A Missense Variant in the IL-6 Receptor and Protection from Peripheral Artery Disease

Interleukin-6 (IL-6) represents a key effector of immune activation and systemic inflammation, which have been linked to adverse cardiovascular outcomes. Observational genetic studies have linked a common (minor allele frequency [MAF] = 0.33) missense variant (rs2228145; p.Asp358Ala) in the IL-6 receptor (IL6R) with reduced risk of coronary artery disease (CAD), abdominal aortic aneurysm, ischemic stroke, and peripheral artery disease (PAD) (1, 2). This common variant impairs classical IL-6 signaling by reducing membrane-bound IL6R, increasing circulating levels of IL6R (+34.3% per minor allele), and dampening the inflammatory response (1, 3). The effects of this variant on trans-signaling are less certain (3). Mendelian randomization (MR) analyses have similarly highlighted IL6 as a potential therapeutic target, providing a causal link between genetic downregulation of IL-6 and decreased risk of CAD, stroke, and abdominal aortic aneurysm (4). Although PAD represents a common form of atherosclerotic cardiovascular disease associated with substantial morbidity and mortality, whether systemic downregulation of IL-6-mediated inflammation represents a causal therapeutic target for PAD remains unknown. To investigate, we performed three complementary analyses utilizing data from the VA Million Veteran Program (MVP).

First, we studied 35,042 PAD cases and 247,115 controls with available rs2228145 (IL6R Asp358Ala) genotype and electronic health record data, with phenotyping, genotyping, and quality control performed as previously described, with diagnosis codes extracted from the VA electronic health record (5). Logistic regression (unless otherwise specified) was performed within each genetically-determined ancestral population, adjusted for age, sex, 5 genetic principal components, and genotyping batch, and combined using fixed-effects meta-analysis to test the association between rs2228145 genotype and each PAD-related outcome. For comparison, we studied 77,241 CAD cases and 139,284 controls.

Among the overall cohort, IL6R Asp358Ala was associated with decreased risk of PAD, without evidence of significant heterogeneity across ancestries (I² = 0.21; Cochran’s Q p-value = 0.28) (Figure 1A). Results were similar when meta-analyzed with PAD GWAS from BioBank Japan and FinnGen, and in models adjusted for each traditional risk factor (hypertension, diabetes, smoking, LDL-cholesterol) among MVP participants. No interactions between IL6R Asp358Ala and each risk factor or statin-use were detected. Results were also similar for CAD (Figure 1A), and in multinomial logistic regression examining increasingly severe manifestations of PAD in MVP (Figure 1B).

FIGURE.

Associations between IL6R Asp358Ala and A) PAD (MVP: 35,042 cases [27,013 European; 5,939 African; 2,090 Hispanic] and 247,115 controls [176,750 European; 49,403 African; 20,962 Hispanic]; and fixed-effects meta-analysis of MVP, BioBank Japan [BBJ], and FinnGen release 5 [FNG]) and CAD (MVP: 77,241 CAD cases [62,060 European; 10,775 African; 4,406 Hispanic] and 139,284 controls [97,640 European; 28,265 African; 13,379 Hispanic]), and B) increasingly-severe manifestations of PAD (claudication; chronic limb threatening ischemia [CLTI]: rest pain, gangrene, ulcer; and amputation) among MVP participants. C) Plot comparing the p-values for genetic variants associated with sIL6R (IMPROVE) and PAD (MVP) at the +/− 1Mb region surrounding the lead sIL6R pQTL (rs4129267; located on chromosome 1, identifying rs7521458 as the variant with the lowest combined p-value as determined by LocusCompare (https://github.com/boxiangliu/locuscompare) and D) regional association plots for IL6R (IMPROVE) and PAD (MVP) at this locus, with colocalization identifying a high probability of a single shared causal genetic association. E) MR results for effect of sIL6R (IMPROVE) on PAD (MVP). As genetic instruments we utilized 15 independent (r² < 0.1) SNPs associated with log-transformed sIL6R at genome-wide significance (p < 5 × 10⁻⁸) in the IMPROVE study. Mean F-statistic was 174.5 (range 31.6–1456.9). Unadjusted p-values are presented for all analyses.

Next, we tested for the presence of a shared genetic signal influencing circulating (soluble) IL-6R (sIL6R) and PAD using colocalization (coloc; https://chr1swallace.github.io/coloc/). A protein quantitative trait locus (pQTL) associated with sIL6R was previously identified among up to 3394 participants of the IMPROVE study (https://doi.org/10.5281/zenodo.264128). We identified strong evidence for a single shared causal genetic signal for sIL6R and PAD (posterior probability 0.93) for the region +/− 1 Mb surrounding the lead sIL6R pQTL variant (rs4129267) (Figure 1C–D).

Finally, we performed two-sample MR to further assess the causal role between IL-6 signaling and PAD. The hypothesized mechanism for CAD risk-reduction is shedding of the IL6 receptor from the cell surface into circulation, leading to decreased classical IL6 signaling (1, 3). Higher circulating levels of sIL6R would therefore be expected to decrease classical IL6 signaling, leading to reduced risk of PAD.

Indeed, in inverse variance weighted MR, increased levels of sIL6R were associated with significantly decreased risk of PAD (Figure 1E). We did not detect evidence of directional pleiotropy (Egger intercept p-value > 0.05) or significant heterogeneity (Cochran’s Q p-value > 0.05). Results were consistent across MR methods that make different assumptions about the presence of pleiotropy (weighted median and MR-PRESSO).

Overall, these results implicate IL-6 signaling in the pathogenesis of PAD, across a combination of observational and causal genetic analyses. These results are consistent with a prior phenome-wide association study of rs2228145 in MVP (2). We extend these findings to 1) link IL-6 signaling to PAD, PAD subtypes, and risk of amputation; and 2) prioritize IL6R as a causal therapeutic target to reduce risk of PAD. Consistent findings across both PAD and CAD represent evidence for a role of inflammation in systemic atherosclerosis. Future study is necessary to determine whether rs2228145 may be associated with decreased CAD burden/severity (considering our finding of reduced risk of severe PAD/amputation). This large, multi-ancestry study in the MVP population had a high prevalence of PAD (12.5%), in contrast to a prior study of European UK Biobank participants which failed to detect an association between IL-6 signaling and PAD (4). Limitations include the high proportion of men among MVP participants (>90%). Recent randomized trials focused on CAD outcomes have demonstrated beneficial effects of targeting systemic inflammation with canakinumab and colchicine, and monoclonal antibodies targeting IL6 signaling (tocilizumab, sarilumab, siltuximab) are clinically approved for other indications. Phenome-wide association and Mendelian randomization studies demonstrate a similar safety profile for genetically-proxied decreases in classical IL6 signaling and therapeutic modulation of IL6/IL6R (2, 4). Trials targeting systemic inflammation may be warranted to characterize the net benefits in the prevention of PAD specifically.