Genome-wide association assessment between immune cells and osteoarthritis: a bidirectional Mendelian randomization study

Introduction

Osteoarthritis (OA) is a chronic degenerative disorder which primarily affects the joints, particularly the knees, hips, hands, and spine (1,2). An estimated 595 million people worldwide suffer from symptomatic OA, imposing substantial socioeconomic burdens (3-5). Beyond impeding daily activities, OA leads to an elevated number of disability-adjusted life years. Emerging evidence further suggests OA may induce systemic health risks beyond joint degeneration (6,7). However, current therapeutic strategies remain limited to intraarticular corticosteroid/hyaluronic acid injections for symptom management and end-stage joint replacement (1,8). There is a paucity of interventions to decelerate or reverse the degenerative processes of OA. This therapeutic gap underscores the urgent need to elucidate OA’s complex pathogenesis for target discovery (9).

Historically, OA was simply perceived as a wear-and-tear-related cartilage degradation, resulting in anatomical remodeling and functional impairment of joint. But recent studies have revealed OA’s multifactorial etiology involving genetic, biological, and biomechanical components (10-12). Evidence from studies in both patients and animal models has indicated that the progression of OA may be driven by chronic inflammation (13-15). The prevailing “chronic wound” hypothesis conceptualizes OA as sustained intra-articular injury, highlighting the immune microenvironment’s critical role in regulating cartilage damage-repair homeostasis (16). Various immune cells are involved in this process. As first responders, neutrophils initiate chondrocyte apoptosis and extracellular matrix degradation through elastase secretion while releasing pro-inflammatory mediators. Then, M1 macrophages are activated in response to the inflammatory cascade induced by interleukin (IL)-1β, tumor necrosis factor (TNF)-α, and IL-6, and finally result in the aggravation of chondrolysis. In contrast, M2 macrophages exhibit anti-inflammatory effects for localizing inflammation and producing pro-cartilage repairing cytokines such as IL-10 and transforming growth factor (TGF)-β (17,18). Besides, there are interaction between chondrocytes and synovial macrophages. Degenerating chondrocytes release damage-associated molecular patterns, which in turn activate macrophages, creating a vicious circle (19).

While prior research has established the immunological framework of OA, conflicting evidence from clinical and experimental studies hinder consensus on disease pathogenesis (20,21). A key challenge lies in clinical research is that the temporal relationship between immune infiltration and OA progression remains elusive, especially compounded by confounding factors such as aging and obesity. A systematic investigation is needed to delineate the roles of specific immune cell subsets in OA pathogenesis. Mendelian randomization (MR) overcomes inherent limitations of observational studies by mitigating confounding, reverse causality, and measurement bias (22). This method employs genetic variants as instrumental variables (IVs) to infer causal relationships between exposures and outcomes. Harnessing the rapid progress in large-scale genome-wide association studies (GWAS) and the availability of their summary results, numerous MR analyses have identified OA-related causal factors through secondary analysis of summary statistics (23-26). To pinpoint the key players that regulating the inflammatory processes in the progression of OA, we conducted this bidirectional MR analysis for comprehensively evaluation of the causal link between 731 immune cell traits and OA. We present this article in accordance with the STROBE-MR reporting checklist (available at https://aoj.amegroups.com/article/view/10.21037/aoj-24-60/rc).

Methods

Study protocol

Figure 1 demonstrates the fundamental steps of our MR analysis. We acquired single-nucleotide polymorphisms (SNPs) to serve as genetic IVs from GWAS databases. This selection progress of IVs was based on three core assumptions of MR analysis: relevance, independence, and exclusion restriction. The immune cells’ traits were regarded initially as exposure factors to detect a causal relationship with OA. A reverse MR analysis was subsequently performed for investigating the causal effects of OA on these immune cell phenotypes. In addition, we performed a replication analysis to further validate the positive results in primary analysis. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Figure 1 Overview of bidirectional MR analysis between immune cell traits and OA. The solid line indicates that the IVs correctly reveal the causal association between exposure and outcome. The dotted line represents potential pleiotropic or direct causal effects between IVs and outcome. FDR, false discovery rate; IV, instrumental variable; MR, Mendelian randomization; OA, osteoarthritis; SNP, single-nucleotide polymorphism.

Data source

We obtained GWAS data of 731 immune cell traits from a publicly available GWAS catalog (https://www.ebi.ac.uk/gwas/publications/32929287) (27,28). The genetic signature in immune cells was derived from flow cytometry analysis in a cohort comprising 3,757 Sardinians. The peripheral blood samples were collected via standard venipuncture and immediately stained with pre-optimized antibody panels. Flow cytometric data were acquired using two BD FACSCanto II flow cytometers with standardized configurations (29). A total of 731 immune features included 118 absolute cell counts, 389 median fluorescence intensity (MFI) of surface antigens, 32 morphological parameters, and 192 intercellular relative counts. These cell phenotypes can be divided into seven subgroups, including B cells, conventional dendritic cells (cDCs), mature T cells, monocytes, bone marrow cells, T cells, B cells, natural killer (NK) cells (TBNKs), and regulatory T cells (Tregs; morphological parameters features contain only cDCs and TBNKs). All assessed traits were normalized using inverse normal transformations and adjusted for sex, age, and age² as covariates.

The collective GWAS statistics of OA, serving as the outcomes in primary analysis, were derived from a secondary analysis on the UK Biobank dataset involving 407,746 individuals (https://www.ebi.ac.uk/gwas/studies/GCST90013881) (30). A newly proposed machine learning method named REGENIE was employed for fitting quantitative and binary phenotypic genome-wide regression models. In addition, an approximate Firth regression method was applied for unbalanced case-control phenotypes in order to eliminate most of the bias in the maximum likelihood estimates in logistic regression models. Disease-related information was extracted from self-reported status questionnaire and the Hospital Episode Statistics data. Participants were defined as OA cases according to the International Classification of Diseases version 10 code (ICD-10). In the replication analysis, we applied a cohort encompassing 63,556 individuals of European descent, including 12,658 OA cases (measured by self-reported questionnaire) and 50,898 controls. This summary GWAS statistics was retrieved from the GWAS catalog (https://www.ebi.ac.uk/gwas/studies/GCST005811) (31).

Selection criteria of IVs

Initially, in order to comply with the core principles of MR analysis, a genome-wide significance threshold of P<5×10⁻⁸ was applied to select SNPs strongly associated with immune cell traits. Due to the insufficient number of SNPs available as alternative risk factors for some immune cell traits, a less stringent threshold of P<1×10⁻⁵ was adopted. This adjustment, with reference to similar studies, was made to broaden the pool of SNPs and facilitate a more robust result (32-36). Considering the interference arising from linkage disequilibrium, we employed the clump method, with parameters set at kb=10,000 and r²=0.001, to reduce redundancy and enhance the independence of selected SNPs. Randomizing harmonise approach was used to minimize the interference of confounding factors on the results. What’s more, to ensure that the chosen IVs exhibit no apparent association with the outcome, the SNPs correlated with outcomes were excluded, with a correlation threshold of P<5×10⁻⁸. The F-statistics, calculated as the ratio of the variance of the genetic variant’s association with the exposure to the variance of its association with the outcome, was applied to assess the strength of each SNP. The SNPs with F-statistics below 10 were removed to ensure that only robust tools with sufficient strength and relevance are retained in later analysis.

Statistical analysis

All the MR analysis was based on the packages “TwoSampleMR” in R version 4.2.0. Inverse variance weighted (IVW) is the primary method used in our MR analysis. Based on the assumption that all selected IVs are valid and independent, the efficacy of IVW is highly susceptible to heterogeneity and pleiotropic variation (genetic variation affecting multiple traits). Hence, we also performed other MR analysis methods (MR egger, weighted median, simple mode, weighted mode) to increase confidence in the validity of causal reasoning. Furthermore, we combine the P values derived from MR Egger and IVW methods to gauge the heterogeneity among SNPs. Horizontal pleiotropy can affect the results of MR analysis by violating the exclusionary assumptions and introducing bias into the estimates of causal effects (37). Therefore, we introduced MR Egger intercept test in this study to detect horizontal pleiotropy. The MR-PRESSO test was applied to explore horizontal pleiotropic outliers (38). And we would repeat the MR analysis after removing the identified outliers. Sensitivity analysis by means of the leave-one-out method was also conducted to confirm the robustness of the results. The nominal significant results were defined as follows: (I) meet the significance threshold (P<0.05) by the IVW method; (II) without evidence of horizontal pleiotropy or heterogeneity; (III) the causal estimates by other methods were consistent with the IVW method in direction (39,40). To mitigate the risk of false positives, the nominal significant results would undergo the false discovery rate (FDR) correction (adjusted by five MR methods). Besides, a replication analysis was conducted to further explore and validate the causal relationship between immune cell phenotypes and OA.

We also downloaded a gene expression profiling data related to OA from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE55457) to further validate the consistency between our MR analysis results and actual immune infiltration in synovial tissue (41,42). The specific infiltrations of 64 immune and stromal cell types were analyzed by xCell algorithm (https://xcell.ucsf.edu/) (43), with a boxplot for depicting differences of cell subsets within OA vs. normal synovial tissue. In addition, we preliminarily investigated the differential expression of genes for characteristic immune markers in inflammatory synovium and normal tissue.

Results

Causal effect of immunophenotypes on OA

The MR results between OA and 731 immune cells traits at nominal significance level were depicted in Figures 2,3. We evaluated the reliability and validity of the SNPs by calculating the F-statistics to avoid weak instrument bias. The F-statistics of each SNPs used in this study ranged from 19.54 to 2,380.4, with an average of 34.4. Through the IVW method, we identified 37 immune cell traits that showed a significant causal relationship with OA, which belonged to seven subgroups including B cell (n=12), cDC (n=3), maturation stages of T cell (n=1), monocyte (n=5), myeloid cell (n=1), TBNK (n=8), Treg (n=7). Among them, 16 traits were found to have a risk-increasing effect on OA, while the other 21 trait showed a protective effect on the joints. After FDR correction, 13 immune cell traits remained significant causal relationships with OA, which came from six panels including B cell (n=2), cDC (n=3), monocyte (n=3), myeloid cell (n=1), TBNK (n=2), Treg (n=2). The odds ratio (OR) obtained by each of the five methods were consistently in the same direction. Strongest effects were found in “CD64 on CD14⁻ CD16⁺ monocyte” [OR: 1.044; 95% confidence interval (CI): 1.012–1.076] and “CD16⁺ monocyte %monocyte” (OR: 0.948; 95% CI: 0.916–0.980). Figure 4 showed the protective effects of three typical immune cell phenotypes (from monocytes and B cells) against OA: “CD14⁻ CD16⁺ monocyte %monocyte” (OR: 0.970; 95% CI: 0.944–0.996), “CD20 on IgD⁺ CD38^dim” (OR: 0.974; 95% CI: 0.974–0.989), and “Memory B cell %B cell” (OR: 0.975; 95% CI: 0.956–0.995).

Figure 2 The distribution of trait types and analytical panels of immune cell traits that were identified to have causal associations with OA at the nominal significance level. (A) The causal effects of immune cell traits on OA. (B) The causal effects of OA on immune cells traits. cDC, conventional dendritic cell; MFI, median fluorescence intensity; OA, osteoarthritis; TBNK, T cell, B cell, natural killer cell; Treg, regulatory T cell.

Figure 3 Positive results of MR analysis to estimate the causal effect of immune cell traits on OA. The nsnp represents the number of SNPs selected as IVs. The P value of each immune trait was assessed by the IVW method. The number in bold font represents results with P_FDR<0.05. cDC, conventional dendritic cell; CI, confidence interval; FDR, false discovery rate; IV, instrumental variable; IVW, inverse variance weighted; MFI, median fluorescence intensity; MR, Mendelian randomization; OA, osteoarthritis; OR, odds ratio; SNP, single-nucleotide polymorphism; TBNK, T cell, B cell, natural killer cell; Treg, regulatory T cell.

Figure 4 Genetic effects of three typical phenotypes of monocyte and B cell on OA. (A) Scatter plot of “CD14⁻ CD16⁺ monocyte %monocyte”; (B) scatter plot of “CD20 on IgD⁺ CD38dim”; (C) scatter plot of “Memory B cell %B cell”. MR, Mendelian randomization; OA, osteoarthritis; SNP, single-nucleotide polymorphism.

Table 1 displays the results of multiple sensitivity analyses with respect to heterogeneity and horizontal pleiotropy, which providing an endorsement for the reliability of our study. In this study, all intercepts converged to zero and no statistical significance was detected (P>0.05), implying that horizontal pleiotropy was almost negligible. What’s more, MR-PRESSO approach identified no outlier in the positive results. It is worth noting that the causal effects estimates (beta values) obtained by MR-PRESSO were consistent with the other five methods. At the same time, the heterogeneity test evaluated through the IVW and MR Egger methods showed that there was no significant evidence of heterogeneity in MR results (P>0.05). Besides, the leave-one-out analysis did not detect any variation that biased the population estimate. The funnel plot showed that the distribution of SNPs was almost symmetric on both IVW and MR Egger analysis.

Causal effect of OA on immunophenotypes

In reverse MR analysis, we identified genetic associations between OA and 25 immune cell traits through five SNPs (depicted in Figures 2,5). These immune cells could be subdivided into six groups, including B cell (n=1), cDC (n=2), monocyte (n=7), myeloid cell (n=2), TBNK (n=10), Treg (n=3). Most of these immune cell traits were negatively associated with OA (n=21). After adjustment for FDR, the results of the IVW analysis suggested that the presence of OA downregulated the expression on “CD14⁻ CD16⁺ monocyte absolute count” (OR: 0.607; 95% CI: 0.417–0.885), “CD8^dim T cell %leukocyte” (OR: 0.616; 95% CI: 0.414–0.918), “CD8^dim NK T absolute count” (OR: 0.548; 95% CI: 0.375–0.801), “CD8^dim NK T %T cell” (OR: 0.552; 95% CI: 0.377–0.810), and “CD8^dim NK T %lymphocyte” (OR: 0.544; 95% CI: 0.372–0.795).

Figure 5 Positive results of MR analysis to estimate the causal effect of OA on immune cell traits. The nsnp represents the number of SNPs selected as IVs. The P value of each immune trait was assessed by the IVW method. The number in bold font represents results with P_FDR<0.05. cDC, conventional dendritic cell; CI, confidence interval; FDR, false discovery rate; IV, instrumental variable; IVW, inverse variance weighted; MFI, median fluorescence intensity; MR, Mendelian randomization; OA, osteoarthritis; OR, odds ratio; SNP, single-nucleotide polymorphism; TBNK, T cell, B cell, natural killer cell; Treg, regulatory T cell.

We did not find any SNPs in sensitivity analysis that significantly interfered with this result. Besides, subsequent analyses observed no evidence of significant heterogeneity or horizontal pleiotropy. The funnel plots indicated that most of the distribution of SNPs was symmetrical, while a slight deviation was observed in some immunophenotype (because of fewer numbers of SNPs).

Results of replication analysis

We identified 34 immune cell traits that had significant casual effects on OA by IVW method (Figure 6). These immunophenotypes were divided into seven subgroups including B cell (n=15), cDC (n=5), maturation stages of T cell (n=1), monocyte (n=2), myeloid cell (n=3), TBNK (n=4), Treg (n=4). Twelve of them meet the significance threshold after FDR correction, which came from seven panels including B cell (n=4), cDC (n=2), maturation stages of T cell (n=1), monocyte (n=1), myeloid cell (n=1), TBNK (n=2), Treg (n=1). And results of sensitivity analysis showed that heterogeneity and horizontal pleiotropy had little effect on causal estimate (Table 2). It is worth noting that “CD64 on CD14⁻ CD16⁺ monocyte” (OR: 1.102; 95% CI: 1.046–1.161; P_FDR<0.001) and “HLA DR⁺ NK %NK” (OR: 0.945; 95% CI: 0.908–0.983; P_FDR=0.03) also appeared in the positive results of our primary analysis.

Figure 6 Positive result of replication MR analysis to estimate the causal effect of immune cell traits on OA. The nsnp represents the number of SNPs selected as IVs. The P value of each immune trait was assessed by the IVW method. The number in bold font represents results with P_FDR<0.05. cDC, conventional dendritic cell; CI, confidence interval; FDR, false discovery rate; IV, instrumental variable; IVW, inverse variance weighted; MFI, median fluorescence intensity; MR, Mendelian randomization; OA, osteoarthritis; OR, odds ratio; SNP, single-nucleotide polymorphism; TBNK, T cell, B cell, natural killer cell; Treg, regulatory T cell.

Immune infiltration between OA and normal synovial tissue

Figure 7 demonstrated the immune cell infiltration in synovial tissue of OA patients and healthy people. Through the analysis by the xCell algorithm, we identified four immune cells (“CD8⁺ naive T cells”, “CD8⁺ Tcm”, “memory B cells”, “pDC”) and six interstitial cells (“lymphatic endothelial cells”, “megakaryocyte-erythroid progenitor”, “microvascular endothelial cells”, “neurons”, “osteoblast”, “preadipocytes”) with significant differences between the two samples types. Further, an upgrade expression of genes, including “ENTPD1”, “CD86”, “CX3CR1”, “CD27”, and “MS4A1”, was identified in synovial tissue of OA patients (Figure 8). It represents an enrichment of immune cells carrying characteristic biomarkers such as Treg (CD39), cDC (CD86), monocyte (CX3CR1), and B cell (CD27 and CD20).

Figure 7 Boxplot describing the difference in infiltration of immune cells or stromal cells in normal synovium vs. synovium of OA. The * and ** signs represent significant difference (P<0.05 and P<0.01) between two types of tissue. OA, osteoarthritis.

Figure 8 Boxplot describing the difference in gene expression of characteristic immune markers in normal synovium vs. synovium of OA. The * and ** signs represent significant difference (P<0.05 and P<0.01) between two types of tissue. OA, osteoarthritis.

Discussion

We conducted this bidirectional MR analysis to explore the evidence of potential causal relationship and the risk/protective effect between numerous immune cell traits and OA. Totally, we identified 13 immune cell traits that had significant causal associations with the OA after FDR correction. Among them, CD64 on CD14⁻ CD16⁺ monocyte has a significant risk on OA, while high proportion of CD16⁺ monocyte in monocyte plays the role of protector. Besides, immune cell infiltration analysis based on GEO database provided supplementary validation to our results.

The long-held view that OA is a non-inflammatory degenerative disease solely induced by mechanical wear has been fundamentally revised through accumulating evidence of immune involvement. This pathological process involves a repetitive cycle of cartilage injury and repair. It ultimately drives irreversible articular cartilage loss, subchondral bone remodeling, and osteophyte formation. Low-grade persistent synovitis plays the key role in this process, which resulted from the crosstalk between immune mediators, biomechanical stress, and metabolic dysregulation (14,44,45). Previous studies have indicated that a variety of immune cells can interfere with the metabolism of chondrocytes by secreting cytokines. For instance, IL-1β produced by macrophages can stimulate the liberation of nitric oxide (NO) and prostaglandin E2 (PGE2) within synovial joints. These mediators upregulate matrix metalloproteinase (MMP) activity, curtail the production of anabolic macromolecules such as collagen and proteoglycan, and thus contribute to chondrocyte degradation and apoptosis (46).

Monocytes and macrophages are the protagonists in regulating synovial homeostasis and cartilage function (18,47,48). Peripheral monocytes are typically categorized into classical (CD14⁺ CD16⁻), intermediate (CD14⁺ CD16⁺), and non-classical (CD14^dim/− CD16⁺) cells determined by their CD14 and CD16 expression levels. A cohort study showed that CD14⁺ CD16⁺ macrophage subpopulation constituted 35% of the total macrophages in synovial fluid. The proportion of CD14⁺ macrophages to total macrophages served as a predictor of Knee Injury and Osteoarthritis Outcome Score (KOOS) and Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), regardless of CD16 expression in this subpopulation of macrophages (49). Our study indicated an inverse association between the proportion of CD16⁺ macrophages to total macrophages and the risk of OA, suggesting the dominance of classical macrophages in OA progression. Notably, our study identified “CD64 on CD14⁻ CD16⁺ monocyte” as a significant pathogenic factor associated with OA. Teunissen van Manen et al.’s study demonstrated a link between CD64 and the destructive ability of synovitis (50). They found that elevated expression of synovial CD64 in OA was associated with the expression of MMPs and inflammatory markers related to structural damage in OA, and lead to the over-production of stroma-degrading proteins in synovial fibroblasts. In addition, research by Culemann et al. revealed that CX3CR1⁺ macrophages exhibit the typical characteristics of epithelial cells, physically isolating joints and limiting inflammation by forming dynamic membrane-like structures within the synovial membrane (51). This explains the protective effect we observed of CX3CR1 on CD14⁺ CD16⁺ monocyte against OA.

T cells also constitute a significant component of synovial-infiltration cells in OA patients (52). Studies have demonstrated a higher presence of CD4⁺ T cells in the sublining layer of synovial tissue among OA patients compared to normal individuals (53). Increased CD4⁺ T cells are linked to macrophage activation and MMP-9 secretion, which can accelerate cartilage degradation. In addition, macrophage inflammatory protein (MIP)-1γ induced by CD4⁺ T cells resulted in an augmentation of osteoclast numbers in the joint (54). HLA DR⁺ is regarded as a sign of CD4⁺ T cells and NK cell activation. However, our study noted a decrease in the risk of OA with the heightened proportion of HLA DR⁺ CD4⁺ cells within lymphocytes. At the same time, it was observed that a higher percentage of HLA DR⁺ NK cells correlated with a reduced occurrence of OA. This association may be attributed to the potential of NK cells to facilitate cartilage repair indirectly through the removal of necrotic tissue and the promotion of mesenchymal stem cell (MSC) recruitment via the secretion of the chemokine CXCL7 (55). Tregs, characterized by CD4⁺ and CD25⁺ subsets, possess anti-inflammatory properties (56). Sohn et al. transferred CD4⁺ CD25⁺ T cells isolated from the spleen of mice in the treatment group to mice in the OA group, and observed inhibition in cartilage destruction compared with control group (57). Similarly, all Tregs found in our study showed inhibitory effects on OA. It is worth noting that a Treg-subtracted T cells subgroup (CD25 on CD45RA⁺ CD4 not Treg) was categorized as a risk correlate for OA.

Our study also identified other immune components that have a causal effect on OA. Two types of B cell traits (“CD20 on IgD⁺ CD38^dim” and “Memory B cell %B cell”) were found to exert a protective effect against OA, while two types (“IgD on unsw mem” and “CD24 on IgD⁺ CD38⁻ unsw mem”) may play the opposite role. Recent studies have revealed that B cells have diverse effects in the pathogenesis of. On the one hand, antigen-specific B cells can activate and differentiate aggregation-specific CD4⁺ T cells by acquiring proteoglycan aggregates in the cartilage extracellular matrix, contributing to OA progression (58). On the other hand, regulatory B cells can secrete IL-10, inhibit the proliferation of autologous T cells and reduce the expression of IFN-γ, so as to suppress the progression of inflammation (59). As for cDC, studies have also shown that they are protective immune regulatory cells generated after cartilage injury, which can not only promote the proliferation of Treg, but also accelerate the differentiation of MSC cartilage by secreting IL-10 (16,60). In contrast, mature cDC promote cartilage degradation. In our study, CD86 on cDC was found to contribute to the increased risk of OA, possibly due to its interaction with CD28 for T cell activation and the initiation of the immune response.

Our work has several key advantages as follows. First, this MR analysis leveraged the summary GWAS statistics of the most comprehensive collection on immune cell phenotypes to date and the largest scale genetic dataset for OA. Compared to previous studies (61), which concentrated on singular immune biomarkers or specific subsets of immune cells, this study considered a broader spectrum of immune cell traits, providing a more complete picture of the complex interplay between the immune system and OA pathogenesis. Secondly, the design of bidirectional MR analysis not only avoided the interference of confounding factors on causal inference in observational studies, but also reduced the potential bias introduced by reverse causality. Thirdly, by exploiting the replication analysis, we managed to improving the robustness of our findings. At last, an immune infiltration analysis on GEO database was conducted as the supplementary validation on the relationship between immune cells and OA.

Several limitations in our study may affect the reliability of MR analysis. First of all, the GWAS data of immune cells was extracted originally from a cohort of Sardinians, whereas the genetic data of OA belong to the white British descent. For one thing, we’re not sure if this slight inconsistency interferes with the findings. On the other hand, the monotony of the sources raises concerns about potential population-specific effects and makes it necessary to be cautious when extrapolating our conclusion to groups other than the European population. Secondly, when screening suitable IVs from immune cell GWAS data, an aggressive threshold (1×10⁻⁵) was set in the initial analysis to ensure sufficient SNPs, which may increase the risk of false positive associations and overfitting. In addition, due to the large total number of SNPs, we failed to individually exclude SNPs that might be associated with potential confounders, which could introduce bias and reduce the ability to detect true causality. The pleiotropic effects come from unknown pathways connecting genetic variations, immune cell traits, and OA may violate assumptions of MR analysis. Therefore, we conducted a series of sensitivity analysis to confirm the robustness of our results. At last, while MR Analysis can offer evidence of potential causality, it is short at making further clarity about whether only the synovial-specific immune responses or the interactions between immune cells and multiple tissues are at work. It also fails to capture the time series of disease progression or the dynamic nature of the immune response to OA. Our MR results should be considered as complementary to, rather than a replacement for, evidence derived from observed trials.

Conclusions

In summary, we identified a total of 13 risk associations between hundreds of immunophenotypes and OA through a systematic MR Analysis, shedding light on the complex interactions between the immune microenvironment and OA. This may provide researchers with new clues to explore potential therapeutic targets for OA, and thus aiding in the deceleration or reversal of cartilage degeneration.

Acknowledgments

We gratefully acknowledge the public available GWAS data provided by Orru et al. on immune cell traits and Mbatchou et al. on osteoarthritis. We also acknowledge Woetzel et al. for their uploading of genome-wide transcriptome dataset. And we are grateful to all the participants and investigators of the European Bioinformatics Institute, UK-Biobank, NHGRI-EBI Catalog, GEO database for providing summary data used in this study.

Footnote

Reporting Checklist: The authors have completed the STROBE-MR reporting checklist. Available at https://aoj.amegroups.com/article/view/10.21037/aoj-24-60/rc

Peer Review File: Available at https://aoj.amegroups.com/article/view/10.21037/aoj-24-60/prf

Funding: This study was supported by the Natural Science Foundation of Guangdong Province (Nos. 2022A1515012334, 2024A1515012811, and 2025A1515010517), the Science and Technology Program of Guangzhou (Nos. 2023A03J0705, 2024A03J0844, and 2024A04J4690), and the Sun Yat-sen Memorial Hospital Clinical Research 5010 Program (No. SYS-5010-202403).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://aoj.amegroups.com/article/view/10.21037/aoj-24-60/coif). All authors report that this study was supported by the Natural Science Foundation of Guangdong Province (Nos. 2022A1515012334, 2024A1515012811, and 2025A1515010517), the Science and Technology Program of Guangzhou (Nos. 2023A03J0705, 2024A03J0844, and 2024A04J4690), and the Sun Yat-sen Memorial Hospital Clinical Research 5010 Program (No. SYS-5010-202403). The authors have no other conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Katz JN, Arant KR, Loeser RF. Diagnosis and Treatment of Hip and Knee Osteoarthritis: A Review. JAMA 2021;325:568-78. [Crossref] [PubMed]
2. Kloppenburg M, Namane M, Cicuttini F. Osteoarthritis. Lancet 2025;405:71-85. [Crossref] [PubMed]
3. Allen KD, Thoma LM, Golightly YM. Epidemiology of osteoarthritis. Osteoarthritis Cartilage 2022;30:184-95. [Crossref] [PubMed]
4. Minnig MCC, Golightly YM, Nelson AE. Epidemiology of osteoarthritis: literature update 2022-2023. Curr Opin Rheumatol 2024;36:108-12. [Crossref] [PubMed]
5. Courties A, Kouki I, Soliman N, et al. Osteoarthritis year in review 2024: Epidemiology and therapy. Osteoarthritis Cartilage 2024;32:1397-404. [Crossref] [PubMed]
6. Palazzo C, Nguyen C, Lefevre-Colau MM, et al. Risk factors and burden of osteoarthritis. Ann Phys Rehabil Med 2016;59:134-8. [Crossref] [PubMed]
7. Hall AJ, Stubbs B, Mamas MA, et al. Association between osteoarthritis and cardiovascular disease: Systematic review and meta-analysis. Eur J Prev Cardiol 2016;23:938-46. [Crossref] [PubMed]
8. Abramoff B, Caldera FE. Osteoarthritis: Pathology, Diagnosis, and Treatment Options. Med Clin North Am 2020;104:293-311. [Crossref] [PubMed]
9. Cho Y, Jeong S, Kim H, et al. Disease-modifying therapeutic strategies in osteoarthritis: current status and future directions. Exp Mol Med 2021;53:1689-96. [Crossref] [PubMed]
10. Lieberthal J, Sambamurthy N, Scanzello CR. Inflammation in joint injury and post-traumatic osteoarthritis. Osteoarthritis Cartilage 2015;23:1825-34. [Crossref] [PubMed]
11. Wang T, He C. Pro-inflammatory cytokines: The link between obesity and osteoarthritis. Cytokine Growth Factor Rev 2018;44:38-50. [Crossref] [PubMed]
12. Wei G, Lu K, Umar M, et al. Risk of metabolic abnormalities in osteoarthritis: a new perspective to understand its pathological mechanisms. Bone Res 2023;11:63. [Crossref] [PubMed]
13. Motta F, Barone E, Sica A, et al. Inflammaging and Osteoarthritis. Clin Rev Allergy Immunol 2023;64:222-38. [Crossref] [PubMed]
14. Robinson WH, Lepus CM, Wang Q, et al. Low-grade inflammation as a key mediator of the pathogenesis of osteoarthritis. Nat Rev Rheumatol 2016;12:580-92. [Crossref] [PubMed]
15. Scanzello CR. Role of low-grade inflammation in osteoarthritis. Curr Opin Rheumatol 2017;29:79-85. [Crossref] [PubMed]
16. Li M, Yin H, Yan Z, et al. The immune microenvironment in cartilage injury and repair. Acta Biomater 2022;140:23-42. [Crossref] [PubMed]
17. Wang L, He C. Nrf2-mediated anti-inflammatory polarization of macrophages as therapeutic targets for osteoarthritis. Front Immunol 2022;13:967193. [Crossref] [PubMed]
18. Zhang H, Cai D, Bai X. Macrophages regulate the progression of osteoarthritis. Osteoarthritis Cartilage 2020;28:555-61. [Crossref] [PubMed]
19. Fujihara Y, Abe T, Asawa Y, et al. Influence of Damage-Associated Molecular Patterns from Chondrocytes in Tissue-Engineered Cartilage. Tissue Eng Part A 2021;27:1-9. [Crossref] [PubMed]
20. Nedunchezhiyan U, Varughese I, Sun AR, et al. Obesity, Inflammation, and Immune System in Osteoarthritis. Front Immunol 2022;13:907750. [Crossref] [PubMed]
21. Hu K, Song M, Song T, et al. Osteoimmunology in Osteoarthritis: Unraveling the Interplay of Immunity, Inflammation, and Joint Degeneration. J Inflamm Res 2025;18:4121-42. [Crossref] [PubMed]
22. Sekula P, Del Greco M F, Pattaro C, et al. Mendelian Randomization as an Approach to Assess Causality Using Observational Data. J Am Soc Nephrol 2016;27:3253-65. [Crossref] [PubMed]
23. Xu J, Zhang S, Tian Y, et al. Genetic Causal Association between Iron Status and Osteoarthritis: A Two-Sample Mendelian Randomization. Nutrients 2022;14:3683. [Crossref] [PubMed]
24. Ho J, Mak CCH, Sharma V, et al. Mendelian Randomization Studies of Lifestyle-Related Risk Factors for Osteoarthritis: A PRISMA Review and Meta-Analysis. Int J Mol Sci 2022;23:11906. [Crossref] [PubMed]
25. Larsson SC, Burgess S. Appraising the causal role of smoking in multiple diseases: A systematic review and meta-analysis of Mendelian randomization studies. EBioMedicine 2022;82:104154. [Crossref] [PubMed]
26. Cui ZY, Feng H, He BC, et al. Causal associations between serum amino acid levels and osteoarthritis: a Mendelian randomization study. Osteoarthritis Cartilage 2023;31:1121-31. [Crossref] [PubMed]
27. Orrù V, Steri M, Sidore C, et al. Complex genetic signatures in immune cells underlie autoimmunity and inform therapy. Nat Genet 2020;52:1036-45. [Crossref] [PubMed]
28. Buniello A, MacArthur JAL, Cerezo M, et al. The NHGRI-EBI GWAS Catalog of published genome-wide association studies, targeted arrays and summary statistics 2019. Nucleic Acids Res 2019;47:D1005-12. [Crossref] [PubMed]
29. Sidore C, Busonero F, Maschio A, et al. Genome sequencing elucidates Sardinian genetic architecture and augments association analyses for lipid and blood inflammatory markers. Nat Genet 2015;47:1272-81. [Crossref] [PubMed]
30. Mbatchou J, Barnard L, Backman J, et al. Computationally efficient whole-genome regression for quantitative and binary traits. Nat Genet 2021;53:1097-103. [Crossref] [PubMed]
31. Zengini E, Hatzikotoulas K, Tachmazidou I, et al. Genome-wide analyses using UK Biobank data provide insights into the genetic architecture of osteoarthritis. Nat Genet 2018;50:549-58. [Crossref] [PubMed]
32. Perry BI, Upthegrove R, Kappelmann N, et al. Associations of immunological proteins/traits with schizophrenia, major depression and bipolar disorder: A bi-directional two-sample mendelian randomization study. Brain Behav Immun 2021;97:176-85. [Crossref] [PubMed]
33. Bouras E, Karhunen V, Gill D, et al. Circulating inflammatory cytokines and risk of five cancers: a Mendelian randomization analysis. BMC Med 2022;20:3. [Crossref] [PubMed]
34. Gu Y, Jin Q, Hu J, et al. Causality of genetically determined metabolites and metabolic pathways on osteoarthritis: a two-sample mendelian randomization study. J Transl Med 2023;21:357. [Crossref] [PubMed]
35. Ji D, Chen WZ, Zhang L, et al. Gut microbiota, circulating cytokines and dementia: a Mendelian randomization study. J Neuroinflammation 2024;21:2. [Crossref] [PubMed]
36. Yang M, Wan X, Zheng H, et al. No Evidence of a Genetic Causal Relationship between Ankylosing Spondylitis and Gut Microbiota: A Two-Sample Mendelian Randomization Study. Nutrients 2023;15:1057. [Crossref] [PubMed]
37. Davies NM, Holmes MV, Davey Smith G. Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ 2018;362:k601. [Crossref] [PubMed]
38. Verbanck M, Chen CY, Neale B, et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018;50:693-8. [Crossref] [PubMed]
39. Chen X, Kong J, Diao X, et al. Depression and prostate cancer risk: A Mendelian randomization study. Cancer Med 2020;9:9160-7. [Crossref] [PubMed]
40. Zagkos L, Dib MJ, Pinto R, et al. Associations of genetically predicted fatty acid levels across the phenome: A mendelian randomisation study. PLoS Med 2022;19:e1004141. [Crossref] [PubMed]
41. Woetzel D, Huber R, Kupfer P, et al. Identification of rheumatoid arthritis and osteoarthritis patients by transcriptome-based rule set generation. Arthritis Res Ther 2014;16:R84. [Crossref] [PubMed]
42. Barrett T, Wilhite SE, Ledoux P, et al. NCBI GEO: archive for functional genomics data sets--update. Nucleic Acids Res 2013;41:D991-5. [Crossref] [PubMed]
43. Aran D, Hu Z, Butte AJ. xCell: digitally portraying the tissue cellular heterogeneity landscape. Genome Biol 2017;18:220. [Crossref] [PubMed]
44. Terkawi MA, Ebata T, Yokota S, et al. Low-Grade Inflammation in the Pathogenesis of Osteoarthritis: Cellular and Molecular Mechanisms and Strategies for Future Therapeutic Intervention. Biomedicines 2022;10:1109. [Crossref] [PubMed]
45. Coaccioli S, Sarzi-Puttini P, Zis P, et al. Osteoarthritis: New Insight on Its Pathophysiology. J Clin Med 2022;11:6013. [Crossref] [PubMed]
46. Afonso V, Champy R, Mitrovic D, et al. Reactive oxygen species and superoxide dismutases: role in joint diseases. Joint Bone Spine 2007;74:324-9. [Crossref] [PubMed]
47. Fernandes TL, Gomoll AH, Lattermann C, et al. Macrophage: A Potential Target on Cartilage Regeneration. Front Immunol 2020;11:111. [Crossref] [PubMed]
48. Zhao K, Ruan J, Nie L, et al. Effects of synovial macrophages in osteoarthritis. Front Immunol 2023;14:1164137. [Crossref] [PubMed]
49. Gómez-Aristizábal A, Gandhi R, Mahomed NN, et al. Synovial fluid monocyte/macrophage subsets and their correlation to patient-reported outcomes in osteoarthritic patients: a cohort study. Arthritis Res Ther 2019;21:26. [Crossref] [PubMed]
50. Teunissen van Manen IJ, van Kooten NJT, Di Ceglie I, et al. Identification of CD64 as a marker for the destructive potential of synovitis in osteoarthritis. Rheumatology (Oxford) 2024;63:1180-8. [Crossref] [PubMed]
51. Culemann S, Grüneboom A, Nicolás-Ávila JÁ, et al. Locally renewing resident synovial macrophages provide a protective barrier for the joint. Nature 2019;572:670-5. [Crossref] [PubMed]
52. Li YS, Luo W, Zhu SA, et al. T Cells in Osteoarthritis: Alterations and Beyond. Front Immunol 2017;8:356. [Crossref] [PubMed]
53. Ishii H, Tanaka H, Katoh K, et al. Characterization of infiltrating T cells and Th1/Th2-type cytokines in the synovium of patients with osteoarthritis. Osteoarthritis Cartilage 2002;10:277-81. [Crossref] [PubMed]
54. Shen PC, Wu CL, Jou IM, et al. T helper cells promote disease progression of osteoarthritis by inducing macrophage inflammatory protein-1γ. Osteoarthritis Cartilage 2011;19:728-36. [Crossref] [PubMed]
55. Almeida CR, Caires HR, Vasconcelos DP, et al. NAP-2 Secreted by Human NK Cells Can Stimulate Mesenchymal Stem/Stromal Cell Recruitment. Stem Cell Reports 2016;6:466-73. [Crossref] [PubMed]
56. Tanigawa S, Aida Y, Kawato T, et al. Interleukin-17F affects cartilage matrix turnover by increasing the expression of collagenases and stromelysin-1 and by decreasing the expression of their inhibitors and extracellular matrix components in chondrocytes. Cytokine 2011;56:376-86. [Crossref] [PubMed]
57. Sohn HS, Choi JW, Jhun J, et al. Tolerogenic nanoparticles induce type II collagen-specific regulatory T cells and ameliorate osteoarthritis. Sci Adv 2022;8:eabo5284. [Crossref] [PubMed]
58. Ciechomska M, Wilson CL, Floudas A, et al. Antigen-specific B lymphocytes acquire proteoglycan aggrecan from cartilage extracellular matrix resulting in antigen presentation and CD4+ T-cell activation. Immunology 2014;141:70-8. [Crossref] [PubMed]
59. Maerz JK, Trostel C, Lange A, et al. Bacterial Immunogenicity Is Critical for the Induction of Regulatory B Cells in Suppressing Inflammatory Immune Responses. Front Immunol 2019;10:3093. [Crossref] [PubMed]
60. Alahdal M, Zhang H, Huang R, et al. Potential efficacy of dendritic cell immunomodulation in the treatment of osteoarthritis. Rheumatology (Oxford) 2021;60:507-17. [Crossref] [PubMed]
61. Luo H, Zhu Y, Guo B, et al. Causal relationships between CD25 on immune cells and hip osteoarthritis. Front Immunol 2023;14:1247710. [Crossref] [PubMed]