Sex differences in the inflammation-depression link: A systematic review and meta -analysis

Major Depressive Disorder (MDD) is a heterogeneous disorder that affects twice as many women than men. Precluding advances in more tailored and efficacious treatments for depression is the lack of reliable biomarkers. While depression is linked to elevations in inflammatory immune system functioning, this relationship is not evident among all individuals with depression and may vary based on symptom subtypes and/or sex. This systematic review and meta-analysis examined whether inflammatory immune peripheral markers of depression are sex-specific. PRISMA guidelines were followed for the systematic review, and a comprehensive search strategy that identified studies from PubMed and PsycInfo was applied. Studies were included if they reported C-reactive protein (CRP), interleukin (IL)-6, tumor necrosis factor (TNF)- α and/or IL-1 β for males and/or females among depressed and healthy adults. We identified 23 studies that satisfied these inclusion criteria. Random-effects meta-analysis models were fit, and measures of association were summarized between levels of circulating markers of inflammation in depressed and healthy males and females. Sex-based analyses revealed elevated levels of CRP among females with depression (Cohen ’ s d = 0.19) relative to their healthy counterparts (p = 0.02), an effect not apparent among males (Cohen ’ s d = -0.01). Similarly, levels of IL-6 were increased among females with depression compared to healthy controls (Cohen ’ s d = 0.51; p = 0.04), but once again this was not found among males (Cohen ’ s d = 0.16). While TNF-α levels were elevated among individuals with depression compared to controls (p = 0.01), no statistically significant sex differences were found. The meta-analysis for IL-1 β resulted in only three articles, and thus, results are presented in the supplemental section. This meta-analysis advances our understanding of the unique involvement of inflammatory biomarkers in depression among men and women, which may help inform more tailored sex-specific treatment approaches in the future.


Introduction
Major Depressive Disorder (MDD) is estimated to affect over 350 million people worldwide (World Health Organization, 2008), and is projected to have the greatest burden of disease by the year 2030 (World Health Organization, 2008; World Federation for Mental Health (WFMH), 2012).Despite its prevalence and impact, effective treatments lag far behind that of many chronic physical disorders due, in part, to a limited understanding of the pathophysiology of depression (McQuaid, 2021).There are several well-established factors that increase the risk of developing depression, including stressful life experiences, such as early-life trauma (Danese et al., 2007;Negele et al., 2015), psychosocial determinants of health (Remes, Mendes and Templeton, 2021), genetic predispositions (Shadrina et al., 2018), as well as sex and gender (Albert, 2015;Zhou et al., 2023).In fact, depression is at least twice as likely to develop among women compared to men (Cyranowski et al., 2000;Piccinelli and Wilkinson, 2000;Salk, Hyde and Abramson, 2017).Thus, sex-specific factors that influence vulnerability to depression require consideration, and it is suggested that biological processes, such as sex steroid hormones (Albert, 2015) and inflammation (Derry et al., 2015;Kropp and Hodes, 2023), are involved.
There is a well-established link between elevated cytokine production and the pathogenesis of depression (Dantzer et al., 2008;Maes, 1999;Miller and Raison, 2016).Specifically, increased circulating pro-inflammatory markers or cytokines can lead to significant behavioral changes that are consistent with depression symptoms, such as anhedonia, social withdrawal, and fatigue (Moreira et al., 2015;Slavich and Irwin, 2014).Several meta-analyses reveal that individuals with depression display increased circulating levels of C-reactive protein (CRP) and cytokines including interleukin (IL)-6, IL-1β, IL-12, and tumor-necrosis factor (TNF)-α compared to healthy controls (Haapakoski et al., 2015;Howren et al., 2009;Osimo et al., 2019;Valkanova et al., 2013).Moreover, individuals with treatment resistant depression display increased inflammatory biomarker profiles compared to those with depression that is not treatment resistant (Haroon et al., 2018;Strawbridge et al., 2019).Although a number of studies find elevated inflammatory factors and cytokines among individuals with depression, there are reports that do not find this relation (Köhler et al., 2018;Köhler-Forsberg et al., 2017;Zhang et al., 2023).
Inconsistencies in the depression-inflammation relation may be due to the heterogeneous nature of depression, such that individuals can present many different symptom combinations to meet diagnostic criteria (Fried and Nesse, 2015).In fact, it may be specific symptoms or subtypes of depression that are marked by elevated inflammatory profiles (Franklyn et al., 2022;Jokela et al., 2016;Majd et al., 2019), though inconsistencies regarding the role of sex in the inflammation-depression link also exist (Derry et al., 2015;Kim et al., 2021;Knight et al., 2020).Sex differences are evident in associations between systemic inflammation ex vivo immune responses to lipopolysaccharide (LPS) and depressive symptoms.Specifically, following LPS stimulation, higher depressive symptoms related to heightened ex vivo inflammatory responses among men, whereas for women higher depressive symptoms related to attenuated pro-inflammatory responses (Knight et al., 2020;Majd et al., 2018).In contrast to these findings, others have suggested that women are more vulnerable to inflammation-induced mood and behavior changes (Derry et al., 2015;Moieni et al., 2015).In clinical populations, IL-6, IL-1β, and CRP were elevated among women with depression, whereas these markers were not elevated in men, and rather men displayed elevated levels of IL-17 (Kim et al., 2021).It is important to better delineate the contribution of sex differences in the relationship between depression and the immune system (Kropp and Hodes, 2023).
Taken together, the overarching goal of the current systematic review and meta-analysis was to synthesize evidence from the published literature with the aim to differentiate immune inflammatory profiles in depression between males and females.Specifically, we examined differences in CRP, IL-6, TNF-α, and IL-1β concentrations, given that these peripheral inflammatory markers are most commonly cited in the literature in relation to depression (Dowlati et al., 2010;Goldsmith et al., 2016;Haapakoski et al., 2015;Howren et al., 2009;Köhler-Forsberg et al., 2017;Osimo et al., 2020;Raison, Capuron & Miller, 2006).We hypothesized that the contrast between elevated inflammatory profiles in individuals with depression and healthy controls would be more evident in females relative to males.Moreover, as a secondary analysis we further expected to confirm that individuals with depression would display significantly elevated levels of CRP, IL-6, TNF-α, and IL-1β compared to healthy controls.

Search strategy and study selection
This systematic review and meta-analysis was conducted and preregistered in Prospero (# CRD42023366432).While several meta-analyses assessing the inflammation-depression relation have been conducted, to our knowledge, none have examined sex differences.PubMed and PsycInfo databases were used to identify relevant articles.The search was conducted adhering to the Population, Exposure, Comparison, Outcome (PECO) framework, using Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines (Morgan et al., 2018;Page et al., 2021).
Inclusion criteria comprised studies that provided basal inflammatory marker concentrations among depression and healthy control groups separately for adult males and females.Search criteria were restricted to studies published after the year 2000, based on a preliminary literature review, in which no studies were found prior to this date that met inclusion criteria, namely articles considering sex differences in the inflammation-depression relation that provided means for males and females separately.Inflammation was assessed by measurement of peripheral blood levels of four common markers of inflammation including CRP, IL-6, TNF-α, and IL-1β.Depression was assessed based on the Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV or − V criteria (American Psychiatric Association, 2013), as well as validated psychometric scales.Details of which psychometric scales were used per study is included in the supplemental material (Supplement A; Table A.1).
Non-human studies were excluded from our review, in addition to studies with a reported mean participant age below 18 or above 65.Studies that did not report the data needed to calculate means and standard deviations for depressed and control groups for males and/or females were not considered, if we could not obtain this information from the authors.Specifically, authors were contacted to retrieve means and standard deviations separately for males and/or females for research articles that provided all other data required but did not separate by sex.Investigations containing individuals presenting with comorbid physical or mental disorders were excluded, as to strictly investigate the depression-inflammatory link without the impact of confounding influences, with the exception of anxiety disorders, given the high rates of comorbidity between depression and anxiety (Kessler et al., 2015).
The search used in this systematic review was created based on the research question objective and clearly defined inclusion and exclusion criteria.Appropriate MeSH terms and Boolean operators were included in the search to detect studies that were relevant.Our search contained all keywords identified by a previously published meta-analysis (Osimo et al., 2020), for details see supplemental materials (Supplement B).The following keywords were added to further identify articles relevant to our research question: (humans OR "human research" OR "human stud*") AND (woman OR women OR female* OR man OR men OR male*) AND (adults OR 18 OR "18 years old" OR "18 years of age").In addition, specific keywords were used in the search criteria to exclude article titles and abstracts that focused on other common biological factors/influences (i.e., gut, microbiome and pregnancy), substances (i.e., cigarette and cannabis smoking, cocaine use and alcohol use), neurological disorders (i.e., schizophrenia and neurodegenerative disorders) and/or illnesses (i.e., cardiovascular disease, diabetes, and flu) that may affect the relationship between inflammation and depression.The search criteria were used in the PubMed and PsycInfo databases according to each database's specific search procedures.All identified papers were then imported into Covidence for title and abstract screening by two researchers (AD and AV), and a third researcher (DJ) resolved conflicting decisions.Full text reviews were conducted of the articles that passed initial screening by two researchers (DJ and AV), and a third researcher (AD) resolved conflicting decisions.The retrieved fulltext articles were examined to identify studies that: (i) reported blood measurements of markers of inflammation; (ii) had both depressed and healthy control groups; (iii) contained means and standard deviations/ standard errors of inflammatory measurements for males and/or females required to calculate the measure of association; and (iv) articles that did not assess comorbid disorders.If articles met all inclusion criteria, except having data separated by males and females, then one researcher (AV) contacted the authors of these studies to request means and standard deviations for both males and/or females separately.The number of references used based on which authors responded and provided data were noted in the PRISMA diagram (Fig. 1).

Assessment of publication bias and heterogeneity
The Newcastle-Ottawa Scale (NOS; Wells et al., 2012) was selected to evaluate the overall quality of the studies considered.The NOS is a ninepoint study assessment tool used to evaluate non-randomized cohort and case-control studies.A point value is assigned to three domains including selection of cohorts or cases and controls (0-4 points), comparability of cohorts or cases and controls (0-2 points), and outcome (0-3 points).A maximum of 9 points may be assigned per study, where studies with ≥7 points were considered as "good", 4-6 points were considered as "fair", and any study with ≤4 points were excluded for being considered of "poor" quality.

Statistical analysis
A meta-analysis was conducted in Stata version 17.0 (StataCorp, 2023) to estimate a summary measure of association (Cohen's d).Cohen's d was calculated using the standardized mean differences to assess differences in inflammation levels between depressed individuals and healthy controls, and to determine mean differences in inflammatory profiles among both males and females.Forest plots were generated to visualize the heterogeneity of the Cohen's d across studies and to provide a summary measure of it.To assess total variance of betweenstudy differences, the measure of heterogeneity that was considered was I 2 , with values greater than 40 % considered moderate, and values greater than 75 % considered to be of high heterogeneity (Borenstein et al., 2009).Summary measures of association were derived using random-effects models for four individual markers of inflammation (CRP, IL-6, TNF-α, and IL-1β), and these four markers were each assessed independently according to sex.To evaluate the observed heterogeneity in the measures of association between studies, we conducted metaregression analyses for study-specific factors of BMI and mean age of participants.Publication bias was assessed through visual inspection of symmetry in funnel plots and testing for statistical significance using Egger's test (Egger et al., 1997;Higgins et al.;2011).All analyses were conducted based on a two-sided alpha level threshold of α = 0.05.

Study selection
From the search criteria, 3,515 papers were retrieved from the PubMed database, and 3911 papers from the PsycInfo database.Of those, 816 duplicates were identified and removed from PubMed, while two duplicates were identified and removed from PsycInfo.In total, 6,610 papers were left for title/abstract screening.Following title/abstract screening, 6,141 papers were excluded, with 8 studies not retrieved, leaving 461 for full-text review.Of the remaining full-text papers, 444 were excluded for failing to meet all the inclusion criteria, 77 of which were excluded as authors who were contacted did not provide data separated by sex (Fig. 1).The final papers identified for analysis included ten papers found through our search strategy (Ho et al., 2017;Huang and Lin, 2007;Hung et al., 2007;Kling et al., 2007;Miller et al., 2005;Pike and Irwin, 2006;Su et al., 2011;Takekawa et al., 2020;Tayefi et al., 2017;Yang et al., 2007), six papers found outside of our search strategy via literature search (Chavda et al., 2011;Kim et al., 2021;Lan et al., 2023;Lu et al., 2019;Renner et al., 2022;Vaccarino et al., 2008), and seven from authors who were emailed and who provided data according to sex, as their original papers included combined measures (Bahrini et al., 2016;de Punder et al., 2018;Eswarappa et al., 2019;Glaus et al., 2018;Podlipný et al., 2010;Sánchez-Carro et al., 2023;Strawbridge et al., 2019).These papers included measurements for at least one (or more) of the four target inflammatory markers (i.e., CRP, IL-6, TNF-α, and IL-1β) for individuals with depression and controls that were reported separately by sex, or only included males or females in the analysis.This allowed for conducting a meta-analysis by sex comparing individual inflammatory or cytokine levels in individuals  with depression and healthy controls.The IL-1β meta-analysis had a limited number of studies that met the inclusion criteria, and as such, the results have been placed in the supplemental material (Supplement C).

Inflammation, depression, and sex differences 3.2.1. C-reactive protein
Females with depression showed statistically significant elevations of CRP compared to controls (95 % CI: 0.04 -0.34, p = 0.02), while no such pattern was observed with males (95 % CI: − 0.14 -0.13, p = 0.94; Fig. 2).While the magnitude of the mean differences was greater in females (Cohen's d = 0.19) compared to males (Cohen's d = -0.01), the sex effect only approached statistical significance (χ 2 (1) = 3.42, p = 0.06).Upon assessing individuals with depression in comparison to healthy controls overall, irrespective of sex, CRP levels were slightly elevated among depressed individuals relative to healthy controls (Cohen's d = 0.11, 95 % CI: 0.00 -0.22, p = 0.06), although this difference only approached significance.Heterogeneity was further assessed, and moderate heterogeneity was observed in females (I 2 = 69.73%), while low heterogeneity was observed in males (I 2 = 39.26 %).A meta-regression accounting for covariate factors, including BMI and age was performed according to sex.BMI did not account for a significant amount of variation among females with depression (z = 0.49, p = 0.63) and healthy control females (z = -1.00,p = 0.32), nor did it account for variation in males with depression (z = 0.93, p = 0.35) and healthy control males (z = 1.53, p = 0.13).Age did not account for variation among females with depression (z = 0.51, p = 0.61) and healthy control females (z = -0.92,p = 0.36), however we see that age does account for some variation in the findings for males with depression (z = 1.97, p = 0.05) and healthy control males (z = -2.30,p = 0.02).

Publication bias
Publication bias for each individual inflammatory biomarker metaanalysis was assessed.Significant publication bias based on Egger's test was observed for CRP (z = 2.36, p = 0.02; Fig. 5.1), however, the funnel plot appeared symmetric with most studies appearing in the center of the plot.Given that CRP showed a significant publication bias, the trimand-fill method was used, however, no studies were imputed for CRP, and thus a new graph and summary measure of association was not produced.For IL-6, the Egger's test revealed no significant publication bias (z = 1.84, p = 0.07; Fig. 5.2) and the funnel plots displayed symmetry.Significant publication bias was observed for TNF-α based on Egger's test (z = 3.75, p = 0.0002; Fig. 5.3), with the funnel plot showing asymmetry.The trim-and-fill method was used for TNF-α, and two studies were imputed for TNF-α to the right of the funnel plot.After assessing the observed and imputed studies, the overall mean difference remained significant for TNF-α, and the summary measure of effect did not differ (Cohen's d = 0.96; 95 % CI: 0.35 -1.58; Fig. 5.4).

Study quality
The study quality scores ranged from 5 to 8 based on the NOS criteria (Wells et al., 2012).Of the 23 studies included in the meta-analysis, 21 were considered good quality, while two were deemed fair quality.For details, see supplemental material (Supplement C; Table A.2). Due to the small number of studies in the 'fair' category, we were unable to assess whether study quality contributed to the heterogeneity in the reported Cohen's d values across all studies.

Meta-analysis of mean differences
This systematic review and meta-analyses had the overarching goal to investigate whether the inflammatory-depression link was sexspecific.Prior meta-analyses have identified mean elevations in a number of inflammatory markers among individuals with depression compared to controls (Goldsmith et al., 2016;Haapakoski et al., 2015;Köhler et al., 2018;Osimo et al., 2020), however; to our knowledge, this is the first meta-analysis to contrast these relationships in males and females.Given the heterogeneous nature of depressive symptom expression, focusing on sex differences may help to better understand the pathophysiology of depression.
The sex-specific analyses revealed that levels of CRP and IL-6 were elevated only among females with depression compared to healthy controls, effects not found among males.Though the IL-1β meta-analysis lacked statistical power due to limited sample size, the findings for IL-1β mirrored those of IL-6.When considering the magnitude of effect sizes, a small effect size relative to CRP levels, and a moderate effect size relative to IL-6 was present for females.The small effect size for CRP among females suggests that only a small percentage of the means across the studies included in the meta-analysis identified elevations in CRP levels among females with depression relative to healthy controls, while a larger percentage of studies found higher IL-6 levels in females with depression in comparison to healthy controls.Among males, however, the effect size is negligible for CRP, while a small effect size was found for IL-6.Elevated CRP and IL-6 in females with depression, and not males, has similarly been reported in investigations focused on identifying sex differences in young populations (Kim et al., 2021;Zajkowska et al., 2023).For TNF-α, no sex effects were apparent, however, females presented with a moderate magnitude of Cohen's d, while males presented with a large effect size.Given the larger effect sizes among both males and females, higher concentrations among all individuals with depression existed compared to controls, even after publication bias Fig. 5. Funnel plot assessing publication bias for studies examining inflammatory levels in depression and controls among males and/or females.A.5.1.Funnel plot for CRP studies; A.5.2.Funnel plot for IL-6 studies; A.5.3.Funnel plot for TNF-α studies; A.5.4.Trim-and-fill funnel plot for TNF-α studies corrected estimates were applied.IL-6 and IL-1β were also elevated for all individuals with depression compared to controls, whereas this was only trending for CRP.In general, these overall effects for IL-6, TNF-α, and IL-1β support previous meta-analytical findings (Dowlati et al., 2010;Goldsmith et al., 2016;Osimo et al., 2019;Osimo et al., 2020).Nonetheless, a high degree of heterogeneity existed in the current study, which is characteristic of this field (Dowlati et al., 2010;Haapakoski et al., 2015).This heterogeneity suggests that differences in biological markers of depression are owing to diverse symptom dimensions.This could help explain the lack of overall effect for CRP, especially given that our lab, and others, have found that CRP, in particular, maps onto specific depressive symptomatologies (Foley et al., 2021;Franklyn et al., 2022;Jokela et al., 2016).

The role of inflammation in depression among the sexes
Females are disproportionately impacted by depression (Acciai and Hardy, 2017;Auerbach et al., 2016), and may also be more likely to experience specific depressive symptom presentations compared to males.In fact, females with depression tend to express somatic or physical symptoms, including sleep disturbances, fatigue, and changes in appetite (Piepenburg et al., 2019;Silverstein et al., 2013).Accordingly, it is suggested that somatic symptoms, but not cognitive symptoms of depression, are accompanied by increased inflammatory responses (Griffis et al., 2013;Hazeltine et al., 2022).A number of investigations have reported that specific depression symptomatologies are related to high CRP levels (Köhler et al., 2018;Mishra et al., 2018;Osimo et al., 2019).Most reliably, CRP has been linked to somatic or physical symptoms of depression, such as sleeping, tiredness, and eating (Case and Stewart, 2014;Jokela et al., 2016), and not to apathy (Maas et al., 2009).In line with these findings, higher CRP levels were observed in atypical depression (Foley et al., 2021).We too have shown that CRP levels were only elevated among clusters of individuals that displayed predominately physical depressive symptoms, including increased sleep and eating patterns, whereas these elevations were not apparent among individuals in clusters marked mainly by anhedonia (Franklyn et al., 2022).In accordance, somatic and cognitive symptoms of depression, including fatigue, quality of life, and psychomotor skills, have been associative with increased levels of IL-6, while no relations were identified among those presenting with anhedonia-like symptoms of depression (Foley et al., 2024).Others have suggested that IL-6 elevations are more commonly reported in melancholic depression (Křenek et al., 2023).Moreover, depression symptom severity, as well as specific symptoms including cognitive symptoms, interest-activity, and suicidality correlated with CRP levels only among females (Köhler-Forsberg et al., 2017).In patients with MDD, elevations in IL-6 were observed among females but not males after controlling for BMI, and females with depression also presented with higher levels of pessimism and lethargy (Birur et al., 2017).Thus, the elevated levels of CRP and IL-6 in females with depression in the current meta-analysis may be owing, in part, to specific depressive symptoms that are more commonly expressed in females.
Sex-specific differences in depressive phenotypes linked with distinct inflammatory profiles may be influenced by a wide array of immune system effector cells and signaling molecules that modulate levels of inflammatory factors circulating in both the central and peripheral nervous system (Wingo et al., 2023).Sex steroid hormones can influence levels of inflammation by impacting the expression of innate and adaptive immune factors, including immune system initiators and immune monitoring cells, as well as immune cells that act to maintain the immune response, slow pathogenicity, and destroy pathogens (Kropp and Hodes, 2023;Wittenberg et al., 2020).Specifically, testosterone has been shown to exhibit anti-inflammatory properties (Bianchi et al., 2018), and males with depression tend to present with lower testosterone and higher CRP levels compared to healthy controls (Lombardo et al., 2024;Zito et al., 2023).Importantly, MDD-associated gene expression patterns among cell types with significant differentially expressed genes were contributed heavily by microglia in females, which was not the case in males (Maitra et al., 2023).Additionally, increases in monocyte activation associated with higher levels of IL-6 has been observed among females compared to males (O'Connor et al., 2007).It is suggested that immune-interactions with other physiological systems may generate sex-dependent sensitivities that give rise to higher levels of inflammation (Kudielka and Kirschbaum, 2005;Lockwood et al., 2016).Together, these findings may explain the elevated basal CRP and IL-6 among females with depression, though in the context of immune challenges, sex-specific effects may differ (Knight et al., 2020;Majd et al., 2018).
In the current meta-analysis, levels of TNF-α were elevated among all individuals with depression compared to healthy controls, however, this relation was not found when assessing males and females separately and no significant sex differences were found.In line with these findings, following low-dose endotoxin administration, TNF-α level changes correlate to depressed mood, however, this relation was not reported when males and females were examined individually (Moieni et al., 2015).Unlike CRP, or IL-6, there are less suggestions that TNF-α relates to particular depressive symptom profiles.In fact, a systematic review examining relations between inflammatory markers and specific depressive subtypes finds little evidence that TNF-α is linked to specific symptomatologies (Křenek et al., 2023).However, sex differences in the relation between TNF-α and depression have been reported previously, in which depressive symptom severity positively correlated with higher levels of TNF-α among females, but not in males (Birur et al., 2017).In contrast, elevations in production of TNF-α levels following endotoxin were associated with greater depressive symptom severity only among males (Majd et al., 2018).While reduced concentrations of TNF-α have been linked with anti-depressant treatments, such as ketamine, these reductions appear to be short-term (Chen et al., 2018), and sex differences were not observed (Sukhram et al., 2022).These findings highlight several complexities that may influence the role of TNF-α expression among individuals with depression, which have yet to be fully elucidated, and prompts further investigation into potential sex differences.Overall, the current study results confirm that depression is linked to a pro-inflammatory state, wherein elevations of CRP and IL-6 are elevated among females with depression, and biological constructs and clinical symptomatologies may underlie these outcomes.

Strengths, limitations and future directions
The main strength of the current study is the focus on sex differences in the relationship between depression and inflammation, as sex is often overlooked when investigating this link.Another strength of the current meta-analysis is the a priori exclusion of studies that comprised individuals with comorbid conditions, with the exception of anxiety disorders.By accounting for many other psychiatric and physical conditions, this allows us to be more certain of the specific link between inflammation and depression.However, approximately 50-70 % of individuals with depression report at least one other psychiatric or medical condition (Steffen et al., 2020), which is further accompanied by heightened inflammatory responses (Halaris, 2017).It is also critical to consider metabolic comorbidity, as increasing evidence suggests that depression may be an immunometabolic disorder (Refisch et al., 2023;Sánchez-Carro et al., 2022).Atypical features of depression are linked to immunometabolic dysregulations (Milaneschi et al., 2020), wherein metabolic markers, such as glucose, insulin, and albumin levels have been implicated in energy-related symptomatologies of depression (Beurel et al., 2020;Chae et al., 2023).While we have excluded a large number of studies due to comorbidities, factoring in comorbid conditions may be an important approach in the future.
A limitation of the current study is the high levels of heterogeneity in the reported measures of associations of published studies.For this reason, we conducted analyses to assess the influence of potential covariate factors including age and BMI, which have been strongly correlated with inflammation (Qin et al., 2017).We found that age contributes to heterogeneity according to the different inflammation biomarkers.It is critical to acknowledge that fluctuations in ovarian hormones throughout the lifespan may influence both peripheral immune functioning and mental health.Taking into consideration that endogenous sex steroid hormones, including estrogen, play a role in the regulation of the immune system (Dragin et al., 2017;Slavich and Sacher, 2019), it is likely that the cyclic nature of a female's menstrual cycle and menopausal status could modulate inflammatory cytokine production (Engler-Chiurazzi et al., 2022).While a limited number of the articles included in the current meta-analysis assessed menopausal status and menstrual cycle phase, no differences were found in inflammatory concentrations with regard to these factors (Tayefi et al., 2017;Yang et al., 2007).Moreover, all studies that were included in the current meta-analysis age-matched their participants in the depression and healthy control groups, which should limit differences in menopausal status across these groups.We also found that BMI accounted for some level of variability in TNF-α levels among females with depression, and therefore it is important to consider BMI as a possible mediator in the relationship between psychiatric disorders and inflammation.Adiposity has been established as a significant contributor to peripheral immune activation (Ray et al., 2023), and has also been associated with the development of depression (Shelton and Miller, 2011).Therefore, it is important to consider age and BMI in future research, given that these individual factors may play a critical role in linking depression to higher levels of inflammatory markers.
Another limitation of the current investigation is that different blood measures for cytokine and CRP determination were used.The majority of studies in the current meta-analysis assessed peripheral markers of inflammation in serum instead of plasma.However, evidence suggests that cytokines are more stable when measured in plasma than in serum (Guo et al., 2013), and that cytokines of lower abundance are more likely to be detected in plasma (Rosenberg-Hasson et al., 2014).As such, this may reflect methodological inconsistencies including differences in assay sensitivity.In this regard, various types of assays were used to quantify concentrations of biomarkers.While reports suggest that some assays may be more sensitive than others, detection of cytokines using the techniques in the studies included in the meta-analysis have proven accurate and reliable (Liu et al., 2021).Additionally, a range of validated scales and tools were used within the articles to assess depression, which may lead to variability and heterogeneity, especially given the low overlap found regarding types of symptoms assessed across scales (Fried, 2017).Furthermore, publication bias was detected, which may threaten validity of results.However, alternate models were run to account for publication bias, and no significant differences in the results were found.Lastly, only inflammatory markers and pro-inflammatory cytokines were considered in the current meta-analyses, and of those considered, a limited number of studies met the inclusion criteria regarding IL-1β.Given that anti-inflammatory markers also relate to depression (Dhabhar et al., 2009;Osimo et al., 2019), investigating sexbased differences in anti-inflammatory cytokine expression will be important.

Conclusion
The current systematic review and meta-analyses reveals sex-specific inflammatory-depression linkages for CRP, IL-6, and IL-1β, but not for TNF-α.High heterogeneity in the current study may suggest that identifying the relations between inflammatory biomarkers and specific symptom subtypes of depression separately in males and females could inform more tailored treatments in the future.In this regard, immune activation is thought to be an underlying factor in treatment resistance (Chamberlain et al., 2019), and with ongoing clinical trials examining the therapeutic benefit of an adjunct anti-inflammatory treatment, the current meta-analytical findings highlight the importance of considering sex differences in the inflammatory-depression relationship.

Fig. 2 .
Fig. 2. Forest plot showing Cohen's d values for standardized mean differences in CRP levels in males and females with depression compared to healthy controls.

Fig. 3 .
Fig. 3. Forest plot showing Cohen's d values for standardized mean differences in IL-6 levels in males and females with depression compared to healthy controls.

Fig. 4 .
Fig. 4. Forest plot showing Cohen's d values for standardized mean differences in TNF-α levels in males and females with depression compared to healthy controls.