Abstract
Background Inflammation is an emerging target for the prevention and treatment of cardiovascular disease (CVD). This drug-target Mendelian randomization (MR) study aimed to predict the on-target effects of IL-18 inhibition on CVD risk. Furthermore, we aimed to explore the effects of IL-18 inhibition on cardio-metabolic traits, cardiac structure, and function, and identify potential adverse outcomes.
Methods We selected five independent circulating IL-18-lowering variants around the IL-18 gene locus from the Systematic and Combined AnaLysis of Olink Proteins (SCALLOP) consortium. We then performed two-sample MR analyses to investigate the association of genetically proxied IL-18-inhibition on downstream inflammatory markers, risk of CVD, cardiac magnetic resonance (CMR) imaging measurements of cardiac structure and function, cardiometabolic traits, and a selection of potential adverse effects. We utilized data from the UK Biobank, Cardiogram, GIGASTROKE, and other large genomic consortia (sample range: 3,301-1,320,016).
Results Following correction for multiple comparisons, one standard deviation (SD) lower in genetically-predicted circulating IL-18 was associated with reductions in downstream biomarkers of IL-18 signaling, including C-reactive protein (SD change –0.02, 95% CI –0.03, –0.02), tumor necrosis factor (SD change –0.19, CI –0.25, –0.14), interferon-gamma (SD change –0.15, CI –0.22, –0.08), and CXCL10 (SD change –0.13, CI –0.16, –0.09). Lower genetically-predicted IL-18 levels were associated with reduced risk of cardio-embolic stroke (Odds Ratio [OR] 0.85, CI 0.79-0.92), but not other stroke subtypes. Furthermore, lower genetically predicted IL-18 levels were associated with reduced risk of peripheral arterial disease (OR 0.91, CI 0.84-0.97), atrial fibrillation (OR 0.94, CI 0.89-0.99), and heart failure (OR 0.84, CI 0.77-0.92), as well as improvements in CMR traits, including a reduction in left atrial volume (β –0.02, CI –0.03, –0.00). Lower genetically-predicted IL-18 levels were associated with lower risk of chronic kidney disease, autoimmune diseases, a favorable cardio-metabolic profile, and higher odds of lung cancer, but not infections.
Conclusions Our study provides genetic support that impaired IL-18 signaling may be causally associated with a lower risk of cardio-embolic stroke, possibly mediated through prevention of cardiac re-modelling, heart failure and atrial fibrillation. IL-18 represents a potential target for anti-inflammatory therapy in stroke and CVD that warrants further investigation in clinical trials.
Using multi-omic data, this Mendelian Randomization study provides evidence that IL-18 lowering is associated with a lower lifetime risk of cardiac remodeling, heart failure, and cardio-embolic stroke.
A significant proportion of the protective effect of impaired IL-18 signaling on cardio-embolic stroke was mediated through a reduced risk of AF.
These data provide compelling evidence that the IL-18 signaling pathway is a promising druggable target for the treatment of heart failure and the prevention of cardio-embolic stroke.
Several monoclonal antibodies targeting IL-18 are in development for the treatment of atopic dermatitis and could be considered for re-purposing for cardiovascular disease.
Introduction
Inflammation has emerged as a promising target for the treatment of atherosclerosis and the secondary prevention of coronary events.1, 2 The first anti-inflammatory trials for secondary prevention after stroke have primarily focused on the therapeutic potential of these agents in patients with atherosclerotic mechanisms and excluded patients with cardio-embolic stroke.3 However, several complementary lines of evidence also support the importance of inflammatory mechanisms in the pathophysiology and progression of heart failure (HF), atrial fibrillation (AF), and cardio-embolic stroke. Animal models provide compelling data that the NOD-like receptor family pyrin domain-containing 3 (NLRP) inflammasome is upregulated in cardiomyocytes in the pathogenesis of AF and promotes structural and electrical atrial remodeling.4 Inflammatory signaling is implicated in several pathways central to the pathophysiology of HF, including mitochondrial dysfunction, calcium homeostasis, cardiac fibrosis, and impaired cardiomyocyte contractility.5, 6 Inflammatory markers are associated with incident AF in population-based studies7, 8 and AF-related stroke.9 Despite these promising data, the results of the first RCTs of anti-inflammatory therapies in HF,10, 11 and AF12 have been neutral.
The NLRP3 inflammasome mediates some of its effects through activation of interleukin-1β (IL-1β) via caspase-1 and leads to downstream IL-6 expression. Pre-clinical,13 observational,14, 15 and genetic studies16 have coalesced to identify the IL-1β-interleukin-6 (IL-6) signaling pathway as a promising target for risk reduction in cardiovascular disease17 and stroke.18 Significantly less attention has been given to the role of interleukin-18 (IL-18), which is also cleaved to its active form by caspase-1 following activation by the NLRP3 inflammasome.13 IL-18 is a potent pro-inflammatory cytokine that promotes atherosclerotic plaque progression in pre-clinical work.13 However, experimental data have also demonstrated that IL-18 signaling plays a key role in cardiomyocyte hypertrophy, ventricular dysfunction, and cardiac fibrosis.19–22 Elevated IL-18 levels are observed in patients with HF and correlate with disease severity, but whether this inflammatory response is a cause or a consequence of the pathogenesis of HF is unknown.23 Some, but not all, prospective cohort studies have reported that IL-18 is associated with incident coronary events24–26 and HF,24, 27 but not incident stroke.24 It remains uncertain whether pharmacological inhibition of the IL-18 signaling pathway is a viable treatment option for a variety of cardiovascular diseases, including coronary disease, stroke, AF, and HF.
In the absence of RCTs of therapeutics targeting the IL-18 signaling pathway for the treatment of cardiovascular disease, genetic studies can offer valuable insights into the on-target effect of these pharmacological agents.28, 29 Drug-target Mendelian Randomization (MR) employs genetic variants associated with the levels or activity of drug-target proteins, thus serving as proxies for pharmacological agents.28 The primary aim of this study is to explore the potential benefit of IL-18 inhibition on cardiovascular outcomes using a drug-target MR approach.
Methods
Study design
We used MR to assess the causal association between genetically proxied downregulation of IL-18 signaling and: 1) cardiovascular disease outcomes, with replication in non-overlapping samples where possible; 2) cardiometabolic and inflammatory traits, including cardiac magnetic resonance (CMR) measures of cardiac structure and function; and 3) potential adverse and beneficial effects associated with reduced IL-18 signaling. An overview of the study design is detailed in Figure 1.
Initially, we examined the validity of our genetic instrument as a pharmacological proxy for downregulated IL-18 signaling by assessing its impact on established downstream biomarkers: IL-6, C-reactive protein (CRP), interferon-gamma (IFN-γ), and tumor necrosis factor. These biomarkers were identified from pre-clinical30 and early-phase clinical studies of therapies targeting the IL-18 signaling pathway.31, 32 Circulating levels of IL-1β were selected as a negative control because as IL-1β functions through an alternative signaling pathway, its levels are not expected to be affected by IL-18 signaling. Two-sample cis-MR was then used to elucidate the relationship between on-target reduced IL-18 signaling and the risk of cardiovascular disease. Subsequent analyses assessed the bidirectional effects between liability to diseases identified in primary analyses and circulating IL-18 levels. We undertook a mediation analysis to identify pathways that contribute to the effect of genetically predicted IL-18 blockade on cardio-embolic stroke. Finally, we explored the effect of downregulated IL-18 signaling on longevity and a selection of clinically relevant outcomes to identify potential adverse outcomes associated with an impaired host immune response. We used parental age at death, a partially heritable trait, as an indirect measure of longevity.33
Data Sources
For our primary CVD outcomes, we obtained summary statistics for stroke from GIGASTROKE (Ncases = 62,100), peripheral arterial disease from the GoLEAD and Summit Consortium (Ncases = 12,086), coronary artery disease from CARDIoGRAMplusC4D (Ncases = 181,522), HF from Aragram et al’s GWAS of ischaemic and non-ischaemic HF in the UK Biobank (UKB) (Ncases = 6,504), dilated cardiomyopathy from Ning et al.’s GWAS in UKB (Ncases = 1,303) and AF from Nielsen’s genome-wide association meta-analysis with UKB excluded (Ncases = 45,800). Summary statistics for lipid traits were derived from the Global Lipids Genetics Consortium (N = 1,320,016) and a UKB GWAS conducted by Richardson et al (N = 115,082). For inflammatory markers, summary statistics were sourced as follows: CRP from Said et al.’s GWAS involving over 500,000 individuals; tumor necrosis factor from The INTERVAL Study (N = 3,301); interferon-gamma from Ahola-Olli et al.’s GWAS (N = 7,701); IL-1 beta from deCODE (N = 35,559); CXCL10 from Gudjonsson et al.’s GWAS (N = 5,366), and IL-6 from a genome-wide meta-analysis by Zhao et al (N = 14,743). Summary statistics for cardiac magnetic resonance (CMR) imaging traits were sourced from four GWAS in UKB (Nrange = 16,920-41,505). Detailed information on all data sources used in this study is available in the Supplementary Methods or Supplementary Table 1.
Instrumental variable selection
Summary statistics for circulating IL-18 concentration were obtained from the Scallop Consortium’s genome-wide association meta-analysis of 90 Olink-measured proteins from 13 contributing cohorts containing 21,758 European individuals.34 We identified common (minor allele frequency >0.01) cis-genetic variants associated with circulating IL-18 concentration at genome-wide significance from within a 100kb flanking region of the IL-18 gene (hg19: chromosome 11: 112013974-112034840). Variants were then clumped using a linkage disequilibrium (LD) threshold r2 <0.1 and a distance threshold of 100kb, utilizing data from the 1000 Genomes Project European reference panel. This methodology selected five partially correlated cis-variants with the strongest association with circulating IL-18 concentration within each LD region from around the IL-18 gene, as described in Supplementary Table 2. Instrument strength was assessed through the calculation of an F-statistic and a mean F-statistic >10 was considered strong. For bidirectional MR analyses, we created instruments for each cardiometabolic disease by selecting genetic variants associated with each outcome at genome-wide significance from throughout the genome. We clumped these variants using an LD threshold of r² < 0.001 and a distance threshold of 10,000 kb. We relaxed our significance threshold for HF to 5 × 10⁻⁷ due to a scarcity of genome-wide significant variants.
Statistical analysis
Mendelian randomization analyses
Causal estimates from MR analyses rely on three key assumptions: (i) relevance, where the genetic instrument must be associated with the exposure; (ii) independence, where the instrument is not associated with any confounders of the exposure-outcome relationship; and (iii) exclusion restriction, where the instrument affects the outcome solely through its effect on the exposure.35 Random-effects inverse-variance weighted (IVW) models were used in primary MR analyses. For all significant associations, we conducted a range of sensitivity analyses to assess for possible violations of the underlying assumptions of MR, including weighted median and mode to assess for consistency of results, MR Egger to assess for directional pleiotropy and MR PRESSO to adjust for the effect of potential outliers.35 We repeated all analyses using generalized inverse variance weighted MR, which adjusts for residual pairwise LD among variants included in our IL-18 instrument.35
Mediation analysis
Following the methodology described by Woolf et al., we employed two-step cis-MR to calculate the proportion of the effect of downregulated IL-18 signaling on cardio-embolic stroke that was mediated by downstream cardiometabolic traits identified in primary analyses.36 This involved estimating the effect of IL-18 on each potential mediator, followed by assessing the impact of each mediator on the outcome while adjusting for IL-18. We then calculated the proportion of IL-18’s overall effect on cardioembolic stroke mediated by each outcome by dividing the indirect effect by the total effect. We used the weighted median estimator for the mediation analysis if heterogeneity was detected in the inverse-variance weighted MR estimate. Standard errors and the 95% CI for the proportion mediated were calculated using the delta method.37
Colocalization and LD check
Bayesian colocalization was used to determine if levels of circulating IL-18 and our primary outcomes share a single causal variant at the IL-18 gene locus, which may indicate a causal pathway between the exposure and the outcome.38 Colocalization involves calculating posterior probabilities (PP) for five hypotheses to assess the likelihood that the genetic variants influencing two traits at a specific locus are overlapping or distinct. H0 suggests that there is no genetic association with either trait; H1 and H2 suggest that only the first or second trait has a genetic association; H3 suggests that both traits are associated but have different causal variants; and H4 suggests that both traits are associated and share a single causal variant. Colocalization analyses rely on a well-powered outcome GWAS and sufficient variant coverage in the region of interest.38, 39 A posterior probability (PP) for H3 (distinct causal variant) and H4 (shared causal variant) less than 0.8, in the context of a statistically significant MR association, may signal an underpowered outcome GWAS to determine if the association is due to a shared causal variant or from a confounding variant in LD (i.e. horizontal pleiotropy). LD Check is an alternative method for co-localization when variant coverage is insufficient for Bayesian colocalization.39 LD Check estimates the LD between the sentinel IL-18 variant and the top 30 variants from the same genomic region in each outcome study. Causality with the sentinel IL-18 variant is suggested if at least one of the top 30 SNPs exhibits strong LD (r2 >0.8).39
Reporting and packages
Results from MR analyses are reported as odds ratios (OR) for binary outcomes and standard deviation (SD) difference for continuous outcomes per SD lower in genetically predicted IL-18 levels. Results from bidirectional MR analyses are reported as the effect of a one-unit increase in the log odds of genetic liability for each cardiometabolic disease. We applied the Benjamini-Hochberg method to control for the false discovery rate (FDR) across all phenotypes and report the unadjusted and FDR-adjusted p-values. Associations with a nominal p-value <0.05 and a Benjamini-Hochberg adjusted p-value >0.05 were deemed suggestive. Associations with a Benjamini-Hochberg adjusted p-value <0.05 were considered significant. Analyses were performed in R (v4.2.1; R Foundation for Statistical Computing) with the TwoSampleMR, MendelianRandomisation, Coloc and TwoStepCisMR packages.
Standard Protocol Approvals, Registrations, and Patient Consents
All contributing studies obtained appropriate informed consent from participants and received ethical approval before commencement. This study was conducted in adherence to the guidelines for Strengthening the Reporting of Observational Studies in Epidemiology – Mendelian randomization (STROBE-MR), as detailed in the Supplement.
Data availability
This study utilizes publicly available, summary-level GWAS data that can be accessed through the GWAS Catalogue (https://www.ebi.ac.uk/gwas/home) and the Integrative Epidemiology Unit (IEU) OpenGWAS project (https://gwas.mrcieu.ac.uk/) or specific cohort portals, including The Scallop Consortium (https://zenodo.org/records/2615265), The Centre for Statistical Genetics (https://csg.sph.umich.edu/willer/public/afib2018/), Pan-UK Biobank (https://pan.ukbb.broadinstitute.org/), The Human Genetics Amplifier (HuGeAMP) (https://kp4cd.org/node/169), The Usher Institute of Population Health Sciences and Informatics (https://datashare.ed.ac.uk/handle/10283/3209), The Cardiovascular Disease Knowledge Portal (https://cvd.hugeamp.org/downloads.html), The NBDC Human Database (https://humandbs.dbcls.jp/en/), The Complex Trait Genetics Lab (https://cncr.nl/research/summary_statistics/) and FinnGen (https://www.finngen.fi/en/access_results).
Results
Genetically downregulated IL-18 signaling and cardiometabolic biomarkers
Our instrument for lower genetically downregulated IL-18 signaling, composed of 5 variants, had a mean F-statistic of 124, indicating an appropriate average instrument strength, and explained 3.64% of the total variation of circulating IL-18 levels in the Scallop Consortium. Lower genetically predicted IL-18 levels were associated with several downstream biomarkers of IL-18 signaling, suggesting that our instruments represented a proxy for IL-18 signaling downregulation. The downstream biomarkers that were significantly associated with genetically proxied IL-18 levels included C-reactive protein (CRP) (SD change –0.02, 95% CI –0.03, –0.02, pfdr <0.001), tumor necrosis factor (SD change –0.19, 95% CI –0.25, – 0.14, pfdr <0.001), interferon-gamma (IFN-y) (SD change –0.15, 95% CI –0.22, –0.08, pfdr <0.001) and CXCL10 (SD change –0.13, 95% CI –0.16, –0.09, pfdr <0.001), but not IL-1β or IL-6 (Figure 2). Lower genetically predicted IL-18 levels were associated with higher diastolic blood pressure (SD change 0.02, 95% CI 0.02, 0.03, pfdr =<0.001) and body mass index (SD change 0.02, 95% CI 0.02, 0.01, pfdr =0.017), but had no association with lipid subfractions.
Genetically downregulated IL-18 signaling and cardiometabolic diseases
Lower genetically predicted IL-18 levels (1 SD decrease in IL-18 levels) were associated with 15% lower odds of cardio-embolic stroke (OR 0.85, 95% CI 0.79-0.92, pfdr <0.001, Figure 3) and 9% lower odds of peripheral arterial disease (OR 0.91, 95% CI 0.84-0.97, pfdr =0.027), but had no association with other stroke subtypes, coronary artery disease or aortic stenosis. In UKB, lower genetically predicted IL-18 levels were also associated with reduced odds of both all-cause HF (OR 0.83, 95% CI 0.75-0.91, pfdr <0.001) and dilated cardiomyopathy (OR 0.74, 95% CI 0.58-0.93, pfdr =0.031). When assessing related cardio-metabolic diseases, lower genetically predicted IL-18 levels were associated with lower odds of chronic kidney disease (CKD) (OR 0.91, 95% CI 0.84-0.99, pfdr=0.027), and type 2 diabetes mellitus (T2DM) (OR 0.97, 95% CI 0.95-0.99, pfdr = 0.027). Additionally, there was a trend towards an association between lower genetically predicted IL-18 levels and reduced risk of AF (OR 0.94, 95% CI 0.89-0.99, pfdr = 0.054), which did not maintain significance following false discovery rate adjustment. Associations for HF were successfully replicated in a non-overlapping sample of American cohort studies (OR 0.85, 95% CI 0.75-0.95, p =0.004, Supplementary Table 3). For AF, results were directionally similar but did not reach statistical significance in the UK Biobank (UKB) (OR 0.96, 95% CI 0.92-1.00, p = 0.061).
In colocalization analyses between IL-18 and AF, peripheral arterial disease, HF, cardio-embolic stroke, CKD and T2DM, the posterior probabilities of H3 and H4 were less than 0.8 and this along with the low minimum p-values in the IL-18 gene region in each of the outcome summary statistics implies suboptimal power for colocalization, the presence of confounding by LD or multiple causal variants within the gene region. We subsequently conducted an LD Check analysis and obtained the LD correlation between the sentinel variant (rs5744249) from the IL-18 gene locus and the top 30 variants within 100kb of rs5744249 with the lowest p-values from the cardio-embolic stroke, AF, peripheral arterial disease, HF, CKD and T2DM summary statistics. At least one top cis-variant from each outcome dataset, except for peripheral arterial disease, exhibited strong LD (r2 > 0.8) with the sentinel variant. This suggests potential causal associations between IL-18 and cardio-embolic stroke, AF, HF, CKD and T2DM
Genetically downregulated IL-18 signaling and cardiac magnetic resonance imaging traits
In UKB, there were directionally consistent improvements in all measures of cardiac structure and function with lower genetically predicted IL-18 levels as displayed in Figure 3, with the exception of left and right ventricular ejection fractions. Following adjustment for multiple testing, lower genetically predicted IL-18 levels were associated with a reduction in left atrial size (SD change –0.02, 95% CI –0.03, –0.00, pfdr = 0.022) and left ventricular end-systolic volume (SD change –0.08, 95% CI –0.13, –0.02, pfdr = 0.039).
Genetically downregulated IL-18 signaling and potential adverse effects
Lower genetically predicted IL-18 levels were associated with a longer parental lifespan, an indirect measure of longevity (SD change 0.02 years, 95% CI 0.01-0.04, pfdr = 0.005), and reduced risk of a range of autoimmune diseases, including atopic dermatitis (OR 0.86, 95% CI 0.83-0.90, pfdr <0.001), psoriasis (OR 0.84-0.89, pfdr <0.001), asthma (OR 0.90, 95% CI 0.86-0.94, pfdr<0.001), and rheumatoid arthritis (OR 0.85, 95% CI 0.81-0.90, pfdr < 0.001, Figure 4). There was also a trend toward an association between lower genetically predicted IL-18 levels and reduced risk of ulcerative colitis (OR 0.88, 95% CI 0.79-0.98, pfdr = 0.051), which did not maintain significance following false discovery rate adjustment. While lower genetically predicted IL-18 levels were not significantly associated with an increased risk of infection, it was associated with higher odds of lung cancer (OR 1.15, 95% CI 1.11-1.20, pfdr <0.001) and multiple sclerosis (OR 1.17, 95% CI 1.03-1.32, pfdr = 0.041).
Bidirectional relationship between cardiometabolic diseases and IL-18 levels
In reverse MR analyses, genetic liability to cardio-embolic stroke was not significantly associated with circulating levels of IL-18 (β representing the effect of one unit increase in the log odds of genetic liability for cardioembolic stroke –0.03, 95% CI –0.08-0.02, p = 0.267). Similarly, circulating levels of IL-18 were not affected by genetic liability to AF (β 0.00, 95% CI –0.03-0.04, p 0.879), HF (β –0.03, 95% CI –0.22-0.16, p =0.743), CKD (β 0.01, 95% CI –0.07-0.06, p = 0.821), peripheral arterial disease (β 0.07, 96% CI –0.26-0.39, p = 0.686), or higher BMI (β 0.05, 95% CI –0.01-0.12, p = 0.104).
Mediation analyses
We performed two-step network MR mediation analyses to quantify the extent to which downstream cardiometabolic traits mediate the relationship between IL-18 signaling and cardio-embolic stroke. We observed that genetic liability to AF accounted for 25% of the total effect of lower genetically predicted IL-18 levels on cardio-embolic stroke (proportion-mediated 25.1%, 95% CI 22.4-27.8%, p <0.001, Supplementary Table 4).
Genetic liability to all-cause HF accounted for 5.6% of the total effect (proportion-mediated 5.3%, 95% CI 1.7-8.8%, p =0.003), while genetic liability to CKD mediated 2.8% of the total effect of downregulated IL-18 signaling on cardio-embolic stroke (proportion-mediated 2.8%, 95% CI 0.1-5.7%, p =0.054). However, BMI had a more modest impact, mediating 1.61% of the total effect (proportion-mediated 1.6%, 95% CI –2.0-5.2%, p =0.383), while genetic liability to T2DM had little impact, mediating only 0.8% of the total effect of lower genetically predicted IL-18 levels on cardio-embolic stroke (proportion-mediated 0.8%, 95% CI –2.8-4.4%, p =0.657
Sensitivity analyses
The MR estimates from generalized IVW, weighted median, weighted mode, MR Egger, and MR PRESSO are reported in Supplementary Tables 5.1-5.24. No heterogeneity was observed across all MR analyses, as indicated by Cochrane’s Q and Rucker’s Q statistics. Estimates were consistent with primary results when using generalized IVW MR to model residual LD, and directionally similar across weighted median and weighted mode methods. The p-values for the MR Egger intercept test and the global MR PRESSO test were not significant, suggesting an absence of marked pleiotropy.
Discussion
Leveraging data from large-scale genomic, proteomic and imaging studies, we have demonstrated new and important evidence linking IL-18 signaling to HF, peripheral arterial disease, atrial cardiomyopathy, and cardio-embolic stroke. Lower genetically predicted IL-18 levels were associated with reduced concentrations of several downstream inflammatory cytokines and exhibited protective effects on cardiac structure and function. Genetic liability to AF and HF mediated a significant proportion of the risk-lowering effect of IL-18 signaling on cardioembolic stroke. Lower genetically predicted IL-18 levels were associated with a favorable metabolic profile overall, with a lower risk of T2DM, CKD, autoimmune disease, and enhanced longevity.
The present data are supported by several lines of evidence from experimental models, observational studies, and emerging data from RCTs implicating the NLRP3-inflammasome and IL-18 activation in HF and AF. The NLRP3-inflammasome is activated by pro-inflammatory mediators, such as IL-1β and TNF-alpha, damage-associated molecular patterns (DAMPs) secondary to ischemia-reperfusion injury, and angiotensin II over-expression from renin-angiotensin activation.4, 21, 40 IL-18 activation ensues secondary to NLRP3 inflammasome formation in leukocytes, endothelial cells, fibroblasts, and cardiomyocytes.21 The binding of IL-18 to its receptor triggers the recruitment of tumor necrosis factor receptor-associated factor-6 (TRAF6), and the activation of nuclear factor Kappa B (NF-κB) and p38 mitogen-activated protein kinase (MAPK).21 This cascade results in the synthesis of secondary inflammatory mediators, including IFN-γ and TNF-α.30 In-vitro and animal models have demonstrated that disrupting the IL-18 signaling pathway can reduce myocardial hypertrophy, improve contractile function, and decrease fibrosis.21, 22 IL-18 mediates β-adrenergic receptor stimulation-induced macrophage infiltration, cardiac inflammation, and fibrosis.20 IL-18 also triggers apoptosis in microvascular endothelial cells through activation of NF-κB and upregulates the expression of osteopontin, fibronectin, and matrix metalloproteinase 2 (MMP-2) in human cardiac fibroblasts.21 In humans, IL-18 levels increase in proportion to disease progression in both HF23 and AF.41, 42 In the Atherosclerosis Risk in Communities (ARIC) study circulating IL-18 in mid-life was associated with future risk of HF.24 Similarly, Henry et al.’s meta-analysis of four population-based studies demonstrated an association between IL-18 levels and risk of incident HF.27
Our results suggest that the IL-18 signaling pathway is a viable treatment target in AF, HF and cardio-embolic stroke. Our data were derived from non-overlapping cohorts and demonstrated biologically consistent results, implicating IL-18 at the center of multiple downstream pro-inflammatory pathways, cardiac remodeling, AF, HF, and ultimately cardio-embolic stroke. Our two-step cis-MR analysis suggests that the reduced risk of cardioembolic stroke associated with downregulated IL-18 signaling is likely mediated through its protective effects against lifetime cardiac remodeling, cardiomyopathy, and AF. Our results are supported by Schmidt et al.’s recent cis-MR study which showed that targeting the IL-18 receptor is associated with four left ventricular CMR traits and risk of dilated cardiomyopathy.43 Furthermore, in their candidate gene study, Wang et al. demonstrated that three variants associated with IL18 gene expression increased risk of AF and left atrial enlargement.44
The global burden of stroke, HF and AF is increasing.45, 46 There is a pressing need to identify and develop disease-modifying therapies that prevent stroke and slow disease progression in both HF and atrial cardiopathy. In atherosclerosis, colchicine and canakuinumab have both shown benefit in reducing major adverse cardiovascular events (MACE) in coronary artery disease.1, 2 Although colchicine did not reduce the risk of MACE after non-cardioembolic stroke, a 20% reduced risk was observed in an on-treatment analysis suggesting a possible biological effect.3 Apart from the treatment of atherosclerosis, there is also increasing interest in targeting inflammation in HF and AF. Whilst there is some evidence that colchicine reduces the risk of postoperative AF following cardiac surgery and decreases AF recurrence after pulmonary vein isolation or ablation47, a recent RCT was neutral.12 Currently, no licensed anti-inflammatory therapies have been proven to prevent hard outcomes in HF. RCTs investigating the potential of TNFα inhibition were neutral.10, 11 Anakinra, an anti-IL-1 monoclonal antibody, improved several surrogate markers of cardiac function but failed to significantly reduce hospitalization for HF or death.48 While genetic studies have identified the IL-6 signaling pathway as a potential druggable target for AF, they do not support a causal role in HF or cardio-embolic stroke.16 Consequently, future experimental work and RCTs should now focus on the potential benefit of IL-18 pathway suppression in the treatment of HF and the prevention of cardio-embolic stroke.
Our results suggest that IL-18 inhibition would likely have a favourable cardio-metabolic profile. Downregulated IL-18 signaling reduced risk of CKD and T2DM, which are key risk factors for cardiovascular disease. However, we did observe that IL-18 suppression was associated with marginal increases in diastolic blood pressure and BMI while having minimal effect on lipid fractions and systolic blood pressure. The clinical relevance of these findings will need to be explored in future studies. We show that downregulated IL-18 signaling is associated with lower risk of atopic disease, which supports the rationale for ongoing clinical trials of IL-18 signaling-blocking therapies in atopic dermatitis. Promising agents for potential repurposing in cardiovascular disease include GSK1070806 and Camoteskimab, which are human monoclonal antibodies targeting IL-18, as well as tadekinig α, a recombinant human IL-18 binding protein that sequesters IL-18. Immunomodulatory therapies carry potential risks, notably with respect to host response to infections and malignancy. Our study shows that downregulated IL-18 signaling increases the risk of lung cancer, but not other cancers or infection. Most importantly, we show that reduced IL-18 signaling is associated with enhanced lifespan, indicative of a net positive effect on all-cause mortality.
Previous data from experimental models and some epidemiological studies suggested that IL-18 was associated with atherosclerosis and coronary disease.26, 49 Apart from a modest association with peripheral arterial disease that did not have LD Check support, we did not find any compelling evidence that IL-18 signaling is causally associated with atherosclerotic endpoints. On the contrary, prior MR studies have implicated IL-6 signaling in coronary artery disease and atherosclerotic stroke.16 Together with data from CANTOS suggesting the potential benefit of IL-1β-IL-6 axis suppression,2 our analysis supports the rationale of ongoing trials focusing on the high potential of IL-6 pathway inhibition in atherosclerosis.17
Our study has several strengths. The MR study design minimizes bias from reverse causation and environmental confounders, while cis-MR provides an evaluation of the effects of directly perturbing a drug target on disease outcomes. We leveraged large multi-omic datasets to uncover new insights into how downregulated IL-18 signaling influences cardiovascular disease. Moreover, we developed a robust genetic cis-instrument that mirrors the effects of reduced IL-18 signaling on downstream biomarkers and conducted a comprehensive set of mediation and sensitivity analyses to improve confidence in our findings. We were also able to validate our findings for HF in independent cohorts. We acknowledge some limitations. Although we utilized large genomic consortia for outcome summary statistics, our analyses may have been underpowered to detect associations with rarer phenotypes. Effect sizes derived from drug-target MR studies reflect lifelong perturbation of the drug target rather than the short-term treatment durations typically seen in clinical practice, potentially leading to significant differences in the observed benefits. Two-sample MR methods assume a linear relationship between genetically proxied exposures and outcomes. In cases of non-linearity, the MR estimates reflect a population average effect rather than capturing potentially varying effects across different levels of exposure. Similarly, in clinical practice, the biological response to drugs often requires surpassing a specific therapeutic dose, a detail that two-sample MR studies might not capture due to their reliance on average effects. Two-step cis-MR relies on several assumptions including linearity, and may yield unreliable estimates when the mediator or outcome is not continuous, potentially impacting the accuracy of our effect size estimates and the interpretation of IL-18 signaling mediation on cardio-embolic stroke through AF, HF, CKD, and T2DM. Our analyses were limited to individuals of European ancestry to minimize confounding by population stratification and the potential impact of downregulated IL-18 signaling on risk of cardio-vascular disease in non-European ancestries remains to be explored. Finally, our MR estimates reflect the impact of downregulated IL-18 signaling on incident disease, and it remains unclear how appliable these findings are to individuals with established disease, who are more likely to be prioritized for treatment and inclusion in clinical trials.
In conclusion, this study used large-scale multi-omic data to explore the effects of genetically downregulated IL-18 signaling on cardiovascular disease. We provide compelling data which suggests the IL-18 pathway is causally implicated in the evolution of cardiomyopathy, from structural remodelling through to an increased risk of HF and cardio-embolic stroke. These data strongly suggest that IL-18 represents a viable target for the prevention and treatment of stroke and heart failure, with a favorable cardiometabolic and autoimmune profile, that warrants further exploration in clinical trials.
Sources of funding
PJK has received funding from the Irish Health Research Board. DG is supported by the British Heart Foundation Centre of Research Excellence (RE/18/4/34215) at Imperial College London.
Disclosures
The authors have no relevant conflicts of interest to disclose.
Abbreviations
- HF
- Heart failure
- NLRP
- NOD-like receptor family pyrin domain-containing
- AF
- Atrial fibrillation
- IL-18
- Interleukin-18
- CMR
- Cardiac magnetic resonance
- CRP
- C-reactive protein
- IFN-γ
- Interferon-gamma
- CVD
- cardiovascular disease
- MR
- Mendelian randomization
- GWAS
- Genome-wide association study