Peripheral levels of C-reactive protein, tumor necrosis factor-α, interleukin-6, and interleukin-1β across the mood spectrum in bipolar disorder: A meta-analysis of mean differences and variability

Importance: It is unclear whether differences exist in the magnitude and variability of pro-inflammatory mediators in the different phases of bipolar disorder (BD) and among subjects with BD, as compared to healthy controls. Objective: To run a comparative meta-analysis of C-Reactive Protein (CRP), IL-1, IL-6, TNF-α in BD vs healthy controls, measuring mean and variability effects on all subjects. Sensitivity analyses include disease activity. Data sources: Systematic review of observational studies in PubMed and PsycInfo up to February 2nd, 2020. Study selection: Case-control studies reporting inflammatory mediators' levels in BD and controls. Data extraction and synthesis: Summary distribution measures of circulating CRP, IL-1β, IL-6, TNF-α in participants with BD and control groups were extracted. Random-effects multivariate meta-analyses were conducted based on individual study/mediator effect sizes (Hedge's g). Main outcomes and measures: Co-primary outcomes were inflammatory mediators' levels (Hedge's g) and variability (coefficient of variance ratio (CVR)) differences between participants with BD across the mood spectrum and controls. Results: Out of the initial 729 papers, 72 were assessed and then excluded after full-text review, and ultimately 53 studies were included in the systematic review, while 49 were included in the meta-analysis. The mean age was 36.96 (SD: 9.29) years, and the mean female percentage was 56.31 (SD: 16.61). CRP (g = 0.70, 95% CI 0.31-1.09, k = 37, BD = 2,215 vs HC = 3,750), IL-6 (g = 0.81, 95% CI 0.46-1.16, k = 45, BD = 1,956 vs HC = 4,106), TNF-α (g = 0.49, 95% CI 0.19-0.78, k = 49, BD = 2,231 vs HC = 3,017) were elevated in subjects with BD vs HC, but not IL-1β (g = -0.28, 95% CI -0.68-0.12, k = 4, BD = 87 vs HC = 66). When considering euthymic, depressive, and manic episodes separately, CRP and TNF-α were elevated in both depressive and manic episodes, but not in euthymia, while IL-6 remained elevated regardless of the disease state. No difference in CVR emerged for CRP, IL-1β, and TNF-α, while a lower CVR was observed for IL-6. When considering disease phases, CVR was higher in BD than in HCs for CRP during depressive episodes, lower for IL-6 during euthymia, and higher during manic episodes for CRP, IL-6, and TNF-α. Sensitivity analyses after excluding outliers identified with funnel plot visual inspection, low-quality studies, and considering only studies matched per body mass index confirmed the main results. Meta-regression showed that age (IL-6, TNF-α), gender (CRP), duration of illness (CRP) moderated elevated individual inflammatory levels. Conclusions and relevance: Peripheral pro-inflammatory marker elevations were confirmed in BD. CRP and TNF-α could represent state markers, as they were only elevated during mood episodes, while IL-6 appeared to be a trait marker for BD. Increased variability of specific inflammatory mediators in specific disease active states suggests that a subset of subjects with BD may exhibit elevated inflammation as part of a manic or depressive episode.