
==== Front
iScience
iScience
iScience
2589-0042
Elsevier

S2589-0042(24)01929-1
10.1016/j.isci.2024.110704
110704
Article
Association of inflammatory cytokines with lung function, chronic lung diseases, and COVID-19
Rontogianni Marina O. 129
Gill Dipender 229
Bouras Emmanouil 129
Asimakopoulos Alexandros-Georgios 1
Tzoulaki Ioanna 12
Karhunen Ville 34
Lehtimäki Terho 567
Raitakari Olli 8910
Wielscher Matthias 2
Salomaa Veikko 11
Jalkanen Sirpa 121314
Salmi Marko 121314
Timonen Markku 1516
Yarmolinsky James 1718
Chen Jing 19
Tobin Martin D. 1920
Izquierdo Abril G. 21
Herzig Karl-Heinz 162223
Ioannides Anne E. 24
Jarvelin Marjo-Riitta 2325263031
Dehghan Abbas 227283031
Tsilidis Konstantinos K. k.tsilidis@imperial.ac.uk
12303132∗
1 Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina, Greece
2 Department of Epidemiology and Biostatistics, School of Public Health, Imperial College London, London, UK
3 Research Unit of Population Health, Faculty of Medicine, University of Oulu, Oulu, Finland
4 Research Unit of Mathematical Sciences, University of Oulu, Oulu, Finland
5 Department of Clinical Chemistry, Faculty of Medicine & Health Technology, Tampere University, Tampere, Finland
6 Finnish Cardiovascular Research Center Tampere, Faculty of Medicine & Health Technology, Tampere University, Tampere, Finland
7 Fimlab Laboratories, Tampere, Finland
8 Centre for Population Health Research, University of Turku and Turku University Hospital, Turku, Finland
9 Research Centre of Applied and Preventive Cardiovascular Medicine, University of Turku, Turku, Finland
10 Department of Clinical Physiology and Nuclear Medicine, Turku University Hospital, Turku, Finland
11 Finnish Institute for Health and Welfare, Helsinki, Finland
12 MediCity Research Laboratory, University of Turku, Turku, Finland
13 Institute of Biomedicine, University of Turku, Turku, Finland
14 InFLAMES Fiagship, University of Turku, Turku, Finland
15 Research Unit of Population Health, University of Oulu, Oulu, Finland
16 Medical Research Center (MRC) and University Hospital, Oulu, Finland
17 MRC Integrative Epidemiology Unit, University of Bristol, Bristol, UK
18 Population Health Sciences, Bristol Medical School, University of Bristol, Bristol, UK
19 Genetic Epidemiology Group, Department of Health Sciences, University of Leicester, Leicester, UK
20 National Institute for Health Research, Leicester Respiratory Biomedical Research Centre, Glenfield Hospital, Leicester, UK
21 Department of Health Sciences, University of Leicester, Leicester, UK
22 Research Unit of Biomedicine and Internal Medicine, Faculty of Medicine, University of Oulu, Oulu, Finland
23 Pediatric Gastroenterology and Metabolic Diseases, Pediatric Institute, Poznan University of Medical Sciences, Poznan, Poland
24 Department of Primary Care and Public Health, School of Public Health, Imperial College London, White City Campus, London, UK
25 Unit of Primary Care, Oulu University Hospital, Oulu, Finland
26 Department of Life Sciences, College of Health and Life Sciences, Brunel University London, Uxbridge, UK
27 Cancer Epidemiology Unit, Nuffield Department of Population Health, University of Oxford, Oxford, UK
28 UK Dementia Research Institute at Imperial College London, London, UK
∗ Corresponding author k.tsilidis@imperial.ac.uk
29 These authors contributed equally

30 These authors contributed equally

31 Senior authors

32 Lead contact

09 8 2024
18 10 2024
09 8 2024
27 10 11070419 2 2024
14 6 2024
6 8 2024
© 2024 The Author(s)
2024
https://creativecommons.org/licenses/by/4.0/ This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
Summary

We investigated the effects of 35 inflammatory cytokines on respiratory outcomes, including COVID-19, asthma (atopic and non-atopic), chronic obstructive pulmonary disease (COPD), and pulmonary function indices, using Mendelian randomization and colocalization analyses. The emerging associations were further explored using observational analyses in the UK Biobank. We found an inverse association between genetically predicted macrophage colony stimulating factor (MCSF), soluble intercellular adhesion molecule-1 (sICAM), and soluble vascular cell adhesion molecule-1 with risk of COVID-19 outcomes. sICAM was positively associated with atopic asthma risk, whereas tumor necrosis factor-alfa showed an inverse association. A positive association was shown between interleukin-18 and COPD risk (replicated in observational analysis), whereas an inverse association was shown for interleukin-1 receptor antagonist (IL-1ra). IL-1ra and monocyte chemotactic protein-3 were positively associated with lung function indices, whereas inverse associations were shown for MCSF and interleukin-18 (replicated in observational analysis). Our results point to these cytokines as potential pharmacological targets for respiratory traits.

Graphical abstract

Highlights

• We assessed the effects of 35 unique cytokines on respiratory outcomes

• Mendelian randomization and colocalization analyses were performed

• The emerging associations were further explored using observational analyses

• Our results implicate several cytokines in respiratory disease risk

Health sciences; Respiratory medicine; Disease; Association analysis

Subject areas

Health sciences
Respiratory medicine
Disease
Association analysis
Published: August 9, 2024
==== Body
pmcIntroduction

Chronic diseases of the respiratory system, including asthma and chronic obstructive pulmonary disease (COPD), represent some of the most common causes of morbidity and mortality worldwide.1 In 2017, approximately 545 million people were living with chronic respiratory disease globally and this accounted for 3.9 million deaths.2 Impairment of lung function, as measured by forced expiratory volume in the first second of exhalation (FEV1), forced vital capacity (FVC) and peak expiratory flow rate (PEF), is a predictor of adverse health outcomes in respiratory disease, as well as in all-cause mortality.3 The importance of respiratory health has been highlighted by the coronavirus disease 2019 (COVID-19) pandemic, which accounted for approximately 18 million excess deaths by the end of 2021.4 Activation of inflammatory pathways plays a crucial role in the pathophysiology of both COVID-195 and chronic respiratory diseases.6 Even though inflammation is a physiological reaction to harmful stimuli, in case of immune dysregulation and hyperinflammation, i.e., a surge of inflammatory factors, it can be detrimental and even life-threatening. Characteristic is the paradigm of the cytokine release syndrome in COVID-19, for which several immunomodulatory drugs are under investigation or approved, such as tocilizumab (interleukin-6 (IL-6) receptor antagonist).5,7

Establishing the role of inflammatory cytokines in impaired lung function and respiratory disease is therefore of direct relevance to understanding the pathophysiology of these conditions, and thus identifying therapeutic opportunities. However, randomized controlled trials can be prohibitively expensive and time-consuming,8 and observational studies have the disadvantage of possible confounding or reverse causation bias. Mendelian randomization (MR) can overcome these limitations by leveraging germline genetic variants as instrumental variables for studying the effect of altering an exposure on a disease outcome.9 The random allocation of genetic variants at meiosis and conception means that their associations with disease outcomes are less likely to be subject to environmental confounding or reverse causation bias. Further, when investigating protein drug target genes,10 colocalization analysis can be used to explore whether any genetic associations are attributable to a common causal variant across protein targets and disease outcomes, rather than distinct variants that are in linkage disequilibrium (LD).11

In this study, we perform MR and colocalization analyses to investigate the effects of 35 unique inflammatory cytokines on FEV1, FVC, the FEV1/FVC ratio, PEF, and risks of asthma (atopic and non-atopic), COPD and COVID-19. Thus, we aim to offer advanced insight into the potential causal role of inflammatory cytokines on respiratory disease and thus identify pharmacological targets for prioritization in clinical studies.

Some of the results of this study have been accepted in the form of a conference abstract.12

Results

Instrument characteristics

We used summary genetic association estimates from a genome-wide association study (GWAS) meta-analysis of circulating levels of 47 inflammatory cytokines.13,14,15,16,17 Among them, instruments were found for 31 and 27 cytokines using the cis-protein quantitative trait locus (cis-pQTL) and cis-expression quantitative trait locus (cis-eQTL) definitions, respectively. These included 35 unique cytokines that are presented in Table S1. Of the 27 eQTL cytokines, 6 had weak instruments (F-statistic<10) and were excluded from the analysis, and one had no instruments after the clumping of the data, resulting in 20 eQTL cytokines that were included in the analysis.

Cytokine associations with outcomes using Mendelian randomization

Coronavirus disease 2019 outcomes

Using the cis-pQTL instrument definition and considering a false discovery rate (FDR) < 5%, genetically proxied concentrations of macrophage colony stimulating factor (MCSF) were inversely associated with the risks of SARS-CoV-2 infection and hospitalization due to COVID-19 (odds ratio [OR]: 0.94, 95% confidence interval [CI]: 0.91 to 0.97, p (not FDR corrected) = 1.0 × 10−4 and OR: 0.90, 95%CI: 0.84 to 0.96, p = 1.4 × 10−3, respectively). Genetically proxied concentrations of soluble intercellular adhesion molecule 1 (sICAM) were inversely associated with the risk of SARS-CoV-2 infection (OR: 0.96, 95%CI: 0.94 to 0.98, p = 3.0 × 10−4), very severe respiratory confirmed COVID-19 (OR: 0.82, 95%CI: 0.73 to 0.91, p = 1.0 × 10−4) and hospitalization due to COVID-19 (OR: 0.89, 95%CI: 0.82 to 0.95, p = 8.0 × 10−4); hospitalization due to COVID-19 was also inversely associated with soluble vascular cell adhesion molecule 1 (sVCAM) (OR: 0.65, 95%CI: 0.48 to 0.88, p = 4.5 × 10−3). Using the cis-eQTL criteria, similar results were found for sICAM (OR: 0.83, 95%CI: 0.74 to 0.93, p = 2.0 × 10−3 for SARS-CoV-2 infection; OR: 0.49, 95%CI: 0.37 to 0.64, p = 1.26 × 10−7 for hospitalization due to COVID-19; OR: 0.39, 95%CI: 0.24 to 0.62, p = 1.0 × 10−4 for very severe COVID-19), and for sVCAM (OR: 0.65, 95%CI: 0.48 to 0.88, p = 4.5 × 10−3 for hospitalization due to COVID-19) (Figures 1, 2, 3, 4, and 5).Figure 1 Heatmap of cytokines (cis-pQTL) with outcomes

Summary of the Mendelian randomization results with the inverse variance weighted method based on the cis-pQTL instrument definition, where squared tiles indicate nominally significant associations (p < 0.05), the asterisk denotes that the association was significant when considering multiple comparison correction (false discovery rate <5%), and associations for which no instrument was available are presented as white tiles. Abbreviations: cis-pQTL, cis-protein quantitative trait locus; activePAI, active plasminogen activator inhibitor-1; CTACK, cutaneous T cell attracting chemokine; GROa, growth regulated oncogene-alpha; HGF, hepatocyte growth factor; IL-12p70, interleukin 12p70; IL 16, interleukin 16; IL 18, interleukin 18; IL1RA, interleukin 1 receptor antagonist; IL2ra, interleukin-2 receptor antagonist; IL6, interleukin 6; IL7, interleukin 7; IL8, interleukin 8; IP10, interferon gamma-induced protein 10; MCP1, monocyte chemotactic protein-1; MCP3, monocyte specific chemokine; MCSF, macrophage colony stimulating factor; MIF, macrophage migration inhibitory factor; MIG, monokine induced by interferon-gamma; MIP1a, macrophage inflammatory protein 1a; MIP1b, macrophage inflammatory protein 1b; PDGFbb, platelet-derived growth factor BB; RANTES, beta-chemokine RANTES; SCF, stem cell factor; SCGFb, stem cell growth factor beta; SeSelectin, soluble Eselectin; sICAM, soluble intercellular adhesion molecule 1; sVCAM, soluble vascular cell adhesion molecule 1; TNF a, tumor necrosis factor a; TRAIL, TNF-related apoptosis inducing ligand; VEGF, vascular endothelial growth factor; COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume measured in the first second of exhalation; and FVC, forced vital capacity.

Figure 2 Heatmap of cytokines (cis-eQTL) with outcomes

Summary of the Mendelian randomization results with the Inverse Variance Weighted method based on the cis-eQTL instrument definition, where squared tiles indicate nominally significant associations (p < 0.05), the asterisk denotes that the association was significant when considering multiple comparison correction (false discovery rate <5%), and associations for which no instrument was available are presented as white tiles. Abbreviations: cis-eQTL, cis-expression quantitative trait locus; CTACK, cutaneous T cell attracting chemokine; FGFBasic, basic fibroblast growth factor ; GROa, growth regulated oncogene-alpha; IL 16, interleukin 16; IL 18, interleukin 18; IL1RA, interleukin 1 receptor antagonist; IL2ra, interleukin-2 receptor antagonist; IL8, interleukin 8; MCP1, monocyte chemotactic protein-1; MCSF, macrophage colony stimulating factor; MIF, macrophage migration inhibitory factor; MIG, monokine induced by interferon-gamma; MIP1a, macrophage inflammatory protein 1a; MIP1b, macrophage inflammatory protein 1b; RANTES, beta-chemokine RANTES; SCGFb, stem cell growth factor beta; sICAM, soluble intercellular adhesion molecule 1; sVCAM, soluble vascular cell adhesion molecule 1; TRAIL, TNF-related apoptosis inducing ligand; COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume measured in the first second of exhalation; and FVC, forced vital capacity.

Figure 3 Forest plot showing the nominally significant (p < 0.05) MR-IVW estimates (beta, 95%CI) per standard deviation (SD) increase in the natural scale of each cytokine based on the pQTL definition

The asterisk denotes that the association was significant when considering multiple comparison corrections (false discovery rate <5%). Abbreviations: MR-IVW, Mendelian randomization Inverse Variance Weighted method; cis-pQTL, cis-protein quantitative trait locus; CI, confidence interval; SNP, single nucleotide polymorphism; COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume measured in the first second of exhalation; FVC, forced vital capacity; MCSF, macrophage colony stimulating factor; sICAM, soluble intercellular adhesion molecule 1; IL 18, interleukin 18; sVCAM, soluble vascular cell adhesion molecule 1; GROa, growth regulated oncogene-alpha; TNF a, tumor necrosis factor a; SeSelectin, soluble E-selectin; VEGF, vascular endothelial growth factor; SCF, stem cell factor; IL1RA, interleukin 1 receptor antagonist; RANTES, beta chemokine RANTES; IL2ra, interleukin-2 receptor antagonist; CTACK, cutaneous T cell attracting chemokine; MCP3, monocyte specific chemokine; PDGFbb, platelet-derived growth factor BB; HGF, hepatocyte growth factor; activePAI, active plasminogen activator inhibitor- 1; and IL 16, interleukin 16.

Figure 4 Forest plot showing the nominally significant (p < 0.05) MR-IVW estimates (beta, 95%CI) per standard deviation (SD) increase in the natural scale of each cytokine based on the eQTL definition

The asterisk denotes that the association was significant when considering multiple comparison correction (false discovery rate <5%). Abbreviations: MR-IVW, Mendelian randomization Inverse Variance Weighted method; cis-eQTL, cis-expression quantitative trait locus; CI, confidence interval; SNP, single nucleotide polymorphism; COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume measured in the first second of exhalation; FVC, forced vital capacity; sICAM, soluble intercellular adhesion molecule 1; MCSF, macrophage colony stimulating factor; sVCAM, soluble vascular cell adhesion molecule 1; FGFBasic, basic fibroblast growth factor ; IL1RA, interleukin 1 receptor antagonist; IL 18, interleukin 18; CTACK, cutaneous T cell attracting chemokine; RANTES, beta-chemokine RANTES; IL2ra, interleukin-2 receptor antagonist; IL 16, interleukin 16; and MIG, monokine induced by interferon-gamma.

Atopic and non-atopic asthma

Using the cis-pQTL instrument definition, an inverse association was found between genetically proxied concentrations of tumor necrosis factor-alfa (TNF-a) and risk of atopic asthma (OR: 0.38, 95%CI: 0.26 to 0.54, p = 9.78 × 10−8) (FDR<5%). Using the eQTL criteria, an FDR significant positive association was found for sICAM (OR: 1.54, 95%CI: 1.24 to 1.91, p = 1.0 × 10−4). Using either the cis-pQTL or the cis-eQTL instrument definition, no FDR significant association was found with non-atopic asthma (Figures 1, 2, 3, and 4).

Chronic obstructive pulmonary disease

Using the cis-pQTL definition, the risk of COPD was higher with increased genetically proxied concentrations of interleukin-18 (IL-18) (OR: 1.13, 95%CI: 1.06 to 1.20, p = 1.0 × 10−4) (FDR<5%). Using the eQTL criteria, the positive and FDR significant association for IL-18 was replicated (OR: 1.13, 95%CI: 1.04 to 1.24, p = 4.2 × 10−3), and additional inverse FDR significant associations were shown for interleukin-1 receptor antagonist (IL-1ra) (OR: 0.86, 95%CI: 0.78 to 0.96, p = 7.0 × 10−3) and cutaneous T cell attracting chemokine (CTACK) (OR: 0.90, 95%CI: 0.84 to 0.97, p = 5.8 × 10−3) (Figures 1, 2, 3, and 4).

Forced expiratory volume in the first second of exhalation

Using the cis-pQTL instrument definition criteria and considering an FDR<5%, a positive association was found between IL-1ra (beta: 0.03, 95%CI: 0.01 to 0.04, p = 3.4 × 10−3), monocyte chemotactic protein-3 (MCP3) (0.01, 95%CI: 0.00 to 0.02, p = 4.7 × 10−3), platelet derived growth factor-BB (PDGFbb) (0.15, 95%CI: 0.04 to 0.25, p = 6.0 × 10−3) and FEV1, whereas a negative association was found for IL-18 (−0.03, 95%CI: −0.04 to −0.01, p = 2.0 × 10−4). Using the eQTL criteria, the positive and FDR significant association of IL-1ra was replicated (0.05, 95%CI: 0.03 to 0.08, p = 3.7 × 10−5) (Figures 1, 2, 3, and 4).

Forced vital capacity

Using the cis-pQTL definition, we found evidence (FDR<5%) of a positive association between genetically proxied concentrations of IL-1ra (beta: 0.03, 95%CI: 0.01 to 0.04, p = 1.0 × 10−4), MCP3 (0.01, 95%CI: 0.00 to 0.02, p = 2.6 × 10−3) and FVC; negative associations were found for IL-18 (−0.03, 95%CI:-0.04 to −0.01, p = 5.0 × 10−4) and MCSF (−0.02, 95%CI: −0.04 to −0.01, p = 2.3 × 10−3). Using the eQTL criteria, similar FDR significant associations were found for IL-1ra (0.04, 95%CI: 0.02 to 0.07, p = 5.0 × 10−4) and MCSF (−0.04, 95%CI: −0.06 to −0.01, p = 3.6 × 10−3) (Figures 1, 2, 3, and 4).

Forced expiratory volume in the first second of exhalation/forced vital capacity

Using the cis-pQTL criteria, no FDR significant association was found between genetically proxied concentrations of cytokines and FEV1/FVC. Using the cis-eQTL criteria, a negative association was found for sICAM (beta: −0.1, 95%CI: −0.16 to −0.04, p = 1.0 × 10−3) (FDR<5%) (Figures 1, 2, 3, and 4).

Peak expiratory flow rate

Using the cis-pQTL and cis-eQTL instrument definition criteria, no FDR significant association was found (Figures 1, 2, 3, and 4).

All significant (FDR<5%) and nominally significant (p < 0.05) associations for all outcomes are shown in Tables S2 and S3.

Analysis showed similar results in weighted-median, MR-Egger and MR-PRESSO for all FDR significant associations, suggesting little evidence for directional pleiotropy driving the results (Tables S2 and S3).

Review of databases for medical drugs

For the 8 cytokines (MCSF, IL-18, IL-1ra, sICAM, MCP3, PDGFbb, sVCAM, TNF-a) that showed evidence of a causal association with COVID-19 outcomes, asthma, COPD or pulmonary function indices in the main (cis-pQTL) MR analyses, records on clinical drug development programs were identified for IL-18, IL-1ra, MCSF, sICAM, sVCAM, PDGFbb, and TNF-a, and some of the drugs have already been marketed. Some of the indications and associated conditions for the investigational and approved drugs are skin disorders, multiple sclerosis, and acute lymphoblastic leukemia (Table S4).

Colocalization

Overall, the findings from colocalization analyses supported most of the FDR significant associations. Poor evidence for shared causal variants was found for MCSF with risk of hospitalization due to COVID-19, CTACK with risk of COPD, IL-1ra with FVC, PDGFbb with FEV1, and sICAM with FEV1/FVC, which might indicate genetic confounding in those associations or be a result of limited power (weak signals) (Tables S5 and S6).

Secondary traits associated with selected instruments and sensitivity analyses

Removal of the variants associated with anthropometric indices and height in the PhenoScanner database18 (Table S7) from the cytokine genetic instruments that showed evidence of associations with any of the examined outcomes, did not alter the results; the only exceptions were the associations between MCSF and FVC, which did not remain significant, and between TNF-a and atopic asthma, for which the only instrument was associated with height, weight and hip circumference (Table S8). In sensitivity analyses adjusting for the slight correlation among variants (r2 < 0.1), no differences were observed (Table S9). Additionally, iterative leave-one-out analysis showed that no association was driven by influential SNPs (Figures S1–S15). The secondary analysis using the eQTL cytokines in lung tissue produced qualitatively similar results to the main eQTL analysis, but associations were nominally significant only for IL-18 in relation to FEV1, FVC, and COPD, and IL-16 in relation to atopic asthma (Table S10).

Observational analysis in the UK biobank

The positive association of IL-18 on COPD (OR: 1.31, 95%CI: 1.21 to 1.42, p < 0.001), and the inverse associations of IL-18 on FEV1 (−0.061, 95%CI: −0.071 to −0.051, p < 0.001) and FVC (−0.079, 95%CI: −0.093 to −0.065, p < 0.001), were replicated in an observational analysis in the UK Biobank. The inverse association of MCSF on FVC (−0.129, 95%CI: −0.152 to −0.105, p < 0.001) was also replicated. The rest of the associations were not replicated either because they were non-significant, or because they were in the opposite direction compared to the MR analysis (Table S11).

Discussion

This hypothesis-free exploratory analysis provides genetic evidence implicating several inflammatory cytokines in respiratory disease risk. The combined consideration of genetic variants that strongly associate with cytokine protein and gene expression maximizes the potential to identify genetic proxies of cytokine effects, with the exploration of effects on measures of respiratory function, respiratory disease, and COVID-19 serving to identify distinct roles across respiratory disease phenotypes.

Regarding COVID-19 risk, we found evidence supporting the protective effects of MCSF, sICAM, and sVCAM. In contrast, a previous conventional epidemiological association study identified associations of sICAM and sVCAM with worse outcomes in COVID-19.19 The discrepancy with our current findings may be attributable to the potential for the previously described epidemiological associations to be affected by confounding factors and reverse causation, while MR is less vulnerable to such sources of spurious associations. However, an alternative and maybe more plausible explanation is that genetically predicted high levels of these cytokines indicate a genetic profile for a strong immune response that may protect from severe disease; but high directly measured cytokine levels in an observational setting indicate the severity of the COVID-19 or the percentage of the lung being involved in the infection. The evidence supporting the protective effect of MCSF is consistent with its biological role in the differentiation of macrophages. A recent MR study found very similar results for COVID-19 outcomes with our work, using a different GWAS for cytokines.20

Considering lung function, IL-1ra is known to modulate IL1-driven inflammation, thus improving FEV1 as suggested by our current genetic analyses. Perhaps a surprising finding in our analyses was the evidence for higher circulating MCP3 levels increasing FEV1 and FVC, as this chemokine is implicated in the recruitment of inflammatory cells including eosinophils to the lung,21 which would be expected to worsen respiratory function. This discrepancy may be reconciled to its effects being locally exerted. Thus, as its circulating levels are increased, this may paradoxically have favorable effects on reducing eosinophil recruitment in the lung environment. IL-18 is an inflammatory cytokine that has previously been implicated in the pathophysiology of COPD.22,23,24 The current genetic and observational analyses support this harmful effect and further identify similar potential effects on reducing FEV1 and FVC, thus additionally highlighting potential disease mechanisms.

The methodological approach taken in these genetic analyses also offers broader mechanistic insights. The complementary consideration of circulating cytokine levels as well as cross-tissue gene expression provides evidence for the role of sICAM and sVCAM in COVID-19 and IL-1ra in FEV1. Further, by incorporating distinct phenotypic traits reflecting various aspects of respiratory function, we were able to unravel specific and widespread effects of cytokines on each of these. Demonstrating this, the consistent evidence for sICAM across the considered COVID-19 phenotypes supports its role in this disease. Furthermore, the opposite effects of sICAM on the risks of COVID-19 and atopic asthma (inverse and positive, respectively), reinforce the negative association between these diseases found in a previous MR study.25

Finally, our triangulation of genetic evidence with existing observational associations provides complementary insight through methodological approaches that vary in their underlying modeling assumptions, thus strengthening the overall evidence for clinical pursuit.26 The presence of existing clinical drug development programs for IL-18, IL-1ra, MCSF, sICAM, sVCAM, PDGFbb, and TNF-a may support the prioritization of these targets for exploration in the remit of chronic respiratory disease.

Limitations of the study

Our work also has limitations. We examined the associations of 35 inflammatory cytokines with respiratory diseases and pulmonary function indices; however, these constitute a small portion of the total number of inflammatory cytokines, and future studies could elucidate associations with further inflammatory cytokines, as their genetic determinants are better researched. The genetic variants used to proxy the perturbation of cytokine effects reflect small, lifelong changes, which may not mimic the effect of a discrete pharmacological intervention in adult life. The availability of genetic variants to serve as instruments for some of the considered cytokines was limited, thus further robust MR analyses could not be pursued. Furthermore, some of the colocalization analyses were likely underpowered and were thus not able to reliably provide any assurance against confounding by LD. We used data for European ancestry individuals, hence our results are not generalizable to other ethnicities. Additionally, previous relevant epidemiological data are relatively scarce; therefore, the validation of our results with further studies is needed. Finally, there remains the prospect that any identified associations of genetic variants proxying cytokine levels may be related to pleiotropic effects unrelated to cytokine effects, thus violating the requisite MR modeling assumptions. While the use of variants at the genes coding for the respective cytokines that are related to their circulating protein levels or gene expression provides some assurances against this, it does remain a possibility. The observational analyses can capture different components of the associations (largely reflecting protein concentrations cross-sectionally) compared to MR (that proxy the effects of the genetically predicted concentrations), and the observational analyses are prone to biases such as confounding or reverse causation that may explain the observed differences in findings from the two approaches.

In conclusion, we have leveraged large-scale genetic association data related to cytokine protein and gene expression to gain novel mechanistic insight into the role of these inflammatory mediators in chronic lung diseases, lung function, and COVID-19. Our findings can be used to prioritize therapeutic targets for further study across respiratory disease areas, potentially increasing the probability of successful drug development, while also reducing the associated time and cost.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Konstantinos K. Tsilidis (k.tsilidis@imperial.ac.uk).

Materials availability

The study did not generate any new materials.

Data and code availability

• All data used in this work are presented in the online data supplement and are available in the original publications.

• The code used for this paper is available in Supplementary Text.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

• The paper has been written in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology using Mendelian Randomization (STROBE-MR) guidelines (Supplementary Text).27

Acknowledgments

This study is co-financed by the 10.13039/501100008530 European Regional Development Fund of the European Union and Greek national funds through the Operational Program “Competitiveness, Entrepreneurship and Innovation (EPAnEK), NSRF 2014-2020 (Project code 10.13039/501100013296 MIS : OΠΣ 5047228 ).” The Young Finns Study has been financially supported by the 10.13039/501100002341 Academy of Finland : grants 322098 , 286284 , 134309 (Eye), 126925 , 121584 , 124282 , 255381 , 256474 , 283115 , 319060 , 320297 , 314389 , 338395 , 330809 , and 104821 , 129378 (Salve), 117797 (Gendi), and 141071 (Skidi); the 10.13039/501100002327 Social Insurance Institution of Finland ; Competitive State Research Financing of the Expert Responsibility area of Kuopio, Tampere and Turku University Hospitals (grant X51001 ); 10.13039/501100004037 Juho Vainio Foundation ; 10.13039/501100008484 Paavo Nurmi Foundation ; 10.13039/501100005633 Finnish Foundation for Cardiovascular Research ; 10.13039/501100003125 Finnish Cultural Foundation ; The Sigrid Juselius Foundation; 10.13039/501100006706 Tampere Tuberculosis Foundation ; 10.13039/501100004756 Emil Aaltonen Foundation ; 10.13039/100010114 Yrjö Jahnsson Foundation ; 10.13039/501100004325 Signe and Ane Gyllenberg Foundation ; Diabetes Research Foundation of Finnish Diabetes Association; EU Horizon 2020 (grant 755320 for TAXINOMISIS and grant 848146 for To Aition); 10.13039/501100000781 European Research Council (grant 742927 for MULTIEPIGEN project); Tampere University Hospital Supporting Foundation, Finnish Society of Clinical Chemistry and the Cancer Foundation Finland. NFBC1966 31-year follow up that included cytokine measurements received financial support from 10.13039/501100006196 University of Oulu Grant no. 65354 , Oulu University Hospital Grant no. 2/97 , 8/97 , Ministry of Health and Social Affairs Grant no. 23/251/97 , 160/97 , 190/97 , National Institute for Health and Welfare, Helsinki Grant no. 54121 , Regional Institute of Occupational Health, Oulu, Finland Grant no. 50621 , 54231 . The data generation, curation and manpower were also supported by the following EU H2020 grants: DynaHEALTH (grant no 633595 ), LifeCycle (733206 ), LongITools (873749 ), EarlyCause (848158 ), EDCMET (825762 ) and the 10.13039/501100000265 Medical Research Council, UK : grant number MRC/BBSRC MR/S03658X/1 (JPI HDHL H2020 ).

Author contributions

K.K.T., M-R.J., and A.D. conceived and designed the study. M.O.R., E.B., and A-G.A. performed the statistical analyses. M.O.R. and E.B. have directly accessed and verified the underlying data. All authors had full access to all the data in the study and interpreted the results. M.O.R. and D.G. drafted the article. All authors critically revised the article for intellectual content. All authors read and approved the final article. All authors accept responsibility to submit for publication.

Declaration of interests

The authors declare the following competing interests: Salomaa Veikko was funded by the Finnish Foundation for Cardiovascular Research and by the Juho Vainio Foundation. Salomaa Veikko has received an honorarium from Sanofi for consulting and has ongoing research collaboration with Bayer Ltd. (All outside the present study). These companies had no role in study design, data collection and analysis, decision to publish, or preparation of the article. Sirpa Jalkanen, Marko Salmi: Financial support for cytokine analyses has come from the Finnish Academy. Abril Izquierdo and Martin Tobin were supported for the present article by Wellcome Trust grants (WT202849/Z/16/Z, WT225221/Z/22/Z) and the National Institute of Health and Care Research (NIHR) Leicester Biomedical Research Center. Martin Tobin has also received an NIHR Senior Investigator Award and had a funded research collaboration with Orion Pharma (the latter outside the scope of this work). Markku Timonen has received payment for one lecture for Otsuka Pharmaceutical and payment for two lectures for H. Lundbeck A/S. Marjo-Ritta Jarvelin is partly funded by the Medical Research Council (MR/S019669/1). James Yarmolinsky is supported by a Cancer Research UK Population Research Postdoctoral Fellowship (C68933/A28534). The remaining authors have no competing interests to declare.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER	
Deposited data	
	
Cytokines	Sun et al.15	https://pubmed.ncbi.nlm.nih.gov/29875488/	
Cytokines	Karhunen et al.16	https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9978757/	
Cytokines	Folkersen et al.14	https://pubmed.ncbi.nlm.nih.gov/33067605/	
COVID-19 outcomes	The COVID-19 host genetics initiative	https://www.covid19hg.org/	
Asthma	Zhu et al.28	https://pubmed.ncbi.nlm.nih.gov/31669095/	
COPD	Sakornsakolpat et al.29	https://pubmed.ncbi.nlm.nih.gov/30804561/	
Pulmonary function indices	Shrine et al.30	https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6397078/	
	
Software and algorithms	
	
R	version 4.0.2.	https://www.r-project.org	
MendelianRandomization	version 0.6.0.	https://cran.r-project.org/	
coloc	version 5.2.3.	https://cran.r-project.org/	

Experimental model and study participant details

Cytokine instrument selection

We used summary genetic association estimates from a GWAS meta-analysis of circulating levels of 47 inflammatory cytokines from the Northern Finland Birth Cohort 1966 (NFBC1966),17 the Cardiovascular Risk in Young Finns (YFS) study, FINRISK 1997 and 2002, the INTERVAL study and the SCALLOP consortium,13,14,15,16 including up to 31,112 individuals (ranging from 3,301 to 31,112) (Table S12; Figure 5). We present the summary data of this GWAS in Table S13 and we provide more details of the GWAS meta-analysis rationale and analytical plan in a recent publication.31Figure 5 Analysis plan

Abbreviations: cis-pQTL, cis-protein quantitative trait locus; cis-eQTL, cis-expression quantitative trait locus; LD, linkage disequilibrium; GTEx, Genotype-Tissue Expression database; COPD, chronic obstructive pulmonary disease; FEV1, forced expiratory volume measured in the first second of exhalation; FVC, forced vital capacity; and MR-IVW, Mendelian randomization Inverse Variance Weighted method.

We used cis instrument definitions, selecting genetic variants located in or close to the coding gene, thus we minimized the probability of horizontal pleiotropy (when variants influence the outcomes of interest through pathways other than the cytokines of interest) and we provided association estimates less likely to be affected by this bias.10,16,32 In the main instrument definition, we selected genetic variants with a minor allele frequency (MAF) > 0.05, within ±500 kilobases (kb) of the corresponding cytokine gene locus, associated with the circulating cytokine concentrations at p < 1 × 10−4; we termed these genetic variants ‘cis-protein quantitative trait locus’ (cis-pQTL) (Table S13). To validate the main analysis and potentially capture additional associations we used a ‘cis-expression quantitative trait locus’ (cis-eQTL) definition, selecting genetic variants within ±500kb of the corresponding cytokine gene locus, associated with both its gene expression aggregated across tissues at p < 1×10-4 (using data from Genotype-Tissue Expression (GTEx) database),33 and its circulating cytokine concentrations at p < 0.05.16 In a secondary analysis and to validate the results of the eQTL analysis, we repeated the analysis using a slightly altered eQTL definition, selecting genetic variants using the same criteria but expressed only in lung tissue.

To avoid the potential loss of causal variants via clumping, we used a lenient clumping r2 threshold of 0.1, and we evaluated the robustness of the significant associations in sensitivity analysis, accounting for the LD structure.34 More details for the methods used are provided in Supplementary Text.

Outcome data

COVID-19

We used summary genetic association estimates from the COVID-19 host genetics initiative (https://www.covid19hg.org/results/) for the risks of four COVID-19 outcomes, namely (i) Any COVID-19, (ii) COVID-19 hospitalization, (iii) Severe COVID-19, (iv) Case only COVID-19 hospitalization, using GWAS data for European ancestry individuals. More details on the COVID-19 outcomes are provided in Supplementary Text.

Asthma, COPD and pulmonary function indices

Genetic association estimates for asthma were obtained from a UK Biobank GWAS, including 24,600 cases and 432,368 controls of European ancestry for atopic asthma, and 29,605 cases and 319,321 controls for non-atopic asthma.28 Estimates for COPD were selected from a GWAS in 35,735 cases and 222,076 controls, using data from 25 studies (including UK Biobank and studies from the International COPD Genetics Consortium).29 Genetic association estimates for indices of pulmonary function, namely FEV1, FVC, FEV1/FVC and PEF, were obtained from a GWAS in individuals of European ancestry (for FEV1, FVC and FEV1/FVC 321,047 samples were available from UK Biobank and 79,055 from the SpiroMeta Consortium; for PEF up to 24,218 samples from SpiroMeta).30

Ethics approval and consent to participate

All studies contributing data to these analyses had the relevant institutional review board approval from each country, in accordance with the Declaration of Helsinki, and all participants provided informed consent.

Method details

Mendelian randomization analyses

We used the inverse-variance weighted (IVW) MR method35 in the case of multiple genetic instruments or the ratio of coefficients method in the case of a single instrument, to investigate the associations of genetically proxied circulating cytokine concentrations in relation to the outcomes. To account for multiple comparisons we estimated the FDR adjusted p values (q-values), as proposed by Benjamini and Hochberg, and considered q-values of less than 5% as statistically significant.36

We assessed the MR assumptions, namely (i) relevance (genetic variants strongly associated with the cytokine concentrations), (ii) exchangeability (no common causes of the instrument-outcome association), and (iii) exclusion restriction (genetic variants only influence outcomes via the inflammatory cytokines) using several approaches. To examine the first assumption, we calculated the F-statistic and proportion of variance explained (r2) that measure the strength of each genetic variant, and excluded weak instruments (i.e. with F-statistic<10) from the analysis (Table S13).37,38 To assess the second and third MR assumptions, we searched the PhenoScanner database18 to identify secondary traits (non-inflammatory, potentially pleiotropic traits) associated with the genetic instruments in previous GWAS. The most common violation of the third MR assumption is horizontal pleiotropy that occurs when a variant influences the outcome through phenotypes other than the exposure being instrumented. We have, in part, adjusted for horizontal pleiotropy “by design” with our cis genetic instrument definition. We further used additional MR analyses that are more robust to violations of the exclusion restriction assumption, namely the weighted-median,39 MR-Egger,40 and MR-PRESSO method,41 with the caveat that these methods operate best in a polygenic MR analysis framework.

Colocalization and leave-one-out analysis

We performed colocalization analyses to investigate whether the observed MR associations may have resulted from confounding by LD (i.e. exposure and outcome not having shared causal variants).42 This method is described in detail in Supplementary Text. To examine whether associations were driven by a single influential SNP in the associations that survived colocalization analysis we performed iterative leave-one-out analysis, iteratively removing one SNP at a time from instruments.

Review of publicly available databases for medication

For the cytokines that emerged via the MR analyses as potentially relevant to our outcomes, we searched the DrugBank database (https://go.drugbank.com/) to identify drugs targeting them, and for the identified drugs, we searched the widely used clinical trial protocol repository of clinicaltrials.gov (https://clinicaltrials.gov/).

Observational analysis in the UK biobank

The associations that emerged in the MR analyses were explored in the UK Biobank,43 which is an ongoing prospective cohort study, which enrolled 502,412 participants aged 40 to 69 years in the UK between 2006 and 2010. Blood plasma samples were randomly selected from 54,306 UK Biobank participants, capturing 1,463 unique proteins.44 The association of circulating plasma protein concentrations and pertinent outcomes was evaluated using regression models. Further details on the analysis are presented in Supplementary Text.

Quantification and statistical analysis

All analyses were performed using R, version 4.0.2.45 Two-sample MR analyses were conducted using the MendelianRandomization package (version 0.6.0.). Colocalization analyses were conducted using the coloc package (version 5.2.3). Figures were constructed using the forestplot (version 3.1.3.) and ggplot2 packages (version 3.4.3) (available from the R CRAN repository (https://cran.r-project.org/)).

MR estimates reflect the change in outcome per one standard deviation (SD) higher concentration in the natural scale of each cytokine.

Supplemental information

Document S1. Figures S1–S15 and Data S1–S3

Table S1. List of cytokines, related to STAR Methods

Table S2. Summary of cis-pQTL results, related to Figures 1 and 3

Table S3. Summary of cis-eQTL results, related to Figures 2 and 4

Table S4. Druggable targets, related to STAR Methods

Table S5. Colocalization analysis of significant cis-pQTLs and cis-eQTLs, related to STAR Methods

Table S6. Colocalization with SuSie, related to STAR Methods

Table S7. Associated traits from Phenoscanner, related to STAR Methods

Table S8. Sensitivity analysis excluding variants associated with somatometric indices or height in Phenoscanner, related to STAR Methods

Table S9. Sensitivity analysis adjusting for the genetic correlation among variants, related to STAR Methods

Table S10. Sensitivity analysis for eQTL cytokines in lung tissue, related to STAR Methods

Table S11. Summary of the results from observational analysis in the UK Biobank, related to STAR Methods

Table S12. Source of GX, related to STAR Methods

Table S13. GX instruments, related to STAR Methods

Supplemental information can be found online at https://doi.org/10.1016/j.isci.2024.110704.
==== Refs
References

1 GBD 2017 Disease and Injury Incidence and Prevalence Collaborators Global, regional, and national incidence, prevalence, and years lived with disability for 354 diseases and injuries for 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017 Lancet (London, England) 392 2018 1789 1858 10.1016/S0140-6736(18)32279-7 30496104
2 GBD Chronic Respiratory Disease Collaborators Prevalence and attributable health burden of chronic respiratory diseases, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017 Lancet Respir. Med. 8 2020 585 596 10.1016/S2213-2600(20)30105-3 32526187
3 Hole D.J. Watt G.C. Davey-Smith G. Hart C.L. Gillis C.R. Hawthorne V.M. Impaired lung function and mortality risk in men and women: findings from the Renfrew and Paisley prospective population study BMJ 313 1996 711 716 10.1136/bmj.313.7059.711 8819439
4 COVID-19 Excess Mortality Collaborators Estimating excess mortality due to the COVID-19 pandemic: a systematic analysis of COVID-19-related mortality, 2020-21 Lancet (London, England) 399 2022 1513 1536 10.1016/S0140-6736(21)02796-3 35279232
5 Fajgenbaum D.C. June C.H. Cytokine Storm N. Engl. J. Med. 383 2020 2255 2273 10.1056/NEJMra2026131 33264547
6 Aghasafari P. George U. Pidaparti R. A review of inflammatory mechanism in airway diseases Inflamm. Res. 68 2019 59 74 10.1007/s00011-018-1191-2 30306206
7 RoActemra | European Medicines Agency [Internet]. https://www.ema.europa.eu/en/medicines/human/EPAR/roactemra.
8 Moore T.J. Zhang H. Anderson G. Alexander G.C. Estimated Costs of Pivotal Trials for Novel Therapeutic Agents Approved by the US Food and Drug Administration, 2015-2016 JAMA Intern. Med. 178 2018 1451 1457 10.1001/jamainternmed.2018.3931 30264133
9 Burgess S. Butterworth A. Malarstig A. Thompson S.G. Use of Mendelian randomisation to assess potential benefit of clinical intervention BMJ 345 2012 e7325 10.1136/bmj.e7325
10 Gill D. Georgakis M.K. Walker V.M. Schmidt A.F. Gkatzionis A. Freitag D.F. Finan C. Hingorani A.D. Howson J.M.M. Burgess S. Mendelian randomization for studying the effects of perturbing drug targets Wellcome Open Res. 6 2021 16 10.12688/wellcomeopenres.16544.2 33644404
11 Zuber V. Grinberg N.F. Gill D. Manipur I. Slob E.A.W. Patel A. Wallace C. Burgess S. Combining evidence from Mendelian randomization and colocalization: Review and comparison of approaches Am. J. Hum. Genet. 109 2022 767 782 10.1016/j.ajhg.2022.04.001 35452592
12 Rontogianni M. Gill D. Bouras E. Asimakopoulos A.G. Tzoulaki I. Karhunen V. Lehtimäki T. Raitakari O. Wielscher M. Salomaa V. Association of Inflammatory Cytokines with Lung Function, Chronic Lung Diseases and COVID-19 2023 Poster presentation in: 11th Annual Meeting of the International Cytokine & Interferon Society
13 Sliz E. Kalaoja M. Ahola-Olli A. Raitakari O. Perola M. Salomaa V. Lehtimäki T. Karhu T. Viinamäki H. Salmi M. Genome-wide association study identifies seven novel loci associating with circulating cytokines and cell adhesion molecules in Finns J. Med. Genet. 56 2019 607 616 10.1136/jmedgenet-2018-105965 31217265
14 Folkersen L. Gustafsson S. Wang Q. Hansen D.H. Hedman Å.K. Schork A. Page K. Zhernakova D.V. Wu Y. Peters J. Genomic and drug target evaluation of 90 cardiovascular proteins in 30,931 individuals Nat. Metab. 2 2020 1135 1148 10.1038/s42255-020-00287-2 33067605
15 Sun B.B. Maranville J.C. Peters J.E. Stacey D. Staley J.R. Blackshaw J. Burgess S. Jiang T. Paige E. Surendran P. Genomic atlas of the human plasma proteome Nature 558 2018 73 79 10.1038/s41586-018-0175-2 29875488
16 Karhunen V. Gill D. Huang J. Bouras E. Malik R. Ponsford M.J. Ahola-Olli A. Papadopoulou A. Palaniswamy S. Sebert S. The interplay between inflammatory cytokines and cardiometabolic disease: bi-directional mendelian randomisation study BMJ Med. 2 2023 e000157 10.1136/bmjmed-2022-000157
17 University of Oulu. Northern Finland Birth Cohort 1966. http://urn.fi/urn:nbn:fi:att:bc1e5408-980e-4a62-b899-43bec3755243..
18 Staley J.R. Blackshaw J. Kamat M.A. Ellis S. Surendran P. Sun B.B. Paul D.S. Freitag D. Burgess S. Danesh J. PhenoScanner: a database of human genotype-phenotype associations Bioinformatics 32 2016 3207 3209 10.1093/bioinformatics/btw373 27318201
19 Tong M. Jiang Y. Xia D. Xiong Y. Zheng Q. Chen F. Zou L. Xiao W. Zhu Y. Elevated Expression of Serum Endothelial Cell Adhesion Molecules in COVID-19 Patients J. Infect. Dis. 222 2020 894 898 10.1093/infdis/jiaa349 32582936
20 Baranova A. Luo J. Fu L. Yao G. Zhang F. Evaluating the effects of circulating inflammatory proteins as drivers and therapeutic targets for severe COVID-19 Front. Immunol. 15 2024 1352583 10.3389/fimmu.2024.1352583
21 Blease K. Lukacs N.W. Hogaboam C.M. Kunkel S.L. Chemokines and their role in airway hyper-reactivity Respir. Res. 1 2000 54 61 10.1186/rr13 11667966
22 Imaoka H. Hoshino T. Takei S. Kinoshita T. Okamoto M. Kawayama T. Kato S. Iwasaki H. Watanabe K. Aizawa H. Interleukin-18 production and pulmonary function in COPD Eur. Respir. J. 31 2008 287 297 10.1183/09031936.00019207 17989120
23 Guo P. Li R. Piao T.H. Wang C.L. Wu X.L. Cai H.Y. Pathological Mechanism and Targeted Drugs of COPD Int. J. Chron. Obstruct. Pulmon. Dis. 17 2022 1565 1575 10.2147/COPD.S366126 35855746
24 Kang M.J. Homer R.J. Gallo A. Lee C.G. Crothers K.A. Cho S.J. Rochester C. Cain H. Chupp G. Yoon H.J. Elias J.A. IL-18 is induced and IL-18 receptor alpha plays a critical role in the pathogenesis of cigarette smoke-induced pulmonary emphysema and inflammation J. Immunol. 178 2007 1948 1959 10.4049/jimmunol.178.3.1948 17237446
25 Baranova A. Cao H. Chen J. Zhang F. Causal Association and Shared Genetics Between Asthma and COVID-19 Front. Immunol. 13 2022 705379 10.3389/fimmu.2022.705379
26 Lawlor D.A. Tilling K. Davey Smith G. Triangulation in aetiological epidemiology Int. J. Epidemiol. 45 2016 1866 1886 10.1093/ije/dyw314 28108528
27 Skrivankova V.W. Richmond R.C. Woolf B.A.R. Yarmolinsky J. Davies N.M. Swanson S.A. VanderWeele T.J. Higgins J.P.T. Timpson N.J. Dimou N. Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization: The STROBE-MR Statement JAMA 326 2021 1614 1621 10.1001/jama.2021.18236 34698778
28 Zhu Z. Guo Y. Shi H. Liu C.L. Panganiban R.A. Chung W. O'Connor L.J. Himes B.E. Gazal S. Hasegawa K. Shared genetic and experimental links between obesity-related traits and asthma subtypes in UK Biobank J. Allergy Clin. Immunol. 145 2020 537 549 10.1016/j.jaci.2019.09.035 31669095
29 Sakornsakolpat P. Prokopenko D. Lamontagne M. Reeve N.F. Guyatt A.L. Jackson V.E. Shrine N. Qiao D. Bartz T.M. Kim D.K. Genetic landscape of chronic obstructive pulmonary disease identifies heterogeneous cell-type and phenotype associations Nat. Genet. 51 2019 494 505 10.1038/s41588-018-0342-2 30804561
30 Shrine N. Guyatt A.L. Erzurumluoglu A.M. Jackson V.E. Hobbs B.D. Melbourne C.A. Batini C. Fawcett K.A. Song K. Sakornsakolpat P. New genetic signals for lung function highlight pathways and chronic obstructive pulmonary disease associations across multiple ancestries Nat. Genet. 51 2019 481 493 10.1038/s41588-018-0321-7 30804560
31 Bouras E. Karhunen V. Gill D. Huang J. Haycock P.C. Gunter M.J. Johansson M. Brennan P. Key T. Lewis S.J. Circulating inflammatory cytokines and risk of five cancers: a Mendelian randomization analysis BMC Med. 20 2022 3 10.1186/s12916-021-02193-0 35012533
32 Schmidt A.F. Finan C. Gordillo-Marañón M. Asselbergs F.W. Freitag D.F. Patel R.S. Tyl B. Chopade S. Faraway R. Zwierzyna M. Hingorani A.D. Genetic drug target validation using Mendelian randomisation Nat. Commun. 11 2020 3255 10.1038/s41467-020-16969-0 32591531
33 GTEx Consortium The GTEx Consortium atlas of genetic regulatory effects across human tissues Science (New York, N.Y.) 369 2020 1318 1330 10.1126/science.aaz1776 32913098
34 Gkatzionis A. Burgess S. Newcombe P.J. Statistical methods for cis-Mendelian randomization with two-sample summary-level data Genet. Epidemiol. 47 2023 3 25 10.1002/gepi.22506 36273411
35 Burgess S. Butterworth A. Thompson S.G. Mendelian randomization analysis with multiple genetic variants using summarized data Genet. Epidemiol. 37 2013 658 665 10.1002/gepi.21758 24114802
36 Benjamini Y. Hochberg Y. Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing J. Roy. Stat. Soc. B 57 1995 289 300 http://www.jstor.org/stable/2346101
37 Palmer T.M. Lawlor D.A. Harbord R.M. Sheehan N.A. Tobias J.H. Timpson N.J. Davey Smith G. Sterne J.A.C. Using multiple genetic variants as instrumental variables for modifiable risk factors Stat. Methods Med. Res. 21 2012 223 242 10.1177/0962280210394459 21216802
38 Brion M.J.A. Shakhbazov K. Visscher P.M. Calculating statistical power in Mendelian randomization studies Int. J. Epidemiol. 42 2013 1497 1501 10.1093/ije/dyt179 24159078
39 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 2016 304 314 10.1002/gepi.21965 27061298
40 Bowden J. Davey Smith G. Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression Int. J. Epidemiol. 44 2015 512 525 10.1093/ije/dyv080 26050253
41 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 2018 693 698 10.1038/s41588-018-0099-7 29686387
42 Kanduri C. Bock C. Gundersen S. Hovig E. Sandve G.K. Colocalization analyses of genomic elements: approaches, recommendations and challenges Bioinformatics 35 2019 1615 1624 10.1093/bioinformatics/bty835 30307532
43 UK Biobank - UK Biobank. https://www.ukbiobank.ac.uk/.
44 Sun B.B. Chiou J. Traylor M. Benner C. Hsu Y.H. Richardson T.G. Surendran P. Mahajan A. Robins C. Vasquez-Grinnell S.G. Genetic Regulation of the Human Plasma Proteome in 54,306 UK Biobank Participants Preprint at bioRxiv 2022 10.1101/2022.06.17.496443
45 R Development Core Team. R: A language and environment for statistical computing. https://www.r-project.org.
