Atypical cytokine profiles in people on the autism spectrum: a comprehensive systematic review and meta-analysis including 54 cytokines

Abstract Atypical peripheral blood cytokine concentrations have been shown in autism, but no clear pattern has been observed. This systematic review and meta-analysis summarised current state of findings, expanded the range of cytokines, accounted for study risk of bias, and examined relations between cytokines and autism traits. Literature comparing peripheral blood cytokine in autistic and non-autistic people was systematically searched in Ovid ® Embase, MEDLINE and APA PsycINFO, Web of Science ™ and Scopus, resulting in 98 studies and 54 cytokines (4236 autistic, 3333 non-autistic controls; age 2 to 65 years) in the meta-analysis. Study risk of bias was assessed using adapted Newcastle-Ottawa Scale. Compared to controls, autistic people had elevated levels of IL1-beta (Hedges’ g = 0.620, 95%CI[0.32, 0.92]), IL4 ( g = 0.245, 95%CI [0.07, 0.42]), IL6 ( g = 0.365, 95%CI [0.011, 0.62]), IL8 ( g = 0.384, 95%CI [0.15, 0.62]), IFN-gamma ( g = 0.404, 95%CI [0.09, 0.72]), TNF-alpha ( g = 0.31, 95%CI [0.11, 0.51]), CXCL1/GRO-α ( g = 0.364, 95%CI [0.058, 0.670]) and MIF ( g = 0.560, 95%CI [0.14, 0.98]). Over a third of studies were classified as having a high risk of bias; their removal revealed higher IL7 and IL1RA in autism relative to controls. Narrative synthesis produced no strong evidence for an association between cytokine and autism traits among autistic individuals. Altogether, our findings support a predominance of pro-inflammatory cytokines, while also indicating potential modulatory contributions from inhibitory cytokines, which reflect group-level differences between autistic and non-autistic individuals, but not variations of autism traits within the autistic population. However, higher-quality studies with low risk of bias are needed before firm conclusions can be drawn.