Integrative mendelian randomization approaches for therapeutic target prioritisation in immune-mediated diseases.

Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Sci Rep . 2026 Mar 3;16:11851. doi: 10.1038/s41598-026-41818-3 Search in PMC Search in PubMed View in NLM Catalog Add to search Integrative mendelian randomization approaches for therapeutic target prioritisation in immune-mediated diseases Maria K Sobczyk Maria K Sobczyk 1 MRC Integrative Epidemiology Unit, University of Bristol, Oakfield House, Oakfield Grove, Bristol, BS8 2BN UK Find articles by Maria K Sobczyk 1, ✉ , Tom R Gaunt Tom R Gaunt 1 MRC Integrative Epidemiology Unit, University of Bristol, Oakfield House, Oakfield Grove, Bristol, BS8 2BN UK 2 NIHR Biomedical Research Centre at the University Hospitals Bristol NHS Foundation Trust and the University of Bristol, Bristol, UK Find articles by Tom R Gaunt 1, 2 Author information Article notes Copyright and License information 1 MRC Integrative Epidemiology Unit, University of Bristol, Oakfield House, Oakfield Grove, Bristol, BS8 2BN UK 2 NIHR Biomedical Research Centre at the University Hospitals Bristol NHS Foundation Trust and the University of Bristol, Bristol, UK ✉ Corresponding author. Received 2024 May 3; Accepted 2026 Feb 23; Collection date 2026. © The Author(s) 2026 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ . PMC Copyright notice PMCID: PMC13065838  PMID: 41775781 Abstract Immune-mediated diseases (IMD) encompass a wide range of autoimmune and inflammatory disorders with aetiology related to immune system dysfunction, signifying a disease area with great potential for drug repurposing. In this study, we employed the genetically informed Mendelian Randomization (MR) method with two distinct exposure types: immune blood cell abundance and protein quantitative trait loci (pQTL) to validate and repurpose 834 drug targets which have been investigated for IMD treatment. Utilizing two-sample MR, we first established causal relationships between major peripheral immune cell types and 14 IMD. Robust associations, particularly with eosinophils, were confirmed across diseases such as asthma, eczema, sinusitis, and rheumatoid arthritis, revealing 59 high-confidence relationships. Intragenic variants associated with causal immune cell types were then extracted to create instruments for 371 existing IMD drug targets (“intermediate trait” MR). In parallel, we leveraged four large blood plasma protein QTL datasets to obtain complementary instruments for 361 targets (“pQTL” MR). In the intermediate trait MR analysis, we identified 811 gene-IMD associations (p-value < 0.05; 137 pairs below Bonferroni-adjusted p-value threshold), 169 of which were supported by strong colocalisation evidence (PP H4 ≥ 0.8). In the pQTL MR analysis, we similarly found 841 protein-IMD associations (p-value < 0.05; 90 pairs below Bonferroni-adjusted p-value threshold), 83 of which were confirmed with colocalization. Comparison with a list of approved drugs indicated low sensitivities across disease outcomes for both exposure types (intermediate trait MR: 0.49 ± 0.23 SD, pQTL MR: 0.28 ± 0.12 SD). Drug targets identified in the pQTL and intermediate trait MR analyses show limited overlap (13% at nominal p-value and 36% at Bonferroni-adjusted p-value threshold), presenting a comprehensive source of drug repurposing opportunities when the two approaches are combined. Supplementary Information The online version contains supplementary material available at 10.1038/s41598-026-41818-3. Keywords: Immune-mediated disease, Immune cells, Protein QTL, Drug target prioritisation, Mendelian randomization, Molecular epidemiology Subject terms: Medical genetics, Data integration, Data mining, Immunological disorders, Biomarkers, Drug development, Epidemiology Introduction Immune-mediated diseases (IMDs) arising from dysregulated immune responses affect up to 10% of the global population, posing a major health burden 1 , 2 . Both innate and adaptive arms of immunity contribute to the pathogenesis of IMD via altered immune cell frequencies and activation states 3 . Dysfunctional immune cells, particularly T cells and B lymphocytes, are central to the pathogenesis of these disorders, leading to the production of inflammatory mediators such as cytokines and autoantibodies. Despite the substantial burden of autoimmune and inflammatory diseases, the pharmaceutical arsenal remains limited, owing to the complexity of the dysfunctional immune cascade underlying these heterogeneous conditions 4 , 5 . The absence of disease-specific biomarkers and need for chronic therapy compound the challenges of developing targeted agents. Therefore, given shared immune pathogenic pathways across IMD, there is substantial interest in repurposing of drugs in this disease category. Drug repurposing involves demonstrating the efficacy of a drug previously tested for safety in one medical condition, for a different indication 6 . For example, adalimumab, an anti-tumour necrosis factor (TNF) monoclonal antibody has been initially approved for treatment of rheumatoid arthritis but has since been extended for use in psoriasis, ankylosing spondylitis, Crohn’s disease, and ulcerative colitis 7 . The prioritization of therapeutic targets can potentially be supported by genetic epidemiological methods, including Mendelian randomization (MR). Since the mid-2000s, genome-wide association studies (GWAS) have uncovered hundreds of disease-associated loci implicating immune-related genes 8 . Uncovering specific proteins driving IMD holds promise for new targeted therapeutics. Human genetic support can more than double the approval odds of drug target in preclinical development, as well as progression along subsequent phases of clinical trials 9 , 10 . In general, the strongest support is provided by variants with impact on protein-coding sequence of gene as they offer the least ambiguous mapping to a drug target, unlike intergenic and intronic variants 11 . MR is an approach that can help prioritise targets in the drug development pipeline by explicitly modelling causal relationships. MR uses GWAS-derived genetic variants as instrumental variables (IV) to study the lifetime effects of genetic perturbations of drug targets. This allows for the examination of causal effects on a chosen outcome of interest, such as any of IMD 12 . By leveraging the natural random assortment of genetic material during meiosis, MR provides a powerful framework to assess causality, mitigating issues of reverse causation and confounding that often plague observational studies 13 . Reliable causal inference in Mendelian Randomization (MR) relies on three core assumptions: (1) genetic variants must be strongly associated with the exposure (relevance), (2) these associations must not be confounded by external factors (independence), and (3) variants must affect the outcome only through the exposure, not via alternative pathways (exclusion restriction). We can identify two main approaches in MR analyses looking at causal support for a given drug target against a disease indication. While the outcome GWAS used involve primarily disease incidence, the exposure GWAS can use genetic variants associated with gene expression (either messenger RNA or protein) in disease-relevant tissue or a downstream biomarker or clinical risk factor (here referred to as “intermediate trait”) as instruments 14 . In the first approach, variants related to two molecular phenotypes: protein and mRNA abundance are known as protein and expression quantitative trait loci (QTL), respectively. Typically, protein QTL are preferable given the closer relationship of protein levels to clinical phenotypes and mechanism of drug action which usually targets proteins 15 . Previous pQTL-based MR studies have established good methodological practice 16 , 17 , which we follow here, but acknowledge that the pQTL approach continues to undergo rapid technical advancement 18 .An important constraint in the analytical application of pQTLs is the potential for confounding arising from artefactual effects caused by differential capture affinity driven by protein-altering variants (PAV) rather than biologically meaningful changes in protein concentration 19 . PAVs, such as missense mutations, can change the structure or epitope of a protein and this can interfere with probe binding in assays used to quantify protein levels, leading to the measured protein concentration reflecting changes in binding affinity rather than true biological abundance. To address these biases, it is recommended to explore multiple independent pQTL datasets, incorporating different assays for the measurement of protein levels whenever possible 20 . In the intermediate trait MR, exposure variants are extracted from GWAS for relevant disease risk factors or biomarkers with available lead variant(s) located in the drug target gene of interest. This approach offers confirmation that the genetic variant indeed influences the clinical outcome of interest 14 . Previously employed intermediate traits in MR include HbA1c for proxying effects of GLP-1 agonists 21 , a class of antidiabetic medication, CRP for mimicking the effect of IL-6 signalling 22 and height for the effect of NPR2 and NPR3 signalling on cardiovascular disease 23 . However, limited use has been made of this approach in evaluating drug targets for IMD. Here, we propose a new intermediate trait category for use in proxying drug targets in IMD: immune cell abundance. There is substantial animal model, observational and MR evidence 24 – 29 implicating immune cell dysregulation in the pathogenesis of immune-mediated disorders, ranging from roles of T H 2 cells, eosinophils, and neutrophils in allergic conditions such as asthma 30 , 31 and eczema 32 , 33 to expansion of certain T helper lymphocyte and B lymphocyte lineages in autoimmune diseases such as systemic lupus erythematosus (SLE) 34 , multiple sclerosis 35 and rheumatoid arthritis 5 . Furthermore, approved drug targets for IMD are usually classed as immunosuppressants or immunomodulators, which alter the balance of immune blood cells in their course of on-target action. For example, corticosteroids reduce the number of various immune cells, including lymphocytes and eosinophils in the blood 36 , while methotrexate promotes monocyte apoptosis 37 and mycophenolate mofetil suppresses T and B lymphocyte proliferation 38 . Here, we implement a multi-pronged MR strategy exploiting genetic instruments from 23 large-scale GWAS resources. Firstly, we examine causal connections between blood cell immunotypes and 14 IMDs. Composition of specialized white cells in peripheral blood underpins immunocompetence, making these genetics-derived perturbations clinically relevant. We present a comprehensive examination of bidirectional relationships between immune cell counts and IMD. In the next stage of our MR analyses, we prioritize genes as potential therapeutic targets for IMD using intragenic variants from immune cell GWAS established as causal, using a novel intermediate trait category in MR studies of autoimmune and inflammatory disease. As proteins constitute ultimate drug-actionable targets, we also employ an alternative approach leveraging blood serum protein QTLs (pQTLs) to instrument targets. To check the robustness of our findings, we include pQTLs derived in 4 largest independent studies so far, which use two different technical assays. Lastly, we confirm all our MR findings with the colocalisation sensitivity analysis. By evaluating concordance and complementarity of these MR approaches, we aim to validate known and discover novel drug target indications for immune-mediated disease. Integrative evaluation of previously implicated genes and their repurposing opportunities using existing drugs can accelerate translation of genetic insights to the clinic. To our knowledge, this is the first study to systematically apply immune-cell–informed MR to prioritize therapeutic targets across multiple IMDs. Materials and methods Immune blood cell exposures We gathered a comprehensive selection of count-based peripheral immune cell GWAS (Supplementary Table 1) derived from complete blood count, which included counts of basophils, eosinophils, granulocytes, lymphocytes, monocytes, neutrophils, myeloid white cells, total white blood cells and also their relative percentages. The GWAS were conducted using mostly the UK Biobank data of 132,959 − 456,785 European ancestry individuals 39 – 41 . Immune-mediated disease outcomes We compiled a selection of 14 IMD GWAS (Supplementary Table 2): ankylosing spondylitis (AS) 42 , asthma 43 , 44 , chronic sinusitis 42 , eczema (atopic dermatitis) 42 , 45 , eosinophilic esophagitis 46 , psoriasis 47 , juvenile idiopathic arthritis 48 , rheumatoid arthritis 49 , inflammatory bowel disease (IBD) 50 , Crohn’s disease 50 , ulcerative colitis 50 , multiple sclerosis (MS) 51 , systemic lupus erythematosus 52 , type 1 diabetes 53 . We employed European-ancestry GWAS with the highest power (as judged by sample size and number of top loci) whenever full summary statistics were available. The biggest GWAS for two IMD: asthma 44 and eczema 45 included a significant contribution from UK Biobank ( N > 300,000). As two-sample MR estimates can be biased by population sample overlap between exposure and outcome GWAS 54 , for these traits we included replicate, independent GWAS 42 , 43 to confirm that the relationships detected in the two-sample MR analyses involving the immune cell exposures and Olink pQTLs ( see below , also derived from UK Biobank). Protein quantitative trait loci (QTL) exposures Blood plasma pQTL data was obtained from large European-ancestry studies with protein abundance measured using two different affinity-based technologies (Supplementary Table 3): aptamer-based SomaScan ver 4 including ~ 4,500 protein targets (ARIC 55 , deCODE 56 , FENLAND 57 and antibody-based Olink assay with ~ 3,000 targets (UK Biobank 58 . Selection of genetic instruments To identify genetic instruments for each exposure, we selected SNPs demonstrating a robust association at the genome-wide significance threshold ( p -value < 5 × 10 − 8 ). Subsequently, we conducted clumping of these SNPs to ensure that the linkage disequilibrium (LD), measured by r 2 , was maintained at less than 0.001 within a 10 Mbp range in the 1000 Genomes European panel 59 . This step aimed to assure that we are not double counting the effects of variants tagging the same underlying causal variant, leading to overly precise confidence intervals. The clumping process was executed using plink version 1.943 60 , facilitated by the ld_clump function in the ieugwasr R package (available at https://mrcieu.github.io/ieugwasr ). In each MR analysis, we extracted and harmonized genetic variant associations for the outcome trait. Subsequently, mean F-statistics and R 2 were computed to assess potential weak instrument bias, and no weak instruments with F-statistics < 10 were used in the study. For intermediate trait instruments, prior to clumping, we obtained all genome-wide significant ( p -value < 5 × 10 − 8 ) hits overlapping genes in the immune cell GWAS showing a robust, high-confidence association with immune-mediated disease. Using Variant Effect Predictor (VEP) 61 annotations, we subsequently selected the variants with the highest priority annotation (reflecting intragenic location and expected severity of functional consequence) available for each gene, prioritising use of missense and protein altering variants whenever possible (Supplementary Table 4). For pQTL exposures we included the following additional steps due to nature of molecular phenotypes resulting in high risk of bias from horizontal pleiotropy 62 . First of all, we only used cis -pQTLs, defined as positioned maximum 1 Mbp away from gene’s TSS and discarded trans-pQTLs as cis- variants are less likely to be affecting the outcome through expression of multiple genes 63 . Secondly, we flagged SNPs (or correlated proxies at r 2 > 0.6) with potentially epitope-altering mutations (such as: “stop_gained”, “stop_lost”, “frameshift_variant”, “start_lost”, “inframe_insertion”, “inframe_deletion”, “missense_variant”, “protein_altering_variant”) which could result in artefactual variation in protein levels. However, we did not remove them due to their high frequency in the exposure datasets and minority of them (~ 25%) having been estimated to result in false positive pQTLs 64 . Two-sample mendelian randomization Two sample MR analyses (Wald ratio for single-SNP instruments or inverse variance weighted regression, IVW, for multi-SNP instruments) were carried out using TwoSampleMR R package 65 . For instances where the instrument comprised three or more SNPs, we conducted sensitivity analyses incorporating MR-PRESSO which can identify and adjust for pleiotropic outlier variants 66 , weighted median, weighted mode, and MR-Egger methods to ensure the consistency of estimates. Additionally, we computed I 2 and Cochran’s Q to investigate the variability of estimates among the variants included in each instrument. The MR-Egger intercept test was applied to assess the potential impact of directional pleiotropy on our results 67 . To evaluate the NOME assumption for MR-Egger, we computed the I 2 GX statistic as an indicator of potential attenuation bias 68 . In the case of all MR analyses evaluating the effect of immune cells on IMD, where we had strong prior evidence of reverse relationships in which IMD may effect immune cell abundance 5 , we conducted bidirectional MR. In addition, to examine whether the potential causal effects were independent of the outcome influencing the exposure, we also included Steiger filtering[30] in all our analyses. Colocalisation We employed the Bayesian colocalisation method coloc 69 to assess whether the nominally significant (p-value < 0.05) MR effects observed for drug targets were indicative of causation or potentially confounded by linkage disequilibrium (LD) 70 . Among the pQTL-instrumented MR analyses, the only three studies (ARIC, deCODE, UK Biobank) with published full summary statistics were used to run coloc with default prior probabilities: p 1 , p 2 = 10 −4 and p 12 = 10 −5 . A 100 kb window around each SNP in the instrumental variable was used to define the colocalisation region and analysis was conducted if at least 50 variants were identified in the specified window. If fewer than 50 SNPs were found in the window, no robust coloc analysis could be done due to inability to capture the local LD structure necessary for inference and resulting instability of estimates 71 . When interpreting the results, posterior probabilities PP H4 > 0.8 were determined to provide robust support for causality. Drug target validation and repurposing Protein target-disease indication pairs for drugs with approved (globally), in active development (preclinical and in trials) and ceased (no development recorded for > 1 year) status were downloaded from manually curated Pharmaprojects database ( https://www.citeline.com/en/products-services/clinical/pharmaprojects ), separately for every IMD included, on 29th August 2023 (Supplementary Table 14). Pharmaprojects has been a continuously updated industry-standard reference for pharmaceutical industry for over 40 years 72 . In cases of multiple drugs targeting the same protein, we selected the drugs with the most advanced development status. In addition, a manually curated list of 4,723 genes involved in immune response derived from the ImmPort 73 project was accessed via InnateDB 74 . Results Bidirectional causal effect of immune blood cell counts on IMD We first wanted to establish robust causal associations between peripheral immune cells and immune-mediated disease using Mendelian Randomization (Supplementary Fig. 1). We ran a number of sensitivity MR analyses, in addition to the baseline IVW analyses involving all genetic instruments (Supplementary Table 5). IVW was also run excluding the major histocompatibility complex (MHC) region located at chromosome 6, between 28.5 Mb–33.5 Mb. This region is prone to skewed results due to complex LD patterns resulting in potential for including multiple correlated variants 75 . In addition, we also conducted outlier-adjusted MR-PRESSO test including and excluding the MHC region (Supplementary Table 9). The nominally significant results (p-value < 0.05) across the 4 analyses were intersected to produce a high-confidence set of 59 immune cell phenotype-> IMD associations (37 associations below Bonferroni-adjusted p-value threshold of 1.8 × 10 − 4 , Supplementary Table 5). For two IMD outcomes (asthma and eczema) we included an independent replicate GWAS due to UK Biobank-derived sample overlap between exposure and outcome GWAS when using the largest available IMD GWAS. Across the duplicate outcome GWAS, we found a directionally robust pattern of associations, regardless of sample overlap presence (Fig. 1 A) so in the subsequent analyses we focussed on results derived from the biggest, single representative GWAS for each disease. Fig. 1. Open in a new tab (A ) A complete-linkage clustered matrix of the central odds-ratio estimates of the associations between peripheral immune cells (x-axis) and immune-mediated disease (IMD y-axis). High-confidence associations with p-values of: < 0.05 (*), < 0.005 (**), < 0.0005 (***) in the IVW analysis excluding the MHC region are highlighted with an asterisk(s). ( B ) Heatmap of the central beta estimates of the associations between immune-mediated disease (IMD, y-axis) and immune cell counts (x-axis). High-confidence associations with p-values of: < 0.05 (*), < 0.005 (**), < 0.0005 (***) in the IVW analysis excluding the MHC region are highlighted with an asterisk(s). Multiple sclerosis is missing from the heatmap as no instrument outside of the MHC region was available for this exposure. Overall, we discovered a strong positive effect of eosinophil phenotypes (count, percentage of white cells, percentage of granulocytes, sum eosinophil basophil counts) on genetic susceptibility to multiple disease (Fig. 1 A). These included disease with atopy component: asthma (OR = 1.72, CI 95% =1.6–1.85, p-value = 3.9 × 10 − 46 ), eczema (OR = 1.25, CI 95% =1.17–1.33, p-value = 1.7 × 10 − 11 ), sinusitis (OR = 1.57, CI 95% =1.45–1.69, p-value = 5.1 × 10 − 31 ) and eosinophilic esophagitis (OR = 1.6, CI 95% =1.23–2.1, p-value = 4.8 × 10 − 4 ), but also rheumatoid arthritis (OR = 1.36, CI 95% =1.18–1.58, p-value = 3.7 × 10 − 5 ), juvenile idiopathic arthritis (OR = 1.51, CI 95% =1.24–1.84, p-value = 3.7 × 10 − 5 ), type 1 diabetes (OR = 1.47, CI 95% =1.26–1.72, p-value = 1.6 × 10 − 6 ) and ulcerative colitis (OR = 1.22, CI 95% =1.06–1.4, p-value=5 × 10 − 3 ). The IVW estimates given correspond to 1 SD change in the exposure, here eosinophil percentage of white cells. Using MR, we also assessed the reverse causal pathway: from IMD to immune cell counts (Fig. 1 B). There, we found some evidence that the relationship between eczema, asthma, sinusitis and eosinophil phenotypes is bidirectional (Supplementary Table 10). One unit increase in the log odds of eczema was associated with increase of eosinophil percentage of white cells (β = 0.103, CI 95% =0.058–0.149, p-value=8 × 10 − 6 ). For asthma and sinusitis, that direction of relationship was not robustly confirmed when Steiger filtering was applied (Supplementary Table 8, Supplementary Table 13), which suggested that the main causal direction was from eosinophils to asthma and sinusitis. We found a positive relationship between neutrophil counts (also neutrophils percentage of white cells) and bowel disease: IBD (OR = 1.29, CI 95% =1.11–1.51, p-value = 9.4 × 10 − 4 ) and Crohn’s disease (OR = 1.44, CI 95% =1.19–1.74, p-value = 1.8 × 10 − 4 ). On the other hand, we found a negative causal relationship between neutrophil percentage of granulocytes and atopic disease (asthma: OR = 0.62, CI 95% =0.55–0.71, p-value = 6.4 × 10 − 13 , eczema: OR = 0.8, CI 95% =0.73–0.88, p-value = 8.8 × 10 − 6 , sinusitis: OR = 0.65, CI 95% =0.58–0.74, p-value = 1.8 × 10 − 12 ), as well as rheumatoid arthritis (OR = 0.76, CI 95% =0.64–0.89, p-value = 1.1 × 10 − 3 ). Similar to eosinophils, bidirectional relationship was only robustly supported for eczema (β=-0.088, CI 95% =[-0.132, -0.044], p-value = 7.5 × 10 − 5 ), but not asthma and sinusitis. Negative effect of immune cell type abundance on IMD was also found for lymphocytes. Lymphocyte percentage of white cells was negatively associated with the odds of asthma (OR = 0.89, CI 95% =0.83–0.95, p-value = 4.5 × 10 − 4 ), eczema (OR = 0.83, CI 95% =0.75–0.93, p-value = 7.9 × 10 − 4 ) and Crohn’s disease (OR = 0.74, CI 95% =0.59–0.93, p-value = 0.01), while lymphocyte count had a negative association with genetic susceptibility to type 1 diabetes (OR = 0.78, CI 95% =0.65–0.94, p-value = 7.3 × 10 − 3 ) and juvenile idiopathic arthritis (OR = 0.74, CI 95% =0.61–0.91, p-value = 4.4 × 10 − 3 ). In addition, we found some evidence (p-values: 0.01–0.04) for bidirectional relationship between lymphocyte percentage of white cells with five IMD (Fig. 1 B). Genetic liability to asthma and sinusitis were revealed as a potential risk factor across a number of immune cell phenotypes (Fig. 1 B). We found a potential negative causal relationship between asthma (β=-0.055, CI 95% =[-0.086, -0.024], p-value = 5.9 × 10 − 4 ), sinusitis (β=-0.042, CI 95% =[-0.067, -0.018], p-value = 6.3 × 10 − 4 ) and monocyte percentage of white cells but the relationship was not replicated using the absolute monocyte counts outcome. Intriguingly, we found no significant genetic support for causal association between any immune cell phenotypes and psoriasis as well as systemic lupus erythematosus (Fig. 1 A). For ankylosing spondylitis and multiple sclerosis, we only found fairly weak statistically significant support (p-value = 0.005–0.05) for immune cell basis of susceptibility (AS - neutrophil count, MS - monocyte count, white blood cell count, and neutrophil percentage of granulocytes). As expected due to genetic complexity of exposure and outcome phenotypes, sensitivity analyses revealed a high amount of heterogeneity 76 in the MR IVW estimates using Cochrane’s Q (Supplementary Tables 6, 11) but not horizontal pleiotropy as measured by MR-Egger intercept test (Supplementary Tables 7, 12). Drug target prioritisation using intermediate trait MR Having established immune blood cells are putatively causally associated with immune-mediated disease, we were then able to prioritise drug targets using intermediate trait MR. In this analysis, we utilised genic variants robustly associated with immune cell abundance (p-value < 5 × 10 − 8 ) as proxy instruments. In total, we were able to instrument 1,081 variants in 371 genes (Supplementary Table 16) out of 834 previously targeted for treatment of any of 12 immune-mediated disease (Supplementary Table 14). A total of 6,127 MR analyses were conducted, with some genes possessing instruments across a number of intermediate traits and so analysed independently (Supplementary Table 15). A quarter of IVW/Wald MR results (1,728) displayed evidence of association at the nominal significance level (p-value < 0.05), which revealed 811 unique gene-IMD associations (137 below Bonferroni-adjusted p-value threshold of 1 × 10 − 7 ). These involved 261 genes with the highest number of hits obtained for asthma (121 genes), inflammatory bowel disease (109 genes), eczema (83 genes) and chronic sinusitis (54 genes). Overall, we found extensive overlap of majority of drug targets with nominally significant MR evidence across IMD (Supplementary Fig. 2). The majority of genes with MR evidence (193, 74%) are known to contribute to immune function, which supports the choice of intermediate immune cell phenotypes. Since MR results are liable to confounding by linkage disequilibrium on their own, we also sought confirmation with the colocalisation approach, which assess the probability of shared genetic signal between the exposure and the outcome (Supplementary Table 20). Figure 2 highlights the MR results for genes with strong colocalisation support (PP H4 > 0.8) for above-mentioned 4 IMD, using instruments sourced from eosinophil count GWAS: A ) asthma – 27 genes, B ) inflammatory bowel disease – 12 genes, C ) sinusitis – 11 genes, D ) eczema – 10 genes. None of the targets were common to all the 4 conditions, but IL3 and GATA3 were shared among IBD, sinusitis and asthma, CDK2 , IL1R1 among eczema, asthma and sinusitis, IL33 and IL7R between asthma and sinusitis, ERBB3 , PRKCQ , STAT6 between asthma and eczema, CSF2 , FADS1 and FAP between IBD and asthma, PPARG between sinusitis and eczema, GPR65 and IL2 between eczema and IBD, and finally PTPN11 was shared between IBD and sinusitis. Many of these are not just confirmatory for existing indications for approved drugs or drugs in development ( PRKCQ , IL1R1 , STAT6 , IL2 ) but also suggesting additional repurposing opportunities for other IMD ( IL7R , CDK2 , GATA3 , IL33 , GPR65 , PTPN11 , ERBB3 , FAP ). However, opposite direction of effect across IMD was found for four targets ( IL3 , CSF2 , FADS1 , PPARG ) indicating lack of feasibility for drug repurposing and even potential adverse side effects. Fig. 2. Open in a new tab Intermediate trait MR results for 4 IMD showing the highest number of hits with strong colocalisation support (PP H4 ≥ 0.8): ( A ) asthma, ( B ) inflammatory bowel disease, ( C ) chronic sinusitis, ( D ) eczema. The x-axis represents the central effect size estimate and y-axis represents the -log10(P-value) in MR analysis, while the gene symbols and colocalisation probabilities of nominally significant MR hits (p-value < 0.05; beta > 0 in red, beta < 0 in blue) with strong colocalisation evidence highlighted in light green and PP H4 values listed in parenthesis. We used eosinophil counts as GWAS source for exposure instruments. Overall, we found 339 MR results at nominally significant level which were supported by strong colocalisation signal. Among 169 unique gene-IMD associations with robust coloc evidence, we found that a minority of associations (62, 36.7%) were confirmatory of existing drug target indications and the majority, 107 (63.3%), were novel. We also provide additional MR sensitivity statistics for heterogeneity (Supplementary Table 17), pleiotropy (Supplementary Table 18) and Steiger filtering (Supplementary Table 19). Since heterogeneity and pleiotropy tests rely on multi-SNP instruments (≥ 2 and ≥ 3 SNPs, respectively), we only obtained their results for a limited number of MR analyses (392 and 45, respectively). High levels of estimate heterogeneity (p-val < 0.05) were only seen for 86 out of 392 associations (22%) and no significant evidence for pleiotropy was found. The main direction of causal effect from immune cell-instrumented drug target to IMD was confirmed for the majority of MR associations (5,352, 87.3%) using Steiger filtering. Drug target prioritisation using protein QTL MR Next, we leveraged an alternative source of instrumental variables for drug targets for IMD: protein QTL corresponding to genetic associations with circulating protein concentration in the blood serum. For comparison purposes, we used three studies which applied the SomaScan technology for protein measurement (ARIC, deCODE and FENLAND), and one study which used the Olink technology (UKBioBank, UKBB). Across the four cohorts combined, we were able to obtain strong instruments for 361 proteins (233, 193, 169, 280 in ARIC, deCODE, FENLAND and UKBB, respectively, Supplementary Table 23) out of 834 with IMD indication in Pharmaprojects, which were used to run 11,370 cohort-specific MR analyses (Supplementary Table 22). Over 10% of IVW/Wald MR results (1,431) showed evidence of association at the nominal significance level (p-value < 0.05), which reflected 841 distinct gene-IMD associations (90 below Bonferroni-adjusted p-value threshold of 1 × 10 − 7 ). These included 284 unique proteins with the highest number of hits returned for inflammatory bowel disease (82 proteins), Crohn’s disease (80 proteins), rheumatoid arthritis (75 proteins) and psoriasis (72 proteins). Similar to intermediate trait MR, we uncovered ubiquitous sharing of drug targets with nominally significant MR evidence across IMD (Supplementary Fig. 3) and a high proportion of targets (78.7%) with immune-related function. Comparison of MR results obtained from the four cohorts using two dimensionality reduction methods (hierarchical clustering- Supplementary Fig. 4 and principal component analysis - Supplementary Fig. 5) showed a strong effect of a particular protein measurement chemistry. SomaScan-sourced pQTLs (ARIC, deCODE, FENLAND) clustered distinctly away from Olink (UKBB), underscoring the importance of inclusion of instruments from diverse sources. When considering overlap of gene-IMD nominally significant associations across all the 4 pQTL sources (Supplementary Fig. 6A), the top 3 categories contained singletons present only in UKBB (273 associations), ARIC (127 associations), deCODE (83 associations), followed by 63 associations shared across the 4 cohorts. However, this result is mostly driven by a priori limited sharing of genetic instruments across cohorts and when considering MR analyses involving only shared drug targets, the top category is composed of significant MR hits across all the four cohorts (Supplementary Fig. 6B). Analogous conclusions can be drawn when sub-setting to gene-IMD associations significant at stringent Bonferroni-corrected p-value threshold (p-value < 10 − 7 , Supplementary Fig. 6C-D). Confirmation of MR results with colocalisation (Supplementary Table 27) revealed strong support (PP H4 ≥ 0.8) for 230 MR associations (16% out of 1,431) comprised of 83 distinct protein-IMD associations. In keeping with results from intermediate trait MR, two-thirds of these associations were novel (53) rather than confirmatory of existing disease indications for drug target (30). Figure 3 highlights the advantage of using multiple pQTL sources ( A – ARIC, B – deCODE, C - UK Biobank) for ulcerative colitis, the IMD showing the highest number of hits with strong colocalisation support. VSIR and CD274 loci robustly colocalised with pQTLs across the 3 cohorts, while evidence for colocalisation for STAT3 was found both in ARIC and deCODE. Additionally, UKBB pQTLs independently prioritised 6 proteins ( CD6 , IL10 , IL10RA , IL1RL1 , ITGAV , OSMR ), none of which were instrumented in the other 3 pQTL studies. Fig. 3. Open in a new tab pQTL MR results for ulcerative colitis, the IMD showing the highest number of hits with strong colocalisation support (PP H4 ≥ 0.8). Plot compares MR results across different blood plasma pQTL exposure sources: ( A ) ARIC, ( B ) deCODE, ( C ) UK Biobank. The x-axis represents the central beta estimate and y-axis represents the -log10(P-value) in MR analysis, while the gene symbols and colocalisation probabilities of nominally significant MR hits (p-value < 0.05; beta > 0 in red, beta < 0 in blue) with strong colocalisation evidence highlighted in light green and PP H4 values listed in parenthesis. Among the MR analyses with sufficient number of exposure SNPs for heterogeneity analysis (Supplementary Table 24), only a small number showed significant evidence (p-value < 0.05) against the null hypothesis of homogeneity (297 out of 2875, 10%). MR Egger intercept test (Supplementary Table 25) revealed only 8 MR results (out of 1,576) with evidence for pleiotropy, 4 of which involved IL5RA pQTL used as exposure. Using Steiger filtering (Supplementary Table 26), the direction of causal effect from protein QTL to IMD was confirmed for all the MR associations, bar 3 for the BRD2 protein. Integrative MR-based validation and repositioning of drug targets We used Pharmaproject’s list containing approved drug targets to disease indication assignments to compare the sensitivity of intermediate trait MR (Supplementary Table 21) and protein QTL MR (Supplementary Table 28) approaches. Sensitivity was generally poor but with a large variance across disease outcomes. Mean intermediate trait MR sensitivity was estimated at 0.49 (± 0.23 SD), while for pQTL MR the mean sensitivity was 0.28 (± 0.12 SD); similar results were obtained when focussing on drug targets in development. Given fairly low predictive power of individual MR approaches, we were interested to compare how much additional information combining them can provide. Intersecting all nominally significant gene-IMD associations found in intermediate trait and pQTL MR (Fig. 4 A) showed only 13% overlap (190 associations). The remainder of associations were symmetrically distributed between intermediate trait MR (621, 42%) and pQTL MR (651, 45%). When sub-setting to MR analyses which instrumented the same genes across both approaches (Fig. 4 B), the fraction of overlap substantially increased (39%), with slightly more associations found only in intermediate trait MR (177, 36%) than pQTL MR (122, 25%). Fig. 4. Open in a new tab Venn diagram showing gene-IMD association overlap between significant intermediate trait MR and pQTL MR results. ( A ) nominal p-value threshold met, including all tested genes with a suitable instrument; ( B ) nominal p-value threshold met, including only overlapping target genes between intermediate and pQTL instrument sources; ( C ) Bonferroni-corrected p-value threshold (p-value < 10 − 7 )* met in either set, including all tested genes with a suitable instrument; ( D ) Bonferroni-corrected p-value threshold met in either set, including only overlapping target genes between intermediate and pQTL instrument sources. *Bonferroni-corrected threshold equalled: 0.05 / (number of proteins tested x number of outcomes tested). Altogether, we counted 167 highest confidence MR hits with p-value < 10 − 7 (Bonferroni-corrected threshold), with 77 (46%) specific to intermediate trait MR, 60 (36%) shared, and 30 (18%) specific to pQTL MR (Fig. 4 C). The vast majority of these singleton associations were due to availability of a given gene exposure only through one approach. Only 9 and 2 gene-IMD associations were found solely via intermediate trait MR and pQTL MR, respectively (Fig. 4 D), when limiting the analysis to the subset of MR results with overlapping target gene exposures. We summarised information regarding the drug targets with the highest confidence evidence present in both intermediate trait MR and pQTL MR in Table 1 . We found genetic evidence for repurposing opportunity in asthma and eczema for already approved rheumatoid arthritis drug anakinra which targets interleukin 1 receptor , type I ( IL1R1 ). For pateclizumab, an antibody directed towards lymphotoxin alpha ( LTα ) which had been halted for development as a rheumatoid arthritis medication due to lack of efficacy 77 , we found genetic evidence for potential in treatment of 3 other IMDs. In addition, we discovered strong MR support for STAT6 inhibition in asthma, currently in preclinical development by Recludix and Sanofi. Table 1. Genetic support for IMD drugs in ongoing clinical trials and potential cross-IMD repurposing opportunities. Summary of the drug targets with stringent support (Bonferroni-corrected p-value threshold < 1e-7) from at least 1 intermediate trait MR and 1 pQTL MR. HGNC Gene ID Gene symbol Genomic location (GRCh38) Disease(s) with MR evidence (protein association with disease: P - positive, N - negative) Drug name(s) or ID Drug mechanism of action Drug status (highest) Current disease indication(s) HGNC:5993 IL1R1 chr2:102,064,544-102,179,874 asthma (P), eczema (P) anakinra antagonist Approved juvenile idiopathic arthritis, rheumatoid arthritis HGNC:6709 LTA chr6:31,560,610-31,574,324 ankylosing spondylitis (P), psoriasis (P), type 1 diabetes (P), Crohn’s disease (N), inflammatory bowel disease (N), multiple sclerosis (N), ulcerative colitis (N) pateclizumab inhibitor Ceased rheumatoid arthritis HGNC:11,368 STAT6 chr12:57,095,408-57,132,139 asthma (P) Recludix and Sanofi (NA) inhibitor Preclinical asthma, eczema HGNC:1103 BRD2 chr6:32,968,594-32,981,505 juvenile idiopathic arthritis (P), multiple sclerosis (P), asthma (N), chronic sinusitis (N), rheumatoid arthritis (N), type 1 diabetes (N) GSK-3,183,475 inhibitor Phase I Clinical Trial psoriasis HGNC:24,235 CD200R1 chr3:112,640,052-112,693,950 eczema (P) ucenprubart agonist Phase I Clinical Trial eczema, inflammatory bowel disease, multiple sclerosis, systemic lupus erythematosus HGNC:11,916 TNFRSF1A chr12:6,328,757-6,342,114 multiple sclerosis (N) atrosab, SBT-104 antagonist Phase I Clinical Trial ankylosing spondylitis, Crohn’s disease, inflammatory bowel disease, multiple sclerosis, psoriasis, rheumatoid arthritis, ulcerative colitis HGNC:5962 IL10 chr1:206,767,602-206,774,541 inflammatory bowel disease (N), ulcerative colitis (N) GM-XANTHO antagonist Phase II Clinical Trial eczema HGNC:6008 IL2RA chr10:6,010,689-6,062,370 eczema (P), multiple sclerosis (P) rezpegaldesleukin, efavaleukin alfa agonist Phase II Clinical Trial eczema, inflammatory bowel disease, multiple sclerosis, psoriasis, systemic lupus erythematosus, rheumatoid arthritis, type 1 diabetes, ulcerative colitis HGNC:6014 IL4 chr5:132,673,986-132,682,678 asthma (P) PF-07275315 antagonist Phase II Clinical Trial eczema HGNC:6024 IL7R chr5:35,852,695-35,879,603 asthma (P), eczema (P), multiple sclerosis (P), type 1 diabetes (P) bempikibart, lusvertikimab antagonist Phase II Clinical Trial Crohn’s disease, eczema, multiple sclerosis, type 1 diabetes, ulcerative colitis HGNC:11,364 STAT3 chr17:42,313,324-42,388,540 Crohn’s disease (P), inflammatory bowel disease (P), ulcerative colitis (P), multiple sclerosis (N) WP-1220 inhibitor Phase II Clinical Trial Crohn’s disease, psoriasis, rheumatoid arthritis HGNC:11,930 TNFSF14 chr19:6,661,253-6,670,588 multiple sclerosis (N) CERC-002 antagonist Phase II Clinical Trial asthma, Crohn’s disease, inflammatory bowel disease, ulcerative colitis HGNC:320 AGER chr6:32,180,968-32,184,322 ankylosing spondylitis (P), juvenile idiopathic arthritis (P), ulcerative colitis (P), multiple sclerosis (P), eczema (N), rheumatoid arthritis (N), type 1 diabetes (N) azeliragon antagonist Phase III Clinical Trial asthma HGNC:2434 CSF2 chr5:132,073,789-132,076,170 asthma (P), eczema (P) lenzilumab, namilumab, plonmarlimab antagonist Phase III Clinical Trial ankylosing spondylitis, asthma, multiple sclerosis, psoriasis, rheumatoid arthritis HGNC:3618 FCGR2B chr1:161,647,243-161,678,654 inflammatory bowel disease (P) obexelimab antagonist Phase III Clinical Trial asthma, eczema, multiple sclerosis, rheumatoid arthritis, systemic lupus erythematosus HGNC:5998 IL1RL1 chr2:102,311,502-102,352,356 Crohn’s disease (P), eczema (P), inflammatory bowel disease (P), asthma (N) astegolimab antagonist Phase III Clinical Trial asthma, chronic sinusitis HGNC:19,100 IL23R chr1:67,138,637-67,265,903 Crohn’s disease (N), inflammatory bowel disease (N) ebdarokimab antagonist Phase III Clinical Trial Crohn’s disease, eczema, inflammatory bowel disease, psoriasis, systemic lupus erythematosus, ulcerative colitis HGNC:16,028 IL33 chr9:6,215,149-6,257,983 asthma (P) itepekimab, tozorakimab antagonist Phase III Clinical Trial asthma, chronic sinusitis, eczema HGNC:6015 IL4R chr16:27,313,668-27,364,778 asthma (N) manfidokimab antagonist Phase III Clinical Trial asthma, chronic sinusitis, eczema, eosinophilic esophagitis Open in a new tab For drugs in phase 1 clinical trials, we identified potential repurposing opportunities across 2 other IMD for a psoriasis target bromodomain-containing 2 ( BRD2) kinase. Interestingly, validation of disease indication was found for CD200 receptor 1 ( CD200R1 ) and eczema, as well as tumor necrosis factor receptor 1 ( TNFRSF1A ) and multiple sclerosis but misaligned direction of effect suggests that the drugs currently being trialled could potentially fail. For drug targets in phase 2 trials, we found confirmatory MR evidence for interleukin-7 receptor ( IL7R ) and eczema, multiple sclerosis, as well as STAT3 and Crohn’s disease. On the other hand, repurposing opportunities were uncovered for interleukin 4 ( IL4 ), IL7R and asthma, STAT3 and ulcerative colitis. Among drugs in the phase 3 clinical trials, we found the highest number of repurposing opportunities (4 IMD) for azeliragon, a small-molecule inhibitor of AGER . We also provide genetic validation for cytokine targets of monoclonal antibodies: interleukin 23 receptor ( IL23R ) for IBD with Crohn’s disease (but featuring misaligned direction of effect relative to currently developed drug), as well as colony stimulating factor 2 ( CSF2 ), interleukin 1 receptor like 1 ( IL1RL1 ), interleukin 33 ( IL33 ) for asthma. For CSF2 and IL1RL1 inhibitors, MR suggests eczema as an additional disease indication, while for IL1RL1 and Fc γ receptor IIb ( FCGR2B ) MR delivered repositioning evidence favouring inflammatory bowel disease. Discussion This study leveraged the Mendelian Randomization method to establish genetically-predicted immune cell mediation of IMDs. We then conducted MR analyses involving immune cell abundance-derived and protein QTL-derived genetic instruments to systematically prioritize indications for approved and in-development IMD drug targets, highlighting avenues for repurposing and providing validation for drugs in preclinical stages and in clinical trials. Integrating multiple MR approaches with colocalisation strengthens confidence in our findings, while indicating that a substantial proportion of MR results may stem from LD patterns between distinct causal variants rather than true causality. Using a crude measure of immune system activation – immune cell abundance (absolute and relative), we sought to establish bidirectional causal relationships of broad immune cell categories with IMD. Notably, eosinophil phenotypes emerged as key players, exhibiting a strong positive effect on diseases with atopic components. The observed bidirectional causal effect between eosinophil phenotypes and specific diseases aligns with prior research implicating eosinophils in the pathogenesis of atopic conditions such as asthma, eczema and eosinophilic esophagitis with their mechanistic contribution to type 2 inflammation 78 , 79 . Tissue eosinophilia is linked to epithelium remodelling along with airway hyperreactivity manifestations in asthma as well as skin and oesophageal mucosa infiltration in eczema and eosinophilic esophagitis, respectively. Reverse positive association of asthma, eczema and sinusitis with eosinophils could be attributed to secondary immunologic activation resulting in positive feedback loop in the atopic march 80 , 81 . Less expected was a robust MR relationship of eosinophiles with a range of autoimmune diseases: rheumatoid arthritis, juvenile idiopathic arthritis, type 1 diabetes and ulcerative colitis, which requires further investigation and triangulation. We found a strong positive association between genetically predicted neutrophil counts and incidence of inflammatory bowel disease and Crohn’s disease. In observational studies of IBD initiation, neutrophils have been found to migrate and accumulate at the site of inflamed mucosa, resulting in microbial dysbiosis, intensified intestinal architectural damage, compromised resolution of inflammation, and an elevated risk of thrombosis 82 . On other hand, functional deficiency of neutrophils can accelerate disease progression, underscoring the multifaceted role of these granulocytes. In addition, our MR analysis found that the “neutrophil percentage of granulocyte” exposure displays a protective effect for allergic disease and rheumatoid arthritis. This phenotype is less straightforward to interpret unlike absolute counts and may be related to changed relative balance of granulocyte types between peripheral and disease injury sites. Negative association of lymphocyte abundance with a number of IMD demonstrated in our MR analysis similarly requires a more fine-grained inspection, given disparate roles of various lymphocyte types (and their subtypes): T helper 83 , T regulatory cells 84 , memory T cells 85 , B lymphocytes 86 etc. across the spectrum of inflammatory and autoimmune disease. Future larger GWAS studies of more fine-grained immunotypes will likely reveal new causal relationships for immune-mediated disease, such as shown recently for SLE 87 . Similarly, leveraging GWAS of specific disease subtypes as they become available will lead to more precise allocation of drug targets to conditions, which recognises their heterogeneity 88 , 89 . Having established putatively causal relationships between immune cell traits and IMD, we then proceeded onto gene-centric MR analysis prioritising drug targets using the intermediate trait approach. Following the paradigm of triangulation 90 , we also corroborated and compared our findings with MR analyses utilising protein QTLs as exposures. The two approaches applied instruments derived from the same tissue – blood and focussed on targets with established roles in immunity, albeit largely non-overlapping between the two analysis sets, with only one-third shared. Altogether, we were able to track MR evidence for 548 drug targets at various stages in the development pipeline. Furthermore, we demonstrated the importance of utilizing multiple protein quantification studies, as unique signals were obtained from each. We also evaluated concordance across aptamer and antibody-based platforms to help derive consistent signals. Within intermediate and pQTL MR, we determined a substantial sharing of drug targets with MR evidence, especially for related atopic disease such as asthma, sinusitis and eczema, or different forms of inflammatory bowel disease: ulcerative colitis and Crohn’s disease. When compared, the two approaches yielded distinct disease sets with the highest number of drug targets with MR evidence, with only IBD overlapping across the top 4 in each method, thus providing complementary MR evidence. Thanks to the utilization of the MR framework, our results are less likely to be affected by environmental confounding and reverse causation bias, factors that can impede causal inference in conventional epidemiological study designs. However, soundness of conclusions from our MR studies is related to fulfilment of main MR assumptions. We satisfied the relevance assumption by using a strict p-value threshold for exposure variants (p-value < 5 × 10 − 8 ) which resulted in strong instrumental variables (min F-stat = 30). We used the colocalisation sensitivity method to test against the second assumption of exchangeability. Colocalisation compares the shape of pQTL or gene-specific intermediate trait association signal with outcome GWAS, which should reduce the number of linkage disequilibrium-related artefacts. Doubts remain around genes with coloc evidence located in the MHC region: AGER , BRD2 and LTA which were found to associate with a number of IMD. The complex LD structure in the region, which extends beyond our colocalisation window 91 may have contributed to some false positives. The colocalisation method employed by us requires the presence of just a single causal variant at the locus, unlike MR. This limitation was off-set by short computation time and lack of requirement for LD matrices for exposure and outcome GWAS 92 . This entails that small PP H4 values may not always represent evidence against colocalization, especially in cases where both PP H3 and PP H4 (both traits are associated, but with a different or shared causal variant, respectively) are small, indicating low power of analysis 70 . An additional measure undertaken against violation of the exchangeability assumption was to restrict our analysis to individuals of European descent, minimizing (but not eliminating 93 the bias associated with population stratification. On the other hand, this may limit the generalizability of our findings to diverse human populations. The exclusion restriction assumption was tested using the Egger intercept test, and it did not reveal much evidence for presence of horizontal pleiotropy, except for IL5RA pQTL. Nevertheless, this assumption cannot be definitively tested here due to the use of mostly single SNP instruments in intermediate trait and pQTL MR. However, restricting variants to cis -SNPs in proximity to the target gene is expected to limit bias from alternative pathways 15 . We encountered some limitations to our chosen study design. First of all, we were only able to find instruments for 548 out of 834 IMD drug targets (66%: 371 in intermediate MR and 361 in protein QTL) which have been deployed for preclinical and clinical investigations of IMD. Second of all, our MR analysis cannot instrument a drug that concurrently affects multiple parallel biological pathways. Multiple targets may be represented individually or in combination through a factorial MR approach 94 or meta-analysis 95 , if only summary GWAS data are available. Since our analysis’ focus was on drug targets rather than mechanism of action for individual drugs, we took the former approach. In general, genetics-based methods such as MR cannot be informative about effectiveness of particular drug molecule chemistry 12 . Thirdly, our MR studies leverages only GWAS of disease incidence as there is a scarcity of genetic studies of disease progression 96 . Focusing solely on incidence may fail to capture genetic factors specifically associated with the severity, rapidity of onset and complications of the disease course. Harnessing drugs which were already approved for use, we were able to quantify the sensitivity of both intermediate MR and pQTL approaches. Despite instrumenting a distinct set of targets, both methods arrived at similarly low average sensitivity (< 0.5). There is a host of possible reasons responsible for this result. In the intermediate trait MR, we may be lacking the most relevant immune cell phenotype to construct an instrumental variable with, such as counts of individual lymphocyte types. While blood is broadly of high relevance to all immune-mediated disease 4 , for individual drug targets it may not always be the most biologically meaningful compartment to proxy drug’s mechanism of action 12 . That said, a number of organs and tissues release their proteins into the blood in health and disease, for example the complement proteins synthesised by the liver 97 . Plasma protein abundance and activation are influenced by post-translational mechanisms such as cleavage and secretion; these processes cannot be modelled using instruments derived simply from protein levels in the plasma. Context-specific QTLs from inflammatory states 98 or single-cell specific QTLs obscured by bulk tissue analysis 99 , 100 could act as more appropriate instruments in future studies. Equally, pQTL analysis can be prone to false positives resulting from protein-altering variants which affect protein epitope; pQTLs detected in this case are going to reflect a mixture of changes due to the epitope effect and actual abundance 19 . Where proteins exist as inactive membrane-bound or circulating precursors as well as activated soluble forms, proteomic assays will be typically unable to clearly separate between different states. Such ambiguities makes interpreting the directionality of associations from MR problematic 98 . For instance, misaligned effect directionality relative to drug’s mechanism of action found in the pQTL MR for the association between TNFRSF1A and multiple sclerosis may stem from the exposure SNPs containing an epitope altering mutation. Furthermore, our pQTL MR results utilising non-PAV linked variants for two cytokine receptors: interleukin 4 receptor ( IL4R ) and interleukin-2 receptor subunit alpha ( IL2RA ) show opposite directions of effect to those established in the appropriate drug trials targeting asthma 101 and eczema 102 , 103 , respectively. We did also find strong evidence supporting truly directionally discordant effect of proteins whose pQTLs were not genetically linked to protein-altering variants. The role of individual proteins in IMD risk can be complex, with the same protein increasing risk for one condition while protecting against another. We replicated previous findings from Ferreira et al. (2013) 104 , Rosa et al. (2019) 105 and Zhao et al. (2023) 98 for IL-6 receptor using cis-pQTL from across 4 pQTL cohorts, with opposing effects found on the risk of atopic disease and rheumatoid arthritis. One plausible explanation is the context-dependent role of IL-6 signaling in different immune pathways. In RA, IL-6 promotes pro-inflammatory cytokine responses and joint inflammation, supporting the therapeutic benefit of IL6R inhibition (e.g., with tocilizumab) 106 . In contrast, in atopic diseases, IL-6 signalling may play a more regulatory or protective role in limiting Th2-driven allergic responses 107 . Thus, blocking IL6R might exacerbate atopic disease risk, consistent with the direction of effect observed in our MR results. In general, immune targets may have divergent effects depending on disease type, which is important for the development of precision therapies and for anticipating unintended consequences of immune modulation. We observed a similar phenomenon for BRD2 and CD40, with genetic predisposition to higher BRD2 increasing risk of multiple sclerosis but protecting against rheumatoid arthritis, and vice versa for CD40. This echoes real-world evidence from TNF-targeting therapies, which are efficacious for rheumatoid arthritis but not multiple sclerosis, where they may actually hasten disease progression 108 . As expected, we also uncovered disease-discordant significant effects using intermediate MR. IL3 , CSF2 and FADS1 were found to display a positive association with asthma but negative with IBD, in the case of CSF2 this could be also verified using pQTL instruments, with direction of effect in agreement. Such findings can diminish genetic support for drug repurposing across IMD. Conclusions The present study utilized Mendelian Randomization (MR) to first clarify bidirectional genetic associations between peripheral immune cell phenotypes and a spectrum of IMD. We found a complex pattern of causal relationships, with elevated eosinophils associated with increased odds of several diseases and increased neutrophil percentage of granulocytes showing a protective effect for atopic conditions. Intermediate trait Mendelian Randomization leveraging immune cell abundance instruments revealed 261 genes associated with immune-mediated diseases, with the highest count of hits for asthma, inflammatory bowel disease, eczema and chronic sinusitis. Colocalisation analysis provided further confirmation for 169 gene-disease pairs, with over 60% representing novel associations beyond existing drug indications. We then explored drug target prioritization utilising genetic associations of circulating protein concentrations as exposure. We compared results from different protein measurement technologies (SomaScan and Olink) which prioritized 284 proteins linked to diseases such as inflammatory bowel disease and rheumatoid arthritis. Colocalisation analysis supported 83 protein-disease pairs, two-thirds of which were novel and may warrant experimental investigation. Integrative analyses combining the intermediate trait and pQTL approaches showed relatively little overlap (13–39%), indicating their complementarity. Among shared signals, both validation for indications undergoing clinical trials and repurposing opportunities were revealed for: bempikibart and lusvertikimab (eczema – validation, asthma - repurposing) as well as lenzilumab, namilumab, plonmarlimab (asthma – validation, eczema – repurposing). New atopic indications for anakinra 109 could progress rapidly given that the drug is already approved. Overall, multi-modal human genetics resources enable comprehensive evaluation of therapeutic candidates for immune-mediated disease. We highlight the importance of considering multiple data sources and methodologies in drug target prioritization for a more robust assessment. Given that ∼57% of pipeline compounds fail due to inadequate efficacy 110 , genetically-informed indication selection holds certain promise to boost success rates. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 (669.3KB, docx) Supplementary Material 2 (14.9MB, xlsx) Author contributions MKS: Conceptualization, Formal analysis, Visualization, Writing – original draft, Writing – review and editingTG: Supervision, Writing – review and editing, Funding acquisition, Resources. Funding This work was funded by the UK Medical Research Council (MRC) as part of the MRC Integrative Epidemiology Unit (MC_UU_00032/03). This study was also supported by the NIHR Biomedical Research Centre at University Hospitals Bristol NHS Foundation Trust and the University of Bristol. The views expressed in this publication are those of the author(s) and not necessarily those of the NHS, the National Institute for Health Research or the Department of Health. Data availability We accessed the following immune cell GWAS summary statistics via OpenGWAS (https://gwas.mrcieu.ac.uk/) 111 : ebi-a-GCST90002292, ebi-a-GCST004634, ebi-a-GCST90002380, ebi-a-GCST90002298, ebi-a-GCST004617, ebi-a-GCST90002382, ebi-a-GCST004614, ebi-a-GCST004608, ebi-a-GCST90002316, ebi-a-GCST90002389, ebi-a-GCST90002340, ebi-a-GCST90002394, ebi-a-GCST004626, ebi-a-GCST90002351, ebi-a-GCST004623, ebi-a-GCST90002399, ebi-a-GCST004620, ebi-a-GCST004624, ebi-a-GCST004613, ebi-a-GCST90002374. We accessed the following immune-mediated disease GWAS summary statistics via OpenGWAS (https://gwas.mrcieu.ac.uk/) 111 : ebi-a-GCST90014325, ebi-a-GCST006862, ebi-a-GCST004132, ebi-a-GCST004131, ieu-b-18, ebi-a-GCST90019016, ebi-a-GCST002318, ebi-a-GCST003156, ebi-a-GCST010681, ebi-a-GCST004133. We obtained the following immune-mediated disease GWAS summary statistics from FinnGen ver. 9 (https://www.finngen.fi/en/access_results) 42 : M13_ANKYLOSPON, J10_CHRONSINUSITIS, L12_ATOPIC and NHGRI-EBI GWAS Catalog (http://www.ebi.ac.uk/gwas) 112 : GCST90244787, GCST90027899, GCST90010715. UKBB pQTL dataset 58 was accessed via [https://metabolomips.org/ukbbpgwas/](https:/metabolomips.org/ukbbpgwas), deCODE pQTL dataset 56 was accessed via [https://www.decode.com/summarydata/](https:/www.decode.com/summarydata) , ARIC pQTL dataset 55 was accessed via [http://nilanjanchatterjeelab.org/pwas/](http:/nilanjanchatterjeelab.org/pwas). Declarations Ethics approval and consent to participate The article uses previously published GWAS summary statistics. No separate ethical approval is required for this study. All subjects provided informed consent in the original studies. No human subjects were directly involved in this study. Competing interests TRG receives funding from Biogen and GlaxoSmithKline for unrelated research. M.K.S. declares no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References 1. He, M. et al. Immune-Mediated Diseases Associated With Cancer Risks. JAMA Oncol. 8 , 209–219 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 2. El-Gabalawy, H., Guenther, L. C. & Bernstein, C. N. Epidemiology of immune-mediated inflammatory diseases: incidence, prevalence, natural history, and comorbidities. J. Rheumatol. Suppl. 85 , 2–10 (2010). [ DOI ] [ PubMed ] [ Google Scholar ] 3. Netea, M. G. et al. Defining trained immunity and its role in health and disease. Nat. Rev. Immunol. 20 , 375–388 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 4. McInnes, I. B. & Gravallese, E. M. Immune-mediated inflammatory disease therapeutics: past, present and future. Nat. Rev. Immunol. 21 , 680–686 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Pisetsky, D. S. Pathogenesis of autoimmune disease. Nat. Rev. Nephrol. 19 , 509–524 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Begley, C. G. et al. Drug repurposing: Misconceptions, challenges, and opportunities for academic researchers. Sci. Transl Med. 13 , eabd5524 (2021). [ DOI ] [ PubMed ] [ Google Scholar ] 7. Petitdemange, A., Blaess, J., Sibilia, J., Felten, R. & Arnaud, L. Shared development of targeted therapies among autoimmune and inflammatory diseases: a systematic repurposing analysis. Ther. Adv. Musculoskelet. Dis. 12 , 1759720X20969261 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. David, T., Ling, S. F. & Barton, A. Genetics of immune-mediated inflammatory diseases. Clin. Exp. Immunol. 193 , 3–12 (2018). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 9. Nelson, M. R. et al. The support of human genetic evidence for approved drug indications. Nat. Genet. 47 , 856–860 (2015). [ DOI ] [ PubMed ] [ Google Scholar ] 10. Minikel, E. V., Painter, J. L., Dong, C. C. & Nelson, M. R. Refining the impact of genetic evidence on clinical success. medRxiv 10.1101/2023.06.23.23291765 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 11. King, E. A., Davis, J. W. & Degner, J. F. Are drug targets with genetic support twice as likely to be approved? Revised estimates of the impact of genetic support for drug mechanisms on the probability of drug approval. PLOS Genet. 15 , 1–20 (2019). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Gill, D. et al. Mendelian randomization for studying the effects of perturbing drug targets. Wellcome open. Res. 6 , 16 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Davies, N. M. & Holmes, M. V. & Davey Smith, G. Reading Mendelian randomisation studies: A guide, glossary, and checklist for clinicians. BMJ 362, (2018). [ DOI ] [ PMC free article ] [ PubMed ] 14. Daghlas, I. & Gill, D. Mendelian randomization as a tool to inform drug development using human genetics. Camb. Prism Precis Med. 1 , e16 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 15. Schmidt, A. F. et al. Genetic drug target validation using Mendelian randomisation. Nat. Commun. 11 , 3255 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Zhao, H. et al. Proteome-wide Mendelian randomization in global biobank meta-analysis reveals multi-ancestry drug targets for common diseases. Cell Genomics 2 , e100195 (2022). [ DOI ] [ PMC free article ] [ PubMed ] 17. Rasooly, D. et al. Genome-wide association analysis and Mendelian randomization proteomics identify drug targets for heart failure. Nat. Commun. 14 , 3826 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Yao, M. et al. Deciphering proteins in Alzheimer’s disease: A new Mendelian randomization method integrated with AlphaFold3 for 3D structure prediction. Cell Genomics 4, (2024). [ DOI ] [ PMC free article ] [ PubMed ] 19. Suhre, K., McCarthy, M. I. & Schwenk, J. M. Genetics meets proteomics: perspectives for large population-based studies. Nat. Rev. Genet. 22 , 19–37 (2021). [ DOI ] [ PubMed ] [ Google Scholar ] 20. Zheng, J. et al. Phenome-wide Mendelian randomization mapping the influence of the plasma proteome on complex diseases. Nat. Genet. 52 , 1122–1131 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. Daghlas, I. et al. Genetic Evidence for Repurposing of GLP1R (Glucagon-Like Peptide‐1 Receptor) Agonists to Prevent Heart Failure. J. Am. Heart Assoc. 10 , e020331 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Georgakis, M. K. et al. Interleukin-6 Signaling Effects on Ischemic Stroke and Other Cardiovascular Outcomes. Circ. Genomic Precis Med. 13 , e002872 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 23. Cronjé, H. T. et al. Genetic evidence implicating natriuretic peptide receptor-3 in cardiovascular disease risk: a Mendelian randomization study. BMC Med. 21 , 158 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 24. Zhou, G. et al. Exploring the association and causal effect between white blood cells and psoriasis using large-scale population data. Front. Immunol. 14 , 1043380 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 25. Pongdee, T., Bielinski, S. J., Decker, P. A., Kita, H. & Larson, N. B. White blood cells and chronic rhinosinusitis: a Mendelian randomization study. Allergy Asthma Clin. Immunol. 18 , 98 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 26. Schuster, C. et al. The Autoimmunity-Associated Gene CLEC16A Modulates Thymic Epithelial Cell Autophagy and Alters T Cell Selection. Immunity 42 , 942–952 (2015). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 27. He, D., Liu, L., Shen, D., Zou, P. & Cui, L. The Effect of Peripheral Immune Cell Counts on the Risk of Multiple Sclerosis: A Mendelian Randomization Study. Front. Immunol. 13 , 867693 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 28. Zeng-Yun-Ou, Z., Zhong-Yu, J. & Wei, L. Bidirectional associations between eosinophils, basophils, and lymphocytes with atopic dermatitis: A multivariable Mendelian randomization study. Front. Immunol. 13 , e1001911 (2022). [ DOI ] [ PMC free article ] [ PubMed ] 29. Guyatt, A. et al. Mendelian randomisation of eosinophils and other cell types in relation to lung function and disease. Thorax 78 , 496–503 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 30. Kim, H. Y., DeKruyff, R. H. & Umetsu, D. T. The many paths to asthma: phenotype shaped by innate and adaptive immunity. Nat. Immunol. 11 , 577–584 (2010). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 31. Hammad, H. & Lambrecht, B. N. The basic immunology of asthma. Cell 184 , 1469–1485 (2021). [ DOI ] [ PubMed ] [ Google Scholar ] 32. Facheris, P., Jeffery, J., Del Duca, E. & Guttman-Yassky, E. The translational revolution in atopic dermatitis: the paradigm shift from pathogenesis to treatment. Cell. Mol. Immunol. 20 , 448–474 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 33. Walsh, C. M. et al. Neutrophils promote CXCR3-dependent itch in the development of atopic dermatitis. Elife 8 , e48448 (2019). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 34. Crow, M. K. Pathogenesis of systemic lupus erythematosus: risks, mechanisms and therapeutic targets. Ann. Rheum. Dis. 82 , 999–1014 (2023). [ DOI ] [ PubMed ] [ Google Scholar ] 35. Baecher-Allan, C., Kaskow, B. J. & Weiner, H. L. Multiple Sclerosis: Mechanisms and Immunotherapy. Neuron 97 , 742–768 (2018). [ DOI ] [ PubMed ] [ Google Scholar ] 36. Williams, D. M. Clinical Pharmacology of Corticosteroids. Respir Care . 63 , 655–670 (2018). [ DOI ] [ PubMed ] [ Google Scholar ] 37. Cronstein, B. N. & Aune, T. M. Methotrexate and its mechanisms of action in inflammatory arthritis. Nat. Rev. Rheumatol. 16 , 145–154 (2020). [ DOI ] [ PubMed ] [ Google Scholar ] 38. Allison, A. C. & Eugui, E. M. Mycophenolate mofetil and its mechanisms of action. Immunopharmacology 47 , 85–118 (2000). [ DOI ] [ PubMed ] [ Google Scholar ] 39. Astle, W. J. et al. The allelic landscape of human blood cell trait variation and links to common complex disease. Cell 167 , 1415–1429 (2016). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 40. Chen, M. H. et al. Trans-ethnic and Ancestry-Specific Blood-Cell Genetics in 746,667 Individuals from 5 Global Populations. Cell 182 , 1198–1213e14 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 41. Vuckovic, D. et al. The Polygenic and Monogenic Basis of Blood Traits and Diseases. Cell 182 , 1214–1231e11 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 42. Kurki, M. I. et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature 613 , 508–518 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 43. Demenais, F. et al. Multiancestry association study identifies new asthma risk loci that colocalize with immune-cell enhancer marks. Nat. Genet. 50 , 42–53 (2018). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 44. Valette, K. et al. Prioritization of candidate causal genes for asthma in susceptibility loci derived from UK Biobank. Commun. Biol. 4 , 700 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 45. Budu-Aggrey, A. et al. European and multi-ancestry genome-wide association meta-analysis of atopic dermatitis highlights importance of systemic immune regulation. Nat. Commun. 14 , 6172 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 46. Chang, X. et al. A genome-wide association meta-analysis identifies new eosinophilic esophagitis loci. J. Allergy Clin. Immunol. 149 , 988–998 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 47. Stuart, P. E. et al. Transethnic analysis of psoriasis susceptibility in South Asians and Europeans enhances fine mapping in the MHC and genome wide. Hum. Genet. Genomics Adv. 3 , 100069 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 48. López-Isac, E. et al. Combined genetic analysis of juvenile idiopathic arthritis clinical subtypes identifies novel risk loci, target genes and key regulatory mechanisms. Ann. Rheum. Dis. 80 , 321–328 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 49. Okada, Y. et al. Genetics of rheumatoid arthritis contributes to biology and drug discovery. Nature 506 , 376–381 (2014). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 50. de Lange, K. M. et al. Genome-wide association study implicates immune activation of multiple integrin genes in inflammatory bowel disease. Nat. Genet. 49 , 256–261 (2017). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 51. Patsopoulos, N. A. et al. Multiple sclerosis genomic map implicates peripheral immune cells and microglia in susceptibility. Science 365 , eaav7188 (2019). [ DOI ] [ PMC free article ] [ PubMed ] 52. Bentham, J. et al. Genetic association analyses implicate aberrant regulation of innate and adaptive immunity genes in the pathogenesis of systemic lupus erythematosus. Nat. Genet. 47 , 1457–1464 (2015). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 53. Forgetta, V. et al. Rare Genetic Variants of Large Effect Influence Risk of Type 1 Diabetes. Diabetes 69 , 784–795 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 54. Burgess, S., Davies, N. M. & Thompson, S. G. Bias due to participant overlap in two-sample Mendelian randomization. Genet. Epidemiol. 40 , 597–608 (2016). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 55. Zhang, J. et al. Plasma proteome analyses in individuals of European and African ancestry identify cis-pQTLs and models for proteome-wide association studies. Nat. Genet. 10.1038/s41588-022-01051-w (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 56. Ferkingstad, E. et al. Large-scale integration of the plasma proteome with genetics and disease. Nat. Genet. 53 , 1712–1721 (2021). [ DOI ] [ PubMed ] [ Google Scholar ] 57. Pietzner, M. et al. Mapping the proteo-genomic convergence of human diseases. Sci. (80-) . 374 , eabj1541 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 58. Sun, B. B. et al. Plasma proteomic associations with genetics and health in the UK Biobank. Nature 10.1038/s41586-023-06592-6 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 59. Delaneau, O. et al. Integrating sequence and array data to create an improved 1000 Genomes Project haplotype reference panel. Nat. Commun. 5 , 3934 (2014). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 60. Chang, C. C. et al. Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience 4 , s13742-015-0047-8 (2015). [ DOI ] [ PMC free article ] [ PubMed ] 61. Thormann, A. et al. Flexible and scalable diagnostic fi ltering of genomic variants using G2P with Ensembl VEP. Nat. Commun. 1–10. 10.1038/s41467-019-10016-3 (2019). [ DOI ] [ PMC free article ] [ PubMed ] 62. Yuan, Z. et al. Testing and controlling for horizontal pleiotropy with probabilistic Mendelian randomization in transcriptome-wide association studies. Nat. Commun. 11 , 3861 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 63. Holmes, M. V., Richardson, T. G., Ference, B. A. & Davies, N. M. Davey Smith, G. Integrating genomics with biomarkers and therapeutic targets to invigorate cardiovascular drug development. Nat. Rev. Cardiol. 18 , 435–453 (2021). [ DOI ] [ PubMed ] [ Google Scholar ] 64. Eldjarn, G. H. et al. Large-scale plasma proteomics comparisons through genetics and disease associations. Nature 10.1038/s41586-023-06563-x (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 65. Hemani, G. et al. The MR-Base platform supports systematic causal inference across the human phenome. Elife 7 , e34408 (2018). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 66. Verbanck, M., Chen, C. Y., Neale, B. & Do, R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat. Genet. 50 , 693–698 (2018). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 67. Bowden, J., Davey Smith, G., Haycock, P. C. & Burgess, S. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genet. Epidemiol. 40 , 304–314 (2016). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 68. Bowden, J. et al. Assessing the suitability of summary data for two-sample Mendelian randomization analyses using MR-Egger regression: the role of the I2 statistic. Int. J. Epidemiol. 45 , 1961–1974 (2016). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 69. Wallace, C. Eliciting priors and relaxing the single causal variant assumption in colocalisation analyses. PLOS Genet. 16 , e1008720 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 70. Zuber, V. et al. Combining evidence from Mendelian randomization and colocalization: Review and comparison of approaches. Am. J. Hum. Genet. 10.1016/j.ajhg.2022.04.001 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 71. Wallace, C. Eliciting priors and relaxing the single causal variant assumption in colocalisation analyses. PLoS Genet. 16 , 1–20 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 72. Mullen, A., Blunck, M., Kalbfleisch, E., Möller, E. & Rohbeck, H. G. Assessment, from an industrial user perspective, of some major competitor information files on pharmaceutical development products. J. Inf. Sci. 23 , 9–23 (1997). [ Google Scholar ] 73. Bhattacharya, S. et al. ImmPort: disseminating data to the public for the future of immunology. Immunol. Res. 58 , 234–239 (2014). [ DOI ] [ PubMed ] [ Google Scholar ] 74. Breuer, K. et al. InnateDB: systems biology of innate immunity and beyond—recent updates and continuing curation. Nucleic Acids Res. 41 , D1228–D1233 (2013). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 75. Chepelev, I., Harley, I. T. W. & Harley, J. B. Modeling of horizontal pleiotropy identifies possible causal gene expression in systemic lupus erythematosus. Front Lupus 1 , e1234578 (2023). [ DOI ] [ PMC free article ] [ PubMed ] 76. Sobczyk, M. K., Richardson, T. G., Leyden, G. M. & Gaunt, T. R. Distinct pathway-based effects of blood pressure and body mass index on cardiovascular traits: comparison of novel Mendelian Randomization approaches. medRxiv (2023). 10.1101/2023.10.31.23297593 [ DOI ] [ PMC free article ] [ PubMed ] 77. Kennedy, W. P. et al. Efficacy and safety of pateclizumab (anti-lymphotoxin-α) compared to adalimumab in rheumatoid arthritis: a head-to-head phase 2 randomized controlled study (The ALTARA Study). Arthritis Res. Ther. 16 , 467 (2014). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 78. Folci, M., Ramponi, G., Arcari, I., Zumbo, A. & Brunetta, E. Eosinophils as Major Player in Type 2 Inflammation: Autoimmunity and Beyond BT - Cell Biology and Translational Medicine, Volume 14: Stem Cells in Lineage Specific Differentiation and Disease. in (ed Turksen, K.) 197–219 (Springer International Publishing, doi:10.1007/5584_2021_640. (2021). [ DOI ] [ PubMed ] 79. Caldwell, J. M., Paul, M. & Rothenberg, M. E. Novel immunologic mechanisms in eosinophilic esophagitis. Curr. Opin. Immunol. 48 , 114–121 (2017). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 80. Elias, P. M. & Steinhoff, M. “Outside-to-Inside” (and Now Back to“Outside”) Pathogenic Mechanisms in Atopic Dermatitis. J. Invest. Dermatol. 128, 1067–1070 (2008). [ DOI ] [ PMC free article ] [ PubMed ] 81. Egawa, G. & Kabashima, K. Multifactorial skin barrier deficiency and atopic dermatitis: Essential topics to prevent the atopic march. J. Allergy Clin. Immunol. 138 , 350–358e1 (2016). [ DOI ] [ PubMed ] [ Google Scholar ] 82. Chen, H., Wu, X., Xu, C., Lin, J. & Liu, Z. Dichotomous roles of neutrophils in modulating pathogenic and repair processes of inflammatory bowel diseases. Precis Clin. Med. 4 , 246–257 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 83. Hirahara, K., Aoki, A. & Nakayama, T. Pathogenic helper T cells. Allergol. Int. 70 , 169–173 (2021). [ DOI ] [ PubMed ] [ Google Scholar ] 84. Sakaguchi, S. et al. Regulatory T Cells and Human Disease. Annu. Rev. Immunol. 38 , 541–566 (2020). [ DOI ] [ PubMed ] [ Google Scholar ] 85. Park, C. O. & Kupper, T. S. The emerging role of resident memory T cells in protective immunity and inflammatory disease. Nat. Med. 21 , 688–697 (2015). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 86. Lee, D. S. W., Rojas, O. L. & Gommerman, J. L. B cell depletion therapies in autoimmune disease: advances and mechanistic insights. Nat. Rev. Drug Discov . 20 , 179–199 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 87. Gu, J. et al. Assessing the causal relationship between immune traits and systemic lupus erythematosus by bi-directional Mendelian randomization analysis. Mol. Genet. Genomics . 298 , 1493–1503 (2023). [ DOI ] [ PubMed ] [ Google Scholar ] 88. Swart, J. F., de Roock, S. & Prakken, B. J. Understanding inflammation in juvenile idiopathic arthritis: How immune biomarkers guide clinical strategies in the systemic onset subtype. Eur. J. Immunol. 46 , 2068–2077 (2016). [ DOI ] [ PubMed ] [ Google Scholar ] 89. Chen, M. et al. Overlap of allergic, eosinophilic and type 2 inflammatory subtypes in moderate-to-severe asthma. Clin. Exp. allergy J. Br. Soc. Allergy Clin. Immunol. 51 , 546–555 (2021). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 90. Lawlor, D. A. & Tilling, K. Davey Smith, G. Triangulation in aetiological epidemiology. Int. J. Epidemiol. 45 , 1866–1886 (2016). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 91. Giambartolomei, C. et al. A Bayesian framework for multiple trait colocalization from summary association statistics. Bioinformatics 34 , 2538–2545 (2018). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 92. Wallace, C. A more accurate method for colocalisation analysis allowing for multiple causal variants. PLoS Genet 17 , e1009440 (2021). [ DOI ] [ PMC free article ] [ PubMed ] 93. Schoeler, T. et al. Participation bias in the UK Biobank distorts genetic associations and downstream analyses. Nat. Hum. Behav. 7 , 1216–1227 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 94. Yarmolinsky, J., Bull, C. J., Walker, V. M., Nounu, A. & Smith, G. D. Mendelian randomization applied to pharmaceutical use: the case of metformin and lung cancer. Int. J. Epidemiol. 49 , 1410–1411 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 95. Zheng, J. et al. Evaluating the efficacy and mechanism of metformin targets on reducing Alzheimer’s disease risk in the general population: a Mendelian randomisation study. Diabetologia 65 , 1664–1675 (2022). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 96. Paternoster, L., Tilling, K. & Davey Smith, G. Genetic epidemiology and Mendelian randomization for informing disease therapeutics: Conceptual and methodological challenges. PLOS Genet. 13 , e1006944 (2017). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 97. Vignesh, P., Rawat, A., Sharma, M. & Singh, S. Complement in autoimmune diseases. Clin. Chim. Acta . 465 , 123–130 (2017). [ DOI ] [ PubMed ] [ Google Scholar ] 98. Zhao, J. H. et al. Genetics of circulating inflammatory proteins identifies drivers of immune-mediated disease risk and therapeutic targets. Nat. Immunol. 24 , 1540–1551 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 99. Yazar, S. et al. Single-cell eQTL mapping identifies cell type-specific genetic control of autoimmune disease. Science 376 , eabf3041 (2022). [ DOI ] [ PubMed ] [ Google Scholar ] 100. Kang, J. B. et al. Mapping the dynamic genetic regulatory architecture of HLA genes at single-cell resolution. Nat. Genet. 55 , 2255–2268 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 101. Harb, H. & Chatila, T. A. Mechanisms of Dupilumab. Clin. Exp. allergy J. Br. Soc. Allergy Clin. Immunol. 50 , 5–14 (2020). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 102. Langowski, J. et al. NKTR-358: a selective, first-in-class IL-2 pathway agonist which increases number and suppressive function of regulatory T cells for the treatment of immune inflammatory disorders. Arthritis Rheumatol. 69 , suppl 10 (2017). 103. Silverberg, J. et al. A Phase 2b, Randomized, Double-Blinded, Parallel-Group, Placebo-Controlled Study to Evaluate the Efficacy and Safety of Rezpegaldesleukin in Adults with Moderate-to-Severe Atopic Dermatitis. Br. J. Dermatol. 191 , ljae266.097 (2024). 104. Ferreira, R. C. et al. Functional IL6R 358Ala Allele Impairs Classical IL-6 Receptor Signaling and Influences Risk of Diverse Inflammatory Diseases. PLOS Genet. 9 , 1–12 (2013). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 105. Rosa, M. et al. A Mendelian randomization study of IL6 signaling in cardiovascular diseases, immune-related disorders and longevity. npj Genomic Med. 4 , 1–10 (2019). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 106. Taylor, P. C. et al. What have we learnt from the inhibition of IL-6 in RA and what are the clinical opportunities for patient outcomes? Ther. Adv. Musculoskelet. Dis. 16 , 1759720X241283340 (2024). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 107. Bachus, H. et al. IL-6 prevents Th2 cell polarization by promoting SOCS3-dependent suppression of IL-2 signaling. Cell. Mol. Immunol. 20 , 651–665 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 108. Group, T. L. M. S. S. & of British Columbia MS/MRI Analysis Group, T. U. TNF neutralization in MS. Neurology 53 , 457 (1999). [ PubMed ] [ Google Scholar ] 109. Atassi, K. T. G., Lio, P. A. & J. Z. N. J. M. & Off-Label studies on anakinra in dermatology: a review. J. Dermatolog Treat. 33 , 73–86 (2022). [ DOI ] [ PubMed ] [ Google Scholar ] 110. Hwang, T. J. et al. Failure of Investigational Drugs in Late-Stage Clinical Development and Publication of Trial Results. JAMA Intern. Med. 176 , 1826–1833 (2016). [ DOI ] [ PubMed ] [ Google Scholar ] 111. Elsworth, B. et al. The MRC IEU OpenGWAS data infrastructure. bioRxiv 10.1101/2020.08.10.244293 (2020). [ Google Scholar ] 112. Sollis, E. et al. The NHGRI-EBI GWAS Catalog: knowledgebase and deposition resource. Nucleic Acids Res. 51 , D977–D985 (2023). [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials Supplementary Material 1 (669.3KB, docx) Supplementary Material 2 (14.9MB, xlsx) Data Availability Statement We accessed the following immune cell GWAS summary statistics via OpenGWAS (https://gwas.mrcieu.ac.uk/) 111 : ebi-a-GCST90002292, ebi-a-GCST004634, ebi-a-GCST90002380, ebi-a-GCST90002298, ebi-a-GCST004617, ebi-a-GCST90002382, ebi-a-GCST004614, ebi-a-GCST004608, ebi-a-GCST90002316, ebi-a-GCST90002389, ebi-a-GCST90002340, ebi-a-GCST90002394, ebi-a-GCST004626, ebi-a-GCST90002351, ebi-a-GCST004623, ebi-a-GCST90002399, ebi-a-GCST004620, ebi-a-GCST004624, ebi-a-GCST004613, ebi-a-GCST90002374. We accessed the following immune-mediated disease GWAS summary statistics via OpenGWAS (https://gwas.mrcieu.ac.uk/) 111 : ebi-a-GCST90014325, ebi-a-GCST006862, ebi-a-GCST004132, ebi-a-GCST004131, ieu-b-18, ebi-a-GCST90019016, ebi-a-GCST002318, ebi-a-GCST003156, ebi-a-GCST010681, ebi-a-GCST004133. We obtained the following immune-mediated disease GWAS summary statistics from FinnGen ver. 9 (https://www.finngen.fi/en/access_results) 42 : M13_ANKYLOSPON, J10_CHRONSINUSITIS, L12_ATOPIC and NHGRI-EBI GWAS Catalog (http://www.ebi.ac.uk/gwas) 112 : GCST90244787, GCST90027899, GCST90010715. UKBB pQTL dataset 58 was accessed via [https://metabolomips.org/ukbbpgwas/](https:/metabolomips.org/ukbbpgwas), deCODE pQTL dataset 56 was accessed via [https://www.decode.com/summarydata/](https:/www.decode.com/summarydata) , ARIC pQTL dataset 55 was accessed via [http://nilanjanchatterjeelab.org/pwas/](http:/nilanjanchatterjeelab.org/pwas). Articles from Scientific Reports are provided here courtesy of Nature Publishing Group ACTIONS View on publisher site PDF (3.1 MB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top