Available online at www.sciencedirect.com

aging 171 (2009) 82–93www.elsevier.com/locate/psychresns

Psychiatry Research: Neuroim

Reduced language lateralization in first-episode schizophrenia: AnfMRI index of functional asymmetry

Maya Bleich-Cohena,b,⁎,1, Talma Hendlera,b,1, Moshe Kotlerc, Rael D. Strousc

aFunctional Brain Imaging Unit, Wohl Institute for Advanced Imaging, Tel Aviv Sourasky Medical Center, IsraelbLevi-Edersheim-Gitter Institute for Human Brain Mapping, affiliated to the Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel

cBeer Yaakov Mental Health Center, Beer Yaakov, Sackler Faculty of Medicine, Tel Aviv University, Israel

Received 20 May 2007; received in revised form 18 December 2007; accepted 4 March 2008

Abstract

Patients with schizophrenia exhibit a decrease or loss of normal anatomical brain asymmetry that also extends to functionallevels. We applied functional magnetic resonance imaging (fMRI) to investigate language lateralization in patients withschizophrenia during their first episode of illness, thus excluding effects of chronic illness and treatment. Brain regions activatedduring language tasks of verb generation and passive music listening were explored in 12 first-episode patients with schizophreniaand 17 healthy controls. Regions of interest corresponded to Broca's area in the inferior frontal gyrus (IFG) and Wernicke's area inthe superior temporal sulcus (STS). Patients with schizophrenia had significantly smaller lateralization indices in language-relatedregions than controls. A similar effect was observed in their IFG and STS regions. There was no difference between the groups inthe auditory cortex for the music task. Patients with schizophrenia demonstrated greater activation than the controls in temporalregions: the difference was larger in patients with more severe positive symptom subscores. In conclusion, patients withschizophrenia demonstrated loss of normal functional brain asymmetry, as reflected in diminished lateralization of language-relatedactivation in frontal and temporal regions. This phenomenon was already present during their first episode of psychosis, possiblyreflecting developmental brain abnormalities of the illness.© 2008 Elsevier Ireland Ltd. All rights reserved.

Keywords: Schizophrenia; Verb generation; Inferior frontal gyrus; Superior temporal sulcus; Broca's area; Wernicke's area

1. Introduction

Brain asymmetry refers to the normal differences inregional hemispheric structure and/or function. This

⁎ Corresponding author. Functional Brain ImagingUnit,Wohl Institutefor Advanced Imaging, Tel Aviv SouraskyMedical Center, 6 WeizmannStreet, Tel Aviv 64239, Israel. Tel.: +972 3 6973094; fax: +972 36973080.

E-mail address: maya44@gmail.com (M. Bleich-Cohen).1 The authors contributed equally.

0925-4927/$ - see front matter © 2008 Elsevier Ireland Ltd. All rights resedoi:10.1016/j.pscychresns.2008.03.002

asymmetry is apparent in brain regions related to languageprocessing, which are usually located in the left hemi-sphere and associated with right-hand dominance in 90%of cases. In contrast, patients with schizophrenia are morefrequently left- or mixed-handed (Yan et al., 1985;Sommer et al., 2001a), and children of mixed-handpreference more often develop schizophrenia in later life(Crow et al., 1996). There is an indication for reducedmotor asymmetry that reflects abnormal brain lateraliza-tion. The establishment of anatomical and functionalasymmetry in the human brain has been indicated as a

rved.

83M. Bleich-Cohen et al. / Psychiatry Research: Neuroimaging 171 (2009) 82–93

component of the normal neurodevelopmental process(Hutsler and Galuske, 2003). Structurally, normallydeveloped right-handed adult individuals show widerright frontal and left occipital lobes, phenomena known asthe “developmental torque”, on computerized tomogra-phy (CT) and magnetic resonance imaging (MRI). Thisopposing regional asymmetry was noted to be inverted inschizophrenia (i.e., wider left frontal and right occipitallobes) by some studies (Luchins et al., 1981; Falkai et al.,1995), but not by others (Andreasen et al., 1982; Jerniganet al., 1982). Asymmetry favoring the left in the planumtemporale and sylvian fissure has also been shown inhealthy right-handed subjects (Barta et al., 1997). Thisstructural asymmetry has been related to languagelateralization and is often missing in schizophrenia(Sommer et al., 2001a).

In addition to decreased anatomical asymmetry, thereis also evidence for decreased functional brain asym-metry in schizophrenia. For example, several studieshave indicated a decrease in language asymmetry,particularly with reference to the fused word and con-sonant vowel tasks, as evidenced by dichotic listeningtests (Sommer et al., 2001a). Brain-imaging techniquesused to investigate functional asymmetry in schizo-phrenia brains have also recently shown decreased leftlateralization in various language tasks. For example, inone functional magnetic resonance imaging (fMRI)study, patients with schizophrenia demonstrated lessoverall left hemispheric lateralization than healthycontrols in performing language tasks. This differencewas accounted for by increased activation in their righthemisphere relative to controls (Sommer et al., 2001b,2003). In addition, Weiss et al. (2004, 2006) demon-strated reduction in language lateralization of the frontalcortex in clinically stable schizophrenia patients (2004)as well as in unmedicated schizophrenia men (2006)compared with controls due to bilateral activation inpatients in comparison to primarily left activation incontrols.

The current fMRI study was designed to furtherinvestigate at an early stage of schizophrenia this redu-ced language-related brain lateralization in terms of itsregional distribution within the hemisphere. We rea-soned that the effects of chronic illness, such as long-term treatment, institutionalization and aging, may belargely excluded by studying relatively young first-episode patients with schizophrenia. For language-rela-ted activation, we alternated the auditory verb genera-tion task with music and silent periods. The paradigm ofauditory verb generation has been used extensively infMRI studies of healthy and neurosurgical populations.It has been shown to be a reliable test for language

lateralization comparable to the gold-standard Wada test(Desmond et al., 1995, Kloppel and Buchel, 2005). Weused a region of interest (ROI) approach to point to thelocalization of the abnormality within the hemispheres,and chose the classical language-related regions,including the inferior frontal gyrus (IFG, representingBroca's area) and the superior temporal sulcus (STS,representing Wernicke's area). Since the language testwas auditory, the primary auditory cortex was examinedfor non-language-related activation.

2. Materials and methods

2.1. Study population

Twelve right-handed patients with schizophrenia(study group) (6M, 6F) and seventeen right-handedapparently healthy volunteers (control group) (10M, 7F)participated in this study. Right-handedness was verifiedusing a dominance questionnaire (consisting of 10questions relating to dominance functioning such aswriting, drawing, and eating). All participants were nativeHebrew speakers. The patients were all hospitalized at alarge state psychiatric referral institution for their first-episode of schizophrenia psychosis. Eligibility for studyparticipation was limited to right-handed patients betweenthe ages of 18 and 45 years. Two board-certifiedpsychiatrists verified the patients' diagnoses, whichwere determined according to the guidelines of theStructured Clinical Interview for DSM-IVAxis I, PatientEdition (First et al., 1994). Patients with any significantmedical or neurological illness or who were pregnant orengaged in any substance abuse were excluded from thestudy. Medical and neurological illnesses were ruled outby physical and neurological examinations, routinelaboratory investigation, reports of the patients' treatingphysicians and medical records. Prior to the imaging, thepatients with schizophrenia were rated by the clinicalassessment scales of the Positive and Negative SyndromeScale (PANSS) (Kay et al., 1987) and the Clinical GlobalImpression-Severity Scale (CGI-S) (Guy 1976) (Table 1).

All study entrants provided written informed consentafter receiving a full explanation of the nature of thestudy and potential risks and benefits of study participa-tion. The studywas approved by the Beer YaakovMentalHealth Center and the Tel Aviv SouraskyMedical CenterInstitutional Review Boards.

2.2. fMRI paradigm

While being scanned, the subjects participated in anauditory language task or listened to periods of music,

Table 1Clinical data of first-episode schizophrenia patient subpopulation

No. Age, years Gender Handedness PANSSpositive

PANSSnegative

PANSSgeneral

PANSStotal

CGI-S

Length ofhospitalization

Medications

1 19 Male Right 17 12 23 52 4 2 weeks Risperidone2 21 Male Right 30 11 32 63 4 l week Risperidone3 34 Female Right 25 15 36 76 5 3 weeks Tispetidone4 19 Male Right 17 17 26 60 4 3 weeks Risperidone5 28 Male Right 32 18 36 84 5 4 weeks Quetiapine6 27 Female Right 24 19 31 74 5 2 weeks Risperidone7 36 Female Right 22 10 33 65 4 1 week Peiphenazine8 21 Male Right 28 17 39 84 5 2 weeks Risperidone9 25 Female Right 25 13 39 77 5 3 weeks Risperidone10 28 Male Right 15 23 32 70 4 2 weeks Olanzapine11 28 Female Right 22 15 36 73 4 2 weeks Risperidone12 26 Female Right 17 17 36 70 3 2 weeks Olanzapine

CGI-S, Clinical Global Impression Severity Scale; PANSS, Positive and Negative Syndrome Scale.

84 M. Bleich-Cohen et al. / Psychiatry Research: Neuroimaging 171 (2009) 82–93

interspersed with periods of no stimuli. All subjectsprior to entering the fMRI phase of the investigationunderwent a preparatory session in which adequatecompliance was documented and assured. Patients forwhatever reason demonstrating lack of complete under-standing and compliance were not included in the study.The language task was composed of 18 spoken Hebrewwords presented through headphones during three 18-speriods. Each block consisted of six different words, at arate of 1/3 Hz. There were 24 s of no stimuli at thebeginning and at the end of the paradigm: although MRIscanning is inherently accompanied by noise, the noiseis constant and monotonous, whereupon “no stimuli”can be considered as “rest” (Fig. 1A). Each verbal periodwas alternated with three 18-s periods of classical musicand five 9-s periods of rest. Two more rest periods of30 s appeared in the beginning and ending of theexperiment. Only one version of the experiment wasused (the order of the music and verb generation taskswas not randomized between the participants).

The verbal and musical periods were recordedseparately and organized in a block paradigm to bepresented by the GoldWave program (5.1.2600.2180;Microsoft Windows). The words were three- to five-letter nouns that described commonly used objects, suchas a brush and a table. The subjects were instructed tothink of but not utter a verb that best described what theycould do with the object that was named. For example,when “cup” was heard, they could think of “drinking”.Each verbal period was terminated with an audible ring.The music condition consisted of an instrumental pieceof classical music composed by Mozart. During theperiods of music, the participants were instructed tolisten passively to the music. They were told to donothing during the rest condition. To avoid contamina-

tion by patient movement, the subject was requested toremain still during scanning. (All the subjects had per-formed with a N90% accuracy score in a prior practicesession in which they were presented with auditorynouns and were instructed to say a verb that best de-scribed what they could do with the object.)

2.3. Behavioral testing

Seven first-episode patients with schizophrenia (4males) and 10 healthy subjects (7 males) (who did notperform the fMRI experiment) were part of the behavioralexperiment outside the scanner. Eighteen spoken Hebrewnouns were presented through headphones. The subjectswere instructed to articulate a verb that best describedwhat they could do with the object that was named.Responses were recorded by Presentation software(Neurobehavioral Systems, Inc., 2003). Reaction timeand accuracy were analyzed. Two-tailed, unpairedStudent's t-tests were used in the satistical analysis.

2.4. Brain scanning

Imaging was performed on a GE 1.5 T Signa HorizonLX 8.25 echo speed scanner (Milwaukee, WI, USA)with a resonant gradient echoplanar imaging system. Allimages were acquired using a standard quadrature headcoil. The scanning session included anatomical andfunctional imaging. The anatomical imaging consistedof 17 contiguous axial T1-weighted slices of 4-mmthickness, with 1-mm gaps that were prescribed basedon a sagittal localizer and covered the whole brain. Inaddition, a three-dimensional (3D) spoiled gradient echo(SPGR) sequence with high resolution (a slice thicknessof 2 mm) was acquired for each subject, in order to

Fig. 1. Experimental paradigm and functional localizer. (A) The experimental paradigm consisted of 3 blocks of verb generation (blue) in which thesubject hears spoken nouns and is instructed to think of but not utter related verbs (VG), 3 blocks of music (yellow) in which the subject passivelylistens to classical music (M), and 7 periods of rest in which the subject is instructed to lie quietly (R). The horizontal line indicates the experiment'sduration in minutes. (B) Three-dimensional parametric activation map obtained from the group of healthy subjects (n=10), shown on sagittal andaxial views (left and right, respectively). The color-coded map presents activation in language-related regions for VG more than Music in Blue-Greenand activation in auditory-related regions for Music more than VG in Yellow-Red. The colors indicate significant activation (Pb0.05) for 1 task overthe other. Arrows point to a priori regions of interest (ROIs) in language-related and auditory areas.

85M. Bleich-Cohen et al. / Psychiatry Research: Neuroimaging 171 (2009) 82–93

allow volumetric statistical analyses of the functionalsignal change. The functional imaging included T2⁎-weighted images that were acquired at the samelocations as the spin-echo T1-weighted anatomicalimages, at runs of 1190 images each (70 images perslice). BOLD contrast was acquired with a gradient echoechoplanar imaging (EPI) sequence (TR/TE/Flipangle=3000/55/90°; with FOV 24×24 cm2; matrixsize 80×80).

2.5. Imaging data analysis

The fMRI data were processed using BrainVoyager4.4 software package (http://www.brainvoyager.com).Preprocessing of functional scans included head move-ment assessment (scans with head movement N1.5 mmwere rejected), high-frequency temporal filtering, andremoval of low-frequency linear trends. To allow for T2⁎

equilibration effects, the first six images of eachfunctional scan were rejected. Pre-processed functionalimages were incorporated into the 3D datasets throughtrilinear interpolation. The complete dataset was trans-formed into Talairach space. Three-dimensional statis-tical parametric maps were calculated separately for eachsubject using a general linear model (GLM) in which allstimuli conditions were positive predictors, with a lag of3–6 s (individual account for the hemodynamic responsedelay). Note that this GLMmodel does not make a prioriassumptions regarding the behavior of the fMRI signal inthe various conditions (i.e., language versus music).

2.6. Regions of interest (ROIs)

Specific effects were studied in pre-determinedregions that are part of the language network andauditory cortex. The regions were defined for each

86 M. Bleich-Cohen et al. / Psychiatry Research: Neuroimaging 171 (2009) 82–93

individual based on commonly used anatomical land-marks that corresponded with activated clusters infunctional imaging of verb generation (VG) versusmusic (Fig. 1B). We focused on two language-relatedareas: one was confined to the inferior frontal gyrus andBrodmann's areas 44 and 45 (IFG). This region repre-sented the frontal system of language (i.e., Broca's area).The second area was confined to the superior temporalsulcus and gyrus and Brodmann's area 41 (STS). Thisregion represented the posterior language area (i.e.,Wernicke's area) (for average Talairach coordinates, seeTable 2).

Language-related activated clusters were obtainedindividually within the described anatomical landmarks(Pb0.05, uncorrected) using the GLM with the languagecondition as a positive predictor and the period of rest as anegative predictor. Music-related activated clusters werecollected using GLM contrasts of the music condition as apositive predictor and the rest condition as a negativepredictor. The number of activated voxels in each of thedefined ROIs on the left and right hemispheres wascounted separately for each subject. These counts wereused to compute a language-related lateralization index(LI) for each region (i.e., LI=L−R/L+R, with L=num-ber of voxels on the left and R=number of voxels on theright). The more positive the number, the more leftlateralized would be the activation, while a negativenumber negative yielded lateralization to the right.

2.7. Statistics

Analysis of variance (ANOVA) was performed toexplore the group differences in lateralization using boththe averaged LIs and number of active voxels in thelanguage-related brain regions using Statistica (version5.0). In addition, time courses were extracted for eachcondition (see details below) from bilateral activatedclusters of at least 100 voxels and applied for furtheranalysis in Excel. The individual's averaged signal wascalculated from all epochs of the same condition andtransformed into percent signal changes relative to the

Table 2Talairach coordinates of activation clusters in regions of interest(ROIs)

ROI Left Right

X Y Z X Y Z

IFG (BA 44,45) −51 16 6 52 15 6STC −42 −34 12 49 −33 12Auditory cortex −45 −15 7 52 −12 7

IFG=inferior frontal gyrus (Broca's area); STS=superior temporalsulcus (Wernicke's area).

averaged baseline signal (i.e., the periods of rest) foreach ROI. Lastly, this averaged signal change was cor-related with the subject's PANSS scores by Pearsoncorrelation in Statistica (version 5.0).

2.8. Voxel-based whole brain analysis

Using random effect models, we compared brain res-ponses of healthy controls and schizophrenia patients totest for any significant differences that were outside our apriori ROIs. We compared responses to the languagecondition (+) with responses to the music condition (−)and responses to themusic condition (+) with responses tothe language condition (−) in the controls and patients at athreshold of Pb.01.

3. Results

3.1. Study sample

Table 1 presents the demographic and clinical data ofthe 12 schizophrenia patients (aged 19–36 years; 6M,6F). The control group consisted of 17 apparently heal-thy age- and gender-matched (aged 22–46 years; 10M,7F) individuals. Clinical ratings of the patients indicatedthe following mean scores: total PANSS=71.9, positivePANSS subscale=23.3, negative PANSS subscale=17.5, general PANSS subscale=30.9, mean hallucina-tions PANSS=2.66, mean delusions PANSS=4.25 andmean CGI-S score=4.45. All patients were receivingmedication at the time of imaging for a mean period of15.75 days (S.D.=12.8). While the time on medicationranged from 7 to 60 days, 11 of the patients were inthe range of 7–27 days and one patient had receivedmedication for 60 days. Medications received by thesubjects include risperidone (N=9, dose range 2–5 mg,mean = 3.2 mg, S.D. = 0.9), quetiapine (N = 1,dose=700 mg), perphenazine (N=1, dose=16 mg) andhaloperidol (N=1, dose=2 mg).

3.2. Behavioral results

Subjects from both groups performed the task withno mistakes (100% accuracy). No difference was foundin reaction time between the two groups (P=0.36).

3.3. Whole brain analyses

We next applied a whole brain analysis in order toexplore task-specific activation using a group randomeffect approach. Tables 3A and 4A show the regionsobtained when contrasting the language condition (+)

Table 3Whole brain analysis in healthy controls (random effect Pb0.01)

ROl Left Right Left peak P value Right peak P value

A. LanguageNmusicIFG (BA 44, 45) −46,13,3 44,19,0 0.000346 0.003275STS (BA 41) −52,−46,10 48,−44,10 0.0 05902 0.001282Insula (BA 13) −32,20,7 32,20,7 0.000950 0.002446Premotor −40,−2,36 0.001956DLPFC −41,23,28 0.000213Putamen −16,4,8 0.002930

B. MusicN languageAuditory cortex −42,−17,9 44,−12,9 0.004179 0.004479Medial frontal gyrus 3,51,17 0.000238

IFG, inferior frontal gyrus; STS, superior temporal sulcus; MTS, middle temporal gyrus; DLPFC, dorsolateral prefrontal cortex.

87M. Bleich-Cohen et al. / Psychiatry Research: Neuroimaging 171 (2009) 82–93

with the music condition (−) in the controls and patients,respectively. These contrasts evoked activation in our apriori language-related ROIs in the IFG and STS in bothgroups. This contrast also evoked activation in the in-sula, premotor, DLPFC and putamen of the controls(Table 3A), and in the MTS, premotor, DLPFC andputamen of the patients (Table 4A). Tables 3B and 4Bpresent the extent of activation that was preferentiallyevoked by contrasting the music condition (+) with thelanguage condition (−) for the two groups: this contrastevoked activation mainly in the auditory cortex for both.It also evoked activation in the medial frontal gyrus ofthe controls (Table 3B) and in the posterior cingulateand medial frontal gyrus of the patients (Table 3B).

3.4. Hemispheric activation

Fig. 2A demonstrates the lateralization of groupactivation obtained for the language task versus rest forthe healthy controls (Fig. 2A1) and for the patients with

Table 4Whole brain analysis in schizophrenia patients (random effect Pb0.01)

ROI Left Right

A. LanguageNmusicIFG (BA 44, 45) −44,14,0 44,19,0STS (BA 41) −54,−44,10 50,−41MTS −51,−36,−2Premotor −43,−4,35DLPFC −41,23,26Putamen −18,8,4

B. MusicN languagePosterior cingulate (BA 23) 4,−51,25Auditory cortex −42,−12,9 44,−12Medial frontal gyrus 3,53,15

IFG=inferior frontal gyrus; STS=superior temporal sulcus; MTS=middle te

schizophrenia (Fig. 2A2). The lateralization of thehealthy controls is clearly greater than that of the pa-tients with schizophrenia for the language task. Fig. 3Aand B shows the difference in lateralization between thegroups across language ROIs (i.e., Broca's and Wer-nicke's areas). This difference is demonstrated by largerand more positive LIs in the controls than in the patients(F(1,54)=22.54; Pb0.00001, Fig. 3A). In order tofurther quantify this group difference, the number ofactivated voxels for language versus rest that wasobtained for each hemisphere was used as a dependentvariable in a two-wayANOVA (group and hemisphere asfactors), and a significant interaction was found showinggreater asymmetry (i.e., left hemisphere dominance) forthe controls than for the patients (F(1,54)=21.41;Pb0.00005). Post-hoc analyses revealed that this maingroup difference was due to less right hemisphereactivation for the controls than for the patients (TukeyHSD post-hoc, Pb0.001) (Fig. 3B). This groupdifference of lateralization was significant for the verbal

Left peak P value Right peak P value

0.000021 0.033275,10 0.000007 0.002688

0.0001340.0039430.0054500.000096

0.000225,9 0.906669 0.001112

0.000238

mporal gyrus; DLPFC=dorsolateral prefrontal cortex.

Fig. 2. Language- and music-related activation maps. Axial views of parametric activation maps obtained during language tasks for the healthy (A1)and the schizophrenia (A2) groups and during music task for the healthy (B1) and the schizophrenia (B2) groups. Colored regions indicate greateractivation during VG than during periods of rest (Pb0.005, uncorrected with random effect).

88 M. Bleich-Cohen et al. / Psychiatry Research: Neuroimaging 171 (2009) 82–93

task but not for the music task in the auditory cortex (Fig.3C). There was, however, an overall greater activationfor music in the auditory cortex of the patients than in thecontrols (Pb0.005) (Fig. 5B), although no significantlateralization effects were found in either group.

A similar analysis for verbal tasks was applied for eachlanguage-related ROI (Fig. 4) and a similar pattern wasobtained for activation per group and hemisphere both inBroca's area (i.e., IFG) and Wernicke's area (i.e., STS).There was a significant interaction in Broca's area(F(1,26)=20.77; Pb0.0001) in which controls, but notpatients, demonstrated a significant left dominance for theSTS (Tukey HSD post-hoc, Pb0.0001) (Fig. 4A). Therewas a significant interaction in Wernicke's area (F(1,26)=5.83; Pb0.05) in which controls, but not thepatients, demonstrated a significant left dominance forthe STS (Tukey HSD post-hoc, Pb0.001) (Fig. 4B).Lastly, the group effect was tested per brain regionacross hemispheres for each task (Fig. 5). For theperiods of verb generation versus rest, the IFG wassimilarly activated in both groups, but a larger

activation was obtained among the patients withschizophrenia than among the controls for the STS(Tukey HSD post-hoc, Pb0.0005; Fig. 5A). Similarly,a larger activation in the auditory cortex in response topassive listening of music was seen for the schizo-phrenia group than for the controls (F(1,52)=8.084;Pb0.005; Fig. 5B).

3.5. Correlations between brain activation and symptomseverity

Analysis of the PANSS subtotal and total symptomscales revealed a positive correlation between the positivesymptom subscale of the PANSS and the averaged per-cent signal change during verb generation in the left(r=0.79; Pb0.05) and right (r=0.85; Pb0.05) STS(Fig. 6A and B, respectively).

Analysis of the PANSS hallucination and delusionscales revealed no correlation with the language later-alization during verb generation. The correlation betweenthe hallucination subscale of the PANSS and the left STS

Fig. 3. Task-related activation per group and hemisphere. (A) Lateralization index (LI) for language task across language-related regions. (B) Averagednumber of activated voxels for VG versus rest across language-related regions. (C) Averaged number of activated voxels for music versus rest in theauditory cortex. Error bars represent standard error of the mean (SEM). Stars indicate significant differences in activation between groups.

89M. Bleich-Cohen et al. / Psychiatry Research: Neuroimaging 171 (2009) 82–93

is −0.45, and the corresponding correlation for the rightSTS is −0.18. Correlations between the delusion subscaleof the PANSS and the left STS and the right STS are 0.16and 0.36, respectively.

4. Discussion

The results of the current fMRI study indicated anoverall reduced functional brain asymmetry in thelanguage-related network that includes Broca's andWernicke's areas (i.e. IFG and STS, respectively) infirst-episode patients with schizophrenia compared withhealthy matched controls. Interestingly, this decrease inbrain asymmetry was due to increased activity in theright hemisphere compared with the findings in healthysubjects rather than to relatively decreased activity in thepatients' own left hemispheres (Fig. 3A and B). In

addition, the patients demonstrated significantly moreactivation in the temporal lobe regions (i.e., STS andauditory cortex) for both linguistic and musical stimulithan the controls (Fig. 5).

4.1. Language-related brain lateralization

Our finding of reduced language-related brainlateralization in first-episode patients with schizophre-nia is in agreement with behavioral studies using audi-tory dichotic tests (Wexler et al., 1991; Bruder et al.,1995) as well as with functional imaging findings inyoung patients with schizophrenia (Luchins et al., 1981;Falkai et al., 1995; Sommer et al., 2001b, 2003). Thespecificity of our findings to language function is evi-denced by the absence of a similar abnormality formusic (Figs. 2B and 3C). Since the establishment of

Fig. 4. Regional language activation (VG versus rest) per hemisphere and group. Averaged number of activated voxels in (A) Broca's area and(B) Wernicke's area. Error bars represent standard error of the mean (SEM). Stars indicate statistical significance. IFG=inferior frontal gyrus;STS=superior temporal sulcus.

90 M. Bleich-Cohen et al. / Psychiatry Research: Neuroimaging 171 (2009) 82–93

brain asymmetry for language is considered to be part ofnormal development, the finding of early stage abnor-malities suggests disordered brain development in schi-zophrenia (Weinberger 1987; Crow et al., 1989; Falkaiand Bogerts, 1992). In support of this view are findingsof reduced right ear advantage for words in a dichoticlistening test in first-degree relatives of patients withschizophrenia (Grosh et al., 1995). Whether asymmetrymeasurements may be applied for early detection of riskfor schizophrenia is still undetermined.

Fig. 5. Task-related regional activation per group across hemispheres. Averageregions and for (B) music versus rest in the auditory cortex. Error bars rsignificance.

Our finding that reduced lateralization in first-episodepatients with schizophrenia is due to a relative increase ofactivity in the right hemisphere is also consistent with thefindings of others in young patients with schizophrenia(Sommer et al., 2001b, 2003). We demonstrate for whatwe believe to be the first time that this “right hemisphereadvantage” can be found during language testing in theright homologues of both Broca's and Wernicke's areas(Fig. 4). It has been suggested that the “right shift factor”,a dominant allele, is responsible for normal cerebral

d number of activated voxels for (A) VG versus rest in language-relatedepresent standard error of the mean (SEM). Stars indicate statistical

91M. Bleich-Cohen et al. / Psychiatry Research: Neuroimaging 171 (2009) 82–93

asymmetry in healthy subjects by disrupting the growth oflanguage-related regions in the right hemisphere (Annett1992). Accordingly, it has been proposed that patientswith schizophrenia might suffer from an abnormality inthis “right shift factor” that is associated with theappearance of psychosis (Crow et al., 1996). Increasedhomologue right-sided activation in the patient groupduring the language task might also reflect known ab-normalities in prosodic processing in schizophrenia(Leitman et al., 2007). Our study demonstrates that arelatively simple language fMRI test can probe anabnormal “right hemisphere advantage” already evidentin young patients with schizophrenia at the time of theirfirst-ever episode, thus excluding long-term effects ofmedication and aging. Future studies should combineimaging and genetic markers to further explore whethersuch a “hemispheric advantage” can be considered as an

Fig. 6. Correlation of language activation with symptom severity in the schizscale of the PANSS and the averaged % signal change in the left (A) and rig

early biological marker of functional resource depletion inschizophrenia.

4.2. Temporal lobe activation

Our study demonstrates that the patients had moreoverall activation in the temporal lobe (STS and auditorycortex) for both language (Fig. 5A) and music (Figs. 2Band 5B) tasks compared with healthy controls. Otherstudies have provided evidence of left planum temporalegraymatter volume reduction and bilateral Heschl's gyrusgraymatter volume reduction in first-episode patientswithschizophrenia compared with healthy controls (Hirayasuet al., 2000). Similarly, Sumich et al. (2002) noted thatright-handedmale patients experiencing their first episodeof psychotic illness had smaller left planum temporalevolumes than healthy subjects. Kasai et al. (2003) reported

ophrenia group. Significant correlation between scores on the positiveht (B) superior temporal sulcus (STS).

92 M. Bleich-Cohen et al. / Psychiatry Research: Neuroimaging 171 (2009) 82–93

that patients with first-episode schizophrenia showedsignificant decreases in gray matter volume over time inthe left superior temporal gyrus compared with patientswith first-episode affective psychosis or healthy subjects.Others have shown volume reduction only in the rightsuperior temporal gyrus in patients with early-onsetschizophrenia in relation to matched controls (Matsumotoet al., 2001). The results from the current study suggestthat these anatomical abnormalities observed onMRImaybe reflected at the functional level as bilateral excessiveactivation extending beyond the STS in the temporal lobe(i.e., the auditory cortex) (Fig. 5). In support of the role ofthe temporal lobe in schizophrenia phenomenology is ourfinding of a significant correlation between signal changesduring language tasks in the STS and the severity ofpositive psychotic symptoms on the PANSS (Fig. 6A andB). This effect, already detectable at the first episodeof schizophrenia, also corresponds with the findings ofothers who have suggested that the temporal lobes areprimarily involved in the appearance of positive symp-toms (Shenton et al., 1992; McGuire et al., 1998; Crespo-Facorro et al., 2004). A possible explanation for thefinding of extended activation in the temporal lobes couldreflect increased anxiety in the patients. This may alsoexplain the correlations with symptom severity. It mayalso be suggested that a further explanation for theincreased temporal activation could be non-selectiverecruitment of temporal brain structures due to impairedspecialization (Dolfus et al., 2005).

Previous imaging reports have found that reducedregional hemispheric volume asymmetry is associatedwith more severe negative symptoms in chronic schizo-phrenia (Bilder et al., 1994). We cannot yet concludewhether our findings of functional asymmetry of languagemay also be associated with negative symptoms, but it ispossible that they may be predictive of a more chronicform of the disorder (i.e., to emerge in the future). A studyof larger patient samples with more variable clinicalpresentations will also allow further exploration of thispossibility. It should be noted that within the rubric of verbgeneration there are several linguistic functions (includinglexical search processes, subvocal representations andprocesses of working memory). While each of these isimportant and would be interesting to examine indi-vidually, the scope of this study did not allow for thebreakdown of “verb generation” into its component parts,and further research should explore these components. Afurther limitation of the study was that despite patientsbeing in their first episode of psychosis, many had alreadyreceived some antipsychotic medication, albeit for a veryshort period of time. While results may have beeninfluenced by an immediate effect of antipsychotic

medication, the short-term use of antipsychotic medica-tion was permitted in this study in these first-episodepsychosis patients for ethical reasons (no delays in clinicalmanagement). Furthermore, precedent exists in manystudies of first-episode psychosis patients to includepatients in investigation despite recent short-term use ofantipsychotic medication in order to maintain patient easeand protocol compliance.

In summary, our fMRI study demonstrates reducedlanguage-related brain asymmetry in Broca's (i.e., IFG)and Wernicke's (i.e., STS) areas already present at thetime of the first episode of schizophrenia psychosis.These observations appeared to be as a result of regionalof increased activity in the right hemisphere rather thandecreased activity in the left hemisphere. In addition,overall activation of the temporal lobe was greater forboth language and music stimuli in patients withschizophrenia than in healthy controls, but only thelanguage-related activation in the temporal lobe regionsdirectly correlated with the severity of positive symp-toms in the patient group. Confirmation of thesefindings with a larger sample size and with broader,more comprehensive neurocognitive reflections oflanguage-related functional asymmetry would berecommended.

Acknowledgments

This study was funded by the National Institute forPsychobiology in Israel (RDS and TH), the Levi-Edersheim-Gitter Institute for Human Brain Mapping(MB) and the Israel Science Foundation (TH). We thankD. Ben-Bashat, I. Sivan and Y. Assaf for the physicsinsights, P. Pianka, T. Gorfein, Y. Yeshurun, A.Mendelsohn and E. Sarig for their constructivecomments, and finally all the subjects who volunteeredto participate in the study.

References

Andreasen, N.C., Dennert, J.W., Olsen, S.A., Damasio, A.R., 1982.Hemispheric asymmetries and schizophrenia. American Journal ofPsychiatry 139, 427–430.

Annett, M., 1992. Parallels between asymmetries of planum temporaleand of hand skill. Neuropsychologia 30, 951–962.

Barta, P.E., Pearlson, G.D., Brill 2nd, L.B., Royall, R., McGilchrist, I.K.,Pulver, A.E., Powers, R.E., Casanova, M.F., Tien, A.Y., Frangou, S.,Petty, R.G., 1997. Planum temporale asymmetry reversal inschizophrenia: replication and relationship to gray matter abnorm-alities. American Journal of Psychiatry 154, 661–667.

Bilder, R.M.,Wu, H., Bogerts, B., Degreef, G., Ashtari,M., Alvir, J.M.,Snyder, P.J., Lieberman, J.A., 1994. Absence of regional hemi-spheric volume asymmetries in first-episode schizophrenia. Amer-ican Journal of Psychiatry 151, 1437–1447.

93M. Bleich-Cohen et al. / Psychiatry Research: Neuroimaging 171 (2009) 82–93

Bruder, G., Rabinowicz, E., Towey, J., Brown, A., Kaufmann, C.A.,Amador, X., Malaspina, D., Gorman, J.M., 1995. Smaller right ear(left hemisphere) advantage for dichotic fused words in patientswith schizophrenia. American Journal of Psychiatry 152, 932–935.

Crespo-Facorro, B., Kim, J.J., Chemerinski, E., Magnotta, V.,Andreasen, N.C., Nopoulos, P., 2004. Morphometry of thesuperior temporal plane in schizophrenia: relationship to clinicalcorrelates. Journal of Neuropsychiatry and Clinical Neuroscience16, 284–294.

Crow, T.J., Colter, N., Frith, C.D., Johnstone, E.C., Owens, D.G.,1989. Developmental arrest of cerebral asymmetries in early onsetschizophrenia. Psychiatry Research 29, 247–253.

Crow, T.J., Done, D.J., Sacker, A., 1996. Cerebral lateralization isdelayed in children who later develop schizophrenia. Schizo-phrenia Research 22, 181–185.

Desmond, J.E., Sum, J.M., Wagner, A.D., Demb, J.B., Shear, P.K.,Glover, G.H., Gabrieli, J.D., Morrell, M.J., 1995. Functional MRImeasurement of language lateralization in Wada-tested patients.Brain 118, 1411–1419.

Dollfus, S., Razafimandimby, A., Delamillieure, P., Brazo, P., Joliot,M., Mazoyer, B., Tzourio-Mazoyer, N., 2005. Atypical hemi-spheric specialization for language in right-handed schizophreniapatients. Biological Psychiatry 57, 1020–1028.

First, M.B., Spitzer, R.L., Gibbon, M., Williams, J.B.W., 1994.Structured Clinical Interview for DSM-IVAxis I Disorders, PatientEdition (SCID-P), Version 2. Biometrics Research, New YorkState Psychiatric Institute, New York.

Falkai, P., Bogerts, B., 1992. Neurodevelopmental abnormalities inschizophrenia. Clin Neuropharmacological Supplementary 1 (Pt A),498A–499A.

Falkai, P., Bogerts, B., Schneider, T., Greve, B., Pfeiffer, U., Pilz, K.,Gonsiorzcyk, C., Majtenyi, C., Ovary, I., 1995. Disturbed planumtemporale asymmetry in schizophrenia. A quantitative post-mortem study. Schizophrenia Research 14, 161–176.

Grosh, E.S., Docherty, N.M., Wexler, B.E., 1995. Abnormal lateralityin schizophrenias and their parents. Schizophrenia Research 14,155–160.

Guy, W., 1976. ECDEU Assessment Manual for Psychopharmacology.U.S. Department of Health, Education, and Welfare Publication(ADM) 76-338. National Institute of Mental Health, Rockville MD,pp. 218–222.

Hirayasu, Y., McCarley, R.W., Salisbury, D.F., Tanaka, S., Kwon, J.S.,Frumin, M., Snyderman, D., Yurgelun-Todd, D., Kikinis, R.,Jolesz, F.A., Shenton, M.E., 2000. Planum temporale and Heschlgyrus volume reduction in schizophrenia: a magnetic resonanceimaging study of first-episode patients. Archives of GeneralPsychiatry 57, 692–699.

Hutsler, J., Galuske, R.A., 2003. Hemispheric asymmetries in cerebralcortical networks. Trends in Neuroscience 26, 429–435.

Jernigan, T.L., Zatz, L.M., Moses Jr., J.A., Cardellino, J.P., 1982.Computed tomography in schizophrenias and normal volunteers. II.Cranial asymmetry. Archives of General Psychiatry 39, 771–773.

Kasai, K., Shenton, M.E., Salisbury, D.F., Hirayasu, Y., Lee, C.U.,Ciszewski, A.A., Yurgelun-Todd, D., Kikinis, R., Jolesz, F.A.,McCarley, R.W., 2003. Progressive decrease of left superiortemporal gyrus gray matter volume in patients with first-episodeschizophrenia. American Journal of Psychiatry 160, 156–164.

Kay, S., Fiszbein, A., Opler, L.A., 1987. The positive and negativesyndrome scale for schizophrenia. Schizophrenia Bulletin 13,261–276.

Kloppel, S., Buchel, C., 2005. Alternatives to the Wada test: a criticalview of functional magnetic resonance imaging in preoperativeuse. Current Opinion in Neurology 18, 418–423.

Leitman, D.I., Hoptman, M.J., Foxe, J.J., Saccente, E., Wylie, G.R.,Nierenberg, J., Jalbrzikowski, M., Lim, K.O., Javitt, D.C., 2007. Theneural substrates of impaired prosodic detection in schizophreniaand its sensorial antecedents. American Journal of Psychiatry 164,474–482.

Luchins, D.J., Morihisa, J.M., Weinberger, D.R., Wyatt, R.J., 1981.Cerebral asymmetry and cerebellar atrophy in schizophrenia:a controlled postmortem study. American Journal of Psychiatry138, 1501–1503.

Matsumoto, H., Simmons, A., Williams, S., Hadjulis, M., Pipe, R.,Murray, R., Frangou, S., 2001. Superior temporal gyrus abnorm-alities in early-onset schizophrenia: similarities and differenceswith adult-onset schizophrenia. American Journal of Psychiatry158, 1299–1304.

McGuire, P.K., Quested, D.J., Spence, S.A., Murray, R.M., Frith, C.D.,Liddle, P.F., 1998. Pathophysiology of ‘positive’ thought disorderin schizophrenia. British Journal of Psychiatry 173, 231–235.

Shenton, M.E., Kikinis, R., Jolesz, F.A., Pollak, S.D., LeMay, M.,Wible, C.G., Hokama, H., Martin, J., Metcalf, D., Coleman, M.,1992. Abnormalities of the left temporal lobe and thought disorderin schizophrenia. A quantitative magnetic resonance imagingstudy. New England Journal of Medicine 27, 604–612.

Sommer, I., Ramsey, N., Kahn, R., Aleman, A., Bouma, A., 2001a.Handedness, language lateralisation and anatomical asymmetry inschizophrenia: meta-analysis. British Journal of Psychiatry 178,344–351.

Sommer, I.E., Ramsey, N.F., Kahn, R.S., 2001b. Language lateraliza-tion in schizophrenia, an fMRI study. Schizophrenia Research 52,57–67.

Sommer, I.E., Ramsey, N.F., Mandl, R.C., Kahn, R.S., 2003.Language lateralization in female patients with schizophrenia: anfMRI study. Schizophrenia Research 60, 183–190.

Sumich, A., Chitnis, X.A., Fannon, D.G., O'Ceallaigh, S., Doku, V.C.,Falrowicz, A., Marshall, N., Matthew, V.M., Potter, M., Sharma,T., 2002. Temporal lobe abnormalities in first-episode psychosis.American Journal of Psychiatry 159, 1232–1235.

Weinberger, D.R., 1987. Implications of normal brain development forthe pathogenesis of schizophrenia. Archives of General Psychiatry44, 660–669.

Weiss, E.M., Hofer, A., Golaszewski, S., Siedentopf, C., Brinkhoff, C.,Kremser, C., Felber, S., Fleischhacker, W.W., 2004. Brainactivation patterns during a verbal fluency test—a functionalMRI study in healthy volunteers and patients with schizophrenia.Schizophrenia Research 70, 287–291.

Weiss, E.M., Hofer, A., Golaszewski, S., Siedentopf, C., Felber, S.,Fleischhacker, W.W., 2006. Language lateralization in unmedi-cated patients during an acute episode of schizophrenia: afunctional MRI study. Psychiatry Research: Neuroimaging 146,185–190.

Wexler, B.E., Giller, E.L., Southwick, S., 1991. Cerebral laterality,symptoms, and diagnosis in psychotic patients. BiologicalPsychiatry 29, 103–116.

Yan, S.M., Flor-Henry, P., Chen, D.Y., Li, T.G., Qi, S.G., Ma, Z.X.,1985. Imbalance of hemispheric functions in the major psychosis:a study of handedness in the People's Republic of China.Biological Psychiatry 20, 906–917.