Correspondence: Sonnig Sue-Whei Chiang, PhD, Institute of Cognitive Neuroscience of National Central University, #300, Jhongda Road, Jhongli City, Taoyuan County 32001, Taiwan. Email: firstname.lastname@example.org
An imbalance between T-helper type 1 (Th1) and type 2 (Th2) cytokines has been implicated in schizophrenia, although empirical evidence is rare. The aim of this study was to examine if a Th1/Th2 imbalance occurs in schizophrenia and schizophrenia-related disorder.
Twenty-six subjects with schizophrenia, 26 subjects with schizophrenia-related disorders, and 26 healthy controls were recruited. The Human Th1/Th2 Cytokine Cytometric Bead Array Kit-II was utilized to assess serum Th1/Th2 cytokines and ratios simultaneously. MANOVA was used to detect differences among the three diagnostic groups in distinct Th1/Th2 cytokines/ratios. Pearson/Spearman correlations were used to examine the relationships between distinct Th1/Th2 cytokines/ratios and clinical/psychopathological data in schizophrenia.
Interferon (IFN)-γ/interleukin (IL)-4, IFN-γ/IL-10, IL-2/IL-4, and tumor necrosis factor (TNF)-α/IL-4 ratios were significantly decreased in schizophrenia, but not in schizophrenia-related disorders compared to healthy controls. IFN-γ/IL-4 and IFN-γ/IL-10 in schizophrenia subjects positively correlated with age, but not in schizophrenia-related disorder subjects or in healthy controls.
A clear Th2 shift was observed in schizophrenia, but not in schizophrenia-related disorders. The Th2 shift in schizophrenia appeared to be an aberrant developmental phenomenon.
SCHIZOPHRENIA (SCH) HAS been associated with immune dysfunction, including aberrant cytokine levels, particularly in the so-called T-helper type 1 (Th1) and type 2 (Th2) cytokines. Th1 and Th2 cytokines have been recently found to regulate neuronal gene expression. The balance between these two Th subpopulations plays a pivotal role in optimizing immune response during viral infection,and diverse viral infections increase the risk of SCH. Prolonged psychosis is associated with interferon (IFN)-α therapy, and IFN-α modulates Th1/Th2 balance. Moreover, the immunological effects of many existing antipsychotics rebalance in part the immune imbalance. Dopamine, the neurotransmitter most frequently implicated in SCH, was also found to polarize Th2 differentiation. Therefore, Th1/Th2 imbalance might be involved in the etiology of SCH through its associations with dopamine and viral infection.
The Th1 cells mainly produce IFN-γ, interleukin-2 (IL-2), and tumor necrosis factor-α (TNF-α), while Th2 lymphocytes predominantly release IL-4, IL-10, and IL-6. Due to the roles of IL-4, IL-10, and IFN-γ in the initiation, development, and regulation of the Th1/Th2 system, the IFN-γ/IL-4 and IFN-γ/IL-10 ratios have been deemed indicators of Th1/Th2 immunity. Although Th1/Th2 imbalance has been implicated in SCH,[14-16] the vast majority of relevant studies examined individual Th1/Th2 cytokines.[15, 17, 18] Only two studies examined IFN-γ/IL-4 in SCH; one found unchanged in vitro ratios, while the other noted increased in vivo ratios.
To date, no studies have compared the Th1/Th2 cytokine ratio in SCH to that in schizophrenia-related disorders (SCH-R). Based on the Th2-shift hypothesis, we examine whether Th1/Th2 imbalance was observed in SCH as well as SCH-R when compared to healthy controls (CON). The indicators of Th1/Th2 balance included IFN-γ/IL-4 and IFN-γ/IL-10 ratios. In addition, IL-2 and IL-4 are probably involved in the regulation of Th1/Th2 balance/activity in the brain and brain function. Thus, the IL-2/IL-4 ratio was also investigated. Furthermore, TNF-α/IL-4was found to be altered in SCH. Therefore, TNF-α/IL-4 ratio was also examined in the present subjects.
Altogether, 78 subjects participated in this study: 26 SCH, 26 SCH-R, and 26 CON. Each group contained 15 women and 11 men. Among the 26 SCH subjects, 21 had paranoid SCH (12 women, 9 men) and five (2 men, 3 women) had disorganized SCH. The SCH-R group consisted of the following: psychotic disorder due to cannabinoid use, n = 1; psychotic disorder due to use of multiple drugs/psychoactive substances, n = 1; persistent delusional disorder, n = 1; schizoaffective disorder, n = 9; adjustment disorder with obsessional thoughts/acts, n = 1; acute polymorphic psychotic disorder with schizophrenic symptoms, n = 4; acute/transient psychotic disorder, n = 4; schizotypal disorder, n = 1; and acute schizophrenia-like disorder, n = 4.
After a complete explanation of the study, the subjects gave written informed consent. A thorough medical–physical examination was then conducted to exclude those subjects suffering from severe medical disease, acute allergy, inflammatory disorder, autoimmune disease, or clinically apparent infection. Criteria for SCH consisted of the following: (i) SCH diagnosis according to ICD-10 and DSM-IV; (ii) no history of psychotropic substance addiction or abuse except nicotine and caffeine; (iii) no personality disorders in line with DSM-IV; and (iv) antipsychotic drug free for at least 1 week. The inclusion criteria for SCH-R were the same as those for SCH except for diagnosis: a diagnosis of SCH-R was mandatory. Furthermore, two essential criteria were required for CON: (i) no psychiatric disorder; and (ii) no first- or second-degree biological relatives with (or had had) any psychiatric disease.
Serum cytokine measurement
Serum samples from each participant were drawn between 8 and 9am. The serum was immediately centrifuged at 6°C at a speed of 3209 g for 10 min using Omnifuge 2.0 RS (Heraeus Sepatech, Hanau, Germany) and stored at −80°C until use. The Human Th1/Th2 Cytokine Cytometric Bead Array (CBA) Kit-II (Becton Dickinson Pharmingen, San Jose, CA, USA) was used to assess serum Th1/Th2 cytokine levels, including IFN-γ, IL-2, TNF-α, IL-4, IL-10, and IL-6. CBA shares many similarities with enzyme-linked immunosorbent assay (ELISA). The advantage of CBA over conventional ELISA is the comparability among the cytokines measured, because the same serum sample is utilized to measure the aforementioned six cytokines simultaneously; that is, there are no inter-assay variances among diverse cytokines for the same subject. CBA is particularly suitable to examine Th1/Th2 balance, which involves the relative relationship between two distinct cytokines. Serum cytokines were assessed, following the procedure as described in the manufacturer's manual. The only minor modifications were to the volumes of serum, capture bead mixture, and phycoerythrin detection reagent. Instead of 50 μL as the manual required, 100 μL of each was utilized, but the volume ratios between the samples and diverse reagents remained unchanged. Moreover, the bottom standard was diluted to two levels lower than that described in the manual.
Data on the symptoms and psychopathology were collected via two structured clinical interviews, the Positive and Negative Syndrome Scale (PANSS) as well as the Clinical Global Impression Scale (CGI). The interviews were conducted by experienced psychiatrists.
The independent variable was the diagnostic group (divided into three levels: SCH, SCH-R, and CON). The dependent variables included the six individual cytokines and various major Th1/Th2 cytokine ratios.
Multivariate analysis of variance (MANOVA) was utilized to investigate the differences among SCH, SCH-R, and CON with regard to the Th1/Th2 cytokines and ratios. One-way ANOVA was used to examine diversity in age. All data for (M)ANOVA analysis were log-transformed in this study. Post-hoc comparisons were performed with Bonferroni adjustment. Pearson/Spearman correlations were applied to investigate the relationships between diverse Th1/Th2 cytokines and ratios and clinical variables/psychopathology. Student t-test was used to examine the differences between male and female SCH subjects with regard to clinical variables, PANSS subscale scores, and CGI scores. The evaluation was performed using SPSS version 13 (SPSS, Chicago, IL, USA).
Demographic and clinical data
The clinical data of some SCH subjects were missing. The average age and available clinical data for the SCH group are summarized in Table 1. One-way ANOVA showed no marked differences among the three groups for age (F(2,75) = 1.65, P = 0.20). Similar outcomes were also found in the three male (F(2,30) = 0.40, P = 0.68) and female subgroups (F(2,42) = 1.31, P = 0.28).
Table 1. Clinical data for SCH subjects
Male + Female
CGI, Clinical Global Impression scale (a, d, and diff, scores at admission, discharge, and difference between admission and discharge); PANSS, Positive and Negative Syndrome Scale (G, N, P, Global, Negative and Positive Psychopathology scale); SCH, schizophrenia.
Onset age (years)
Disease duration (years)
Independent samples t-tests identified no marked differences between both sexes of SCH patient in the length of hospital stay (t(22) = −0.30, P = 0.77), onset age (t(19) = −0.46, P = 0.65), disease duration (t(19) = −0.10, P = 0.92), PANSS-Positive subscale (t(19) = −0.73, P = 0.47), PANSS-Negative subscale (t(19) = 1.62, P = 0.12), PANSS-Global psychopathology (t(19) = −0.41, P = 0.69), CGI at admission (t(18) = 0.00, P = 1.00), CGI at discharge (t(18) = 1.12, P = 0.28), or CGI difference between admission and discharge (t(18) = −1.05, P = 0.31).
Serum Th1/Th2 cytokines and ratios
Data on Th1/Th2 cytokines and ratios are summarized in Table 2.
At the individual cytokine level, SCH, SCH-R, and CON were not significantly distinguishable (IFN-γ: F(2,75) = 1.92, P = 0.15; IL-2: F(2,75) = 1.28, P = 0.29; TNF-α: F(2,75) = 1.72, P = 0.19; IL-10: F(2,75) = 0.05, P = 0.95; IL-6: F(2,75) = 0.07, P = 0.93; IL-4: F(2,75) = 0.61, P = 0.55). These three diagnostic groups were markedly different, however, for IFN-γ/IL-10 (F(2,75) = 4.42, P = 0.02), IFN-γ/IL-4 (F(2,75) = 4.17, P = 0.02), IL-4/IL-2 (F(2,75) = 3.95, P = 0.02), and TNF-α/IL-4 (F(2,75) = 3.90, P = 0.03). Bonferroni post-hoc comparisons identified marked differences between SCH and CON in IFN-γ/IL-10 (P = 0.01), IFN-γ/IL-4 (P = 0.02), IL-4/IL-2 (P = 0.03), and TNF-α/IL-4 (P = 0.02). The differences, however, between SCH and SCH-R as well as those between SCH-R and CON were not significant. The comparisons of the standardized IFN-γ/IL-4, IFN-γ/IL-10, IL-2/IL-4, and TNF-α/IL-4 among these three diagnostic groups are presented in Figure 1.
Similarly, no significant differences were observed among the three male subgroups at the individual cytokine level (IFN-γ: F(2,30) = 2.00, P = 0.15; IL-2: F(2,30) = 1.93, P = 0.16; TNF-α: F(2,30) = 1.64, P = 0.21; IL-10: F(2,30) = 0.66, P = 0.53; IL-6: F(2,30) = 0.61, P = 0.55; IL-4: F(2,30) = 1.43, P = 0.26). Nevertheless, they were significantly distinguishable in IL-2/IL-4 (F(2,30) = 3.75, P = 0.04) and TNF-α/IL-4 (F(2,30) = 3.65, P = 0.04). The differences in IFN-γ/IL-10 (F(2,30) = 3.05, P = 0.06) and IFN-γ/IL-4 (F(2,30) = 3.14, P = 0.06) reached only a marginal level of statistical significance. Post-hoc Bonferroni comparisons identified (i) significantly different IL-4/IL-2 ratios between male SCH and CON (P = 0.05); and (ii) marginally different TNF-α/IL-4 between male SCH and CON (P = 0.07) as well as between male SCH and SCH-R (P = 0.08).
Among the three female subgroups, no clear differences were seen in any of the Th1/Th2 cytokines/ratios (IFN-γ/IL-4: F(2,42) = 1.58, P = 0.22; IFN-γ/IL-10: F(2,42) = 1.90, P = 0.16; IL-2/IL-4: F(2,42) = 1.04, P = 0.36; TNF-α/IL-4: F(2,42) = 2.09, P = 0.14; IFN-γ: F(2,42) = 0.67, P = 0.52; TNF-α: F(2,42) = 0.66, P = 0.52; IL-2: F(2,42) = 0.64, P = 0.53; IL-4: F(2,42) = 1.62, P = 0.21; IL-6: F(2,42) = 0.82, P = 0.45; IL-10: F(2,42) = 0.96; P = 0.39).
Th1/Th2 cytokines/ratios and clinical/psychopathological data in SCH
The relationships between Th1/Th2 cytokines and ratios and distinct clinical variables in SCH were also examined.
In SCH (male + female), marked correlations were found (i) between onset age and IFN-γ (r(19) = 0.52, P = 0.02) as well as IL-2 (r(19) = 0.52, P = 0.02); and (ii) between disease duration and IFN-γ/IL-4 (r(19) = 0.45, P = 0.04). In male SCH, significant associations were observed (i) between onset age and TNF-α (r(7) = 0.71, P = 0.03) as well as IL-2 (r(7) = 0.75, P = 0.02) and (ii) between disease duration and IFN-γ/IL-4 (r(7) = 0.67, P = 0.05), TNF-α/IL-4 (r(7) = 0.67, P = 0.05), as well as IL-2/IL-4 (r(7) = 0.72, P = 0.03). Other than these, no further significant associations were observed between clinical variables and any of the Th1/Th2 cytokines/ratios.
In addition, age appeared to be significantly associated with various Th1/Th2 ratios/cytokines in SCH (IFN-γ: r(25) = 0.51, P = 0.007; TNF-α: r(25) = 0.50, P = 0.01; IL-2: r(25) = 0.38, P = 0.05; IL-6: r(25) = 0.43, P = 0.03; IFN-γ/IL-4: r(25) = 0.50, P = 0.009; IFN-γ/IL-10: r(25) = 0.38, P = 0.05). In male SCH, age markedly correlated with distinct Th1/Th2 cytokines/ratios (IFN-γ: r(10) = 0.75, P = 0.007; TNF-α: r(10) = 0.70, P = 0.02; IL-2: r(10) = 0.91, P < 0.0001; IL-6: r(10) = 0.74, P = 0.009; IFN-γ/IL-4: r(10) = 0.66, P = 0.03; IL-2/IL-4: r(10) = 0.71, P = 0.02; TNF-α/IL-4: r(10) = 0.68, P = 0.02). Nevertheless, no marked correlations were seen in SCH-R or CON or female SCH.
The psychopathological data for some SCH subjects were missing. Spearman correlation coefficients for serum Th1/Th2 ratios and those available PANSS/CGI scores are summarized in Table 3.
Table 3. Correlation between Th1/Th2 ratio and psychopathology in SCH
At the individual cytokine level, significant correlations were found (i) between TNF-α and PANSS-Global (r(19) = 0.42, P = 0.05); (ii) between IL-2 and CGI at admission (r(19) = 0.50, P = 0.02), as well as (iii) between IL-10 and CGI-difference (r(19) = 0.44, P = 0.05) in SCH (male + female). In female SCH, marked associations were observed (i) between CGI-difference and IFN-γ (r(11) = 0.65, P = 0.02) as well as (ii) between CGI-difference and IL-10 (r(11) = 0.59, P = 0.04). Furthermore, marginally significant correlations were also seen (i) between IFN-γ and both PANSS-Positive (r(7) = −0.66, P = 0.06) and -Global subscales (r(7) = −0.65, P = 0.06) in male SCH; (ii) between TNF-α and PANSS-Global (r(11) = 0.56, P = 0.06); (iii) between IFN-γ and CGI at discharge (r(11) = −0.53, P = 0.08); as well as (iv) between IL-2 and CGI at admission (r(11) = 0.51, P = 0.09) in female SCH.
Generally speaking, at the individual serum cytokine level, the outcomes of this study corresponded to the majority of in vivo studies in this regard: SCH had unaltered in vivo Th1/Th2 cytokines.[15, 17, 18] Due to the main focus of this study, more details at the individual cytokine level will not be given here. SCH, however, did have significantly reduced Th1/Th2 ratios including IFN-γ/IL-4, IFN-γ/IL-10, IL-2/IL-4, and TNF-α/IL-4 when compared with CON. Nevertheless, no marked alterations in any of the Th1/Th2 cytokines/ratios were observed in SCH-R. That is, Th1/Th2 imbalance in favor of Th2 shift was seen in SCH, as indicated in previous reports.[14, 21, 25] The Th2 shift appeared to be specific for SCH.
In contrast to Miller et al., we found a Th2 shift in SCH; the reason for the difference might be that SCH in the Miller et al. study had included SCH-R, but SCH in the present study included only paranoid and disorganized SCH. The present finding regarding IFN-γ/IL-4 in SCH contradicts two previous studies. The reason might be that Avgustin et al. examined in vitro IFN-γ/IL-4, while we investigated in vivo ratios. The results from the in vitro production and those from the in vivo level could vary due to different biological systems involved. Although Kim et al. and the present investigated in vivo IFN-γ/IL-4 ratios, opposite results were obtained. In contrast to the Kim et al. study, we applied a modified CBA to measure cytokines (using flow cytometry) and did not have missing data. The Kim et al. results might due to the outcomes of selection because some of their SCH and CON subjects had cytokine levels under the detectable limits and had been, thus, excluded from analysis. Moreover, the Kim et al. SCH subjects were likely to have predominantly negative symptoms because significantly higher IFN-γ/IL-4 has recently been found in major depression. The IFN-γ/IL-4 balance might be involved in the etiology of affective symptoms that are observed in SCH and depression. In contrast to the O'Brien et al. results, we found reduced TNF-α/IL-4 in SCH. The age difference in the studies might be the reason for this. We found a positive correlation between TNF-α and age as well as between disease duration and TNF-α/IL-4 in SCH. Because the O'Brien et al. SCH subjects were on average almost 8 years older than the present subjects and had likely, therefore, longer disease duration, the TNF-α/IL-4 for the O'Brien et al. SCH subjects were correspondingly higher than the present results.
Some might argue that Th2 shift in the present SCH subjects may be a result of antipsychotic treatment, because antipsychotics could influence Th1/Th2 differentiation. The effects of antipsychotics on the Th1/Th2 system, however, are complicated and may vary with the class of drug: some suppress Th2 cytokines, while some others suppress or exert no effects on Th1 cytokines.[30, 31] Also, the cytokine levels reported here were measured at admission. If Th2 shift in the present patients was a result of neuroleptic medication, it must be from previous treatment. Nevertheless, we found a positive correlation between disease duration and Th1/Th2 ratio in SCH. That is, the longer since disease onset, the more the Th1/Th2 imbalance in the present SCH subjects shifted to Th1 than to Th2. Thus, antipsychotic treatment may not be the cause of Th2 shift in the present SCH subjects.
Reduced Th1/Th2 ratios in the present SCH subjects might have a genetic basis because altered gene expression of various Th1/Th2 cytokines has been found in SCH, including IFN-γ, TNF-α, IL-2, and IL-4. Genetic polymorphisms of distinct Th1/Th2 cytokines were also highly correlated with SCH, including IFN-γ, TNF-α, IL-2, and IL-10.[36, 37] In addition, environmental factors such as viral infections could have caused the Th2 shift in the present SCH subjects because viral infections have been closely associated with SCHand often result in activation of the Th1/Th2 systems. Moreover, gender dimorphism was likewise observed in the present study: male SCH subjects had a clearer serum Th2 shift than their female counterparts. This could be partly explained by the diverse roles of estrogen/testosterone in SCHand the altered level of testosterone in male SCH subjects because sex hormones could affect expression of Th1/Th2 cytokines. The significant correlations between the various Th1/Th2 ratios and psychopathology might indicate their roles in SCH due to: (i) the role of IL-2/IL-4 in regulating brain function and Th1/Th2 balance in the central nervous system,[22, 40]as well as (ii) the protective effects of IFN-γ/IL-4 and IFN-γ/IL-10 balance for astrocyte functionand neurite growth. Furthermore, corresponding to a previous study, a positive correlation between TNF-α and PANSS-Global was also found in the present SCH subjects. The link of TNF-α to SCH is likely via its role in neurogenesis/neuron function.
Additionally, we found close relationships between age and various Th1/Th2 cytokines/ratios in SCH, but not in SCH-R or CON. The older the SCH subjects were, the more the Th1/Th2 balance shifted towards Th1. Although SCH was on average the oldest group in this study, Th2 shift was observed. The question remains as to why Th2 shift was not observed in SCH-R. This is the first study examining Th1/Th2 cytokines in SCH/SCH-R as two separate groups. The Miller et al. review included studies investigating SCH/SCH-R and found no Th2 shift. SCH and SCH-R might be two related groups with distinct etiologies. The second reason is likely that SCH-R in this study represents a group of disorders with different etiologies. For example, drug-induced psychosis could involve Th2 shift, while the other(s) may not. Other possible reasons could be expression of different subtypes of IFN-α and/or indoleamine 2,3-dioxygenase (IDO) in SCH/SCH-R due to (i) distinct regulatory effects of diverse IFN-α subtypesand IDO on Th1/Th2 balance, as well as (ii) associations of IFN-α and IDO with psychosis.[7, 17] It is noteworthy that the present findings must be taken with caution due to the small sample size. A larger sample is required to confirm the outcomes. Despite the small sample size, however, Th2 shift was observed in SCH, but not in SCH-R; it may be related to the etiology of, at least, a subgroup of SCH. Questions remain as to how Th1/Th2 balance influences symptom severity and the general psychopathology in SCH.
In summary, serum Th2 shift was sexually dimorphic and observed in SCH, but not in SCH-R when compared to CON; it might be an abnormal development and related to the pathology of SCH.
This study is supported by the Theodore Vada Stanley Research Institute. We also thank Professor Daisy Lan Hung and Professor Chi-Hung Juan of the Institute of Cognitive Neuroscience of National Central University (Taiwan) for their support in publishing this article, and Dr Chun-Ming Shih for his assistance with statistical analysis.