The infectious theory of psychosis, prominent early in the twentieth century, has recently received renewed scientific support. Evidence has accumulated that schizophrenia and bipolar disorder are complex diseases in which many predisposing genes interact with one or more environmental agents to cause symptoms. The protozoan Toxoplasma gondii and cytomegalovirus are discussed as examples of infectious agents that have been linked to schizophrenia and in which genes and infectious agents interact. Such infections may occur early in life and are thus consistent with neurodevelopmental as well as genetic theories of psychosis. The outstanding questions regarding infectious theories concern timing and causality. Attempts are underway to address the former by examining sera of individuals prior to the onset of illness and to address the latter by using antiinfective medications to treat individuals with psychosis. The identification of infectious agents associated with the etiopathogenesis of schizophrenia might lead to new methods for the diagnosis, treatment and prevention of this disorder.
The idea that microbial agents may cause psychotic disorders has a surprisingly lengthy history. As the germ theory of diseases became established in the late-nineteenth century, several European researchers theorized that bacteria might be etiologically linked to dementia praecox or other psychiatric diseases. It was speculated at that time that the bacteria might not infect the brain directly but rather could reside in the intestine or elsewhere and, by a process of autointoxication, affect the brain by secreting toxins.
One of the earliest enthusiasts for this theory was Emil Kraepelin who, in the 1896 edition of his Psychiatrie, reasoned that dementia praecox was ‘a tangible morbid process of the brain’ caused by ‘an autointoxication whose immediate causes lie somewhere in the body’; Kraepelin believed that the sex organs were the most likely site of infection.1 The fact that psychoses occasionally accompanied known bacterial diseases such as typhoid fever, tuberculosis and diphtheria supported the infectious theory; a 1904 review concluded that although ‘insanity following infection is generally of short duration, … in a few cases it does last from several months to years’.2 The following year, a spirochete was discovered to be the cause of syphilis, and by the First World War, infectious theories had joined genetic and endocrine theories in the mainstream of psychiatric research.
The autointoxication theory, however, fell into disrepute following attempts by several researchers to surgically remove the theoretical foci of infection in various organs, often with tragic results.3, 4 Attention then shifted to viruses, after it was noted that infection with the influenza virus caused psychosis in some people.5 Such cases were extensively studied by Karl Menninger following the 1918–1919 influenza pandemic. After observing many cases, Menninger concluded: ‘I am persuaded that dementia praecox (schizophrenia) is at least in most instances a somata-psychosis; the psychic manifestations of an encephalitis (italics in original). The acuteness or chronicity, the benign or malignant nature of this encephalitis perhaps determines the degree of reversibility of the schizophrenia.’6 In recent years, modern imaging techniques have established that schizophrenia is, in most cases, not associated with acute encephalitis, although it remains possible that it is a sequela or recurrence of a past episode of encephalitis.
Interest in infectious theories of psychiatric disorders waned in the United States and Europe in the 1930s as Freudian theories became prominent. Psychiatric research on infectious agents continued sporadically, mostly in Eastern Europe7, 8 and Russia,9 but after the onset of the Cold War, these studies were not widely known in the West. In the 1970s, there was renewed interest in infectious agents in Western Europe10, 11, 12 and the United States,13, 14 culminating in a 1983 World Health Organization symposium, ‘Research on the Viral Hypothesis of Mental Disorders’.15 Since that time, interest in this research area has steadily increased.
This review attempts to summarize such research. It first summarizes general evidence favoring an infectious theory, and then discusses the interaction of microbial agents and genetic factors. Of necessity, the review does not discuss every infectious agent and every psychiatric disorder that have been studied. It rather focuses on psychotic disorders in general, and schizophrenia in particular, as examples, and on the infectious agents that appear to be the most promising etiological candidates: Toxoplasma gondii and cytomegalovirus (CMV). Finally, it concludes with a summary of the current state of this research area, including both its strengths and weaknesses.
Why should microbial agents be considered?
There are several aspects of psychotic disorders that suggest a possible infectious origin. One is the century-old observation, noted above, that individuals sometimes present with the clinical symptoms of schizophrenia, bipolar disorder or depression with psychosis in the course of developing, or shortly after having had, a known microbial disease. The list of infectious agents that have been reported to sporadically cause psychotic symptoms is long but most prominently includes spirochetes such as Treponema pallidum (causing syphilis) and Borrelia burgdorferi (causing Lyme disease); viruses such as herpes simplex viruses (HSV-1 and -2), Epstein–Barr virus, CMV, influenza, measles, rubella, mumps, polio, vaccinia, enteroviruses such as Coxsackie B4, arboviruses such as Eastern equine encephalitis virus, retroviruses such as the human immunodeficiency virus (HIV) or endogenous human retroviruses, and Borna disease virus; and protozoa such as T. gondii.16, 17, 18 Such cases may become evident by the appearance of neurological symptoms shortly after the onset of psychotic symptoms, or they may remain hidden until autopsy.
Epidemiological considerations, however, make it unlikely that many of these microbes are candidates for causing a large number of cases of schizophrenia or other psychoses. Since psychoses are widespread in the world, localized infectious agents such as arboviruses, Borna disease virus and B. burgdorferi are unlikely to be causative agents in different areas of the world. Schizophrenia and bipolar disorder have also been described for at least 200 years, thus ruling out infectious agents that only recently infected human populations, such as HIV. In addition, the incidence of schizophrenia and bipolar disorder has not decreased markedly in recent years, whereas the incidence of syphilis, rubella, measles and mumps has done so as a result of immunization and improved therapies. Thus, relatively few microbes meet epidemiological requirements for serious consideration for causing more than sporadic cases.
Another epidemiological aspect of schizophrenia and bipolar disorder is the fact that both have a modest seasonal birth predominance in the winter and spring months.19 Over 200 studies have demonstrated this, and it is, in fact, one of the most consistently replicated aspects of these disorders. Statistical artifact and parental procreational habits have been ruled out as explanations, but seasonal differences in sunlight, temperature, rainfall, birth complications, toxins and nutrition have all been considered, along with microbes. If an infectious agent is responsible for the seasonality, it would be one that occurs through the year but that has a modest seasonal predominance. The infectious agent would also be one that does not vary markedly in incidence from year to year; this makes influenza an unlikely candidate.
Another epidemiological fact of interest is that being born in, or raised in, an urban environment, compared to a rural environment, approximately doubles a person's risk of later being diagnosed with schizophrenia. The urban risk is clearly associated with schizophrenia but has not been observed for bipolar disorder or other affective disorders.20 A dose–response relationship has been described insofar as ‘the more years lived in the higher degree of urbanization, the greater the risk’.21 In one study, the population-attributable risk for developing schizophrenia due to childhood urban living was 34.6, more than six times greater than the genetic risk conferred by having a first-degree relative with schizophrenia (5.5).22 Socioeconomic status and obstetric complications have been ruled out as explanations of the urban risk factor;23 a greater density of population must be considered, since it is associated with the rapid transmission of many microbes.
An unresolved issue concerning the possible microbial etiology of psychosis is the timing of the infection. In some, but not all, studies, it has been reported that maternal infections with rubella,24 influenza,25 HSV-226 and T. gondii27, 28 increase the incidence of psychotic disorders in the offspring. Other studies have documented an increased rate of schizophrenia in adults who during childhood had viral or bacterial meningitis.29 It is thus possible that microbes could cause psychosis secondary to infections in utero, in infancy or in childhood in addition to in adulthood around the time of onset of the psychosis. It is also possible that an infection occurring in infancy can be reactivated in later life, as is known to occur with the herpes viruses and T. gondii. The outcome of the infection may differ markedly depending on its timing, as is known to be true for human infections such as those caused by the rubella and polio viruses. It is also known that infections at different stages of brain development result in varying degrees of lifelong changes in behavior and cognition.
How genes and microbes interact
Despite more than a century of research, the genetic basis of psychotic disorders is not well understood. For schizophrenia, recent research has focused on the search for genetic mutations and polymorphisms as causes. This research is based on multiple family studies indicating that schizophrenia has a high degree of heritability; it has been stimulated by the wealth of genetic information that has become available as part of the human genome project. The search for specific genes associated with schizophrenia has also been fueled by findings from linkage studies indicating that some regions of the human genome are inherited nonrandomly in family members who have schizophrenia.
Despite a great deal of laboratory research, no single gene has been discovered that has a major effect in different populations. Although several genes have been associated with schizophrenia in case-control and family studies, many have not been found to be operant in different populations. The genes that are consistently associated with schizophrenia in different studies have relatively low odds ratios, generally in the range of 1.1–1.5. Studies reporting higher odds ratios have generally not been reproducible in subsequent studies performed in other populations. An increased understanding of human haplotype structure has also suggested that the strength of some previously reported associations has been overestimated because of haplotype misclassification. In addition, many of the genes that are associated with schizophrenia have similar associations with other psychiatric disorders such as bipolar disorder and unipolar depression, as has recently been described for methylenetetrahydrofolate reductase.30
While the reproducible associations are of great interest to neurobiologists interested in elucidating pathways of neural transmission and other neurological processes, associations at this level are of substantially less value to clinicians and epidemiologists, since they have low positive predictive values and thus provide little in the way of clinical or diagnostic guidance. Furthermore, findings in this range are difficult to duplicate except in studies of large sample sizes. For example, it takes a study of approximately 2000 individuals to detect a genetic factor that is present in the general population at a rate of 10% and that has a relative risk of 1.5 with a power of 0.9. Even larger sample sizes are required to detect genes that are present in lower prevalence or that have lower odds ratios; for example, more than 6100 individuals would be required to detect a gene with a population prevalence of 5% and a relative risk of 1.4.
Methods for genetic analysis are continually evolving. There is currently much interest in methods that can increase the number of genetic loci that can be evaluated in a given period of time. While such methods, which include ‘whole genome association’ studies, have given hope in terms of finding new areas of genetic association, it is difficult to envision how the results from these studies will solve the problem of the low odds ratios and the population diversity that seem to be inherent in the study of schizophrenia. In fact, it can be anticipated that these studies will result in the identification of larger numbers of potential genetic associations that will need to be replicated in very large sample sizes and then integrated into current algorithms. Furthermore, the analyses of larger numbers of potential loci will require the performance of larger numbers of multiple comparisons among genes, making it even more difficult to distinguish positive findings from true associations.
For this reason, it is important to elucidate nongenetic factors that might contribute to schizophrenia. Of particular importance are environmental influences that might interact with genetic factors and thus be applicable to a disease such as schizophrenia, which has a high degree of heritability. The integration of environmental risks might increase the odds ratios of genetic associations in specific populations of exposed individuals and hence increase the likelihood of identifying true positive associations.
The modern era of studying how genes and microbes interact was initiated in 1948 with the observation by Haldane31 that the gene for sickle hemoglobin provides protection against malaria, undoubtedly explaining the persistence of this gene in humans living in malaria endemic areas. More recently, the observation that individuals vary in their response to HIV infection has led to the discovery of a number of genes that are associated with a variable response to retrovirus infection. There are many additional infectious diseases for which a genetic component has been found in repeated studies. These include infections caused by Mycobacterium tuberculosis, M. leprae (leprosy), hepatitis B and C viruses, cholera, Helicobacter pylori, norovirus and Leishmania donovani. In addition, genes are very important in infectious diseases involving multiple organisms, such as bacterial sepsis and otitis media.32
Currently recognized genetic determinants of response to infection include receptors, transcription factors, cytokines, complement components, human lymphocyte antigen and other T-cell determinants, Toll-like receptors and other determinants of innate immunity. As different aspects of the immune system are more precisely defined in humans and in animal models, there is an ever-increasing list of genes that can potentially be associated with host response to infectious agents. In addition, the receptors for specific agents are being increasingly identified through the application of molecular biological techniques, further increasing the number of potential genetic loci defining susceptibility to infectious agents.
As discussed above, there are a number of infectious agents, which have been associated with schizophrenia. This review focuses on Toxoplasma and CMV, since they meet the criteria discussed above and have been associated with schizophrenia in different populations
T. gondii and schizophrenia
T. gondii is a coccidian protozoan of the apicomplexa family, first described in 1908. Felines, including domestic cats, are its definitive host, and the organism can only complete the sexual part of its life cycle within feline hosts. As shown in Figure 1, T. gondii oocysts are excreted in the feces of cats at the time they are initially infected. The oocysts may then become aerosolized and infect humans who are changing cat litter boxes, gardening or playing in sandboxes. The cat may also deposit feces on the ground or in animal feed in barns; domestic animals may eat it, producing T. gondii tissue cysts in their muscles, which then infect humans who eat undercooked meat. Felines can also contaminate water, with subsequent infection of humans through drinking water, eating vegetables washed with contaminated water or eating undercooked fish that had lived in the contaminated water. The relative importance of different modes of transmission depends upon local geographic, cultural and socioeconomic factors.
T. gondii is distributed worldwide and infects between 10 and 80% of the adult population; in the United States, the rate is approximately 25%.33 Of interest in relationship to schizophrenia, the peak ages of seroconversion is between ages 15 and 35;34 adolescent males become infected earlier than females35 and there are inconsistent suggestions of a seasonal occurrence (least in summer)36, 37 and greater urban risk.38
Genetically, numerous animal studies have identified genetic determinants of response to T. gondii infection, especially in immune-compromised hosts. Host genes associated with susceptibility or resistance to Toxoplasma infections include ones that regulate the expression of T-cell determinants,39 nuclear factor-κB transcription,40 interferon gamma,41 lymphotoxinα,42 other cytokines,43 Toll-like receptors44 and chemokine receptors,45 as well as genes of an as yet undefined effect on macrophage function.46 The genes that determine the response to Toxoplasma in immune competent humans have not yet been defined but are the subject of ongoing investigations.
Clinically, most human infections with T. gondii had been thought to be asymptomatic. However, recent studies of individuals infected with Toxoplasma-contaminated water have indicated that a majority of infected immune-competent individuals have clinically apparent symptoms, such as headache, fever, malaise, myalgias, adenitis, anorexia and arthralgias.47 Occasional cases have been described with delusions and hallucinations.48 The microbe is known to be neurotrophic and to infect both neurons and glia.49 Serologically, the first research linking schizophrenia and other psychoses to an increase in antibodies to T. gondii was published in 1953; since then, at least 41 other studies have been carried out. A recent meta-analysis on 23 of these studies reported an odds ratio for schizophrenia risk associated with serological evidence of T. gondii infection as 2.73 (95% confidence interval, 2.10–3.60; P<0.000001).50 While still in the range of moderate association, this odds ratio is substantially higher than that found in most meta-analyses of specific genes associated with schizophrenia risk.
These serological findings are supported by other studies linking T. gondii to schizophrenia. A study of newborn sera and a study of maternal sera of individuals who later developed schizophrenia and schizophrenia spectrum disorders both reported more T. gondii antibodies in cases than in controls.27, 28 In addition, preliminary analysis of a cohort of individuals in the US Military indicates that increased levels of Toxoplasma antibodies can be found in individuals prior to the onset of symptoms, obviating the possibility that the finding of increased levels of antibodies is an epiphenomenon associated with exposure occurring after the onset of disease.51
There are a number of reasons why interest in the study of the role of Toxoplasma and other infectious agents in the etiology of schizophrenia has, until recently, been limited to a small number of investigators. First, until recently, there was little understanding of the molecular biology of T. gondii and the pathophysiological mechanisms by which Toxoplasma infection causes disease. Recently, the techniques that led to the sequencing of the human genome have been applied to the sequencing of protozoan genomes as well, resulting in a corresponding increase in the characterization and understanding of the molecular basis of infection. For example, the characterization of the genome of T. gondii, which was completed in 2005, has led to an increased understanding of the biology of Toxoplasma infections and to the development of assays for antibodies to specific Toxoplasma proteins. The application of these assays has revealed that Toxoplasma organisms, which had previously been thought to be relatively uniform in terms of biological properties, exist in multiple strains with differing degrees of pathogenicity. Of particular importance has been the recent elucidation of specific Toxoplasma protein kinases in some strains, which alter signal transduction in infected hosts and thus modulate humoral immune function. These protein kinases appear to be the major determinants of the pathogenicity of individual Toxoplasma strains.52, 53 Further studies of these and other markers of pathogenicity should lead to improved assays for defining the risk of disease in Toxoplasma-infected individuals.
Another factor that has impeded an understanding of the role of infectious agents as etiological factors in schizophrenia has been the lack of a teleological framework relating to why, from an evolutionary point of view, infection with a microbial agent might lead to the altered behavior that is the hallmark of schizophrenia. However, recently it has been recognized that infectious agents can alter host behavior in a manner than enhances the survival and reproduction of the infecting agent. T. gondii provides a prime example of this type of evolutionary adaptation. As noted above, felines constitute the natural hosts for the protozoan. If a Toxoplasma organism infects a cat or other member of this family, it can complete its entire life cycle, including sexual reproduction. If, however, the organism infects another animal, for example, through the ingestion of Toxoplasma organisms shed by infected cats, the non-feline host is a ‘dead end’ for the Toxoplasma, in that the microbe cannot complete the sexual part of its life cycle and thus remains trapped in cyst form within the tissues of this incomplete host. The organism trapped in a non-feline host has another way out, however; if it can get the host to be eaten by a feline, it will be able to complete its life cycle in the new feline host.
Since all felines are carnivores, the possibility that another animal will be consumed is not remote; if the host is a rodent or other small animal, the possibility is even greater. Recent research has indicated that T. gondii can increase the likelihood of having its rodent intermediate host eaten by a cat by altering the motor behavior of the rodent.54 Specifically, the rodent loses its inherent fear of novel stimuli, including its fear of cats. Behavioral experiments have documented that Toxoplasma-infected rats lose their aversion to cat urine without other alterations in motor behavior, such as lethargy, which might make the rat less desirable prey to felines. In the real world, this alteration is likely to lead to a markedly increased consumption of Toxoplasma-infected rodents by cats, resulting in an infection of the cats and completion of the Toxoplasma life cycle. Since humans are generally not liable to be consumed by felines (hunters and trainers of large cats excepted), the effect of Toxoplasma on humans is largely vestigial. Since the Toxoplasma organism has limited ability to modulate its response in different intermediate hosts, however, it can be postulated that all hosts can be affected by similar pathways. In fact, in humans, two studies have independently reported that young adults with serological evidence of Toxoplasma infection have increased rates of automobile accidents compared to age-and gender-matched controls, a behavior that is likely to be associated with increased risk-taking behavior.55, 56
The neurobiological mechanism by which T. gondii causes psychiatric symptoms and exerts its effect on human behavior is unknown. One possibility is a direct effect of the organism on neurons and/or glia; such an effect may occur at the time of infection or much earlier in life. There are animal models of infections known to occur in utero or early in life that do not become manifest with behavioral changes until the animal reaches maturity.57 In the case of T. gondii, it is known that the organism may affect signal transduction pathways and that it also encodes proteins with hemology to tyrosine hydroxylase and the mammalian D2 receptor, suggesting that it may interfere with dopamine synthesis pathways in human hosts.58, 59 There are also models of infectious agents that affect dopamine pathways in one part of the brain but not in others.60 Alternatively, infectious agents such as T. gondii may cause psychiatric symptoms by the effect of antibodies against the organism or by cytokines elicited by the infection.61 Studies of such mechanisms are ongoing.
The development of hallucinations and other clinical symptoms in an individual infected with Toxoplasma is thus likely to be dependent on both the human and the microbial genomes. In this sense, schizophrenia might be considered not merely a genetic disease but rather a disease resulting from the interaction of multiple genomes. In this context, it is likely that study of both the human and the microbial genomes will be required to come to a complete understanding of complex neuropsychiatric disorders such as schizophrenia.
Cytomegalovirus and schizophrenia
Cytomegalovirus is another example of an infectious agent that has been linked to a specific psychiatric disorder, schizophrenia. It is a β-herpesvirus initially isolated in 1956 from children with congenital mental retardation and hearing loss. CMV occurs worldwide and is spread by personal contact, including saliva, blood, urine, genital secretions and breast milk. It thus spreads more quickly under conditions of poor hygiene; in developed countries, approximately 60% of adults are infected, while in developing countries, the figure approaches 100%. Some observers have said that CMV infections are more common in the winter and spring,62 but others have not observed any seasonal difference.
It is known that resistance to CMV infection is associated with variations in the gene encoding tumor necrosis factor-α;63 variations in this gene have also been associated with an increased risk for schizophrenia in some studies64, 65 but not others.66 Variations in the gene-encoding interleukin-10 have also been associated with susceptibility to both CMV infections and schizophrenia.67, 68
The two most prevalent CMV clinical syndromes are congenital infection and systemic infections in immunosuppressed individuals. Congenital infections may produce mental retardation, hearing loss, visual impairment and other deficits in the newborn. In many cases, the infection is initially asymptomatic but later decreases the child's IQ.69 Once infected with CMV, individuals remain latently infected for life, although they may also become secondarily infected with a different CMV strain.70 Suppression of the immune system, as occurs in transplantation and HIV infection, often leads to reactivation of CMV infection with a range of clinical consequences, including encephalitis, pneumonia, hepatitis, enteritis and retinitis.
In immune-competent individuals, most initial CMV infections are asymptomatic. If symptomatic, the most common syndrome is fever and enlarged lymph nodes, and CMV is thought to be responsible for approximately 10% of mononucleosis-like syndromes in young adults. Occasional cases of encephalitis have been reported. In one case, a 30-year-old male developed chronic CMV encephalitis with ‘deficits in concentration, memory, manipulation of knowledge, humor and emotional expression’.71 Similarly, a study of CMV antibodies in 323 individuals with schizophrenia reported that those who had primarily deficit symptoms (n=88) were significantly more likely to have CMV antibodies compared to those who had nondeficit symptoms (n=235; P=0.006).72 Another study of individuals with schizophrenia reported that ‘higher levels of CMV antibodies were associated with decreased performance on baseline measures of verbal learning and memory.73 Also of interest is a case study of a young woman who presented with auditory hallucinations, delusions, tangential thinking and flattened affect and was diagnosed with schizophrenia; while hospitalized, she died and was then diagnosed with CMV encephalitis on the basis of retrospective antibody titers and post-mortem findings.16
Neuropathologically, CMV is known to have an affinity for the limbic system74 and to evoke a strong glial response, producing cytokines and chemokines.75 It is unclear whether CMV's effects on the brain are primarily due to the direct effects of the virus or are indirectly mediated by the immune response. The histopathologic hallmark of CMV infections are intranuclear inclusion bodies and scattered glial nodules. Most neuropathological studies of CMV have been done on immune-compromised individuals, and it is unclear whether immune-competent individuals exhibit similar findings. In animal studies, infecting rats with a rat CMV shortly after birth was said to produce ‘a deficit in sensorimotor gating upon dopamine stimulation, supporting a possible link between viral infection and schizophrenia’.76
Many studies have looked for evidence of CMV infection in the blood, CSF and brain tissue of individuals with schizophrenia. Fourteen serological studies carried out prior to 1994, using less sensitive assays and mostly patients with chronic schizophrenia, were negative (reviewed in Yolken and Torrey77). More recently, five studies using more sensitive assays and patients with a more recent onset of schizophrenia have all reported a higher rate of CMV seropositivity in the patients compared to controls (reviewed in Torrey et al.78).
Of particular note was the study by Leweke et al.79 in Germany. They studied 36 first-episode, never-treated individuals with schizophrenia; 10 individuals with schizophrenia who had been treated in the past but were medication-free at the time of the study; 39 individuals with schizophrenia with recent onset but who were on medication at the time of the study and 73 unaffected controls. As shown in Figure 2, CMV immunoglobulin G antibody levels were significantly higher in both the serum (P<0.001) and cerebrospinal fluid (CSF) (P<0.004) in individuals with schizophrenia who had never been treated compared to the unaffected controls. Noteworthy was the stepwise gradient of antibody levels in both serum and CSF from those who had never been treated, those who had been treated in the past, those who were presently being treated and the unaffected controls, suggesting that antipsychotic medication may have been decreasing the antibodies.
Other CMV studies of the CSF in individuals with schizophrenia have been less conclusive. Since 1980, eight studies have been done; four reported significantly increased CMV antibodies in the CSF of patients compared to controls, while four others did not (reviewed in Torrey et al.78). Nine studies have also attempted to identify the CMV genome in post-mortem brain tissue from individuals with schizophrenia, using hybridization or PCR techniques. Eight of the studies were unable to detect CMV in the brain tissue (reviewed in Torrey et al.78); the other reported that ‘a clear hybridization signal was detected with DNA from the temporal cortex of a young man with the full picture of schizophrenia’.80The Leweke et al. study, cited above, suggests that treating individuals with schizophrenia with antipsychotic drugs may reduce antibodies to CMV, presumably by suppressing the infection. This observation is consistent with studies reporting that some antipsychotic drugs suppress the replication of some infectious agents.81 Based on this observation, two studies have also been carried out using adjunct medications that suppress CMV replication in individuals with schizophrenia. In the first, valacyclovir, a known antiviral drug, was used for a 16-week, double-blind trial for 65 outpatients with schizophrenia. Among the 21 patients who were CMV seropositive, there was ‘a significant improvement in overall symptoms (P<0.0005)’.82 No such improvement was noted for patients who were seropositive for other herpes viruses.
In the other treatment study, 50 outpatients with an acute exacerbation of schizophrenia were treated with adjunctive celecoxib in a double-blind trial. Those receiving the drug had a significant improvement in their symptoms, especially between 2 and 4 weeks (P=0.001).83 Celecoxib is an inhibitor of cyclooxygenase (COX-2),84 and it is known that inhibition of COX-2 blocks the replication of CMV.85 Based on the results of these two treatment trials, additional tests are underway.
T. gondii and CMV are examples of infectious agents that may be plausibly linked to schizophrenia. They are a good fit epidemiologically, including a worldwide distribution and a peak exposure. They are known to be neurotrophic and to alter human behavior. They are capable of infecting individuals in early life and are thus consistent with neurodevelopmental theories of schizophrenia. They are also capable of reactivation in early adulthood, the peak time for the onset of the symptoms of schizophrenia. And they are known to be suppressed, to some degree, by antipsychotic medications. They thus represent examples of microbes that may have an etiological relationship to a major psychiatric disorder.
There are two major limitations to any infectious theory of schizophrenia: timing and causality: Since most studies rely on the detection of infection in individuals who already have schizophrenia, there has been difficulty in distinguishing whether an increased rate of exposure to an infectious agent is a cause or an effect of the altered behavior that is characteristic of schizophrenia. For example, it might be argued that some of the concomitants of schizophrenia, such as hospitalization, homelessness and the administration of medications, might alter the host immune response and result in an increased exposure to an infectious agent independent of the factors that led to the initiation of the disease itself. This problem is generic to all potential associations between infection and chronic disease defined in cross-sectional case control studies, since it can be difficult to determine if the measured exposure occurred before, during or after the initiation of the disease process. The problem can be compounded by the difficulty in identifying unaffected controls who do not have the disease but who are likely to be similar to the cases in terms of social and demographic factors that are related to exposures to infectious agents.
The problem of timing has been partly addressed by the evaluation of individuals with ‘recent onset’ schizophrenia who have not been previously hospitalized or, in some cases, have not received medications. Several studies in such populations have found increased evidence of exposure to T. gondii as well as to other infectious agents. These studies, however, might also be affected by exposures occurring around the time of disease onset, particularly in individuals with prolonged periods of undiagnosed and untreated psychosis.
For this reason, there has been an increased recognition of the importance of prospective cohort studies for the evaluation of the risk of environmental exposure in complex disorders. Such cohorts have proven to be invaluable in addressing the long-term effects of diet, smoking and other toxic exposures on chronic diseases such as heart disease, stroke and cancer. Of particular importance in terms of the study of the role of infectious agents in schizophrenia are cohorts in which blood samples have been obtained and stored, since such samples can be retrieved after the diagnosis of the disease and used to assess pre-disease exposure to infectious agents by the measurement of specific antibodies.
One type of cohort study that has been employed to examine the role of exposures in the subsequent development of schizophrenia has involved the study of pregnant women and their offspring. These have either been cohorts specifically set up for the study of diseases or population-based cohorts that have made use of neonatal blood samples obtained for the analysis of neonatal metabolic diseases and then stored for future studies. In two prospective neonatal cohorts, it was found that perinatal exposure to T. gondii resulted in an increased risk of schizophrenia in later life. Additional perinatal exposures that have been found to be associated with increased risk of subsequent schizophrenia in cohort studies include rubella virus, influenza viruses, enteroviruses and HSV-2 in some populations but not others. Increased risk of schizophrenia has also been associated with nonspecific markers of inflammation such as cytokines and with other inflammatory conditions such as pre-eclampsia.
Perinatal cohort studies have been important in identifying exposures in the prenatal period associated with risk of schizophrenia in later life. However, these studies cannot evaluate exposures beyond the neonatal period. Similarly, cohorts of older adults, such as the Framingham or Cache County studies, are directed largely at the study of diseases that occur in later life and enroll the majority of their subjects after the age of risk of schizophrenia. Fortunately, there are cohorts of young adults that can be employed to study diseases such as schizophrenia with onset in young adulthood. The most widely employed to date has been the cohort of individuals in the US Military maintained by the Army Medical Surveillance Activity. This cohort has been employed to study the role of Epstein–Barr infection in multiple sclerosis as well as other associations with infectious diseases. Recently, researchers have identified individuals who developed schizophrenia while they were in the military and measured antibodies contained in archived blood samples obtained both before and after the onset of their symptoms. As noted previously, preliminary analysis of these samples documented an increased level of antibodies to T. gondii in samples obtained approximately 2 years prior to the diagnosis of schizophrenia. This study, combined with the perinatal studies and the studies of recent-onset individuals cited above, strongly suggests that a substantial part of the exposure to Toxoplasma in individuals with schizophrenia occurs prior to the onset of schizophrenia and is not solely a result of post hoc exposures. Such studies provide the framework for attempts to prevent schizophrenia through the prevention of Toxoplasma infection or the treatment of infected individuals.
The other major limitation is related to the difficulty of proving causality. The interactions between infectious agents and host factors resulting in complex disorders such as schizophrenia generally do not follow the rules laid down by Robert Koch and his co-workers at the end of the nineteenth century. These rules, generally known as Koch's postulates, require that there be a one-to-one correspondence between an infectious agent and a disease process. It is clear, however, that the relationship between infectious agents and chronic diseases often does not follow the straightforward associations defined by Koch's postulates but rather follows a more complicated course comprising multiple pathways leading to diverse clinical outcomes. These pathways can involve more than one infectious agent resulting in the same disease, as well as many individuals who are infected but who do not develop the target disease in their lifetime. This latter phenomenon is based on a number of factors, including the timing of infection, the nature of the infecting strain and the genetic makeup of the host.
Guidelines for evaluating the relationship between an infectious agent and a chronic disorder to supplant Koch's postulates have been devised by a number of authors. The most widely cited are those devised by British statistician Austin Bradford Hill published in 1985, which identify quantitative degrees of association between exposure and disease and incorporate biological plausibility and coherence into the criteria.86 The extent to which the association between Toxoplasma infection and schizophrenia fulfill the Hill criteria has been the subject of a recent review.87
Specifically, with regard to T. gondii, the association between Toxoplasma infection and schizophrenia fulfills many of the criteria, including strength, consistency, temporality, plausibility and analogy. The main factor that is missing is direct evidence that the replication of Toxoplasma organisms is associated with ongoing symptoms of schizophrenia. The most straightforward way to test this hypothesis would be to demonstrate that the inhibition of Toxoplasma replication by pharmacological means would result in a decrease in the symptoms of schizophrenia and an alteration in the clinical course of the disease. In the past, this approach has not been feasible because of the lack of a safe and effective method for the treatment of Toxoplasma that would be suitable for administration to individuals with schizophrenia. Recently, however, it has been shown that derivatives of artemisinin, which has proven to be safe and effective for the treatment of malaria, also inhibit the replication of Toxoplasma, which shares a substantial portion of its genome with malaria organisms.88 A number of clinical trials examining the efficacy of artemisinin derivatives in individuals with schizophrenia are in progress. Their successful conclusion would provide the most direct evidence for a role of Toxoplasma in the etiology of schizophrenia and would open a new era in methods for the diagnosis, prevention and treatment of this devastating illness.
About this article
Translational Psychiatry (2017)