Inflammatory Cytokine Alterations in Schizophrenia:
A Systematic Quantitative Review
Stéphane Potvin, Emmanuel Stip, Amir A. Sepehry, Alain Gendron, Ramatoulaye Bah,
and Edouard Kouassi
Background: Cytokines play an important role in infection and inflammation and are crucial mediators ofthe cross-talk between the brain
and the immune system. Schizophrenia would be associated with an imbalance in inflammatorycytokines, leading to a decrease in Th1 and
an increase in Th2 cytokine secretion. However, data published so far have been inconsistent. The primary objective of the present
meta-analysis was to verify whether the cytokine imbalance hypothesis of schizophrenia is substantiated by evidence.
Methods: Cross-sectional studies were included ifthey assessed in vivo plasma or serum cytokine concentrations and/or in vitro secretion
of cytokines by peripheral blood leukocytes from schizophrenia patients and healthy volunteers.
Results: Data from 62 studies involving a total sample size of 2298 schizophrenia patients and 1858 healthy volunteers remained for
analysis. Ten cytokines were assessed, including the prototypic Th1 and Th2 cytokines gamma interferon (IFN-?) and interleukin 4 (IL-4) as
well as IL-2, soluble IL-2 receptor (sIL-2R), IL-1?, IL-1 receptor antagonist (IL-1RA), tumor necrosis factor-alpha (TNF-?), IL-6, soluble IL-6
schizophrenia. No significant effect sizes were obtained for the other cytokines.
develops in schizophrenia.
Key Words: Autoimmunity, inflammatorycytokines,meta-analysis,
been proposed (1,2). One of the approaches to investigate this
literature is to examine evidence based on studies measuring
cytokines. Considered as the hormones of the immune system,
cytokines include a large and expanding number of proteins,
which are involved in regulation of immunologic and inflamma-
tory responses in physiologic and pathologic conditions (3).
Cytokines are also important mediators of the cross-talk between
the central nervous system (CNS) and the immune system, which
might have implications for clinical psychiatry (4,5). Examples of
cytokines include interleukins (IL), interferons (IFN), tumor
necrosis factors (TNF), transforming growth factors (TGF), and
chemokines. These molecules are synthesized and secreted by a
variety of cell types, including not only immune cells such as T
lymphocytes, natural killer (NK) cells, dendritic cells, polymor-
phonuclear leukocytes, monocytes/macrophages, and microglia
but also non-immune cells, such as fibroblasts, endothelial cells,
adipocytes, and neurons. Cytokines are key players in the
coordinate responses of cells of the innate immune system (e.g.,
mong etiological explanations of schizophrenia, several
hypotheses concerning immune-related disorders such as
infections and autoimmune inflammatory diseases have
polymorphonuclear leukocytes, monocyte/macrophages, and
NK cells) and those of the adaptive immune system (e.g., T and
B lymphocytes). Most cytokines exert pleiotropic and overlap-
ping effects through interaction with specific receptors expressed
on different target cells. Cytokine receptors also exist in soluble
forms, such as the soluble IL-2 receptor (sIL-2R) which is shed
from the membrane surface of activated immune cells and can
inhibit the biological activity of IL-2, an important T-cell growth
factor, by preventing its binding to membrane-anchored recep-
tors. Therefore, sIL-2R is viewed as a marker of immune activa-
tion. In contrast, binding of soluble IL-6 receptor (sIL-6R) to IL-6
forms a complex that enhances the biological activity of IL-6.
Cytokine actions can be inhibited also by naturally occurring
cytokine receptor antagonists such as the IL-1 receptor antago-
nist (IL-RA), which competes with the physiologic ligand for
binding to membrane IL-1 receptors. The IL-1RA is produced in
response to several inflammatory stimuli, including IL-1? and
IL-6, and is elevated in a variety of infections and inflammatory
diseases (6). The IL-1?, IL-6, TNF-?, and IFN-? are considered
pro-inflammatory, in the sense that they augment the immune
response to infection and inflammation by promoting leukocyte
recruitment to inflammatory sites and/or by activating inflamma-
tory cells. The IL-1RA, IL-4, and IL-10 are anti-inflammatory
cytokines that contribute to dampen the immune and inflamma-
The cytokine production profile of CD4? T helper (Th)
lymphocytes allowed the identification of at least two distinct
subsets: Th1 and Th2 (7). The Th1 cells produce IFN-?, a potent
activator of cell-mediated immunity, and Th2 cells produce IL-4,
IL-5, and IL-13, potent inducers of B-cell immunoglobulin (Ig)
isotype switching to IgE and activators of eosinophil recruitment
(7,8). The cytokine environment produced by dendritic cells and
other cell sources during inflammatory responses is essential in
driving the differentiation of naive CD4? T cells into Th1 or Th2
lineages. The IL-12 produced by antigen-presenting cells (e.g.,
dendritic cells and macrophages) acts via the transcription factors
From the Department of Psychiatry (SP, ES, AAS), Faculty of Medicine, Uni-
versity of Montreal, Fernand-Seguin Research Center, Louis-H Lafon-
taine Hospital; Department of Medicine (RB, EK), Faculty of Medicine,
University of Montreal, Maisonneuve-Rosemont Hospital Research Cen-
ter, Montreal, Quebec; and AstraZeneca Pharmaceuticals (AG), Missis-
sauga, Ontario, Canada.
Address reprint requests to Edouard Kouassi, Ph.D., Maisonneuve-Rose-
mont Hospital Research Center; 5415 Assomption Boulevard, Montreal,
Quebec, Canada, H1T 2M4; E-mail: email@example.com.
Received March 30, 2007; revised July 30, 2007; accepted September 29,
BIOL PSYCHIATRY 2008;63:801–808
© 2008 Society of Biological Psychiatry
Stat4 and T-bet to promote the development of Th1 cells,
whereas IL-4 produced by activated T cells and cells of the innate
immune system acts via Stat6 and GATA3 to drive the differen-
tiation of Th2 cells. The IFN-?–producing Th1 cells are respon-
sible for eradication of intracellular pathogens (e.g., intracellular
bacteria, viruses, and some protozoa), whereas IL-4-producing
Th2 cells enhance elimination of extracellular organisms (e.g.,
helminths) and are involved in allergy responses and asthma.
Autoimmune inflammatory diseases (e.g., rheumatoid arthritis,
type 1 diabetes, and multiple sclerosis) do not fit very well into
the Th1/Th2 paradigm, and recent work indicates that two other
T cell subsets might play a role in these cases, namely IL-17–
producing T cells (Th17) that induce autoimmunity and Foxp3?
regulatory T cells (Treg) that inhibit autoimmune tissue injury.
The development of Th17 cells is induced by TGF-?1 in the
presence of IL-6, and their expansion depends on IL-23. The
TGF-?1 also promotes the development of Treg cells, by induc-
ing the transcription factor Foxp3, an outcome that is inhibited in
the presence of IL-6 (8–11).
Schizophrenia might be associated with an imbalance in
Th1/Th2 cytokines, with a shift toward the Th2 system (12–15).
Empirical evidence in support of the Th1/Th2 imbalance has
been inconsistent. To date, the most frequently studied cytokines
in schizophrenia have been IL-2 and IL-6. Although there is
evidence of decreased IL-2 levels in schizophrenia (16,17), some
studies do not confirm this trend (18), whereas other studies
actually show the reverse relationship (19). As for the predicted
increase in IL-6 levels in schizophrenia, results reported so far
have been more consistent, but not all groups have replicated
this observation (20–22). As for other cytokines, the lack of
consistency of results has been noticed by many authors
(4,13,23). Another drawback in the field has been the discovery
of the impact of antipsychotic drugs on immune parameters
(24,25), which has fed skepticism about the conceptualization of
schizophrenia as an autoimmune disorder. Hinze-Selch and
Pollmächer (26) performed an extensive review of the data on in
vitro cytokine secretion in patients with schizophrenia but found
out the available literature as published by March 2001 was
plagued with considerable heterogeneity, not amenable to meta-
analytic treatment. Since then, numerous cross-sectional in vivo
and in vitro studies have been published on the topic, making
the feasibility of meta-analysis more likely. The current meta-
analysis was conducted to verify whether the cytokine imbalance
hypothesis of schizophrenia is substantiated by evidence.
Methods and Materials
A search of computerized literature databases (PubMed and
EMBASE) was conducted, with the following key words: “schizo-
phrenia,” “cytokine,” “interleukin,” “interferon,” and “tumor ne-
crosis factor.” Studies were also identified by cross-referencing of
included studies. A consensus has been reached among authors
on the studies retained or discarded, on the basis of the following
inclusion and exclusion criteria.
Inclusion criteria were: 1) patients with a schizophrenia
spectrum disorder: schizophrenia, schizoaffective disorder, and
schizophreniform disorder; 2) cross-sectional studies comprising
a control group of healthy volunteers; 3) studies assessing
circulating cytokine levels with plasma or serum samples (in vivo
studies) or in vitro cytokine secretion by peripheral blood
leukocytes stimulated or not by mitogens (in vitro studies); 4)
studies assessing one or more of the prototypic Th1 and/or Th2
cytokines, soluble cytokine receptors, endogenous cytokine
antagonists, and other inflammatory cytokines; 5) studies in
English language published before 52nd week of 2005. Exclusion
criteria were: 1) studies assessing cytokine genes; 2) studies
assessing immune markers other than cytokines (e.g., immuno-
globulins); and 3) prospective studies (mostly pharmacologic
studies) with no control group.
Data Extraction and Quantitative Data Synthesis
Two reviewers independently extracted data (n, mean & SD)
to avoid potential mistakes. Discrepancies in data entry were
double-checked by the two reviewers with the original published
data, and a consensus was reached. With Comprehensive Meta-
Analysis (27), effect size estimates of the differences in cytokine
levels between schizophrenia patients and healthy volunteers
were calculated. Effect size estimates were calculated from
means and SD or t test (for two studies, effect sizes were derived
from Dunn t tests). In the case of studies with analyses of
variance, the effect size estimate was calculated with D-STAT
(28). Within a random effect model, effect size estimates were
derived with Hedges’s g (29), which provides unbiased effect
sizes adjusted for sample size. Taking into account variability in
effect size estimates, random effect models allow population-
level inferences and are more stringent than fixed effect models
(30). The direction of the effect size was positive if schizophrenia
patients showed: 1) lower Th1 cytokine levels than healthy
volunteers, 2) higher Th2 cytokine levels, and 3) higher levels of
all other inflammatory cytokines or immune activation markers.
The level of significance for the effect size estimates was set at
p ? .05.
Obviously, the studies retained in the meta-analysis did not
assess all of the aforementioned cytokines. For each study, an
effect size estimate was calculated for each cytokine assessed. A
composite effect size estimate was then calculated by pooling
effect size estimates across studies, cytokine by cytokine. Publi-
cations pertaining to the same research group and studying the
same cytokine were checked for potential data overlap. In this
latter case, corresponding authors were contacted, and the
publication with the most complete data set was retained for the
meta-analysis (Supplement 1).
For some studies, the cytokine levels were only provided for
specific subgroups of patients (e.g., acute versus remission/
medicated versus non-medicated). To calculate the effect size for
these studies, the subgroup cytokine levels were collapsed
(weighted average) with D-STAT. The subgroup cytokine levels
were used, however, for secondary analyses (see Results sec-
For studies assessing cytokines in vitro, cytokine secretion
was unstimulated and/or stimulated with diverse mitogens,
therefore providing diverse cytokine levels. Furthermore, some
in vitro studies were done on whole blood and others on isolated
blood lymphocytes or mononuclear cells. To calculate the effect
size estimates for these studies, the cytokine levels were col-
lapsed (weighted average), again with D-STAT.
Homogeneity of Effect Size Estimates
It is more legitimate to aggregate effect size estimates when
they are homogeneous. Thus, we have calculated the Q statistic
for the effect size estimates of the studies included in the
meta-analysis. Level of significance of heterogeneity was set at
p ? .1, as recommended by Song et al. (31).
802 BIOL PSYCHIATRY 2008;63:801–808
S. Potvin et al.
A total of 544 studies were identified; 457 studies were
rejected for the following reasons: 1) type of article (e.g., review,
letter, case report; 223 studies); 2) methodology (e.g., no
control group, immunoglobulin assessment; 92 studies); 3)
study population (e.g., depression, bipolar mania, twin sub-
jects; 57 studies); 4) type of study (e.g., immunotherapy,
genetic study; 75 studies); 5) cytokines were measured in the
cerebrospinal fluid; 4 studies); and 6) foreign language (6
Eighty-seven articles responded to our search criteria. Of 87
studies identified, 25 articles could not be entered in the meta-
analysis, for the following reasons: 1) data were not available,
even after contacting the authors (6 studies); 2) data were
incomplete (6 studies); 3) schizophrenia patients suffered from
malignant tumors (1 study); 4) cytokine levels were not detect-
able (2 studies); and 5) articles—from foreign countries—were
not retrievable, even after contacting authors (10 studies).
Thus, 62 articles were available for meta-analysis (16–22,
25,32–85) for a total sample size of 2298 schizophrenia patients
and 1858 healthy volunteers. For these 62 studies, very few data
were missing or unfit for meta-analytic treatment (e.g., range
instead of mean). As per our own estimate, more than 95% of the
relevant data were retrievable. Even so, authors were contacted
to gather missing data. As described in Supplement 1, studies
included were heterogeneous in terms of cytokines assessed,
medication (antipsychotic, non-medicated, or mixed), psychiat-
ric state (acute/inpatient, non-acute/outpatient, or mixed), ex-
perimental condition (in vivo or in vitro), and location of study
(Europe, Asia, and so forth). Cytokines assessed in three studies
or fewer (e.g., IL-5, IL-8, IL-12, IL-13, IL-18, and TGF-?1) were
not retained in the meta-analysis. The most frequently assessed
cytokines were IL-2, sIL-2R, and IL-6, and the most frequently
used assay was enzyme-linked immunosorbent assay (ELISA;
83% of the studies), whereas radio-immunoassay (RIA) and
bioassays were used in only 9% and 8% of the studies, respec-
tively (Supplement 1).
Quantitative Data Synthesis: Composite Effect Size Estimates
On the basis of the 62 studies included in the composite
analysis, an effect size estimate was produced for 10 cytokines
including the prototypic Th1 and Th2 cytokines, IFN-? and IL-4,
respectively, as well as the following inflammatory cytokines,
their soluble receptors, or natural antagonists: IL-1?, IL-1RA, IL-2,
sIL-2R, TNF-?, IL-6, sIL-6R, and IL-10, with the relevant studies.
The respective numbers of in vivo / in vitro studies responding to
our inclusion criteria were as follows: IFN-? (3/10), IL-4 (2/4),
IL-1? (8/2), IL-1RA (7/0), IL-2 (10/20), sIL-2R (17/2), TNF-? (8/2),
IL-6 (19/1), sIL-6R (7/0), and IL-10 (2/6). Therefore, the available
studies were predominantly in vivo (IL-1?, IL-1RA, sIL-2R,
TNF-?, IL-6, and sIL-6R) or in vitro (IFN-?, IL-2, IL-4, and IL-10),
depending on the cytokines. In vivo and in vitro data were kept
separate, assuming that they do not represent the same biological
mechanisms (Table 1). Some articles contributed for both in vivo
and in vitro studies on the same cytokine. In this case, the data
for each experimental condition were considered separately, but
each article was counted only once in the total number of 62
For studies assessing in vivo peripheral levels of cytokines, a
moderate and highly significant effect size emerged for IL-1RA (7
studies), IL-6 (19 studies), and sIL-2R (17 studies), suggesting an
increase in these cytokines in schizophrenia (Table 1; Figures 1
and 2). Nonsignificant effect size estimates were obtained for
IL-1? (8 studies), sIL-6R (7 studies), TNF-? (8 studies), and IL-2
(10 studies) (Table 1). For studies assessing in vitro cytokine
secretion, a significant moderate effect size was obtained for IL-2
(20 studies), suggesting a decrease in IL-2 secretion in schizo-
Table 1. Composite Effect Size Estimates/Cytokine in In Vivo and In Vitro Studies
CytokineStudies (n) Subjects (n)Effect Sizep 95% CIQ-Testp
CI, confidence interval; IL, interleukin; IFN, interferon; RA, receptor antagonist; TNF, tumor necrosis factor.
aEffect size was significant at p ? .05, and heterogeneity (Q-test) was significant at p ? .1.
schizophrenia relative to control subjects. CI, confidence interval.
S. Potvin et al.
BIOL PSYCHIATRY 2008;63:801–808 803
phrenia patients. Nonsignificant effect size estimates were ob-
tained for IFN-? (10 studies), IL-4 (5 studies), and IL-10 (6
studies) (Table 1).
Sets of studies used for the composite analysis were hetero-
geneous for all cytokines, except for IL-1RA, as reflected by the
p values of the Q-test (Table 1). For this reason, secondary
analyses were performed in subsets of studies on the basis of
antipsychotic medication (medicated or not), psychiatric setting
(acute or not), and location of study (Supplement 1). For location
of study, we treated studies of patients recruited in Middle-East
and Asian countries (mostly non-Caucasian) and studies of
patients recruited in European and North American countries
(mostly Caucasian) separately.
When an analysis was performed in the subset of studies
assessing in vivo sIL-2R in European and North American p a -
tients receiving antipsychotic treatment (n ? 7), heterogeneity
was eliminated (Table 2). With this subset of studies, a
moderate and very significant effect size was obtained. This
latter result highlights an increase in in vivo sIL-2R in schizo-
phrenia patients, relative to healthy volunteers, insofar as
patients are European or North American and treated with an
As for in vivo IL-6, heterogeneity was substantially reduced
when secondary analyses were conducted with studies includ-
ing: 1) European or North American patients (15 studies); and 2)
non-medicated patients (13 studies). These subanalyses pro-
duced very significant moderate effect size estimates, suggesting
an increase in IL-6 levels in European or North American and/or
non-medicated patients, relative to healthy volunteers (Table 2).
In the case of in vitro IL-2 secretion, heterogeneity was not
corrected by any of the secondary analyses performed (Table 2).
Furthermore, the level of significance of the effect size remained
moderate with subsets of studies specific to European or North
American patients (16 studies), whereas it was higher in patients
receiving antipsychotic treatment (12 studies).
The current meta-analysis was conducted to verify whether
the cytokine imbalance and Th2 predominance hypotheses of
schizophrenia are substantiated by evidence. The results suggest
an increase in in vivo peripheral levels of IL-1RA, sIL-2R, and IL-6
and a decrease in in vitro IL-2 secretion in schizophrenia patients.
No significant effect sizes were obtained for IFN-?, IL-4, IL-1?,
TNF-?, sIL-6R, and IL-10. The absence of significant changes of
IFN-? and IL-4 does not support the Th2 shift hypothesis of
schizophrenia, at least in the peripheral blood. Until recently,
IL-2 and IL-6 were classified as Th1 and Th2 type cytokines,
respectively (7,86,87). This probably contributed to the formula-
tion of the hypothesis of a shift from Th1 to Th2 cytokines on the
basis of studies showing decreased in vitro IL-2 secretion and
increased in vivo circulating sIL-2R and IL-6 levels in schizophre-
nia patents (14). However, classification of cytokines is being
re-examined, because new CD4? T cell subsets, such as Th17
and Treg (8,11) are emerging. Unfortunately and because of the
novelty of the findings, the data on Th17 and Treg defining
cytokines in schizophrenia are still lacking or insufficient at this
time point. As a consequence, speculations on a specific devia-
tion of CD4? T cell differentiation in schizophrenia, if any, await
The significant increases in in vivo peripheral levels of IL-1RA,
sIL-2R, and IL-6 in schizophrenia patients as shown in this
meta-analysis provide evidence of immune activation and in-
flammatory syndrome in schizophrenia, as pointed out by sev-
eral previous studies (18,45,52,63,74,76). Most studies assessing
combinations of IL-1RA, sIL-2R, and IL-6 in the same schizophre-
nia patients have found concordant upregulation of these cyto-
kines (18,21,45,55,82), indicating some correlation in their dys-
regulation. Interestingly, these cytokines, notably IL-1RA and
IL-6, are predominantly produced by cells of the innate immu-
nity, suggesting primary alterations of this arm of the immune
system in schizophrenia patients. The elevated circulating levels
of IL-1RA in schizophrenia is particularly intriguing in view of its
known protective effects against rheumatoid arthritis (6) and the
Figure 2. Forest plot of the effect size estimates of in vivo interleukin (IL)-1
receptor antagonist (RA) in schizophrenia relative to control subjects. The
arrow corresponding to the study of Maes et al. (20) shows that the confi-
dence interval expands beyond the limit for the effect size range (?2.00 to
2.00). CI, confidence interval.
Table 2. Subanalyses of In Vivo sIL-2R, In Vivo IL-6, and In Vitro IL-2 in Schizophrenia
Cytokine Studies (n) Subjects (n) Subgroup Effect Sizep 95% CIQ-Testp
E or NA
E or NA
E or NA
E or NA, European or North American; Antipsychotic, receiving antipsychotic medication; Non-medicated, non-
medicated at the moment of blood collection (?1 week), other abbreviations as in Table 1.
aEffect size was significant at p ? .05, and heterogeneity (Q-test) was significant at p ? .1.
804 BIOL PSYCHIATRY 2008;63:801–808
S. Potvin et al.
reportedly lower than expected prevalence of this autoimmune
disease in schizophrenia patients as shown in a number of
settings (88–90). This raises the possibility that elevated IL-1RA is
involved in the negative relationship between schizophrenia and
The findings of the current meta-analysis must be interpreted
cautiously. Alterations in the cytokine network could be related
to the physiopathology of schizophrenia or even its etiology.
Alternatively, these alterations might also be the byproduct of the
stress associated with the disorder. Schizophrenia patients are
sensitive to stress, which is known to trigger psychotic relapses
(91). It is likely that the elevated peripheral levels of inflammatory
cytokines are induced by stress as observed in other major psychi-
atric disorders such as depression, by mechanisms involving defec-
tive glucocorticoid-mediated feedback inhibition and/or exagger-
ated sympathetic nervous system-mediated activation of immune
responses (92,93). Similarly, cytokine alterations in schizophrenia
might be secondary to weight gain. Patients with schizophrenia
are at greatest risk for metabolic disorders and obesity (94).
Obesity in schizophrenia is accentuated by poor dietary condi-
tions, sedentary lifestyles, and antipsychotic drugs, especially
second-generation antipsychotics (95). Of interest, some cyto-
kines, namely TNF-? and IL-1RA, seem to contribute—either as a
cause or consequence—to weight gain in psychiatric as well as
non-psychiatric conditions (96,97), and adipocytes are a major
source of inflammatory cytokines production, including IL-6 (98).
Alterations in the cytokine network in schizophrenia could be
secondary also to antipsychotic treatment, which has been
regularly shown to significantly impact on the immune system
(24,25,99). Subanalyses conducted in the current meta-analysis
did substantiate the importance of antipsychotic treatment. For
instance, the decrease in in vitro IL-2 secretion was only true for
patients treated with antipsychotic medication, not for non-
medicated patients at the moment of sample collection, although
at least some studies found in vitro IL-2 abnormalities in drug-
naïve patients as well (17,36,40,43,47,65,78,83). Similarly, the
increase in in vivo sIL-2R levels was only true for patients treated
with antipsychotic drugs. This strongly suggests that IL-2 and
sIL-2R alterations in schizophrenia are directly related to medi-
cation and not to the disease per se. However, for IL-6, the
reverse pattern was observed. Indeed, the increase in IL-6 levels
in schizophrenia was only applicable to untreated patients, not to
patients receiving antipsychotic medication at the moment of
blood collection. Thus, it seems more than likely that IL-6
alterations in schizophrenia are related to the disease in itself, not
to the effects of medication. However, it must be considered that
only 3 of 13 “non-medicated patient” studies actually involved
drug-naïve patients (21,36,42). Most of these studies included
patients unmedicated at the time of blood collection (time span:
1 week to 6 months). Like the results for IL-6, the results obtained
for IL-1RA did not seem to be related to antipsychotic medica-
tion. To verify this, a subanalysis was conducted with IL-1RA
studies of treated and untreated patients. In both cases, suba-
nalyses with homogeneous sets of studies produced very signif-
icant moderate effect size estimates.
Apart from medication, psychiatric state of patients (acutely ill
or not) did not appear to impact on the strength and significance
of effect size estimates, for any cytokines. This suggests that
cytokine alterations in schizophrenia are not state-dependent.
Conversely, there was a relationship between location of study
and abnormal cytokine concentrations in schizophrenia patients.
For instance, control for location of study substantially reduced
heterogeneity of studies assessing in vivo IL-6 or in vivo sIL-2R.
Despite the extent of controls performed in the current
meta-analysis, heterogeneity remained a concern. Yet uncon-
trolled factors include smoking, dietary habits, type of antipsy-
chotic drug (first- or second-generation), duration of treatment,
concomitant use of other medications, substance abuse, gender,
weight gain, and age, which are all factors potentially influencing
the immune system (26,100). It must be mentioned that a
regression analysis was performed for each cytokine, to control
for age. For each study, the mean age of the patient group was
used as the explanatory variable. No relationship was found
between age and any cytokine (data not reported here). As for
body mass index, it was only reported by a minority of studies,
insufficient for meta-analytic treatment.
The meta-analysis comprised limitations. First, some data
were not available, whereas some articles—mostly from Eastern
Europe journals—could not be retrieved, even after contacting
authors and journal editors. One might think that authors of
negative findings were less likely to respond to queries than
authors of studies with statistically significant findings. However,
the fail-safe number of additional negative studies (i.e., studies
reporting no cytokine differences between patients and control
subjects) required to nullify the significance of our composite
analyses was high: 34 studies for IL-1RA, 244 for IL-2, 195 for
sIL-2R, and 229 studies for IL-6. Another limitation was related to
heterogeneity. Heterogeneity was, after controlling for medica-
tion and location of study, substantially reduced or eliminated for
studies assessing sIL-2R and IL-6. However, these factors did not
explain the heterogeneity in studies assessing IFN-?, IL-4, IL-1?,
IL-2, TNF-?, sIL-6R, and IL-10. In addition, study heterogeneity
will need to be further tracked down by searching for potential
confounding factors, such as weight gain. Another potential
limitation is that the grouping of some studies does not take into
account methodological differences in sample handling, storage
conditions, and methods of cytokine assessment, which can
influence cytokine levels and the overall results. However,
Hinze-Selch and Pollmächer have previously shown that variabil-
ity in methodology does not explain inconsistency of results, at
least for in vitro cytokine secretion studies in individuals with
To our knowledge, this is the first meta-analysis to be
conducted on cytokines in schizophrenia, at the level of the
proteins, although a number of studies have been performed on
cytokine genes (101,102). The meta-analysis was conducted with
a large set of studies (n ? 62), involving a very large sample size
(2298 schizophrenia patients and 1858 healthy volunteers). Fur-
thermore, effect size estimates were produced for seven cyto-
kines (IL-1?, IL-2, IFN-?, TNF-?, IL-4, IL-6, and IL-10), one
cytokine receptor antagonist (IL-1RA), and two soluble cytokine
receptors (sIL-2R and s-IL6R). In sets of homogeneous studies, a
very significant increase in IL-1RA, sIL-2R, and IL-6 levels was
observed in schizophrenia, supporting the existence of an ongo-
ing inflammatory process. In the case of IL-6 and most probably
IL-1RA, the alterations were not related to antipsychotic medica-
tion. Thus, the current findings lead us to propose that IL-6 and
IL-1RA alterations in schizophrenia are linked to the physiopa-
thology of schizophrenia or to phenotypic traits of the disorder
yet to be characterized. Although the classical actions of these
molecules have been described in the immune system, it is
possible that they might have as yet unknown actions in the
nervous system unrelated to immunity and inflammation. In the
same line, it must be pointed out that, even in studies showing
cytokine alterations, most of the subjects with schizophrenia
showed no clinical evidence of immunologic dysfunction. This
S. Potvin et al.
BIOL PSYCHIATRY 2008;63:801–808 805
suggests that, if immunologic processes do play a role in schizo-
phrenia, this is likely to be the case in only a relatively small
subgroup of the individuals who clinically meet the criteria for the
diagnosis. In another valuable finding, the current meta-analysis
helps to identify two factors explaining part of study heteroge-
neity, namely location of study and antipsychotic medication,
and it emphasizes the need for the design of a more focused and
refined research study to elucidate the role of cytokine imbalance
This work was supported in part by an operating Grant,
MOP-67183 from the Canadian Institutes of Health Research
(CIHR), to Edouard Kouassi. Emmanuel Stip is holder of the Eli
Lilly Chair of Schizophrenia from the University of Montreal.
Stéphane Potvin is holder of a scholarship from the CIHR. Drs.
Potvin, Stip, Gendron, and Kouassi, Mr. Sepehry, and Mrs. Bah
reported no biomedical financial interest or potential conflict of
We would like to thank Dr. Y.K. Kim and Dr. J. Kowalski for
generously providing their data. We would also like to thank Dr.
Ron Sullivan for his comments and Karine Damiens for her
contribution to data compilation.
Supplementary material cited in this article is available
2. Eaton WW, Byrne M, Ewald H, Mors O, Chen CY, Agerbo E, et al. (2006):
Association of schizophrenia and autoimmune diseases: Linkage of
Danish national registers. Am J Psychiatry 163:521–528.
3. Steinke JW, Borish L (2006): Cytokines and chemokines. J Allergy Clin
4. Kronfol Z, Remick DG (2000): Cytokines and the brain: Implications for
5. Raison CL, Miller AH (2004): Brain-immune system interaction: Rele-
vance to the pathophysiology and treatment of neuropsychiatric dis-
American Psychiatric Press, 147–162.
Two types of murine helper T cell clone. I. Definition according to
8. Dong C (2006): Diversification of T-helper-cell lineages: Finding the
family root of IL-17-producing cells. Nat Rev Immunol 6:329–333.
9. Harrington LE, Mangan PR, Weaver CT (2006): Expanding the effector
CD4 T-cell repertoire: The Th17 lineage. Curr Opin Immunol 18:349–
10. McKenzie BS, Kastelein RA, Cua DJ (2006): Understanding the IL-23-
IL-17 immune pathway. Trends Immunol 27:17–23.
11. Tato CM, Laurence A, O’Shea JJ (2006): Helper T cell differentiation
enters a new era: Le roi est mort; vive le roi! J Exp Med 203:809–812.
12. Muller N, Riedel M, Gruber R, Ackenheil M, Schwarz MJ (2000): The
13. Muller N, Riedel M, Ackenheil M, Schwarz MJ (1999): The role of im-
mune function in schizophrenia: An overview. Eur Arch Psychiatry Clin
Neurosci 249(suppl 4):62–68.
14. Schwarz MJ, Muller N, Riedel M, Ackenheil M (2001): The Th2-hypoth-
nia caused by immune mechanisms. Med Hypotheses 56:483–486.
15. Schwarz MJ, Chiang S, Muller N, Ackenheil M (2001): T-helper-1 and
T-helper-2 responses in psychiatric disorders. Brain Behav Immun 15:
16. Arolt V, Rothermundt M, Wandinger KP, Kirchner H (2000): Decreased
in vitro production of interferon-gamma and interleukin-2 in whole
17. Bessler H, Levental Z, Karp L, Modai I, Djaldetti M, Weizman A (1995):
Cytokine production in drug-free and neuroleptic-treated schizo-
phrenic patients. Biol Psychiatry 38:297–302.
18. Kaminska T, Wysocka A, Marmurowska-Michalowska H, Dubas-Slemp
H, Kandefer-Szerszen M (2001): Investigation of serum cytokine levels
and cytokine production in whole blood cultures of paranoid schizo-
19. Ebrinc S, Top C, Oncul O, Basoglu C, Cavuslu S, Cetin M (2002): Serum
al. (2000): Effects of atypical antipsychotics on the inflammatory re-
sponse system in schizophrenic patients resistant to treatment with
typical neuroleptics. Eur Neuropsychopharmacol 10:119–124.
tic administration. Schizophr Res 37:97–106.
Serum interleukin-6 in schizophrenic patients. Life Sci 49:661–664.
23. Gladkevich A, Kauffman HF, Korf J (2004): Lymphocytes as a neural
probe: Potential for studying psychiatric disorders. Prog Neuropsycho-
24. Ghezzi P, Garattini S, Mennini T, Bertini R, Delgado Hernandez R, Be-
nigni F, et al. (1996): Mechanism of inhibition of tumor necrosis factor
production by chlorpromazine and its derivatives in mice. Eur J Phar-
in schizophrenia: Comparison to normal controls and effects of cloza-
26. Hinze-Selch D, Pollmacher T (2001): In vitro cytokine secretion in indi-
viduals with schizophrenia: results, confounding factors, and implica-
tions for further research. Brain Behav Immun 15:282–318.
27. Borenstein M, Rothstein H (1999): Comprehensive Meta-Analysis: A
28. Johnson BT (1989): D-STAT. Software for Meta-Analytic Review of Re-
search Literatures. Hillsdale, New Jersey: Laurence Erlbaum.
York: Russell Sage Foundation Publications.
30. DerSimonian R, Laird N (1986): Meta-analysis in clinical trials. Control
31. Song F, Sheldon TA, Sutton AJ, Abrams KR, Jones DR (2001): Methods
reactivity with relative Th2 dominance in patients with acute exacer-
bation of schizophrenia. Croat Med J 46:268–274.
33. Baker I, Masserano J, Wyatt RJ (1996): Serum cytokine concentrations
in patients with schizophrenia. Schizophr Res 20:199–203.
34. Barak V, Barak Y, Levine J, Nisman B, Roisman I (1995): Changes in
serum of schizophrenic patients. J Basic Clin Physiol Pharmacol 6:
35. Cazzullo CL, Scarone S, Grassi B, Vismara C, Trabattoni D, Clerici M
without paranoid behaviour. Prog Neuropsychopharmacol Biol Psychi-
36. Cazzullo CL, Sacchetti E, Galluzzo A, Panariello A, Colombo F, Zagliani
A, et al. (2001): Cytokine profiles in drug-naive schizophrenic patients.
et al. (2002): Cytokine profiles in schizophrenic patients treated with
risperidone: A 3-month follow-up study. Prog Neuropsychopharmacol
38. el-Mallakh RS, Suddath RL, Wyatt RJ (1993): Interleukin-1 alpha and
interleukin-2 in cerebrospinal fluid of schizophrenic subjects. Prog
M (1997): Interleukin-6-(IL-6) plasma levels in depression and schizo-
806 BIOL PSYCHIATRY 2008;63:801–808
S. Potvin et al.
40. Ganguli R, Rabin BS, Belle SH (1989): Decreased interleukin-2 produc-
tion in schizophrenic patients. Biol Psychiatry 26:427–430.
41. Ganguli R, Rabin BS (1989): Increased serum interleukin 2 receptor
42. Ganguli R, Yang Z, Shurin G, Chengappa KN, Brar JS, Gubbi AV, et al.
(1994): Serum interleukin-6 concentration in schizophrenia: Elevation
associated with duration of illness. Psychiatry Res 51:1–10.
43. Ganguli R, Brar JS, Chengappa KR, DeLeo M, Yang ZW, Shurin G, et al.
(1995): Mitogen-stimulated interleukin-2 production in never-medi-
cated, first-episode schizophrenic patients. The influence of age at
onset and negative symptoms. Arch Gen Psychiatry 52:668–672.
44. Gattaz WF, Dalgalarrondo P, Schroder HC (1992): Abnormalities in
serum concentrations of interleukin-2, interferon-alpha and interfer-
on-gamma in schizophrenia not detected. Schizophr Res 6:237–241.
45. Haack M, Hinze-Selch D, Fenzel T, Kraus T, Kuhn M, Schuld A, et al.
(1999): Plasma levels of cytokines and soluble cytokine receptors in
psychiatric patients upon hospital admission: Effects of confounding
factors and diagnosis. J Psychiatr Res 33:407–418.
46. Katila H, Hurme M, Wahlbeck K, Appelberg B, Rimon R (1994): Plasma
and cerebrospinal fluid interleukin-1 beta and interleukin-6 in hospi-
talized schizophrenic patients. Neuropsychobiology 30:20–23.
47. Kim YK, Lee MS, Suh KY (1998): Decreased interleukin-2 production in
Korean schizophrenic patients. Biol Psychiatry 43:701–704.
48. Kim YK, Kim L, Lee MS (2000): Relationships between interleukins,
neurotransmitters and psychopathology in drug-free male schizo-
phrenics. Schizophr Res 44:165–175.
and Th3 cytokine alteration in schizophrenia. Prog Neuropsychophar-
50. Kowalski J, Blada P, Kucia K, Madej A, Herman ZS (2001): Neuroleptics
normalize increased release of interleukin- 1 beta and tumor necrosis
factor-alpha from monocytes in schizophrenia. Schizophr Res 50:169–
51. Kowalski J, Blada P, Kucia K, Lawniczek T, Madej A, Belowski D, et al.
(2000): In-vitro immunomodulatory effects of haloperidol and pera-
52. Lin A, Kenis G, Bignotti S, Tura GJ, De Jong R, Bosmans E, et al. (1998):
nia: Increased serum interleukin-6. Schizophr Res 32:9–15.
53. Maes M, Meltzer HY, Buckley P, Bosmans E (1995): Plasma-soluble
interleukin-2 and transferrin receptor in schizophrenia and major de-
54. Maes M, Bosmans E, Calabrese J, Smith R, Meltzer HY (1995): Interleu-
kin-2 and interleukin-6 in schizophrenia and mania: Effects of neuro-
leptics and mood stabilizers. J Psychiatr Res 29:141–152.
et al. (2002): Increased serum interleukin-8 and interleukin-10 in
schizophrenic patients resistant to treatment with neuroleptics and
the stimulatory effects of clozapine on serum leukemia inhibitory fac-
tor receptor. Schizophr Res 54:281–291.
57. Maes M, Bosmans E, Ranjan R, Vandoolaeghe E, Meltzer HY, De Ley M,
et al. (1996): Lower plasma CC16, a natural anti-inflammatory protein,
nia: Effects of antipsychotic drugs. Schizophr Res 21:39–50.
58. Mahendran R, Chan YH (2004): Interleukin-2 levels in chronic schizo-
Effects of clozapine treatment. Psychiatry Res 71:11–17.
schizophrenic patients. Eur Psychiatry 12:294–299.
tic treatment increases soluble IL-2 receptors and decreases soluble
IL-6 receptors in schizophrenia. Eur Arch Psychiatry Clin Neurosci 247:
62. Naudin J, Mege JL, Azorin JM, Dassa D (1996): Elevated circulating
levels of IL-6 in schizophrenia. Schizophr Res 20:269–273.
63. Naudin J, Capo C, Giusano B, Mege JL, Azorin JM (1997): A differential
role for interleukin-6 and tumor necrosis factor-alpha in schizophre-
nia? Schizophr Res 26:27–233.
64. Nimgaonkar VL, Yang ZW, Zhang XR, Brar JS, Chakravarti A, Ganguli R
(1995): Association study of schizophrenia and the IL-2 receptor beta
(1996): Increased production of interleukin-2 (IL-2) but not soluble
interleukin-2 receptors (sIL-2R) in unmedicated patients with schizo-
phrenia and schizophreniform disorder. Psychiatry Res 65:171–178.
66. Rapaport MH, McAllister CG, Pickar D, Nelson DL, Paul SM (1989):
neuroleptic-naive schizophrenic subjects and in medicated schizo-
phrenic subjects with and without tardive dyskinesia. Acta Psychiatr
(1994): Increased serum soluble interleukin-2 receptors in Caucasian
and Korean schizophrenic patients. Biol Psychiatry 35:767–771.
69. Rothermundt M, Arolt V, Leadbeater J, Peters M, Rudolf S, Kirchner H
(2000): Cytokine production in unmedicated and treated schizo-
phrenic patients. Neuroreport 11:3385–3388.
70. Rothermundt M, Arolt V, Weitzsch C, Eckhoff D, Kirchner H (1998):
Immunological dysfunction in schizophrenia: A systematic approach.
71. Rothermundt M, Arolt V, Weitzsch C, Eckhoff D, Kirchner H (1996):
Production of cytokines in acute schizophrenic psychosis. Biol Psychi-
nity in schizophrenia. J Autoimmun 9:661–666.
tibodies against 60-kDa heat shock protein in schizophrenia. Eur Arch
administration on serum levels of soluble IL-2 receptor-alpha and IL-1
AD, Papamichail MP, et al. (2001): Cytokine serum levels, autologous
icated and chronically medicated schizophrenic patients. Schizophr
76. van Kammen DP, McAllister-Sistilli CG, Kelley ME, Gurklis JA, Yao JK
77. Villemain F, Chatenoud L, Guillibert E, Pelicier Y, Bach JF (1987): De-
78. Villemain F, Chatenoud L, Galinowski A, Homo-Delarche F, Ginestet D,
Loo H, et al. (1989): Aberrant T cell-mediated immunity in untreated
schizophrenic patients: Deficient interleukin-2 production. Am J Psy-
79. Wilke I, Arolt V, Rothermundt M, Weitzsch C, Hornberg M, Kirchner H
(1996): Investigations of cytokine production in whole blood cultures
of paranoid and residual schizophrenic patients. Eur Arch Psychiatry
81. Yang ZW, Chengappa KN, Shurin G, Brar JS, Rabin BS, Gubbi AV, et al.
(1994): An association between anti-hippocampal antibody concen-
nia. Psychol Med 24:449–455.
tic-free schizophrenia: Association with psychopathology. Schizophr
83. Zhang XY, Zhou DF, Cao LY, Zhang PY, Wu GY (2002): Decreased
medication-free patients with schizophrenia. J Psychiatr Res 36:331–
S. Potvin et al.
BIOL PSYCHIATRY 2008;63:801–808 807
84. ZhangXY,ZhouDF,CaoLY,ZhangPY,WuGY,ShenYC(2004):Changes Download full-text
with risperidone and haloperidol: Relationship to outcome in schizo-
85. Zhang XY, Zhou DF, Cao LY, Wu GY, Shen YC (2005): Cortisol and
Association with psychopathology and response to antipsychotics.
87. Lucey DR, Clerici M, Shearer GM (1996): Type 1 and type 2 cytokine
dysregulation in human infectious, neoplastic, and inflammatory dis-
88. Eaton WW, Hayward C, Ram R (1992): Schizophrenia and rheumatoid
arthritis: A review. Schizophr Res 6:181–192.
89. Mors O, Mortensen PB, Ewald H (1999): A population-based register
study of the association between schizophrenia and rheumatoid ar-
thritis. Schizophr Res 40:67–74.
90. Oken RJ, Schulzer M (1999): At issue: Schizophrenia and rheumatoid
The stress cascade and schizophrenia: Etiology and onset. Schizophr
patients with major depression and increased early life stress. Am J
93. Bierhaus A, Wolf J, Andrassy M, Rohleder N, Humpert PM, Petrov D, et
al. (2003): A mechanism converting psychosocial stress into mononu-
94. Henderson DC (2005): Schizophrenia and comorbid metabolic disor-
95. Wirshing DA (2004): Schizophrenia and obesity: Impact of antipsy-
C, et al. (2003): Adipose tissue is a major source of interleukin-1 recep-
tor antagonist: Upregulation in obesity and inflammation. Diabetes
97. Kraus T, Haack M, Schuld A, Hinze-Selch D, Koethe D, Pollmacher T
(2002): Body weight, the tumor necrosis factor system, and leptin
production during treatment with mirtazapine or venlafaxine. Phar-
98. Trayhurn P, Wood IS (2004): Adipokines: Inflammation and the pleio-
tropic role of white adipose tissue. Br J Nutr 92:347–355.
99. Szuster-Ciesielska A, Slotwinska M, Stachura A, Marmurowska-Micha-
lowska H, Kandefer-Szerszen M (2004): Neuroleptics modulate cyto-
kine and reactive oxygen species production in blood leukocytes of
Smoking affects plasma-soluble interleukin-2 receptor levels in pa-
tients with schizophrenia. Arch Gen Psychiatry 54:89–90.
101. Duan S, Xu Y, Chen W, Liu Z, Guo T, Gao J, et al. (2004): No association
between the promoter variants of tumor necrosis factor alpha (TNF-
alpha) and schizophrenia in Chinese Han population. Neurosci Lett
of IL10, IL1beta, and IL1RN and schizophrenia with tag SNPs from a
comprehensive database: Suggestive association with rs16944 at
IL1beta. Schizophr Res 88:235–244.
808 BIOL PSYCHIATRY 2008;63:801–808
S. Potvin et al.