frontal regions.Thus, our findings might suggest that people with VCFS have abnormalitiesin frontal maturation, and this could predispose to the development of schizophrenia.We therefore suggest that 22q11 deletion is associated with the presence ofa generalized disturbance in brain development that increases the liabilityfor developing schizophrenia. However, schizophrenia may develop only whenregionally specific neurodevelopmental differences of frontal regions alsooccur, leading to changes in the volume and tissue composition of frontalregions.

We did not find support for our hypothesis that people with S-VCFS havea reduced hippocampal volume, but we found deficits in temporal gray matterin our voxelwise analysis and reduced temporal lobe volume compared with thehealthy control group. This finding is in agreement with results by Chow etal,³² who found reduced temporal gray-mattervolumes in their sample of adults with VCFS and schizophrenia, compared withnormal-IQ controls, a finding also frequently reported in non–VCFS-associatedschizophrenia.

The cause of the differences in brain anatomy of people with VCFS, itsrelation to schizophrenia, and the time course of their development is unknown.Our study was cross-sectional, and therefore we cannot determine whether thesedifferences in brain anatomy change across the lifespan. Nonetheless, we foundthat sulcal CSF volume was increased in adults with S-VCFS, and this addstentative support to the hypothesis that schizophrenia is associated withan abnormal neurodevelopmental process that is not limited to an aberrationin prenatal brain development, but may be progressive. Also, our results donot allow us to draw conclusions as to whether schizophrenia in VCFS is dueto a primary abnormality in gray-matter development (eg, differences in programmedcell death [apoptosis]), with secondary changes of white-matter structure,or vice versa. There is preliminary evidence to suggest that differences inprogrammed cell death may underlie the differences in brain anatomy we foundin our study. For example, recent studies report that PRODH, a gene located at 22q11, is implicated in apoptosis⁶⁷ andthat variation within the PRODH/DGR6 locus at 22q11might contribute to VCFS-associated schizophrenia,⁶⁸ althoughthe latter study has not been replicated.⁶⁹ Alternatively,schizophrenia in VCFS may be related to developmental differences in white-matterstructure (eg, possibly as a result of abnormal myelination), with consequentdisturbances in connectivity between neocortical and limbic gray-matter regions.Schizophrenia in the general population is increasingly seen by some as asupraregional disorder involving differences in interconnecting white-mattertracts³⁷ and neuronal connectivity⁷⁰ possibly associated with oligodendroglial dysfunctionresulting in abnormalities in myelin maintenance and repair.²³ Infavor of this argument is the finding that in children with VCFS (who areat risk of developing schizophrenia) white matter may be more compromisedthan gray matter, and WMHIs are common.^{30 - 31} Also,we found overall white matter to be compromised in S-VCFS, whereas there wereno between-group differences in total gray-matter volume.

Another potential candidate gene for schizophrenia in VCFS is the catechol O-methyltransferase gene (COMT;the enzyme that degrades dopamine). The COMT genelies within the 22q11 region, and therefore people with VCFS have a reducedgene dose of COMT. Also, a recent study demonstratedthat variation in activity of COMT may have neurobiologicaleffects specific to the prefrontal cortex and may modulate the risk of schizophreniain the general population.⁷¹ Moreover, othershave suggested that in schizophrenia, neurodevelopmental abnormalities ofprefrontal dopaminergic systems result in enhanced vulnerability to sensitizationduring late adolescence and early adulthood.⁷² Althoughthere is little evidence yet that COMT plays a majorrole in the development of schizophrenia in VCFS,^{73 - 74} itcannot be excluded that reduced gene dose of COMT affectsthe integrity of the dopaminergic system, including its prefrontal projections,and so increases susceptibility to schizophrenia in people with VCFS.

Our sample size and the cross-sectional design are limitations of thisstudy, but we were still able to detect several significant differences witha large effect size and sufficient power; as noted by others,²⁸ theeffect size for structural brain abnormalities in this group of patients isrelatively large. In addition, we carried out multiple statistical comparisonsand thereby increased the risk of a type I error (false-positive outcomes).However, we believe this is unlikely to fully explain our results. After adjustingfor the number of tests carried out, the group differences in volumes of totalwhite matter, total and sulcal CSF, and cerebellum remained. Also, we foundsignificant effects on manually traced brain volumes (eg, reduced cerebellarvolume) that were consistent with our computerized voxelwise analysis. Moreover,in our computerized voxelwise analysis of gray- and white-matter volumes,the level of significance adopted was chosen specifically to yield less thanone false-positive cluster. We did not control for effects of neurolepticmedication, parental IQ, and parental head circumference, and therefore wecannot exclude these as potential confounds of our results. All subjects inthe S-VCFS group were taking antipsychotic medication, whereas none of thesubjects in the other 2 groups did. We did not find a sex effect except inICV. This may be due to the small sample size of our study. Alternatively,sex differences in raw volumes may have disappeared after controlling forICV. Also, we did not quantify white-matter lesions and did not relate themto other outcome variables.

In conclusion, our results, although preliminary, suggest that structuralbrain abnormalities present in people with S-VCFS are partially similar tothose seen in people with schizophrenia in the general population. Our findingsare compatible with an abnormal neurodevelopmental process that may be progressive.We suggest that this results in generalized abnormalities in brain anatomy,but particularly affecting white matter and frontal brain regions. Largerand longitudinal studies are planned to replicate these findings, and to determinehow genetic and environmental variables are related to the development ofschizophrenia in VCFS.

Submitted for Publication: November 1, 2002;final revision received April 5, 2004; accepted April 21, 2004.

Correspondence: Therese van Amelsvoort,MD, PhD, MRCPsych, Department of Psychiatry, Academic Medical Centre, Tafelbergweg25, Amsterdam 1105 BC, the Netherlands (t.a.vanamelsvoort@amc.uva.nl).

Funding/Support: This study was funded in partby The Stanley Foundation (Muscatine, Iowa) and the Medical Research Council(London, England).

Previous Presentations: This study was presentedin part at the Third International VCFS Meeting, June 8, 2002; Rome, Italy;and was a poster presentation at the meeting of the Society of BiologicalPsychiatry; May 15-17, 2003; San Francisco, Calif.