Available online at www.sciencedirect.com

103 (2008) 1–10www.elsevier.com/locate/schres

Schizophrenia Research

Corpus callosum shape alterations in individuals prior to theonset of psychosis

Mark Walterfang a,b,c,g,⁎, Alison Yung e,f, Amanda G. Wood d, David C. Reutens d,Lisa Phillips f, Stephen J. Wood a,b,g, Jian Chen d, Dennis Velakoulis a,b,c,

Patrick D. McGorry e, Christos Pantelis a,b,h

a Melbourne Neuropsychiatry Centre and Department of Psychiatry, University of Melbourne, Australiab North Western Mental Health Program, Sunshine Hospital, Australia

c Royal Melbourne Hospital, Melbourne, Australiad Department of Neurosciences, Monash Medical Centre, Melbourne, Australia

e ORYGEN Research Centre, Early Psychosis Prevention and Intervention Centre (EPPIC),Personal Assistance and Crisis Evaluation (PACE) Clinic, Australia

f Department of Psychiatry University of Melbourne, Australiag Department of Psychology, University of Melbourne, Australiah Centre for Neuroscience and Howard Florey Institute, Australia

Received 13 February 2008; received in revised form 20 April 2008; accepted 28 April 2008Available online 18 June 2008

Abstract

Reductions in the size of the anterior callosum have been described for both first-episode schizophrenia-spectrum psychosis andestablished schizophrenia, but have not been examined in individuals at ultra-high risk for psychosis (UHR). We compared 100 UHRindividuals (27 of whom later developed psychosis) with 38 age-matched control subjects on measures of size and shape of the corpuscallosum to determine if changes previously demonstrated in first-episode and established schizophrenia are present in the pre-psychotic phase. Each individual's callosum was extracted from the mid-sagittal slice from T1-weighted magnetic resonance images,and total area, length and curvature of the callosum was compared using one-way ANOVA, and 39 regional thicknesses via a non-parametric permutation method to account for non-independence of adjacent measures. Total area, length and curvature did not differbetween the groups. Compared to both the UHR-NP group and controls, the UHR-P group showed significant regional reductions inthe region of the anterior genu of the callosum. The UHR-NP group did not differ from controls. Positive and negative symptoms didnot affect regional thickness in either of the patient groups. Cox regression showed that mean anterior genu thickness was highlypredictive of a transition to psychosis. Reductions in the thickness of the anterior callosum differentiate between high-risk individualswho transition to psychosis and those who do not, and is highly predictive of transition. These changes may reflect primary pathologyof orbitofrontal and medial frontal cortex, or deficits in anterior interhemispheric myelination.© 2008 Elsevier B.V. All rights reserved.

Keywords: Corpus callosum; Psychosis; High risk; Schizophrenia

⁎ Corresponding author. Melbourne Neuropsychiatry Centre and Department of Psychiatry, University of Melbourne, Australia. Tel.: +61 393428750; fax: +61 3 93428483.

E-mail address: mark.walterfang@mh.org.au (M. Walterfang).

0920-9964/$ - see front matter © 2008 Elsevier B.V. All rights reserved.doi:10.1016/j.schres.2008.04.042

2 M. Walterfang et al. / Schizophrenia Research 103 (2008) 1–10

1. Introduction

The identification of individuals, who later developpsychotic illness, whilst in the pre-psychotic phase mayallow for the targeting of interventions to prevent, delayor attenuate the course of a psychotic disorder (Wyatt,1991). This recognition of the role that early interventionmay have on modifying the illness course has promptedthe search for factors that may identify individuals whoare at very high risk of psychotic illness (Yung et al.,1998), enabling “indicated prevention” to be undertakenin this group through intervention in the pre-psychoticphase (Mrazek and Haggerty, 1994). In 1994, thePersonal Assessment and Crisis Evaluation (PACE)Clinic was established in Melbourne, Australia tofacilitate prospective study of the development ofpsychotic illnesses, and uses a ‘close-in strategy’ toidentify combinations of putative state and trait riskfactors that define a target population at ‘ultra-high risk’(UHR) of an impending psychotic episode (McGorryet al., 2001; Yung et al., 1995). This strategy has beenshown to identify a group of young people with a 40%chance of developing a psychotic illness within a 12-month period (Yung et al., 2003a). Subjects receive anMRI brain scan at baseline and are followed clinically fora minimum of 1 year, and a range of neurobiologicalindices are measured at baseline and follow-up with theaim of identifying factors that predict a transition fromthe UHR phase to frank psychotic illness.

Through careful follow-up of UHR individuals in thePACE cohort, factors such as olfactory identification(Brewer et al., 2003), verbal memory and spatialworking memory function (Brewer et al., 2005; Woodet al., 2003), and hypothalamic–pituitary axis function(Thompson et al., 2007) have been shown to be at leastpartially predictive of the transition to psychosis.Additionally, we have been able to identify a numberof neuroimaging indices that differentiated patients inthe UHR group who later became psychotic (UHR-P)from those who remained non-psychotic at follow-up(UHR-NP), including pituitary volume (Garner et al.,2005) and thickness of the anterior cingulate cortex(Fornito et al., 2007, in press). In a longitudinalneuroimaging study that followed a group of UHRindividuals from the pre-psychotic phase, we showedthat those in the UHR-P group showed right medial andlateral temporal, right inferior frontal, and bilateralcingulate cortex reductions when compared to the UHR-NP group (Pantelis et al., 2003), and have also shownthat accelerated grey matter loss occurs in prefrontalcortical regions from the pre-psychotic period throughthe transition to psychosis (Sun et al., 2007a,b).

The common neuroanatomical origin of the majorityof these indices, which appear to be strong neurocognitiveand neuroimaging trait markers for the development ofpsychotic illness (Brewer et al., 2006; Pantelis et al.,2007), appears to be in the anterior cortex. We haverecently shown that reductions in the genu of the corpuscallosum are present in both first-episode and establishedschizophrenia patients (Walterfang et al., in press), whichcarry fibres that connect contralateral inferior frontal andprefrontal regions. Subtle reductions in callosal size arewell-described in schizophrenia (Woodruff et al., 1995),although no studies have examined this structure in thepre-psychotic phase. Given our findings of inferior frontalreductions in those UHR individuals who later becamepsychotic compared to those who did not (Pantelis et al.,2003), we speculated that these anterior callosal changesseen at the first-episode of schizophrenia may be presentin pre-psychotic individuals and may differentiate themfrom individuals identified in the UHR group who do notprogress to psychosis.

2. Methods

2.1. Subjects

The ultra-high risk group (N=100) was recruited fromthe Personal Assessment and Crisis Evaluation (PACE)Clinic, Melbourne, Australia (McGorry et al., 2001; Yunget al., 2003b) and had not experienced a previouspsychotic episode. UHR identification criteria have beenpreviously described (Yung et al., 2003b) and subjectswere included in the study if they were between the agesof 14–30, psychotropic-naïve at study entry and had beenfollowed up for at least 12 months in order to determinewhether they developed a psychotic illness.Of this cohort,27 individuals developed a psychotic illness (UHR-P) and73 did not (UHR-NP) over the period of follow-up. ThisUHR cohort has been previously described in other work(Garner et al., 2005; Velakoulis et al., 2006). The controlparticipants (n=38) were recruited by approachingancillary hospital staff and through advertisements.These subjects were recruited from similar socio-demo-graphic areas as the patients. Demographic data (age andgender) was obtained on all subjects, and handednessrated with the Edinburgh Handedness Inventory (Old-field, 1971) (Table 1). Pre-morbid intelligence quotient(IQ) was rated using the National Adult Reading Test(NART) (Nelson and Willison, 1991). The patient groupwas rated on the Brief Psychiatric Rating Scale (BPRS)(Lukoff et al., 1986; Overall and Gorham, 1962) and theScale for the Assessment of Negative Symptoms (SANS)(Andreasen, 1983).

Table1

Dem

ographic

andmaincallo

saldata

onmainparticipantgroups

Sub

ject

grou

pNum

ber

(M/F)

Age

(years)

Height(cm)

Pre-m

orbidIQ

Handedn

ess

(R/L/M

/U)

Callosalarea

(mm

2)

Callosal

leng

th(m

m)

Callosalbending

angle(degrees)

Meancallo

sal

width

(mm)

BPRStotal

score

SANSTo

tal

Score

UHRtotal

100(59/41

)20

.18(3.29)

171.93

(9.25)

96.53(13.57

)87

/8/2/3

662.17

(94.26

)10

1.28

(7.03)

94.58(6.16)

7.22

(0.90)

18.76(6.51)

23.57(16.44

)UHRpsycho

tic27

(18/9)

18.72(2.60)

172.22

(10.29

)94

.17(13.52

)24

/2/0/1

648.06

(107.02)

100.53

(6.75)

93.92(6.89)

7.13

(0.94)

19.16(6.02)

28.89(16.52

)UHRno

n-psychotic

73(41/32

)20

.72(3.37)

171.29

(8.93)

97.25(13.77

)63

/6/2/2

667.39

(89.32

)10

1.56

(7.15)

94.82(5.90)

7.25

(0.89)

18.61(6.74)

21.51(16.11)

Con

trol

subjects

33(23/15

)21

.02(3.40)

174.15

(9.06)

101.77

(10.66

)33

/4/1/0

656.26

(97.16

)10

4.45

(16.12

)95

.07(7.80)

7.14

(1.14)

N/A

N/A

Ingend

er,M

=maleandF=female.In

handedness,R

=right,L=left,M

=mixed,U

=un

know

n.BPR

S=BriefPsychiatricRatingScale.S

ANS=ScheduleforAssessm

ento

fNegativeSym

ptom

s.

3M. Walterfang et al. / Schizophrenia Research 103 (2008) 1–10

Subjects were screened for co-morbid medical andpsychiatric conditions by clinical assessment, andphysical and neurological examination. Writteninformed consent was obtained from all subjects. Thestudy was approved by the local Research and EthicsCommittee. Exclusion criteria for patients were: ahistory of significant head injury, seizures, neurologicaldiseases, impaired thyroid function, steroid use or DSM-IIIR criteria of alcohol or substance dependence.Control subjects with a personal history of psychiatricillness or family history of psychosis were excluded.

2.2. Magnetic resonance scanning acquisition

All subjects were scanned on a 1.5 T GE Signa MRImachine. A three-dimensional volumetric spoiled gra-dient recalled echo in the steady state sequence generated124 contiguous, 1.5 mm coronal slices. Imagingparameters were: time-to-echo, 3.3 ms; time-to-repeti-tion, 14.3ms; flip angle, 30°; matrix size, 256×256; fieldof view, 24 × 24 cm matrix; voxel dimensions,0.938×0.938×1.5 mm. Head movement was minimisedby foam padding and velcro straps across the foreheadand chin. This scanner was calibrated fortnightly usingthe same proprietary phantom to ensure stability andaccuracy of measurements. A numerical code was usedto ensure blind analysis of data.

2.3. Image processing

The brain was automatically segmented from the restof the head (Smith, 2002). Using the software packageAutomated Image Registration (Woods et al., 1998),images were registered to a template image comprisingthe average of 152 normal T1-weighted MRI scanspreviously placed in stereotaxic coordinate space. A 9-parameter linear transformation was used which allowedtranslation, rotation and scaling along each of the threeprincipal axes. The mid-sagittal slice was identified andinterpolated to a voxel dimension of 0.5 mm×0.5 mm inthe y and z planes. White matter voxels in the mid-sagittal slice were identified using a histogram segmen-tation procedure (Otsu, 1979). Non-callosal voxels werethen removed manually. A measure of callosal area intotal mm2 was then generated. To measure regionalcallosal thickness, voxels at the edge of the callosumwere identified and upper and lower edges were definedaccording to anterior and posterior end points. Aniterative search for optimum end points that maximisedthe length of a line segment traversing the centre of thecallosum was then performed (Fig. 1). The line segmentwas defined by dividing the upper and lower surfaces of

Fig. 1. Extraction of the corpus callosum from mid-sagittal image. Topshows mid-sagittal callosal image, middle shows binarised callosalimage, and bottom shows mid-spline traversing callosum between twoendpoints and callosal thicknesses orthogonal to equidistant pointsalone the mid-spline.

Fig. 2. Major callosal metrics. Top, total callosal area; middle, mid-spline length of the callosum; bottom, callosal bending angle.

4 M. Walterfang et al. / Schizophrenia Research 103 (2008) 1–10

the callosum into 40 equidistant portions by 39 nodes.The midpoints between corresponding nodes on theupper and lower surfaces were identified. The linesegment was created by joining endpoints and succes-sive midpoints. Once the optimum endpoints andcorresponding midpoints were identified, a smoothcurve joining them was obtained with cubic splineinterpolation, and the anteroposterior length of thiscurve was measured (callosal length). This curve wasdivided into 40 segments of equal lengths by 39 nodes.At each node, the line orthogonal to the curve wascalculated. The distance between its intersection withthe dorsal and ventral surfaces of the callosumrepresented regional callosal thickness at these 39points, and the mean of these 39 regional thicknessesrepresented mean callosal thickness. Finally, a simplemeasure of curvature, the callosal bending angle, wasobtained by measuring the angle between two vectors,

each joining each endpoint of the callosum to themidpoint, along the mid-spline of the callosum. Thechief measure types are illustrated in Fig. 2.

2.4. Statistical analysis

Analysis for between-group differences in demo-graphic variables was undertaken with Chi-squareanalyses for gender and handedness, and t-tests fortwo-group comparisons of age in years, years ofeducation, and pre-morbid IQ measured using theNational Adult Reading Test (NART). Unitary callosalmeasures such as total callosal area, callosal length andcallosal bending angle were compared between groupsusing t-tests. For regional callosal thickness, a non-parametric permutation method of 20,000 randomisa-tions was used for all group comparisons, to account fornon-independence between adjacent thickness measure-ments and for multiple comparisons (Holmes et al.,1996). A non-parametric multiple analysis of variancewas used to look for an overall effect of group. Step-down t-testing to determine which regions showedsignificant change was planned to localise between-

Fig. 3. Mean callosal slice thickness (from slice 1, anterior, to slice 39,posterior) in controls, psychotic and non-psychotic high-risk individuals.

5M. Walterfang et al. / Schizophrenia Research 103 (2008) 1–10

group differences in regional callosal thickness, andnon-parametric multiple regression was used to deter-mine the effect of symptom variables on regionalcallosal thickness. Statistical inference was based onthe method of Holm, which controls for multiplecomparisons of non-independent measures by control-ling the family-wise error rate without assumingindependence (Holm, 1979). A Cox regression analysiswas performed to determine whether any differencesfound in the between-group comparison of regionalthickness were predictive of later psychosis, usingtransition to psychosis as the study variable and regionalthickness and symptom measures as covariates.

3. Results

3.1. Demographic data

When the UHR group as a whole was compared tothe control group, no differences were seen in measuresof age (t=1.32, p=0.19), although the PACE groupshowed a trend to having a lower IQ (t=1.86. p=0.07).There were no differences in gender (χ2 =0.03, p=0.87)and handedness (χ2 =0.22, p=0.89). When the UHR-Pand UHR-NP groups were compared, the only sig-nificant difference was age, with the UHR-P being amean of two years younger (t=−3.13, pb0.005).

3.2. Major callosal metrics

When controls were compared against the UHRgroup as a whole, no between-group differences wereseen in callosal area (t=−0.11, p=0.91), length(t=1.34, p=0.18), bending angle (t=1.00, p=0.32) ormean thickness (t=−0.41, p=0.68). UHR-P and UHR-NP groups did not differ on measures of callosal area(t=−0.91, p=0.37), length (t=−0.64, p=0.52), meanthickness (t=−0.66, p=0.51) or bending angle (t=−0.65, p=0.52). When we included gender in theanalysis, there was no effect, and no group by genderinteraction, on any of these variables.

3.3. Regional callosal thickness

Mean callosal thicknesses of the three main groups(controls, UHR-P and UHR-NP) are shown in Fig. 3.When the UHR-P group was compared to controls,significant reductions were seen in slices 1–3, again inthe anterior genu, and slice 15, in the posterior genu(pb0.05, Fig. 4). When the UHR-P and UHR-NPgroups were compared, the UHR-P showed significantlysmaller anterior callosal thicknesses in slices 1–4, in the

anterior genu (pb0.05, Fig. 4). Although in the latteranalysis differences were not significant at slice 15, theeffect size was identical at this slice in each analysis(Cohen's d=0.46 and 0.47 respectively), suggestingthat this is related to reduced sample size and power inthe latter analysis.

When the UHR-P group was restricted to those whodeveloped a schizophrenia-spectrum illness (UHR-Scz,schizophrenia, schizophreniform disorder or schizoaf-fective disorder, n=17), reductions were seen comparedto controls at slice 1 (pb0.01, Fig. 4), and compared tothe UHR-NP group at the same slice (pb0.01).Although the significance in this restricted analysiswas not seen in slices 2–4, in all UHR-P/UHR-Scz vsUHR-NP/control analyses, effect size was between0.55–0.75 for slices 1–4 in all analyses. No significantchanges in any callosal region were seen when theUHR-NP group was compared to controls.

Because of the small but significant group differ-ences in age, a regression analysis was used todetermine the effect of age on regional thickness ineach group. In controls, age was positively correlatedwith thickness at slices 7–9 and 30–36 at the pb0.05level. In the UHR-P group, a similar positive effect wasseen at slices 5–8 and 35–37, and in the UHR-NPgroup, slices 5–9 and 34–38. These changes, in the mid-genu and splenium, generally excluded the area ofbetween-group difference in genu thickness, suggestingthese changes were not due to age alone.

Because differences were found in slices 1–4between UHR-P and UHR-NP groups, we constructeda mean thickness across slices 1–4 as the main covariatefor a Cox regression survival analysis to predicttransition to psychosis, and BPRS and SANS totalscores. Mean thickness in the anterior genu was a

Fig. 4. Reductions in between-group comparisons. Top, reductions inslices 1–4 in UHR-P compared to UHR-NP. Middle, reductions inslices 1–3 and slice 15 in UHR-P compared to controls. Bottom,reductions in UHR-Scz in slice 1 compared to controls.

6 M. Walterfang et al. / Schizophrenia Research 103 (2008) 1–10

significant predictor of transition to psychosis(Wald=11.19, p=0.001, β=0.52, 95% CI=0.35–0.76), suggesting that for every 1 mm reduction inmean thickness of this region, there was a 52% increasein risk for transition. The symptom scales SANS(p=0.54) or BPRS (p=0.65) were not predictive oftransition. When transition was restricted to those in theUHR-P group who developed a schizophrenia-spectrumillness (n=17), mean thickness across slices 1–4remained predictive (Wald=5.47, p=0.019, β=0.57,95% CI=0.35–0.91).

When the effect of symptom variables on regionalcallosal thickness was examined using multiple regres-sion, there was no relationship between BPRS andthickness in the UHR-P (p=0.34) or UHR-NP (p=0.33)groups, and no relationship between SANS andthickness in the UHR-P (p=0.28) or UHR-NP(p=0.74) groups.

4. Discussion

In this study of pre-psychotic individuals at ultra-high risk for developing psychosis, we found reductions

in the thickness of the genu of the corpus callosum inthose subjects who later developed a first-episodepsychosis (FEP) when compared to clinically similarsubjects who did not, and when compared to controls.This relationship held when the analysis was confined toindividuals who later developed a schizophrenia-spectrum illness. Additionally, pre-psychotic indivi-duals also showed reductions in the posterior genuwhen compared to controls.

A further significant finding in this study was thatreductions in the thickness of the anterior genu werepredictive of transition to psychosis in the high-riskgroup. The observed predictive validity of anteriorgenu thickness is unlikely to be due to baselinevariations in the UHR-P and UHR-NP groups' baselineclinical characteristics, since genu thickness waspredictive of transition to psychosis independent ofshared variance with baseline symptom ratings. Giventhat many of the fibres passing through the anteriorgenu connect bilateral orbitofrontal cortices, thesefindings are consistent with findings of impairedolfactory identification in the UHR-P group comparedto both UHR-NP individuals and healthy controls(Brewer et al., 2003). As both olfactory identificationability and genu thickness are related to orbitofrontalgrey matter, it may be that each of these measuresseparately indexes aberrant development of orbitofron-tal circuitry that precedes the onset of psychosis. Wecurrently have few neuroimaging predictors of transi-tion to psychosis (Pantelis et al., 2008); we havepreviously shown that pituitary volume (Garner et al.,2005) and thickness of the right anterior cingulategyrus (Fornito et al., 2007) are also predictive oftransition, and others have shown that medial temporalstructures are predictive in a genetically high-risksample (Lawrie et al., 2002). A number of thesepredictors are undergoing dynamic change over theadolescent and early adult period, or may be changingwith emergence of psychosis (Pantelis et al., 2008).This is best exemplified by changes in the pituitary, aslarger pituitary size correlates with reduced proximityto transition to psychosis (Garner et al., 2005). Thissuggests that a number of disparate brain regions andneurodevelopmental processes may be affected by theemergence of, or interaction with, psychosis (Panteliset al., 2007, 2005).

Our cross sectional study design does not allow us todefine the timing of the callosal changes. Thus, ourfindings could be explained by early neurodevelop-mental insults on the callosum. The callosum developsrelatively late during gestation compared to otherstructures, with the body appearing at 12 weeks and

7M. Walterfang et al. / Schizophrenia Research 103 (2008) 1–10

the genu at approximately 16 weeks (Kier and Truwit,1996). Intrauterine brain insults occurring at thebeginning of the second trimester could lead to impairedcallosal development with sparing of the callosal bodyand regional genu thinning. A significant body ofevidence suggests that neurodevelopmental insults inthe second trimester of pregnancy may explain somelater structural abnormalities in those developingschizophrenia (Avila et al., 2003; Marenco andWeinberger, 2000). This association is perhaps strongestfor maternal influenza infection (Brown et al., 2004,2000; Limosin et al., 2003; Shi et al., 2003). Recentanimal research has shown that mid-pregnancy sub-lethal infection with human influenza virus results insignificant alterations in expression of axon guidancegenes and white matter development (Fatemi et al.,2005, 2008), and adolescent offspring demonstrateenlarged ventricular volumes, reduced brain volumeand thinning of the corpus callosum (Fatemi et al.,2008). These animal studies suggest that prenatal insultssuch as influenza may interrupt crucial early callosaldevelopment, including axonal guidance across themidline, and/or myelination to result in callosalthinning.

Alternatively, it is possible that the observed callosalreductions are the result of post-natal rather thanprenatal neurodevelopmental disturbance. Job et al.showed that progressive reductions in pre-psychotichigh-risk individuals occur for up to three years prior tothe onset of frank psychosis (Job et al., 2005),suggesting that some of the findings in the pre-psychotic group may represent a progressive processoccurring in the adolescent neurodevelopmental phase.Neuroimaging and post-mortem analyses have shownthat white matter continues to develop throughout thisperiod, particularly in cortical regions such as thefrontal lobes, where white matter volume increases asgrey matter volume decreases (Sowell et al., 1999).This has been very well-defined in long associationtracts including the callosum (Benes et al., 1994; Pauset al., 1999). The callosum increases in the mid-sagittalarea into the third decade (Giedd et al., 1999; Pujol etal., 1993), with increased size being predominantly dueto increased myelination rather than axonal size(Aboitiz et al., 1992; LaMantia and Rakic, 1992). Asdevelopment of the callosum occurs rostrocaudally, thegenu matures later than the splenium (Lebel et al.,2008; Thompson et al., 2000), suggesting that itsdevelopmental cycle is longer than posterior regions. Ifthese developmental trajectories are altered or inter-rupted, altered adolescent and early adult myelinationprofiles of the callosum may be the result, which may

be manifest as subtle regional changes in callosalthickness.

Examining patients at different illness stages canprovide clues as to the timing of structural changes. Wehave previously shown that thickness reductions in theventral genu are present in both first-episode schizo-phrenia-spectrum psychosis and established schizophre-nia (Walterfang et al., 2008), and that these reductions arenot seen in first-episode affective psychosis (Walterfang etal., 2006). Additional reductions in the posterior genu/anterior midbody and isthmus were seen in subjects withestablished illness, suggestive of progressive change withongoing illness, although this may also represent a subsetof the first-episode sample that go on to develop a morechronic condition (Walterfang et al., 2006). The fact thatwe found posterior genu reductions in pre-psychoticindividuals is consistent with the hypotheses that eitherprogressive extragenual changes occur in establishedillness, or that extragenual changes may be a marker for amore severe or chronic illness course. Longitudinalstudies of pre-psychotic individuals are needed to assessthese possibilities.

While there are no current longitudinal studies ofcallosal thickness to help clarify many of the issuesraised above, recent studies have examined longitudinalbrain changes across the whole brain using a voxel-based approach. Using this approach in the ultra-highrisk group we have demonstrated pre-psychotic fronto-temporal grey matter reductions (Pantelis et al., 2003)and, more recently, pre-psychotic white matter expan-sions in frontotemporal and frontoparietal associationtracts bilaterally, suggesting that reductions in greymatter structures are not merely mirrored by reductionsin white matter (Walterfang et al., in press). More recentdevelopments in voxel-based approaches in whitematter that examine structural integrity as opposed tovolume, and that account for the inter-individualvariability of white matter tracts, suggest that volumetricreductions in grey matter are accompanied by reductionsin anatomically associated white matter tracts (Douaudet al., 2007). However, the nature, degree or direction-ality of causality cannot yet be inferred from thesestudies or their results.

The current study describes structural changes in thecallosum but does not allow us to specifically define thecause of these changes. Reductions in thickness of thegenu may represent a reduction in the degree ofmyelination of axons or reduced axonal size (whichwould increase axonal fibre density independent ofaxonal number), a reduced number of axons in the genu(which itself would not alter fibre density), or acombination of both. Previous reports of abnormalities

8 M. Walterfang et al. / Schizophrenia Research 103 (2008) 1–10

of T2 relaxation time and n-acetylaspartate concentra-tion in the genu in schizophrenia (Aydin et al., 2007;Flynn et al., 2003) provide some support for a reductionin myelination. Post-mortem studies have tended to findno difference in fibre number or density betweenschizophrenia patients and controls (Casanova et al.,1990; Machiyama et al., 1987; Nasrallah et al., 1983),although more recent work has found a reversal of theusual femaleNmale difference in fibre density across thecallosum in schizophrenia patients, and a reduction infibre number in females rather than males (Highley etal., 1999). However, we were unable to demonstrate anygroup by gender interactions with regard to callosalstructure in our sample.

In conclusion, our results identified pre-psychoticreductions in thickness of the genu of the corpuscallosum that are consistent with reductions seen in first-episode schizophrenia-spectrum (but not affective)psychosis and established schizophrenia. This suggeststhat these changes are present prior to the onset of frankpsychotic illness and may be markers for neurodevelop-mental insult to anterior brain regions. Additionally,measures of thickness of the genu may increase ourability to predict which individuals identified as beingprodromal or at ultra-high risk for the development ofpsychosis will develop a psychotic episode — particu-larly when combined with other neuroimaging measuressuch as pituitary volume or thickness of the cingulatecortex. Enhancing our capacity to detect individualswho will later develop psychosis may allow us to targetfuture interventions appropriately to this critical phaseof illness, and modify the illness course of psychoticdisorders.

Role of funding sourceThis research was supported by project grants from the National

Health & Medical Research Council (NHMRC; grant ID numbers:970598, 981112), Ian Potter Foundation, Woods Family Trust, and aprogram grant from the Victorian Health Promotion Foundation. Noneof the funding sources had any further role in study design; in thecollection, analysis and interpretation of data; in the writing of thereport; and in the decision to submit the paper for publication.

ContributorsDr Walterfang designed the study, wrote the protocol, performed

the analysis and prepared the first draft of the manuscript. Assoc ProfYung, Dr Phillips and Prof McGorry developed the PACE clinic,designed the entry criteria and undertook clinical assessments of theparticipants. The image analysis protocol and statistical analysis wasdeveloped and performed by Prof Reutens and Drs AWood and Chen.Dr S Wood, Dr Velakoulis and Prof Pantelis assisted with data analysisand interpretation of study findings, and assisted with re-drafting of themanuscript. All authors contributed to and approved the final

manuscript.

Conflict of interestAll authors declare they have no conflicts of interest which are

relevant to this manuscript.

AcknowledgementsThis research was supported by project grants from the National

Health & Medical Research Council (NHMRC; grant ID numbers:970598, 981112), Ian Potter Foundation, Woods Family Trust, anNHMRC Program Grant (350241) and a program grant from theVictorian Health Promotion Foundation. Dr Walterfang was supportedby a Stanley Research Centre Grant. Dr AWood was supported by anNHMRC Clinical Research Training Fellowship (251755). Dr S Woodwas supported by an NHMRC Clinical Career Development Award.Professor McGorry was supported by a NARSAD DistinguishedInvestigator Award. Dr Walterfang takes responsibility for the integrityof the data and the accuracy of the data analysis. All authors had fullaccess to all the data in the study.

Appendix A. Supplementary data

Supplementary data associated with this article canbe found, in the online version, at doi:10.1016/j.schres.2008.04.042.

References

Aboitiz, F., Scheibel, A., Fisher, R., et al., 1992. Fiber composition ofthe human corpus callosum. Brain Res. 598, 143–153.

Andreasen, N., 1983. The Scale for the Assessment of NegativeSymptoms (SANS). University of Iowa, Iowa City.

Avila, M., Sherr, J., Valentine, L., et al., 2003. Neurodevelopmentalinteractions conferring risk for schizophrenia: a study ofdermatoglyphic markers in patients and relatives. Schizophr.Bull. 29, 595–605.

Aydin,K.,Ucok,A.,Cakis, S., 2007.Quantitative protonMRSspectroscopyfindings in the corpus callosum of patients with schizophrenia suggestcallosal disconnection. Am. J. Neuroradiol. 28, 1968–1974.

Benes, F., Turtle, M., Khan, Y., et al., 1994. Myelination of a key relayzone in the hippocampal formation occurs in the human brainduring childhood, adolescence, and adulthood. Arch. Gen.Psychiatry 51, 477–484.

Brewer, W., Francey, S., Wood, S., et al., 2005. Memory impairmentsidentified in people at ultra high-risk for psychosis who laterdevelop first episode psychosis. Am. J. Psychiatry 162, 71–78.

Brewer, W., Wood, S., McGorry, P., et al., 2003. Impairment of olfactoryidentification ability in individuals at ultra-high risk for psychosiswho later develop schizophrenia. Am. J. Psychiatry 160, 1790–1794.

Brewer, W., Wood, S., Phillips, L., et al. (2006) Generalized and specificcognitive performance in clinical high-risk cohorts: a reviewhighlighting potential vulnerability markers for psychosis. 32.

Brown, A., Begg, M., Gravenstein, S., et al., 2004. Serologic evidenceof prenatal influenza in the etiology of schizophrenia. Arch. Gen.Psychiatry 61, 774–780.

Brown, A., Schaefer, C., Wyatt, R., et al., 2000. Maternal exposure torespiratory infections and adult schizophrenia spectrum disorders:a prospective birth cohort study. Schizophr. Bull. 26, 287–295.

Casanova, M., Sanders, R., Goldberg, T., et al., 1990. Axonal counts ofthe corpus callosum of schizophrenic patients. J. Neurol. Neurosurg.Psychiatry 53, 416–421.

9M. Walterfang et al. / Schizophrenia Research 103 (2008) 1–10

Douaud, G., Smith, S., Jenkinson, M., et al., 2007. Anatomicallyrelated grey and white matter abnormalities in adolescent-onsetschizophrenia. Brain 130, 2375–2386.

Fatemi, S., Pearce, D., Brooks, A., et al., 2005. Prenatal viral infectionin mouse causes differential expression of genes in brains of mouseprogeny: a potential animal model for schizophrenia and autism.Synapse 57, 91–99.

Fatemi, S., Reutiman, T., Folsom, T., et al., 2008. Maternal infectionleads to abnormal gene regulation and brain atrophy in mouseoffspring: implications for genesis of neurodevelopmental dis-orders. Schizophr. Res. 99, 56–70.

Flynn, S., Lang, D., Mackay, A., et al., 2003. Abnormalities ofmyelination in schizophrenia detected in vivo with MRI, and post-mortem with analysis of oligodendrocyte proteins. Mol. Psychiatry8, 811–820.

Fornito, A., Philips, L., Yucel, M., et al., 2007. Anatomicalabnormalities of the anterior cingulate cortex and paracingulatecortex prior to the onset of schizophrenia or affective psychosis.NeuroImage 36, S77.

Fornito, A., Yung, A., Wood, S., et al. (in press) Anatomical abnormalitiesof the anterior cingulate cortex prior to psychosis onset: an MRI studyof ultra-high risk individuals. Arch Gen Psychiatry.

Garner, B., Pariante, C., Wood, S., et al., 2005. Pituitary volumepredicts future transition to psychosis in individuals at ultra-highrisk of developing psychosis. Biol. Psychiatry 58, 417–423.

Giedd, J., Blumenthal, J., Jeffries, N., et al., 1999. Development of thehuman corpus callosum during childhood and adolescence: alongitudinal MRI study. Prog. Neuropsychopharmacol. Biol.Psychiatry 23, 571–588.

Highley, J., Esiri, M., McDonald, B., et al., 1999. The size and fibercomposition of the anterior commissure with respect to gender andschizophrenia. Biol. Psychiatry 45, 1120–1127.

Holm, S., 1979. A simple sequentially rejective multiple testprocedure. Scand. J. Statist. 6, 65–70.

Holmes, A., Blair, R., Watson, J., et al., 1996. Nonparametric analysisof statistic images from functional mapping experiments. J. Cereb.Blood Flow Metab. 16, 7–22.

Job, D., Whalley, H., Johnstone, E., et al., 2005. Grey matter changesover time in high risk subjects developing schizophrenia. Neuro-Image 25, 1023–1030.

Kier, E., Truwit, C., 1996. The normal and abnormal genu of thecorpus callosum: an evolutionary, embryologic, anatomic, and MRanalysis. AJNR 17, 1631–1641.

LaMantia, A., Rakic, P., 1992. Axon overproduction and elimination inthe corpus callosum of the developing rhesus monkey. J. Neurosci.10, 2156–2175.

Lawrie, S., Whalley, H., Abukmeil, S., et al., 2002. Temporal lobevolume changes in people at high risk of schizophrenia withpsychotic symptoms. Br. J. Psychiatry 181, 138–143.

Lebel, C., Walker, L., Leemans, A., et al., 2008. Microstructuralmaturation of the human brain from childhood to adulthood.NeuroImage 40, 1044–1055.

Limosin, F., Rouillon, F., Payan, C., et al., 2003. Prenatal exposure toinfluenza as a risk factor for adult schizophrenia. Acta Psychiatr.Scand. 107, 331–335.

Lukoff, D., Nuechterlein, K., Ventura, J., 1986. Manual for expandedBrief Psychiatric Rating Scale (BPRS). Schizophr. Bull. 12, 594–602.

Machiyama, Y., Watanabe, Y., Machiyama, R., 1987. Neuroanatomicalstudies of the corpus callosum in schizophrenia: evidence ofaberrant interhemispheric fibre connection. In: Takahashi, R., Flor-Henry, P. (Eds.), Cerebral Dynamics, Laterality and Psychopathol-ogy. Elsevier, Amsterdam, pp. 411–421.

Marenco, S., Weinberger, D., 2000. The neurodevelopmental hypoth-esis of schizophrenia: following a trail of evidence from cradle tograve. Dev. Psychopathol. 12, 501–527.

McGorry, P., Phillips, L., Yung, A., 2001. Recognition and treatmentof the pre-psychotic phase of psychotic disorders: frontier orfantasy? In: Johanssen, J. (Ed.), Early Intervention in PsychiatricDisorders. Kluwer, Amsterdam.

McGorry, P., Yung, A., Phillips, L., 2001. Ethics and early interventionin psychosis: keeping up the pace and staying in step. Schizophr.Res. 51, 17–29.

Mrazek, P., Haggerty, R., 1994. Reducing Risks for Mental Disorders:Frontiers for Preventive Intervention Research. National AcademyPress, Washington DC.

Nasrallah, H., McCalley-Whitters, M., Bigelow, L., et al., 1983. Ahistological study of the corpus callosum in chronic schizophrenia.Psychiatry Res. 8, 251–260.

Nelson, H., Willison, J., 1991. The Revised National Adult ReadingTest: Test Manual. NFER, Windsow.

Oldfield, R., 1971. The assessment and analysis of handedness: theEdinburgh inventory. Neuropsychologia 9, 97–113.

Otsu, N., 1979. A threshold selection method from gray levelhistograms. IEEE Trans. Syst. Man Cybern. 9, 62–66.

Overall, J., Gorham, D., 1962. The brief psychiatric rating scale.Psychol. Rep. 10, 799–812.

Pantelis, C., Velakoulis, D., McGorry, P., et al., 2003. Neuroanatomicalabnormalities before and after onset of psychosis: a cross-sectionaland longitudinal MRI comparison. Lancet 361, 281–288.

Pantelis, C., Velakoulis, D., Wood, S., et al., 2007. Neuroimaging andemerging psychotic disorders: the Melbourne ultra-high riskstudies. Int. Rev. Psychiatry 19, 371–379.

Pantelis, C., Yucel, M., Wood, S., et al., 2008. Neurobiologicalendophenotypes of psychosis and schizophrenia: are therebiological markers of illness onset? In: Jackson, H., McGorry, P.(Eds.), Recognition and Management of Early Psychosis: aPreventive Approach. Cambridge University Press, Cambridge.

Pantelis, C., Yucel, M., Wood, S., et al., 2005. Structural brain imagingevidence for multiple pathological processes at different stages ofbrain development in schizophrenia. Schizophr. Bull. 31,672–696.

Paus, T., Zijdenbos, A., Worsley, K., et al., 1999. Structural maturationof neural pathways in children and adolescents: an in-vivo study.Science 283, 1908–1911.

Pujol, J., Vendrell, P., Junque, C., et al., 1993. When does human braindevelopment end? Evidence of corpus callosum growth up toadulthood. Ann. Neurol. 34, 71–75.

Shi, L., Fatemi, S., Sidwell, R., et al., 2003. Maternal influenzainfection causes marked behavioural and pharmacological changesin the offspring. J. Neurosci. 23, 297–302.

Smith, S., 2002. Fast robust automated brain extraction. Hum. BrainMapp. 17, 143–155.

Sowell, E., Thompson, P., Holmes, C., et al., 1999. In-vivo evidencefor post-adolescent brain matural in frontal and striatal regions.Nat. Neurosci. 2, 859–861.

Sun, D., Phillips, L., Velakoulis, D., et al., 2007a. Progressive brainstructural changes mapped as psychosis develops in ‘at risk’individuals. Schizophr. Bull. 33, 356–357.

Sun, D., Velakoulis, D., Phillips, L., et al. (2007) Brain structuralchanges mapped as psychosis develops: A longitudinal MRIstudy., In submission.

Thompson, K., Phillips, L., Komesaroff, P., et al., 2007. Stress andHPA-axis functioning in young people at ultra high risk forpsychosis. J. Psychiatr. Res. 41, 561–569.

10 M. Walterfang et al. / Schizophrenia Research 103 (2008) 1–10

Thompson, P., Giedd, J., Woods, R., et al., 2000. Growth patterns inthe developing brain detected by using continuum mechanicaltensor maps. Nature 404, 190–193.

Velakoulis, D., Wood, S., Wong, M., et al., 2006. Hippocampal andamygdala volumes differ according to psychosis stage and diagnosis:an MRI study of chronic schizophrenia, first-episode psychosis andultra-high risk subjects. Arch. Gen. Psychiatry 63, 139–149.

Walterfang, M., McGuire, P., Yung, A., et al. (in press) White mattervolume changes in people who develop psychosis. Br J Psychiatry.

Walterfang, M., Wood, A., Reutens, D., et al., 2008. Morphology ofthe corpus callosum at different stages of schizophrenia: a cross-sectional study in first-episode and chronic illness. Br J Psychiatry.192, 429–434.

Walterfang, M., Wood, A., Reutens, D., et al., 2006. Cross-sectionaland longitudinal callosal morphometry in high-risk individuals,first-episode psychosis, chronic schizophrenia and controls.Schizophr. Res. 81, S29–S30.

Wood, S., Pantelis, C., Proffitt, T., et al., 2003. Spatial workingmemory ability is a marker of risk-for-psychosis. Psychol. Med.33, 1239–1247.

Woodruff, P., McManus, I., David, A., 1995. Meta-analysis of corpuscallosum size in schizophrenia. J. Neurol. Neurosurg. Psychiatry58, 457–461.

Woods, R., Grafton, S., Holmes, C., et al., 1998. Automated imageregistration I: general methods and intrasubject, intramodalityvalidation. J. Comp. Assist. Tomog. 22, 155–165.

Wyatt, R., 1991. Neuroleptics and the natural course of schizophrenia.Schizophr. Bull. 17, 325–351.

Yung, A., McGorry, P., McFarlane, C., et al., 1995. The PACE clinic:development of a clinical service for young people at high risk ofpsychosis. Aust. Psychiatry 3, 345–349.

Yung, A., Phillips, L., McGorry, P., et al., 1998. Can we predict theonset of first-episode psychosis in a high-risk group? Int. Clin.Psychopharmacol. 13, s23–s30.

Yung, A., Phillips, L., Yuen, H., et al., 2003a. Psychosis prediction:12 month follow-up of a high risk (‘prodromal’) group. Schizophr.Res. 60, 21–32.

Yung, A., Phillips, L., Yuen, H., et al., 2003b. Psychosis prediction:12-month follow up of a high-risk (“prodromal”) group. Schizophr.Res. 60, 21–32.