Cytokines in bipolar disorder: A systematic review and meta-analysis

Journal of Affective Disorders 144 (2013) 16–27

Contents lists available at SciVerse ScienceDirect

Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad

Review

Cytokines in bipolar disorder: A systematic review and meta-analysis Klaus Munkholm n, Maj Vinberg, Lars Vedel Kessing Psychiatric Center Copenhagen, Rigshospitalet, University of Copenhagen, Denmark

a r t i c l e i n f o

a b s t r a c t

Article history: Received 29 March 2012 Received in revised form 8 June 2012 Accepted 9 June 2012 Available online 30 June 2012

Background: Current research and hypothesis regarding the pathophysiology of bipolar disorder suggests the involvement of immune system dysfunction that is possibly related to disease activity. Our objective was to systematically review evidence of cytokine alterations in bipolar disorder according to affective state. Methods: We conducted a systemtic review of studies measuring endogenous cytokine concentrations in patients with bipolar disorder and a meta-analysis, reporting results according to the PRISMA statement. Results: Thirteen studies were included, comprising 556 bipolar disorder patients and 767 healthy controls, evaluating 15 different cytokines-, cytokine receptors- or cytokine antagonists. The levels of tumor necrosis factor-a (TNF-a), the soluble tumor necrosis factor receptor type 1 (sTNF-R1) and the soluble inlerleukin-2 receptor (sIL-2R) were elevated in manic patients compared with healthy control subjects (p o 0.01 for each). Levels of sTNF-R1 and TNF-a were elevated in manic patients compared to euthymic patients (p ¼ 0.01 and p ¼0.04, respectively). sTNF-R1 levels were elevated in euthymic patients compared with healthy control subjects (po 0.01). There were no significant findings for other comparisons, including intra-individual alterations of cytokine levels. Limitations: Stratification according to mood state resulted in small study numbers for some cytokines. Findings were limited by heterogeneity, small sample sizes and a lack of control for confounding factors in individual studies. Conclusions: This meta-analysis found some support for immune dysregulation in bipolar disorder. Future research is warranted to elucidate the role of endogenous cytokine alterations in bipolar disorder. Clinical studies examining longitudinal changes within individuals are recommended. & 2012 Elsevier B.V. All rights reserved.

Keywords: Cytokines Bipolar disorder Meta-analysis Review Inflammation Biomarker

Contents 1. 2.

3.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Methods and materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.1. Eligibility criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.2. Information sources and search. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.3. Study selection and data extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.1. Study selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.2. Study characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.3. Risk of bias and quality in individual studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3.4. Synthesis of results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.5. Manic patients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3.6. Depressed patients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.7. Euthymic patients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.8. Differences in cytokine levels between affective states . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.9. Intra-individual alterations of cytokine levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.10. Correlation with clinical features . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.11. Heterogeneity and sensitivity analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

n Corresponding author. Tel.: þ45 38 64 70 75/þ 45 38 64 70 80; fax: þ 45 35 45 62 38. E-mail address: [email protected] (K. Munkholm).

0165-0327/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jad.2012.06.010

K. Munkholm et al. / Journal of Affective Disorders 144 (2013) 16–27

4.

Discussion . . . . . . . . . . Role of funding source . Conflict of interest . . . . Acknowledgments . . . . References . . . . . . . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

1. Introduction Multiple lines of evidence indicate that bipolar disorder is a systemic disease, with widespread biochemical alterations occurring in and beyond the central nervous system (Berk et al., 2010; Kapczinski et al., 2008). Current hypotheses regarding the neurobiological background for bipolar disorder point towards defects in both cellular energy regulation, the immune system and expression of neurotrophic factors along with epigenetic alterations as core elements in the pathophysiology of the disorder (Gardner and Boles, 2010; Grande et al., 2011). These components, along with epigenetic alterations, have accordingly been proposed to be central to the neuroprogressive changes observed in bipolar disorder (Berk, 2009; Berk et al., 2010). Several areas of research have pointed to immune system dysregulation and inflammation in bipolar disorder (Goldstein et al., 2009). More specifically, immune system aberrations have been demonstrated in both in-vitro studies (Kim et al., 2007; Knijff et al., 2007) and in clinical studies showing both alterations of peripheral markers of inflammation (Brietzke et al., 2009b; Cunha et al., 2008; Dickerson et al., 2007) and alterations of inflammation related gene signatures (Drexhage et al., 2010b; Padmos et al., 2008). In addition, preclinical studies have indicated anti-inflammatory properties as a possible role of action for mood stabilizers (Bosetti et al., 2002; Lee et al., 2008; Maes et al., 1999; Rapaport and Manji, 2001). Cytokines are key signalling molecules in inflammation, exerting a regulatory effect in both the innate and the adaptive immunological response. They are produced by immune cells as well as non-immune cells and exert their effects beyond strictly the immune system. Cytokines bind to either specific cellular receptors or soluble receptors capable of modulating the immunological effect of cytokines. The immune response is further moderated by cytokine receptor antagonists, also binding to cytokine receptors (Drexhage et al., 2010a). Importantly, the role of cytokines in metabolism extend beyond the inflammatory system, impacting also on neurotransmitter metabolism, neurogenesis and the neuroendocrine system (Haroon et al., 2012). Alterations of the inflammatory system appear to be related to disease state. Peripheral markers related to inflammation, oxidative stress and neurotrophins have been proposed as mediators of systemic toxicity, specifially related to mood episodes and illness activity (Grande et al., 2011; Kapczinski et al., 2010a, 2010b). In unipolar depression, inflammatory markers elevated in depressed patients have been demonstrated to normalize after successful treatment (Miller et al., 2009) and in schizophrenia some cytokines have been indicated to be state markers for acute exacerbations (Miller et al., 2011). While mood-state related inflammatory system alterations have been demonstrated in individual clinical studies in bipolar disorder, mainly indicating elevated levels of pro-inflammatory markers during manic and depressive episodes, results have been inconsistent and the association between altered cytokine levels and mood state remains unclear for a number of cytokines (Goldstein et al., 2009). Taken together, current evidence suggests that core pathological processes underlying bipolar disorder and the possible detrimental effects resulting from mood episodes are closely related to illness activity and alterations between affective states. Meta-analysis has the potential to bring further clarity to an area of research where single studies are generally of low power

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

17

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. . . . .

. 23 . 25 . 25 . 25 . 25

and to improve the strength of evidence while also examining sources of heterogeneity among studies and whether this influences the observed effects. The purpose of the present study was therefore to perform a meta-analysis of peripheral blood cytokine-, cytokine receptor and cytokine antagonist levels in bipolar disorder according to clinical state. First, we examined alterations of endogenous immune activity in bipolar disorder patients according to clinical state (i.e., manic, depressive and euthymic state) in comparison to healthy controls. Comparisons between bipolar patients in different clinical states and evaluation of intra-individual alterations of immune activity were also performed. Second, we performed a qualitative assessment of the effect of symptom severity, medication status and clinical characteristics (i.e., smoking status and body mass index) on cytokine-, cytokine receptor and cytokine antagonist levels. We specifically refrained from including a global analysis of cytokine levels in bipolar patients compared with healthy controls, as this does not inform on cytokine levels according to mood state and illness activity. This is the first meta-analysis of cytokines in bipolar disorder.

2. Methods and materials The meta-analysis was conducted and reported according to the PRISMA statement (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) (Moher et al., 2009). A protocol for the review was prepared and is available by contact to the first author. 2.1. Eligibility criteria Original studies reporting on cytokine concentrations, cytokine receptor concentrations or cytokine antagonist concentrations in bipolar disorder were eligible for review. Further criteria were (1) adult subjects 418 years of age, meeting DSM-III-R, DSM-IV or ICD-10 criteria for bipolar disorder; studies confirming diagnosis by several individual trained psychiatrists were also evaluated; (2) cross-sectional studies comparing levels in bipolar disorder patients with either control subjects or bipolar disorder patients in a different affective state (case-control studies) or longitudinal studies comparing cytokine-, cytokine receptor or cytokine antagonist levels in bipolar disorder patients in a specified state and subsequent remission; where studies evaluated a therapeutic intervention the baseline clinical state and corresponding cytokine parameter levels were used for comparison with healthy controls if a control group was studied; (3) clinical status of bipolar disorder patients in a well defined affective state (euthymia, hypomania, mania, depression or remission); (4) studies assessing blood levels of cytokine-, cytokine receptor or cytokine antagonist levels; (5) studies published in English. If the study population included patients in various affective states, authors were contacted in order to obtain stratified data if these were not available in the manuscript. Exclusion criteria were (1) studies of in vitro cytokine production in stimulated peripheral blood mononuclear cells as they assess the response to immune challenge and do not likely inform about endogenous immune activity; (2) studies not reporting mean and standard deviation of cytokine parameter levels. If these data were not available in the

18

K. Munkholm et al. / Journal of Affective Disorders 144 (2013) 16–27

manuscript the authors were contacted to obtain these data. (3) Where more than one publication reported on results on identical cytokine parameters from the same study population, the publication with the smaller study population was excluded; (4) studies not comparing two groups of patients or control subjects. A cytokine parameter was omitted from analysis if the reported mean level was below the detection limit of the assay. 2.2. Information sources and search Studies were identified from searching MEDLINE and PubMed (January 1950–January 2012), Embase (1974–January 2012) and PsycINFO (1806–January 2012). The search strategy was inflammation or cytokine or interleukin or inflammatory markers or tumor necrosis factor and bipolar disorder or mixed state or rapid cycling or mania. In addition, references lists of retrieved articles were searched by hand. 2.3. Study selection and data extraction From the primary literature search each study title and abstract was screened for eligibility by one researcher (K.M.). Full text articles of all potentially relevant studies were subsequently

retrieved and further assessed for eligibility. A flow diagram giving an overview of the study selection process was created (Fig. 1). Data were extracted from the relevant articles, without blinding to authorship, independently by one researcher (K.M.) using a form designed specifically for the study and subsequently confirmed by two other researchers independently (L.V.K. and M.V.). In case of uncertainty or missing data in the primary publication study authors were contacted. Disagreements regarding inclusion or study data extraction were resolved by discussion between the three researchers (K.M., L.V.K. and M.V.). Specifically, data were extracted for cytokine-, cytokine receptor and cytokine antagonist levels (mean and standard deviation) for each group of study participants. Additionally, clinical data for study participants were collected (age, sex, smoking status, BMI, pharmacological treatment, bipolar type) along with assay type and blood fraction analyzed. One study included a group of patients consisting of medication free patients and patients on Lithium monotherapy and provided stratified demographic data and results, and for this study data were combined for the two groups (Guloksuz et al., 2010). A study evaluated manic patients before and after valproate treatment, here the levels before valproate treatment were chosen for comparison with healthy controls (Maes et al., 1995). Lastly, one study included patients in an

Fig. 1. Flow chart of the study selection process.

K. Munkholm et al. / Journal of Affective Disorders 144 (2013) 16–27

‘‘elevated’’ state and a ‘‘depressed’’ state and did not state whether these patients fulfilled diagnostic criteria for mood episodes, but instead included symptom rating scale results which were estimated to fulfil criteria and data were included (Hope et al., 2011a). Risk of bias in individual studies was assessed by evaluation of (1) selection bias (clear description of inclusion- and exclusion criteria, matching of control subjects with patients, control for confounders); (2) outcome assessment and measurement (standardization of blood sampling, reporting of assay sensitivity). For the analysis we used Review Manager v. 5 (Cochrane Collaboration, Oxford, United Kingdom). Because of different assay types used (i.e., Flow Cytometry and ELISA) in evaluating cytokine-, cytokine receptor and cytokine antagonist concentrations and because different blood fractions were analyzed in the included studies, we chose to calculate a standardized mean difference (SMD) for the principal summary measure and this was done for every cytokine in each included study. The calculation of the standardized mean difference was done using a random-effects model described by DerSimonian and Laird. This model is more appropriate than a fixed-effect model when the true effect is likely to vary between studies due to heterogeneity not readily explained, and was therefore chosen, since this was apparently the case for the included studies. A random-effects model yields wider confidence intervals and is likely to produce a more conservative estimate. Where cytokines parameters were measured in only one study, these were not included in the quantitative analysis. Heterogeneity was evaluated for combined measures by the chi-square test where a p-value of 0.10 or less was considered significant, because of small sample sizes and a relatively low study number. In addition, inconsistency was calculated using the I-square value, indicating the impact of heterogeneity on the metaanalysis. Sources of heterogeneity were evaluated and risk of bias across studies, publication bias and selective reporting was assessed. Sensitivity analysis was performed where possible.

3. Results 3.1. Study selection Fig. 1 shows the PRISMA flow chart describing the study selection process. The initial literature search identified 195 articles from Medline, 400 articles from Embase, 92 articles from PsycINFO after limiting the search strategy to human studies published in English and one article was additionally identified by hand search. After eliminating duplicates, 58 potential studies were eligible for meta-analysis (Barbosa et al., 2011; Bellani et al., 2010; Boufidou et al., 2004; Breunis et al., 2003; Brietzke et al., 2009a, 2011, 2009b; Brietzke and Teixeira, 2010; Cunha et al., 2008; De Berardis et al., 2008; Dickerson, 2010; Dickerson et al., 2009, 2007; Drexhage et al., 2011a, 2011b, 2010a, 2010c; Guloksuz et al., 2010; Himmerich et al., 2005; Hope et al., 2011a, 2009, 2011b; Hornig et al., 1998; Huang and Lin, 2007; Kapczinski et al., 2011, 2009; Kauer-Sant’Anna, 2011, 2009; Kim et al., 2007, 2004, 2002; Knijff et al., 2007, 2006; Kunz et al., 2011, 2010; Legros et al., 1985; Liu et al., 2004; Maes et al., 1995, 1997; Middle et al., 2000; O’Brien et al., 2006; Ortiz-Dominguez et al., 2007; Pandey, 2011; Prossin et al., 2010; Rapaport, 1994; Rapaport et al., 1999; Remlinger-Molenda et al., 2010; Remlinger et al., 2009; Su et al., 2002; Taylor, 2011; Taylor et al., 2009; Teixeira et al., 2011; Tsai et al., 1999, 2010, 2012, 2001; Vuksan-Cusa et al., 2010; Wadee et al., 2002). Of these a total of 45 studies were excluded based on the publication type being a review or a conference abstract (Bellani et al., 2010; Brietzke et al., 2011; Dickerson, 2010; Dickerson et al., 2009; Drexhage et al., 2011a, 2010a; Hope et al., 2011b; Kapczinski

19

et al., 2009; Kauer-Sant’Anna, 2011; Kunz et al., 2010; Pandey, 2011; Prossin et al., 2010; Remlinger-Molenda et al., 2010; Remlinger et al., 2009; Taylor, 2011; Taylor et al., 2009; Teixeira et al., 2011; Tsai et al., 2010) (n¼18), data were not stratified according to affective state (Himmerich et al., 2005; Rapaport et al., 1999) (n¼2), only reporting in vitro cytokine production (Kim et al., 2007; Knijff et al., 2007, 2006; Liu et al., 2004; Rapaport, 1994; Su et al., 2002; Tsai et al., 1999) (n¼7), not measuring cytokine-, cytokine receptor or cytokine antagonist levels (Brietzke et al., 2009a; Cunha et al., 2008; De Berardis et al., 2008; Dickerson et al., 2007; Drexhage et al., 2010c; Hornig et al., 1998; Huang and Lin, 2007; Legros et al., 1985; Maes et al., 1997; Middle et al., 2000; Vuksan-Cusa et al., 2010; Wadee et al., 2002) (n¼12), significant overlap of study populations (Hope et al., 2009; Kunz et al., 2010) (n¼2), cytokine parameter levels were below detection limit of assay (Boufidou et al., 2004) (n¼1) and mean and/or standard deviation (SD) values of cytokine levels were not available (Breunis et al., 2003; Brietzke et al., 2009b; Kauer-Sant’Anna et al., 2009) (n¼3). Where needed, study authors were contacted for additional data or clarification (n¼7) and all but three replied. Thus, a total of 13 studies were included in the metaanalysis (Barbosa et al., 2011; Brietzke and Teixeira, 2010; Drexhage et al., 2011b; Guloksuz et al., 2010; Hope et al., 2011a; Kapczinski et al., 2011; Kim et al., 2004, 2002; Maes et al., 1995; O’Brien et al., 2006; Ortiz-Dominguez et al., 2007; Tsai et al., 2012, 2001). 3.2. Study characteristics Characteristics of the 13 included studies are described in Table 1. The studies comprised a total of 556 bipolar disorder patients and 767 healthy controls. Eight studies were exclusively cross-sectional (Barbosa et al., 2011; Brietzke and Teixeira, 2010; Drexhage et al., 2011b; Guloksuz et al., 2010; Hope et al., 2011a; Kapczinski et al., 2011; O’Brien et al., 2006; Ortiz-Dominguez et al., 2007) and five studies included a follow-up arm in addition to a cross-sectional design, examining patients in an acute illness phase (mania) and subsequent remission (euthymia) or after treatment (Kim et al., 2004, 2002; Maes et al., 1995; Tsai et al., 2012, 2001). No studies evaluated cytokines-, cytokine receptors- or cytokine antagonists in a hypomanic or a mixed state and there were no follow-up studies of depressed patients. Notably no studies evaluated intra-individual alterations of cytokine parameters between a depressed state and a hypomanic or manic state. In all, 15 different cytokines-, cytokine receptors- or cytokine antagonists were evaluated: Interleukin (IL)-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-12, IL-1b, IL-1 receptor antagonist (IL1RA), soluble IL-2 receptor (sIL-2R), sIL-6R, interferon-g (IFN-g), transforming growth factor-b1 (TGF-b1), soluble tumor necrosis factor receptor type 1 (sTNFR1), sTNFR2 and tumor necrosis factora (TNF-a). Five studies measured serum levels and 8 studies measured plasma levels. All studies used an enzyme-linked immunosorbent assay (ELISA) or enzyme immunoassay (EIA) except two, which used a flow cytometric assay (Drexhage et al., 2011b; Guloksuz et al., 2010). 3.3. Risk of bias and quality in individual studies Inclusion- and exclusion-criteria were described in all of the included studies, and in all cases sampling and assessment were done prospectively. Matching of study population with control group was successful regarding age in most studies and overall no significant differences were seen in the gender distribution between patients and healthy controls. Thus 43% of the patients were male whereas 40% of healthy controls were male. Only five studies reported smoking status among the participants (Hope et al., 2011a; Kim et al., 2004, 2002; Tsai et al., 2012, 2001), and five studies reported BMI of the participants (Guloksuz et al., 2010; Hope et al., 2011a; Kim et al., 2004, 2002; Tsai et al., 2012).

20

K. Munkholm et al. / Journal of Affective Disorders 144 (2013) 16–27

Table 1 Characteristics of studies included in the meta-analysis of cytokine levels in bipolar disorder. Agea

Cytokines measured

Assay

Blood fraction

Medicated

(21–12) (21–12) (21–12) (8–12) (4–16) (8–12) (32–48) (9–29)

31.67 6.0 31.67 6.0 28.97 3.9 37.97 12.1 46.17 9.3 46.67 12.7 40.77 12.5 41.17 9.6

IL1-Ra, sTNF-R1,

ELISA

Plasma

Yes

TNF-a, IL-6, IL-10

ELISA

Serum

Yes

IFN-g, IL-17A, IL-10, IL-6, IL-4, IL-5, IL-8, TNF-a, IL-1b, sIL-2R

Flow cytometry, ELISA for sIL-2R

Serum

Yes

(3–19) (6–11) (23–35) (12–14) (105–134) (13–21)

41.37 9.5 397 9 357 11 367 14 367 10 49.67 14.2

sTNF-R1, IL1-Ra, IL-6

EIA

Plasma

Yes

TNF-a, sTNFR1 and sTNFR2

ELISA

Plasma

Yes

IL-2, IL-4, IL-5, IL-10, IFN-g, TNF-a

Flow cytometry

Serum

16 medication free, 15 on Li monotherapy

sTNFR1, sTNFR2

ELISA

Serum

N.A.

TNF-a, IL-6, IL-1b, IL2, IL-4

ELISA

Serum

No

IL-6, IL-8, IL-10, TNFa, sIL-6R

ELISA

Plasma

Yes

IFN-g, IL-4, TGF-b1

ELISA

Plasma

No

IL-12

ELISA

Plasma

No

sIL-2R, sIL-6R

ELISA

Plasma

13 drug free in acute mania, 18 medicated

IL-6, slL-6R, slL-2R

EIA

Plasma

No (wash-out period followed by Valproate monotherapy on follow-up)

Author, year

Subjects

(Tsai et al., 2012)

Mania Remissionb Control Mania Depression Euthymia Control Euthymia

33 33 33 20 20 20 80 38

(Barbosa et al., 2011)

Control Elevated Depression Euthymia Control Mania

22 17 58 26 239 34

(Guloksuz et al., 2010)

Euthymia Control Euthymia

19 (8–11) 38 (18–20) 31 (23–8)

44.57 10.9 42.97 9.7 32.067 6.69

Control Euthymia

16 (12–4) 30 (N.A.)

31.87 4.8 N.A.

Control Mania

30 (N.A.) 10 (3–7)

N.A. 28.97 8.45

Depression Control Mania

10 (2–8) 33 (5–28) 12 (5–7)

39.77 11.43 32.37 10.8 41,17 (19–64)

(Tsai et al., 2001)

Depressed Control Mania Control Mania Control Mania

9 21 70 96 25 85 31

46,78 (19–68) 36,52 (21–51) 32.97 12.2 30.37 10.1 28.67 8.0 32.57 9.6 31.97 10.0

(Maes et al., 1995)

Remissionb Control Mania

31 (15–16) 31 (15–16) 10 (3–7)

31.97 10.0 33.17 8.7 40.57 13.4

Control

21 (8–13)

33.07 10.1

(Kapczinski et al., 2011)

(Drexhage et al., 2011b)

(Hope et al., 2011a)

(Brietzke and Teixeira, 2010) (Ortiz-Dominguez et al., 2007)

(O’Brien et al., 2006)

(Kim et al., 2004) (Kim et al., 2002)

N (male– female)

(4–5) (9–12) (27–43) (42–54) (12–13) (28–57) (15–16)

ELISA, enzyme-linked immunosorbent assay; EIA, enzyme immunoassay; IL, interleukin; IL1-Ra, IL-1 receptor antagonist; sTNF-R1, soluble tumor necrosis factor receptor 1; sTNF-R2, soluble tumor necrosis factor receptor-2; TNF-a, tumor necrosis factor-a; IFN-g, interferon-g; sIL-2R, soluble IL-2 receptor; sIL-6R, soluble IL-6 receptor; TGFb1, transforming growth factor-b1. a b

Mean7 SD. Remitted after manic episode.

On an outcome level, five of the included studies reported both assay sensitivity and coefficients of variation (Kim et al., 2004; O’Brien et al., 2006; Ortiz-Dominguez et al., 2007; Tsai et al., 2012, 2001). Standardization of the sampling conditions were variably described, five studies performed blood sampling in a fasting state (Kim et al., 2004, 2002; Maes et al., 1995; Tsai et al., 2012, 2001) with the remaining studies failing to describe the fasting state. All but two studies described time of day of blood sampling, of which all were in the morning (Barbosa et al., 2011; Drexhage et al., 2011b; Guloksuz et al., 2010; Kim et al., 2004, 2002; Maes et al., 1995; O’Brien et al., 2006; Ortiz-Dominguez et al., 2007; Tsai et al., 2012, 2001) except one study where blood sampling was done throughout the day (Hope et al., 2011a). 3.4. Synthesis of results The standardized mean differences with 95% confidence intervals for duplicated cytokines, cytokine receptors and cytokine

antagonists are presented in Fig. 2 for each cytokine parameter comparing bipolar disorder patients in a specified affective state with healthy control subjects and listed for each affective state separately. Comparisons between bipolar disorder patients in various affective states are presented in Tables 3–5. Cytokine parameters that have been evaluated in only one study, stratified according to affective state, are omitted from meta-analysis, however, for a full overview of the research area they are included in the Table 2. 3.5. Manic patients Levels of TNF-a (SMD 3.68, 95% CI: 1.35 to 6.01, p o0.01), sIL2R (SMD 0.85, 95% CI: 0.42 to 1.29, p o0.01) and sTNFR1 (SMD 0.87, 95% CI: 0.44 to 1.31, po0.01) were overall significantly increased in bipolar disorder patients in a manic state compared with healthy control subjects. The findings were replicated in two (O’Brien et al., 2006; Ortiz-Dominguez et al., 2007) of four studies

K. Munkholm et al. / Journal of Affective Disorders 144 (2013) 16–27

21

(p ¼0.43), IL-5 (p ¼0.66), sTNFR2 (p¼ 0.40) and TNF-a (p ¼0.64) between euthymic patients and healthy control subjects (Table 2). 3.8. Differences in cytokine levels between affective states The findings of replicated studies of cytokine levels between bipolar disorder patients in various affective states are described in Tables 3–5. Comparisons between euthymic patients and manic patients were replicated for four cytokine parameters. Significantly lower levels of TNF-a (SMD  0.57, CI:  1.10 to  0.03, p ¼0.04) and sTNFR1 (SMD  0.52, CI: 0.94 to  0.10, p¼0.01) in euthymic patients compared with manic patients were found, whereas findings for IL-6 (p ¼0.88) and IL-1RA (p ¼0.10) were not significant. Comparisons between euthymic patients and depressed patients were replicated only for IL-6 with insignificant results overall (p ¼0.25). Comparisons between depressed patients and manic patients were replicated for IL-6 (p ¼0.08) and TNF-a (p ¼0.27), but with insignificant findings. 3.9. Intra-individual alterations of cytokine levels Fig. 2. Cytokine-, cytokine receptor-, or antagonist levels in bipolar disorder according to affective state. Standardized mean difference for individual cytokine parameters are presented for manic patients (yellow bars), depressed patients (red bars) and euthymic patients (blue bars) compared to healthy control subjects. Positive standardized mean differences, represented by bars going to the right, reflect higher levels of cytokines in bipolar patients compared with healthy control subjects, whereas negative standardized mean differences indicate higher levels of cytokines in healthy control subjects. IL, interleukin; IL1-Ra, IL-1 receptor antagonist; sTNF-R1, soluble tumor necrosis factor receptor -1; sTNF-R2, soluble tumor necrosis factor receptor-2; TNF-a, tumor necrosis factor-a; IFN-g, interferon-g; sIL-2R, soluble IL-2 receptor; sIL-6R, soluble IL-6 receptor; TGF-b1, transforming growth factor-b1. (For interpretation of the references to color in this figure legend, the reader is reffered to the web version of this article.)

Only alterations of sIL-2R and sIL-6R were investigated in multiple studies, however these studies were substantially heterogenous. Specifically, one study investigated levels of sIL-2R and sIL-6R before and after treatment with valproate in medication free manic patients (Maes et al., 1995) whereas the other study investigated the same cytokine receptors in manic patients and subsequent remission (Tsai et al., 2001). Overall, findings in the meta-analysis were insignificant for both sIL-2R (SMD 0.23, CI: 0.21 to 0.66, p ¼0.31) and sIL-6R (SMD 0.05, CI: 0.38 to 0.49, p ¼0.81). 3.10. Correlation with clinical features

for TNF-a, two (Barbosa et al., 2011; Hope et al., 2011a) of three studies of sTNFR1 and both of the studies investigating sIL-2R (Maes et al., 1995; Tsai et al., 2001). For other cytokine parameters investigated in multiple studies, there were no overall significant differences in levels of IL-1RA (p ¼0.09), IL-6 (p ¼0.23), IL-4 (p ¼0.22) and sIL-6R (p ¼0.20) between manic patients and healthy control subjects, the studies all demonstrating trends toward higher levels for manic patients (Table 2). 3.6. Depressed patients Cytokine levels among depressed bipolar disorder patients were only replicated for two cytokines with overall increased levels among depressed patients compared with healthy control subjects, however none of these findings were significant. TNF-a was investigated in three studies of which two (O’Brien et al., 2006; Ortiz-Dominguez et al., 2007) found significantly elevated levels among depressed patients, however the overall standardized mean difference was insignificant (p ¼0.08). IL-6 was investigated in three studies of which one (Ortiz-Dominguez et al., 2007) found significantly altered levels between depressed patients and healthy control subjects for an overall insignificant standardized mean difference (p ¼0.20) (Table 2).

Four studies provided data on correlation between smoking status and levels of cytokines-, cytokine receptor- or antagonists, however only IL-1Ra and sTNFR1 were investigated across studies and findings were not replicated (Hope et al., 2011a; Tsai et al., 2012). The same studies were the only to investigate possible associations between identical cytokine parameters and body mass index (BMI), with partial replication of findings of a correlation for IL-1Ra but not for the remaining cytokine parameters investigated. Correlations between cytokine parameter levels and severity of affective symptoms were reported in multiple studies for IL-6, sIL2R, sIL-6R and TNF-a, however no positive finding were replicated (Barbosa et al., 2011; Hope et al., 2011a; Kim et al., 2004, 2002; Maes et al., 1995; O’Brien et al., 2006; Tsai et al., 2001). Likewise, there were no replicated correlation between duration of illness and cytokine parameter levels among the studies evaluating such a correlation (Barbosa et al., 2011; Guloksuz et al., 2010; Kim et al., 2004, 2002, 2001). Of the eight studies reporting on the association between cytokine parameter levels and medication status, four did not find significant correlation between medication status and cytokine levels (Barbosa et al., 2011; Maes et al., 1995; Tsai et al., 2012, 2001) and the remaining four studies demonstrated correlations for individual cytokines parameters but none of these were replicated (Guloksuz et al., 2010; Hope et al., 2011a; Kim et al., 2004, 2002).

3.7. Euthymic patients 3.11. Heterogeneity and sensitivity analysis Only sTNFR1 was significantly altered between euthymic patients and healthy control subjects, with levels increased overall among bipolar patients (SMD 0.58, CI: 0.25 to 0.92, po0.01), a finding that was replicated in two of four studies (Barbosa et al., 2011; Tsai et al., 2012). By contrast there was no overall difference in levels of IL-1RA (p¼0.24), IL-10 (p ¼0.66), IL-6

For at majority of the comparisons, significant heterogeneity in the standardized mean difference was present, supporting the use of a random effects model. For the sensitivity analysis, removing the two studies using flow cytometry yielded no difference in the standardized mean

22

Table 2 Cytokine levels in bipolar disorder patients according to affective state in comparison to healthy control subjects. Outcome

N studies

Depressed sIL-2R sIL-6R IFN-g TNF-a sTNFR1 sTNFR2 TGF-b1 IL-1b IL-2 IL-4 IL-5 IL-6 IL-8 IL-10 IL-12 IL-1RA

patients – – – 3 1 – – 1 1 1 – 3 1 1 – 1

Euthymic patients sIL-2R 1 sIL-6R 1 IFN-g 1 TNF-a 4 sTNFR1 4 sTNFR2 2 – TGF-b1 – IL-1b IL-2 1 IL-4 1 IL-5 2 IL-6 2 IL-8 1 IL-10 3 IL-12 – IL-1RA 2

Healthy controls

SMD (95% CI)

p value

Heterogeneity X2

p value

I2

References

41 41 70 76 84 34 70 10 10 80

52 52 96 172 310 38 96 33 33 129

0.85 (0.42, 1.29) 0.58 ( 0.30, 1.46) 0.15 ( 0.16, 0.45) 3.68 (1.35, 6.01) 0.87 (0.44, 1.31) 0.31 ( 0.15, 0.78)  0.59 ( 0.90,  0.28)  6.62 ( 8.25,  4.99)  3.08 ( 4.06,  2.10) 3.75 (  2.31, 9.81)

o 0.01 0.20 NA o 0.01 o 0.01 NA NA NA NA 0.22

0.35 3.50 NA 104.12 4.67 NA NA NA NA 49.89

0.55 0.06 NA o0.01 0.10 NA NA NA NA o0.01

0 71 NA 97 57 NA NA NA NA 98

(Maes et al., 1995; Tsai et al., 2001) (Maes et al., 1995; Tsai et al., 2001) (Kim et al., 2004) (Barbosa et al., 2011; Kapczinski et al., 2011; O’Brien et al., 2006; Ortiz-Dominguez et al., 2007) (Barbosa et al., 2011; Hope et al., 2011a; Tsai et al., 2012) (Barbosa et al., 2011) (Kim et al., 2004) (Ortiz-Dominguez et al., 2007) (Ortiz-Dominguez et al., 2007) (Kim et al., 2004; Ortiz-Dominguez et al., 2007)

37 12 20 25 50

319 21 80 85 272

0.22 ( 0.13, 0.56) 4.61  0.04 (  0.53, 0.45) 0.51 (0.06, 0.96) 0.55 ( 0.08, 1.17)

0.23 NA NA NA 0.09

0.05 NA NA NA 11.41

0.83 NA NA NA 0.02

0 NA NA NA 65

(Hope et al., 2011a; Kapczinski et al., 2011) (O’Brien et al., 2006) (Kapczinski et al., 2011) (Kim et al., 2002) (Hope et al., 2011a; Tsai et al., 2012)

39 58

134 239

4.31 (  0.57, 9.19) 0.12 ( 0.17, 0.41)

0.08 NA

89.86 NA

o0.01 NA

98 NA

(Kapczinski et al., 2011; O’Brien et al., 2006; Ortiz-Dominguez et al., 2007) (Hope et al., 2011a)

10 10 10

33 33 33

 0.51 ( 1.22, 0.21)  3.03 ( 4.00,  2.05)  0.62 ( 1.35, 0.10)

NA NA NA

NA NA NA

NA NA NA

NA NA NA

(Ortiz-Dominguez et al., 2007) (Ortiz-Dominguez et al., 2007) (Ortiz-Dominguez et al., 2007)

56 9 20

352 21 80

1.04 ( 0.54, 2.62) 7.14 (5.05, 9.23) 0.38 ( 0.11, 0.87)

0.20 NA NA

44.78 NA NA

o0.01 NA NA

96 NA NA

(Hope et al., 2011a; Kapczinski et al., 2011; Ortiz-Dominguez et al., 2007) (O’Brien et al., 2006) (Kapczinski et al., 2011)

58

239

 0.08 (  0.37, 0.21)

NA

NA

NA

NA

(Hope et al., 2011a)

31 31 31 108 108 49

31 31 16 156 340 68

0.56 (0.05, 1.07) 0.19 ( 0.31, 0.69) 0.26 ( 0.34, 0.87)  0.06 (  0.34, 0.21) 0.58 (0.25, 0.92) 0.57 ( 0.77, 1.90)

NA NA NA 0.64 o 0.01 0.40

NA NA NA 3.06 5.48 11.62

NA NA NA 0.38 0.14 o0.01

NA NA NA 2 45 91

(Tsai et al., 2001) (Tsai et al., 2001) (Guloksuz et al., 2010) (Barbosa et al., 2011; Drexhage et al., 2011b; Guloksuz et al., 2010; Kapczinski et al., 2011) (Barbosa et al., 2011; Brietzke and Teixeira, 2010; Hope et al., 2011a; Tsai et al., 2012) (Barbosa et al., 2011; Brietzke and Teixeira, 2010)

45 31 69 46 38 89

41 16 38 319 22 118

0.04 ( 0.56, 0.65) 0.38 ( 0.22, 0.99)  0.01 (  0.52, 0.49) 0.47 ( 0.69, 1.64) 0.57 ( 0.77, 1.90)  0.08 (  0.44, 0.28)

NA NA 0.96 0.43 NA 0.66

NA NA 1.58 13.17 NA 2.69

NA NA 0.21 o0.01 NA 0.26

NA NA 37 92 NA 26

(Guloksuz et al., 2010) (Guloksuz et al., 2010) (Drexhage et al., 2011b; Guloksuz et al., 2010) (Hope et al., 2011a; Kapczinski et al., 2011) (Drexhage et al., 2011b) (Drexhage et al., 2011b; Guloksuz et al., 2010; Kapczinski et al., 2011)

59

272

0.20 ( 0.13, 0.53)

0.24

1.11

0.29

10

(Hope et al., 2011a; Tsai et al., 2012)

2

2

SMD, standardized mean difference; CI, confidence interval; X , chi squared; I , Isquared; IL, interleukin; IL1-Ra, IL-1 receptor antagonist; sTNF-R1, soluble tumor necrosis factor receptor -1; sTNF-R2, soluble tumor necrosis factor receptor-2; TNF-a, tumor necrosis factor-a; IFN-g, interferon-g; sIL-2R, soluble IL-2 receptor; sIL-6R, soluble IL-6 receptor; TGF-b1, transforming growth factor-b1.

K. Munkholm et al. / Journal of Affective Disorders 144 (2013) 16–27

Manic patients sIL-2R 2 sIL-6R 2 1 IFN-g TNF-a 4 sTNFR1 3 sTNFR2 1 1 TGF-b1 1 IL-1b IL-2 1 IL-4 2 IL-5 – IL-6 2 IL-8 1 IL-10 1 IL-12 1 IL-1RA 2

Bipolar patients

K. Munkholm et al. / Journal of Affective Disorders 144 (2013) 16–27

23

Table 3 Inter-individual cytokine alterations between manic patients and euthymic patients. Outcome

TNF-a sTNFR1 IL-6 IL-1RA

N studies

2 2 2 2

Euthymic patients

39 45 46 59

Manic patients

SMD (95% CI)

54 51 37 50

 0.57  0.52 0.07  0.32

p value

( 1.10, ( 0.94, ( 0.80, ( 0.70,

 0.03)  0.10) 0.93) 0.07)

0.04 0.01 0.88 0.10

Heterogeneity X2

p value

1.54 0.28 3.87 0.24

0.21 0.60 0.05 0.62

I2

References

35 0 74 0

(Barbosa (Barbosa (Hope et (Hope et

et al., 2011; Kapczinski et al., 2011) et al., 2011; Hope et al., 2011a) al., 2011a; Kapczinski et al., 2011) al., 2011a; Tsai et al., 2012)

SMD, standardized mean difference; CI, confidence interval; X2, chi squared; I2, Isquared; IL, interleukin; IL1-Ra, IL-1 receptor antagonist; sTNF-R1, soluble tumor necrosis factor receptor -1; TNF-a, tumor necrosis factor-a.

Table 4 Inter-individual cytokine alterations between depressed patients and euthymic patients. Outcome

N studies

Euthymic patients

Depressed patients

SMD (95% CI)

p value

Heterogeneity 2

IL-6

2

46

78

0.48 (  0.34, 1.31)

0.25

X

p value

4.44

0.04

I2

References

100

(Hope et al., 2011a; Kapczinski et al., 2011)

SMD, standardized mean difference; CI, confidence interval; X2, chi squared; I2, Isquared; IL, interleukin.

Table 5 Inter-individual cytokine alterations between depressed patients and manic patients. Outcome N studies

Depressed patients

Manic patients

SMD (95% CI)

p value

Heterogeneity I2 X2

p value

TNF-a

3

39

42

 0.55 (  1.53, 0.43) 0.27

8.22

0.02

IL-6

2

46

37

 0.40 (  0.84, 0.04) 0.08

0.03

0.87

2

References

76 (Kapczinski et al., 2011; O’Brien et al., 2006; Ortiz-Dominguez et al., 2007) 0 (Hope et al., 2011a; Kapczinski et al., 2011)

2

SMD, standardized mean difference; CI, confidence interval; X , chi squared; I , Isquared; IL, interleukin; TNF-a, tumor necrosis factor-a.

difference or heterogeneity in the comparison of TNF-a between euthymic patients and healthy controls and the finding remained insignificant (Drexhage et al., 2011b; Guloksuz et al., 2010). Removing one study in the comparison of TNF-a between depressed patients and healthy controls the standardized mean difference became significant (Kapczinski et al., 2011) whereas removing one of two studies in the same comparison for manic patients made results insignificant (Barbosa et al., 2011; Kapczinski et al., 2011), in both cases without any effect on heterogeneity. For sTNFR1, the heterogeneity was no longer significant when removing one study in the comparison between manic patients and controls (Tsai et al., 2012) and in the comparison between euthymic patients and healthy controls (Barbosa et al., 2011). In the comparison of IL-6 levels between depressed patients and healthy controls, the heterogeneity was no longer significant but the standardized mean difference remained insignificant (Ortiz-Dominguez et al., 2007). Likewise, removing one study for IL-10 comparisons between euthymic patients and healthy controls, the standardized mean difference remained insignificant but the heterogeneity was no longer significant (Guloksuz et al., 2010). Other sensitivity analyses were not possible due to low study numbers. Overall, the results of the sensitivity analysis were not influenced by studies of questionable quality or eligibility, but could rather reflect the limited study numbers. Publication bias was difficult to assess, since there are no databases of conducted studies in the area and statistical analysis of publication bias was considered of low value because of the low study number.

4. Discussion This systematic review and meta-analysis of cytokine-, cytokine receptor and cytokine antagonist levels in bipolar disorder during different affective phases of the illness included 13 studies, comprising 556 bipolar disorder patients and 767 healthy control subjects. It is the first meta-analysis of cytokine levels in bipolar disorder according to affective state. Overall, the meta-analysis demonstrated elevated levels of sTNF-R1, TNF-a and sIL-2R in manic patients compared with healthy control subjects as well as elevated levels of sTNF-R1 and TNF-a in manic patients compared with euthymic patients. sTNFR1 levels were elevated in euthymic patients compared with healthy control subjects. The results lend support to the theory of an activated immune response system in bipolar disorder. TNF-a levels showed a near significant trend towards elevation among depressed patients compared with healthy controls, suggesting TNF-a could possibly represent a state marker in the disease, reflecting overall disease activity, regardless of polarity. There were no significant findings for comparisons between cytokine parameter levels in manic patients and depressed patients, just as there were no significant findings regarding intra-individual alterations of cytokine parameter levels between mood states or as a result of treatment. However, the findings should be interpreted with caution because of significant heterogeneity among studies and low study numbers. A strength in the present study is that the results were stratified according to affective state. For other well-studied

24

K. Munkholm et al. / Journal of Affective Disorders 144 (2013) 16–27

biomarkers in bipolar disorder, such as brain derived neurotrophic factor (BDNF), markers related to oxidative stress and cortisol, state-related alterations have been demonstrated (Andreazza et al., 2008; Daban et al., 2005; Kapczinski et al., 2010b; Lin, 2009). Further, clinical status has been demonstrated to influence cytokine parameter levels in other major psychiatric disorders such as schizophrenia. A recent meta-analysis demonstrated alterations of cytokine parameter levels in acute exacerbations of the disorder, normalizing following antipsychotic treatment (Miller et al., 2011). Taken together, research indicates that biological changes are related to disease activity and highlight the need for elucidating alterations of cytokine parameter levels in bipolar disorder in various affective states. Accordingly, we decided to include only a mood-specific analytic approach in comparing cytokine parameter levels between bipolar disorder patients and healthy controls, as an analysis of global measures of cytokines across all mood states crucially does not inform on state related changes of cytokine parameters. Stratification according to mood state may thus also help prevent type 2 errors, if there are indeed true differences in cytokine parameter levels between mood states, such as indicated by current research. Meta-analyses of observational studies potentially present certain challenges because of inherent biases and differences in study designs (Huston and Naylor, 1996). However, they may also provide additional tools for helping to understand and quantify sources of variability in results across studies (Egger et al., 1998), and the inclusion of a meta-analysis was considered a strength in the present study. There are a number of limitations in the present study. Some studies were excluded for not reporting cytokine parameter levels stratified according to affective state or because they did not report summary data of mean cytokine parameter levels and standard deviation and their inclusion in the meta-analysis could potentially have altered the results, however the direction of such inclusion is uncertain and most likely the addition of studies would allow more robust conclusions due to larger study numbers. While stratifying results according to mood state gave valuable information about the alterations of cytokine parameters in relation to disease activity, it also resulted in relatively low study numbers for some cytokines. Substantial heterogeneity was observed for most cytokine parameters and comparisons, which could be due to a number of factors. First, control for potential confounders of cytokine parameter levels was limited, with only 38% of the studies reporting on BMI or smoking status, none of these producing significant replicated findings. BMI and smoking status along with age and gender have been shown to influence cytokine levels (Esposito et al., 2003; Haack et al., 1999) and the limited control for these confounders could therefore account for some of the heterogeneity observed. Furthermore, other clinical parameters like blood pressure, physical activity and alcohol intake can influence inflammatory markers and thus confound results as these factors were not considered in the included studies (Chae et al., 2001; Imhof et al., 2001; Pischon et al., 2003). Second, while almost all studies performed blood sampling in the morning, only 38% of the studies did so in a fasting state which is important since levels of markers of inflammation differ between the fasting and post prandial state (Manning et al., 2008; Poppitt et al., 2008). Third, we included both medicated and nonmedicated patients. Even though our analysis did not find significant correlations between medication status and cytokine levels, individual studies did in fact find such correlations and other lines of evidence also indicate that antipsychotic medication as well as mood stabilizers can potentially alter cytokine levels (Pollmacher et al., 2000; Rybakowski, 2000), although it is not possible to discriminate between the direct effect of medication and the effect of achieving remission of symptoms. Low study

numbers did not allow selection of only studies examining medication free subjects, though. Fourth, studies with varying assay procedures were included, which potentially could induce heterogeneity and bias results. However, in the sensitivity analysis heterogeneity was not altered when removing the studies not using ELISA, although this analysis was limited because of low study numbers. It is likely though, considering that the included studies reported intra-assay variability of up to 11% and inter assay variability of up to 10%, that the laboratory methods and technical analysis introduce substantial heterogeneity when combining results from different labs and different studies. None of the studies reported on the duration of the current episode. The temporal relation between cytokine parameter alterations during affective episodes and subsequent remission in bipolar disorder has received little attention, and the scarce evidence has been conflicting. One study evaluated levels of IL1Ra and sTNFR1 in manic patients and subsequent partial remission and full remission, finding levels of both elevated in mania with IL-1Ra but not sTNF-R1 levels normalizing upon remission (Tsai et al., 2012). One other study of manic patients found levels of sIL-2R but not sIL-6R to normalize in remission (Tsai et al., 2001). There is a need to clarify whether the observed alterations of cytokine parameter levels in acute affective episodes normalize in remission and also to elucidate the temporal relationship of cytokine parameter alterations and mood symptoms in order to examine whether cytokines could potentially serve as state markers or predictors of affective episodes. It is thus of interest whether the observed alterations of inflammatory markers precede changes in affective state or whether the alterations appear concurrently with alterations in affective state. Among the hypotheses behind the involvement of the inflammatory system in bipolar disorder, the macrophage-T-lymphocyte theory, initially proposed for depression (Leonard, 2001; Smith, 1991), has been extended to bipolar disorder. This theory postulates a chronically activated immune response system mediated by macrophages (and in the brain, microglia) and T-lymphocytes as a basis for the disease, in which their production of cytokines and inflammatory substances impart a vulnerability and destabilization of the brain function, rendering it susceptible to environmental stressors which, in turn, result in mood disturbances. Current hypothesis of bipolar disorder, which extends to neurotrophic alterations and mitochondrial disturbances, could place inflammatory system aberrations as a central integrating mechanism in mood disorders. Impaired neuroplasticity has been proposed as a pathophysiological mechanism in mood disorders (Duman, 2004; Duman and Monteggia, 2006; Manji et al., 2000), which could, in part, be mediated by cytokines. Decreased levels of BDNF, which plays a central role in synaptic plasticity and neuronal survival (Poo, 2001), have been documented in acute phases of bipolar illness, with levels appearing to normalize upon treatment after acute mania (Fernandes et al., 2011). The specific association between peripheral levels of cytokines and neurotrophins needs further clarification. One study of adult patients in early- and late-stage bipolar disorder found BDNF levels positively associated with IL-6 levels but negatively associated with TNF-a (Kauer-Sant’Anna et al., 2009), while a negative association between BDNF and IL-6 levels was found in a study of adolescents with bipolar disorder in various affective states (Goldstein et al., 2011). Inflammation has widespread and negative effects on brain neurogenesis (Ekdahl et al., 2003). This is mediated largely by activated microglia (Monje et al., 2003), which, formed by tissue macrophages and immature dendritic cells, constitute the inflammatory system in the brain and have also been demonstrated to produce neurotrophic factors in steady state (Kim and de Vellis, 2005). Both post-mortem and

K. Munkholm et al. / Journal of Affective Disorders 144 (2013) 16–27

in-vivo studies support the role for glial pathology in mood disorders, forming the basis of a glial theory for mood disorders (Rajkowska, 2000; Schroeter et al., 2011). Multiple proinflammatory cytokines produced by activated microglia, including IL-6, TNF-a, IL-1b and IFN-g have antineurogenic properties (Ben-Hur et al., 2003; Ekdahl et al., 2003; Liu et al., 2005; Monje et al., 2003). In support of this, we found elevated levels of TNF-a and its soluble receptor, sTNF-R1 in bipolar patients; however, the findings for IL1b and IFN-g were not replicated. We also found significantly elevated levels of sIL-2R, indicating T-cell activation during mania. Emerging evidence points to the involvement of oxidative stress and mitochondrial dysfunction as possible pathophysiological mechanisms in bipolar disorder (Ng et al., 2008; Steckert et al., 2010). Specifically, elevated levels of antioxidative enzymes have been demonstrated in affective episodes while elevated markers of lipid peroxidation have been found independently of the phase of illness (Andreazza et al., 2007; Machado-Vieira et al., 2007). Also, alterations of the activity and function of the mitochondrial electron transport chain have been demonstrated in bipolar disorder (Andreazza, 2009; Konradi et al., 2004). The proinflammatory cytokine IL-6, produced to a large extent by activated glia, has been demonstrated to stimulate production of reactive oxygen species in the brain (Behrens et al., 2008) and another proinflammatory cytokine, TNF-a has been shown to induce mitochondrial damage via suppression of the activity of mitochondrial complexes (Zell et al., 1997). Recent evidence indicate that mitochondrial reactive oxygen species act as signalling molecules in triggering proinflammatory cytokine production (Naik and Dixit, 2011), pointing to a bidirectional relationship between the inflammatory system and mitochondrial dysfunction. Abnormalities of the hypothalamic-pituitary-adrenal (HPA) axis have repeatedly been demonstrated in bipolar disorder (Daban et al., 2005) and cytokines interact with the HPA system in several ways, possibly by exerting negative effects on neurogenesis (Cameron and Gould, 1994). The function and expression of the glucocorticoid receptor can be modulated by cytokines (Miller et al., 1999) and cytokines released as part of the stress response can similarly directly stimulate the hypothalamic release of corticotropin releasing hormone (CRH) and pituitary release of adrenocorticotropic hormone (ACTH) (Black, 1994; Chrousos, 1995). It will be important for future research to explore the extent of state related alterations of cytokines in bipolar disorder in order to elucidate whether the inflammatory changes observed reflect a specific temporal relationship with alterations in psychopathology and, importantly, whether cytokines could serve as markers of disease activity or markers of vulnerability. Stratifying patients by clinical status is therefore recommended. Diminishing heterogeneity within studies and between studies is of great importance, and in this regard careful control for confounders known to affect cytokine levels (O’Connor et al., 2009) are of paramount importance as are high levels of standardization of blood sampling and laboratory procedures. Considering that actual differences in cytokine concentrations between groups were rather small compared with substantial variation in levels observed in individual studies alongside considerable intra-assay variation coefficients, true differences may easily be overlooked. This could be particularly true of casecontrol studies, where phenotypical and genetic differences among participants can be substantial. It is therefore recommended that future studies also include longitudinal aspects, evaluating intraindividual alterations of cytokine parameter levels across affective states, which is a study design that could prove more sensitive in uncovering subtle alterations in immune status. Along the same lines, it is surprising that in bipolar disorder, a disease that is characterized by recurrent shifts between affective states, very few

25

studies have explored intra-individual differences in cytokine levels between different affective states. In conclusion, we found evidence of increased levels of the proinflammatory cytokine TNF-a and it is receptor, sTNF-R1 along with increased levels of sIL-2R in manic patients and increased levels of sTNF-R1 in euthymic patients, supporting the macrophage-T-lymphocyte theory of bipolar disorder. Overall, results were limited by low study numbers, a lack of replicability and heterogeneity between studies. Future studies in this important area of research are warranted.

Role of funding source The study was supported by grants from the Lundbeck Foundation, Denmark (R34-A3696) and the Danish Council for Independent Research 9 Medical Sciences (09-073972).

Conflict of interest Maj Vinberg has been a consultant for Eli Lilly, AstraZeneca, Servier and Janssen-Cilag. Lars Vedel Kessing has been a consultant for Bristol-Myers Squibb, Eli Lilly, Lundbeck, AstraZeneca, Pfizer, Wyeth, Servier and Janssen-Cilag. Klaus Munkholm has no conflicts of interest to disclose.

Acknowledgments This study was supported by grants from the Lundbeck Foundation, Denmark (R34-A3696) and the Danish Council for Independent Research 9 Medical Sciences (09-073972).

References Andreazza, A.C., 2009. S02-02 mitochondrial dysfunction and oxidative stress in bipolar disorder. European Psychiatry 24 (Supplement 1), S15. Andreazza, A.C., Cassini, C., Rosa, A.R., Leite, M.C., de Almeida, L.M., Nardin, P., Cunha, A.B., Cereser, K.M., Santin, A., Gottfried, C., Salvador, M., Kapczinski, F., Goncalves, C.A., 2007. Serum S100B and antioxidant enzymes in bipolar patients. Journal of Psychiatric Research 41, 523–529. Andreazza, A.C., Kauer-Sant’anna, M., Frey, B.N., Bond, D.J., Kapczinski, F., Young, L.T., Yatham, L.N., 2008. Oxidative stress markers in bipolar disorder: a metaanalysis. Journal of Affective Disorders. Barbosa, I.G., Huguet, R.B., Mendonca, V.A., Sousa, L.P., Neves, F.S., Bauer, M.E., Teixeira, A.L., 2011. Increased plasma levels of soluble TNF receptor I in patients with bipolar disorder. European Archives of Psychiatry and Clinical Neuroscience 261, 139–143. Behrens, M.M., Ali, S.S., Dugan, L.L., 2008. Interleukin-6 mediates the increase in NADPH-oxidase in the ketamine model of schizophrenia. Journal of Neuroscience 28, 13957–13966. Bellani, M., Bergami, A., Tomelleri, L., Perlini, C., Cerruti, S., Ferro, A., Marinelli, V., Tansella, M., Furlan, R., Martino, G.V., Brambilla, P., 2010. Altered mRNA levels of chemokines and cytokines in schizophrenia and bipolar disorder. Schizophrenia Research 117 (2–3), 251–252. Ben-Hur, T., Ben-Menachem, O., Furer, V., Einstein, O., Mizrachi-Kol, R., Grigoriadis, N., 2003. Effects of proinflammatory cytokines on the growth, fate, and motility of multipotential neural precursor cells. Molecular and Cellular Neurosciences 24, 623–631. Berk, M., 2009. Neuroprogression: pathways to progressive brain changes in bipolar disorder. International Journal of Neuropsychopharmacology 12, 441–445. Berk, M., Kapczinski, F., Andreazza, A.C., Dean, O.M., Giorlando, F., Maes, M., Yucel, M., Gama, C.S., Dodd, S., Dean, B., Magalhaes, P.V., Amminger, P., McGorry, P., Malhi, G.S., 2010. Pathways underlying neuroprogression in bipolar disorder: focus on inflammation; oxidative stress and neurotrophic factors. Neuroscience & Biobehavioral Reviews. Black, P.H., 1994. Immune system-central nervous system interactions: effect and immunomodulatory consequences of immune system mediators on the brain. Antimicrobial Agents and Chemotherapy 38, 7–12. Bosetti, F., Rintala, J., Seemann, R., Rosenberger, T.A., Contreras, M.A., Rapoport, S.I., Chang, M.C., 2002. Chronic lithium downregulates cyclooxygenase-2 activity and prostaglandin E(2) concentration in rat brain. Molecular Psychiatry 7, 845–850. Boufidou, F., Nikolaou, C., Alevizos, B., Liappas, I.A., Christodoulou, G.N., 2004. Cytokine production in bipolar affective disorder patients under lithium treatment. Journal of Affective Disorders 82, 309–313. Breunis, M.N., Kupka, R.W., Nolen, W.A., Suppes, T., Denicoff, K.D., Leverich, G.S., Post, R.M., Drexhage, H.A., 2003. High numbers of circulating activated T cells and raised levels of serum IL-2 receptor in bipolar disorder. Biological Psychiatry 53, 157–165.

26

K. Munkholm et al. / Journal of Affective Disorders 144 (2013) 16–27

Brietzke, E., Kauer-Sant’Anna, M., Teixeira, A.L., Kapczinski, F., 2009a. Abnormalities in serum chemokine levels in euthymic patients with bipolar disorder. Brain, Behavior, and Immunity 23, 1079–1082. Brietzke, E., Stabellini, R., Grassi-Oliveira, R., Lafer, B., 2011. Cytokines in bipolar disorder: recent findings, deleterious effects but promise for future therapeutics. CNS Spectrums, 16. Brietzke, E., Stertz, L., Fernandes, B.S., Kauer-Sant’anna, M., Mascarenhas, M., Escosteguy Vargas, A., Chies, J.A., Kapczinski, F., 2009b. Comparison of cytokine levels in depressed, manic and euthymic patients with bipolar disorder. Journal of Affective Disorders 116, 214–217. Brietzke, E., Teixeira, A.L., 2010. Similar immune profile in bipolar disorder and schizophrenia: selective increase in soluble tumor necrosis factor receptor I and von Willebrand factor. Bipolar Disorder 12, 453–454. Cameron, H.A., Gould, E., 1994. Adult neurogenesis is regulated by adrenal steroids in the dentate gyrus. Neuroscience 61, 203–209. Chae, C.U., Lee, R.T., Rifai, N., Ridker, P.M., 2001. Blood pressure and inflammation in apparently healthy men. Hypertension 38, 399–403. Chrousos, G.P., 1995. The hypothalamic-pituitary-adrenal axis and immunemediated inflammation. New England Journal of Medicine 332, 1351–1362. Cunha, A.B., Andreazza, A.C., Gomes, F.A., Frey, B.N., da Silveira, L.E., Goncalves, C.A., Kapczinski, F., 2008. Investigation of serum high-sensitive C-reactive protein levels across all mood states in bipolar disorder. European Archives of Psychiatry and Clinical Neuroscience 258, 300–304. Daban, C., Vieta, E., Mackin, P., Young, A.H., 2005. Hypothalamic-pituitary-adrenal axis and bipolar disorder. Psychiatric Clinics of North America 28, 469–480. De Berardis, D., Conti, C.M., Campanella, D., Carano, A., Scali, M., Valchera, A., Serroni, N., Pizzorno, A.M., D’Albenzio, A., Fulcheri, M., Gambi, F., La Rovere, R., Cotellessa, C., Salerno, R.M., Ferro, F.M., 2008. Evaluation of C-reactive protein and total serum cholesterol in adult patients with bipolar disorder. International Journal of Immunopathology and Pharmacology 21, 319–324. Dickerson, F., 2010. Immunologic abnormalities associated with mania. Brain, Behavior, and Immunity 24, S54. Dickerson, F., Copp, C., Stallings, C., Origoni, A., Khushalani, S., Yolken, R., 2009. C-Reactive protein in acute mania: a longitudinal study. Bipolar Disorders 11 (S1), 34. Dickerson, F., Stallings, C., Origoni, A., Boronow, J., Yolken, R., 2007. Elevated serum levels of C-reactive protein are associated with mania symptoms in outpatients with bipolar disorder. Progress in Neuro-Psychopharmacology & Biological PsychiatryisBiological Psychiatry 31, 952–955. Drexhage, H.A., Drexhage, R.C., Mesman, E., Van Beveren, N., Hillegers, M., Nolen, W.A., 2011a. Immune activation in psychiatric and pre-psychiatric disease. European Psychiatry, 26. Drexhage, R.C., Hoogenboezem, T.H., Versnel, M.A., Berghout, A., Nolen, W.A., Drexhage, H.A., 2011b. The activation of monocyte and T cell networks in patients with bipolar disorder. Brain, Behavior, and Immunity 25, 1206–1213. Drexhage, R.C., Knijff, E.M., Padmos, R.C., Heul-Nieuwenhuijzen, L., Beumer, W., Versnel, M.A., Drexhage, H.A., 2010a. The mononuclear phagocyte system and its cytokine inflammatory networks in schizophrenia and bipolar disorder. Expert Review of Neurotherapeutics 10, 59–76. Drexhage, R.C., van der Heul-Nieuwenhuijsen, L., Padmos, R.C., van Beveren, N., Cohen, D., Versnel, M.A., Nolen, W.A., Drexhage, H.A., 2010b. Inflammatory gene expression in monocytes of patients with schizophrenia: overlap and difference with bipolar disorder. A study in naturalistically treated patients. International Journal of Neuropsychopharmacology, 1–13. Drexhage, R.C., van der Heul-Nieuwenhuijsen, L., Padmos, R.C., van Beveren, N., Cohen, D., Versnel, M.A., Nolen, W.A., Drexhage, H.A., 2010c. Inflammatory gene expression in monocytes of patients with schizophrenia: overlap and difference with bipolar disorder. A study in naturalistically treated patients. International Journal of Neuropsychopharmacology 13, 1369–1381. Duman, R.S., 2004. Role of neurotrophic factors in the etiology and treatment of mood disorders. Neuromolecular Medicine 5, 11–25. Duman, R.S., Monteggia, L.M., 2006. A neurotrophic model for stress-related mood disorders. Biological Psychiatry 59, 1116–1127. Egger, M., Schneider, M., Davey Smith, G., 1998. Spurious precision? Meta-analysis of observational studies. British Medical Journal 316, 140–144. Ekdahl, C.T., Claasen, J.H., Bonde, S., Kokaia, Z., Lindvall, O., 2003. Inflammation is detrimental for neurogenesis in adult brain. Proceedings of the National academy of Sciences of the United States of America 100, 13632–13637. Esposito, K., Pontillo, A., Di Palo, C., Giugliano, G., Masella, M., Marfella, R., Giugliano, D., 2003. Effect of weight loss and lifestyle changes on vascular inflammatory markers in obese women: a randomized trial. JAMA 289, 1799–1804. Fernandes, B.S., Gama, C.S., Cereser, K.M., Yatham, L.N., Fries, G.R., Colpo, G., de Lucena, D., Kunz, M., Gomes, F.A., Kapczinski, F., 2011. Brain-derived neurotrophic factor as a state-marker of mood episodes in bipolar disorders: a systematic review and meta-regression analysis. Journal of Psychiatric Research 45, 995–1004. Gardner, A., Boles, R.G., 2010. Beyond the serotonin hypothesis: Mitochondria, inflammation and neurodegeneration in major depression and affective spectrum disorders. Progress in Neuro-Psychopharmacology & Biological Psychiatry. Goldstein, B.I., Collinger, K.A., Lotrich, F., Marsland, A.L., Gill, M.K., Axelson, D.A., Birmaher, B., 2011. Preliminary findings regarding proinflammatory markers and brain-derived neurotrophic factor among adolescents with bipolar spectrum disorders. Journal of Child and Adolescent Psychopharmacology 21, 479–484.

Goldstein, B.I., Kemp, D.E., Soczynska, J.K., McIntyre, R.S., 2009. Inflammation and the phenomenology, pathophysiology, comorbidity, and treatment of bipolar disorder: a systematic review of the literature. Journal of Clinical Psychiatry 70, 1078–1090. Grande, I., Magalhaes, P.V., Kunz, M., Vieta, E., Kapczinski, F., 2011. Mediators of allostasis and systemic toxicity in bipolar disorder. Physiology & Behavior. Guloksuz, S., Cetin, E.A., Cetin, T., Deniz, G., Oral, E.T., Nutt, D.J., 2010. Cytokine levels in euthymic bipolar patients. Journal of Affective Disorders 126, 458–462. Haroon, E., Raison, C.L., Miller, A.H., 2012. Psychoneuroimmunology meets neuropsychopharmacology: translational implications of the impact of inflammation on behavior. Neuropsychopharmacology 37, 137–162. Himmerich, H., Koethe, D., Schuld, A., Yassouridis, A., Pollmacher, T., 2005. Plasma levels of leptin and endogenous immune modulators during treatment with carbamazepine or lithium. Psychopharmacology (Berlin) 179, 447–451. Hope, S., Dieset, I., Agartz, I., Steen, N.E., Ueland, T., Melle, I., Aukrust, P., Andreassen, O.A., 2011a. Affective symptoms are associated with markers of inflammation and immune activation in bipolar disorders but not in schizophrenia. Journal of Psychiatric Research 45, 1608–1616. Hope, S., Melle, I., Aukrust, P., Steen, N.E., Birkenaes, A.B., Lorentzen, S., Agartz, I., Ueland, T., Andreassen, O.A., 2009. Similar immune profile in bipolar disorder and schizophrenia: selective increase in soluble tumor necrosis factor receptor I and von Willebrand factor. Bipolar Disorder 11, 726–734. Hope, S., Melle, I., Ueland, T., Lorentzen, S., Steen, N.E., Berg, A.O., Dieset, I., Agartz, I., Aukrust, P., Andreassen, O.A., 2011b. Interleukin 1 receptor antagonist and tumor necrosis factor receptor 1 are associated with general severity and psychotic symptoms in schizophrenia and bipolar disorder. Schizophrenia Bulletin 37, 115–116. Hornig, M., Goodman, D.B., Kamoun, M., Amsterdam, J.D., 1998. Positive and negative acute phase proteins in affective subtypes. Journal of Affective Disorders 49, 9–18. Huang, T.L., Lin, F.C., 2007. High-sensitivity C-reactive protein levels in patients with major depressive disorder and bipolar mania. Progress in Neuro-Psychopharmacology & Biological Psychiatry 31, 370–372. Huston, P., Naylor, C.D., 1996. Health services research: reporting on studies using secondary data sources. CMAJ: Canadian Medical Association Journal¼ Journal de l’Association Medicale Canadienne 155, 1697–1709. Haack, M., Hinze-Selch, D., Fenzel, T., Kraus, T., Kuhn, M., Schuld, A., Pollmacher, T., 1999. Plasma levels of cytokines and soluble cytokine receptors in psychiatric patients upon hospital admission: effects of confounding factors and diagnosis. Journal of Psychiatric Research 33, 407–418. Imhof, A., Froehlich, M., Brenner, H., Boeing, H., Pepys, M.B., Koenig, W., 2001. Effect of alcohol consumption on systemic markers of inflammation. Lancet 357, 763–767. Kapczinski, F., Dal-Pizzol, F., Teixeira, A.L., Magalhaes, P.V., Kauer-Sant’anna, M., Klamt, F., Moreira, J.C., Augusto de Bittencourt Pasquali, M., Fries, G.R., Quevedo, J., Gama, C.S., Post, R., 2010a. Peripheral biomarkers and illness activity in bipolar disorder. Journal of Psychiatric Research.. Kapczinski, F., Dal-Pizzol, F., Teixeira, A.L., Magalhaes, P.V., Kauer-Sant’Anna, M., Klamt, F., Moreira, J.C., de Bittencourt Pasquali, M.A., Fries, G.R., Quevedo, J., Gama, C.S., Post, R., 2011. Peripheral biomarkers and illness activity in bipolar disorder. Journal of Psychiatric Research 45, 156–161. Kapczinski, F., Dal-Pizzol, F., Teixeira, A.L., Magalhaes, P.V., Kauer-Sant’Anna, M., Klamt, F., Pasquali, M.A., Quevedo, J., Gama, C.S., Post, R., 2010b. A systemic toxicity index developed to assess peripheral changes in mood episodes. Molecular Psychiatry 15, 784–786. Kapczinski, F., Dias, V.V., Kauer-Sant’Anna, M., Brietzke, E., Vazquez, G.H., Vieta, E., Berk, M., 2009. The potential use of biomarkers as an adjunctive tool for staging bipolar disorder. Progress in Neuro-Psychopharmacology & Biological Psychiatry 33, 1366–1371. Kapczinski, F., Frey, B.N., Kauer-Sant’Anna, M., Grassi-Oliveira, R., 2008. Brainderived neurotrophic factor and neuroplasticity in bipolar disorder. Expert Review of Neurotherapeutics 8, 1101–1113. Kauer-Sant’Anna, M., 2011. Brain-derived neurotrophic factor and inflammatory markers in patients with early-vs. late-state bipolar disorder. Biological Psychiatry 1, 123S. Kauer-Sant’Anna, M., Kapczinski, F., Andreazza, A.C., Bond, D.J., Lam, R.W., Young, L.T., Yatham, L.N., 2009. Brain-derived neurotrophic factor and inflammatory markers in patients with early- vs. late-stage bipolar disorder. International Journal of Neuropsychopharmacology 12, 447–458. Kim, S.U., de Vellis, J., 2005. Microglia in health and disease. Journal of Neuroscience Research 81, 302–313. Kim, Y.K., Jung, H.G., Myint, A.M., Kim, H., Park, S.H., 2007. Imbalance between proinflammatory and anti-inflammatory cytokines in bipolar disorder. Journal of Affective Disorders 104, 91–95. Kim, Y.K., Myint, A.M., Lee, B.H., Han, C.S., Lee, S.W., Leonard, B.E., Steinbusch, H.W., 2004. T-helper types 1, 2, and 3 cytokine interactions in symptomatic manic patients. Psychiatry Research 129, 267–272. Kim, Y.K., Suh, I.B., Kim, H., Han, C.S., Lim, C.S., Choi, S.H., Licinio, J., 2002. The plasma levels of interleukin-12 in schizophrenia, major depression, and bipolar mania: effects of psychotropic drugs. Molecular Psychiatry 7, 1107–1114. Knijff, E.M., Breunis, M.N., Kupka, R.W., de Wit, H.J., Ruwhof, C., Akkerhuis, G.W., Nolen, W.A., Drexhage, H.A., 2007. An imbalance in the production of IL-1beta and IL-6 by monocytes of bipolar patients: restoration by lithium treatment. Bipolar Disorder 9, 743–753.

K. Munkholm et al. / Journal of Affective Disorders 144 (2013) 16–27

Knijff, E.M., Breunis, M.N., van Geest, M.C., Kupka, R.W., Ruwhof, C., de Wit, H.J., Nolen, W.A., Drexhage, H.A., 2006. A relative resistance of T cells to dexamethasone in bipolar disorder. Bipolar Disorder 8, 740–750. Konradi, C., Eaton, M., MacDonald, M.L., Walsh, J., Benes, F.M., Heckers, S., 2004. Molecular evidence for mitochondrial dysfunction in bipolar disorder. Archives of General Psychiatry 61, 300–308. Kunz, M., Cereser, K.M., Goi, P.D., Fries, G.R., Teixeira, A.L., Fernandes, B.S., Belmonte-de-Abreu, P.S., Kauer-Sant’Anna, M., Kapczinski, F., Gama, C.S., 2011. Serum levels of IL-6, IL-10 and TNF-alpha in patients with bipolar disorder and schizophrenia: differences in pro- and anti-inflammatory balance. Revista Brasileira de Psiquiatria 33, 268–274. Kunz, M., Cereser, K.M.M., Fries, G.R., Teixeira, A., Fernandes, B.S., Kauer-Sant’Anna, M., Kapczinski, F., Gama, C.S., 2010. Serum levels of Il6, Il10 and TNF-alpha in patients with bipolar disorder in euthymia and with schizophrenia: differences in pro and anti-inflammatory balance. Biological Psychiatry 1, 132S. Lee, H.J., Ertley, R.N., Rapoport, S.I., Bazinet, R.P., Rao, J.S., 2008. Chronic administration of lamotrigine downregulates COX-2 mRNA and protein in rat frontal cortex. Neurochemical Research 33, 861–866. Legros, S., Mendlewicz, J., Wybran, J., 1985. Immunoglobulins, autoantibodies and other serum protein fractions in psychiatric disorders. European Archives of Psychiatry and Neurological Sciences 235, 9–11. Leonard, B.E., 2001. The immune system, depression and the action of antidepressants. Progress in Neuro-Psychopharmacology & Biological Psychiatry 25, 767–780. Lin, P.Y., 2009. State-dependent decrease in levels of brain-derived neurotrophic factor in bipolar disorder: a meta-analytic study. Neuroscience Letters 466, 139–143. Liu, H.C., Yang, Y.Y., Chou, Y.M., Chen, K.P., Shen, W.W., Leu, S.J., 2004. Immunologic variables in acute mania of bipolar disorder. Journal of Neuroimmunology 150, 116–122. Liu, Y.P., Lin, H.I., Tzeng, S.F., 2005. Tumor necrosis factor-alpha and interleukin-18 modulate neuronal cell fate in embryonic neural progenitor culture. Brain Research 1054, 152–158. Machado-Vieira, R., Andreazza, A.C., Viale, C.I., Zanatto, V., Cereser Jr., V., da Silva Vargas, R., Kapczinski, F., Portela, L.V., Souza, D.O., Salvador, M., Gentil, V., 2007. Oxidative stress parameters in unmedicated and treated bipolar subjects during initial manic episode: a possible role for lithium antioxidant effects. Neuroscience Letters 421, 33–36. Maes, M., Bosmans, E., Calabrese, J., Smith, R., Meltzer, H.Y., 1995. Interleukin-2 and interleukin-6 in schizophrenia and mania: effects of neuroleptics and mood stabilizers. Journal of Psychiatric Research 29, 141–152. Maes, M., Delange, J., Ranjan, R., Meltzer, H.Y., Desnyder, R., Cooremans, W., Scharpe, S., 1997. Acute phase proteins in schizophrenia, mania and major depression: modulation by psychotropic drugs. Psychiatry Research 66, 1–11. Maes, M., Song, C., Lin, A.H., Pioli, R., Kenis, G., Kubera, M., Bosmans, E., 1999. In vitro immunoregulatory effects of lithium in healthy volunteers. Psychopharmacology (Berlin) 143, 401–407. Manji, H.K., Moore, G.J., Rajkowska, G., Chen, G., 2000. Neuroplasticity and cellular resilience in mood disorders. Molecular Psychiatry 5, 578–593. Manning, P.J., Sutherland, W.H., McGrath, M.M., de Jong, S.A., Walker, R.J., Williams, M.J., 2008. Postprandial cytokine concentrations and meal composition in obese and lean women. Obesity (Silver Spring) 16, 2046–2052. Middle, F., Jones, I., Robertson, E., Lendon, C., Craddock, N., 2000. Tumour necrosis factor alpha and bipolar affective puerperal psychosis. Psychiatric Genetics 10, 195–198. Miller, A.H., Maletic, V., Raison, C.L., 2009. Inflammation and its discontents: the role of cytokines in the pathophysiology of major depression. Biological Psychiatry 65, 732–741. Miller, A.H., Pariante, C.M., Pearce, B.D., 1999. Effects of cytokines on glucocorticoid receptor expression and function. Glucocorticoid resistance and relevance to depression. Advances in Experimental Medicine and Biology 461, 107–116. Miller, B.J., Buckley, P., Seabolt, W., Mellor, A., Kirkpatrick, B., 2011. Meta-analysis of cytokine alterations in schizophrenia: clinical status and antipsychotic effects. Biological Psychiatry 70, 663–671. Moher, D., Liberati, A., Tetzlaff, J., Altman, D.G., 2009. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. British Medical Journal 339, b2535. Monje, M.L., Toda, H., Palmer, T.D., 2003. Inflammatory blockade restores adult hippocampal neurogenesis. Science 302, 1760–1765. Naik, E., Dixit, V.M., 2011. Mitochondrial reactive oxygen species drive proinflammatory cytokine production. Journal of Experimental Medicine 208, 417–420. Ng, F., Berk, M., Dean, O., Bush, A.I., 2008. Oxidative stress in psychiatric disorders: evidence base and therapeutic implications. International Journal of Neuropsychopharmacology 11, 851–876. O’Brien, S.M., Scully, P., Scott, L.V., Dinan, T.G., 2006. Cytokine profiles in bipolar affective disorder: focus on acutely ill patients. Journal of Affective Disorders 90, 263–267. O’Connor, M.F., Bower, J.E., Cho, H.J., Creswell, J.D., Dimitrov, S., Hamby, M.E., Hoyt, M.A., Martin, J.L., Robles, T.F., Sloan, E.K., Thomas, K.S., Irwin, M.R., 2009. To assess, to control, to exclude: effects of biobehavioral factors on circulating inflammatory markers. Brain, Behavior, and Immunity 23, 887–897. Ortiz-Dominguez, A., Hernandez, M.E., Berlanga, C., Gutierrez-Mora, D., Moreno, J., Heinze, G., Pavon, L., 2007. Immune variations in bipolar disorder: phasic differences. Bipolar Disorder 9, 596–602.

27

Padmos, R.C., Hillegers, M.H., Knijff, E.M., Vonk, R., Bouvy, A., Staal, F.J., de Ridder, D., Kupka, R.W., Nolen, W.A., Drexhage, H.A., 2008. A discriminating messenger RNA signature for bipolar disorder formed by an aberrant expression of inflammatory genes in monocytes. Archives of General Psychiatry 65, 395–407. Pandey, G.N., 2011. Proinflammatory cytokines and brain-derived neurotrophic factor in patients with pediatric bipolar disorders. Biological Psychiatry 1, 123S. Pischon, T., Hankinson, S.E., Hotamisligil, G.S., Rifai, N., Rimm, E.B., 2003. Leisuretime physical activity and reduced plasma levels of obesity-related inflammatory markers. Obesity Research 11, 1055–1064. Pollmacher, T., Haack, M., Schuld, A., Kraus, T., Hinze-Selch, D., 2000. Effects of antipsychotic drugs on cytokine networks. Journal of Psychiatric Research 34, 369–382. Poo, M.M., 2001. Neurotrophins as synaptic modulators. Nature Reviews Neuroscience 2, 24–32. Poppitt, S.D., Keogh, G.F., Lithander, F.E., Wang, Y., Mulvey, T.B., Chan, Y.K., McArdle, B.H., Cooper, G.J., 2008. Postprandial response of adiponectin, interleukin-6, tumor necrosis factor-alpha, and C-reactive protein to a highfat dietary load. Nutrition 24, 322–329. Prossin, A., McInnis, M., Zalcman, S., Ellingrod, V., 2010. IL-6 is associated with metabolic syndrome in bipolar disorder. Brain, Behavior, and Immunity 24, S47. Rajkowska, G., 2000. Postmortem studies in mood disorders indicate altered numbers of neurons and glial cells. Biological Psychiatry 48, 766–777. Rapaport, M.H., 1994. Immune parameters in euthymic bipolar patients and normal volunteers. Journal of Affective Disorders 32, 149–156. Rapaport, M.H., Guylai, L., Whybrow, P., 1999. Immune parameters in rapid cycling bipolar patients before and after lithium treatment. Journal of Psychiatric Research 33, 335–340. Rapaport, M.H., Manji, H.K., 2001. The effects of lithium on ex vivo cytokine production. Biological Psychiatry 50, 217–224. Remlinger-Molenda, A., Wojciak, P., Kostrzewa, A., Karczewski, J., Wiktorowicz, K., Rybakowski, J.K., 2010. Cytokine profiles in bipolar affective disorder during acute episodes and in remission. European Neuropsychopharmacology 20, S415–S416. Remlinger, A., Wojciak, P., Karczewski, J., Kostrzewa, A., Wiktorowicz, K., Rybakowski, J.K., 2009. The effect of mania on cytokine status in bipolar illness. Journal of Cancer Education 24, S462. Rybakowski, J.K., 2000. Antiviral and immunomodulatory effect of lithium. Pharmacopsychiatry 33, 159–164. Schroeter, M.L., Steiner, J., Mueller, K., 2011. Glial pathology is modified by age in mood disorders—a systematic meta-analysis of serum S100B in vivo studies. Journal of Affective Disorders 134, 32–38. Smith, R.S., 1991. The macrophage theory of depression. Medical Hypotheses 35, 298–306. Steckert, A.V., Valvassori, S.S., Moretti, M., Dal-Pizzol, F., Quevedo, J., 2010. Role of oxidative stress in the pathophysiology of bipolar disorder. Neurochemical Research 35, 1295–1301. Su, K.P., Leu, S.J., Yang, Y.Y., Shen, W.W., Chou, Y.M., Tsai, S.Y., 2002. Reduced production of interferon-gamma but not interleukin-10 in bipolar mania and subsequent remission. Journal of Affective Disorders 71, 205–209. Taylor, V., 2011. Alterations in cytokine and adipokine levels over two years of follow up in first episode patients with bipolar disorder. Bipolar Disorders 13, 98–99. Taylor, V.H., MacQueen, G., Maconald, K., McKinnion, M., 2009. Alterations in cytoKines and adiponectine levels over two years of follow up in patients with mood disorders. Biological Psychiatry 1, 67S. Teixeira, A.L., Barbosa, I.G., Huguet, R.B., Neves, F.S., Abreu, M.N., Bauer, M.E., 2011. Investigation of immunologic parameters and neurotrophic factors in bipolar disorder: Trait or state? Bipolar Disorders 13, 99. Tsai, S.Y., Chen, K.P., Yang, Y.Y., Chen, C.C., Lee, J.C., Singh, V.K., Leu, S.J., 1999. Activation of indices of cell-mediated immunity in bipolar mania. Biological Psychiatry 45, 989–994. Tsai, S.Y., Chung, K.H., Huang, S.H., 2010. No state-dependent alternation of immunological parameters, insulin, and leptin during bipolar depression. European Neuropsychopharmacology 20, S419. Tsai, S.Y., Chung, K.H., Wu, J.Y., Kuo, C.J., Lee, H.C., Huang, S.H., 2012. Inflammatory markers and their relationships with leptin and insulin from acute mania to full remission in bipolar disorder. Journal of Affective Disorders 136, 110–116. Tsai, S.Y., Yang, Y.Y., Kuo, C.J., Chen, C.C., Leu, S.J., 2001. Effects of symptomatic severity on elevation of plasma soluble interleukin-2 receptor in bipolar mania. Journal of Affective Disorders 64, 185–193. Vuksan-Cusa, B., Sagud, M., Jakovljevic, M., 2010. C-reactive protein and metabolic syndrome in patients with bipolar disorder compared to patients with schizophrenia. Psychiatria Danubina 22, 275–277. Wadee, A.A., Kuschke, R.H., Wood, L.A., Berk, M., Ichim, L., Maes, M., 2002. Serological observations in patients suffering from acute manic episodes. Human Psychopharmacology 17, 175–179. Zell, R., Geck, P., Werdan, K., Boekstegers, P., 1997. TNF-alpha and IL-1 alpha inhibit both pyruvate dehydrogenase activity and mitochondrial function in cardiomyocytes: evidence for primary impairment of mitochondrial function. Molecular and Cellular Biochemistry 177, 61–67.