Schizophrenia Research 125 (2011) 1–12

Contents lists available at ScienceDirect

Schizophrenia Research

j ourna l homepage: www.e lsev ie r.com/ locate /schres

Are cavum septum pellucidum abnormalities more common inschizophrenia spectrum disorders? A systematic review and meta-analysis

Clarissa Trzesniak a,b, Irismar R. Oliveira c, Matthew J. Kempton b, Amanda Galvão-de Almeida c,d,Marcos H.N. Chagas a, Maria Cecília F. Ferrari a, Alaor S. Filho a, Antonio W. Zuardi a,Daniel A. Prado a, Geraldo F. Busatto e, Phillip K. McGuire b,Jaime E.C. Hallak a, José Alexandre S. Crippa a,⁎a Department of Neuroscience and Behavior, Medical School, University of São Paulo (USP), Ribeirão Preto (SP), Brazil and INCT Translational Medicine, Brazilb Department of Psychosis Studies, Institute of Psychiatry, King's College London, UKc Affective Disorders Center, Universidade Federal da Bahia (UFBA), Salvador (BA), Brazild Post Graduation Program, Universidade Federal de São Paulo (UNIFESP), and Laboratório Interdisciplinar de Neurociências Clínicas (LiNC), São Paulo (SP), Brazile Department of Psychiatry, Medical School, University of São Paulo (USP), São Paulo (SP), Brazil

a r t i c l e i n f o

⁎ Corresponding author. Departamento de NeurocComportamento — FMRP-USP, Avenida BandeirantRibeirão Preto, SP, Brazil. Tel.: +55 16 3602 2201; fax

E-mail addresses: clarissaf@hotmail.com (C. Trzesnirismar.oliveira@uol.com.br (I.R. Oliveira), matthew.ke(M.J. Kempton), amandacgalvao@yahoo.com.br (A. Gasetroh@hotmail.com (M.H.N. Chagas), cecilia_pqu@ya(M.C.F. Ferrari), alaorsantos@hotmail.com (A.S. Filho)awzuardi@fmrp.usp.br (A.W. Zuardi), alprado_psi@ya(D.A. Prado), geraldo.busatto@hcnet.usp.br (G.F. Busaphilip.mcguire@kcl.ac.uk (P.K. McGuire), jhallak@fmrpjcrippa@fmrp.usp.br (J.A.S. Crippa).

0920-9964/$ – see front matter © 2010 Elsevier B.V.doi:10.1016/j.schres.2010.09.016

a b s t r a c t

Article history:Received 9 June 2010Received in revised form 7 September 2010Accepted 23 September 2010Available online 20 October 2010

Magnetic resonance imaging (MRI) studies have reported a variety of brain abnormalities inassociation with schizophrenia. These include a higher incidence of cavum septum pellucidum(CSP), which is consistent with a neurodevelopmental model for this disorder. In this meta-analytic review, we describe and discuss the main CSP MRI findings in schizophrenia spectrumdisorders (SSDs) to date. We adopted as keywords cavum and schizophrenia or psychosis, andthe inclusion criteria were articles in English, with samples of SSD patients compared to healthysubjects, which used MRI to assess CSP, without time limit. From 18 potential reports, fifteenwere eligible to be part of the current review. These studies included 1054 patients with SSDand 866 healthy volunteers. Six out of 15 studies pointed to a higher prevalence of CSP of anysize in SSD patients, while five out of 15 showed that subjects with SSD had a greateroccurrence of a large CSP than healthy individuals. However, the meta-analysis demonstratedthat only the incidence of a large CSP was significantly higher in SSD relative to healthycomparisons (odds ratio=1.59; 95%CI 1.07–2.38; p=0.02). Overall our results suggest thatonly a large CSP is associated with SSD while a small CSP may be considered a normalneuroanatomical variation. Our review revealed a large degree of variability in the methodsemployed across the MRI studies published to date, as well as evidence of publication bias.Studies in large, community-based samples with greater standardization of methods shouldclarify the true significance of CSP in SSD.

© 2010 Elsevier B.V. All rights reserved.

Keywords:Cavum septum pellucidumMagnetic resonance imagingSchizophreniaMeta-analysis

iências e Ciências does 3900, 14048-900: +55 16 3602 2544.iak),mpton@iop.kcl.ac.uklvão-de Almeida),hoo.com.br,hoo.com.brtto),.usp.br (J.E.C. Hallak),

All rights reserved.

1. Introduction

Despite more than a century of research, the etiology ofschizophrenia remains unknown. It is widely accepted thatthe interaction between environmental and biological (i.e.,genetic, biochemical, physiological and developmental) fac-tors is crucial to the manifestation of this disabling disorder.Among several etiological hypotheses for schizophrenia, theneurodevelopmental model is one of the most prominant. Inits simplest form the model posits that schizophrenia is thebehavioral outcome of an aberration in neurodevelopmental

2 C. Trzesniak et al. / Schizophrenia Research 125 (2011) 1–12

processes that begins long before the onset of clinicalsymptoms and is caused by a combination of environmentaland genetic factors (Murray and Lewis, 1987; Weinberger,1987). Several post-mortem and neuroimaging studiessupport this hypothesis by demonstrating that a significantnumber of patients with schizophrenia show brain develop-mental alterations, such as agenesis of the corpus callosum,arachnoid cysts, and other abnormalities (Hallak et al., 2007;Kuloglu et al., 2008). Additionally, animal models of schizo-phrenia suggest that rats with early neurodevelopmentallesions show delayed aberrant behavior occurring duringadolescence (for review, see Shenton et al., 2001). Conversely,in recent years, longitudinal brain imaging studies of bothearly and adult onset populations indicate that brain changesare more dynamic than previously thought, with gray mattervolume loss particularly striking in adolescence and appear-ing to be an exaggeration of the normal developmentalpattern, as well as progressive volumetric reductions in adultpatients. Indeed a recent meta-analysis has confirmedprogression of lateral ventricular volume in patients withschizophrenia (Kempton et al., 2010). Some investigatorshave suggested that, for a proportion of individuals develop-ing schizophrenia, there are disruptions in early (pre- andperinatal) and late (postpubertal) neurodevelopment, basedon evidence of observed changes occurring before, during,and after illness onset (Pantelis et al., 2005). Thus, schizo-phrenia may be a disorder of early and late brain develop-ment (Cannon et al., 2003; Pantelis et al., 2003), in whichgenetic and nongenetic influences are important in under-standing the brain structural abnormalities observed (DeLisi,1997; Harrison andWeinberger, 2005; Rapoport et al., 2005).This progressive neurodevelopmental model may also beuseful in understanding the neuropsychological deficits andbehavioral manifestations seen in the disorder.

The septum pellucidum, a component of the limbicsystem, is a thin plate of two laminae that forms the medialwalls of the lateral ventricles. When these laminae fail to fuse,they form a cavity known as cavum septum pellucidum (CSP)(Sarwar, 1989). The CSP is present in 100% of fetuses andpremature infants, but the posterior half of the leaves arenormally fused by the age 3–6 months (Shaw and Alvord,1969). The presence of a CSP later in life might reflectdevelopmental abnormalities of structures bordering theseptum pellucidum, such as the corpus callosum andhippocampus (Rakic and Yakovlev, 1968). Thus, the CSP canbe considered a marker of limbic system dysgenesis, a formefruste of midline abnormalities, or both (Nopoulos et al.,1998). The complete nonfusion of the two leaflets of theseptum pellucidum, an anomaly termed combined CSP andcavum vergae (CV), is considered the most extreme form ofCSP (Fig. 1).

Magnetic resonance imaging (MRI) is an importantmethod for investigating strutural brain abnormalities inschizophrenia spectrum disorders (SSD), such as the CSP.Nevertheless, there are some difficulties in interpreting thepublished literature to date. In 2001, Shenton et al. reviewed12 published MRI studies reporting CSP prevalence inschizophrenia. The finding that 11 of the studies reported ahigher prevalence of CSP (large/any size) in schizophreniacompared to controls, led the authors to conclude that theCSPwas one of themost robustMRI findings in schizophrenia.

However, 13 new studies have been performed in the lastnine years, with contrasting results both in relation to thepresence and the length of CSP.

The application of recent meta-analytic methods offer astrategy for combining data from multiple brain imagingstudies for qualitative and quantitative analyses. Someadvantages of this approach include: 1) increment instatistical power; 2) the possibility of teasing apart theinfluence of factors causing heterogeneity in the resultsreported by individual studies; 3) the opportunity to combinefindings from all studies, improving the estimation of theoverall effect size (Thompson et al., 1997). Taking advantageof this, we present herein a systematic review and meta-analysis providing a qualitative and quantitative integrationof the MRI findings of CSP in SSD published to date.

2. Methods

Searches were performed in the Medline and Web ofScience databases using the keywords cavum and schizo-phrenia or psychosis, with no time limit. References ofselected articles were also hand-searched for additionalcitations. All studies involving human patient samplesreporting MRI data of CSP in psychotic disorders wereincorporated in this analysis.

To be included in this review, the studies had to meet thefollowing criteria: 1) to be published in English; 2) in apopulation predominantly diagnosed as having schizophre-nia or schizoaffective disorder; 3) to use MRI to assess theCSP; 4) to have a comparison group of healthy subjects. As thebrain is still developing in childhood (Pratt, 2002), we haveexcluded one study that included only children in its sample(Nopoulos et al., 2000).

In order to select the potentially relevant studies, threeindependent reviewers (AGA, CT, and MHNC) evaluated theabstracts identified in the literature search. Next, the samethree reviewers, working independently, decided which ofthose papers fulfilled the inclusion criteria. Disagreement atany stage was resolved by consensus. For each studyinvestigated, a data collection form was used, and thesedata were obtained independently by the three reviewers.The MRI studies included in this review are summarized inTable 1, which lists the sociodemographic characteristics ofeach study sample by first author and year.

Statistical analyses were performed using the Compre-hensive Meta-Analysis Software, version 2.2.048, November7, 2008 (Borenstein et al., 2009). For each study, odds ratiosand 95% confidence intervals were calculated. We used arandom effects model that weighted the studies according tothe inverse variance, and provided the odds ratio and thecorresponding confidence interval.

In order to ascertain whether pooling was viable amongthe selected studies (pN0.05), a Q test of homogeneity for theodds ratios across studies was calculated. The proportion ofpatients from the included studies who presented CSPcompared to those who did not was broken down into 2×2contingency tables. We used Egger's regression test, which isa formal method of estimating publication bias (Egger et al.,1997). The effect of the year of publication was assessed in arandom effects meta-regression model by using the

BA

Fig. 1. Complete fusion defect of the septum pellucidum leaflets in coronal (A) and axial (B) MR images showing (arrow) the presence of Cavum Septi Pellucidi/Cavum Vergae severely enlarged in a patient with schizophrenia (28-years-old) (de Souza Crippa et al., 2006).

3C. Trzesniak et al. / Schizophrenia Research 125 (2011) 1–12

appropriated command in the software. The default optionusing residual maximum likelihood (REML) was selected.

3. Results

Twenty studies that examined the prevalence of CSP inschizophrenia and schizoaffective disorder fulfilled inclusioncriteria for this meta-analysis. Among these, one study wasexcluded due to inconsistency regarding CSP prevalence, i.e.presentation of two different values along the report(Rajarethinam et al., 2008). We were unable to make contactwith the author for the correct data. Three articles (DeGreefet al., 1992b; Fukuzako et al., 1996; Hagino et al., 2001) wereexcluded because their samples were already included inother studies (DeGreef et al., 1992a; Fukuzako and Kodama,1998; Takahashi et al., 2007, respectively). The study by Scottet al. (1993) was rejected for not including a healthycomparison group. Thus 15 studies provided comparativedata from patients with and without CSP (Table 1).

Among the 15 studies analyzed in this review, allinvestigated the presence of CSP regardless of its size. Theincidence of CSP ranged from 1.1% (Shioiri et al., 1996) to89.7% (Takahashi et al., 2008) in healthy volunteers, and from10.0% (Keshavan et al., 2002) to 89.5% (Takahashi et al., 2008)in patients with SSD. An increased occurrence of CSP of anysize in patients with schizophrenia was observed both inchronic (Degreef et al., 1992a; Galarza et al., 2004; Jurjuset al., 1993; Rajarethinam et al., 2001; Shioiri et al., 1996) andfirst-episode SSD subjects (Degreef et al., 1992a; DeLisi et al.,1993). However, nine out of 15 studies failed to findsignificant differences in the prevalence of CSP between SSDpatients and healthy subjects (de Souza Crippa et al., 2006;Flashman et al., 2007; Fukuzako and Kodama, 1998; Kasaiet al., 2004; Keshavan et al., 2002; Kwon et al., 1998;Nopoulos et al., 1997; Takahashi et al., 2007, 2008). Asshown in Fig. 2, the presence of CSP of any size did not differbetween individuals with SSDs in comparison with healthysubjects (odds ratio=1.19; 95%CI 0.82–1.72; p=0.37). There

was significant between-study heterogeneity for the studiesincluded in the present meta-analysis (I2=56.7%, Q=32.3,pb0.01), and no evidence of publication bias (p=0.30).

Contradictory results were also evident when consideringthe prevalence of large CSPs: rates varied from 0 (Keshavanet al., 2002) to 11.5% (Takahashi et al., 2008) in healthyindividuals, and from 1.2% (DeLisi et al., 1993) to 26.7%(Kwon et al., 1998) in SSD patients. Using a qualitativeassessment that consisted of visual inspection (see below),six (DeGreef et al., 1992a; DeLisi et al., 1993; Flashman et al.,2007; Galarza et al., 2004; Jurjus et al., 1993; Keshavan et al.,2002) out of seven articles did not observe differences in thepresence of large CSP between patients and healthy subjects,while only Shioiri et al. (1996) reported significant results. Onthe other hand, with a more quantitative approach, four (deSouza Crippa et al., 2006; Kasai et al., 2004; Kwon et al., 1998;Nopoulos et al., 1997) out of nine (de Souza Crippa et al.,2006; Flashman et al., 2007; Fukuzako and Kodama, 1998;Kasai et al., 2004; Kwon et al., 1998; Nopoulos et al., 1997;Rajarethinam et al., 2001; Takahashi et al., 2007, 2008)studies observed an increased incidence of large CSP in SSDpatients, although the prevalence of CSP of any size did notdiffer between patients and healthy volunteers in severalstudies (de Souza Crippa et al., 2006; Kasai et al., 2004; Kwonet al., 1998; Nopoulos et al., 1997). In the random effectsquantitative meta-analysis, patients with SSDs had signifi-cantly increased rates of large CSP relative to the healthygroup (odds ratio=1.59; 95%CI 1.07–2.38; p=0.02), withpatients with schizophrenia or schizoaffective disorderpresenting 1.59 times the incidence of large CSP than healthyindividuals (Fig. 3). Although there was no between-studyheterogeneity (I2=0, Q=13.7, p=0.47), we found evidenceof publication bias (pb0.01) across the studies.

We also identified differences in the degree of significanceof between-group differences in regard to the presence of CSPof any size (Fig. 4), but not specifically in regard to theprevalence of a large CSP. By carrying out a meta-regressiontaking year of publication as a moderator, we verified that,

Table 1Demographic and clinical characerization of the samples.

Reference Subjects N (M/F) Mean age+SD(years)

Duration ofillness (years)

Antipsy-chotics

Diagnosticcriteria

Comments / Healthy controls origin

Degreef et al. (1992a) FE SCHZ 62 (33/29) 24.1±5.8 1.0±1.7 drug-naïve

RDC CTRL: members of the community, hospital staff

CHR SCHZ 19 (17/2) 29.1±5.7 9.19±5.6 Yes DSM-III-RCTRL 46 (22/24) 28.8±7.5 – –

Jurjus et al. (1993) CHR SCHZ 67 (49/18) ND ND ND DSM-III-R CTRL: NDCTRL 37 (ND) ND

DeLisi et al. (1993) FE SCHZ+SCHZA+SCHZT+NOS

85 (48/37) 26.6±7.3 ND ND DSM-III-R CTRL: members of the community

CTRL 47 (29/18) 26.6±6.6Shioiri et al. (1996) CHR SCHZ 40 (21/19) 32.2±11.3 8.7±9.5 ND DSM-III-R age: difference between groups CTRL: hospital staff and

members from the comunityCTRL 92 (34/98) 38.8±14.0 –

Nopoulos et al. (1997) CHR SCHZ+SCHZA 55 (37/18) 29.7±9.0 ND ND DSM-III-R CTRL: members of the communityCTRL 75 (39/36) 27.3±7.8

Fukuzako and Kodama (1998) CHR SCHZ 72 (53/19) 28.7±9.1 ND Yes DSM-III-R CTRL: members of the community, hospital staff, university studentsCTRL 41 (30/11) 32.0±7.9 –

Kwon et al. (1998) CHR SCHZ a 15 (15/0) ND ND ND DSM-III-R CTRL: hospital staffCTRL a 15 (15/0) ND

Rajarethinam et al.(2001)

CHR SCHZ 73 (56/17) 35.3±11.5 ND ND DSM-III-R CTRL: members of the communityCTRL 43 (21/22) 35.6±13.1

Keshavan et al. (2002) FE SCHZ+FE SCHZA 40 (25/15) 24.7±7.5 NA drug-naïve

DSM-III-R CTRL: members of the community

CTRL 59 (29/30) 21.4±7.5 – – DSM-IVGalarza et al. (2004) CHR SCHZ 32 (0/32) 52.9±9.2 28.0±9.8 Yes DSM-III-R CTRL: ND

CTRL 19 (0/19) 51.1±12.7 –

Kasai et al. (2004) FE SCHZ 33 (28/5) 24.7±6.5 NA Yes DSM-IV CTRL: members of the communityCTRL 56 (44/12) 24.0±3.9 – –

De Souza Crippa et al.(2006)

CHR SCHZ 38 (26/12) 29.9±10.0 10.3±8.2 Yes DSM-IV CTRL: members of the communityCTRL 38 (26/12) 29.7±9.7 –

Flashman et al. (2007) CHR SCHZ+SCHZA+NOS 77 (57/20) 34.3±10.5 ND ND DSM-IV CTRL: NDCTRL 55 (32/33) 32.7±11.0

Takahashi et al. (2007) CHR SCHZ 154 (74/80) 28.0±7.8 5.1±5.5 Yes ICD-10 CTRL: members of the community, hospital staff, university studentsCTRL 163 (97/66) 27.0±8.0 – – DSM-IV

Takahashi et al. (2008) FE SCHZ 103 (76/27) 21.2±3.3 0.2±0.2 Yes DSM-III-R gender: MNF in CHR compared with other all groups age:CHR SCHZNother groups CTRL: similar sociodemographic areas as pacintesby hospital staff; members of the community

CHR SCHZ 89 (76/13) 34.9±9.6 12.8±9.9 YesCTRL 87 (55/32) 26.9±10.1 NA –

CHR, chronic; CTRL, controls; CSP, cavum septum pellucidum; DSM, Diagnostic Statistical Manual; F, female; FE, first episode; ICD, International Classification of Diseases; M, male; NA, not applied; ND, not described; NOS,patients with psychosis not otherwise specified; RDC, Research Diagnostic Criteria; SCHZ, schizophrenics; SCHZA, schizoaffectives; XCHZT, schizotypicals; SD, standard deviation.

a Number of subjects in the study who were not included in posterior studies in the list.

4C.Trzesniak

etal./

SchizophreniaResearch

125(2011)

1–12

sr Odds ratio and 95% CIsr

Fig. 2. The odds ratio of the 15 studies that investigated the prevalence of cavum septum pellucidum of any size in patients with schizophrenia spectrum disordeand in healthy volunteers. The mean data shows no significant differences between patients and healthy comparison subjects (pN0.05). CI, confidence intervalCSP, cavum septum pellucidum; OR, odds ratio.

5C. Trzesniak et al. / Schizophrenia Research 125 (2011) 1–12

during the 1990s, the majority of the findings showed higherprevalence of CSP of any size in SSD patients when comparedto healthy volunteers (Fig. 4). On the other hand, thissignificant difference in the incidence rates between patientsand controls disappears when considering all the studiesconducted in 2000s.

Odds ratio and 95% CI

Healthy volunteers Patients

OR=1.59; 95%CI 1.07-2.38; p=0.02 Test for heterogeneity: I2=0; Q=13.7; p=0.48

reetnuloVyhtlaeHstneitaPydutSDegreef et al. (1993) 2 / 81 0 / 46

Jurjus et al. (1993) 4 / 67 2 / 37

DeLisi et al. (1993) 1 / 85 0 / 47

Shioiri et al. (1996) 2 / 40 0 / 92

Nopoulos et al. (1997) 6 / 55 1 / 75

Fukuzako et al. (1998) 7 / 72 2 / 41

Kwon et al. (1998) 4 / 15 1 / 15

Rajarethinam et al. (2001) 3 / 73 1 / 43

Keshavan et al. (2002) 1 / 40 0 / 59

Galarza et al., (2004) 1 / 32 0 / 19

Kasai et al. (2004) 6 / 33 4 / 56

De Souza Crippa et al. (2006)

Flashman et al. (2007)

Takahashi et al. (2007) 10 / 154 12 / 163

Takahashi et al. (2008) 19 / 192 10 / 87

Total

Large CSP / Total (N)

Odds ratio and 95% CI

OR=1.59; 95%CI 1.07-2.38; p=0.02 Test for heterogeneity: I2=0; Q=13.7; p=0.48

sreetnuloVyhtlaeHstneitaPydutSDegreef et al. (1993) 2 / 81 0 / 46

Jurjus et al. (1993) 4 / 67 2 / 37

DeLisi et al. (1993) 1 / 85 0 / 47

Shioiri et al. (1996) 2 / 40 0 / 92

Nopoulos et al. (1997) 6 / 55 1 / 75

Fukuzako et al. (1998) 7 / 72 2 / 41

Kwon et al. (1998) 4 / 15 1 / 15

Rajarethinam et al. (2001) 3 / 73 1 / 43

Keshavan et al. (2002) 1 / 40 0 / 59

Galarza et al., (2004) 1 / 32 0 / 19

Kasai et al. (2004) 6 / 33 4 / 56

De Souza Crippa et al. (2006) 8 / 38 1 / 388 / 38 1 / 38

Flashman et al. (2007) 11 / 77 5 / 5511 / 77 5 / 55

Takahashi et al. (2007) 10 / 154 12 / 163

Takahashi et al. (2008) 19 / 192 10 / 87

Total 85 / 1054 39 / 86685 / 1054 39 / 866

Large CSP / Total (N)

0.010.01 0.10 1.00 10.00 100.00

Fig. 3. The odds ratio of the 15 studies that investigated the prevalence of large cavum septum pellucidum in patients with schizophrenia spectrum disorder and inhealthy volunteers. The mean data shows significant differences between patients and healthy comparison subjects (pb0.05). CI, confidence interval; CSP, cavumseptum pellucidum; OR, odds ratio.

r;

4. Discussion

This meta-analysis demonstrates the considerable effortthat has been directed towards investigating the relationshipbetween CSP and SSD. The results indicate that the occur-rence of a small-sized CSP may simply reflect normal

1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010

Year

3.00

2.50

2.00

1.50

1.00

0.50

0.00

-0.50

-1.00

-1.50

Meta-regression of year on Log odds ratio

Log

odds

rat

io

Fig. 4. Meta-regression taking year of publication as moderator showing the odds ratio of the prevalence of the cavum septum pellucidum of any size in patientwith schizophrenia spectrum disorder and healthy comparison subjects (Slope=−0.082; ±s.e.=0.029; z=−2.828; p=0.005). The circle size indicates theweight of each study; the bigger the stronger.

6 C. Trzesniak et al. / Schizophrenia Research 125 (2011) 1–12

anatomical variability, since our statistical comparisons failedto reveal significant differences in the prevalence of CSP ofany size between SSD patients and healthy subjects. On theother hand, our analyses showed that the incidence of a largeCSP was higher in SSD patients than in healthy volunteers.We therefore propose that the clinical significance of a CSPmay depend more on its size than whether it is present orabsent, as suggested by Nopoulos et al. (1998) and de SouzaCrippa et al. (2006). In the same way, based on the findingthat an enlarged CSP was more prevalent in individuals atultra-high risk for psychosis (Choi et al., 2008) it could also bespeculated that alterations in the cavity indicates anincreased susceptibility to psychosis. Despite this, it isimportant to emphasize that an enlarged CSP occurs in onlya subgroup of individuals with schizophrenia, with preva-lence rates ranging from 4% (Rajarethinam et al., 2001) to 30%(Kwon et al., 1998), respectively. Therefore, this midlinestructural abnormality should at most be regarded as an earlyneurodevelopmental risk factor that may be related to thepresence of schizophrenia in a subgroup of patients, ratherthan being a strong causative determinant of the disorder.

It is well known that fusion of the septi pellucidi isassociated with rapid growth of the hippocampus, corpuscallosum and other midline structures (Sarwar, 1989; Shawand Alvord, 1969) that have consistently been linked toschizophrenia (Shenton et al., 2001). Therefore, the pres-ence of CSP in patients with schizophrenia could representan early marker of a developmental defect involving thesebrain regions. Similarly, it is unlikely that a disturbance in alocalized structure such as the CSPwould lead to widespreadmanifestations of schizophrenia and other disorders. A moreplausible possibility is that abnormalities in the formationand maturation of the septum pellucidum are a marker of anoverall aberrance of early neurodevelopment of morewidespread proportions and etiopathological significanceto psychosis.

It is also important to stress that CSP abnormalities are notspecific to schizophrenia or psychosis in general, since thesefindings are also observed in other disorders of developmen-tal origin, such as fetal alcohol syndrome (prevalence of 20%in patients; Swayze et al., 1997), Sotos (38–40% of thepatients have CSP, being 94–100% CV, Lim and Yoon, 2008;

s

Schaefer et al., 1997), and Apert's syndrome (prevalence of15% in patients; Renier et al., 1996). In addition, although thesize and prevalence of CSP are normal in subjects with otherpsychiatric conditions such as panic disorder (patients have76% of any CSP; Crippa et al., 2004), there have been reports ofa higher prevalence of CSP in subjects with first-episodeaffective psychosis (any CSP: 81%; large CSP: 15%, Kasai et al.,2004); bipolar affective disorder (BAD) (any CSP: 7%, Shioiriet al., 1996) and schizotypal personality disorder (SPD) (largeCSP: 20%, Dickey et al., 2007). Based on this, some authorshave speculated that psychosis associated with schizophreniaand BAD may share, at least to some extent, neurodevelop-mental abnormalities involving midline structures (Kasaiet al., 2004) and that SPD may be a milder form on acontinuum of SSD (Kwon et al., 1998).

4.1. Methodological aspects: sample characteristics

The wide discrepancy in the reported prevalence and sizeof CSP in association with SSD may be, in great part,explained by differences in the methodology among thestudies published to date. Similarly, this methodologicalvariation may have also be the reason for the differences inthe prevalence of CSP of any size during the 1990s and the2000s. First, not all of the studies used homogeneoussamples of patients (DeLisi et al., 1993; Flashman et al.,2007; Jurjus et al., 1993; Keshavan et al., 2002; Nopouloset al., 1997); i.e., the patient groups in these studies hadmore than one disorder and it was not possible to evaluatethe rates in each diagnostic category separately. Besidesschizophrenia, most of the samples also included otherconditions related to psychosis, such as schizotypal person-ality (DeLisi et al., 1993) and schizoaffective disorder (DeLisiet al., 1993; Flashman et al., 2007; Jurjus et al., 1993;Keshavan et al., 2002; Nopoulos et al., 1997). Although someauthors hypothesized that psychosis in general may share,at least to some extent, neurodevelopmental abnormalitiesinvolving midline structures (Kasai et al., 2004; Kwon et al.,1998), the variability of the diagnoses across differentstudies makes it difficult to compare their results. Themethods for recruitment of healthy comparison groupscould also be considered a source of bias, since most studies

7C. Trzesniak et al. / Schizophrenia Research 125 (2011) 1–12

to date recruited hospital staff, university students; ormembers of community by advertisements (Table 1). More-over, some studies did not detail the recruitment of theirhealthy control samples (Flashman et al., 2007; Galarzaet al., 2004; Jurjus et al., 1993).

In all but one of the MRI studies to date, the samples of SSDpatients and controls were not matched by sex (Takahashiet al., 2008). Although the authors of these studies did notattribute their negative findings to this unbalance, two articlesfound that male individuals showed higher incidence of CSP ofany size than female subjects (Jurjus et al., 1993; Nopouloset al., 1997). There is also evidence that male patients maypresent a higher occurrence of any (Rajarethinam et al., 2001)or large (de Souza Crippa et al., 2006; Nopoulos et al., 1997)CSP than male healthy subjects. In contrast to such gender-related differences, the age of subjects did not seem toinfluence the variability of CSP prevalence in samples of SSD,since none of the MRI studies to date showed significantcorrelations between this variable and CSP measures.

There is evidence of volumetric differences in brainstructures in the vicinity of the CSP, such as reduced volumeof the hippocampus and reduced cross-sectional area of thecorpus callosum, when first-episode psychosis patients aredirectly compared with chronic patients with SSD (Ellison-Wright et al., 2008; Meisenzahl et al., 2008). This indicatesthat it is also important to control for the duration of illness inMRI studies when evaluating the CSP in psychotic disorders.Given that the fusion of the septi pellucidi probably resultsfrom the rapid growth of the hippocampus (Sarwar, 1989;Shaw and Alvord, 1969), it could be speculated thatvolumetric alterations in this brain structure over the courseof time could also lead to changes in morphologicalparameters related to the CSP, i.e. length, area and volume.Indeed, although the septum pellucidum fuses early in life,there is evidence of higher prevalence of CSP in professionalboxers, probably due to repeated head trauma (Aviv et al.,2010; Zhang et al., 2003). None of the eight (de Souza Crippaet al., 2006; DeGreef et al., 1992a; Galarza et al., 2004; Kasaiet al., 2004; Keshavan et al., 2002; Shioiri et al., 1996;Takahashi et al., 2007, 2008) studies that controlled theirfindings for duration of illness in this review foundsignificant correlations between this variable and measuresof CSP. However, it has been reported that patients withchronic schizophrenia display higher incidences of large CSPthan healthy individuals, while patients with first-episodepsychosis show an intermediate prevalence (Kwon et al.,1998).

There is no mention about the use of antipsychotics inseven studies (DeLisi et al., 1993; Flashman et al., 2007;Jurjus et al., 1993; Kwon et al., 1998; Nopoulos et al., 1997;Rajarethinam et al., 2001; Shioiri et al., 1996). This lack ofinformation regarding treatment status possibly relates tothe view that the presence of CSP is entirely establishedduring early life, and therefore CSP incidence indices wouldnot be affected by the use of antipsychotic medications.Nonetheless, Dickey et al. (2007) suggest that these drugsmay influence measures of CSP, given that the use of thesedrugs in SSD have already been related to morphologicalalterations in brain regions related to the CSP, such as thehippocampus (Chakos et al., 2005; McClure et al., 2006) andcorpus callosum (Girgis et al., 2006).

4.2. Methodological aspects: experimental designs andcharacterization of CSP

The variation in the acquisition parameters across differ-ent MRI studies may influence the results obtained in thestudies of CSP in SSDs. In relation to the intensity of themagnetic field, only one study used a 1.0 T scanner (DeGreefet al., 1992a). In contrast, the other 14 studies usedequipment with magnetic field of 1.5 T, which probablycontributed to improve the quality of the MRI data acquired.

There has been a great degree of variability in thethickness of MRI slices acquired across different studies todate, ranging from 0.94 mm (Kasai et al., 2004; Takahashiet al., 2008) to 5.0 mm (DeLisi et al., 1993; Jurjus et al., 1993;Shioiri et al., 1996). There is also variation in the inter-slicegaps across MRI studies, ranging from 1.5 mm (Shioiri et al.,1996) to 5 mm (Jurjus et al., 1993). Thinner and contiguousslices are considered the ‘gold-standard’ method, since theseallow more accurate estimates not only of the prevalence ofCSPs, but also their size (Nopoulos et al., 1997). It is well-known that with thicker slices, such as 3.0 to 5.0 mm, gapsmay occasionally miss small CSPs, as well as leading to partialvolume effects. This factor may have contributed to thediscrepancy in the findings of CSP of any size between thestudies conducted in the 1990s and 2000s, given that the firststudies were carried out with thick and non contiguous slices(Table 2).

There is also great variation in the way that the studiesdetermined the presence of the CSP, as well as the parametersfor evaluating whether this cavity is classified as large or not(Table 2). Concerning the criteria for detection, many authorshave categorized the CSP using a qualitative approach,consisting of both visual inspection and ranking findings ongrades of severity. Scales of severity are often created for eachstudy individually, and are established from previous knowl-edge about characteristics of the CSPs. These scales may rangefrom 0 (absent CSP) to 3 (severe CSP) (DeGreef et al., 1992a;DeLisi et al., 1993); from 1 (absent CSP) to 3 (severe CSP)(Keshavan et al., 2002); and from 0 (absent CSP) to 4 (severeCSP), with five distinctive levels (Flashman et al., 2007; Jurjuset al., 1993; Shioiri et al., 1996). Using a different approach,Galarza et al. (2004) used embryological types of CSP (from Ito III) as parameters for rating the severity of the cavity.Thereafter, each CSP of the sample is classified according to itssize, and placed in the appropriated category. Limitations insuch qualitative forms of assessing CSPsmay contribute to thecontradictory findings among the articles published to date,as the scales differ across studies and the ratings given byeach examiner are somewhat subjective.

In order to minimize these discrepancies, recent studieshave adopted more quantitative methods for classifying theCSP, by counting the number of slices in which the cavityclearly appears, especially on coronal MRI views. When thistechnique is used, the number of slices can be multiplied bytheir thickness, allowing the calculation of the anterior-to-posterior length of the CSP. This methodology, now consid-ered the state-of-art technique to assess the presence andseverity of CSP, was first used by Nopoulos et al. (1996), andemployed in most of the subsequent studies (de Souza Crippaet al., 2006; Flashman et al., 2007; Kasai et al., 2004; Kwonet al., 1998; Nopoulos et al., 1997; Rajarethinam et al., 2001;

Table 2Magnetic resonance imaging studies of the csp in schizophrenia spectrum.

References MagneticField (T)

Slice length /gap (mm)

Criteria ofdetection of CSP

Criteria oflarge CSP

Subjects Prevalenceof anyCSP (%) a

Prevalenceof largeCSP (%) b

Measuresof CSP(prevalence)

Findings Comments

Degreef et al.(1992a)

1.0 3.1 Visual inspection basedon a grade (0–3)

Grade 3 FE SCHZ 34.5 3.2 any CSP ↑ in FE and inCHR x

CHR SCHZ 26.3 0 Grade 3 CTRLCTRL 15.2 0 NS

Jurjus et al.(1993)

1.5 5.0 / 5.0 Visual inspection basedon a grade (0–4)

Grade 4 CHR SCHZ 25.4 6.0 any CSP NS male had higher rates of CSP thanfemale subjectsCTRL 18.9 5.4 Grade 3–4 NS

DeLisi et al.(1993)

1.5 5.0 / 2.0 Visual inspection basedon a grade (0–3)

Grade 3 FE SCHZ+SCHZA+SCHZT+NOS

44.7 1.2 any CSP ↑ in psychotics NS differences between CSP gradesand the size of ventricles, TC andCC or asymetries

CTRL 29.8 0 Grade 3 NSShioiri et al.(1996)

1.5 5.0 / 1.5 Visual inspection basedon a grade (0–4)

Grade 4 CHR SCHZ 17.5 5.0 any CSP ↑ in SCHZ x CTRLCTRL 1.1 0 Grade 3–4 ↑ in SCHZ x CTRL

Nopoulos et al.(1997)

1.5 1.5 Number of 1.5 mm slices inwhich CSP was visualized

N4 slices CHR SCHZ+SCHZA

58.2 10.1 any CSP NS male had higher incedence of anyCSP than female subjects malepsychotics had higher incidence oflarge CSP than male CTRL

(N6 mm) CTRL 58.7 1.3 large CSP ↑ in psychotics

Fukuzako andKodama (1998)

1.5 1.0 Number of 1 mm slices inwhich CSP was visualized

N5 mm CHR SCHZ 47.2 9.7 any CSP NS SCHZ with long-term admissionshad higher incidence of CSPCTRL 39.0 4.9 large CSP NS

Kwon et al.(1998)

1.5 1.5 Number of 1.5 mm slicesin which CSP was visualized

N4 slices CHR SCHZ 73.3 26.7 any CSP NS number of slices of CSP negativelycorrelated with left and right vol.of hippocampus in CHR SCHZ

Nopouloset al. (1997)criteria

CTRL 86.7 6.7 large CSPNopouloscriteria

↑ in SCHZ x CTRL ↑in SCHZ x CTRL(trend)

Rajarethinamet al. (2001)

1.5 1.0 Number of 1 mmslices in which CSPwas visualized

N6 slices CHR SCHZ 60.3 4.1 any CSP ↑ in SCHZ male SCHZ had higher incidence ofany CSP than male CTRL

(N6 mm) CTRL 41.9 2.3 large CSP NS male CTRL

8C.Trzesniak

etal./

SchizophreniaResearch

125(2011)

1–12

Keshavanet al. (2002)

1.5 3.0 Visual inspection basedon a grade (1–3)

Grade 3 FE SCHZ+FE SCHZA 10.0 2.5 any CSP NSCTRL 11.9 0 Grade 3 NS

Galarza et al.(2004)

1.5 1.0 Based on embryologicaltypes (I–III)

NA CHR SCHZ 43.8 NA any CSP ↑ in SCHZCTRL 10.5 NA Type I NS

Type II NSType III NS

Kasai et al.(2004)

1.5 0.94 Number of 0.94 mm slicesin which CSP was visualized

N6 slices FE SCHZ 59.0 18.2 any CSP NS SCHZ: number of slices of CSPpositively correlated with thinkingdisturbance; CSP rating negativelycorrelated with left PHG volume

(N5.6 mm) CTRL 81.7 7.1 large CSP ↑ in SCHZ × CTRL

De Souza Crippaet al. (2006)

1.5 1.0 Number of 1 mm slices inwhich CSP was visualized

N6 slices CHR SCHZ 78.9 21.1 any CSP NS 1 SCHZ had carvum vergae Large CSPcommoner in male SCHZ than maleCTRL Large CSP not associated withabsent adhesio interthalamic

(N6 mm) CTRL 86.8 2.6 length NSlarge CSP ↑ in SCHZvol NS

Flashmanet al. (2007)

1.5 1.0 Visual inspection basedon a grade (0–4) Numberof 1 mm slices in whichCSP was visualized

Grade 4N CHR SCHZ+SCHZA+NOS

68.8 14.3 any CSP NS large CSP psychotics showed worseperformance on CVLT-II CSP lengthpositively correlated with negativesymptoms (SANS) CSP length positivelycorrelated with reaction time andvigilance of the CPT in psychotics;negatively correlated in CTRL

6 slices(N6 mm)

CTRL 76.4 9.1 length NS

grade 4 NSlarge CSP NS

Takahashi et al.(2007)

1.5 1.0 Number of 1 mm slices inwhich CSP was visualized

N6 slices CHR SCHZ 76.0 6.5 any CSP NS large CSP psychotics showed ↓bilateral amygdala and left posteriorPHG vol. x psychotics with small CSPCSP length negatively correlated withamygdala and left posterior PHG vol.in psychotics

(N6 mm) CTRL 81.6 7.4 length NSlarge CSP NS

Takahashi et al.(2008)

1.5 0.94 Number of 0.94 mm slicesin which CSP was visualized

N6 slices FE SCHZ 89.5 9.3 any CSP NS(N5.6 mm) CHR SCHZ 87.6 11.2 length NS

CTRL 89.7 11.5 large CSP NS

↓, decreased; ↑, increased; CC, corpus callosum; CHR, chronic; CPT, Continuous Performance Test; CSP, cavum septum pellucidum; CTRL, controls; CVLT-II, The California Verbal Learning Test II; FE, first episode; L, left; mm,millimeter; NA, not applied; ND, not described; NOS, patients with psychosis not otherwise specified; NS, no significant; PHG, parahippocampal gyrus; R, right; SANS, scale for the assessment of negative symptoms; SCHZ,schizophrenics; SCHZA, schizoaffectives; T, Tesla; TC, temporal cortex; vol., volume.

a The prevalence was calculated as follows: 100×(number of subjects with any CSP/number of all subjects).b The prevalence was calculated as follows: 100×(number of subjects with large CSP/number of all subjects).

9C.Trzesniak

etal./

SchizophreniaResearch

125(2011)

1–12

10 C. Trzesniak et al. / Schizophrenia Research 125 (2011) 1–12

Takahashi et al., 2007, 2008). This has proven to be the mostsensitive and reliable way to classify CSP (Kwon et al., 1998;Nopoulos et al., 1997).

Despite the advances brought by this more quantitativemethod, there are still conflicting results among the studiesthat employed such technique. Six out of nine researchreports observed higher prevalence of large CSP in SSDpatients relative to healthy individuals (de Souza Crippa et al.,2006; Fukuzako and Kodama, 1998; Kasai et al., 2004; Kwonet al., 1998; Nopoulos et al., 1997). Such disagreement couldbe explained by the variability of criteria for determining thedefinition of a large CSP across studies, as well as clinical anddemographic differences among samples. Fukuzako andKodama (1998) defined a large CSP as a cavity visible inmore than five 1 mm MRI slices (i.e., N5 mm). However,based on post-mortem findings (Shunk, 1963), most authorshave defined CSPs with anterior-to-posterior length spanning6 or more millimeters as large. Therefore, in both studies inwhich the thickness of the contiguous slices was 1.5 mm, theauthors considered cavities greater than or equal to 4 slices aslarge (≥6 mm; Kwon et al., 1998; Nopoulos et al., 1997).When the thickness of the slices was 0.94 (Kasai et al., 2004;Takahashi et al., 2008) or 1.0 mm contiguous, CSPs wereclassified as large when greater or equal to 6 slices (≥5.6–6.00 mm; de Souza Crippa et al., 2006; Flashman et al., 2007;Kasai et al., 2004; Rajarethinam et al., 2001; Takahashi et al.,2007, 2008).

4.3. Correlations with quantitative and qualitative variables

Considering all the CSP studies in SSD, including post-mortem investigations, there have been reports correlatingthe presence of CSP in SSD patients to poor prognosis(Fukuzako and Kodama, 1998) and a family history ofpsychosis (Uematsu and Kaiya, 1989). Moreover, small CSPshave been previously linked to decreased volume of thefrontal and temporal lobe tissues (Nopoulos et al., 1996).Although not always replicated, the presence of large CSP inSSD has also been associatedwith reduced IQ (Nopoulos et al.,2000), more negative symptoms (Flashman et al., 2007) andmore severe thinking disturbance (Kasai et al., 2004), highersuicide rates (Filipović et al., 2005), higher rates of exposureto in utero infection (Brown et al., 2009) and greater cognitivedeficits (Flashman et al., 2007; Nopoulos et al., 2000).Moreover, although our group failed to find an associationbetween the presence of CSP and patterns of autonomicsystem activity in subjects with schizophrenia (Crippa et al.,2007), it was demonstrated that patients with large CSP hadmore pronounced rightN left brain volume asymmetry andreduced volumes of the left temporal lobe (Nopoulos et al.,1996), bilateral hippocampus (Kwon et al., 1998), bilateralamygdala and left parahippocampal gyrus relative to controls(Kasai et al., 2004; Takahashi et al., 2007). Taken together,these findings suggest that patientswith a large CSPmay havegreater psychopathological impairment, possibly reflectingdistinct patterns of disturbed brain morphology.

4.4. Limitations

The main difficulty in analyzing the 15 articles included inthe meta-analysis arose from the wide variation in the

methodological variables across studies, as discussed above.Additionally, even though these MRI investigations wererigorous in many ways—particularly the recent ones—few ofthem reached satisfactory criteria for inclusion in the meta-analysis, both in respect to the clinical and demographiccharacteristics of the samples studied, and in the parametersof image acquisition. The less than ideal use of thick slices(3.0–5.0 mm) and inter-slice gaps in the imaging acquisitionprotocols, variability in the criteria used for categorizing CSP,and the use of samples unmatched for critical demographicvariables are among the factors likely to add variability to thefindings reported across different studies. Using optimalinclusion criteria regarding the number of SSD patientsincluded (N20 subjects), an adequate matching between thesamples of patients and controls, acquisition of thinner andcontiguous slices (≤1.0 mm), and quantitative analysis of CSPsize, five well-designed studies can be selected from theliterature (de Souza Crippa et al., 2006; Flashman et al., 2007;Kasai et al., 2004; Rajarethinam et al., 2001; Takahashi et al.,2007). Four out of five of these studies failed to find a higherincidence of CSP of any size in SSD compared to healthyindividuals (Table 2). This reinforces the proposition that thepresence of a small CSP should be considered part of thenormal neuroanatomical variability (Nopoulos et al., 1997).However, only two of these papers observed higher preva-lence of large CSP in patients (de Souza Crippa et al., 2006;Kasai et al., 2004).

It is important to highlight some limitations of the presentreview. As with all meta-analyses, the findings are dependenton the quality of the primary studies. We excluded duplicatedsamples and results when there was indication of more thanone publication based on the same dataset. However, wecannot exclude the possibility that our findings wereinfluenced by confounding factors, or methodological het-erogeneity among the articles. In addition, several types ofstudy bias could arise during publication of the primarystudies (Naylor, 1997). In our meta-analysis, we detectedevidence of publication bias across the studies concerning theprevalence of large CSP. This phenomenon is related to thefact that articles which report positive or significant resultsare more likely to be published than studies reportingnegative findings (Dwan et al., 2008). As observed in Fig. 3,although most of the 15 studies included in this review pointto higher prevalence of large CSP in SSP, the last two(Takahashi et al., 2007, 2008), which included a large numberof patients, show negative results. We believe that this is thereason for the significant value related to the publication biaswe found. Although this finding does not invalidate ouranalysis, it casts doubts on the real implication of large CSP inSSD.

4.5. Conclusion

In summary, the results of the present meta-analysissuggest that there is no increase in the prevalence of CSP ofany size in SSD patients. In terms of a large CSP, the publishedliterature suggests this is more common in SSD patients,however there is also evidence that studies finding theopposite result may have not been published. Thus, furtherinvestigations are warranted to reconcile some conflictingfindings in the literature to date. First, there is a need for

11C. Trzesniak et al. / Schizophrenia Research 125 (2011) 1–12

studies evaluating the prevalence of psychosis in large,community-based samples of individuals presenting withCSP. Second, since many clinical aspects of SSD are hetero-geneous, standardizantion of MRI methodology and clearreporting of clinical variables such as duration of illness andmedication use is advisible. Given that the fusion of the septipellucidi is associated with hippocampal growth, volumetricalterations in the later, over the course of time or due toantipsychotic use could also lead to changes in morphologicalparameters related to the CSP, even in adulthood. Finaly, amore judicious selection of the subjects—such as in firstepisode psychosis population-based investigations withlongitudinal designs, is likely to generate more reliableconclusions concerning the role of CSP in SSD, as well aselucidating the complex clinical and functional effects of thisneurodevelopmental anomaly.

Role of funding sourceSupported in part by grants from ‘Coordenação de Aperfeiçoamento de

Pessoal de Nível Superior’(CAPES-Brazil, process number 5797/09-8),‘Conselho Nacional de Desenvolvimento Científico e Tecnológico’ (CNPq-Brazil) and from ‘Fundação de Amparo à Pesquisa do Estado de São Paulofellowship’ (FAPESP-Brazil). JASC, GFB, JECH and AWZ are recipients of aCNPq (Brazil) fellowship Award. The funding agencies had no further role instudy design; in the collection, analysis and interpretation of data; in thewriting of the report; and in the decision to submit the paper for publication.

ContributorsCT participated the literature review, the data extraction and drafted the

first manuscript. JASC and JECH participated in the conception and providedscientific supervision. IRO and MJK provided statistical advice and scientificsupervision. MCFF, ASF and DAP helped to draft manuscript. AGA and MHNCcontributed data extraction, literature review and made contributions to thewritten manuscript. MJK, PKM, GBF and AWZ contributed content knowl-edge and made contribuitions to the written manuscript. All authors readand approved the final manuscript.

Conflict of interestThe authors declare that they have no competing interests.

AcknowledgementWe thank to Prof. Tsutomu Takahashi and Kiyoto Kasai, who kindly

provided the raw data necessary for our analysis.

References

Aviv, R.I., Tomlinson, G., Kendall, B., Thakkar, C., Valentine, A., 2010. Cavumsepti pellucidi in boxers. Can. Assoc. Radiol. J. 61 (1), 29–32.

Borenstein, M., Hedges, L., Higgins, J., Rothstein, H., 2009. Comprehensivemeta-analysis. Available at: http://www.meta-analysis.com/.

Brown, A.S., Deicken, R.F., Vinogradov, S., Kremen, W.S., Poole, J.H., Penner, J.D.,Kochetkova, A., Kern, D., Schaefer, C.A., 2009. Prenatal infection and cavumseptum pellucidum in adult schizophrenia. Schizophr. Res. 108, 285–287.

Cannon, T.D., van Erp, T.G., Bearden, C.E., Loewy, R., Thompson, P., Toga, A.W.,Huttunen, M.O., Keshavan, M.S., Seidman, L.J., Tsuang, M.T., 2003. Earlyand late neurodevelopmental influences in the prodrome to schizophre-nia: contributions of genes, environment, and their interactions.Schizophr. Bull. 29 (4), 653–669.

Chakos, M.H., Schobel, S.A., Gu, H., Gerig, G., Bradford, D., Charles, C.,Lieberman, J.A., 2005. Duration of illness and treatment effects onhippocampal volume in male patients with schizophrenia. Br. J.Psychiatry 186, 26–31.

Choi, J.S., Kang, D.H., Park, J.Y., Jung, W.H., Choi, C.H., Chon, M.W., Jung, M.H.,Lee, J.M., Kwon, J.S., 2008. Cavum septum pellucidum in subjects at ultra-high risk for psychosis: compared with first-degree relatives of patientswith schizophrenia and healthy volunteers. Prog. Neuropsychopharma-col. Biol. Psychiatry 32 (5), 1326–1330.

Crippa, J.A., Uchida, R., Busatto, G.F., Guimarães, F.S., Del-Ben, C.M., Zuardi, A.W.,Santos, A.C., Araújo, D., McGuire, P.K., Graeff, F.G., 2004. The size and

prevalence of the cavum septum pellucidum are normal in subjects withpanic disorder. Braz. J. Med. Biol. Res. 37 (3), 371–374.

Crippa, J.A., Waldo Zuardi, A., Trzesniak, C., Hallak, J.E., Busatto, G.F., McGuire,P.K., 2007. No association between electrodermal hyporesponsivity andenlarged cavum septi pellucidi in schizophrenia. Schizophr. Res. 95 (1–3), 256–258.

de Souza Crippa, J.A., Zuardi, A.W., Busatto, G.F., Sanches, R.F., Santos, A.C.,Araújo, D., Amaro, E., Hallak, J.E., Ng, V., McGuire, P.K., 2006. Cavumseptum pellucidum and adhesio interthalamica in schizophrenia: an MRIstudy. Eur. Psychiatry 21 (5), 291–299.

Degreef, G., Bogerts, B., Falkai, P., Greve, B., Lantos, G., Ashtari, M., Lieberman,J., 1992a. Increased prevalence of the cavum septum pellucidum inmagnetic resonance scans and post-mortem brains of schizophrenicpatients. Psychiatry Res. 45 (1), 1–13.

Degreef, G., Lantos, G., Bogerts, B., Ashtari, M., Lieberman, J., 1992b.Abnormalities of the septum pellucidum on MR scans in first-episodeschizophrenic patients. AJNR Am. J. Neuroradiol. 13 (3), 835–840.

DeLisi, L.E., 1997. Is schizophrenia a lifetime disorder of brain plasticity,growth and aging? Schizophr. Res. 23 (2), 119–129.

DeLisi, L.E., Hoff, A.L., Kushner, M., Degreef, G., 1993. Increased prevalence ofcavum septum pellucidum in schizophrenia. Psychiatry Res. 50 (3),193–199.

Dickey, C.C., McCarley, R.W., Xu, M.L., Seidman, L.J., Voglmaier, M.M.,Niznikiewicz, M.A., Connor, E., Shenton, M.E., 2007. MRI abnormalitiesof the hippocampus and cavum septi pellucidi in females withschizotypal personality disorder. Schizophr. Res. 89, 49–58.

Dwan, K., Altman, D.G., Arnaiz, J.A., Bloom, J., Chan, A.W., Cronin, E., Decullier,E., Easterbrook, P.J., Von Elm, E., Gamble, C., Ghersi, D., Ioannidis, J.P.,Simes, J., Williamson, P.R., 2008. Systematic review of the empiricalevidence of study publication bias and outcome reporting bias. PLoS ONE28 (3), e3081 (8).

Egger, M., Davey Smith, G., Schneider, M., Minder, C., 1997. Bias inmetaanalysis detected by a simple, graphical test. Bmj 315 (7109),629–634.

Ellison-Wright, I., Glahn, D.C., Laird, A.R., Thelen, S.M., Bullmore, E., 2008. Theanatomy of first-episode and chronic schizophrenia: an anatomicallikelihood estimation meta-analysis. Am. J. Psychiatry 165 (8),1015–1023.

Filipović, B., Kovacević, S., Stojicić, M., Prostran, M., Filipović, B., 2005.Morphological differences among cavum septi pellucidi obtained inpatients with schizophrenia and healthy individuals: forensic implica-tions. A post-mortem study. Psychiatry Clin. Neurosci. 59 (1), 106–108.

Flashman, L.A., Roth, R.M., Pixley, H.S., Cleavinger, H.B., McAllister, T.W.,Vidaver, R., Saykin, A.J., 2007. Cavum septum pellucidum in schizophre-nia: clinical and neuropsychological correlates. Psychiatry Res. 154 (2),147–155.

Fukuzako, H., Kodama, S., 1998. Cavum septum pellucidum in schizophrenia.Biol. Psychiatry 43 (6), 467.

Fukuzako, T., Fukuzako, H., Kodama, S., Hashiguchi, T., Takigawa, M., 1996.Cavum septum pellucidum in schizophrenia: a magnetic resonanceimaging study. Psychiatry Clin. Neurosci. 50 (3), 125–128.

Galarza, M., Merlo, A.B., Ingratta, A., Albanese, E.F., Albanese, A.M., 2004.Cavum septum pellucidum and its increased prevalence in schizophre-nia: a neuroembryological classification. J. Neuropsychiatry Clin. Neu-rosci. 16 (1), 41–46.

Girgis, R.R., Diwadkar, V.A.,Nutche, J.J., Sweeney, J.A., Keshavan,M.S.,Hardan,A.Y.,2006. Risperidone in first-episode psychosis: a longitudinal, exploratoryvoxel-based morphometric study. Schizophr. Res. 82 (1), 89–94.

Hagino, H., Suzuki, M., Kurokawa, K., Mori, K., Nohara, S., Takahashi, T.,Yamashita, I., Yotsutsuji, T., Kurachi, M., Seto, H., 2001. Magneticresonance imaging study of the cavum septi pellucidi in patients withschizophrenia. Am. J. Psychiatry 158 (10), 1717–1719.

Hallak, J.E., Crippa, J.A., Pinto, J.P., Machado de Sousa, J.P., Trzesniak, C.,Dursun, S.M., McGuire, P., Deakin, J.F., Zuardi, A.W., 2007. Total agenesisof the corpus callosum in a patient with childhood-onset schizophrenia.Arq. Neuropsiquiatr. 65 (4B), 1216–1219.

Harrison, P.J., Weinberger, D.R., 2005. Schizophrenia genes, gene expression,and neuropathology: on the matter of their convergence. Mol. Psychiatry10 (1), 40–68 (image 5).

Jurjus, G.J., Nasrallah, H.A., Olson, S.C., Schwarzkopf, S.B., 1993. Cavumseptum pellucidum in schizophrenia, affective disorder and healthycontrols: a magnetic resonance imaging study. Psychol. Med. 23 (2),319–322.

Kasai, K., McCarley, R.W., Salisbury, D.F., Onitsuka, T., Demeo, S., Yurgelun-Todd, D., Kikinis, R., Jolesz, F.A., Shenton, M.E., 2004. Cavum septipellucidi in first-episode schizophrenia and first-episode affectivepsychosis: an MRI study. Schizophr. Res. 71 (1), 65–76.

Kempton, M.J., Stahl, D., Williams, S.C., DeLisi, L.E., 2010. Progressive lateralventricular enlargement in schizophrenia: a meta-analysis of longitudi-nal MRI studies. Schizophr. Res. 120 (1–3), 54–62.

12 C. Trzesniak et al. / Schizophrenia Research 125 (2011) 1–12

Keshavan, M.S., Jayakumar, P.N., Diwadkar, V.A., Singh, A., 2002. Cavum septipellucidi in first-episode patients and young relatives at risk forschizophrenia. CNS Spectr. 7 (2), 155–158.

Kuloglu, M., Caykoylu, A., Yilmaz, E., Ekinci, O., 2008. A left temporal lobearachnoid cyst in a patient with schizophrenia-like psychosis: a casereport. Prog. Neuropsychopharmacol. Biol. Psychiatry 32 (5),1353–1354.

Kwon, J.S., Shenton, M.E., Hirayasu, Y., Salisbury, D.F., Fischer, I.A., Dickey, C.C.,Yurgelun-Todd, D., Tohen, M., Kikinis, R., Jolesz, F.A., McCarley, R.W.,1998. MRI study of cavum septi pellucidi in schizophrenia, affectivedisorder, and schizotypal personality disorder. Am. J. Psychiatry 155 (4),509–515.

Lim, J.J., Yoon, S.H., 2008. The first neurosurgical analysis of 8 Korean childrenwith Sotos syndrome. J. Korean Neurosurg. Soc. 44 (4), 240–244.

McClure, R.K., Phillips, I., Jazayerli, R., Barnett, A., Coppola, R., Weinberger, D.R.,2006. Regional change in brain morphometry in schizophrenia associatedwith antipsychotic treatment. Psychiatry Res. 148 (2–3), 121–132.

Meisenzahl, E.M., Koutsouleris, N., Bottlender, R., Scheuerecker, J., Jäger, M.,Teipel, S.J., Holzinger, S., Frodl, T., Preuss, U., Schmitt, G., Burgermeister,B., Reiser, M., Born, C., Möller, H.J., 2008. Structural brain alterations atdifferent stages of schizophrenia: a voxel-based morphometric study.Schizophr. Res. 104 (1–3), 44–60.

Murray, R.M., Lewis, S.W., 1987. Is schizophrenia a neurodevelopmentaldisorder? Brit. Med. J. 295 (6600), 681–682.

Naylor, C.D., 1997. Meta-analysis and the meta-epidemiology of clinicalresearch. Br. Med. J. 315, 617–619.

Nopoulos, P., Swayze, V., Andreasen, N.C., 1996. Pattern of brain morphologyin patients with schizophrenia and large cavum septi pellucidi. J.Neuropsychiatry Clin. Neurosci. 8 (2), 147–152.

Nopoulos, P., Swayze, V., Flaum, M., Ehrhardt, J.C., Yuh, W.T., Andreasen, N.C.,1997. Cavum septi pellucidi in normals and patients with schizophreniaas detected by magnetic resonance imaging. Biol. Psychiatry 41 (11),1102–1108.

Nopoulos, P.C., Giedd, J.N., Andreasen, N.C., Rapoport, J.L., 1998. Frequencyand severity of enlarged cavum septi pellucidi in childhood-onsetschizophrenia. Am. J. Psychiatry 155 (8), 1074–1079.

Nopoulos, P., Krie, A., Andreasen, N.C., 2000. Enlarged cavum septi pellucidiin patients with schizophrenia: clinical and cognitive correlates. J.Neuropsychiatry Clin. Neurosci. 12 (3), 344–349.

Pantelis, C., Yücel, M., Wood, S.J., McGorry, P.D., Velakoulis, D., 2003. Earlyand late neurodevelopmental disturbances in schizophrenia and theirfunctional consequences. Aust. N.Z.J. Psychiatry 37 (4), 399–406.

Pantelis, C., Yücel, M., Wood, S.J., Velakoulis, D., Sun, D., Berger, G., Stuart, G.W.,Yung, A., Phillips, L., McGorry, P.D., 2005. Structural brain imaging evidenceformultiple pathological processes at different stages of brain developmentin schizophrenia. Schizophr. Bull. 31 (3), 672–696.

Pratt, H.D., 2002. Neurodevelopmental issues in the assessment andtreatment of deficits in attention, cognition, and learning duringadolescence. Adolesc. Med. 13 (3), 579–598.

Rajarethinam, R., Miedler, J., DeQuardo, J., Smet, C.I., Brunberg, J., Kirbat, R.,Tandon, R., 2001. Prevalence of cavum septum pellucidum in schizo-phrenia studied with MRI. Schizophr. Res. 48 (2–3), 201–205.

Rajarethinam, R., Sohi, J., Arfken, C., Keshavan, M.S., 2008. No difference inthe prevalence of cavum septum pellucidum (CSP) between first-episode schizophrenia patients, offspring of schizophrenia patients andhealthy controls. Schizophr. Res. 103 (1–3), 22–25.

Rakic, P., Yakovlev, P.I., 1968. Development of the corpus callosum andcavum septi in man. J. Comp. Neurol. 132 (1), 45–72.

Rapoport, J.L., Addington, A.M., Frangou, S., Psych, M.R., 2005. Theneurodevelopmental model of schizophrenia: update 2005. Mol.Psychiatry 10 (5), 434–449.

Renier, D., Arnaud, E., Cinalli, G., Sebag,G., Zerah,M.,Marchac, D., 1996. Prognosisfor mental function in Apert's syndrome. J. Neurosurg. 85 (1), 66–72.

Sarwar, M., 1989. The septum pellucidum: normal and abnormal. Am. J.Neuroradiol. 10 (5), 989–1005.

Schaefer, G.B., Bodensteiner, J.B., Buehler, B.A., Lin, A., Cole, T.R., 1997. Theneuroimaging findings in Sotos syndrome. Am. J. Med. Genet. 68 (4),462–465.

Scott, T.F., Price, T.R., George, M.S., Brillman, J., Rothfus, W., 1993. Midlinecerebral malformations and schizophrenia. J. Neuropsychiatry Clin.Neurosci. 5 (3), 287–293.

Shaw, C.M., Alvord Jr., E.C., 1969. Cava septi pellucidi et vergae: their normaland pathogical states. Brain 92 (1), 213–223.

Shenton, M.E., Dickey, C.C., Frumin, M., McCarley, R.W., 2001. A review of MRIfindings in schizophrenia. Schizophr. Res. 49 (1–2), 1–52.

Shioiri, T., Oshitani, Y., Kato, T., Murashita, J., Hamakawa, H., Inubushi, T.,Nagata, T., Takahashi, S., 1996. Prevalence of cavum septum pellucidumdetected by MRI in patients with bipolar disorder, major depression andschizophrenia. Psychol. Med. 26 (2), 431–434.

Shunk, H., 1963. Congenital dilatations of the septi pellucidum. Radiology 81,610–618.

Swayze II, V.W., Johnson, V.P., Hanson, J.W., Piven, J., Sato, Y., Giedd, J.N.,Mosnik, D., Andreasen, N.C., 1997. Magnetic resonance imaging of brainanomalies in fetal alcohol syndrome. Pediatrics 99 (2), 232–240.

Takahashi, T., Suzuki, M., Hagino, H., Niu, L., Zhou, S.Y., Nakamura, K., Tanino,R., Kawasaki, Y., Seto, H., Kurachi, M., 2007. Prevalence of large cavumsepti pellucidi and its relation to the medial temporal lobe structures inschizophrenia spectrum. Prog. Neuropsychopharmacol. Biol. Psychiatry31 (6), 1235–1241.

Takahashi, T., Yung, A.R., Yücel,M.,Wood, S.J., Phillips, L.J., Harding, I.H., Soulsby,B., McGorry, P.D., Suzuki, M., Velakoulis, D., Pantelis, C., 2008. Prevalence oflarge cavum septi pellucidi in ultra high-risk individuals and patients withpsychotic disorders. Schizophr. Res. 105 (1–3), 236–244.

Thompson, S.G., Smith, T.C., Sharp, S.J., 1997. Investigating underlying risk asa source of heterogeneity in meta-analysis. Stat. Med. 16, 2741–2758.

Uematsu, M., Kaiya, H., 1989. Midsagittal cortical pathomorphology ofschizophrenia: a magnetic resonance imaging study. Psychiatry Res. 30(1), 11–20.

Weinberger, D.R., 1987. Implications of normal brain development for thepathogenesis of schizophrenia. Arch. Gen. Psychiatry 44 (7), 660–669.

Zhang, L., Ravdin, L.D., Relkin, N., Zimmerman, R.D., Jordan, B., Lathan, W.E.,Uluğ, A.M, 2003. Increased diffusion in the brain of professional boxers: apreclinical sign of traumatic brain injury? AJNR Am. J. Neuroradiol. 24(1), 52–57.