Schizophrenia Research Forum - A Catalyst for Creative Thinking

Forum Discussion: Antibodies to Toxoplasma gondii in Patients with Schizophrenia: A Meta-Analysis

View article

In our Forum discussion “journal club” series, the editors of Schizophrenia Bulletin or Schizophrenia Research provide access to the full text of a recent article. A short introduction by a journal editor gets us started, and then it's up to our readers to share their ideas and insights, questions, and reactions to the selected paper. So read on…

Torrey EF, Bartko JJ, Lun ZR, Yolken RH. Antibodies to Toxoplasma gondii in Patients With Schizophrenia: A Meta-Analysis. Schizophr Bull. 2006 Nov 3.

View Comments By:
Chris Carter — Posted 12 March 2007
Fuller Torrey — Posted 13 March 2007
John McGrath — Posted 14 March 2007
Mary Reid — Posted 29 March 2007
Huan Ngo — Posted 9 April 2007
Fuller Torrey — Posted 11 April 2007
Mary Reid — Posted 14 April 2007
Carla Gallo — Posted 22 June 2007

Background Text
by Gunvant Thaker, Maryland Psychiatric Research Center, and Associate Editor, Schizophrenia Bulletin

In an upcoming issue of Schizophrenia Bulletin, Torrey, Bartko, Lun, and Yolken report results from a meta-analysis of studies examining association between antibodies to Toxoplasma gondii and schizophrenia. One of the unique aspects of the meta-analysis is that the authors include all available datasets, including many that were published in languages other than English. The analysis includes studies from 17 countries, many from Eastern Europe and China, and only six studies had been written in English. This attempt to include all available datasets is one of the methodological strengths of the meta-analysis carried out by Torrey and colleagues. The results suggest that schizophrenia patients are more than two times likely to have antibodies to T. gondii than the comparison subjects (odds ratio of 2.79). However, these studies are not informative about the timing of infection that contributed to the development of the disorder. Previous studies that noted associations between schizophrenia and other infections such as influenza suggested exposure during the prenatal period, usually at the beginning of the second trimester, being the critical time. Almost all of the studies reviewed by Torrey and colleagues examined antibodies to T. gondii in patients who already had developed the illness. Findings in the first episode cohorts are reassuring and suggest that the exposure likely occurred prior to the onset of the psychotic symptoms. Authors also cite publications that implicate maternal T. gondii infection in the etiology of schizophrenia. This raises the question whether the presence of antibodies in adults reflects prenatal infection, or whether postnatal infection with T. gondii is also a risk factor.

This is an interesting report identifying another environmental risk factor for schizophrenia with public health implications of reducing the risk. However, the meta-analysis also raises several questions. The evidence provided by meta-analysis is indirect, supporting an association but raising the question: Is it causal? Robert Schwarcz and Christopher Hunter speculate that the answer is yes, and in an accompanying "At Issue" piece provide a plausible causal pathway involving astrocyte-derived kynurenic acid. Schwarcz and Hunter review the recent literature that suggests that there is massive astrocyte activation and dramatic increase in the brain content of kynurenic acid in animals infected with T. gondii. Furthermore, data suggest that elevation of brain kynurenic acid levels play a role in the pathophysiology of schizophrenia. The same issue of Schizophrenia Bulletin includes findings from two new studies that further confirm a link between schizophrenia and T. gondii (Dickerson et al., 2007; Mortensen et al., 2007).

The next question is: when is the exposure critical in causing the harm? This is relevant for public health goals of reducing risk, as well as for therapeutic strategies in individual patients. If the prenatal period is the critical vulnerable time for the exposure to have any effect, then treatment against T. gondii in adulthood will be ineffective. What are the underlying mechanisms by which exposure to T. gondii leads to schizophrenia vulnerability, and how do these interact with other risk factors, including genetic factors? These are some of the questions that will need to be addressed, and hopefully will draw comments and discussion from other readers.

Comments on Online Discussion
Comment by:  Chris Carter
Submitted 10 March 2007
Posted 12 March 2007

Toll-like receptors (TLRs) play an important role in activating the immune response defense mechanisms to pathogens, including T. gondii. In mice, TLR9 appears to be an important defense against this parasite (Minns et al., 2006). T. gondii also appears to have evolved a counterattack mechanism, through which it blocks the effects of Toll receptor signaling (TLR2, TLR4 and TLR9) (Bennouna et al., 2006). Toll-like receptors activate a number of signaling pathways via the adaptor proteins MYD88 (myeloid differentiation primary response gene 88) and TIRAP (toll-interleukin 1 receptor [TIR] domain containing adaptor protein). These pathways include activation of the double-stranded DNA responsive eIF2-α kinase PKR (EIF2AK2) (Horng et al., 2001). This viral-activated kinase is one of a series of stress-responsive kinases that shut down protein synthesis via inhibition of the translation initiation factor EIF2B, a pathway that may well play a role in the oligodendrocyte malfunction observed in schizophrenia (Carter, 2007). Mutations in EIF2B (subunits 1-5) are responsible for vanishing white matter disease that primarily affects oligodendrocytes and astrocytes (van der Knaap et al., 2006). Oligodendrocyte cell loss plays a key role in schizophrenia (Uranova et al., 2007). In the rat, T. gondii infects a variety of cells, including astrocytes, neurons, and microglia (Luder et al., 1999). The effects of the parasite on oligodendrocytes are less well characterized, although toxoplasmosis can induce demyelination in humans (Bertrand et al., 1998).

T. gondii thus inputs into an eIF2-α kinase-signaling network also activated by other environmental risk factors (famine, rubella, influenza). This network, which can be linked to a large number of schizophrenia susceptibility genes (Carter, 2007), controls the viability of glial cells, again placing them firmly at the focus of the pathology of schizophrenia (Moises et al., 2002).

Perhaps we should also consider submitting these types of papers to veterinary journals. The vaccination of cats against toxoplasmosis and other pathogens might well be an effective means of reducing the incidence of many of the diseases, perhaps of zoonotic origin, that plague humanity (Olsen, 1999).

View all comments by Chris CarterComment by:  Fuller Torrey
Submitted 13 March 2007
Posted 13 March 2007

Chris Carter eloquently illustrates the many places where genes may interact with an infectious agent such as T. gondii. Regarding veterinary journals, we have in fact had interest in our work from this group but have elected to submit the majority of our papers to infectious disease journals.

View all comments by Fuller TorreyComment by:  John McGrath, SRF Advisor
Submitted 12 March 2007
Posted 14 March 2007

This paper is a good demonstration of the utility of systematic reviews and meta-analysis (Torrey et al., 2006). The review has identified several previously neglected studies. Furthermore, the synthesis of the data clearly shows that we have a good epidemiological signal here. Those with schizophrenia are significantly more likely to have serological markers of past T. gondii exposure.

As an aside, the data is also a testament to the tenacity and skills of the authors of this study—an initial hypothesis related to cat exposure has been buttressed by a convincing array of data (42 studies from 17 countries). This is an impressive collection of data for any field of risk factor epidemiology. While new hypotheses often strike ideological resistance, good data are harder to ignore.

Not only are individuals with schizophrenia more likely than controls to be seropositive for T. gondii, there is evidence that early life exposure is associated with an increased risk of later schizophrenia. A U.S. study based on banked maternal sera (Brown et al., 2005) and a Danish study based on neonatal dried blood spots (Mortensen et al., 2007) both lend weight to this association.

As the target article states, the biological plausibility of the exposure is beyond doubt—this is a parasite that impacts on brain function. From a public health perspective, T. gondii is an attractive candidate—interventions related to hygiene, contact with animals, eating uncooked meat, etc., should be able to reduce exposure to the candidate. T. gondii is associated with a range of adverse neurological outcomes (most notably congenital toxoplasmosis). If public health interventions can identify points for intervention (ideally ones that are cheap and safe), then we may be rewarded with reductions in the incidence of various adverse neuropsychiatric outcomes.

However, epidemiology is a notoriously blunt and imprecise instrument. In recent years, well-conducted observational studies have provided apparently robust associations that have subsequently been rejected when examined with randomized controlled trials. Think, for example, of the association between the use of hormone replacement therapy and reduced risk of heart disease that emerged from the U.S. Nurses Study (an observational study) and then the refutation from the subsequent Women’s Health Initiative (an RCT). Associations that emerge from observational studies may operate via confounding factors that were not considered in the original research design. These back-flips have been unsettling and sobering for the research community. Here is the take-home message: researchers must remain vigilant for residual confounding. In the absence of supportive data from RCTs, we must actively explore our findings looking for potential confounds.

When discussing risk factors for schizophrenia, I recall Fuller Torrey reassuring me that researchers are each allowed one delusion. His is cats—mine is vitamin D! In the spirit of friendly encouragement, I would like to offer a potential confound for the authors to consider that may provide alternative explanations for the association between schizophrenia and T. gondii. Vitamin D status can influence a broad range of immune markers (Cantorna et al., 2004; DeLuca and Zierold, 1998; Hewison, 1992). More specifically, based on animal studies and in-vitro studies, the administration of the active form of vitamin D has been shown to reduce intracellular growth of T. gondii (Rajapakse et al., 2005; Rajapakse et al., 2007). With respect to prenatal exposure to T. gondii, seasonal fluctuations in vitamin D levels may be one potential mechanism underpinning the seasonal fluctuations in acute toxoplasmosis in pregnant women (Logar et al., 2005). But, is there any evidence that prenatal vitamin D impacts on brain development? Indeed, there is robust evidence from animal experiments showing that developmental vitamin D deficiency is associated with a range of adverse brain outcomes of interest to schizophrenia research (Eyles et al., 2003; Feron et al., 2005; Kesby et al., 2006).

With respect to the association between adults with schizophrenia and serological markers of T. gondii, this may also be related to vitamin D status. Individuals with mental illness (or, indeed, impaired health in general) tend to have low vitamin D stores (Schneider et al., 2000). To date, there is no evidence linking adult vitamin D levels and adverse psychiatric outcomes. However, it is feasible that both low vitamin D and its associated nonspecific changes in immune markers are both merely downstream consequences of altered behavior after the onset of schizophrenia. Neither adult vitamin D levels nor T. gondii exposure would be causally related to schizophrenia in this scenario.

Could altered immune markers to T. gondii be proxy indicators of other, yet to be identified risk modifying factors for schizophrenia? Creative researchers might like to postulate other mechanisms linking their favorite nongenetic risk factors for schizophrenia and T. gondii. (Hint: prenatal folate levels appear to be associated with schizophrenia [Brown et al., 2007; Gilbody et al., 2007; McClellan et al., 2006; Stahl et al., 2005], and T. gondii has some novel features related to folate metabolism [Massimine et al., 2005]). If the association between T. gondii and risk of schizophrenia can withstand scrutiny with respect to known potential confounds, then this will be calmative for anxious epidemiologists. The T. gondii hypothesis has withstood the scrutiny of a systematic review and meta-analysis. By any standards, this is a gold-plated candidate risk factor. It would be negligent to ignore this candidate in future studies examining nongenetic risk factors for schizophrenia.

View all comments by John McGrathComment by:  Mary Reid
Submitted 27 March 2007
Posted 29 March 2007

A study by Tabbara and colleagues (Tabbara et al., 2001) reports an increased risk of Toxoplasma gondii infection in those with glucose-6-phosphate dehydrogenase deficiency. In view of the fact that glucose-6-phosphate dehydrogenase (G6PD) deficiency has been associated with acute psychosis, catatonic schizophrenia, and bipolar disorders (Bocchetta, 2003), I wonder whether it is actually the enzyme deficiency that causes these signs or the consequence of the infection. Chaudhary et al. comment (Chaudhary et al., 2005) on the fact that T.gondii cannot synthesize purines de novo and possesses two redundant purine salvage pathways involving the enzymes hypoxanthine-xanthine-guanine phosphoribosyltransferase (HXGPRT)1 and adenosine kinase. It's interesting that reduced adenosine A2A receptor expression with greatly increased adenosine A1A expression is reported in Lesch-Nyhan disease, which occurs due to deficiency of hypoxanthine-guanine phosphoribosyltransferase (Bertelli et al., 2006). The adenosine A1 receptor locus 1q32.1 has also been implicated in schizophrenia (Jang et al., 2006). Perhaps a T.gondii infection leads to purinergic dysfunction.

Chemokine receptor CCR5 deficiency is associated with susceptibility to infection with T. gondii. Rasmussen and colleagues report that a CCR5 32-bp deletion allele is a susceptibility factor for schizophrenia with late onset (Khan et al., 2006; Rasmussen et al., 2006).

Further evidence to support a role for the A2A adenosine receptor in the development of schizophrenia is that it is known to regulate cyclic AMP levels. DISC1 interacts with phosphodiesterase (PDE) 4B, which degrades cyclic AMP. Depression, mania, schizophrenia, paranoia, anxiety, and obsessive compulsive disorders are reported in patients with Huntington disease (Barquero-Jimenez and Gomez-Tortosa, 2001). Tarditi and colleagues (Tarditi et al., 2006) suggest that alteration of A2A receptor signaling is present in HD. Wang et al. (Wang et al., 2003) report reduced prepulse inhibition and startle habituation in mice lacking the A2A adenosine receptor.

Also of interest, the Lewandowski group (Lewandowski et al., 2007) reports that 22q11 deletion syndrome participants exhibited deficits in intelligence, achievement, sustained attention, executive functioning, and verbal working memory compared to controls. It's interesting that eczema and asthma are seen in chromosome 22q11.2 deletion syndrome (Staple et al., 2005). The adenosine A2A receptor is found at 22q11.2 and deficient A2A receptor activity leads to airway inflammation in an asthma model (Nadeem et al., 2007). This would seem to indicate increased A2B receptor activity as a consequence.

A patent issued on October 24, 2006 (, describes an A2B antagonist for the treatment of type 1 hypersensitivity disorders, such as asthma, hay fever, and atopic eczema. They also state, "Another adverse biological effect of adenosine acting at the A.sub.2B receptor is the overstimulation of cerebral IL-6, a cytokine associated with dementias and Alzheimer's disease. Inhibiting the binding of adenosine to A.sub.2B receptors would therefore mitigate those neurological disorders that are produced by IL-6." Carta et al. (Carta et al., 2002) find a correlation between the degree of mental retardation and IL-6 in patients with Down syndrome. I wonder whether there might also be a similar correlation in those with 22q11.2 deletions. Perhaps an A2B adenosine receptor antagonist may help alleviate the deficits in intelligence as well as the hypersensitivity disorders associated with this condition.

It's interesting that adenosine signaling by A2B receptor might stimulate hair growth through FGF-7 upregulation in dermal papilla cells as abundant scalp hair has been reported in DiGeorge chromosome region deletions (Ravnan et al., 1996; Iino et al., 2007). Deletions at 22q11 have also been associated with autosomal dominant polycystic kidney disease in which increased FGF-7 expression is reported (Gogusev et al., 2003; Mei et al., 2005). Perhaps the A2A adenosine receptor may be a candidate. In view of the association with 22q11 deletions and the increased risk of schizophrenia, it's also of interest that translin-associated protein X (TRAX) has been found to interact with the A2A adenosine receptor. The DISC1/TRAX locus has been implicated in several studies (Sun et al., 2006; Cannon et al., 2005).

View all comments by Mary ReidComment by:  Huan Ngo
Submitted 9 April 2007
Posted 9 April 2007

I am delighted to read this discussion, because as captivated as I have been by the proposed hypothesis of chronic Toxoplasma infection as a risk factor of schizophrenia, it brings to light some of the areas that need to be addressed, such as....

1. Prevalence of Toxoplasma seropositivity differs significantly across the globe, up to 75 percent in France, ~30 percent in the US, ~10 percent in Norway (Sukthana, 2006), depending on the cultural and hygienic practices. Why is the frequency of schizophrenia consistently ~1 percent across the board according to conventional dogma—but see current SRF Live Discussion. If an answer is genetic susceptibility, then how? If the next answer is epigenetic susceptibility, then which regions of which chromosomes are affected by Toxoplasma infection that contains the relevant clusters of risk alleles?

2. Since there is a complex array of susceptibility and risk genes associated with this disease, how does the host gene response induced by Toxoplasma interface with the abnormal expression of these candidates, and at which levels, ranging from molecule to molecule, cell to cell, region to region, circuit to circuit, or organ to organ?

3. Since Toxoplasma can infect all nucleated cells in the brain parenchyma (Carruthers and Suzuki, 2007) and does not appear to show highly specific trophism to brain regions (Vyas et al., 2007)—dissemination is by hijacked monocytes and dendritic cells via the cerebrovascular system (Courret et al., 2006)—which infected brain structure and which type of infected cell contribute to the eventual synaptic disconnectivities that are associated with psychosis?

4. Since Toxoplasma infection can be in a stable dormant stage for most of our lifetime as cysts, but sporadic reactivation of tachyzoites can occur during short periods of immunochallenges, which is more relevant to the schizophrenia pathology? If current proposals suggest a neonatal insult, why do most congenital toxoplasmosis cases not develop schizophrenia? If the neurodevelopmental defect is due to transplacental anti-Toxoplasma IgG from maternal sources during mid-pregnancy, from either an acute or chronic maternal infection, then what is the specific mechanism, and would you not expect to see a significantly higher schizophrenia rate in France?

5. How does Toxoplasma infection weigh in relation to so many other proposed environmental risk factors, for example, exposure to heavy metals, winter viral infection, and paternal age, etc.? Do all of these environmental risks converge on a common systemic effect, for example, Toll receptor activation, neuroendocrine depletion, folate deficiency, etc.?

6. Toxoplasma is a protozoan obligate auxotroph, requiring the host organism to provide specific biosynthetic precursors. If folate and purine scavenging is the cause of psychosis, then there would have to be a significant amount of parasites proliferating in the brain to deplete detectable plasma levels of vitamin D and brain levels of adenosine…an unlikely scenario, unless you have already developed detectable massive brain lesions, such as those found in HIV patients. If the auxotrophic hypothesis is accurate, then the parasite must be infecting critical nodes of circuitries at a critical window of neurodevelopment. Where and when?

The systemic review and meta-analysis are suggestive, but until we have more mechanistic models that can be experimentally tested, it remains on the fringe of being one more provocative proposal to the etiology of schizophrenia. Any Toxoplasma-related model must take into account other genetic and environmental risk factors, with refined sophistication at the elementary level of where, when and how?

As an armchair scientist preparing to reenter research science to engage in this exciting hypothesis, I have assimilated a comprehensive list of ~650 host genes that have been shown to be altered in gene microarray studies by Toxoplasma infection. The completed process of analyzing them, one at a time, as a possible lead into the neuropathological mechanism, has taught me that we can project all kinds of theoretical modeling—some are more mainstream and others are more obscure. I think that the answer may boil down to answering the "where" and "when" first; only then will the "how" be much easier to solve.

View all comments by Huan NgoComment by:  Fuller Torrey
Submitted 11 April 2007
Posted 11 April 2007

Dr. Ngo raises many questions, for most of which there are currently no good answers. There are several possible reasons why the incidence of schizophrenia may not correlate well with the incidence of individuals who have antibodies to this organism. Genetic predisposition, as Dr. Ngo notes, is certainly one, and it is of interest that many of the genes that have been identified in relationship to schizophrenia are genes that concern immunological function. Another is strain differences in T. gondii, with some strains being more neurotropic than others. Another variable is mode of infection, as it is unclear if infection by tissue cyst (e.g., eating undercooked lamb) produces the same result as infection by oocyst (e.g., inhaling spores from dried cat feces).

A potentially important variable is the timing of the infection. Polio is a classic example of an infectious agent that causes no sequelae when infection takes place in early childhood but may cause severe problems when infection takes place in late childhood. Finally is possible co-infection, since there are known human diseases (e.g., hepatitis D) that require two infectious agents working together.

View all comments by Fuller TorreyComment by:  Mary Reid
Submitted 12 April 2007
Posted 14 April 2007

Reply to Dr. Ngo:

You have asked why do most congenital toxoplasmosis cases not develop schizophrenia. Levy and colleagues (2006) report relatively high adenosine concentrations in neonatal blood plasma. If the auxotroph hypothesis is correct, then I wonder whether these high levels are protective?

It begs the question, does it require a double hit as suggested by Dr Torrey? Perhaps increased adenosine deaminase activity due to a further immune challenge depletes adenosine stores with an eventual immunodeficiency state.

View all comments by Mary ReidComment by:  Carla Gallo
Submitted 22 June 2007
Posted 22 June 2007

As stated in the previous commentaries, the article by Torrey and collaborators (Torrey et al., 2007) constitutes a step forward in the knowledge of the biological relationship between T. gondii and schizophrenia, for it brings solid evidence on the association of seropositivity to toxoplasma and the risk to develop schizophrenia. Nevertheless, it is a consensus that several important questions remain to be answered. These questions start with causality, and include timing of exposure, the underlying molecular mechanisms, and their modulating factors.

Important contributions to clarifying the timing issues are those of Brown et al. (2005) and Mortensen et al. (2007) on the role of in utero and early exposure to T. gondii for schizophrenia risk. Also, it should be very important to perform studies like those of Joram Feldon’s group that, simulating the immune challenge by certain viruses and bacteria through the activation of Toll-like receptor 3 (TLR3) during critical periods of fetal development in rodents, have demonstrated the appearance of schizophrenia-like behaviors in later life. Immune activation in early/mid-pregnancy leads to a variety of abnormalities associated with positive symptoms in schizophrenia, whereas immune activation in late gestation results in behavioral and pharmacological dysfunctions associated with negative and cognitive symptoms (Meyer et al., 2006). Interestingly, adoption of newborn mice to immune-stressed lactating mothers until weaning also produced neuroanatomical and pharmacological effects on the offspring (Meyer et al., 2007). Studies simulating the T. gondii immune challenge in rodents—for instance, through the activation of either TLR2, TLR9, and TLR11, or different combinations of them—would generate information useful for identifying the potential critical periods in humans, and will certainly be complementary to those performed by Brown et al. (2005) and Mortensen et al. (2007).

Other important approaches to the timing issue are the currently ongoing treatment trials of anti-toxoplasmosis drugs as adjunctive medication (Torrey and Yolken, 2007) in individuals with recent-onset or already established schizophrenia, and the follow-up studies in high-risk individuals. The results of these studies will allow assessing the role of parasite exposure or reactivation in adulthood, for the risk of developing schizophrenia.

In relation to the localization and the cellular/molecular mechanisms involved, it will be important to take into consideration that very fine tuning mechanisms of the membrane signaling machinery of the host—such as membrane lipid composition or lipid precursor availability—can be affected by T. gondii (Coppens, 2006). In consequence, the invasion or reactivation of the parasite, if occurring during critical periods of neurodevelopment or later life, could severely interfere with the membrane machinery organization—critical for signaling events—allowing the occurrence of the observed deficits in schizophrenia. These events could affect more severely certain cell types, depending on their reliance on these components at those critical time frames.

Finally, the point raised by Dr. Ngo about the relative weight of environmental risk factors is a crucial one. In this regard, it is very important that future studies with patients, associating clinical characteristics to toxoplasma seropositivity or antibody levels, analyze the information accounting for potential confounders—such as exposure to herpesviruses—with the idea of adjusting the results for their influence on the observed effects of toxoplasma. Our ongoing studies have been designed in this line. It will be also useful to identify populations having only one or few known risk factors; this will allow the dissection of the individual contribution of these factors with smaller population samples.

View all comments by Carla Gallo