Causal relationship between the blood immune cells and intervertebral disc degeneration: univariable, bidirectional and multivariable Mendelian randomization

Background Intervertebral disc degeneration (IVDD) is a prominent contributor to chronic low back pain, impacting millions of individuals annually. Current research on disc degeneration is placing a growing emphasis on the role of the immune system in this process. Nevertheless, the precise relationship between immunity and disc degeneration remains to be fully elucidated. Method We obtained GWAS data for immune cells from the latest summary-level GWAS, including 6,620 individuals from Sardinian and 746,667 individuals from five global populations. Summary results for IVDD were sourced from the FinnGen consortium, comprising 20,001 cases and 164,682 controls. We conducted a comprehensive univariable Mendelian randomization (MR) analysis to explore the potential causal relationship between immune cells and IVDD. Primary estimation was carried out using Inverse-Variance Weighting (IVW). To ensure robustness, we employed additional MR methods such as MR-Egger, Weighted Median, Weighted Mode, and Simple Mode. Various tests were employed to assess pleiotropy and heterogeneity, including the Cochran Q test, leave-one-out test, MR-Egger intercept analysis and MR-PRESSO test. To account for potential confounding factors among the immune cells, we conducted a multivariable MR analysis. Finally, we investigated the possibility of a reverse association between immune cells and IVDD through bidirectional MR. Result In total, our study identified 15 immune cells significantly associated with IVDD through univariable MR. Among these, 9 immune cell types were indicated as potential contributors to IVDD, while 6 were found to have protective effects. Importantly, we observed no evidence of heterogeneity or pleiotropy, signifying the robustness of our results. To mitigate confounding among immune cells, we utilized multivariable MR, leading to the discovery that only 9 immune cell types exerted independent effects on IVDD. These encompassed 7 as risk factors and 2 as protective factors. Additionally, our analysis revealed a bidirectional causal relationship between CD39+ CD4+ T cell %CD4+ T cell and IVDD. Conclusion Our findings suggest a connection between immune cells and the risk of IVDD, shedding light on potential therapeutic avenues for modulating immune cell function in individuals with IVDD. However, the specific underlying mechanisms warrant further investigation in future experiments.


Introduction
Intervertebral disc degeneration (IVDD) is a prevalent degenerative disease, with an incidence of 40-50 cases per 10,000 people (1).IVDD is a pathological condition that can lead to persistent back pain or neurological deficits.The incidence of IVDD is gradually increasing, posing an escalating healthcare burden and economic pressure (2, 3).Unfortunately the etiology of IVDD still remains unclear (4)(5)(6).It is reported that intervertebral disc belongs to the immune-privileged tissue (7)(8)(9).Extrinsic factors, such as external forces and mechanical loads, play a role in annulus fibrosus rupture, with the subsequent immune response triggered by nucleus pulposus antigen exposure serving as the initiator of disc degeneration (10-13).Nevertheless, the association between immune cells and IVDD remains unconfirmed due to the absence of randomized controlled trials (RCTs).
Mendelian Randomization (MR), a method examining potential causal relationships between exposure factors and disease outcomes, has gained popularity in recent years (14).Consequently, we chose to employ univariable, multivariable, and bidirectional MR approaches to identify potential protective and risk factors for IVDD.This study focused on exploring the possible causal links between immune cells and IVDD, with the aim of uncovering novel targets for IVDD prevention and treatment.

Study design
In this study, we conducted univariable MR to investigate potential causal associations between various immune cells and IVDD, utilizing genome-wide association studies (GWAS) summary statistics (Figure 1A).To ensure unbiased causal effects, MR analysis must adhere to three core assumptions: (I) the relevance assumption, where genetic variants are strongly associated with the exposure of interest; (II) the independence assumption, ensuring genetic variants are not associated with potential confounders; and (III) the exclusion restriction, stipulating that genetic variants affect outcomes solely through the exposure of interest (15).To further explore the link between immune cells and IVDD, we employed multivariable MR, assessing the independent beta of different immune cells, as these cells can influence each other's effector values (Figure 1B).Additionally, we conducted bidirectional MR to investigate the possibility of reverse causality between the exposure and the outcome (14) (Figure 1C).

GWAS data source
The GWAS data on immune cells were obtained from the latest summary-level GWAS, which involved 6,620 individuals from Sardinian descent within the European population and 746,667 individuals from five diverse global populations (16,17).The former study comprised 3,757 cases and 3,027 controls, with a gender distribution of 43% males and 57% females, and age ranges spanning from 18 to 102 years (16).The latter dataset included 408,961 individuals of European ancestry from the UK Biobank (UKBB), 143,988 individuals of Japanese ancestry from the Biobank of Japan (BBJ), and 5,275 African Americans from the Vanderbilt University Biobank (BioVU) (17).

Instrumental variable selection
Genetic instrumental variables for immune cell exposures were carefully chosen based on a genome-wide significance threshold (p<5×10 -8 ) (14).We employed the PLINK clumping method to assess the linkage disequilibrium (LD) among the selected SNPs, with LD defined as SNPs having an R 2 >0.001 and being within a 10,000 kb physical distance (14).SNPs identified to be in LD were subsequently excluded from further analysis.Additionally, to eliminate weak instrumental variables, we calculated F-statistics for the SNPs using a previously described approximation method (18,19).Importantly, all included exposures exhibited F-statistics exceeding 10 [Supplementary Table 1].

Statistical analysis
We primarily employed the inverse-variance weighted (IVW) method as our main approach in Mendelian randomization (MR).In addition, we conducted MR-Egger, weighted median, weighted mode, and simple mode analyses to ensure the robustness of our results.To assess heterogeneity among summary estimates, Cochran's Q test was utilized.Moreover, we utilized the MR-Egger intercept to address and account for pleiotropy (14).To further assess the robustness of our findings, we implemented a leave-one-out analysis.The MR-PRESSO test was employed to identify and address pleiotropy, and to remove outliers (20).
W e c o n d u c t e d a l l s t a t i s t i c a l a n a l y s e s u s i n g t h e "TwoSampleMR" (version 0.5.7)packages and MR-PRESSO package within the R statistical software (version 4.3.1).A significance level of p < 0.05 was chosen to denote statistical significance.
Given the interrelated nature of various immune cells, they have the potential to act as confounding factors for each other.Hence, we employed multivariable MR to estimate the independent direct effect of each exposure on the outcome.Our analysis revealed that only 9 immune cell types maintained an independent role when accounting solely for the effect of each exposure on the outcome (Table 1; Figure 3).CD39+ CD4+ T cell %T cell increase the risk of IVDD significantly (OR:1.406,95%CI: 1.133-1.745).CD39+ CD8+ T cell % T cell increase the risk of IVDD significantly (OR: 1.810, 95%CI: 1.282-2.555).CD39+ CD8+ T cell Absolute Count decrease the risk of IVDD significantly (OR: 0.551, 95%CI: 0.383-0.793).CD28-CD8+ T cell Absolute Count increase the risk of IVDD significantly (OR: 1.394, 95%CI: 1.204-1.512).CD19 on IgD+ CD38dim B cell increase the risk of IVDD significantly (OR: 1.400, 95%CI: 1.056-1.856).CD19 on unswitched memory B cell decrease the risk of IVDD significantly (OR: 0.786, 95%CI: 0.664-0.932).CD25 on CD39+ CD4+ T cell increase the risk of IVDD significantly (OR: 1.461, 95%CI: 1.255-1.700).CD4 on CD39+ CD4+ T cell increase the risk of IVDD Forest plots of effects of immune cells on the risk of IVDD by using univariable MR.  2, 3).We conducted a reverse MR analysis and identified reverse causality specifically between CD39+ CD4+ T cells %CD4+ T cells and IVDD.Additionally, we observed that other intervertebral disc disorders significantly increase the risk of CD39+ CD4+ T cell % CD4+ T cell (OR: 1.409, 95%CI: 1.053-1.885,p=0.021) (Table 2; Figure 4).The causal relationship between these factors appears to be bidirectional.

Sensitivity analyses
To ensure the robustness of our MR estimates, we conducted several sensitivity analyses.We employed Cochran's Q test, MR-Egger intercept test, and a leave-one-out test.Notably, all MR-Egger intercept tests yielded P-values greater than 0.05, indicating the absence of pleiotropy.Additionally, the Cochran's Q test revealed no significant heterogeneity between immune cells and IVDD (Table 3).Furthermore, the results remained stable in the leave- Forest plots of effects of immune cells on the risk of IVDD by using multivariable MR. one-out test (Figure 5).The MR-PRESSO analysis revealed the absence of outliers and pleiotropy, as illustrated in Table 3.However, the MR-PRESSO test could not be applied to the CD28-CD8+ T cell Absolute Count due to the limited number of SNPs.In summary, the results of the univariable MR analysis are deemed acceptable.
For the bidirectional MR analysis between CD39+ CD4+ T cell %CD4+ T cell and other intervertebral disc disorders, both the pvalues of Cochran's Q and MR-Egger intercept were greater than 0.05 (Table 4), suggesting the absence of heterogeneity and pleiotropy.The leave-one-out test reaffirmed the stability of these results (Figure 6).Pleiotropy and outliers were not observed in our results (Table 4).In summary, the bidirectional results remained stable.

Discussion
In our comprehensive study, we conducted MR to investigate the causal relationship between 744 immune cells and IVDD.Our analysis identified 15 immune cells significantly associated with IVDD, encompassing 6 protective factors and 9 risk factors.We utilized     (30).CD19 is increasing in the patient of IVDD (31).However, unswitched memory B cells lack the ability to produce antibodies, making them incapable of initiating an immune attack on the disc.CD38 facilitates inflammation, cell migration, phagocytosis, and antigen presentation (32).CD19 on IgD+ CD38dim B cells appears to activate the immune response to the disc, thereby contributing to IVDD.
In our study, we found that a type of immune cell count called CD28-CD8+ T cells could increase the risk of IVDD.CD28 usually helps activate CD8 cells to boost the immune response (33).Surprisingly, even without CD28, CD8+ still contributes to IVDD (34).So we suppose that CD8 plays a more crucial role and CD28 might not be the main factor in IVDD.
Adjusted by multivariate MR, CD39+ CD8+ T cell Absolute Count and CD25 on CD39+ CD4+ T cells changed from protective to facilitating factors.CD19 on IgD+ CD38dim B cells changed from facilitating to protective factors.This may be due to the fact that the interference between immune cells leads to the masking of their true role for IVDDD.Consequently, it was only through multivariate MR that we unveiled their true independent effects.
We also discovered a bidirectional potential causal link between CD39+ CD4+ T cell %CD4+ T cell and IVDD.It is suggested that there is a vicious cycle between immune cells and IVDD.This is a consequence of the immune system being initiated by the deteriorated disc and the activated immune system have irreversible damage to the disc (10, 12,35).
Our MR analysis offers several advantages.Firstly, we employed univariable, multivariable, and bidirectional MR, mitigating confounding factors and reverse causality.Secondly, we conducted numerous sensitivity analyses to validate our hypotheses and minimize bias.Thirdly, we addressed population stratification issues by restricting the test population.
However, our study has limitations.It may not be directly applicable to other populations.The sample size was limited, potentially introducing bias, necessitating larger samples for robust results.Through we have utilized the univariable MR, bidirectional MR and multivariable MR, there is also unmeasured pleiotropy.Lastly, we refrained from using multiple analysis corrections as our primary goal was to identify potential biomarkers or therapeutic targets for IVDD.We opted not to apply the Bonferroni correction in our analysis since our primary focus was to identify potential therapeutic and prophylactic targets associated with IVDD.Bonferroni's criteria are quite stringent and could have resulted in the exclusion of meaningful indicators.

Conclusion
Our comprehensive MR analysis revealed 15 immune cells associated with IVDD risk through univariable MR.After accounting for confounding effects between immune cells, our multivariable MR identified only 9 immune cells as independent factors for IVDD, comprising 7 risk factors and 2 protective factors.These 9 immune cells can serve as potential biomarkers for IVDD, offering new perspectives on its treatment and prevention.Nonetheless, further experiments are needed to elucidate the underlying mechanisms.guardians/next of kin in accordance with the national legislation and the institutional requirements.

Exposure
FIGURE 3

TABLE 1
Multivariable MR immune cells on IVDD.

TABLE 3
Results of sensitivity analysis of blood immune cells on the risk of intervertebral disc disorder.

TABLE 2
Forest plots of the bidirectional MR between CD39+ CD4+ T cell % CD4+ T cell and IVDD.

TABLE 4
Results of sensitivity analysis between other intervertebral disc disorder and the risk of CD39+ CD4+ T cell %CD4+ T cell.