Predictors of successful anti-inflammatory drug trials in patients with schizophrenia: A meta-regression and critical commentary

Given evidence pointing toward a role for immune dysregulation in the pathogenesis of schizophrenia, anti-inflammatory agents are promising adjunctive treatments that have potential to support a causal relationship for inflammation and psychopathology and lead to novel treatments for individuals. Indeed, previous meta-analyses have demonstrated small-to-medium effect sizes (ES) in favor of various anti-inflammatory agents, though there is significant heterogeneity and challenges in the interpretation of this literature. Identifying predictors, including sociodemographic variables, trial duration, and/or symptoms themselves, of successful anti-inflammatory trials may help identify which patients who might benefit from these compounds. We performed a meta-regression analysis of 63 adjunctive anti-inflammatory trial arms (2232 patients randomized to adjunctive anti-inflammatory agents and 2207 patients randomized to placebo).Potential predictors of effect size estimates for changes in psychopathology scores from baseline to endpoint included geography, trial duration, sample size, age, sex, race, smoking, body mass index, illness duration, age of onset of psychosis, study quality score and psychopathology scores (total and subscale) at baseline. Geography (β = 0.31, p = 0.011), smaller sample size (β = 0.33, p = 0.009), and higher study quality score (β = 0.44, p < 0.001) were significant predictors of larger ES estimates for change in total psychopathology in favor of anti-inflammatory agents. Smaller sample size (β = 0.37, p = 0.034) and higher study quality score (β = 0.55, p = 0.003) were significant predictors of larger ES estimates for change in negative psychopathology in favor of anti-inflammatory agents. Higher study quality score (β = 0.46, p = 0.019) was a significant predictor of larger ES estimates for change in general psychopathology in favor of anti-inflammatory agents. These findings should be interpreted with caution given concerns of publication bias regarding the geographic differences and small study effects. The lack of an association with other demographic variables should be seen as a primary limitation of the literature that needs to be considered in future studies. The association with study quality score suggests that future anti-inflammatory trials must consider demographic variables known to be associated with inflammation (e.g., BMI and smoking) and evidence of increased baseline inflammation should be incorporated in study design. Moreover, evidence of target engagement and endpoints thoughts to be associated with increased inflammation should be considered as well.