Genetically determined circulating levels ofcytokines and the risk of rheumatoid arthritis

doi:10.21203/rs.3.rs-80168/v1

Download PDF

Research article

Genetically determined circulating levels ofcytokines and the risk of rheumatoid arthritis

https://doi.org/10.21203/rs.3.rs-80168/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background: Accumulation of inflammatory leukocytes in articular tissues is the hallmark feature of rheumatoid arthritis (RA). Increasing evidence from observational studies have suggested that several cytokines are relevant in the disease process of RA. However, traditional observational studies are susceptible to bias from confounding and reverse causation; therefore, the potential causal association of individual cytokines with the risk of RA remains elusive. Our study aimed to evaluate whether genetically determined circulating levels of cytokines are associated with the risk of RA by conducting a two-sample Mendelian randomization (MR) study.

Methods: We identified single nucleotide polymorphisms (SNPs) associated with circulating levels of cytokines and growth factors from a genome-wide association study (GWAS) including 8,293 participants of European ancestry as instrumental variables (IVs). The association estimates of these IVs with RA were obtained from a GWAS meta-analysis including 14,361 RA cases and 43,923 controls of European ancestry. MR analyses were performed using the random-effects inverse variance-weighted, weighted-median, MR-Egger and MR pleiotropy residual sum and outlier tests. Sensitivity analyses were further performed using restricted IVs excluding potential pleiotropic SNPs.

Results: In the primary MR analysis, a total of 270 SNPs associated with circulating levels of 27 cytokines were identified and used as IVs. We found a suggestive inverse association between genetically determined circulating level of macrophage inflammatory protein-1β (MIP-1b) and the risk of RA [odds ratio (OR): 0.97, 95% confidence interval (CI) = 0.92-0.99, P = 0.016]. The association remained statistically significant in alternative MR analyses, and the causal effect estimate from sensitivity analysis using the restricted IVs was similar.

Conclusions: Genetically determined elevated circulating level of MIP-1b was associated with a lower risk of RA. Further studies are warranted to determine how this MIP-1b and related pathways contribute to the development of RA.

Orthopedics

Rheumatology

Cytokine

genome-wide association study

macrophage inflammatory protein-1β

Mendelian randomization

rheumatoid arthritis

single nucleotide polymorphisms.

Rheumatoid arthritis (RA) is a chronic systemic inflammatory disease affecting approximately 1% of the global population (1). Lymphocytic promotion of auto-antibodies and inflammatory cytokines are central to its pathology. Genome-wide association studies (GWASs) have confirmed RA-associated single nucleotide polymorphisms (SNPs) enrichment in or related to genes active in CD4⁺ T cells (2). Improved understanding in these areas have translated to successful therapies that have revolutionized its treatment. Targeted immune therapies, such as tumor necrosis factor (TNF) inhibitors, have been widely used in RA treatment (3). However, less than a quarter of patients achieve remission at 6 months (4), and up to 10% remain “resistant” to multiple conventional synthetic and biologic disease modifying antirheumatic drugs (DMARDs) (5). Alternative cytokine-pathways (and therefore potential therapies) may be relevant in these patients.

Accumulation of inflammatory leukocytes in articular tissues is the hallmark feature of RA. Why articular tissues are preferentially affected, and the process of cellular recruitment into the inflamed tissue, remains unclear. Increasing evidence from observational studies have suggested that several cytokines, such as monocyte chemoattractant protein-1 and interleukin-1 α, are relevant in the disease process (6, 7). However, observational studies are susceptible to bias from confounding and reverse causation; therefore, the potential causal association of individual cytokines with the risk of RA remains elusive.

Mendelian randomization (MR) is a method that uses genetic variants as instrumental variables (IVs) for the exposures and assesses their associations with the outcomes (8). Unlike observational epidemiological studies, the use of MR analyses can reduce biases, such as reverse causation (because genetic variant are not affected by diseases) and residual confounding (because alleles are randomly assigned to offspring) (8). Comprehensive data on a range of phenotypes provided by GWASs enables the exploration of associations for which there is a paucity of observational evidence, as is the case for cytokines and RA.

In this study, we evaluated the potential causal associations between circulating levels of cytokines and the risk of RA by conducting a two-sample MR study in a large GWAS meta-analysis of 14,361 participants with RA and 43,923 control participants (2).

Study design

The overall design of the study is summarized in Fig. 1. Briefly, this study consisted of three sequential steps. First, we identified SNPs associated with circulating levels of cytokines to be used as IVs from summary statistics of 8,293 participants of Finnish ancestry (9). Second, we performed a series of MR analyses to explore the association between circulating levels of individual cytokines and the risk of RA using the summary statistics including 58,284 participants of European ancestry (2). Third, we scanned for potential secondary phenotypes (e.g. platelet count and white blood cell count) associated with each IV, to assess the effect of pleiotropy (i.e., genetic variants influencing RA risk through alternative pathways from the exposure of interest), and reran the MR analyses using updated IVs excluding potentially pleiotropic SNPs. Since all MR analyses were based on the publicly summary statistics, no ethic approval or consent to human participants was required.

Source of Outcomes

The descriptive characteristics of the GWAS summary statistics used for our MR analyses are listed in detail in Supplementary Table 1 along with the links for downloading these statistics. Briefly, summary statistics for RA were obtained from a GWAS meta-analysis including 14,361 participants with RA and 43,923 control participants of European ancestry (2). All RA cases met the 1987 RA-related diagnostic criteria of the American College of Rheumatology, or were diagnosed by a board-certified rheumatologist (2).

Selection of SNPs

In our primary analyses, the IVs for circulating levels of cytokines were taken from a GWAS consisting 8,293 Finnish participants (9). Specifically, we selected SNPs reaching genome-wide level of significance (P < 5 × 10^− 8) for their associations with circulating levels of cytokines. Then, for each cytokine, SNPs used as IVs were pruned for linkage disequilibrium (i.e. genetic variants correlation) at stringent threshold (r² < 0.1), and those with a larger P-value for exposure of interest were excluded. We identified a total of 958 SNPs associated with circulating levels of 28 cytokines. We excluded SNPs for which associations of these SNPs and their proxies (SNPs correlated at r² > 0.9) with circulating cytokines were not available, leaving a total of 270 SNPs (including one proxy) as the final IV set. (Supplementary Table 2).

In the secondary analysis, to minimize the effect of potential pleiotropic IVs on causal estimates, we restricted the selection of IVs to exclude potentially pleiotropic SNPs. We first excluded 32 SNPs associated with circulating levels of more than one cytokine at genome-wide significance threshold. We further excluded 16 SNPs showing associations of P < 5 × 10^− 8 with potential confounding traits using the GWAS Catalog (https://www.ebi.ac.uk/gwas/, Accessed on February 5th, 2020). The causal estimates between circulating levels of cytokines and the risk of RA were replicated using the remaining SNPs (n = 222) as the restricted IVs (Supplementary Table 2).

Statistical analysis

After extraction of the association estimates of SNP-cytokine and SNP-RA and harmonization of the direction of estimates by effect alleles, we estimated the proportions of variances in circulating levels of cytokines explained by our primary genetic IVs (R²) (10). F-statistics were calculated to quantify the strength of the IVs using the method previously described (11). We then computed individual MR estimates with the Wald estimator and standard error with the Delta method (12). In our primary analyses, we pooled these individual MR estimates using the random-effects inverse-variance-weighted (IVW) method (12). To assess the robustness of these association estimates, we further performed alternative MR analyses as sensitivity analyses including the weighted-median method, MR-Egger regression and MR pleiotropy residual sum and outlier (MR-PRESSO) test. Specifically, the weighted-median method can provide a valid estimate under the assumption that less than 50% of the weights are invalid (13). Additionally, the intercept of the MR-Egger regression can be used as an indicator of directional pleiotropy (P-value for intercept < 0.05 indicated the presence of directional pleiotropy) (14). The possible pleiotropic outliers can also be identified using the MR-PRESSSO test, and outliers were excluded from the primary genetic IVs and the causal effect estimates were reassessed using the remaining IVs (15).

For all analyses, we corrected for multiple comparisons with the Bonferroni approach and set statistical significance at a P-value < 0.002 (0.05/27) based on the number of cytokines included in the primary MR analyses. Associations above 0.002but P < 0.05, were considered suggestive. All MR analyses were performed in R software (v 3.6.2) using MendelianRandomization and MRPRESSO packages.

The primary MR results from the random-effects IVW analyses for 27 cytokines are presented in Fig. 2. Overall, there were no circulating levels of cytokines showing statistically significant associations with risk of RA after Bonferroni correction for multiple comparisons (P-value < 0.002).

Notably, circulating level of one cytokine, macrophage inflammatory protein-1β (MIP-1b/CCL4), showed suggestive inverse association with the risk of RA [odds ratio (OR): 0.95, 95% confidence interval (CI) = 0.92–0.99, P = 0.016] using the IVW method (Fig. 3). The detailed information of the IVs related to the circulating level of MIP-1b was depicted in Supplementary Table 3.

Additionally, in the weighted-median method, the association remained stable, though with wider confidence intervals (OR: 0.94, 95% CI = 0.88–0.99, P = 0.034). Moreover, there was no evidence showing the presence of directional pleiotropy effects as assessed by the intercept from MR-Egger regression (P for intercept = 0.859), and no outliers were detected with the MR-PRESSO test (Supplementary Table 4).

In sensitivity analyses using the restricted IVs, after excluding one pleiotropic SNP, rs225285, the association estimate of genetically determined circulating level of MIP-1b with the risk of RA remained consistent (OR: 0.95, 95% CI = 0.91–0.99, P = 0.014) (Fig. 3).

In this MR study, we used a two-sample MR approach and systematically screened the potential causal association between circulating levels of cytokines and the risk of RA. Our study found a suggestive inverse association between genetically determined circulating level of MIP-1b and the risk of RA. Moreover, the results for the inverse association were stable in alternative MR analyses and in sensitivity analyses using the restricted IVs. Collectively, these findings provided evidence for a potential protective role of circulating MIP-1b on the risk of RA.

The directionality of the MR association between genetically determined circulating level of MIP-1b and risk of RA we obtained is inconsistent with the results from previous observational epidemiological studies. For example, a case-control study including 14 RA patients and 27 controls found that RA patients did not have a higher serum level of MIP-1b [log transformed level (range): 4.92 (4.19–5.89)] compared to healthy controls [log transformed level (range): 4.74 (3.14–5.72)] (16). While another case-control study including 43 untreated early RA patients and 14 healthy controls found that plasma levels of MIP-1b were significantly higher in RA patients than in healthy controls (P < 0.01) (17). Differences in these findings may be explained by several reasons. First, these observational studies included smaller samples of RA cases and controls, which might result in inadequate statistical power. In addition, the observed associations are susceptible to biases inherent in the observational study design, such as confounding and reverse causality. Our MR study utilized the summary data from published GWAS meta-analysis with the largest sample sizes of RA patients and controls to date including more 58,000 participants, with an estimated F-statistic of 66.07 to overcome the weak instruments bias and achieve sufficient statistical power (91.84%). Moreover, as genetic variants are presumed to be randomly allocated during meiosis and unaffected by the onset of diseases (e.g. RA), confounding factors are anticipated to be equally distributed among different genotypes. Therefore, MR study is more reliable in causal inference, since it minimizes the influence of reverse causation and confounding bias (8).

One fundamental assumption for MR study to ensure a valid causal estimate is that genetic variants used as IVs did not directly act on the target outcome, neither affect the outcome through pathways other than through the exposure (8). To verify this model assumption and to guarantee the validity of results, we first performed a series of MR analyses, such as the weighted-median, MR-Egger regression, and MR-PRESSO tests. The results obtained from these analyses were robust and did not provide evidence of potential pleiotropy. In addition, we manually scanned the SNPs used as IVs for circulating level of MIP-1b for their potential associations with secondary traits. We found that none of them are associated with RA at genome-wide significance level, and the causal estimate remained stable in sensitivity analysis using the restricted IVs excluding pleiotropic SNPs, suggesting the robustness of our findings.

The underlying biological mechanism of MIP-1b in the pathogenesis of RA remains unclear. One possible explanation has been suggested is that MIP-1b is involved in the regulation of Th1 cell trafficking (18). Evidence from a case-control study of 24 RA patients and 22 controls showed that the mean (± SEM) level of MIP-1b was higher in the synovial fluids of RA patients (738 ± 282 pg/ml), while it was lower in the serum of RA patients (189 ± 23 pg/ml), as compared to controls (39 ± 6 pg/ml in synovial fluids; 244 ± 68 pg/ml in serum) (18). The concentration of MIP-1b in synovial tissues and fluids was elevated to traffic Helper T 1 cells into inflamed synovium (18), thus contributed to the development of RA by inducing osteoclastogenesis (19). Further studies were warranted to clarify the potential mechanism of circulating MIP-1b in the pathogenesis of RA.

There were limitations to this study. First, our results may not be generalizable to other populations, since the ancestry of participants included in our MR study was restricted to European. Second, the effects of genetic variations (e.g. elevated MIP-1b level due to CCL4 gene polymorphisms) on normal development may influence the expression of other genes, which in turn compensate for the influence of lifelong higher circulating level of MIP-1b (8). Hence, the likelihood that the MR association reflecting the true potential causal effect might be reduced. Third, since the IVs for several cytokines were not available and the statistical power of genetic associations for some cytokines might be insufficient; it is possible that we might miss potential weak associations of these cytokines with the risk of RA. Further individual level-based MR studies with a prospective design are needed to address these issues.

Our MR study found suggestive evidence that genetically determined elevated circulating level of MIP-1b is associated with a reduced risk of RA. More studies on the potential biological mechanism underlying MIP-1b and risk of RA are warranted to explore whether MIP-1b or related pathways could be potential therapeutic targets in the prevention of RA.

CI: confidence interval; DMARDs, disease modifying antirheumatic drugs; GWAS: genome-wide association studies; IV: instrumental variable; IVW: inverse-variance-weighted; MIP-1b/CCL4: macrophage inflammatory protein-1β; MR: Mendelian randomization; MR-PRESSO:MR Pleiotropy RESidual Sum and Outlier; OR: odds ratio; RA: rheumatoid arthritis; SNP: single nucleotide polymorphism; TNF, tumor necrosis factor.

Ethics approval and consent to participate

Studies participating in the meta-analysis of GWASs have received ethical approval from relevant institutional review boards and informed consent was obtained from each study participant. As we used publicly available summary statistics from the published GWAS meta-analysis, no additional ethical approval was required.

Consent for publication

Not applicable.

Availability of data and materials

Summary statistics for RA are available at http://plaza.umin.ac.jp/~yokada/datasource/software.htm.

Competing interest

The authors declare that they have no competing interests.

Funding

This work was supported by grants of National Natural Science Foundation of China (81973663, 81873269 and 81602917), the National Key R&D Program of China (2018YFC1705500), and the Talent Project of Zhejiang Association for Science and Technology (2018YCGC003). The funders had no role in the study design, data analysis, interpretation of data, or preparation of the manuscript.

Authors' contributions

MY, QY and JX contributed conception and design of the study; QY, ZS, HZ and WC analysed and interpreted the data; QY wrote the first draft of the manuscript; QY, YD, JX and ZS wrote sections of the manuscript. All authors contributed to manuscript revision, read and approved the submitted version.

Acknowledgments

The authors thank the researchers of the previous GWASs to share the summary statistics.

Smolen JS, Aletaha D, Barton A, Burmester GR, Emery P, Firestein GS, et al. Rheumatoid arthritis. Nat Rev Dis Primers. 2018;4:18001.
Okada Y, Wu D, Trynka G, Raj T, Terao C, Ikari K, et al. Genetics of rheumatoid arthritis contributes to biology and drug discovery. Nature. 2014;506(7488):376-81.
McInnes IB, Schett G. Pathogenetic insights from the treatment of rheumatoid arthritis. Lancet. 2017;389(10086):2328-37.
Hetland ML, Christensen IJ, Tarp U, Dreyer L, Hansen A, Hansen IT, et al. Direct comparison of treatment responses, remission rates, and drug adherence in patients with rheumatoid arthritis treated with adalimumab, etanercept, or infliximab: results from eight years of surveillance of clinical practice in the nationwide Danish DANBIO registry. Arthritis Rheum. 2010;62(1):22-32.
de Hair MJH, Jacobs JWG, Schoneveld JLM, van Laar JM. Difficult-to-treat rheumatoid arthritis: an area of unmet clinical need. Rheumatology. 2018;57(7):1135-44.
Rantapaa-Dahlqvist S, Boman K, Tarkowski A, Hallmans G. Up regulation of monocyte chemoattractant protein-1 expression in anti-citrulline antibody and immunoglobulin M rheumatoid factor positive subjects precedes onset of inflammatory response and development of overt rheumatoid arthritis. Ann Rheum Dis. 2007;66(1):121-3.
Jorgensen KT, Wiik A, Pedersen M, Hedegaard CJ, Vestergaard BF, Gislefoss RE, et al. Cytokines, autoantibodies and viral antibodies in premorbid and postdiagnostic sera from patients with rheumatoid arthritis: case-control study nested in a cohort of Norwegian blood donors. Ann Rheum Dis. 2008;67(6):860-6.
Smith GD, Ebrahim S. 'Mendelian randomization': can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol. 2003;32(1):1-22.
Ahola-Olli AV, Wurtz P, Havulinna AS, Aalto K, Pitkanen N, Lehtimaki T, et al. Genome-wide Association Study Identifies 27 Loci Influencing Concentrations of Circulating Cytokines and Growth Factors. Am J Hum Genet. 2017;100(1):40-50.
Park JH, Wacholder S, Gail MH, Peters U, Jacobs KB, Chanock SJ, et al. Estimation of effect size distribution from genome-wide association studies and implications for future discoveries. Nature genetics. 2010;42(7):570-5.
Burgess S, Thompson SG, Collaboration CCG. Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol. 2011;40(3):755-64.
Burgess S, Butterworth A, Thompson SG. Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol. 2013;37(7):658-65.
Bowden J, Davey Smith G, Haycock PC, Burgess S. Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genet Epidemiol. 2016;40(4):304-14.
Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol. 2015;44(2):512-25.
Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nature genetics. 2018;50(5):693-8.
Barra L, Summers K, Bell D, Cairns E. Serum cytokine profile of unaffected first-degree relatives of patients with rheumatoid arthritis. J Rheumatol. 2014;41(2):280-5.
Pandya JM, Lundell AC, Andersson K, Nordstrom I, Theander E, Rudin A. Blood chemokine profile in untreated early rheumatoid arthritis: CXCL10 as a disease activity marker. Arthritis Res Ther. 2017;19(1):20.
Patel DD, Zachariah JP, Whichard LP. CXCR3 and CCR5 ligands in rheumatoid arthritis synovium. Clinical immunology. 2001;98(1):39-45.
Kotake S, Nanke Y, Mogi M, Kawamoto M, Furuya T, Yago T, et al. IFN-gamma-producing human T cells directly induce osteoclastogenesis from human monocytes via the expression of RANKL. European journal of immunology. 2005;35(11):3353-63.

Supplementarymaterial.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Genetically determined circulating levels ofcytokines and the risk of rheumatoid arthritis

Status:

Version 1

Abstract

Figures

Background

Methods

Results

Discussion

Conclusions

Abbreviations

Declarations

References

Supplementary Files

Status:

Version 1