Meta-analysis

Beyond the association. Toxoplasma gondiiin schizophrenia, bipolar disorder, andaddiction: systematic review andmeta-analysis

Sutterland AL, Fond G, Kuin A, Koeter MWJ, Lutter R, van Gool T,Yolken R, Szoke A, Leboyer M, de Haan L. Beyond the association.Toxoplasma gondii in schizophrenia, bipolar disorder, and addiction:systematic review and meta-analysis.

Objective: To perform a meta-analysis on studies reporting prevalenceof Toxoplasma gondii (T. gondii) infection in any psychiatric disordercompared with healthy controls. Our secondary objective was toanalyze factors possibly moderating heterogeneity.Method: A systematic search was performed to identify studies intoT. gondii infection for all major psychiatric disorders versus healthycontrols. Methodological quality, publication bias, and possiblemoderators were assessed.Results: A total of 2866 citations were retrieved and 50 studiesfinally included. Significant odds ratios (ORs) with IgG antibodieswere found in schizophrenia (OR 1.81, P < 0.00001), bipolar disorder(OR 1.52, P = 0.02), obsessive–compulsive disorder (OR 3.4,P < 0.001), and addiction (OR 1.91, P < 0.00001), but not for majordepression (OR 1.21, P = 0.28). Exploration of the associationbetween T. gondii and schizophrenia yielded a significant effect ofseropositivity before onset and serointensity, but not IgM antibodiesor gender. The amplitude of the OR was influenced by region andgeneral seroprevalence. Moderators together accounted for 56% ofthe observed variance in study effects. After controlling forpublication bias, the adjusted OR (1.43) in schizophrenia remainedsignificant.Conclusion: These findings suggest that T. gondii infection is associatedwith several psychiatric disorders and that in schizophrenia reactivationof latent T. gondii infection may occur.

A. L. Sutterland1, G. Fond2,3,A. Kuin1, M. W. J. Koeter1,R. Lutter4, T. van Gool5,R. Yolken6, A. Szoke2,3,M. Leboyer2,3, L. de Haan11Department of Psychiatry, Academic Medical Centre(AMC), Amsterdam, the Netherlands, 2AP-HP, DHUPe-PSY, Pole de Psychiatrie et d’addictologie desHopitaux Universitaires H Mondor, INSERM U955, Eq 15Psychiatrie Translationnelle, Universit�e Paris Est-Cr�eteil,3Fondation Fondamental, Cr�eteil, France, 4Departmentsof Experimental Immunology and Respiratory Medicine,Academic Medical Centre (AMC), 5Department ofParasitology, Academic Medical Centre (AMC),Amsterdam, the Netherlands and 6Stanley NeurovirologyLaboratory, Johns Hopkins School of Medicine,Baltimore, MD, USA

Key words: bipolar disorder; meta-analysis;schizophrenia; substance abuse disorder; Toxoplasmagondii

Arjen L. Sutterland, Department of Psychiatry, AcademicMedical Centre, Amsterdam, Meibergdreef 5, 1100 DDAmsterdam, the Netherlands.E-mail: a.l.sutterland@amc.uva.nl

Accepted for publication March 17, 2015

Summations

• Toxoplasma gondii infection is not only associated with schizophrenia, but also with bipolar disorder,possibly addiction, and OCD, but not with major depression.

• The association increases with higher antibody titers, but decreases with higher baseline exposure toT. gondii of the population, although it remains significant.

• It appears that in schizophrenia reactivation of latent T. gondii infection may occur. Acute toxoplas-mosis does not seem to be implicated.

161

Acta Psychiatr Scand 2015: 132: 161–179 © 2015 John Wiley & Sons A/S. Published by John Wiley & Sons LtdAll rights reservedDOI: 10.1111/acps.12423

ACTA PSYCHIATRICA SCANDINAVICA

Considerations

• After controlling for evidence of publication bias, the association of Toxoplasma gondii infection withschizophrenia seems to be smaller than previously estimated and serointensity became an uncertainmoderator.

• Although region-specific effects were found, there are insufficient studies looking at a T. gondii strain-specific risk on psychiatric disorders.

• Effect of disease state could not be analyzed, which could account for the considerable residual heter-ogeneity found in our analysis.

Introduction

The intracellular parasite Toxoplasma gondii(T. gondii) is a common pathogen within the pop-ulation, with a varied worldwide prevalence (1). Itcan cause asymptomatic latent infections, exceptduring pregnancy or in immunocompromisedpatients (2). The parasite has a complex life cycle,whereby it needs to end up in the intestines of cats(felines) to be able to sexually reproduce (3). Inother hosts than cats (intermediate hosts, all warm-blooded animals), it invades the body after inges-tion, whereby it quickly duplicates until theimmune system of the host subdues the parasite(4). The parasite then transforms into a slow repli-cating stage and forms intracellular cysts in muscle,liver, and neuronal cells to reside in a latent stagepotentially for the rest of the host’s life (5). To endup in the cat’s intestine again, research has shownthat a latent T. gondii infection is able to modulatethe behavior of intermediate hosts like mice,increasing the chance to be caught by cats (6, 7).

Evidence has emerged that latent toxoplasmosisnot only causes behavioral changes in mice, butprobably also in humans (8, 9). Moreover, there issubstantial evidence suggesting an increased preva-lence of T. gondii infection in schizophrenia. Meta-analyses showed that the odds ratio (OR) of beingseropositive to T. gondii is 2.7 in patients withschizophrenia compared to healthy controls (10, 11).

Several hypotheses, assuming causality, havebeen postulated to explain this association. First,T. gondii may increase dopamine production,which has been shown in vitro by Prandovsky andcolleagues (12). Second, the type of immuneresponse in patients with schizophrenia to T. gon-dii infection has been postulated to lead to deleteri-ous effects (13, 14). An extensive literaturedemonstrating immunological abnormalities inpsychoses is available, including changed serumlevels of cytokines in schizophrenia patients (15),changes in enzymes of the immune-modulatory ky-nurenine pathway (16, 17), activated glia in vivo

(18), the immunosuppressive effects and/or anti-infectious effects of antipsychotics (19), activationof human endogenous retrovirus (20), whileimmuno-genetic markers associated with schizo-phrenia (21–23) also support this assumption.Major perturbation induced by T. gondii on gutmicrobiota is another issue recently recognized inschizophrenia (24).

However, dopamine and immunological distur-bances are not unique to schizophrenia. Both havebeen reported in other psychiatric disorders (25–29), including several immuno-genetic functionalmarkers known to reduce anti-infectious response,in particular with bipolar disorder (30–32). There-fore, T. gondii could be involved in other psychiat-ric disorders besides schizophrenia. Associationsof T. gondii infection with other psychiatric disor-ders, such as bipolar disorder (33) and obsessive–compulsive disorder (OCD) (34), have indeed beenreported but hitherto no analysis has looked sys-tematically at studies linking T. gondii infectionwith different psychiatric disorders.

Moreover, although an association between IgGantibodies against T. gondii and schizophrenia hasbeen found, considerable heterogeneity betweenstudies was found and debate exists on the natureof the association, whether the T. gondii infectiontruly precedes the disease and whether severity ofthe infection is relevant (35). For ideas about pop-ulation attributable risk, it would be informativeto what extent the exposure to T. gondii in thepopulation influences the magnitude of the associ-ation with psychiatric disorders (36). Furthermore,it has been postulated that gender could moderatethe association between T. gondii exposure andpsychiatric disorders (37, 38). Lastly, Arias et al.(11) found in their meta-analysis some evidence forpublication bias, which deserves further attention.

Aims of the study

The primary objective was to evaluate whetherToxoplasma gondii infection is more prevalent in

162

Sutterland et al.

one or more psychiatric disorders compared withhealthy controls by performing a systematic searchand meta-analysis. Our secondary objectives wasto explain heterogeneity between studies by look-ing at various potential moderators, to assess pub-lication bias and if applicable correct for that.

Material and methods

We followed the systematic review guidelines pro-vided by the PRISMA statement (39). We con-ducted a systematic search (Appendix 1) lookingfor studies analyzing T. gondii infection in any psy-chiatric disorder compared to healthy controls inMEDLINE, PsychInfo, and EMBASE until Sep-tember 19, 2013. Inclusion criteria were as follows:(i) original research papers; (ii) published in Eng-lish, French, Spanish, or Dutch; (iii) analysis ofT. gondii infection; (iv) a group with a distinct psy-chiatric disorder defined by a DSM-III (or ICD-9)or higher classification method; (v) a group ofhealthy control subjects; and (vi) use of one of thefollowing diagnostic assays: Sabin-Feldman dyetest, complement fixation (CF), immune hemagglu-tination (IHA), immune fluorescence (IF), orenzyme-linked immunosorbent assay (ELISA). Ifmultiple assays were used in a study, we utilizedthe results of the assay that showed the smallestdifference.

Additional clinical trials were sought in trial regis-ters and in references of all reviews on this topic.Screening of the search results by manuscript titlesand abstracts was performed independently by tworesearchers (AS and AK). As a final screening, thewhole manuscript was read to validate inclusion. Allauthors were approached for additional data if nec-essary on method information, possible moderators,or data to reach inclusion criteria. Also, all authorswere asked whether data of published work over-lapped (leading to exclusion of one of the papers)and whether they were aware of relevant (un)pub-lished work of themselves or colleagues.

Meta-analysis of eligible studies

For all studies, an odds ratio (OR) was calculated.ORs adjusted for confounders were used if avail-able. As considerable heterogeneity has beenshown in previous meta-analyses (10, 11), the ran-dom effects model was used in all analyses for con-servative purposes, which assumes there aredifferences among the effect sizes as a result of vari-ations in study characteristics. I-squared (I2) wascalculated to estimate the amount of heterogeneity.Meta-analytical calculations were carried out withComprehensive Meta-analysis Software 3.0 (40).

If 10 or more studies were available, additionalanalysis of moderators was calculated if available(study quality, gender, high antibody titers, meanage, disease phase-specific rates, seroprevalence ofhealthy control population, region where studywas performed, (un)published, age-adjustedrates).

All selected articles were screened on study qual-ity independently by two researchers (AS and GF)following Cochrane criteria of quality on case–control or cohort studies (41). If there was a differ-ence in quality score, the difference was discussed.If consensus could not be reached, a third opinionwas asked (LdH) for a final decision. The continu-ous scores were used for analysis of heterogeneity.

It is well known that there is an increased preva-lence of T. gondii infection with increasing age,which could account for some of the heterogeneity(42), so average age of patients and controls wereanalyzed as moderators, as well as age adjustment.

Higher antibody titers compared to controls hasbeen called serointensity, whereby numbers onhigh versus low, but still positive, antibody titerswere used to see if this influenced magnitude of theOR. Disease state (i.e., acutely ill versus remitted)-specific analysis was not possible, as most studiesdid not provide individual patient data.

Another possible moderator on the magnitudeof the association is strain type of the T. gondiiparasite prevalent in that region. As only one studyhas looked into that so far (43), we looked at anindirect measure with categorizing the studyaccording to its region and looking at the ampli-tude of the OR. Research has shown that there is adifference in strain prevalence in each region (44),whereby we identified the following regions: Northand South America, Europe, the Middle East,Africa, and Asia.

Examination of the residual heterogeneity wasanalyzed to estimate the impact of the moderatorsseparately. Where applicable meta-regressionanalysis (for continuous data) or subgroup analy-sis (for categorical data) was performed with themethods of moments and the mixed effects modelrespectively.

To assess to which extent moderators togetherinfluenced variance of the true effect, expressed bytau2, a regression analysis was performed investi-gating multiple moderators as covariates. By inves-tigating whether each moderator contributed tothe variation, a model was constructed explainingthe maximal amount of variance. The amount ofvariance caused by the moderators was expressedby R-squared.

Also, IgM antibodies against T. gondii in casesand controls were analyzed separately. If both

163

T. gondii & psychiatric disorders, a meta-analysis

groups had no IgM antibodies, no OR was calcu-lated, because the OR would then only be depen-dent on difference between numbers of cases andcontrols.

The potential for publication bias was assessedby examination of funnel plots and by the Egger’stest (which was considered significant if the one-sided P value was ≤0.10). If applicable, Duval andTweedie’s trim and fill method was used for a bet-ter estimation of the true OR (45).

Results

Following our systematic search 2866 studieswere found, from which 50 studies were finallyincluded (see flowchart in Fig. 1) (33, 34, 42, 43,46–87). Of these studies, four were unpublished(Hammer C, Ehrenreich H. 2014, personal com-munication, Torrey E. 2003, personal communi-cation, Gallo C. 2005, personal communication,and Dickerson F. 2005, personal communica-tion). Some studies reported seroprevalence ofT. gondii in more than one psychiatric disordercompared to healthy controls, resulting in uniquedatasets for schizophrenia (n = 42), bipolar disor-der (n = 11), major depression (n = 10), addiction(n = 4), and OCD (n = 2). No other psychiatricdisorders meeting inclusion criteria were found.A total of 12 009 cases and 71 441 healthy con-trols were analyzed. See Table 1 for a summaryof all studies found.

Schizophrenia

An overall OR of 1.81 (95% CI 1.51–2.16,P < 0.00001) was found, indicating a substantialassociation of schizophrenia with T. gondii infec-tion (Fig. 2). With the exception of Conejero-Goldberg et al. (56), who tried without success tofind DNA of T. gondii in postmortem brain sam-ples in patients and controls, all other studieslooked at IgG antibodies against T. gondii inserum and one in CSF (67). Evidence for publica-tion bias was shown by a significant regressionslope by the Egger’s test: P < 0.01. The funnel plotconfirmed this. Duval and Tweedie’s trim and fillwas used to correct for this publication bias, add-ing 11 imputed studies with a low OR showing anadjusted OR of 1.43 (95% CI 1.21–1.70).

A very high heterogeneity (I2 = 82%) was foundbetween studies. No clear pattern emerged wheninvestigating studies with high and low ORs.Impact of subsequent moderators was assessed.

Study quality

A meta-regression with the precise estimates ofquality did not show a significant effect on the OR(P = 0.53; data available upon request).

Mean age

When the studies reporting the mean age ofpatients or those of healthy controls (n = 27) were

2866 Citations from Medline, Embase and PsychInfo

141 Selected for detailed assessment of whole manuscript

(including reference screening)

55 Studies Selected for further analysis

2725 Excluded after title & abstract screening

43 Excluded following inclusion criteria

43 Reviews selected for reference screening:1 Additional Study added for further analysis

50 Studies Selected for Final analysis with data on:

42 Schizophrenia11 Bipolar Disorder10 Major Depression4 Addiction2 Obsessive Compulsive Disorder

After contact with authors for additional information:

2 Excluded due to overlapping cases in studies5 Excluded for not meeting diagnostic specificity2 Excluded for inadequate Healthy Control group1 Excluded for not providing Toxoprevalence data

11 Studies provided additional information4 Unpublished studies retrieved1 Published study added by author

Fig. 1. Flowchart of the study selectionprocess.

164

Sutterland et al.

Table 1. Summary of included studies

Authors Year CountryPatients (n) +diagnosis

Controls(n)

Toxodetermination

Age Pt(mean, years)

Age HC(mean, years)

1. Ahmad (46) 2010 Iran 80 SZ 99 ELISA 33.0* 33.82. Alipour (47) 2011 Iran 62 SZ 62 ELISA 37.5* 37.23. Alvarado-Esquivel (48) 2006 Mexico 38 SZ

6 BD3 MD25 AD

180 ELISA 43.3 (MD)55.5 (SZ)34 (BD)NA (AD)

42

4. Alvarado-Esquivel (49) 2011 Mexico 50 SZ 150 ELISA 45.1* 45.15. Bamne (50) 2012 USA 604 SZ 404 ELISA 38.5 41.06. Blomstrom (51) 2013 Sweden 47 SZ 124 ELISA NA* NA7. Brown (52) 2005 USA 63 SZ 123 IHA & SFDT 24.2* 25.28. Brown (53) 2013 USA 85 BD 170 IHA & SFDT NA* NA9. Buchy (54) 2002 Vietnam 300 AD 150 Indirect dye EIA 39.3 2810. Cetinkaya (55) 2007 Turkey 100 SZ

50 MD50 ELISA 37.3 (SZ)*

37.4 (MD)*37.2

11. Conejero-Goldberg (56) 2002 USA 14 SZ 26 PCR 39.9 48.712. Dickerson (unpub.) 2005 USA 405 SZ 170 ELISA NA NA13. Dickerson (57) 2014 USA 172 ROS 314 ELISA 24.4* 32.014. Dogruman (58) 2009 Turkey 77 SZ

11 ROS88 ELISA 38.5 (Total)

40.3 (SZ)25.9 (ROS)*

29.7 (total)26.2 (matched ROS)

15. El-Sayed (59) 2012 Egypt 60 SZ30 MD

20 ELISA 38.0 (SZ)*37.0 (MD)*

37.8

16. Emelia (60) 2012 Malaysia 144 SZ 144 ELISA NA NA17. Gallo (unpub.) Unpub. Peru 119 SZ 119 ELISA NA (18–65)* NA18. Gutierrez (61) 1996 Spain 647 AD 6454 ELISA 20–35 years 20–35 years19. Hamdani (33) 2013 France 141 BD

91 SZ171 ELISA 44.8 (BD)*

NA (SZ)37.8

20. Hamidinejat (62) 2010 Iran 98 SZ46 MD

48 ELISA 33 (median) 33 (median)

21. Hammer (unpub.) Unpub. Germany 876 SZ60 BD88 MD

1271 ELISA NA* NA

22. Hinze-Selch (42) 2007 Germany 277 SZ464 MD

214 IF 37,4 (SZ)*46 (MD)*

38.9

23. Horacek (63) 2012 Czech Republic 44 SZ 56 ELISA 30.8 27.924. Khademvatan (64) 2013 Iran 117 BD 200 ELISA 33.9 33.825. Khademvatan (65) 2014 Iran 100 SZ 95 ELISA NA NA26. Krause (66) 2010 Germany 31 SZ 30 ELISA 36.7 33.727. Leweke (67) 2004 Germany 113 SZ 73 ELISA 29.1 drugnaive

33.7 drugfree30.5 treated

28

28. Li (68) 2013 USA 855 SZ 1165 ELISA NA* NA29. Liu (69) 2011 China 477 SZ 210 ELISA 29.9 28.130. Martinez (70) 1995 Germany 37 AD 145 Autopsy; cerebral

toxoplasmosisNA NA (overall 41.4)

31. Miman (34) 2010 Turkey 42 OCD 100 ELISA & IFA 34.1 (median) 38.1 (median)32. Mortensen (71) 2007 Denmark 71 SZ 684 ELISA NA* NA33. Mortensen (72) 2011 Denmark 127 BD 127 ELISA NA* NA34. Nascimento (73) 2012 Brazil 41 SZ 95 ELISA 32.7 38.235. Niebuhr (74) 2008 USA 180 SZ 532 ELISA NA* NA36. Park (75) 2012 South Korea 96 SZ 50 ELISA and IHA 46.1* 44.837. Pearce (77) 2012 USA 41 BD

515 MD7440 ELISA NA* NA

38. Pearce (76) 2013 USA 183 SZ 137 ELISA 44.9 37.739. Pedersen (78) 2011 Denmark 80 SZ 44436 ELISA NA NA40. Saraei-Sahnesaraei (79) 2009 Iran 103 SZ 114 ELISA NA NA41. Tamer (80) 2008 Turkey 40 SZ 37 ELISA 33 (median) NA

42. Tanyuksel (81) 2010 Turkey 73 ROS 40 ELISA & SFDT 23.4 30.343. Tedla (82) 2011 Ethiopia 214 SZ

171 BD71 ELISA 32.7 (SZ)*

31.6 (BD)*30.4

44. Torrey (unpub.) Unpub. Ireland 55 SZ10 BD

20 ELISA NA* NA

45. Wang (83) 2006 China 600 ROS 200 ELISA 22.6* 22.546. Xiao (43) 2009 USA 120 SZ 346 ELISA NA* NA

165

T. gondii & psychiatric disorders, a meta-analysis

Table 1. (Continued)

Authors Year CountryPatients (n) +diagnosis

Controls(n)

Toxodetermination

Age Pt(mean, years)

Age HC(mean, years)

47. Xiao (84) 2010 China 291 SZ49 BD29 MD12 OCD

2634 ELISA NA NA

48. Yolken (85) 2001 Germany 38 ROS 27 EIA 27 median* NA49. Yuksel (86) 2010 Turkey 300 SZ 150 ELISA & SFDT 42.6* 42.650. Zhou (87) 2011 China 1301 SZ

419 ROS356 BD237 MD

1552 ELISA 30.8 (SZ)*32.0 (BD)37.0 (MD)24.4 (ROS)

30.0

SZ, schizophrenia; BD, bipolar disorder; MD, major depression; AD, addiction disorder; ROS, recent onset schizophrenia; OCD, obsessive-compulsive disorder; SFDT, Sabin-Feld-man dye test; IHA, immune hemagglutination; IF, immune fluorescence; EILSA, enzyme-linked immunosorbent assay; NA, not available; Unpub., unpublished data.*Age-adjusted analysis used.

Fig. 2. Schizophrenia and prevalence of IgG antibodies against Toxoplasma gondii. Analysis with random effects model. unpub.,unpublished.

166

Sutterland et al.

analyzed with a metaregression analysis, no effecton OR was found (both P = 0.78). There was nosignificant difference between studies that usedage-adjusted (and/or age-matched) analysis andthose that did not (P = 0.45), heterogeneityremained high.

Disease phase

Another potential moderator is phase-specificORs: Seven prospective studies reported an associ-ation of T. gondii infection before the onset ofschizophrenia (43, 51, 52, 68, 71, 74, 78). Heteroge-neity was moderate (I2 = 42%). Overall, a signifi-cant OR of 1.30 (95% CI 1.05–1.61, P = 0.017)was found, indicating an increased risk of T. gondiiinfection before developing schizophrenia. Fur-thermore, it was possible to discern 10 studies withrecent onset schizophrenia (defined as <2 years ofillness duration). Heterogeneity was moderate(I2 = 43%). A substantial OR was found of 2.18(95% CI 1.58–3.01, P < 0.00001). The other stud-ies (n = 28) mainly investigated chronic schizo-phrenia cases, showed very high heterogeneity(I2 = 83%), and rendered an OR of 1.88 (95% CI1.46–2.42, P < 0.000001). There was a significantdifference between groups (P = 0.01).

Serointensity

Apart from associations on seroprevalencebetween cases and controls, several studiesreported a higher serointensity as well (42, 88).Most studies looked at serointensity in schizophre-nia (n = 13), and just one in major depression (42)

and bipolar disorder (33). Consequently, effect ofserointensity was analyzed in schizophrenia only.Studies used different cut-offs for high antibodytiters at 60 IU/ml (60), 100 IU/ml (52), 128 IU/ml(42), or 250 IU/ml (81). Other authors set the cut-off at the 75th or 90th percentile of seropositivecontrol subjects to distinguish subjects with hightiters (71, 78). We calculated ORs using above-mentioned studies and six studies that did not pub-lish their serointensity rates (33, 49, 50) (Torrey E.2003, personal communication, Gallo C. 2005, per-sonal communication), whereby we used the 75thand 90th percentile cut-offs of the provided data.

Using the 75th percentile in unpublished studies,a significant OR was found of 1.85 (95% CI 1.11–3.10, P = 0.019). When the 90th percentile ofunpublished studies was used, the OR was 2.27(95% CI 1.35–3.47, P = 0.001). Heterogeneityremained significant (I2 = 66%). One study had anOR of 17 (46), which made it a clear outlier. Whenexcluding this study, the OR was 1.79 (95% CI1.29–2.47, P < 0.001) indicating a significant asso-ciation with high antibody titers and schizophreniaand heterogeneity statistically became low(I2 = 0%), although the forest plot still showedconsiderable variation in ORs.

There was no evidence of publication bias withEgger’s test (one-sided P = 0.48). However, whenstudies were grouped into published and unpub-lished data a significant group effect was found(Q = 11.2, P = 0.001), with unpublished studiesshowing no effect (OR 0.79) and published studiesa significant effect (OR 2.17) of serointensity(Fig. 3). Using the 90th percentile in the unpub-lished studies, the difference was less pronounced

Fig. 3. Schizophrenia and association with high IgG antibodies to Toxoplasma gondii. Analysis with random effects model.0 = Unpublished; 1 = Published.

167

T. gondii & psychiatric disorders, a meta-analysis

and trend significant (OR 1.11 vs. 2.17, Q = 3.28,P = 0.07).

Gender

In 21 studies, gender-specific seroprevalence rateswere provided (46–51, 55, 58, 63, 64, 72, 77, 78, 86,87), fourteen studies on schizophrenia, four onbipolar disorder, and three on major depression.

No significant gender effect on the OR ofT. gondii infection in schizophrenia was found(P = 0.90) and heterogeneity remained high.

Seroprevalence of the control population

In all studies but two (57, 68), the percentage ofhealthy controls with T. gondii antibodies wasgiven, which showed a broad range (0–87%) withan average seroprevalence of 24%. The effect of se-roprevalence on ORs was studied by metaregres-sion. In schizophrenia, a significant decline of theOR was observed (OR intercept = 2.51,slope = �0.01, P = 0.036) with increasing seropre-valence. When splitted in groups using a cut-off of24%, a significant difference between groups wasfound (Q-value between = 5.5, P = 0.02) withstudies with ‘low’ general seroprevalence showingan OR of 2.34 (95% CI 1.74–3.13) and studies withhigh general seroprevalence 1.49 (95% CI 1.19–1.88), remaining significant nonetheless. Heteroge-neity remained high in both categories.

Region specificity

For every region/continent, a significant OR wasfound with high ORs in Africa, South America,the Middle East and Asia and more modest ORs inEurope and North America (Table 2). Asignificant between group effect was found(Q = 17.3, P = 0.004). The analysis was not feasi-ble for the other psychiatric disorders separately,

but when all psychiatric disorders were combined,yielding 69 datasets, a similar pattern was seen.

Analysis of covariates together

When all moderators were entered in a regressionmodel as covariates of the true effect and, they ableto maximally explain 56% of the variance(s2 = 0.27, R2 = 0.56) using data of 40 studies. Theoptimal model contained the following modera-tors: seroprevalence of the control population,region, and age adjustment. When a regression lineof this model was entered in a scatterplot with se-roprevalence of the control population on the x-axis, it showed that with 0–5% of the populationaffected with T. gondii (taking into account ageadjustment and region) the OR would be around3.3. Conversely, the OR would on averageapproach 1.0 when the seroprevalence of the popu-lation reached 40% (Fig. 4).

IgM antibodies

IgM antibodies against T. gondii often indicate anacute infection (89) and this was explored in 22studies, most of them (n = 19) in schizophrenia.Because four studies did not find IgM antibodiesin cases and controls (52, 58, 71, 73), fifteen studiesremained for analysis (46, 48, 49, 55, 62, 65, 66, 75,79-81, 83, 85, 87) (Gallo C. 2005, personal commu-nication). Except for some studies in China andIran (46, 79, 83, 87), most studies found very fewsubjects with IgM antibodies. Consequently, theconfidence intervals of the latter studies were large,giving a low heterogeneity statistic, although theOR ranged from 0.2 to 15.5. A non-significant ORof 1.24 (95% CI 0.98–1.57, P = 0.08) was found.Indication of publication bias was found (Egger’stest P = 0.04) and the Duval’s trim and fill methodled to a new estimate of 1.08 (95% CI 0.81–1.46).Five studies evaluated IgM antibodies in recentonset schizophrenia (81, 83, 85, 87), showing anon-significant OR of 1.58 (95% CI 0.80–3.11,P = 0.19). Here, publication bias was suggested aswell (Egger’s test P = 0.05) and the Duval andTweedie’s trim and fill method showed an adjustedOR of 0.93 (95% CI 0.45–1.91).

Bipolar disorder

A significant OR of 1.52 (95% CI 1.06–2.18,P = 0.02) was found, indicating a significant asso-ciation of T. gondii infection with bipolar disorder(Fig. 5). Heterogeneity was high (I2 = 67%). Nomoderator significantly impacted the OR sepa-rately (data available upon request). However, the

Table 2. Region-specific associations of Toxoplasma gondii with schizophrenia

RegionStudies(n)

Pointestimate

Lowerlimit

Upperlimit P-value

Africa 2 3.74 1.70 8.23 0.001South America 4 2.48 1.71 3.58 0.000Asia 6 2.06 1.13 3.77 0.018Middle East 11 2.04 1.30 3.22 0.002Europe 12 1.53 1.11 2.10 0.010North America 7 1.26 1.03 1.54 0.022Overall 41 1.60 1.39 1.84 0.000Total between 0.004

Studies grouped by region investigating IgG antibodies to T. gondii in schizophrenia.Analysis performed with the mixed effects analysis.

168

Sutterland et al.

regression analysis using several moderators ascovariates, the moderator (un)published data andage adjustment explained 82% of the variance intrue effect (R = 0.82). Nevertheless, the Egger’stest was not significant, indicating no significantpublication bias. The funnel plot showed an evenlydistribution (data available on request).

Major depression

A non-significant OR of 1.21 (95% CI 0.86–1.70,P = 0.28) was found, indicating no overall differ-ence of T. gondii infection between healthy con-trols and subjects with depression. Theheterogeneity was high (I2 = 64%). The heteroge-neity is also reflected in the studies evaluating largecohorts, where Zhou and colleagues in China,found a high OR of 2.41, while Hammer and Eh-renreich (personal communication) and Pearceet al. found a low OR of 0.66 and 0.8 respectively(77, 87). This difference could be due to the factthat Zhou analyzed subjects with depression with

psychotic features (personal communication) andthat these data were not yet adjusted for age andgender, as they were with Hammer and Ehrenreich(personal communication) and Pearce et al. With-out the study of Zhou et al., the heterogeneity wasconsiderably lower (I2 = 16%). The OR attenuatedto 1.01 (95% CI 0.80–1.28, P = 0.95). This wasconfirmed by a significant effect of studies thatadjusted (or matched) for age in studies looking atmajor depression (Q = 5.0, P = 0.03) and thosethat did not, the latter showing a higher OR.

Also, meta-regression with the estimates of qual-ity was analyzed with a significant decline of theOR (P < 0.001) with increasing quality (data avail-able upon request), mainly due to the above-men-tioned studies. Not surprisingly, age adjustment asa covariate explained 73% of the variance. How-ever, quality of studies alone explained all the vari-ance of the true effect (R2 = 1.00).

The Egger’s test was not significant (P = 0.36),and the funnel plot showed an evenly distribution(data available on request).

Fig. 4. Meta-regression analysis of the odds ratio of schizophrenia and prevalence of IgG antibodies against Toxoplasma gondii.Meta-regression model whereby the log odds ratio is plotted against the prevalence of antibodies against T. gondii in the healthy con-trols in each study (seroprevalence), taking into account the covariate region and age adjustment of study. Y = intercept of log oddsratio if seroprevalence is zero. Age adjustment 0 = not adjusted (reference variable) and 1 = adjusted. The reference variable ofregion is Europe.

Fig. 5. Bipolar disorder and prevalenceof IgG antibodies against Toxoplasmagondii. Analysis with random effectsmodel. unpub., unpublished.

169

T. gondii & psychiatric disorders, a meta-analysis

Obsessive–compulsive disorder

Two studies reported on OCD compared with con-trols (34, 84), both finding an elevated OR, with anoverall OR of 3.4 (95% CI 1.73–6.68, P = 0.0004).

Addiction

Four studies were included, from which threereported seroprevalence rates of T. gondii in publichealth studies evaluating healthy subjects as well asdrug abusing patients (54, 61, 70). Although thestudies were not designed as case–control studieswith a hypothesis about a possible relationshipbetween T. gondii infection and addiction, it waspossible to discern ‘healthy’ subjects [e.g., pregnantmothers (54, 61)] and addicted subjects. Most iden-tified studies concerned intravenous drug usingpatients (heroin), which was not operationalizedfurther following diagnostic criteria. Nevertheless,the chance that these patients would not have anaddiction following diagnostic criteria was consid-ered to be so low that it was decided to includethese studies. A significant OR of 1.91 (95% CI1.49–2.44, P < 0.00001) was found, indicating anincreased prevalence of T. gondii infection in (her-oin) addiction (Fig. 6). The heterogeneity of thisfew studies was low (I2 = 12%).

Egger’s test was trend significant (one tailedP = 0.10) indicating possible publication bias. Thefunnel plot showed that the OR could even beunderestimated (data available upon request).

Discussion

This is the first study that systematically analyzedthe association of T. gondii within different psychi-atric disorders. It showed that T. gondii infectionis not only significantly associated with schizophre-nia, but also with bipolar disorder, addiction, andOCD, but not with major depression.

In schizophrenia, the evidence of an associationwith T. gondii is overwhelming, despite evidence ofpublication bias. When corrected for this bias anOR of 1.43 (1.21–1.70) remained, which is lowerthan reported in previous meta-analyses (10, 11),

who did not correct for potential publication bias.Torrey et al. found far more publications in otherlanguages, but we chose to conduct a thoroughand replicable systematic review following thePRISMA statement (39), which is consideredmethodologically superior.

Heterogeneity was generally high in our analy-sis, which is a problem when trying to draw con-clusions based on the overall OR.

In schizophrenia, it was possible to assess sev-eral moderators, giving us more information onthe nature of the variance of effect sizes betweenstudies. Importantly, phase-specific ORs showedthat a T. gondii infection seems to precede the dis-order and that the association gets stronger duringthe illness. Second, serointensity was more preva-lent in schizophrenia cases compared to controls,pointing at either a more active infection or astronger immune response. These findings (tempo-rality and biological gradient) give an indicationthat T. gondii infection (and/or its immuneresponse against it) may be a causal factor in theetiology of developing schizophrenia.

On the other hand, some pre-onset studies werebirth-cohort studies (43, 51, 52, 90), which showantibodies of the mother in a newborn child (as achild is not yet able to form his own antibodies orthis is masked by the antibodies transmitted by themother) and are thereby indirect indications of aT. gondii infection before onset of the disorder.Transplacental transmission can occur with anacute T. gondii infection during pregnancy, but isconsidered to be a rare event (91). The other threestudies could be considered true pre-onset studies,as they measured antibodies in adults who laterdeveloped schizophrenia. These did not find a sig-nificant association overall (data not shown) (68,74, 78). The increased prevalence of antibodies inmothers of offspring could at one hand beexplained by increased immune activation duringpregnancy, which is a known risk factor for laterdevelopment of schizophrenia (92–94). However,Mortensen et al. (71) did not find increased generalIgG levels to be associated with an increased riskfor later development of schizophrenia in contrastwith IgG antibodies against T. gondii. Alterna-

Fig. 6. Addiction and prevalence ofIgG antibodies against Toxoplasmagondii. Analysis with random effectsmodel. unpub., unpublished.

170

Sutterland et al.

tively, a reactivation of a latent T. gondii infectionduring pregnancy occurs, possibly affecting theunborn child. That serointensity of these four stud-ies increased the OR to 1.91 (95% CI 1.20–3.04)could be seen as confirmation of this hypothesis.

The fact that there was a significant differencebetween published and unpublished studies on ser-ointensity, also raises doubt on the true effect, eventhough overall a significant effect was shown and ahigher cut-off in the unpublished data appeared toincrease the OR. Therefore, we argue that uncer-tainty concerning the support for a causal etiologystill remains.

Gender on the other hand did not prove to be amoderator in T. gondii and psychiatric disorders inthis meta-analysis. This finding does not seem to fitwell with animal and human research showing gen-der-specific effects of T. gondii on behavior (95–97). Different neurobiological effects have beenreported as well (38, 95), both in acute and chronicinfection. It might be that gender differential ani-mal behavior changes are not translatable into dif-ferences in psychiatric categories. Alternatively,Lindova and colleagues have shown that the differ-ences in behavior alterations caused by T. gondiibetween men and women are probably causedmore by gender differences in response to stressitself, than by different effects of the parasite in thebrain of a male or a female (98). This supports ourfinding that there is no gender specificity.

Seroprevalence of T. gondii infection in the con-trol population was a relevant moderator for themagnitude of the OR in schizophrenia, which is thusalso responsible for some of the heterogeneity. Per-haps this is not surprising, because when the expo-sure of a factor in the control population is higher, itis more difficult to find a significant difference. Sta-tistically, the finding does show us that T. gondiidoes not cause an absolute increase in risk, becausethen there would not be this ceiling effect. Wehypothesize that there is a limited population at risk,by which T. gondii infection plays a role.

Nonetheless, even in countries with high sero-prevalence of T. gondii, significant differences havebeen found and the group of studies with a higherthan average (24%) seroprevalence still showed asignificantly increased OR. As latent T. gondiiinfection is presumed to persist life long in the hostand IgG levels likely decrease below detection lev-els with time. The enhanced presence of antibodiestherefore may be indicative of re-activation ofT. gondii and/or of an enhanced immune responsein these patients with high higher seroprevalence(81, 99).

Finally, we showed significant variation of theOR according to region, which could be seen as an

indication of strain-specific effects of T. gondii onthe human brain (43, 44, 100). Regional differencesin strain prevalence have been reported (44, 100),and strain-specific effects have been found in affec-tive psychosis as well as on anxiety and depressionlevels (43, 101). This strain hypothesis could onlybe indirectly analyzed in schizophrenia specificallyin our study. Interestingly, the type II and IIIstrains, which were not associated with psychiatricdisorders in these studies, are more commonlyfound in North America and Europe where moremodest ORs were found.

A meta-regression analysis showed that severalmoderators together explained about 56% of thevariance of the true effect in schizophrenia. Thisshowed that regional differences, age adjustment instudies and especially seroprevalence of T. gondiiin the population matters in the amplitude of theOR. Nevertheless, considerable variance remains,urging us to keep evaluating other factors poten-tially responsible for this variation. Disease statewould be a interesting factor, as state-specificimmune abnormalities have been reported (27, 57).Also studies have reported that Rhesus factorcould have a role in the immunity against T. gondiiinfection and can protect against possible deleteri-ous cognitive effects (102), but so far no study hasreported that this may be relevant in the suscepti-bility for psychiatric disorders.

Bipolar disorder showed a similar associationwith T. gondii infection with a significant OR of1.52 (1.06–2.19), which shows that T. gondii infec-tion is not only associated with schizophrenia.There was no evidence of publication bias. How-ever, a model with age adjustment and whether ornot studies were published accounted for themajority of the variance in true effect. This waspartly caused by a large negative study of Hammerand Ehrenreich (personal communication) usingdata from the GRAS study (103), who found sig-nificantly negative associations of T. gondii infec-tion with bipolar disorder, but also schizophreniaand major depression after controlling for age andgender. Another large study of Hamdani et al.(33), however, did show a significant OR after thesame corrections, whereby general seroprevalenceof the control population was high as well.

Factors possibly explaining the remaining heter-ogeneity could be the strain hypothesis. Xiao et al.(43) has shown that T. gondii strain type I is espe-cially relevant in affective psychotic disorders.However, due to a relatively limited amount ofstudies, regional differences could not be checkedin this analysis.

Major depression was not related to IgGantibodies against T. gondii, even though a

171

T. gondii & psychiatric disorders, a meta-analysis

toxoplasmosis infection has frequently been associ-ated with fatigue (104). This implies that T. gondiidoes not have a general association with psychiat-ric disorders, but that there is a degree of specific-ity.

The highly significant association with addictionshould be interpreted with caution, as the studieson addiction were not case–control studies, andtherefore, there was no adequate matching of casesand controls. The quality of the studies was there-fore low. Another confounding factor could bethat the studies reported prevalence of T. gondiiinfection in intravenous drug abusers. Concerningthe lifestyle of these patients, one could argue thata T. gondii infection is a consequence rather thanrisk factor in this type of addiction. On the otherhand, following animal research on changes ofrisk-taking behavior in mice with T. gondii infec-tion (3, 7) and human studies on suicide attempts(105) and traffic accidents (9, 106), we hypothesizethat T. gondii infection may modulate risk/aversivebehavior, which may increase the risk of addictionas a result (107). We consider this association to beintriguing and think it deserves further investiga-tion. If more evidence arises on this association,one could argue that there should be a correctionfor addiction in investigating T. gondii infection inother psychiatric disorders versus healthy controls,as addiction is a known risk factor in, for example,schizophrenia (108).

The increased OR of T. gondii antibodies in OCDmust be interpreted with caution as well, as it isbased on merely two average-sized studies (34, 84).For example, Yuksel et al. (86) also evaluated OCDsubjects, but unfortunately not separated depressionand anxiety disorders, making it unsuitable for fur-ther analysis. Nevertheless, this group did not showa different seroprevalence from healthy controls. Infavor of the possible role in anxiety disorders are therecent studies finding a significant association ofT. gondii with Generalized Anxiety Disorder (109).Clearly, more studies are needed in this disorderconsidering T. gondii infection.

No studies were found reporting an associationof T. gondii infection with other psychiatric disor-ders such as autism, eating disorders, and atten-tion-deficit disorder (ADHD). Autism has beenpostulated earlier to be involved in congenitaltoxoplasmosis (110, 111), where earlier studies onmental retardation (112) in humans could supportthis assumption. Neurocognitive effects of T. gon-dii infection in animals and the increased suicidali-ty and traffic accidents in humans also point toattentional deficits and increased impulsivity (7).Together these findings could suggest a possiblelink with ADHD, personality disorders, and

T. gondii infection, whereby an increased preva-lence of personality disorders has been reported inseropositive patients with psychiatric disorders(113).

IgM antibodies, a marker of acute/recent infec-tion, did not seem to be associated with schizo-phrenia. This is in sharp contrast with a recentmeta-analysis of Monroe and colleagues whoreported an OR of 1.6 of IgM antibodies againstT. gondii with psychosis, whereby this OR washigher for recent onset/acute psychosis (114). Inour analysis, we also saw a modest, but non-signifi-cant, association with recent onset schizophrenia.A possible explanation for this difference is that wealso included data that had not been published(87) (Gallo C 2003, personal communication). Onestudy found by Monroe et al. was not found by us(115), and another was excluded for not fitting ourinclusion criteria (116). We did not include thestudy of Liu et al. because the status of the IgMseropositivity of the healthy controls did notbecome clear to us. However, according to Mon-roe and colleagues, Liu found a staggering OR of48.6 (95% CI 3.0–792.3) making it a clear outlierfrom other studies. When our analysis wasrepeated with the two other studies (without Liuet al.), our finding remained the same. We there-fore think that our finding is closer to the trueeffect, especially because we did find indications ofpublications bias and corrected for that using theDuval and Tweedie’s trim and fill method. Eye-balling the funnel plot of Monroe and colleagues,one could argue that this should have been carriedout in their analysis as well.

This implies that acute infection does not seemto be more prevalent in schizophrenia, pointing ata chronic latent infection, which started beforeonset of the diagnosis of schizophrenia and is pos-sibly reactivated if we consider the serointensityfinding valid. Animal research has also showedthat a chronic infection seems to alter the host’sbehavior (3, 117, 118). This could either point to adirect pathophysiology caused by the parasite, oralternatively that the T. gondii infection is reacti-vated by alterations in the immune system or byanother infection.

Considering the neuroinflammation hypothesis,microglial activation has been shown in vivo inschizophrenia (18) and the possible relevance ofmicroglia in the pathophysiology of schizophreniais increasingly recognized (119–123). Microgliaplay a role in controlling a T. gondii infection inthe brain (124), possibly through the kynureninepathway (125), whereby differences of the micro-glial immune response to different strains ofT. gondii have been found (126). Alternatively, IL-6

172

Sutterland et al.

production of macrophages reverts the inhibitionof T. gondii replication caused by astrocytes andmicroglial cells and may be involved in the mecha-nism of reactivation of the infection in patientswith AIDS (127–129). IL-6 has repeatedly beenfound to be raised in schizophrenia patients (15).

Another hypothesis links CD8 T-cell downregu-lation with the mental effects of toxoplasmosis(130), which can be caused by the kynurenine path-way. Also a considerable amount of schizophreniasusceptibility genes are implicated in the life cycleof pathogens including T. gondii (13), wherebythese genes have a role in immunity, but also regu-late binding to the integrin system which couldlead to hypofunction of NMDA-receptors (13), aswell as influencing AKT1 signaling (preventinghost cell death, but also regulation of dopamine-dependent behavior) (131). Another infectionmight reactivate a T. gondii infection. In supportof this hypothesis, several studies have found anincreasing risk with increased number of neuro-tropic infections (66, 68).

From our analysis, the population seropreva-lence and regional differences stood out as relevantmoderators in the observed variance of reportedeffects. Apart from the strain hypothesis, theimmune protection against a T. gondii infection ina population where this parasite is abundant mightbe different to regions where the population hasless parasite-pressure.

The impact of T. gondii infection on increasingthe dopamine levels was demonstrated in vitro andin mice (12, 132). It is therefore tempting to look atthe dopamine hypothesis, as this hypothesis isimplicated in schizophrenia (133), bipolar disorder(134), OCD (135), and addiction (136). Also withinschizophrenia higher severity of positive symptomsin T. gondii-positive patients has been reported (83,137). Schizophrenia, bipolar disorder, and OCDcan be treated effectively with dopamine-blockingagents (133, 138, 139). Addiction, however, can beworsened by these medicines (140) and low dopa-mine release has been associated with this disorder(141). Clearly, more research is needed in this areafocusing on the effect of toxoplasmic infection onthe risk of addiction, symptom dimensions, anddopamine neurotransmission in vivo is needed.

Toxoplasma gondii infection as a risk factor maybe independent from other risk factors for psycho-sis and addiction: not more often found in men,not more often in urban environments (78),although possibly more often in mothers with apsychiatric history (78). It should be noted in thisrespect that a stronger immune reaction to anothermicrobiotic environment could be a risk factor inmigration and psychiatric illness (142) although a

study of Blomstrom et al. (51) did not find a rela-tion between migration and infections in schizo-phrenia.

If the association should prove to be causal,these findings may give further clues about howT. gondii infection can possibly modulate the riskof specific psychiatric disorders. As these disordersare classifications based on symptom profiles, thechallenge will be to translate this to underlyingneurobiological mechanisms (143). This in turncan give an indication concerning the mechanismby which T. gondii could increase the vulnerabilityfor several psychiatric disorders. To discern a sub-group, where immune alterations and neurotropicinfections are relevant can help us delivering tar-geted treatment (119, 144, 145). We suggest that itcould be valuable to see raised IgG prevalenceagainst T. gondii as one of the expressions of thissubgroup.

Limitations

Unfortunately, current available studies did notallow us to analyze disease states (i.e., acutely illversus remitted or chronically ill). There are someindications in the literature that this might be rele-vant (57, 146). Staging in psychiatric disorderscould show to be of interest in the field of neuroim-munology and could give an indication how (reac-tion to) a T. gondii infection behaves during thedevelopment of a psychiatric disorder. It couldalso provide insights into the moderators like sero-intensity and general T. gondii prevalence in a pop-ulation. The same applies for identifyingsubgroups in these disorders.

Furthermore, indications of publication biaswere found in studies reporting an association withschizophrenia and on the moderator serointensityand IgM antibodies. Publication bias remains amajor challenge in science (147), and this meta-analysis confirms that. However, we carefully cor-rected for this bias in our analysis and thoroughlysearched for unpublished studies to avoid publica-tion bias as much as possible. Compared to previ-ous meta-analyses, our method rendered moremodest ORs, which could be seen as some externalvalidation of the reliability of our estimations (10,11, 114).

A clear difference with the first meta-analysis onschizophrenia and T. gondii infection is that wedid not include studies written in Chinese or East-ern European languages (10). Excluding studiesbased on language is considered to be methodolog-ically less strong. Nevertheless, we wanted to per-form a replicable systematic search, which isconsidered the gold standard. In accordance with

173

T. gondii & psychiatric disorders, a meta-analysis

Torrey et al., our association has been shown to beconsistent in almost every continent of the world(with the exception of Oceania, where no studieshave been reported).

Studying antibodies against T. gondii remainsan indirect measure of infection and does not givemuch information about the activity or location ofT. gondii parasites. The only study trying to findT. gondii parasites in the brain failed to do so incases and controls (56). Considering the averageold-age of the subjects in that study, this could beseen as an unfortunate negative finding in a coun-try with a general seroprevalence of 7–14% (148)or as evidence that we have not discovered theright methods yet to show a latent T. gondii infec-tion in the brain.

Despite our careful analysis of possible modera-tors, it was clear that significant heterogeneityremained. With the software we used, it was possi-ble to perform a regression analysis with differentmoderators, which indicated that 56% of the vari-ance in OR found in schizophrenia could beaccounted for by the moderators, leaving consider-able variance unexplained. This urges to continueto explore additional moderators of the associa-tion.

In conclusion, our meta-analysis suggests that aT. gondii infection is not only associated withschizophrenia, but also with bipolar disorder andpossibly addiction and OCD. Furthermore, onanalysing heterogeneity by several moderators, itappears that in schizophrenia a reactivation of alatent T. gondii infection occurs and that generalseroprevalence and in which region the populationresides matters. Acute toxoplasmosis does notseem to be implicated. The strength of the associa-tion increases with higher antibody titers, butdecreases with higher baseline exposure to T. gon-dii of the population. T. gondii infection does notseem to show a gender differential effect. Aftercareful controlling for indications of publicationbias, the association seems to be smaller than pre-viously estimated and serointensity was an uncer-tain moderator. Pathophysiologically, immuneresponse to the infection and direct neurotransmit-ter effects by the T. gondii parasite could beresponsible for these associations. Further effortson identifying a subgroup of psychiatric disorders,where immune alterations are implicated, areneeded to improve our treatment practices.

Acknowledgements

We thank our clinical librarian Joost Daams at the Universityof Amsterdam for his help on designing the systematic searchin the different databases. We also would like to express our

gratitude to the authors Carla Gallo, E. Fuller Torrey,Christian Hammer, Hannelore Ehrenreich, and Faith Dicker-son for sharing their unpublished data, as well as all theauthors who shared extra data on their published work (47, 48,50, 51, 58, 68, 72, 74, 75, 77, 87) making better analysis ofmoderators possible.

Declaration of interest

This work was supported by Inserm, Assistance publique–Hopitaux de Paris, RTRS Sante Mentale (Fondation Fonda-mental) and by Agence Nationale Pour la Recherche (ANR:NEURO 2009, V.I.P. project). This work was supported (inpart) by the Investissements d’Avenir program managed by theANR under reference ANR-11-IDEX-0004-02 and ANR-10-COHO-10-01. We declare no potential conflict of interestrelevant to this article.

References

1. Yolken RH, Dickerson FB, Fuller Torrey E. Toxo-plasma and schizophrenia. Parasite Immunol2009;31:706–715.

2. Montoya JG, Liesenfeld O. Toxoplasmosis. Lancet2004;363:1965–1976.

3. Afonso C, Paixao VB, Costa RM. Chronic Toxoplasmainfection modifies the structure and the risk of hostbehavior. PLoS One 2012;7:e32489.

4. Maubon D, Ajzenberg D, Brenier-Pinchart MP, Darde

ML, Pelloux H. What are the respective host and para-site contributions to toxoplasmosis? Trends Parasitol2008;24:299–303.

5. Dubey JP. Advances in the life cycle of Toxoplasma gon-dii. Int J Parasitol 1998;28:1019–1024.

6. Berdoy M, Webster JP, Macdonald DW. Fatal attractionin rats infected with Toxoplasma gondii. Proc Biol Sci2000;267:1591–1594.

7. Webster JP. The effect of Toxoplasma gondii on animalbehavior: playing cat and mouse. Schizophr Bull2007;33:752–756.

8. Yagmur F, Yazar S, Temel HO, Cavusoglu M. MayToxoplasma gondii increase suicide attempt-preliminaryresults in Turkish subjects? Forensic Sci Int2010;199:15–17.

9. Kocazeybek B, Oner YA, Turksoy R et al. Higherprevalence of toxoplasmosis in victims of traffic acci-dents suggest increased risk of traffic accident in Toxo-plasma-infected inhabitants of Istanbul and its suburbs.Forensic Sci Int 2009;187:103–108.

10. Torrey EF, Bartko JJ, Lun ZR, Yolken RH. Antibodiesto Toxoplasma gondii in patients with schizophrenia: ameta-analysis. Schizophr Bull 2007;33:729–736.

11. Arias I, Sorlozano A, Villegas E et al. Infectious agentsassociated with schizophrenia: a meta-analysis. Schiz-ophr Res 2012;136:128–136.

12. Prandovszky E, Gaskell E, Martin H, Dubey JP, Webster

JP, McConkey GA. The neurotropic parasite Toxo-plasma gondii increases dopamine metabolism. PLoSOne 2011;6:e23866.

13. Carter CJ. Schizophrenia susceptibility genes directlyimplicated in the life cycles of pathogens: cytomegalovi-rus, influenza, herpes simplex, rubella, and Toxoplasmagondii. Schizophr Bull 2009;35:1163–1182.

14. Flegr J, Striz I. Potential immunomodulatory effects oflatent toxoplasmosis in humans. BMC Infect Dis2011;11:274.

174

Sutterland et al.

15. Miller BJ, Buckley P, Seabolt W, Mellor A, KirkpatrickB. Meta-analysis of cytokine alterations in schizophre-nia: clinical status and antipsychotic effects. Biol Psychi-atry 2011;70:663–671.

16. Myint AM, Schwarz MJ, Verkerk R et al. Reversal ofimbalance between kynurenic acid and 3-hydroxykynur-enine by antipsychotics in medication-naive and medica-tion-free schizophrenic patients. Brain Behav Immun2011;25:1576–1581.

17. Schwarcz R, Notarangelo FM, Hunter CA.Kynureninepathway metabolism reveals molecular links betweenimmune system dysregulation and psychiatric diseases.14th International Congress on Schizophrenia Research,ICOSR 2013 Orlando, FL United States, 2013;39:S125–S126.

18. Van Berckel BN, Bossong MG, Boellaard R et al. Mi-croglia activation in recent-onset schizophrenia: a quan-titative (R)-[11C]PK11195 positron emissiontomography study. Biol Psychiatry 2008;64:820–822.

19. Fond G, Macgregor A, Tamouza R et al. Comparativeanalysis of anti-toxoplasmic activity of antipsychoticdrugs and valproate. Eur Arch Psychiatry Clin Neurosci2014;264:179–183.

20. Leboyer M, Tamouza R, Charron D, Faucard R, PerronH. Human endogenous retrovirus type W (HERV-W) inschizophrenia: a new avenue of research at the gene–environment interface. World J Biol Psychiatry2013;14:80–90.

21. Stefansson H, Ophoff RA, Steinberg S et al. Commonvariants conferring risk of schizophrenia. Nature2009;460:744–747.

22. Jia P, Wang L, Meltzer HY, Zhao Z. Common variantsconferring risk of schizophrenia: a pathway analysis ofGWAS data. Schizophr Res 2010;122:38–42.

23. Corvin A, Morris DW. Genome-wide association stud-ies: findings at the major histocompatibility complexlocus in psychosis. Biol Psychiatry 2014;75:276–283.

24. Dinan TG, Borre YE, Cryan JF. Genomics of schizo-phrenia: time to consider the gut microbiome? Mol Psy-chiatry 2014;19:1252–1257.

25. Postolache TT. Toxoplasma gondii, the immune systemand suicidal behavior: An update. 31st InternationalWinter-Workshop on Clinical, Chemical, and Biochemi-cal Aspects of Pteridines StChristoph, Arlberg Austria.2012;23:72–74.

26. Reininghaus EZ, McIntyre RS, Reininghaus B et al. Tryp-tophan breakdown is increased in euthymic overweightindividuals with bipolar disorder: a preliminary report.Bipolar Disord 2014;16:432–440.

27. Severance EG, Gressitt KL, Yang S et al. Seroreactivemarker for inflammatory bowel disease and associationswith antibodies to dietary proteins in bipolar disorder.Bipolar Disord 2014;16:230–240.

28. Gray SM, Bloch MH. Systematic review of proinflam-matory cytokines in obsessive–compulsive disorder.Curr Psychiatry Rep 2012;14:220–228.

29. Mesman E, Hillegers MH, Ambree O, Arolt V, NolenWA, Drexhage HA. Monocyte activation, brain-derivedneurotrophic factor (BDNF), and S100B in bipolar off-spring: a follow-up study from adolescence into adult-hood. Bipolar Disord 2015;17:39–49.

30. Oliveira J, Busson M, Etain B, Jamain S, Hamdani N, Bou-kouaci W, et al. Polymorphism of Toll-like receptor 4gene in bipolar disorder. J Affect Disord 2014;152–154:395–402.

31. Oliveira J, Hamdani N, Busson M, Etain B, Bennabi M,Amokrane K, et al. Association between toll-like recep-

tor 2 gene diversity and early-onset bipolar disorder. JAffect Disord 2014;165:135–141.

32. Debnath M, Busson M, Jamain S, Etain B, Hamdani N,Oliveira J, et al. The HLA-G low expressor genotype isassociated with protection against bipolar disorder.Hum Immunol 2013;74:593–597.

33. Hamdani N, Daban-Huard C, Lajnef M et al. Relation-ship between Toxoplasma gondii infection and bipolardisorder in a French sample. J Affect Disord2013;148:444–448.

34. Miman O, Mutlu EA, Ozcan O, Atambay M, Karlidag R,Unal S. Is there any role of Toxoplasma gondii in the eti-ology of obsessive–compulsive disorder? Psychiatry Res2010;177:263–265.

35. Selten JP, Kahn RS. Schizophrenia after prenatal expo-sure to Toxoplasma gondii? Clin Infect Dis 2002;35:633–634.

36. Smith G. Estimating the population attributable fractionfor schizophrenia when Toxoplasma gondii is assumedabsent in human populations. Prev Vet Med2014;117:425–435.

37. Flegr J, Lindova J, Kodym P. Sex-dependent toxoplasmo-sis-associated differences in testosterone concentrationin humans. Parasitology 2008;135:427–431.

38. Gatkowska J, Wieczorek M, Dziadek B, Dzitko K, Dlu-gonska H. Sex-dependent neurotransmitter level changesin brains of Toxoplasma gondii infected mice. Exp Paras-itol 2013;133:1–7.

39. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferredreporting items for systematic reviews and meta-analy-ses: the PRISMA statement. J Clin Epidemiol2009;62:1006–1012.

40. BorensteinM, Hedges L, Higgins J, RothsteinH. Compre-hensive Meta-Analysis Version 2. Biostat, Englewood,NJ 2005. www.Meta-Analysis.com.

41. Higgins JPT, Altman DG, Sterne JAC. Chapter 8: Asses-sing risk of bias in included studies. In: Higgins JPT,Green S, eds. Cochrane Handbook for SystematicReviews of Interventions. Version 5.1.0 [updated March2011]. The Cochrane Collaboration, 2011. Availablefrom www.cochrane-handbook.org.

42. Hinze-Selch D, Daubener W, Eggert L, Erdag S, Stolten-berg R, Wilms S. A controlled prospective study ofToxoplasma gondii infection in individuals with schizo-phrenia: beyond seroprevalence. Schizophr Bull2007;33:782–788.

43. Xiao J, Buka SL, Cannon TD et al. Serological patternconsistent with infection with type I Toxoplasma gondiiin mothers and risk of psychosis among adult offspring.Microbes Infect 2009;11:1011–1018.

44. Shwab EK, Zhu XQ, Majumdar D et al. Geographicalpatterns of Toxoplasma gondii genetic diversity revealedby multilocus PCR-RFLP genotyping. Parasitology2014;141:453–461.

45. Duval S, Tweedie R. Trim and fill: a simple funnel-plot-based method of testing and adjusting for publicationbias in meta-analysis. Biometrics 2000;56:455–463.

46. Ahmad D, Mehdi S, Sayed HH, Sayed AK, Shirzad G.Serological survey of Toxoplasma gondii in schizophre-nia patients referred to Psychiatric Hospital, Sari City,Iran. Trop Biomed 2010;27:476–482.

47. Alipour A, Shojaee S, Mohebali M, Tehranidoost M, AbdiMasoleh F, Keshavarz H. Toxoplasma infection inschizophrenia patients: a comparative study with controlgroup. Iran J Parasitol 2011;6:31–37.

48. Alvarado-Esquivel C, Alanis-Quinones OP, Arreola-

Valenzuela MA et al. Seroepidemiology of Toxoplasma

175

T. gondii & psychiatric disorders, a meta-analysis

gondii infection in psychiatric inpatients in a northernMexican city. BMC Infect Dis 2006;6:178.

49. Alvarado-Esquivel C, Urbina-Alvarez JD, Estrada-

Martinez S et al. Toxoplasma gondii infection andschizophrenia: a case control study in a low Toxoplasmaseroprevalence Mexican population. Parasitol Int2011;60:151–155.

50. Bamne M, Wood J, Chowdari K et al. Evaluation of HLApolymorphisms in relation to schizophrenia risk andinfectious exposure. Schizophr Bull 2012;38:1149–1154.

51. Blomstrom A, Karlsson H, Wicks S, Yang S, Yolken RH,Dalman C. Maternal antibodies to infectious agents andrisk for non-affective psychoses in the offspring – amatched case–control study. Schizophr Res2012;140:25–30.

52. Brown AS, Schaefer CA, Quesenberry CP Jr, Liu L, Bab-ulas VP, Susser ES. Maternal exposure to toxoplasmosisand risk of schizophrenia in adult offspring. Am J Psy-chiatry 2005;162:767–773.

53. Brown AS, Montoya JG, Bao Y, Shen L, Schaefer CA.Specificity of maternal Toxoplasma gondii to risk ofschizophrenia in offspring. Schizophr Bull 2013;39:S59.

54. Buchy P, Follezou JY, Lien TX et al. Serological studyof toxoplasmosis in Vietnam in a population of drugusers (Ho Chi Minh City) and pregnant women (NhaTrang). Bull Soc Pathol Exot (1990) 2003;96:46–47.

55. Cetinkaya Z, Yazar S, Gecici O, Namli MN. Anti-Toxo-plasma gondii antibodies in patients with schizophrenia–preliminary findings in a Turkish sample. Schizophr Bull2007;33:789–791.

56. Conejero-Goldberg C, Torrey EF, Yolken RH. Herpesvi-ruses and Toxoplasma gondii in orbital frontal cortex ofpsychiatric patients. Schizophr Res 2003;60:65–69.

57. Dickerson F, Stallings C, Origoni A et al. Antibodies toToxoplasma gondii in individuals with mania. BipolarDisord 2014;16:129–136.

58. Dogruman-Al F, Aslan S, Yalcin S, Kustimur S, Turk S.A possible relationship between Toxoplasma gondii andschizophrenia: a seroprevalence study. Int J PsychiatryClin Pract 2009;13:82–87.

59. El-Sayed NM, Ismail KA, Ahmed SAEG, Ezz-El-din

HM, Azzam MEED. Possible association between Toxo-plasma gondii infection and schizophrenia: Egyptianstudy. Infect Dis Clin Pract 2012;20:394–399.

60. Emelia O, Amal RN, Ruzanna ZZ et al. Seroprevalenceof anti-Toxoplasma gondii IgG antibody in patients withschizophrenia. Trop Biomed 2012;29:151–159.

61. Gutierrez J, Roldan C, Maroto MC. Seroprevalence ofhuman toxoplasmosis. Microbios 1996;85:73–75.

62. Hamidinejat H, Ghorbanpoor M, Hosseini H et al. Toxo-plasma gondii infection in first-episode and inpatientindividuals with schizophrenia. Int J Infect Dis 2010;14:e978–e981.

63. Horacek J, Flegr J, Tintera J et al. Latent toxoplasmosisreduces gray matter density in schizophrenia but not incontrols: voxel-based-morphometry (VBM) study.World J Biol Psychiatry 2012;13:501–509.

64. Khademvatan S, Khajeddin N, Izadi S, Saki J. Study ofToxoplasma gondii infection in patients with bipolar dis-order. J Med Sci 2013;13:215–220.

65. Khademvatan S, Khajeddin N, Izadi S, Yousefi E. Investi-gation of anti-Toxocara and anti-Toxoplasma antibod-ies in patients with schizophrenia disorder. SchizophrRes Treatment 2014;2014:230349.

66. Krause D, Matz J, Weidinger E et al. The association ofinfectious agents and schizophrenia. World J Biol Psy-chiatry 2010;11:739–743.

67. Leweke FM, Gerth CW, Koethe D et al. Antibodies toinfectious agents in individuals with recent onsetschizophrenia. Eur Arch Psychiatry Clin Neurosci2004;254:4–8.

68. Li Y, Weber NS, Fisher JA et al. Association betweenantibodies to multiple infectious and food antigens andnew onset schizophrenia among US military personnel.Schizophr Res 2013;151:36–42.

69. Liu M, Wang T, Li H et al. Analysis of the antibodiesanti-Toxoplasma gondii by ELISA based on two diag-nostic antigens: rSAG1 and rBAG1. Acta Parasitol2011;56:353–359.

70. Mart�ınez AJ, Sell M, Mitrovics T et al. The neuropa-thology and epidemiology of AIDS. Pathol Res Pract1995;191:427–443.

71. Mortensen PB, Norgaard-Pedersen B, Waltoft BL et al.Toxoplasma gondii as a risk factor for early-onset schizo-phrenia: analysis of filter paper blood samples obtainedat birth. Biol Psychiatry 2007;61:688–693.

72. Mortensen PB, Pedersen CB, McGrath JJ et al. Neonatalantibodies to infectious agents and risk of bipolar disor-der: a population-based case–control study. Bipolar Dis-ord 2011;13:624–629.

73. Nascimento FS, De Rosalmeida Dantas C, NettoMP et al.Prevalence of antibodies to Toxoplasma gondii inpatients with schizophrenia and mood disorders. Schiz-ophr Res 2012;142:244–245.

74. Niebuhr DW, Millikan AM, Cowan DN, Yolken R, Li Y,Weber NS. Selected infectious agents and risk of schizo-phrenia among U.S. military personnel. Am J Psychiatry2008;165:99–106.

75. Park MH, Kwon YJ, Jeong HY et al. Associationbetween intracellular infectious agents and schizo-phrenia. Clin Psychopharmacol Neurosci 2012;10:117–123.

76. Pearce BD, Hubbard S, Rivera HN et al. Toxoplasmagondii exposure affects neural processing speed as mea-sured by acoustic startle latency in schizophrenia andcontrols. Schizophr Res 2013;150:258–261.

77. Pearce BD, Kruszon-Moran D, Jones JL. The relation-ship between Toxoplasma gondii infection and mood dis-orders in the third National Health and NutritionSurvey. Biol Psychiatry 2012;72:290–295.

78. Pedersen MG, Stevens H, Pedersen CB, Norgaard-Peder-sen B, Mortensen PB. Toxoplasma infection and laterdevelopment of schizophrenia in mothers. Am J Psychia-try 2011;168:814–821.

79. Saraei-Sahnesaraei MSF, Hashemi HJ, Khabbaz F, Ali-zadeh SA. Relation between Toxoplasma gondii infec-tions and schizophrenia. Iran J Psychiatry Clin Psychol2009;15:3–9.

80. Tamer GS, Dundar D, Yalug I, Caliskan S, Yazar S, AkerA. The schizophrenia and Toxoplasma gondii connec-tion: infectious, immune or both? Adv Ther 2008;25:703–709.

81. Tanyuksel M, Uzun O, Araz E, Koru O, Babur C. Possi-ble role of toxoplasmosis in patients with first-episodeschizophrenia. Turk J Med Sci 2010;40:399–404.

82. Tedla Y, Shibre T, Ali O et al. Serum antibodies toToxoplasma gondii and Herpesvidae family viruses inindividuals with schizophrenia and bipolar disorder: acase–control study. Ethiop Med J 2011;49:211–220.

83. Wang HL, Wang GH, Li QY, Shu C, Jiang MS, Guo Y.Prevalence of Toxoplasma infection in first-episodeschizophrenia and comparison between Toxoplasma-seropositive and Toxoplasma-seronegative schizophre-nia. Acta Psychiatr Scand 2006;114:40–48.

176

Sutterland et al.

84. Xiao Y, Yin J, Jiang N et al. Seroepidemiology of humanToxoplasma gondii infection in China. BMC Infect Dis2010;10:4.

85. Yolken RH, Bachmann S, Ruslanova I et al. Antibodiesto Toxoplasma gondii in individuals with first-episodeschizophrenia. Clin Infect Dis 2001;32:842–844.

86. Yuksel P, Alpay N, Babur C et al. The role of latenttoxoplasmosis in the aetiopathogenesis of schizophre-nia–the risk factor or an indication of a contact withcat? Folia Parasitol (Praha) 2010;57:121–128.

87. Yonghua Z, Yue C, Lei Z, QiG. Analysis of probable cor-relation between Toxoplasma gondii and schizophrenia:a seroepidemiological longitudinal investigation from2002 to 2007 in Suzhou and Wuxi regions, Jiangsu,China. Trop Med Int Health 2011;16(suppl 1):97–384.

88. Okusaga O, Langenberg P, Sleemi A et al. Toxoplasmagondii antibody titers and history of suicide attempts inpatients with schizophrenia. Schizophr Res 2011;133:150–155.

89. Meek B, van Gool T, Gilis H, Peek R. Dissecting the IgMantibody response during the acute and latent phase oftoxoplasmosis. Diagn Microbiol Infect Dis 2001;41:131–137.

90. Mortensen PB, Norgaard-Pedersen B, Waltoft BL,Sorensen TL, Hougaard D, Yolken RH. Early infectionsof Toxoplasma gondii and the later development ofschizophrenia. Schizophr Bull 2007;33:741–744.

91. Guerina NG, Hsu HW, Meissner HC et al. Neonatalserologic screening and early treatment for congenitalToxoplasma gondii infection. The New England Regio-nal Toxoplasma Working Group. N Engl J Med1994;330:1858–1863.

92. Anderson G, Maes M. Schizophrenia: linking prenatalinfection to cytokines, the tryptophan catabolite (TRY-CAT) pathway, NMDA receptor hypofunction, neuro-development and neuroprogression. ProgNeuropsychopharmacol Biol Psychiatry 2013;42:5–19.

93. Meyer U. Prenatal poly(i:C) exposure and other devel-opmental immune activation models in rodent systems.Biol Psychiatry 2014;75:307–315.

94. Zuckerman L, Rehavi M, Nachman R, Weiner I. Immuneactivation during pregnancy in rats leads to a postpuber-tal emergence of disrupted latent inhibition, dopaminer-gic hyperfunction, and altered limbic morphology in theoffspring: a novel neurodevelopmental model of schizo-phrenia. Neuropsychopharmacology 2003;28:1778–1789.

95. Xiao J, Kannan G, Jones-Brando L et al. Sex-specificchanges in gene expression and behavior induced bychronic Toxoplasma infection in mice. Neuroscience2012;206:39–48.

96. Flegr J, Lenochova P, Hodny Z, Vondrova M. Fatalattraction phenomenon in humans: cat odour attractive-ness increased for toxoplasma-infected men whiledecreased for infected women. PLoS Negl Trop Dis2011;5:e1389.

97. Hodkova H, Kolbekova P, Skallova A, Lindova J, FlegrJ. Higher perceived dominance in Toxoplasma infectedmen–a new evidence for role of increased level oftestosterone in toxoplasmosis-associated changes inhuman behavior. Neuro Endocrinol Lett 2007;28:110–114.

98. Lindova J, Kubena AA, Sturcova H et al. Pattern ofmoney allocation in experimental games supports thestress hypothesis of gender differences in Toxoplasmagondii-induced behavioural changes. Folia Parasitol2010;57:136–142.

99. Van Gool T, Gilis H, Haddad D. Serologische diagnos-tiek van toxoplasmose bij immuuncompetente pati€entenen zwangeren. Tijdschr Infect 2006;1:190–198.

100. Morisset S, Peyron F, Lobry JR et al. Serotyping ofToxoplasma gondii: striking homogeneous patternbetween symptomatic and asymptomatic infectionswithin Europe and South America. Microbes Infect2008;10:742–747.

101. Groer MW, Yolken RH, Xiao JC et al. Prenataldepression and anxiety in Toxoplasma gondii-positivewomen. Am J Obstet Gynecol 2011;204:433e1–433e7.

102. Novotna M, Havlicek J, Smith AP et al. Toxoplasma andreaction time: role of toxoplasmosis in the origin, preser-vation and geographical distribution of Rh blood grouppolymorphism. Parasitology 2008;135:1253–1261.

103. Ribbe K, Friedrichs H, Begemann M et al. The cross-sec-tional GRAS sample: a comprehensive phenotypicaldata collection of schizophrenic patients. BMC Psychia-try 2010;10:91.

104. Wong WK, Upton A, Thomas MG. Neuropsychiatricsymptoms are common in immunocompetent adultpatients with Toxoplasma gondii acute lymphadenitis.Scand J Infect Dis 2013;45:357–361.

105. Arling TA, Yolken RH, Lapidus M et al. Toxoplasmagondii antibody titers and history of suicide attempts inpatients with recurrent mood disorders. J Nerv MentDis 2009;197:905–908.

106. Yereli K, Balcioglu IC, Ozbilgin A. Is Toxoplasma gon-dii a potential risk for traffic accidents in Turkey? Foren-sic Sci Int 2006;163:34–37.

107. Mitchell MR, Weiss VG, Beas BS, Morgan D, Bizon JL,Setlow B. Adolescent risk taking, cocaine self-adminis-tration, and striatal dopamine signaling. Neuropsycho-pharmacology 2014;39:955–962.

108. Dekker N, Meijer J, Koeter M et al. Age at onset ofnon-affective psychosis in relation to cannabis use,other drug use and gender. Psychol Med 2012;42:1903–1911.

109. Markovitz AA, Simanek AM, Yolken RH et al. Toxo-plasma gondii and anxiety disorders in a community-based sample. Brain Behav Immun 2015;43:192–197.

110. Hamlyn J, Duhig M, McGrath J, Scott J. Modifiable riskfactors for schizophrenia and autism–shared risk factorsimpacting on brain development. Neurobiol Dis2013;53:3–9.

111. Brown AS. Epidemiologic studies of exposure to prena-tal infection and risk of schizophrenia and autism. DevNeurobiol 2012;72:1272–1276.

112. Nizel-SiminskaW. The mental state of children with posi-tive titres in serological examinations for toxoplasmosis.[Polish]. Psychiatr Pol 1980;14:241–246.

113. Hinze-Selch D, Daubener W, Erdag S,Wilms S. The diag-nosis of a personality disorder increases the likelihoodfor seropositivity to Toxoplasma gondii in psychiatricpatients. Folia Parasitol (Praha) 2010;57:129–135.

114. Monroe JM, Buckley PF, Miller BJ. Meta-analysis ofanti-Toxoplasma gondii IgM antibodies in acute psycho-sis. Schizophr Bull 2014; pii: sbu159. [Epub ahead ofprint].

115. Mahmoud SSHM. Seroprevalence of toxoplasmosisamong Schizophrenic patients. Yemeni J Med Sci2009;1:1–7.

116. James BO, Agbonile IO, Okolo M, Lawani AO, Om-

oaregba JO. Prevalence of Toxoplasma gondii infectionamong individuals with severe mental illness in Nigeria:a case control study. Pathog Glob Health 2013;107:189–193.

177

T. gondii & psychiatric disorders, a meta-analysis

117. Haroon F, Handel U, Angenstein F et al. Toxoplasmagondii actively inhibits neuronal function in chronicallyinfected mice. PLoS One 2012;7:e35516.

118. Henriquez SA, Brett R, Alexander J, Pratt J, RobertsCW. Neuropsychiatric disease and Toxoplasma gon-dii infection. NeuroImmunoModulation 2009;16:122–133.

119. Hashimoto K. Microglial activation in schizophrenia andminocycline treatment. Prog NeuropsychopharmacolBiol Psychiatry 2008;32:1758–1759; author reply 60.

120. Wake H, Moorhouse AJ, Miyamoto A, Nabekura J. Mi-croglia: actively surveying and shaping neuronal circuitstructure and function. Trends Neurosci 2013;36:209–217.

121. Gos T, Myint AM, Schiltz K et al. Reduced microglialimmunoreactivity for endogenous NMDA receptor ago-nist quinolinic acid in the hippocampus of schizophreniapatients. Brain Behav Immun 2014;41:59–64.

122. Ribeiro BM, Do Carmo MR, Freire RS et al. Evidencesfor a progressive microglial activation and increase iniNOS expression in rats submitted to a neurodevelop-mental model of schizophrenia: reversal by clozapine.Schizophr Res 2013;151:12–19.

123. Acharjee S, Branton WG, Vivithanaporn P et al. HIV-1Nef expression in microglia disrupts dopaminergic andimmune functions with associated mania-like behaviors.Brain Behav Immun 2014;40:74–84.

124. Blanchard N, Dunay IR, Schluter D. Persistence ofToxoplasma gondii in the central nervous system: a fine-tuned balance between the parasite, the brain and theimmune system. Parasite Immunol 2015;37:150–158.

125. Notarangelo FM, Wilson EH, Horning KJ et al. Evalua-tion of kynurenine pathway metabolism in Toxoplasmagondii-infected mice: implications for schizophrenia.Schizophr Res 2014;152:261–267.

126. Glaser KC, Hagos B, Molestina RE. Effects of Toxo-plasma gondii genotype and absence of host MAL/Myd88on the temporal regulation of gene expression in infectedmicroglial cells. Exp Parasitol 2011;129:409–413.

127. Da Silva RC, Langoni H. Toxoplasma gondii: host–para-site interaction and behavior manipulation. ParasitolRes 2009;105:893–898.

128. Suzuki Y. Host resistance in the brain against Toxo-plasma gondii. J Infect Dis 2002;185(Suppl 1):S58–S65.

129. Luft BJ, Kansas G, Engleman EG, Remington JS. Func-tional and quantitative alterations in T lymphocyte sub-populations in acute toxoplasmosis. J Infect Dis1984;150:761–767.

130. Bhadra R, Cobb DA, Weiss LM, Khan IA. Psychiatricdisorders in toxoplasma seropositive patients-The CD8connection. Schizophr Bull 2013;39:485–489.

131. Tan HY, Nicodemus KK, Chen Q et al. Genetic variationin AKT1 is linked to dopamine-associated prefrontalcortical structure and function in humans. J Clin Invest2008;118:2200–2208.

132. Skallova A, Kodym P, Frynta D, Flegr J. The role ofdopamine in Toxoplasma-induced behavioural altera-tions in mice: an ethological and ethopharmacologicalstudy. Parasitology 2006;133(Pt 5):525–535.

133. Van Os J, Kapur S. Schizophrenia. Lancet 2009;374:635–645.

134. Berk M, Dodd S, Kauer-Sant’anna M et al. Dopaminedysregulation syndrome: implications for a dopaminehypothesis of bipolar disorder. Acta Psychiatr ScandSuppl 2007;434:41–49.

135. Kim CH, Koo MS, Cheon KA, Ryu YH, Lee JD, Lee HS.Dopamine transporter density of basal ganglia assessed

with [123I]IPT SPET in obsessive–compulsive disorder.Eur J Nucl Med Mol Imaging 2003;30:1637–1643.

136. Volkow ND, Wang GJ, Tomasi D, Baler RD. Obesityand addiction: neurobiological overlaps. Obes Rev2013;14:2–18.

137. Holub D, Flegr J, Dragomirecka E, et al. Differences inonset of disease and severity of psychopathologybetween toxoplasmosis-related and toxoplasmosis-unre-lated schizophrenia. Acta Psychiatr Scand 2013;127:227–238.

138. Ducasse D, Boyer L, Michel P et al. D2 and D3 dopa-mine receptor affinity predicts effectiveness of antipsy-chotic drugs in obsessive–compulsive disorders: ametaregression analysis. Psychopharmacology 2014;231:3765–3770.

139. Kasteng F, Eriksson J, Sennfalt K, Lindgren P. Meta-bolic effects and cost-effectiveness of aripiprazole versusolanzapine in schizophrenia and bipolar disorder. ActaPsychiatr Scand 2011;124:214–225.

140. Machielsen M, Beduin AS, Dekker N et al. Differences incraving for cannabis between schizophrenia patientsusing risperidone, olanzapine or clozapine. J Psycho-pharmacol. 2012;26:189–195.

141. Martinez D, Saccone PA, Liu F et al. Deficits in dopa-mine D(2) receptors and presynaptic dopamine in heroindependence: commonalities and differences with othertypes of addiction. Biol Psychiatry 2012;71:192–198.

142. Torrey EF, Yolken RH. The urban risk and migrationrisk factors for schizophrenia: are cats the answer?Schizophr Res 2014;159:299–302.

143. Kapur S, Phillips AG, Insel TR. Why has it taken so longfor biological psychiatry to develop clinical tests andwhat to do about it? Mol Psychiatry 2012;17:1174–1179.

144. Muller N, Myint AM, Krause D, Weidinger E, SchwarzMJ. Anti-inflammatory treatment in schizophrenia. ProgNeuropsychopharmacol Biol Psychiatry 2013;42:146–153.

145. Schwarz E, Van Beveren NJ, Ramsey J et al. Identifica-tion of subgroups of schizophrenia patients withchanges in either immune or growth factor and hor-monal pathways. Schizophr Bull 2014;40:787–795.

146. Yolken RH, Torrey EF. Are some cases of psychosiscaused by microbial agents? A review of the evidence.Mol Psychiatry 2008;13:470–479.

147. Lynch JR, Cunningham MR, Warme WJ, Schaad DC,Wolf FM, Leopold SS. Commercially funded and UnitedStates-based research is more likely to be published;good-quality studies with negative outcomes are not. JBone Joint Surg Am 2007;89:1010–1018.

148. Dubey JP, Jones JL. Toxoplasma gondii infection inhumans and animals in the United States. Int J Parasitol2008;38:1257–1278.

Appendix 1: Search strategy

Medline

Search (toxoplasmosis OR toxoplasmoses ORtoxoplasmic OR toxoplasma OR Neurotoxoplas-mosis OR t gondii OR toxopl*) AND (mental dis-order OR “mental disorders” (Mesh) OR“Schizophrenia and Disorders with PsychoticFeatures” (Mesh) OR paranoia OR paranoid ORpsychotic OR psychosis OR psychoses OR schizo-

178

Sutterland et al.

phren* OR bipolar OR mania OR manic ORdepression OR “Mood Disorders” (Mesh) ORobsessive OR compulsive OR OCD).

EMBASE

1. (toxoplasmosis or Toxoplasma gondii or toxo-plasma).ab,kw,sh,ti.

2. (toxoplasmoses or toxoplasmic or Neurotoxo-plasmosis or t gondii or toxopl*).ab,kw,ti.

3. 1 or 2 (t gondii sensitief)4. exp schizophrenia/5. exp bipolar disorder/6. exp mood disorder/7. exp anxiety disorder/8. exp mental disease/9. (paranoia or psychotic or psychosis or psycho-

ses or depression or mania or bipolar or manicor schizophren*).ab,sh,ti.

10. exp alcohol abuse/11. alcohol abus*.ab,kw,ti.12. (alcoholic? or alcoholism).ab,kw,ti.13. alcoholism/14. alcohol dependence.ab,kw,ti.15. binge drink*.ab,kw,ti.16. drug abus*.ab,kw,ti.17. drug misuse*.ab,kw,ti.18. exp drug abuse/19. drug dependence.ab,kw,sh,ti.20. substance abuse.ab,kw,sh,ti.21. substance dependence.ab,kw,ti.22. binge eating disorder/23. binge eat*.ab,kw,ti.24. bulimia.ab,kw,sh,ti.25. increased appetite/26. appetite.ab,kw,ti.27. exp obesity/28. (obesity or obese or overweight).ab,kw,ti.29. tobacco dependence.ab,kw,sh,ti.30. tobacco.ab,ti.

31. exp smoking/32. smoking.ab,kw,ti.33. nicotin*.ab,kw,ti.34. narcotic.ab,kw,ti.35. narcotic dependence/36. (narcotic dependence or narcotic abuse or nar-

cotic addict*).ab,kw,ti.37. (crack or cocaine or marihuana or marijuana

or cannabis or pot or hashish or heroin or di-amorphine or morphine or morfin?).ab,kw,ti.

38. cocaine/39. diamorphine/40. morphine/41. cannabis/42. opiate/43. opiate?.ab,kw,ti.44. alcohol consumption/45. drinking behavior/46. (abstinence symptom or abstinence syndrome

or craving or drug withdrawal syndrome orsubstance withdrawal syndrome or withdrawalsymptom or withdrawal syndrome).ab,kw,ti.

47. withdrawal syndrome.ab,kw,sh,ti.48. autism/or asperger syndrome/or childhood dis-

integrative disorder/or “pervasive developmen-tal disorder not otherwise specified”/

49. (autism or autistic or (asperger? adj2 (syn-drome or disorder))).ab,kw,ti.

50. attention deficit disorder/51. (attention deficit or adhd).ab,kw,ti.52. or/4-5153. 3 and 52

PsycINFO

1. (toxoplasmosis or toxoplasmoses or toxoplas-mic or toxoplasma or Neurotoxoplasmosis or tgondii or toxopl*).mp. (mp=title, abstract,heading word, table of contents, key concepts,original title, tests & measures).

179

T. gondii & psychiatric disorders, a meta-analysis