Is Relapse in Schizophrenia an Immune-Mediated Effect?

Schizophrenia is commonly a chronic, debilitating disorder with life-long consequences for affected individuals. Schizophrenia is heterogeneous with respect to clinical presentation, disease course, and outcome (1). However, the clinical course is often characterized by recurrent relapses, which are associated with adverse outcomes, including treatment-resistant symptoms, cognitive decline, and functional disability. Thus, to better understand relapse in schizophrenia is a compelling opportunity and a public health priority.

Immune system abnormalities in schizophrenia, including inflammation, have been one of the more enduring findings in the field, albeit with significant heterogeneity in the results, including negative studies. Nonetheless, multiple lines of evidence support an association between immune system dysfunction and schizophrenia. Polymorphisms in immune system-related genes, including cytokines (2) and the major histocompatibility complex (3–5) are associated with increased risk of schizophrenia. Prenatal maternal infections with a variety of agents are a replicated risk factor for schizophrenia (6). There is also an increased prevalence of autoimmune disease in both patients with schizophrenia and their first-degree relatives (7).

In this review, we will not focus on the relationship between the immune system and risk of schizophrenia. Rather, we ask a related, but important question that has not been reviewed in the literature: regardless of the underlying etiology, is relapse in schizophrenia an immune mediated effect? A number of recent findings support the plausibility of this association in some patients with schizophrenia. Several randomized, double-blinded trials in relapsed patients found that adjunctive nonsteroidal anti-inflammatory drug (NSAID) treatment significantly improved psychopathology (8–12), and that blood cytokine levels were a predictor of treatment response (10, 13). In a meta-analysis, we found that serum levels of some cytokines are increased in patients with first-episode drug-naïve psychosis (FEP) and acute relapse of schizophrenia, and decreased following treatment for relapse, suggesting a state-related effect that is independent of antipsychotic medications (14). Another recent study found that some patients with FEP have potentially pathogenic autoantibodies to central nervous system (CNS) antigens in the absence of overt signs of encephalitis (15). These associations raise the possibility of immune-based treatments for relapse (and/or relapse prevention) in a subset of patients with schizophrenia.

In this paper, we present a selected review of studies of immune system abnormalities in acute psychosis, in patients with FEP and/or relapse of chronic schizophrenia, including cytokines, the acute phase response, leukocyte subsets, autoantibodies, and markers of blood-brain barrier (BBB) dysfunction. As a comprehensive review of each of these areas of immune function is beyond the scope of the present paper, we will instead focus on replicated positive findings. We present a theoretical framework that attempts to integrate these findings and suggest potential mechanisms whereby immune system dysfunction might mediate relapse in some patients with schizophrenia. We also discuss limitations of the current literature and suggest future research directions.

Cytokines are key regulators of inflammation—the complex response of blood vessels to injury—that involves activation and recruitment of immune cells, and increased blood supply and vascular permeability. They coordinate both innate (e.g. granulocytes, monocytes/macrophages, and natural killer cells) and adaptive (e.g. B- and T-lymphocytes) arms of the immune system. Cytokines are key signaling molecules of the immune system that exert effects by binding specific cytokine receptors on a variety of target cells in the periphery and the brain. Soluble forms of cytokine receptors can either inhibit (e.g. soluble interleukin-2 receptor [sIL-2R]) or enhance (e.g. sIL-6R) the biological activity of cytokines. Endogenous cytokine receptor antagonists (e.g. IL-1 receptor antagonist [IL-1RA]) compete with cytokines for membrane receptors. Cytokines are also key regulators of the acute phase response (detailed below).

In a meta-analysis of 29 studies, we found that effect sizes for differences in serum cytokine levels between patients and controls were similar in magnitude and direction for relapse of chronic schizophrenia and FEP, suggesting an association that is independent of antipsychotic medications (14). IL-1β, IL-6, and TGF-β appeared to be state markers of acute psychosis, as levels were increased in acute psychosis, and decreased with antipsychotic treatment for relapse. In contrast, IL-12, interferon-γ (IFN-γ), tumor necrosis factor- α (TNF-α), and sIL-2R appeared to be trait markers, as levels remained elevated in acute psychosis and following antipsychotic treatment.

The findings in acute psychosis were most replicated for IL-6, which was significantly increased in 5 of 6 studies of relapsed patients and 4 of 4 studies of FEP, and significantly decreased following antipsychotic treatment in 3 of 5 studies in the meta-analysis. Serum IL-6 levels were significantly positively correlated with total psychopathology scores at baseline and following antipsychotic treatment in two studies (16, 17). A recent study of relapsed patients with schizophrenia replicated the pattern of results from the meta-analysis for serum IL-6 levels (18), and another study found increased IL-6 messenger ribonucleic acid (mRNA) levels in acute psychosis (19). Cerebrospinal fluid (CSF) IL-6 levels are also increased in relapsed patients with schizophrenia, particularly those with delayed response to antipsychotic treatment (20). Interestingly, unaffected adult carriers of a mutation in neuregulin 1 (NRG1) associated with schizophrenia have abnormal immune function, including increased in vitro IL-6 production and IL-6 mRNA (21). Increased cytokine levels were a predictor of response to adjunctive treatment with NSAIDs in two studies (10, 13). Further investigation of these relationships may even pave the way for targeted immune-based therapeutic interventions for acute psychosis, such as anti-IL-6 biopharmaceuticals.

Changes in serum cytokine levels may also precede relapse in schizophrenia. In 36 patients with schizophrenia who underwent weekly assessments for 1 year, in vitro IL-2 production plus antihippocampal immunoglobulin G (IgG) levels from the previous week significantly predicted relapse in three of seven patients (22). Similarly, among 64 male subjects with schizophrenia, increased CSF IL-2 levels following haloperidol withdrawal were a significant predictor of acute psychotic relapse (23). Longitudinal, intraindividual changes in other serum cytokine levels in stable patients have not been explored as a predictor of relapse in schizophrenia. If changes in serum cytokine levels are found to predict relapse, this marker could be evaluated in a variety of relapse prevention efforts.

The acute phase response is the body’s compensatory reaction to disturbances in homeostasis due to factors such as infection, tissue injury, and even psychosocial stress. At the local site of abnormal homeostasis, the acute phase response includes activation of leukocytes, fibroblasts, and vascular endothelium. These activated cells produce proinflammatory cytokines, including IL-6 and IL-1β, which can mediate a systemic response that may include hypercortisolemia, leukocytosis, complement activation, increased secretion of immunoglobulins, and hepatic synthesis of acute phase proteins (24).

Several replicated findings are consistent with an acute phase response in patients with FEP and/or relapse of chronic schizophrenia. Hypercortisolemia has been reported in patients with FEP (25, 26). Following antipsychotic treatment of patients with FEP, one study reported a significant decrease in serum cortisol levels (26), and another found a significant positive correlation between decreases in serum cortisol levels and improvements in psychopathology (27). Importantly, in this study there was no correlation between perceived stress and serum cortisol levels, suggesting that the findings are not due to potential increased stress associated with acute psychosis.

Increased peripheral blood white blood cell (WBC) counts have been reported in acute psychosis compared with control subjects (28, 29), and in one of these studies there was a significant decrease in WBC counts following antipsychotic treatment for relapse. Increased serum levels of complement C3 (30-32) and C4 (31, 32) have also been reported. One study found that C3 levels were significantly positively correlated with Positive and Negative Syndrome Scale (PANSS) negative subscale scores (33). Three studies have reported increased total serum IgM and IgG levels (34–36), and lower IgG levels may be a predictor of greater clinical improvement (34). One study found increased erythrocyte sedimentation rate (ESR) with no known cause in 17% of patients with acute psychosis (37). Following eight weeks of antipsychotic treatment for relapse, ESR normalized in two-thirds of these patients, concomitant with decreased psychopathology.

Serum levels of acute phase proteins are also abnormal in patients with acute psychosis. In a meta-analysis of 8 studies, we found increased CRP levels in patients with schizophrenia (38). Since IL-6 and, to a lesser extent, IL-1β are known inducers of CRP, we hypothesized that CRP levels would be increased in acute psychosis, which is supported by several studies. Shcherbakova et al (39) found significantly increased CRP levels in acutely relapsed males compared with controls. Ohaeri et al. (40, 41) found significant decreases in CRP levels following resolution of acute psychosis. Mazzarello et al. (42) found higher mean CRP levels among patients with an episodic illness course compared with continuously ill subjects. However, no studies have simultaneously measured blood CRP and cytokine levels in acute psychosis. Although the findings have not been replicated, levels of several other proteins consistent with an acute phase response are also abnormal in acute psychosis, including increased alpha-1-antitrypsin (39), ceruloplasmin (43), and haptoglobin (32), and decreased albumin (44). In another study, ceruloplasmin levels were significantly positively correlated with PANSS negative and general symptom subscales (33). Taken together, many findings in relapse of schizophrenia are consistent with changes seen in the acute phase response.

In addition to increased WBC counts (28, 29), there are a number of replicated abnormalities of blood leukocyte subsets in FEP and/or relapse of schizophrenia. Increased blood monocytes have been reported in acute psychosis (29, 45), and another study of intraindividual changes in blood monocyte levels found an association between monocytosis and worsening of psychotic symptoms (46). Consistent with these findings, two studies found an increased proportion of (monocyte-derived) macrophages in the CSF during acute psychosis compared with controls (47, 48), that normalized in some patients following antipsychotic treatment for relapse (47). As described above, monocyte activation and cytokine production are part of the acute phase response.

Blood levels of lymphocyte subsets are also abnormal in acute psychosis. Three studies have found increased antibody-producing blood CD19+B-lymphocytes in acute psychosis (49–51), with a significant decrease in levels following antipsychotic treatment for relapse (49, 51). Blood total (CD3+) T-lymphocytes are also increased in acute psychosis (28, 52), and are further increased following treatment for relapse (49, 50, 52). One study found that in patients with clinically significant positive symptoms, a higher proportion of lymphocytes in the WBC differential predicted clinically significant improvement (53). There is also evidence for activated lymphocytes in the CSF during acute psychosis (48). Furthermore, dopamine can directly activate T-lymphocytes by binding T-cell D2 and D3 receptors, thereby potentially modulating T-lymphocytes in the CNS (54).

There are also alterations in T-lymphocyte subsets in acute psychosis. CD4+T-helper lymphocytes (Th cells) initiate and maintain immune system responses, including maturation of B-lymphocytes and activation of cytotoxic T-lymphocytes and macrophages. CD8+cytotoxic T-cells (Tc cells) can destroy virally infected cells and tumor cells. When activated, natural killer T-lymphocytes (NK cells) can perform functions ascribed to both Th and Tc cells

Three studies have found increased blood levels of Th cells in acute psychosis (28, 52, 55), although changes in levels following antipsychotic treatment for relapse have been inconsistent. The ratio of blood CD4+/CD8+T-lymphocytes is also increased in acute psychosis (28, 52). Consistent with this observation, another study found a significantly lower percentage of Tc cells in the CSF of relapsed patients compared with controls, although the authors noted a wide distribution of the proportion of Th and Tc cells in the patient group (56). Increased blood levels of NK cells during acute psychosis have also been found (28, 57), although changes in levels following antipsychotic treatment for relapse have varied. Consistent with these findings, serum levels of neopterin, a marker of cell-mediated immunity, including activation of macrophages and NK cells, are increased in patients with acute psychosis (58, 59), and decreased following antipsychotic treatment in one study (58). Few studies have simultaneously measured blood cytokines and leukocyte subsets (an important source of serum cytokines), which limits the ability to make broader inferences regarding immune function.

Both patients with schizophrenia and their first-degree relatives have an increased prevalence of autoimmune disease, which are associated with autoantibodies (7). Even in the absence of comorbid autoimmune disease, there is evidence for increased blood autoantibody levels in acute psychosis, consistent with increased levels of antibody-producing blood CD19+B-lymphocytes described above. Serum levels of platelet autoantibodies (PAA) are increased in both younger and adult patients with acute psychosis (60, 61). There is also a case report of a patient with chronic schizophrenia treated with the immunosuppressant azathioprine who had significant clinical improvement preceded by a decrease in PAA levels (62).

Although not replicated, there is evidence for increased serum antinuclear (ANA), smooth muscle (SMA; 63), thyroid peroxidase (TPO; 64), and heat shock proteins 70 kDa (HSP70; 65) autoantibodies in relapse of schizophrenia. HSP70 antibody titers significantly decreased after 6 weeks of antipsychotic treatment, and patients with higher baseline titers had higher baseline Brief Psychiatric Rating Scale scores and greater clinical improvement. Interestingly, patients with schizophrenia and a history of obstetric complications have a significantly higher prevalence of serum autoantibodies, including ANA, SMA, and TPO (66). Furthermore, unaffected carriers of the NRG1 mutation described above also had an increased prevalence of ANA, HSP, and TPO autoantibodies (21). Another study found that patients with positive ANA or rheumatoid factor titers were more likely to have clinically significant negative symptoms (53).

In a sample of 46 patients with FEP, Zandi et al. (15) retrospectively found four patients with either NMDA receptor or voltage-gated potassium channel autoantibodies, which are associated with limbic encephalitis. However, all four cases fulfilled DSM-IV criteria for schizophrenia, and none of these patients had any neurological signs or symptoms. These findings, particularly if replicated, have important potential implications for the evaluation and treatment of a subset of patients with FEP. Indeed, in conditions outside of schizophrenia other autoantibodies are associated with psychosis, such as antiribosomal P antibodies in lupus psychosis (67). Interestingly, patients with active (versus inactive) systemic lupus erythematosus (including central nervous system disease) have significantly higher blood IL-6 and IFN-γ levels (68).

Some serum cytokines, particularly IL-1RA and IL-6 can cross the BBB and bind target cells and affect CNS function (69). Serum S100B is a marker of astrocyte activation and BBB dysfunction (70). Elevated serum levels of S100B have been reported in 5 studies of patients with acute psychosis compared with controls (70–74). S100B levels were significantly positively correlated with PANSS total scores in two studies (71, 73), and PANSS negative subscale scores in one study (71). Also, serum S100B levels significantly decreased after 6 weeks of treatment for acute psychosis in one study (72), and the subgroup of patients with persistently elevated S100B levels had significantly higher PANSS negative subscale scores. Furthermore, increased CSF levels of S100B have been reported in acute psychosis (73, 75). Another marker of BBB disruption is an elevated CSF/serum albumin ratio, which was in three studies was abnormal in 19% (76), 22% (77), and 53% (78) of patients with acute psychosis, respectively.

Soluble intracellular adhesion molecule 1 (sICAM-1) is a marker of BBB damage and intrathecal immune activation. Paradoxically, three studies by the same group found decreased sICAM-1 in relapse of schizophrenia (77, 79, 80), although there are several caveats to these findings. In one study, patients with elevated HSP 60 kDa antibody titers had higher levels of sICAM-1 and sIL-2R (80). Another study found a significant positive correlation between sICAM-1 levels and PANSS negative subscale scores and a trend for higher sICAM-1 levels in patients with BBB impairment based on lumbar puncture (elevated CSF/serum albumin; 77). Lastly, a third study found higher levels of lymphocyte-function-associated antigen 1 (LFA-1+) Th cells, which binds sICAM-1, in patients with relapse of schizophrenia and BBB impairment (76). Thus, these findings are largely consistent with other evidence for BBB dysfunction in a subset of patients with acute psychosis.

In order to summarize the studies reviewed above, Figure 1 presents a theoretical framework that attempts to integrate findings and relate them to potential mechanisms whereby immune system dysfunction might mediate relapse in some patients with schizophrenia. Replicated findings, as well as immune parameters that are significantly correlated with psychopathology in acute psychosis and/or “normalize” following treatment for relapse, are highlighted. This model is not intended to be comprehensive, but rather to serve as a catalyst for critical thinking and to suggest potential areas for future research. In brief, abnormal homeostasis results in cellular activation and proinflammatory cytokine production, which in turn stimulates an acute phase response. In the setting of increased BBB permeability, increased autoantibodies may directly cross-react with CNS antigens, or cytokine abnormalities may directly modulate dopaminergic neurotransmission or indirectly modulate glutamatergic neurotransmission through tryptophan catabolism, resulting in acute psychosis.

The potential for cross-reactivity between autoantibodies with CNS antigens has already been described above. Cytokines may also mediate acute psychosis. Proinflammatory cytokines can modulate neurotransmitter function, as systemic increases in serum IL-6 in adult rodents modulate dopamine turnover and sensitization to amphetamine-induced locomotion (81–83). By contrast, indoleamine 2,3-dioxygenase (IDO), the rate-limiting enzyme in tryptophan catabolism, is also expressed in astrocytes and microglia, and its activity can be modulated by cytokines. IDO induction results in increased production of kynurenine, which is converted in astrocytes to the NMDA receptor antagonist kynurenic acid (KYN-A). NMDA receptor hypofunction has been implicated in the pathophysiology of schizophrenia (84, 85). Previous studies have found increased blood (86), CSF (87), and postmortem brain (88) levels of KYN-A, as well as increased IDO activity (89) in patients with schizophrenia.

An important consideration in this field is, “Acute psychosis is stressful, and stress can alter immune function, so are the observed abnormalities due to increased stress?” Furthermore, a related concern is, “Many individual studies did not control for potential factors known to influence various immune parameters, including antipsychotic medications, body mass index, and smoking (90), so is the observed immune dysfunction due to confounding by these factors?” Several lines of evidence argue against this reasoning. Many of the immune abnormalities described in this review have been found in drug-naïve patients with FEP, suggesting an effect that is independent of antipsychotic medications. Furthermore, some studies have found immune abnormalities in acute psychosis after controlling for multiple potential confounding factors, for example increased IL-6 (91) and CRP (92). Several studies reviewed here also found significant correlations between immune abnormalities and psychopathology, particularly total and negative symptoms, and we have reviewed putative mechanisms by which immune dysfunction could impact on psychopathology. Longitudinal studies described here also suggest that immune abnormalities normalize to some extent following antipsychotic treatment for relapse, implying that certain immune parameters may be state-dependent markers of acute psychosis. An intriguing study found potentially pathogenic CNS autoantibodies in a subset of patients with FEP, suggesting the potential for immune-based treatments in these patients (15). Lastly, adjunctive NSAIDs have been efficacious in improving psychopathology in relapsed patients (8–12). Taken together, these findings support the plausibility that acute psychosis in some patients with schizophrenia may be an immune mediated effect.

Despite the plausibility of our hypothesis, the literature in this field is fraught with significant heterogeneity, including contradictory findings. As an example, one study found increased CD3+T-lymphocytes that decreased with treatment in patients with FEP (28), whereas two other studies found decreased CD3+T-lymphocytes that increased with treatment (49, 51). How then do we reconcile and interpret these discrepant results? Importantly, it should be emphasized that schizophrenia is a very heterogeneous disorder. The extant literature suggests that immune dysfunction may mediate relapse in a subset of patients with schizophrenia; thus, future research should aim to further characterize this group of patients, which can reduce the “signal-to-noise” ratio in future analyses. Several approaches will facilitate accomplishing this task. Future studies of immune function in schizophrenia must control for potential confounding factors, including age, race, sex, BMI, smoking, duration of illness, and psychotropic medications. Additional studies in drug-naïve patients with FEP will also be indispensable. Most existing studies have focused on a single immune parameter (i.e. cytokines, acute phase reactants, or leukocyte subsets only). Future studies should concomitantly measure multiple immune parameters in individual patients, to begin to explore the validity of potential causal pathways. Longitudinal studies with serial measurements of immune parameters in stable patients (to explore potential relapse-predictive markers) as well as patients with acute psychosis (as a marker of response to treatment) are also needed.

Bilbo and Schwarz (93) hypothesized that maternal inflammation during critical periods of neurodevelopment may sensitize neural substrates and permanently alter the “set-point” of the immune system in adult offspring with schizophrenia. Indeed, a rodent prenatal maternal inflammation model found age-dependent increases in serum levels of IL-2, IL-6, and TNF-α, in the exposed offspring, which were decreased with haloperidol treatment (94, 95). Birth cohorts with archived maternal serum and other prenatal data—and the ability to identify cases of schizophrenia through register linkage—would afford unique opportunities to characterize patients in which relapse may be immune mediated. Future studies could also investigate the effects on immune function of genes associated with both schizophrenia and the immune system, including brain-derived neurotrophic factor, V-akt murine thymoma viral oncogene homolog 1, methylenetetrahydrofolate reductase, and phosphodiesterase 4B (2).

Recently, an increased understanding of the complex interactions between immune dysfunction and the brain in other chronic diseases has better informed this relationship in schizophrenia. There is growing empirical support that immune dysfunction may mediate relapse and response to treatment in some patients with schizophrenia. These findings also fit into the broader context of how a better understanding of the interface between immunology and chronic disease can guide and drive the delivery of clinical care, the advent of monoclonal antibody therapies for several human cancers being a prime example (96). The characterization of a subset of patients with schizophrenia in whom relapse is immune-mediated could be used to assess treatment effectiveness, advance relapse prevention efforts, and potentially even lead to future immune-based therapeutic interventions, thereby reducing the burden of relapse and improving clinical care of patients with schizophrenia.