The inflammatory proteome and risk of atrial fibrillation: genetic insights into causal relevance

Abstract Background Cardiac and systemic inflammation have been associated with a higher risk of atrial fibrillation (AF) in both animal models and human studies. Causality, however, has been questioned due to inconsistent efficacy of anti-inflammatory treatments in preventing incident or recurrent AF, and potentially confounding associations with AF risk factors such as ischaemic heart disease. Inflammatory processes are highly complex and involve a large number of proteins, but the individual relevance and causal role of these for AF remains unclear. Purpose To explore the nature of the relationship between the inflammatory proteome and risk of AF. Methods Associations between 26 inflammatory proteins and AF were assessed in a two-sample Mendelian randomization (MR) framework, with primary MR causal estimates with AF estimated among 339,214 White British UK Biobank (UKB) participants (AF cases=30,630). MR analyses for each inflammatory protein included independent variants previously associated with the relevant protein at genome-wide significance (p<5x10-8) across the genome (trans-protein quantitative trait loci [pQTLs]), or restricted to at least 50kb from the closest upstream gene and 50 kb downstream from the target gene (cis-pQTLs). Sensitivity analyses to assess potential pleiotropy were conducted (MR-Egger, weighted median, and weighted mode). Associations passing multiple testing threshold (p<0.002) in UKB were further validated in meta-analyses with independent datasets including over 50,000 AF cases (FinnGen and AFGen). Research undertaken using UKB application 14568. Results This inflammatory protein-wide study identified three proteins that have support for causal associations with AF. Genetically-predicted levels of IL6R, CD40, and Chitinase-3-like protein 1 (YKL40) each showed a negative association with risk of AF, with up to a 5% lower AF risk per standard deviation higher genetically-predicted levels of these proteins (Figure 1). These findings were further confirmed in meta-analysis with data from AFGen and FinnGen. No evidence of pleiotropy was identified, and results were also confirmed in cis-pQTL analyses, which were fully consistent with the trans-pQTL analyses (Table 1). Analyses stratified by various baseline characteristics highlighted stronger causal estimates in those with higher levels of C-reactive protein (with interaction p-values <0.001). Conclusions Genetic approaches support the relevance of lifelong differences in the inflammatory proteome for AF, and causal relationships between specific inflammatory markers and risk of AF. Greater soluble IL6R and CD40 as measured in proteomic analyses are both associated with lower levels of inflammation. As previous work indicates that genetically upregulated CD40 and IL6R are similarly associated with a lower risk of ischemic stroke, the risk-benefit of directly targeting these pathways needs further consideration.Figure 1Table 1