Inflammatory protein markers and treatment outcomes across mood and psychotic disorders: a systematic review and meta-analysis

Poor treatment outcomes worsen prognosis in schizophrenia, bipolar disorder (BD), and major depressive disorder (MDD) and likely involve shared mechanisms, including alterations in inflammatory protein markers. However, the role of these markers on treatment outcomes was not previously studied cross-diagnostically, and to fill this gap, we performed a systematic review and meta-analysis. We searched major scientific databases for studies providing quantitative data on the association between peripheral inflammatory protein markers and psychopharmacological treatment outcomes in schizophrenia, BD, and MDD. Cross-diagnostic random-effect meta-analyses were performed, with heterogeneity, publication bias, and study quality evaluation. Sensitivity analyses and meta-regressions were performed. The study protocol followed the PRISMA guideline and was preregistered. A total of 105 and 42 studies entered the systematic review and meta-analysis, respectively. Sixty-one samples (N = 5787) considered MDD, 26 samples (N = 2113) schizophrenia, and BD was the least represented diagnosis (12 samples, N = 342). IL-6 was the most investigated biomarker, with no evidence of association with treatment outcomes. For TNF-α, responders demonstrated greater baseline-endpoint concentration decrease (SMD = 0.75, 95% CI [0.14-1.35]; k = 7). CRP low vs high baseline levels were associated with remission (OR = 1.84, 95%CI [1.33, 2.54]; k = 6), with cut-off definitions varying across studies. For other markers no associations emerged except that for IL-8, which was higher at baseline in non-responders when considering mood disorders (SMD = -0.38; 95%CI [-0.73, -0.03]; k = 6). Heterogeneity was high across most analyses. Sensitivity analyses and meta-regressions generally did not add meaningful findings. Inflammatory biomarkers still offer limited predictive value for treatment outcomes across psychiatric disorders and the available evidence is substantially limited to MDD. Standardised methodological workflows and a shift from candidate proteins to hypothesis-free proteomics are key considerations for future research.