CRP, IL-6 and depression: A systematic review and meta-analysis of longitudinal studies

Journal of Affective Disorders 150 (2013) 736–744

Contents lists available at ScienceDirect

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

Review

CRP, IL-6 and depression: A systematic review and meta-analysis of longitudinal studies Vyara Valkanova, Klaus P. Ebmeier, Charlotte L. Allan n Department of Psychiatry, University of Oxford, Warneford Hospital, Oxford OX3 7JX, United Kingdom

art ic l e i nf o

a b s t r a c t

Article history: Received 29 January 2013 Received in revised form 4 June 2013 Accepted 5 June 2013 Available online 17 July 2013

Background: Inflammatory markers are raised in cross-sectional studies of depressed patients and may represent an important mediating factor for behaviour, neural plasticity and brain structure. Methods: We undertook a systematic review of longitudinal studies, investigating whether raised inflammatory markers indicate an increased risk of subsequent depressive symptoms. We searched three databases (1970–2012) for longitudinal studies with repeat data on CRP or IL-6 levels and subsequent depressive symptoms. We calculated effect sizes using a mixed-effects model, with separate meta-analyses for inflammatory markers and age groups. Results: We identified eight papers for CRP (14,832 participants) and three for IL-6 (3695 participants). There was a significant association between increased CRP and depressive symptoms (weighted-mean effect size ‘unadjusted r’ ¼0.069, p o0.0005; ‘adjusted r’¼ 0.046, po 0.0005), with moderate heterogeneity between studies (Q ¼11.21, p¼ 0.08, I2 ¼46.5). For IL-6 the weighted-mean effect size was smaller (‘unadjusted r’¼ 0.045, p-value ¼0.007; ‘adjusted r’ ¼0.097, p-value ¼0.06). Limitations: The meta-analysis was based on a relatively small number of studies (particularly for IL-6) and only two inflammatory markers. There was moderate heterogeneity between studies and some evidence of publication bias. Conclusions: Raised inflammatory markers have a small but significant association with the subsequent development of depressive symptoms. This is a robust effect which remains significant after adjustment for age and a wide range of factors associated with risk for depression. Our results support the hypothesis that there is a causal pathway from inflammation to depression. & 2013 Elsevier B.V. All rights reserved.

Keywords: Inflammation CRP IL-6 Depressive Longitudinal Meta-analysis

Contents 1. 2.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Search criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Inclusion criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Exclusion criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. Data extraction and analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Systematic review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Meta-analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Role of funding source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conflict of interest. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

n

Corresponding author. Tel.: +44 1865 223639; fax: +44 1865 793101. E-mail address: [email protected] (C.L. Allan).

0165-0327/$ - see front matter & 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jad.2013.06.004

737 738 738 738 738 738 738 738 739 739 743 743 743 743

V. Valkanova et al. / Journal of Affective Disorders 150 (2013) 736–744

1. Introduction Depression is a common, heterogeneous condition with a multi-faceted aetiology. Greater understanding of the role of biological factors, such as inflammation and vascular risk, is recognised as increasingly important to enhance our understanding of the aetiology and pathophysiology of depression, with the ultimate aims of early intervention and personalised therapy. Evidence for the role of raised inflammatory markers has been building over the last 20 years and many inflammation measures have been identified as potentially significant, including cytokines (e.g. interleukin-6, IL-6, interleukin-1, IL-1 and tumour necrosis factor, TNF-α), chemokines (e.g. monocyte chemotactic protein-1, MCP-1) and acute phase reactant proteins (e.g. C-reactive protein, CRP) (Krishnadas and Cavanagh, 2012; Lanquillon et al., 2000; Maes et al., 1990, 1991, 1993; Sluzewska et al., 1994, 1996). The exact mechanisms linking inflammation to depression are still unclear, but include effects of psychological stressors, sensitisation of cells to neurotoxic peptides, oxidative and nitrosative stress, autoimmune response to oxidatively and nitrosatively damaged molecules, lowered omega-3 and antioxidants (Maes et al., 2011a; Moylan et al., 2013; Riemer et al., 2010). For instance, chronic inflammation and cell-mediated immune cytokines can activate the enzyme degrading tryptophan (indoleamine 2,3 dioxygenase) leading to serotonin depletion and impaired antioxidant defences. In parallel, cell-mediated immune cytokines increase the synthesis of neurotoxic tryptophan catabolytes (TRYCATs) which lead to oxidative stress, impair the mitochondrial metabolism and trigger apoptosis (Maes et al., 2011c). TRYCATs also lead to excitotoxicity by exerting NMDA receptor agonistic effects, inhibiting the uptake of glutamate and impairing the expression of excitatory amino-acid transporter (McNally et al., 2008; Moylan et al., 2013). The increased production of pro-inflammatory cytokines has also been associated with glucocorticoid resistance and hyperactivity of the HPA-axis (Danese et al., 2008; Maes et al., 1993). There is increasing evidence that inflammation plays a role in the chronic course of depression, increasing vulnerability for further episodes and precipitating progressive illness course. Previous depressive episodes increase the likelihood of new depressive episodes by amplifying the responses of depressogenic cytokines, contributing to the staging and recurrence of depression and inducing neuroprogressive processes (Moylan et al., 2013). Proinflammatory cytokines are important in normal synaptic and neural plasticity, however, increased levels may have an adverse effect on cellular processes leading to neuronal loss, neuronal atrophy and altered gene expression, subsequently affecting signal transduction and behavioural systems (Khairova et al., 2009). Several lines of evidence support a link between inflammation and depression. The phenomenological similarities between sickness behaviour and clinical depression including fatigue, reduced activity, social withdrawal, anorexia and altered sleep pattern suggest that although both conditions are completely different they may share a common pathophysiological pathway i.e. increased proinflammtory cytokines (Dantzer, 2004; Maes et al., 1990, 1993). High rates of depression are found in systemic diseases such as cancer, cardiovascular, metabolic and neurodegenerative diseases, all of which are associated with activation of (neuro) inflammatory, oxidative and nitrosative stress pathways (Dantzer et al., 2008; Maes et al., 2011b; Yirmiya, 1997). In addition many inflammatory conditions such as the postnatal period, haemodialysis, and IFN-α based immunotherapy may trigger clinical depression in patients who are vulnerable due to different factors including decreased levels of omega-3 and dypeptidil peptidase IV or presence of specific single nucleotide polymorphisms in oxidative and nitrosative genes (Moylan et al., 2013). Activation of these (neuro) inflammatory pathways may explain not only the above co-morbidities of clinical depression,

737

but also the association between psychological risk factors and depression. For instance, well-known precipitants of depression, such as negative emotions and stress, increase the production of pro-inflammatory cytokines (Maes et al., 1998; Kiecolt-Glaser et al., 2003; Steptoe et al., 2007). Inflammatory markers are increased in clinical depression even in the absence of major systemic (autoimmune or infectious) disease. Although this increase is modest compared to the one observed in acute inflammatory conditions such as rheumatoid arthritis, low grade inflammation over a prolonged period of time can have significant effects (Raison and Miller, 2011). Over a life-time, subtle elevations in inflammatory markers may influence neural plasticity and subsequent risk of depression (Khairova et al., 2009). Prior depressive episodes sensitise the immune system leading to enhanced inflammatory response and possibly increased vulnerability to further depressive episodes. For instance women with a lifetime history of depression showed increased IL-6 and sIL-1 in the early puerperum, compared to controls (Maes et al., 2001), while in a community sample of older adults, those with more depressive symptoms had a greater IL-6 increase after an annual influenza vaccination compared to those with fewer symptoms suggesting that even modest elevations in depressive symptoms may sensitise the inflammatory response (Glaser et al., 2003). In cross-sectional studies depressed adults show raised inflammatory markers, including CRP, TNF-α, IL-1, IL-6 and IL-2R, when compared with controls (Dowlati et al., 2010; Howren et al., 2009; Liu et al., 2012). This effect is relevant across the age span, with raised levels of IL-1β found in those over 60 years old with major depression (Thomas et al., 2005). Furthermore, consistent with the theory that inflammatory cytokines contribute to depressive symptoms, successful anti-depressant treatment for major depressive disorder reduces serum cytokine levels of IL-1β, and possibly IL-6, with SSRIs having a specific effect on lowering levels of IL-6 and TNF-α (Hannestad et al., 2011). Treatment with the antiinflammatory drug infliximab resulted in a significant decrease in depressive symptoms in patients with treatment resistant depression and increased CRP, but not in the depressed patients without increased inflammatory markers (Raison et al., 2012). In addition, immunotherapy with interferon-α (Bonaccorso et al., 2002) increases depressive symptoms, while withholding of cytokine treatment or adding antidepressants improve symptoms (Capuron et al., 2002). Further support for a role of inflammation in the pathogenesis of depression is provided by neuroimaging. PET studies demonstrated increased metabolic activation of the subgenual anterior cingulate cortex (sgACC) during episodes of major depression compared with remission in the same subjects (Drevets et al., 2008), while fMRI studies showed enhanced activation in the same area in response to treatment with IFN-α (Capuron et al., 2005) and as a result of vaccine-induced increase in the levels of IL-6 (Harrison et al., 2009). Investigating whether there is a longitudinal relationship between raised inflammatory markers and depression is a more complex question, which requires longitudinal studies, preferably with large numbers of participants. Longitudinal studies are vulnerable to confounding factors, and present difficulties in establishing causality, yet given the growing evidence for the role of pro-inflammatory cytokines in depression we sought to investigate whether a cumulative effect of raised inflammatory markers (as markers of chronic lowgrade inflammation) is associated with depressive symptoms. Considering that three recent meta-analyses found an association between inflammation and depression in cross-sectional studies (Dowlati et al., 2010; Howren et al., 2009; Liu et al., 2012) we hypothesised that raised inflammatory markers also indicate an increased risk of subsequent depressive symptoms. Since ageing has been associated with a pro-inflammatory state (Alexopoulos

738

V. Valkanova et al. / Journal of Affective Disorders 150 (2013) 736–744

and Morimoto, 2011) characterised by increased levels of proinflammatory cytokines (Krabbe et al., 2004), increased numbers of activated microglia and decrease in anti-inflammatory molecules (Sparkman and Johnson, 2008), the inflammation hypothesis of depression may be particularly relevant to late-life depression. We therefore considered the effect of age on the relationship between inflammation and depression, hypothesising that older participants would show a stronger relationship between raised inflammation and depression.

2. Methods 2.1. Search criteria We systematically searched Embase, Medline and PsychINFO from 1970–August 2012 for the terms [depression OR depressive] AND [Inflammatory OR interleukin OR IL-1 OR IL-6 OR tumour necrosis factor OR TNF OR C-Reactive OR CRP OR cytokine] AND [longitudinal]. We reviewed titles and abstracts to select papers which were potentially relevant. Following this screening process, we reviewed the full text papers. If there was doubt about the suitability of the paper based on the abstract alone, the full text was reviewed. We hand-searched reference lists of papers finally selected for inclusion to check for any additional studies.

2.2. Inclusion criteria We included English-language papers that used a cohort methodology, as part of a longitudinal study. All included papers measured depression using a standardised score or diagnostic criteria. In addition, papers were required to report measures of inflammatory markers at baseline and depression scores at followup. Through our systematic search strategy we aimed to include multiple inflammatory markers, however, after review of full-text articles, we decided to focus on CRP and IL-6 due to the limited number of studies considering other markers. The search revealed only one study on IL-1receptor antagonist (Milaneschi et al., 2009), one on soluble TNF-alfa receptor 1 (Matheny et al., 2011) and one on fibrinogen (Von Kanel et al., 2009); there were no relevant papers on other markers e.g. IL-1. Where there were insufficient data for meta-analysis, we contacted the authors directly to obtain this information.

2.4. Data extraction and analysis Data were extracted by two independent raters (VV and CA), with disagreements settled by consensus and discussion. Information was extracted in a systematic fashion as follows: (a) study characteristics; (b) study design and length of follow-up; (c) sample source; (d) sample characteristics; (e) inclusion and exclusion criteria; (f) inflammatory markers (type, measures); (g) measures of depression; (h) analysis (statistical models, measures of effect size); and (i) confounding factors. Meta-analysis was undertaken using Comprehensive MetaAnalysis Software version 2.2 (Englewood, 2006). We computed the correlation coefficient ‘r’ as a measure of effect size calculated from the sample size and p-value associated with the t-test. When this information was not available we obtained ‘r’ from the F-statistic, or mean difference and standard deviation. For one study we asked the authors for additional information and they computed the effect size for us (Hamer et al., 2009). Most studies reported unadjusted and adjusted versions for a number of different variable values. To make these more comparable, we split them into two groups: ‘unadjusted r’ (which included unadjusted ‘r’ or ‘r’ adjusted only for socio-demographic variables, when both were reported we used the value adjusted for socio-demographic variables) and ‘adjusted r’ (when more than socio-demographic variables were considered e.g. smoking, alcohol consumption, body-mass index, cholesterol level, physical activity, medication use or chronic illness). We undertook two meta-analyses, first to compare separate inflammatory markers (CRP and IL-6), and second to compare age groups (under or over 50 years). The latter was for studies of CRP only, as all studies on IL-6 used participants over the age of 50. We present data relating to mixed-effects models as opposed to a fixed-effect model. These provide a more conservative estimate of effect size, which is necessary given the different methodologies employed by these studies. To assess heterogeneity across studies we used the Cochrane Q statistic with p o0.10 considered to represent statistically significant heterogeneity and the I2 statistic with 25%, 50% and 75% considered to indicate low, medium and high heterogeneity respectively. We used Begg and Mazumdar's rank correlation and a funnel plot of standard error and Fisher's Z to assess publication bias. In addition, we used Egger's regression intercept and Duval and Tweedie's trim and fill as further tests of bias.

3. Results 3.1. Systematic review

2.3. Exclusion criteria We excluded papers that involved treatment trials, lacked sufficient information to calculate outcome measures or dual publications (if the same sample was used in more than one publication, the study that provided stronger evidence was considered for analysis). We excluded reviews, conference abstracts, editorials and letters, as well as animal studies. We excluded studies that focussed on specific populations (e.g. women postpartum, patients after gastric bypass surgery) in order to reduce the effects of confounding variables in the analysis (there were not enough studies in each specific patient group to do subgroup analysis). We excluded patients with chronic inflammatory conditions (e.g. rheumatoid arthritis) and studies involving stimulated cytokine production because we were interested in the effects of chronic low-grade inflammation, as opposed to acute inflammatory response. At the final stage we used the STROBE statement (Strengthening the reporting of observational studies in epidemiology) to assess study quality (von Elm et al., 2007). No exclusions were made because of poor quality.

The search identified 332 papers, which reduced to 210 papers once duplicate publications were removed. Following a review of titles and abstracts, 178 papers were excluded (89, unrelated; 40, conference abstracts; 19, review papers; 14, treatment trials; eight, not longitudinal; five, letter/editorial/case reports; three animal studies) leaving 32 papers for full-text review (Fig. 1). Following full text review, 24 papers were excluded (eleven for lack of relevant data, e.g. inflammatory markers were not measured at baseline or depression was not assessed as an outcome; four for cross-sectional analysis of data from a longitudinal study; three for dual publications; three for chronic inflammatory conditions; two for specific patient populations; one for stimulated cytokine production), leaving two papers, which investigated both CRP and IL-6 (Gimeno et al., 2009; Stewart et al., 2009), five papers which investigated CRP only (Copeland et al., 2012; Deverts et al., 2010; Hamer et al., 2009; Matthews et al., 2010; Pasco et al., 2010), and one paper which considered IL-6 only (Matheny et al., 2011). One additional paper was identified through a hand-search of reference lists (van den Biggelaar

V. Valkanova et al. / Journal of Affective Disorders 150 (2013) 736–744

739

Fig. 1. Search strategy.

et al., 2007). The median duration of follow-up was five years. The primary purpose of the studies included in this meta-analysis was to investigate the directionality of the relationship between inflammatory markers and depression (Copeland et al., 2012; Deverts et al., 2010; Gimeno et al., 2009; Matthews et al., 2010; Stewart et al., 2009), to determine whether subclinical systemic inflammation is associated with an increased risk of depression (Matheny et al., 2011; Pasco et al., 2010; van den Biggelaar et al., 2007) or to examine whether inflammatory markers play a mediating role in the association between physical activity and depression (Hamer et al., 2009). Baseline depressive symptoms were accounted for by adjusting for them in the analysis (Copeland et al., 2012; Deverts et al., 2010; Gimeno et al., 2009; Matthews et al., 2010; Stewart et al., 2009), excluding participants with depressive symptoms (Hamer et al., 2009; van den Biggelaar et al., 2007) or a lifetime history of depressive disorder according to the Structured Clinical Interview for DSM-IV (SCID)(Pasco et al., 2010). The baseline characteristics of all included studies are illustrated in Table 1. 3.2. Meta-analysis Separate analyses for specific inflammatory markers were performed using unadjusted and adjusted data for CRP and IL-6. Information regarding CRP was available from eight studies involving a total of 14,832 participants (Copeland et al., 2012; Deverts et al., 2010; Gimeno et al., 2009; Hamer et al., 2009; Matthews et al., 2010; Pasco et al., 2010; Stewart et al., 2009; van den Biggelaar et al., 2007). The mixed-model weighted-mean effect size for six studies reporting ‘unadjusted r’ showed a small but significant association between baseline increased levels of CRP and depressive symptoms at follow-up (mixed-effects pooled r ¼0.069; p o0.0005) (Copeland et al., 2012; Gimeno et al., 2009; Hamer et al., 2009; Matthews et al., 2010; Stewart et al., 2009; van den Biggelaar et al., 2007). When the seven studies that adjusted for more than socio-demographic variables were considered, the strength of the association decreased slightly but remained statistically significant (mixed-effects pooled r ¼0.046; p o0.0005) (Copeland et al., 2012; Deverts et al., 2010; Gimeno et al., 2009; Matthews et al., 2010; Pasco et al., 2010; Stewart et al., 2009; van den Biggelaar et al., 2007) (Fig. 2). There was a moderate degree of heterogeneity between studies (Q¼ 11.21, p ¼0.08,

I2 ¼ 46.5) suggesting that systematic differences exist between them (Table 2). Publication bias assessed with Begg and Mazumdar rank correlation was non-significant (Kendall's τ (with continuity correction) ¼ 0.48; two-tailed p ¼ 0.13). However the more sensitive Egger's regression intercept showed a significant degree of publication bias (t ¼2.79; df¼5; two-tailed p ¼0.04). Analysis using Duval and Tweedie's procedure trimmed and filled three studies, resulting in a more conservative corrected pooled effect size of r ¼0.030 (95% CI ¼0.003 to 0.058). Information regarding IL-6 was available in three studies, involving a total of 3695 participants (Gimeno et al., 2009; Matheny et al., 2011, Stewart et al., 2009). The pooled mixed-model effect size was 0.045 (p-value¼0.007) for the two studies that report unadjusted values (Gimeno et al., 2009, Stewart et al., 2009), and 0.097 (p-value¼0.06) for the pooled r available from all three adjusted studies (Table 2, Fig. 2).The weighted-mean effect size for an association between IL-6 and depressive symptoms was smaller compared with the effect size for CRP. After considering only adjusted correlations it became statistically non-significant. This suggests a stronger link between CRP and depression, compared with IL-6. Separate analyses for CRP for age groups (subjects under vs. over 50 years) showed a higher correlation where participants were older, but the effect size remained significant in both groups. The weighted-mean mixed-model effect size for participants over the age of 50 was 0.079 (p¼ 0.02) (Gimeno et al., 2009; Matheny et al., 2011; Stewart et al., 2009; van den Biggelaar et al., 2007), compared with 0.030 (p ¼0.003) for participants under the age of 50 (Copeland et al., 2012; Deverts et al., 2010; Gimeno et al., 2009; Matthews et al., 2010; Pasco et al., 2010) (Table 2). However the total between group heterogeneity was not significant (Q¼1.85, p¼ 0.17), suggesting that the effect is relevant in both age groups.

4. Discussion Our results suggest that raised inflammatory markers have a small but significant association with subsequent depressive symptoms. This was a robust effect that remained significant after adjustment for a wide range of factors associated with raised inflammatory markers and increased risk of depression. Our

740

Author and year

Study design Length of FU

Sample type

Participants N

Exclusion criteria

Age

Gender (% male)

Inflammatory marker

Depression measure

Type Measurement N measurements

Copeland et al. (2012) (29)

cohort

1–3 years

community 1334

mean age[SD]: 14.2 [3.19]

51.3

CRP410 mg/l

CRP

Deverts et al. (2010) (30)

cohort

5 years

community 2544

33–45 years

45

CRP410 mg/l

CRP

Gimeno et al. (2009) (27)

cohort

12 years

community CRP: 3339; IL-6: 3298

70 mean age[SD]: men 49.3 [6.1]; women 50.4[6.1]

Hamer et al. (2009) (26)

cohort

2 years

community 4660

mean age[SD]: 63.4 +/-9.7

Matheny et al. (2011) (22) Matthews et al. (2010) (31)

cohort

1 year

community 134

cohort

7 years

community 1781

median: 5 serum equivalents were calculated from blood spot measures plasma 1

Confounders that were controlled

Country

SDV

Other

SCID

age, sex, race,SES

BMI, medication use, substance use, recent physical illness and chronic illness

CES-D

USA BMI, medication use, age, sex, glucose, insulin, HDL,LDL,TG, race, education smoking, alcohol consumption, physical activity, oral contraceptive, number of medical conditions UK alcohol, diet, physical age, sex, ethnicity, activity, smoking, BMI, HDL, Cholesterol, blood pressure, SES chronic diseases(CHD, diabetes, respiratory illness), medication use

CRP410 mg/l; reported cold or ‘flu’ CRP; in the last two weeks IL-6

serum

1

47.4

CRP410 mg/l; depression at baseline

CRP

serum

1

GHQ-30(4 items assessing cognitve symptoms of depression) CES-D

65+ years

0

MMSEo 20; contraindication for exercise;

IL-6

serum

1

GDS-15

42–52 years

0

CRP amenorrhoea; reproductive hormones use within 3 months prior to study; missing data; stroke; heart condition; diabetes, arthritis; osteoporosis; taking inhaler/ steroids; CRP 410 mg/l

not reported

1

CES-D

age, sex, social class, marital status age

BMI, chronic illness, smoking, alcohol, physical activity,

USA

UK

Charlson Comorbidity Index, USA intervention group(exercise/ no exercise), MMSE, BMI

age, race, BMI, medication use, selfeducation reported health history, hormone therapy, menopausal status

USA

V. Valkanova et al. / Journal of Affective Disorders 150 (2013) 736–744

Table 1 Baseline characteristics of studies.

FU—follow up; CRP—C-reactive protein; IL-6—Interleukin-6; BMI—body-mass index; SDV—socio demographic variables; SES—socioeconomic status; CHD—Coronary Heart Disease; HDL—high density lipoprotein; TG—triglycerides; MMSE—mini-mental state examination; SCID—Structured Clinical Interview for DSM-IV; GHQ—General Health Questionnaire; CES-D—Center for Epidemiologic studies depression scale; GDS—Geriatric Depression Scale; BDI—Beck Depression Inventory.

GDS-15 serum 5 years van den Biggelaar et al. (2007) (33)

cohort;

community 267

85+years 37

MMSEo 24; GDS≥3; use of corticosteroids

CRP

median: 3

BDI-II 1 serum CRP; IL-6 history of chronic disease; CRP410 mg/l 48.3 50–70 years 6 years cohort

community 263

10 years

Pasco et al. (2010) (32) Stewart et al. (2009) (28)

retrospective cohort

community 644

20–84 years

0

depression at baseline; inability to provide informed consent

CRP

serum

1

SCID

age, SES

AU alcohol, smoking, activity level, BMI, chronic diseases (pernicious anaemia, cancer, SLE,RA),medication use USA alcohol, smoking, physical age, sex, activity level, history of race, education chronic diseases, HDL, TG, glucose, BMI, Beck Anxiety Inventory, Cook-Medley Hostility Scale NL sex, smoking, stroke, chronic education diseases, BMI, MMSE score, disability in daily functioning, albumin

V. Valkanova et al. / Journal of Affective Disorders 150 (2013) 736–744

741

results suggest that raised inflammatory markers precede the development of depressive symptoms and provide support for a pathway from inflammation to depression. However, we cannot be certain whether raised CRP and IL-6 represent an association, a mediating risk factor, or a causal factor for depression. The small size of the effect between baseline levels of inflammation and depressive symptoms may simply indicate that the contribution of inflammation to depression is small. Alternatively, the association may be attenuated because depression is a heterogeneous condition and inflammation is aetiologically related to some, but not to all cases of depression. The extent to which different factors, such as inflammation, contribute to the symptoms of depression is likely to vary between patients and a challenge for future research is to reliably stratify patients in subgroups with the aim of an aetiologically specific treatment. A further explanation for a small effect size may be the duration of follow up. If the effect of chronic inflammation on the brain increases over time our median follow up of only five years may result in an attenuated effect size (likewise, if the effect decreases over time, this length of follow-up may also lead to an attenuated effect). It is also possible that the observed relationship between inflammation and depression was attenuated because CRP and IL6 are not the most adequate biomarkers for depression. They not only have a short half-life, making their detection more difficult, but both are specific inflammatory markers. In the light of the evidence for the role of cell-mediated immune activation in depression better markers might be neopterin, sCD8 and tryptophan, all of which measure the cell-mediated immune response (Maes, 2011). Further, as inflammatory responses are accompanied by induction of oxidative and nitrosative stress pathways, the concentration of acute phase proteins such as haptoglobin, albumin and ceruloplasmin which function as antioxidants, together with autoimmune markers of oxidative and nitrosative responses may be more relevant than the inflammatory markers measured by this meta-analysis (Maes et al., 2011a). Our results represent a conservative estimate of the effect size due to the type of samples included and the method of measuring inflammatory markers. All included studies were of communitybased samples, however, a recent cross-sectional meta-analysis showed substantially larger associations in clinical samples (Howren et al., 2009). This effect may be further emphasised in longitudinal clinical samples. Our meta-analysis focussed on chronic low-grade inflammation, and therefore excluded studies of patients with chronic diseases (e.g. cancer, heart disease) and inflammatory conditions (e.g. rheumatoid arthritis). As a result the confounding impact of chronic illnesses, which are already associated with higher rates of depression, was minimised. The studies included in this meta-analysis used a single measurement of inflammatory markers, and when repeated measurements were available they were considered as separate observations in the analysis. However, a single measurement of inflammatory markers is not only more prone to a measurement error, but in addition, does not account for the fluctuations in the levels of inflammatory markers and therefore may not adequately reflect chronic inflammatory processes. This is illustrated in a recent study which found that the longitudinal association between IL-6 and common mental disorder was four times stronger when the serum IL-6 level was determined based on the average of two examinations at different time points (Kivimäki et al., 2013). We demonstrated a stronger relationship for elevated CRP and subsequent depressive symptoms, compared with IL-6. Although this may be due to the larger number of studies or other methodological issues, it remains possible that CRP is one of the key drivers in the association between inflammatory processes and depressive symptoms. CRP is strongly associated with different interleukins, including IL-6, perhaps providing a measure of the

742

V. Valkanova et al. / Journal of Affective Disorders 150 (2013) 736–744

Fig. 2. Mixed-effects model for inflammatory markers and depressive symptoms (adjusted values).

Table 2 Weighted-mean effect sizes and measures of heterogeneity. Inflammatory marker

Subgroups

CRP CRP* old young IL-6 IL-6*

N studies

6 7 3 4 2 3

Correlation (weighted mean mixed effects model)

0.069 0.046 0.079 0.030 0.045 0.097

95% CI lower limit

0.036 0.021 0.012 0.010 0.013 -0.005

95% CI upper limit

0.103 0.070 0.144 0.050 0.078 0.198

p-value

o 0.0005 o 0.0005 0.02 0.003 0.007 0.06

Heterogeneity Q-value

p-value

I-squared

19.23 11.22 5.70 3.19 0.54 5.2

0.002 0.08 0.06 0.36 0.46 0.08

74.00 46.50 64.89 6.09 0.00 61.37

CRP—C-Reactive protein; IL—6-Interleukin-6; * fully adjusted (e.g. body-mass index, smoking, use of medication, chronic illness etc.).

cumulative effect of a range of inflammatory markers. Our finding raises questions about the relationship between different inflammatory markers and depression. Although there are different hypotheses about the effects of cytokines on the brain, the exact mechanisms have not been identified. While IL-6 has direct effects on the brain and CRP is a more general marker of inflammatory processes (Dantzer and Kelley, 2007), there are a number of indirect mechanisms through which cytokines can signal to the brain and stimulate microglia, leading to neuro-inflammation (Capuron and Miller, 2011). The nature of the association between peripheral and brain levels of cytokines, and mediating factors on this association requires further clarification. There may also be other inflammatory markers which are relevant to the development of depressive symptoms, and though small numbers of such studies precluded inclusion in this paper, they warrant further research. The effect of raised inflammatory markers on depression was present regardless of age. Although our results could be explained by methodological issues such as the relatively small number of studies in each age group, they are consistent with a meta-analysis of crosssectional studies that includes a large number of studies (Howren et al., 2009). There is evidence suggesting that a stronger association between inflammation and depression might be expected in older adults. With ageing, the peripheral immune activation produces an exaggerated central inflammatory response (Dilger and Johnson, 2008) and the brain is in a chronic state of neuro-inflammation (Sparkman and Johnson, 2008). Conditions such as ischaemic stroke and cardiovascular diseases are associated with elevated levels of circulating cytokines, have increased prevalence with age and are comorbid with depression in later life (Apostolakis et al., 2008; Pascoe et al., 2011). As yet, it is not clear whether this effect of inflammation is specific to older age groups. Certain limitations of meta-analytic methodology should be remembered when interpreting these results. Our findings are based on a relatively small number of studies (particularly for IL-6) and only

two inflammatory markers. There was moderate heterogeneity between studies and some evidence of publication bias which led to a smaller effect size when Duval and Tweedie's procedure was used. A particular difficulty of longitudinal studies is the issue of confounding variables, and there are many different aspects (biological, psychological and social) that may contribute to the development of depression and of raised inflammatory markers, which are impossible to control for. Though adjusted variables have taken this into account and provided more conservative effect sizes, the range of considered confounding factors and the quality of their assessment vary between studies and the effects are unlikely to have been eliminated. In these studies depressive symptoms were measured using a range of standardised measures including the Structured Clinical Interview for DSM-IV (SCID), General Health Questionnaire (GHQ), Center for Epidemiologic studies depression scale (CES-D), Geriatric depression scale (GDS) and Beck depression Inventory (BDI). Although these are well-validated measures, commonly used in longitudinal studies, only the SCID provides a clinical diagnosis. High scores on the other measures are suggestive of depressive symptoms, but do not necessarily equate with clinical depression. Despite these limitations, the finding that there is an association between longitudinal inflammatory markers and depressive symptoms is interesting and raises several questions for future research, not least the issue of how raised inflammatory markers lead to depressive symptoms and whether this is a causal relationship. Use of neuroimaging in longitudinal samples could be one way of investigating whether there is a correlation between persistently raised inflammatory measures and changes in brain structure and function, which may increase susceptibility to depression. It is possible that raised inflammatory markers are mediating factors which help understand and explain the vascular depression hypothesis, or that investigation of hippocampal regions can elucidate whether raised inflammatory markers are factors mediating between depression and dementia (e.g. via raised amyloid). Neuroimaging could also increase

V. Valkanova et al. / Journal of Affective Disorders 150 (2013) 736–744

our understanding of the relationship between peripheral and central immune activation, as in vivo visualisation of activated microglia is now possible with PET and SPECT (Lautner et al., 2011). Finally, there are many questions to be asked about the nature of inflammatory markers involved (is this a general effect, specific to CRP and IL-6 or relevant in other inflammatory markers?), the position of inflammation in the sequence of events leading to depression, whether length of exposure to inflammatory markers is important, and whether lowgrade inflammation is as important as high-grade inflammation (e.g. related to chronic disease). Investigation of these questions will help to clarify the biological mechanisms between stress, chronic illness and depression. Role of funding source Grant and other support to KPE from the Medical Research Council (UK), Gordon Edward Small Charitable Trust, Norman Collisson Foundation, HDH Wills 1965 Charitable Trust, and National Institute for Health Research (England). CLA is a Clinical Research Fellow funded by the OHSRC/BRC/NOF/OUCAGS.

Conflict of interest VV and CLA have no possible conflict of interest, financial or otherwise, related directly or indirectly to this work. KPE has received grant and other support from the Medical Research Council (UK), Gordon Edward Small Charitable Trust, Norman Collisson Foundation, HDH Wills 1965 Charitable Trust, and National Institute for Health Research (England). KPE has received educational expenses grants to Department of Psychiatry for NHS CPD from Lundbeck & Jansen-Cilag.

Acknowledgements We thank Dr. Mark Hamer, UCL, for providing additional data.

References Alexopoulos, G.S., Morimoto, S.S., 2011 Nov. The inflammation hypothesis in geriatric depression. International Journal of Geriatric Psychiatry, http://dx. doi.org/10.1002/gps.2672, in press. Apostolakis, S., Vogiatzi, K., Krambovitis, E., Spandidos, D.A., 2008. IL-1 cytokines in cardiovascular disease: diagnostic, prognostic and therapeutic implications. Cardiovascular and Hematological Agents in Medicinal Chemistry 6, 150–158. Bonaccorso, S., Marino, V., Biondi, M., Grimaldi, F., Ippoliti, F., Maes, M., 2002. Depression induced by treatment with interferon-alpha in patients affected by hepatitis C virus. Journal of Affective Disorders 72, 237–241. Capuron, L., Gumnick, J.F., Musselman, D.L., Lawson, D.H., Reemsnyder, A., Nemeroff, C.B., Miller, A.H., 2002. Neurobehavioral effects of interferon-alpha in cancer patients: phenomenology and paroxetine responsiveness of symptom dimensions. Neuropsychopharmacology 26, 643–652. Capuron, L., Miller, A.H., 2011. Immune system to brain signaling: neuropsychopharmacological implications. Pharmacology and Therapeutics 130, 226–238. Capuron, L., Pagnoni, G., Demetrashvili, M., Woolwine, B.J., Nemeroff, C.B., Berns, G. S., Miller, A.H., 2005. Anterior cingulate activation and error processing during interferon-alpha treatment. Biological Psychiatry 58, 190–196. Copeland, W.E., Shanahan, L., Worthman, C., Angold, A., Costello, E.J., 2012. Cumulative depression episodes predict later C-reactive protein levels: a prospective analysis. Biological Psychiatry 71, 15–21. Danese, A., Moffitt, T.E., Pariante, C.M., Ambler, A., Poulton, R., Caspi, A., 2008. Elevated inflammation levels in depressed adults with a history of childhood maltreatment. Archives of General Psychiatry 65, 409–416. Dantzer, R., 2004. Cytokine-induced sickness behaviour: a neuroimmune response to activation of innate immunity. European Journal of Pharmacology 500, 399–411. Dantzer, R., Kelley, K.W., 2007. Twenty years of research on cytokine-induced sickness behavior. Brain, Behavior, and Immunity 21, 153–160. Dantzer, R., O’Connor, J.C., Freund, G.G., Johnson, R.W., Kelley, K.W., 2008. From inflammation to sickness and depression: when the immune system subjugates the brain. Nature Reviews Neuroscience 9, 46–56. Deverts, D.J., Cohen, S., DiLillo, V.G., Lewis, C.E., Kiefe, C., Whooley, M., Matthews, K. A., 2010. Depressive symptoms, race, and circulating C-reactive protein: the coronary artery risk development in young adults (CARDIA) study. Psychosomatic Medicine 72, 734–741. Dilger, R.N., Johnson, R.W., 2008. Aging, microglial cell priming, and the discordant central inflammatory response to signals from the peripheral immune system. Journal of Leukocyte Biology 84, 932–939. Dowlati, Y., Herrmann, N., Swardfager, W., Liu, H., Sham, L., Reim, E.K., Lanctot, K.L., 2010. A meta-analysis of cytokines in major depression. Biological Psychiatry 67, 446–457. Drevets, W.C., Savitz, J., Trimble, M., 2008. The subgenual anterior cingulate cortex in mood disorders. CNS Spectrums 13, 663–681. Englewood, N.J., 2006. Comprehensive Meta-Analysis 2.2.048. Biostat.: USA.

743

Gimeno, D., Kivimaki, M., Brunner, E.J., Elovainio, M., De Vogli, R., Steptoe, A., Kumari, M., Lowe, G.D., Rumley, A., Marmot, M.G., Ferrie, J.E., 2009. Associations of C-reactive protein and interleukin-6 with cognitive symptoms of depression: 12-year follow-up of the Whitehall II study. Psychological Medicine 39, 413–423. Glaser, R., Robles, T.F., Sheridan, J., Malarkey, W.B., Kiecolt-Glaser, J.K., 2003. Mild depressive symptoms are associated with amplified and prolonged inflammatory responses after influenza virus vaccination in older adults. Archives of General Psychiatry 60, 1009–1014. Hamer, M., Molloy, G.J., de Oliveira, C., Demakakos, P., 2009. Persistent depressive symptomatology and inflammation: to what extent do health behaviours and weight control mediate this relationship? Brain, Behavior, and Immunity 23, 413–418. Hannestad, J., DellaGioia, N., Bloch, M., 2011. The effect of antidepressant medication treatment on serum levels of inflammatory cytokines: a meta-analysis. Neuropsychopharmacology 36, 2452–2459. Harrison, N.A., Brydon, L., Walker, C., Gray, M.A., Steptoe, A., Critchley, H.D., 2009. Inflammation causes mood changes through alterations in subgenual cingulate activity and mesolimbic connectivity. Biological Psychiatry 66, 407–414. Howren, M.B., Lamkin, D.M., Suls, J., 2009. Associations of depression with C-reactive protein, IL-1, and IL-6: a meta-analysis. Psychosomatic Medicine 71, 171–186. Khairova, R.A., Machado-Vieira, R., Du, J., Manji, H.K., 2009. A potential role for proinflammatory cytokines in regulating synaptic plasticity in major depressive disorder. International Journal of Neuropsychopharmacology 12, 561–578. Kiecolt-Glaser, J.K., Preacher, K.J., MacCallum, R.C., Atkinson, C., Malarkey, W.B., Glaser, R., 2003. Chronic stress and age-related increases in the proinflammatory cytokine IL-6. Proceedings of the National Academy of Sciences USA 100, 9090–9095. Kivimäki, M., Shipley, M.J., Batty, G.D., Hamer, M., Akbaraly, T.N., Kumari, M., Jokela, M., Virtanen, M., Lowe, G.D., Ebmeier, K.P., Brunner, E.J., Singh-Manoux, A., 2013 Apr 9. Long-term inflammation increases risk of common mental disorder: a cohort study. Molecular Psychiatry, http://dx.doi.org/10.1038/mp.2013.35, in press. Krabbe, K.S., Pedersen, M., Bruunsgaard, H., 2004. Inflammatory mediators in the elderly. Experimental Gerontology 39, 687–699. Krishnadas, R., Cavanagh, J., 2012. Depression: an inflammatory illness? Journal of Neurology, Neurosurgery and Psychiatry 83, 495–502. Lanquillon, S., Krieg, J.C., Bening-Abu-Shach, U., Vedder, H., 2000. Cytokine production and treatment response in major depressive disorder. Neuropsychopharmacology 22, 370–379. Lautner, R., Mattsson, N., Scholl, M., Augutis, K., Blennow, K., Olsson, B., Zetterberg, H., 2011. Biomarkers for microglial activation in Alzheimer's disease. International Journal of Alzheimer's Disease 2011, 939426. Liu, Y., Ho, R.C., Mak, A., 2012. Interleukin (IL)-6, tumour necrosis factor alpha (TNFalpha) and soluble interleukin-2 receptors (sIL-2 R) are elevated in patients with major depressive disorder: a meta-analysis and meta-regression. Journal of Affective Disorders 139, 230–239. Maes, M., 2011. Depression is an inflammatory disease, but cell-mediated immune activation is the key component of depression. Progress in NeuroPsychopharmacology and Biological Psychiatry 35, 664–675. Maes, M., Bosmans, E., Suy, E., Vandervorst, C., De Jonckheere, C., Raus, J., 1990. Immune disturbances during major depression: upregulated expression of interleukin-2 receptors. Neuropsychobiology 24, 115–120. Maes, M., Bosmans, E., Suy, E., Vandervorst, C., DeJonckheere, C., Raus, J., 1991. Depression-related disturbances in mitogen-induced lymphocyte responses and interleukin-1 beta and soluble interleukin-2 receptor production. Acta Psychiatrica Scandinavica 84, 379–386. Maes, M., Capuron, L., Ravaud, A., Gualde, N., Bosmans, E., Egyed, B., Dantzer, R., Neveu, P.J., 2001. Lowered serum dipeptidyl peptidase IV activity is associated with depressive symptoms and cytokine production in cancer patients receiving interleukin-2-based immunotherapy. Neuropsychopharmacology 24, 130–140. Maes, M., Galecki, P., Chang, Y.S., Berk, M., 2011a. A review on the oxidative and nitrosative stress (O&NS) pathways in major depression and their possible contribution to the (neuro)degenerative processes in that illness. Progress in Neuro-Psychopharmacology and Biological Psychiatry 35, 676–692. Maes, M., Kubera, M., Obuchowiczwa, E., Goehler, L., Brzeszcz, J., 2011b. Depression's multiple comorbidities explained by (neuro)inflammatory and oxidative & nitrosative stress pathways. Neuroendocrinology Letters 32, 7–24. Maes, M., Leonard, B.E., Myint, A.M., Kubera, M., Verkerk, R., 2011c. The new ‘5-HT’ hypothesis of depression: cell-mediated immune activation induces indoleamine 2,3-dioxygenase, which leads to lower plasma tryptophan and an increased synthesis of detrimental tryptophan catabolites (TRYCATs), both of which contribute to the onset of depression. Progress in NeuroPsychopharmacology and Biological Psychiatry 35, 702–721. Maes, M., Scharpe, S., Meltzer, H.Y., Bosmans, E., Suy, E., Calabrese, J., Cosyns, P., 1993. Relationships between interleukin-6 activity, acute phase proteins, and function of the hypothalamic-pituitary-adrenal axis in severe depression. Psychiatry Research 49, 11–27. Maes, M., Song, C., Lin, A., De Jongh, R., Van Gastel, A., Kenis, G., Bosmans, E., De Meester, I., Benoy, I., Neels, H., Demedts, P., Janca, A., Scharpe, S., Smith, R.S., 1998. The effects of psychological stress on humans: increased production of pro-inflammatory cytokines and a Th1-like response in stress-induced anxiety. Cytokine 10, 313–318. Matheny, M.E., Miller, R.R., Shardell, M.D., Hawkes, W.G., Lenze, E.J., Magaziner, J., Orwig, D.L., 2011. Inflammatory cytokine levels and depressive symptoms in

744

V. Valkanova et al. / Journal of Affective Disorders 150 (2013) 736–744

older women in the year after hip fracture: findings from the Baltimore hip studies. Journal of the American Geriatrics Society 59, 2249–2255. Matthews, K.A., Schott, L.L., Bromberger, J.T., Cyranowski, J.M., Everson-Rose, S.A., Sowers, M., 2010. Are there bi-directional associations between depressive symptoms and C-reactive protein in mid-life women? Brain, Behavior, and Immunity 24, 96–101. McNally, L., Bhagwagar, Z., Hannestad, J., 2008. Inflammation, glutamate, and glia in depression: a literature review. CNS Spectrums 13, 501–510. Milaneschi, Y., Corsi, A.M., Penninx, B.W., Bandinelli, S., Guralnik, J.M., Ferrucci, L., 2009. Interleukin-1 receptor antagonist and incident depressive symptoms over 6 years in older persons: the InCHIANTI study. Biological Psychiatry 65, 973–978. Moylan, S., Maes, M., Wray, N.R., Berk, M., 2013. The neuroprogressive nature of major depressive disorder: pathways to disease evolution and resistance, and therapeutic implications. Molecular Psychiatry 18, 595–606. Pasco, J.A., Nicholson, G.C., Williams, L.J., Jacka, F.N., Henry, M.J., Kotowicz, M.A., Schneider, H.G., Leonard, B.E., Berk, M., 2010. Association of high-sensitivity C-reactive protein with de novo major depression. British Journal of Psychiatry 197, 372–377. Pascoe, M.C., Crewther, S.G., Carey, L.M., Crewther, D.P., 2011. Inflammation and depression: why poststroke depression may be the norm and not the exception. International Journal of Stroke 6, 128–135. Raison, C.L., Miller, A.H., 2011. Is depression an inflammatory disorder? Current Psychiatry Reports 13, 467–475. Raison, C.L., Rutherford, R.E., Woolwine, B.J., Shuo, C., Schettler, P., Drake, D.F., Haroon, E., Miller, A.H., 2012. A randomized controlled trial of the tumor necrosis factor antagonist infliximab for treatment-resistant depression: the role of baseline inflammatory biomarkers. Archives of General Psychiatry, 1–11. Riemer, S., Maes, M., Christophe, A., Rief, W., 2010. Lowered omega-3 PUFAs are related to major depression, but not to somatization syndrome. Journal of Affective Disorders 123, 173–180.

Sluzewska, A., Nowakowska, E., Gryska, K., Mackiewicz, A., 1994. Haptoglobin levels in a chronic mild stress model of depression in rats before and after treatment. European Neuropsychopharmacology 4, 302. Sluzewska, A., Rybakowski, J., Bosmans, E., Sobieska, M., Berghmans, R., Maes, M., Wiktorowicz, K., 1996. Indicators of immune activation in major depression. Psychiatry Research 64, 161–167. Sparkman, N.L., Johnson, R.W., 2008. Neuroinflammation associated with aging sensitizes the brain to the effects of infection or stress. Neuroimmunomodulation 15, 323–330. Steptoe, A., Hamer, M., Chida, Y., 2007. The effects of acute psychological stress on circulating inflammatory factors in humans: a review and meta-analysis. Brain, Behavior, and Immunity 21, 901–912. Stewart, J.C., Rand, K.L., Muldoon, M.F., Kamarck, T.W., 2009. A prospective evaluation of the directionality of the depression-inflammation relationship. Brain, Behavior, and Immunity 23, 936–944. Thomas, A.J., Davis, S., Morris, C., Jackson, E., Harrison, R., O’Brien, J.T., 2005. Increase in interleukin-1beta in late-life depression. American Journal of Psychiatry 162, 175–177. van den Biggelaar, A.H., Gussekloo, J., de Craen, A.J., Frolich, M., Stek, M.L., van der Mast, R.C., Westendorp, R.G., 2007. Inflammation and interleukin-1 signaling network contribute to depressive symptoms but not cognitive decline in old age. Experimental Gerontology 42, 693–701. von Elm, E., Altman, D.G., Egger, M., Pocock, S.J., Gotzsche, P.C., Vandenbroucke, J.P., 2007. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet 370, 1453–1457. Von Kanel, R., Bellingrath, S., Kudielka, B.M., 2009. Association between longitudinal changes in depressive symptoms and plasma fibrinogen levels in school teachers. Psychophysiology 46, 473–480. Yirmiya, R., 1997. Behavioral and psychological effects of immune activation: implications for ‘depression due to a general medical condition’. Current Opinion in Psychiatry 10, 470–476.