Inflammatory Biomarkers and Risk of Schizophrenia

Key Points Question What is the effect of increased inflammatory biomarkers on the risk of developing schizophrenia? Findings In this 2-sample mendelian randomization study using summary gene-biomarker association results estimated in pooled samples ranging from 1645 to more than 80 000 individuals, 2-fold increments in circulating levels of C-reactive protein and soluble interleukin-1 receptor levels were associated with a 10% reduction and a 6% increase in the lifetime odds of developing schizophrenia. Meaning We found that blockade of interleukin-6 effects and low C-reactive protein levels might increase schizophrenia risk, possibly due to increased susceptibility to early life infection.


Datasets of summary association results
The datasets of summary genetic associations datasets (shown in eTables 1 and 2) used in this work as described below.
 Circulating C reactive protein (CRP) Two sets of CRP-associated SNPs were used. One included 18 independent SNPs identified in a GWAS in ≤82,725 European ancestry individuals. 1 SNP-CRP linear regression coefficients (changes in ln-transformed circulating CRP levels) and standard errors were obtained assuming additive genetic effects and adjusting for sex, age, recruitment site (if necessary) and relatedness (for family studies). Given that a statistical criterion was used to select these 18 instruments, they are hereafter referred to as the liberal set.
The second set included four variants in the CRP gene, which explain 98% of the genetic variation in this locus in European ancestry populations and have been shown to regulate CRP expression levels without changing protein sequence. 2 SNP-CRP linear regression coefficients (changes in lntransformed circulating CRP levels) and standard errors for these variants were obtained from the CRP Coronary Heart Disease Genetics Collaboration in a sample mostly of European ancestry. The analyses were adjusted for ancestry to minimize residual confounding due to population stratification. 3 These four instruments are hereafter referred to as conservative because they are located in the CRP gene region and their selection incorporated biological criteria. Therefore, we considered this approach less likely to be biased by horizontal pleiotropy. These variants are in partial linkage disequilibrium with one another (eTable 3). A single missense SNP (Asp358Ala) in the IL6R gene was selected as the IL-6R instrument. Binding of IL-6 to its membrane-bound receptor, IL-6R, is necessary for triggering classical IL-6 effects on hepatocytes and some leukocytes. 6,7 The Asp358Ala SNP increases cleave of membrane-bound IL-6R cleavage and blockade of IL-6 classical cell signaling. Therefore, downstream effects of IL-6 are attenuated in Ala-allele carriers, despite increased levels of IL-6 and soluble IL-6R (sIL-6R) 8 IL6R is a target for drugs like tocilizumab, a monoclonal antibody used to reduce systemic and articular inflammation in patients with rheumatoid arthritis. 9 A large scale Mendelian randomization study showed that this genetic variant was not associated with a wide range of biomarkers, except for those that are widely known downstream consequences of IL-6 (i.e. CRP and fibrinogen). In addition the effects of this genetic variant were consistent with the effects of tocilizumab. 10 SNP-sIL6R associations were estimated in 1,645 individuals included in a large collaborative MR study. 11 These estimates were presented in percentage differences, which were converted to lntransformed units.
 Schizophrenia SNP-schizophrenia ln(odds ratio) and standard errors were downloaded from the Psychiatric Genomics Consortium website (http://www.med.unc.edu/pgc/downloads). Logistic regression assuming an additive genetic effect and adjusting for ancestry-informative principal components were performed. Analyses included 34,241 cases and 45,604 ancestry-matched controls (most of European ancestry), as well as three family-based studies of 1,235 European ancestry trios. 12

Testing for influential genetic variants
To identify potentially influential instruments in the liberal set of CRP instruments, two tests of influence (based on studentized residuals or Cook's distance) were applied separately for IVW or MR-Egger regression. 38,39 P-values for the studentized residuals test were obtained from a Student's t-distribution with degrees of freedom equal to L−2 (for IVW) or L−3 (for MR-Egger regression), with L being the number of genetic instruments. The F distribution with joint degrees of freedom equal to (1, L−1) (for IVW) or (1, L−2) (for MR-Egger regression) was used for the Cook's distance test. We then applied three different statistical significance criteria to classify SNPs as potentially influential: P<0.01, P<0.05 n or P<0.1 in at least one of the influence tests.

Mediation analysis
The expected effect assuming full mediation ( ) can be estimated as