Methods
A two-sample MR study was conducted to investigate the association of genetically proxied inhibition of IL-6 signalling with different measures of renal function: estimated glomerular filtration rate (eGFR), CKD and blood urea nitrogen (BUN). Two-sample refers to the fact that the instrument-exposure and instrument-outcome estimate are obtained from two different genome wide association study (GWAS), increasing the statistical power of the MR study.
Genetic variants for downregulated IL-6 signalling were selected as independent (linkage disequilibrium r2< 0.1) single-nucleotide polymorphism (SNPs) within 300 kB of the IL-6 receptor gene (IL6R ) that strongly associated with circulating CRP (p < 5×10-8 with independent associations as described by Georgakis et al8), a downstream target of IL-6 signalling. The association between the IL6R genetic variants and CRP level were taken from the summary data of the Cohorts for Heart and Aging Research in Genomic Epidemiology (CHARGE) Inflammation Working Group GWAS of 204,402 individuals of European ancestry.13 In this GWAS, association estimates are expressed as change in natural log transformed CRP (mg/L). The variance in CRP levels explained by the genetic variants,R2 , was calculated using the formula: R2 = [2 × MAF × (1–MAF) × β2]/var(log(CRP)), where MAF is the minor allele frequency and β is the effect estimate of the SNP on CRP levels. F-statistics, a measure of instrument strength in MR, were calculated using the formula: F = 𝑅2×(𝑁−2)/(1−𝑅2) whereR 2 is the variance of CRP explained by the specific variant and N the number of individuals in the GWAS analysis.
Summary GWAS data from the Chronic Kidney Disease Genetics (CKDGen) Consortium meta-analyses were used for the primary outcome associates of log eGFR (61 studies, 765,348 participants), BUN (33 studies, 416,178 participants), CKD (30 studies, 64,164 cases and 561,055 controls).14 Log eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation in adults and using the Schwartz formula for participants who were 18 years or younger. CKD was constructed as a binary outcome based on an eGFR < 60 ml min−1 per 1.73 m2. BUN was calculated as 2.8 x blood urea (mg/dl).
Data for the exposure and outcome were harmonised according to the effect allele and no exclusions were made for palindromic variants. Random-effects inverse-variance weighted method was used as the primary MR analysis. To assess potential pleiotropy, we conducted sensitivity analyses using the simple median, weighted median and MR-Egger methods. The median methods are robust even with up to 50% of the contribution to the MR estimates coming from invalid instrumental variables.15 MR-Egger provides robust estimates even when all instrumental variables are invalid, as long as the INstrument Strength Independent of Direct Effect (INSIDE) assumption holds - that any pleiotropic effect of the variants on the outcome are independent of the strength of their association with the exposure.15The estimated MR-Egger intercept is indicative of the average pleiotropic effect of the variants used.15 We tested for such pleiotropy by assessing whether our intercept was significantly different from zero.15 Results are presented as effect estimates and corresponding 95% confidence intervals in forest plots. For eGFR and BUN respectively, estimates represent the change in log eGFR or blood urea nitrogen per genetically predicted decrease in natural log of CRP. For CKD the results are expressed as odds ratio of CKD, with 95% confidence intervals. All data analyses were performed by the “TwoSampleMR” package version 4.26 in R software.