The effect of genetically proxied IL-6 signalling on severe malaria: A Mendelian randomisation analysis

IntroductionSevere malaria remains a deadly disease for many young children in low- and middle-income countries. Levels of Interleukin-6 (IL-6) have been shown to identify cases of severe malaria and associate with severity, but it is unknown if this association is causal, or whether manipulation of IL-6 might alter outcomes in severe malaria. MethodsA single nucleotide polymorphism (SNP, rs2228145) in the IL-6 receptor (IL6R) was chosen as a genetic variant that is known to alter IL-6 signalling. We measured the association between the minor allele of this SNP (C) and C-reactive protein (CRP) levels, a marker of IL-6 signalling in the non-European ancestry population recruited to UK Biobank. We then took this forward as an instrument to perform Mendelian randomisation (MR) in MalariaGEN, a large cohort study of patients with severe malaria at eleven worldwide sites. As a secondary approach, we identified cis protein quantitative trait loci (cis-pQTL) for IL6R itself and other markers of IL-6 signalling in a recently published GWAS of the plasma proteome performed in African Americans. We then performed MR using these instruments in the African MalariaGEN sites (9/11). Analyses were performed at each site, and meta-analysed using inverse variance weighting. Additional analyses were performed for specific sub-phenotypes of severe malaria: cerebral malaria and severe malarial anaemia. ResultsThe minor allele (C) of rs2228145 was associated with decreased CRP across all tested continental ancestries in UK Biobank. There was no evidence of heterogeneity of effect and a large overall effect (beta -0.11 per standard deviation of normalised CRP per C allele, p = 7.55 x 10-255) In Mendelian randomisation studies using this SNP, we did not identify an effect of decreased IL-6 signalling on severe malaria case status (Odds ratio 1.14, 95% CI 0.56 - 2.34, p = 0.713). Estimates of the association with any severe malaria sub-phenotype were similarly null although there was significant imprecision in all estimates. Using an alternative instrument (cis-pQTLs for IL6R), which included 3 SNPS (including rs2228145), we identified the same null effect, but with greater precision (Odds ratio 1.02, 95% CI 0.95 - 1.10), and no effect on any severe malaria subtypes. ConclusionsMendelian randomisation analyses using a SNP in the IL-6 receptor known to alter IL-6 signalling do not support a causal role for IL-6 signalling in the development of severe malaria, or any severe malaria sub-phenotype. This result suggests IL-6 may not be causal for severe outcomes in malaria, and that therapeutic manipulation of IL-6 may not be a suitable treatment for severe malaria.