A mendelian randomization study on the association between 731 types of immune cells and 91 types of blood cells with venous thromboembolism

Abstract

Background

Venous thromboembolism (VTE) is a grave medical condition characterized by the blockage of distant blood vessels due to blood clots or detached vessel wall fragments, leading to ischemia or necrosis of the affected tissues. With the recent introduction of immunothrombosis, the significance of immune cells in the process of thrombus formation has gained prominent attention. Complex cross-talk occurs between immune cells and blood cells during infection or inflammation, with immune cells actively participating in blood clot formation by promoting platelet recruitment and thrombin activation. Nevertheless, comprehensive studies on the genetic association between immune cells phenotypes and VTE remain scarce. This article employed Mendelian randomization (MR) to investigate the association between the incidence of VTE and a range of 731 immune cell types, along with 91 blood cell perturbation phenotypes, utilizing single nucleotide polymorphisms as instrumental variables.

Methods

Through the utilization of publicly available genetic data, a two-sample bi-directional MR analysis was conducted. Sensitivity analyses included Cochran’s Q test, MR-Egger intercept test, MR-pleiotropy residual sum and outlier (MR-PRESSO) and leave-one-out analysis. For significant associations, replication analysis was conducted using GWAS data from deep vein thrombosis (DVT) and pulmonary embolism (PE).

Results

We firstly investigated the causal relationship between 731 immune cells and VTE risk. All the GWAS data were obtained from European populations and from men and women. The IVW analysis revealed that CD20 on naive-mature B cell, CD20 on IgD- CD38dim B cell, and CD20 on unswitched memory B cell may increase the risk of VTE (P < 0.05). CD28- CD8dim T cell %T cell, CD64 on monocyte and CD64 on CD14 + CD16- monocyte may be protective factors against DVT (P < 0.05). Then disturbed blood cells types as exposure were analyzed to examine its association with occurrence of VTE. Initial and replication analysis both revealed that environmental KCl-impacted red blood cells and butyric acid-impacted platelet accelerated incidence of VTE (P < 0.05), while colchicine -impacted eosinophil, KCl-impacted reticulocyte and Lipopolysaccharide (LPS) -impacted neutrophil reduced VTE risk (P < 0.05). Sensitivity analyses confirmed the robustness and reliability of these positive findings.

Conclusions

Our study presents evidence of a causal link between six immune cell phenotypes and VTE. Additionally, we have identified two types of blood cells that are associated with both VTE and DVT, and three types of blood cells that are relevant to both VTE and PE.

Clinical trial number

Not applicable.

Introduction

Venous thromboembolism (VTE), including deep vein thrombosis (DVT) and pulmonary embolism (PE), stands as a pressing global health concern, ranking as the third most prevalent cause of mortality across the world [1, 2]. Most risk factors of this disease have been widely recognized [3], and its pathophysiological mechanisms are often explained through the traditional Virchow’s triad, comprising hypercoagulability, venous stasis, and endothelial injury [4]. Furthermore, age, obesity, varicose veins, myelodysplastic syndromes [5] are also considered significant risk factors for VTE [6]. Iatrogenic factors such as improper medication use and handling during surgical procedures, may also lead to incidence of VTE [7]. Considering the severe peril VTE still presents to public health currently [8], a deep dive into the molecular mechanism and pathogenesis is exceedingly vital.

The concept of immune thrombosis has been brought to light recently [9,10,11], which explains a process in which platelets attach to the vessel wall, promoting the recruitment of immune cells especially for neutrophils and monocytes [12]. This process is bolstered by immune cells and thrombosis-related molecules, leading to construction of an intravascular scaffold. Besides, the focus of academic and pharmacological research has shifted from therapies targeting the traditional “cascade” reaction of coagulation to the immune components that drive coagulopathies [13]. Therefore, further research is necessary to investigate the impact of immune cells and blood cells related to thrombosis on VTE. This not only helps us better understand origins of the disease but also offers a scientific rationale for developing more effective prevention and treatment strategies.

Blood cellls including platelets [14], red blood cells [15] and white blood cells [16], play significant roles in venous thromboembolism, participating in the formation of blood clots and inflammatory processes. In the past, we usually focused on static clinical parameters such as total blood cell count, white blood cell volume and distribution, while the dynamic responses of blood cells to environmental conditions were neglected. Some studies show external environmental factors such as high altitude [17, 18], the use of specific hormonal medications [19, 20] and intestinal flora imbalance [21, 22] have an influence on VTE. However, it is currently unclear whether changes in blood characteristics and internal environment caused by environmental interference [23] can promote the formation and development of thrombosis. Hence, addressing these issues concerning causal relationships is vital for enhancing our comprehension of VTE and fostering the development of innovative prevention and treatment approaches.

Mendelian randomization (MR) refers to a statistical analysis method utilized to assess the causality of observed associations between beneficial exposures or risk factors and clinical outcomes [24]. Although Randomized Controlled Trials (RCTs) are regarded as the gold standard design for inferring causality, they are often expensive, time-consuming, and impractical to conduct. Similar to RCTs, the random segregation of genetic alleles during gametogenesis, when DNA is passed from parents to offspring, is expected to reduce confounding in genetic associations [25]. The large-scale genome-wide association studies conducted over the past decade have identified numerous genetic variations associated with cardiovascular metabolic traits, providing a vast amount of data sources for MR studies [26,27,28,29]. Our study utilizes MR analysis to explore the potential impact of immune and blood cells on VTE, with the objective of gaining novel insights into the pathogenesis of VTE.

Methods

Study design

The causal impact of 731 immune cells and 91 blood cells on VTE was explored in this study through MR analysis. A schematic outline of the study is depicted in Fig. 1. The MR analysis must meet the following conditions: (1) The instrumental variables (IVs) must be closely tied to exposure; (2) The IVs must be independent of confounders; (3) The IVs can only affect the outcome through exposure. This study was a secondary data analysis from the original study and didn’t need further ethical approval [24].

Fig. 1

The schematic diagram of the study’s design

Data sources

The study methods adhered to the STROBE-MR checklist [30]. Exposure data on 731 immune cell types were sourced from public GWAS databases, covering IDs GCST90001391 to GCST90002121 based on a sample of 3757 Europeans [31].The genetic data for 91 blood cell perturbation phenotypes were sourced from experiments conducted by Homilius et al. [23], which involved treating isolated human peripheral blood cells under 36 perturbation conditions designed to stimulate various physiological stress responses and expose the cells to chemical stressors, gut microbiota metabolites and drugs. The underlying cellular processes were investigated through functional readings obtained via flow cytometry and combinations of specific light scattering including side scatter (SSC) for internal complexity and granularity, forward scatter (FSC) for cell size and side fluorescence (SFL) for the presence and concentration of intracellular fluorescent markers [32]. Besides, using four fluorescence dyes containing white cell differential channel by fluorescence (WDF), white count and nucleated red blood cells (WNR), reticulocyte (RET) and platelet F (PLT-F)), the morphological and intracellular properties of blood cells were quantified statistically by cell counts, as well as Median (Med), Coefficient of Variation (CV) and Standard Deviation (SD) for each blood cell population. Based on these induced characteristics, a genome-wide association study was subsequently conducted on a sample of up to 2,600 individuals [33].

The primary outcome was incident VTE, a composite outcome including incident DVT and incident PE [26]. The secondary outcomes included incident DVT and PE respectively (Table S1). The GWAS summary data of VTE, DVT and PE was collected from the FinnGen database.

IVs selection

The IVs were selected according to the following criteria [34]: (1) Single nucleotide polymorphisms.

(SNPs) exhibiting significant associations with immune(p<5e-8) cells and blood cells(p<1e-5) were considered as potential IVs; (2) Independent SNPs were acquired with specific parameters to minimize LD effects, ensuring an r2 value less than 0.001 and a clumping distance of 10,000 kb; (3) SNPs that were palindromic (having A/T or G/C alleles) and those not appearing in the outcome were excluded; (4) Only SNPs with an F-statistic greater than 10 were retained to prevent weak instrument bias.

MR analysis

The five principal analysis methods used in MR studies are inverse variance weighting (IVW), MR‒Egger, weighted mode, simple mode and weighted median methods. The IVW method was the main method used in our MR analysis. The weighted median assumes that at least 50% of the weight comes from valid SNPs [35]. The weighted mode assumes that the most common effect is consistent with the true causal effect [36]. The slope of the MR‒Egger regression represents the causal estimate, but the estimation accuracy is very low [37]. The weighted median provides a precise estimation under the premise that at least half of the IVs are valid [35]. While the simple mode may not be as potent as IVW, it offers robustness against pleiotropy. The weighted mode is susceptible to the challenge of selecting an appropriate bandwidth for mode estimation [36].

Sensitivity analysis

Several sensitivity analyses were conducted in our study to guarantee the robustness of the results. The Cochran’s Q statistic was utilized to assess the heterogeneity of each SNP, and a P value less than 0.05 was considered indicative of significant variability among the estimates of the included SNPs [38]. The MR‒Egger intercept was used as an indicator of horizontal pleiotropy [39]. In addition, the MR- Pleiotropy Residual Sum and Outlier (MR-PRESSO) method was used to discern outliers and potential horizontal pleiotropy [40]. The detected outliers were excluded to yield a more precise corrected estimate. Furthermore, a leave-one-out analysis was performed to assess the stability of the MR estimates by sequentially excluding each selected SNP one by one.

Statistics analysis

The analyses were conducted utilizing the statistical software R, specifically version 4.3.3, which incorporated the TwoSampleMR version 0.6.8 for Mendelian randomization studies, along with the MRPRESSO package for additional robustness checks [40].

Results

Selection of IVs

In this study, the GWAS data underwent screening for IVs, ensuring that all selected IVs exhibited F values exceeding 10, thereby eliminating any concern for weak IVs bias. Following rigorous filtering criteria to maintain high quality, a total of 1892 SNPs associated with 731 immune cell subtypes and 987 SNPs related to 91 blood cell perturbations were selected as IVs. Detailed data on the IVs can be found in Table S2-S3.

MR analysis results

MR analysis of 731 immune cells on VTE, DVT and PE

MR analysis was conducted to investigate the causal impacts of immune cells on VTE, employing the IVW method as the primary analytical tool. Utilizing the finngen_R11_I9_VTE dataset, the MR analysis uncovered that 40 immune cell types had significant associations with VTE. Among these, 12 immune cell types were identified as risk factors for VTE. Notably, CD20 was detected on naive-mature B cells (odds ratio (OR) = 1.089, 95%CI = 1.0.38–1.142, P < 0.001) and on IgD + B cells (OR = 1.083, 95% CI = 1.030–1.138, P < 0.05), indicating as increased risk factors. Conversely, 28 immune cell types, exemplified by HLA-DR on CD33dim HLA DR + CD11b- cells (OR = 0.954, 95% CI = 0.929–0.981, P < 0.001) and CD127 on CD45RA + CD4 + T cells (OR = 0.923, 95% CI = 0.881–0.968, P < 0.001), were found to potentially decrease the risk of VTE. (Fig. 2, Table S4)

Fig. 2

A circus plot showcases the causal impacts of 731 immune cells on VTE, with five methodologies displayed in layers from the outer circle to the inner circle: inverse variance weighted, weighted median, simple mode, weighted mode, and MR Egger. The shades of color denote the size of the p-values

The MR analysis based on the finngen_R11_I9_PHLETHROMBDVTLOW dataset revealed that 33 immune cells were significantly associated with DVT. Among these, CD20 on naive-mature B cell (OR = 1.158, 95% CI = 1.064–1.260, P < 0.001) and CD20 on IgD + B cell (OR = 1.150, 95% CI = 1.054–1.255, P < 0.05) were identified as two of the 15 immune cells that may increase the risk of DVT. Additionally, CD45RA + CD8 + T cell Absolute Count (OR = 0.741, 95%CI = 0.639–0.858, P < 0.001) and CD28 on CD45RA- CD4 not regulatory T cell (OR = 0.878, 95% CI = 0.801–0.963, P < 0.05) were identified as two of the 18 immune cells that may decrease the risk of DVT (Fig. 3, Table S5).

Fig. 3

A circus plot illustrates the causal relationships of 731 immune cells with DVT, with five methods depicted in concentric circles from the outer circle to the inner circle: inverse variance weighted, weighted median, simple mode, weighted mode, and MR Egger. The color shades indicate the statistical significance of the p-values

The MR analysis based on the finngen_R11_I9_PULMEMB dataset revealed that 18 immune cells were significantly associated with PE. Among these, HLA DR on CD14- CD16- (OR = 1.124, 95% CI = 1.041–1.124, P < 0.05) and CD20 on IgD- CD38dim B cell (OR = 1.094, 95% CI = 1.015–1.179, P < 0.05) were identified as two of the 6 immune cells that may increase the risk of PE. Additionally, CD28 on CD39 + CD4 + T cell (OR = 0.947, 95%CI = 0.909–0.987, P < 0.05) and CD28- CD8dim T cell %T cell (OR = 0.884, 95% CI = 0.802–0.975, P < 0.05) were identified as two of the 12 immune cells that may decrease the risk of PE (Fig. 4, Table S6).

Fig. 4

The circus plot presents the causal relationships between 731 immune cells and PE, with five methods displayed in layers from the outermost layer to the innermost layer: inverse variance weighted, weighted median, simple mode, weighted mode, and MR Egger. The color shades indicate the p-value magnitudes, reflecting the statistical significance of the results

To ensure the quality and robustness of our study, we chose 6 shared immune cells from the three datasets in our MR analysis results (Supplementary Fig. 1). CD20 on naive-mature B cell, CD20 on IgD- CD38dim B cell and CD20 on unswitched memory B cell are three immune cells that may be risk factors for VTE. Besides, the following another three immune cells may serve as protective factors for VTE: CD28- CD8dim T cell %T cell, CD64 on monocyte and CD64 on CD14 + CD16- monocyte(Fig. 5).

Fig. 5

Forest plot of the causal effects of shared immune cells which have significant influence on VTE, DVT and PE. nSNP: number of single nucleotide polymorphisms; OR: odds ratio, 95% CI: 95% confidence interval

MR analysis of 91 blood cells on VTE, DVT and PE

After applying strict filtering criteria to control quality, a total of 987 SNPs associated with 91 blood cell perturbations were selected as IVs. All relevant SNPs had F statistics greater than 10, indicating that the IVs are highly effective. Using these IVs in an IVW analysis, we initially identified ten blood cell perturbation phenotypes with potential causal effects on VTE.

The IVW estimates for these phenotypes were all statistically significant (Fig. 6). The phenotypes identified were as follows: Platelet perturbation response (platelet count in response to butyric acid perturbation measured by platelet-F dye), abbreviated as Platelet-ba (P = 0.011, OR = 1.054, 95% CI = 1.012–1.098); Red blood cell perturbation response (side fluorescence coefsficient of variation of RBC in response to H2O2 perturbation measured by platelet-F dye), abbreviated as RBC-H2O2 (P = 0.027, OR = 1.039, 95% CI = 1.004–1.075); Red blood cell perturbation response (forward scatter standard deviation of RBC in response to nigericin perturbation measured by platelet-F dye), abbreviated as RBC-nigericin (P = 0.021, OR = 0.961, 95% CI = 0.929–0.994); Red blood cell perturbation response (forward scatter standard deviation of RBC 1 in response to KCl perturbation measured by reticulocyte dye), abbreviated as RBC1-KCl (P = 0.011, OR = 1.054, 95% CI = 1.012–1.098); Reticulocyte perturbation response (side scatter standard deviation of reticulocyte 1 in response to KCl perturbation measured by reticulocyte dye), abbreviated as Reticulocyte- KCl (P = 0.034, OR = 0.963, 95% CI = 0.931–0.997); Platelet perturbation response (side scatter standard deviation of platelet in response to Pam3CSK4 perturbation measured by reticulocyte dye), abbreviated as Platelet-Pam3CSK4 (P = 0.017, OR = 1.014, 95% CI = 1.002–1.026); Red blood cell perturbation response (side fluorescence median of RBC 2 in response to TMAO perturbation measured by reticulocyte dye), abbreviated as RBC2-TMAO (P = 0.026, OR = 1.021, 95% CI = 1.002–1.040); Eosinophil perturbation response (side fluorescence median of eosinophil 2 in response to colchicine perturbation measured by WDF dye), abbreviated as Eosinophil-colchicine (P = 0.015, OR = 0.955, 95% CI = 0.920–0.991); Neutrophil perturbation response (side fluorescence coefficient of variation of neutrophil 4 in response to Lipopolysaccharide/LPS perturbation measured by WDF dye), abbreviated as Neutrophil-LPS (P = 0.036, OR = 0.968, 95% CI = 0.940–0.998); Neutrophil perturbation response (neutrophil 2/neutrophil 4 ratio in response to water perturbation measured by WDF dye), abbreviated as Neutrophil2/4-water (P = 0.022, OR = 0.961, 95% CI = 0.929–0.994).

Based on our results, simulated physiological stressors such as KCl and water primarily alter the osmolarity of the extracellular fluid, thereby disrupting homeostasis within the internal environment. These changes can affect cellular morphology, granularity and the number of red blood cells and neutrophil and lead to blood stasis or turbulence, which can promote the formation of blood clots. Chemical stressors mimic the onset of inflammation, for instance, H2O2 induces oxidative stress in red blood cells and stimulates inflammatory responses. LPS activates neutrophil chemotaxis, phagocytosis and the release of cytokines to inhibit thrombosis. Pam3CSK4, a synthetic triacylated lipopeptide (LP) and a ligand for TLR2/TLR1, serves as an effective activator of the pro-inflammatory transcription factor NF-κB. Pam3CSK4 may accelarate platelet aggregation and activation to promote production of thrombosis. Drugs such as colchicine, used in the treatment of gout, can exhibit anti-inflammatory effects. Gut microbiota metabolites, like butyric acid, may promote platelet production, adhesion and aggregation. Nigericin causes oxidative damage to red blood cells, leading to their death and decrease the incidence of thrombosis. Additionally, TMAO, trimethylamine N-oxide, potentially alters the ratio of phospholipids to proteins on the red blood cell membrane, resulting in decreased membrane fluidity or increased membrane fragility and leading to the release of free hemoglobin, which may increase blood viscosity, thereby affecting blood flow and thrombosis.

Fig. 6

Forest plot of the causal effects of 91 blood cells on VTE. nSNP: number of single nucleotide polymorphisms; OR: odds ratio, 95% CI: 95% confidence interval; RBC1-KCl, Red blood cell perturbation response (forward scatter standard deviation of RBC 1 in response to KCl perturbation measured by reticulocyte dye); Platelet-ba, Platelet perturbation response (platelet count in response to butyric acid perturbation measured by platelet-F dye); Eosinophil-colchicine, Eosinophil perturbation response (side fluorescence median of eosinophil 2 in response to colchicine perturbation measured by WDF dye); Platelet-Pam3CSK4, Platelet perturbation response (side scatter standard deviation of platelet in response to Pam3CSK4 perturbation measured by reticulocyte dye); RBC-nigericin, Red blood cell perturbation response (forward scatter standard deviation of RBC in response to nigericin perturbation measured by platelet-F dye); Neutrophil2/4-water, Neutrophil perturbation response (neutrophil 2/neutrophil 4 ratio in response to water perturbation measured by WDF dye); RBC2-TMAO, Red blood cell perturbation response (side fluorescence median of RBC 2 in response to TMAO perturbation measured by reticulocyte dye); RBC-H2O2, Red blood cell perturbation response (side fluorescence coefsficient of variation of RBC in response to H2O2 perturbation measured by platelet-F dye); Reticulocyte- KCl, Reticulocyte perturbation response (side scatter standard deviation of reticulocyte 1 in response to KCl perturbation measured by reticulocyte dye); Neutrophil-LPS, Neutrophil perturbation response (side fluorescence coefficient of variation of neutrophil 4 in response to LPS perturbation measured by WDF dye)

To enhance the credibility of our estimation results, we replicated MR analyses separately using DVT and PE outcome GWAS datasets. The results indicate that six types of blood cell perturbation phenotypes are significantly associated with DVT. As expected, Platelet-ba (P = 0.009, OR = 1.05, 95%CI = 1.01–1.08) and RBC1-KCl (P = 0.031, OR = 1.08, 95% CI = 1.01–1.14) also displayed similar results in DVT. (Fig. 7)

Fig. 7

Forest plot of the causal effects of 91 blood cells on DVT. nSNP: number of single nucleotide polymorphisms; OR: odds ratio, 95% CI: 95% confidence interval; Platelet-ba, Platelet perturbation response (platelet count in response to butyric acid perturbation measured by platelet-F dye); Immature platelet, Immature platelet fraction perturbation response (side fluorescence standard deviation of IPF at baseline measured by platelet-F dye); RBC1-KCl, Red blood cell perturbation response (forward scatter standard deviation of RBC 1 in response to KCl perturbation measured by reticulocyte dye); WBC, White blood cell perturbation response (forward scatter median of WBC 1 at baseline measured by platelet-F dye); Neutrophil3, Neutrophil perturbation response (side fluorescence coefficient of variation of neutrophil 3 in response to empagliflozin perturbation measured by WDF dye); Neutrophil4, Neutrophil perturbation response (side fluorescence coefficient of variation of neutrophil 4 at baseline measured by WDF dye)

When the outcome is changed to PE the results show that six types of blood cell perturbation phenotypes are significantly associated with PE. Specifically, Neutrophil-LPS (P = 2.24E-04, OR = 0.92, 95%CI = 0.88–0.96), Reticulocyte-KCl (P = 1.09E-02, OR = 0.95, 95% CI = 0.91–0.99) and Eosinophil-colchicine (P = 4.59E-02, OR = 0.94, 95% CI = 0.89-1.00) also exhibited comparable findings in PE. (Fig. 8)

Fig. 8

Forest plot of the causal effects of 91 blood cells on PE. nSNP: number of single nucleotide polymorphisms; OR: odds ratio, 95% CI: 95% confidence interval; Neutrophil-LPS, Neutrophil perturbation response (side fluorescence coefficient of variation of neutrophil 4 in response to LPS perturbation measured by WDF dye); Monocyte, Monocyte perturbation response (side scatter median of monocyte 2 at baseline measured by WDF dye); Reticulocyte- KCl, Reticulocyte perturbation response (side scatter standard deviation of reticulocyte 1 in response to KCl perturbation measured by reticulocyte dye); Reticulocyte-rotenone, Reticulocyte perturbation response (side fluorescence coefficient of variation of reticulocyte 1 in response to rotenone perturbation measured by reticulocyte dye); Neutrophil-colchicine, Neutrophil perturbation response (side fluorescence coefficient of variation of neutrophil 4 in response to colchicine perturbation measured by WDF dye); Eosinophil-colchicine, Eosinophil perturbation response (side fluorescence median of eosinophil 2 in response to colchicine perturbation measured by WDF dye)

In summary, we found that Platelet-ba and RBC1-KCl both are risk factors for VTE. However, Neutrophil-LPS, Reticulocyte-KCl and Eosinophil-colchicine are all protective for VTE. Based on initial readings, we investigate the genetic regions linked to traits exhibiting significant responses to perturbations both in primary and second analyses, which participate in transcription regulation, cell apoptosis, metabolic processes and so on. Subsequently, potential cellular responses under various perturbation conditions are presented as follows (Table 1).

Table 1 Candidate genes and celluar responses of significant blood perturbation phenotypes

Reverse analysis

To assess whether VTE (comprising DVT and PE) can induce alterations in immune cell subtypes and blood cell abnormalities, we considered venous thromboembolism as the exposure factor, with the resulting changes in immune cell subtypes and blood cell disturbances serving as the outcomes. After selecting SNPs significantly correlated with VTE, DVT and PE as IVs (Table S7-S9), we performed a reverse MR analysis on the six main significant immune cells and five primary blood cell disturbances identified in the initial and second MR analysis. The results (Table 2 Reverse mendelian randomization analysis results) showed that VTE has no causal effects on CD20 on naive-mature B cell (OR = 0.912, 95% CI = 0.828–1.004, P = 0.060), CD20 on IgD- CD38dim B cell (OR = 0.999, 95% CI = 0.909–1.099, P = 0.987), CD20 on unswitched memory B cell (OR = 0.955, 95% CI = 0.869–1.050, P = 0.340), CD28- CD8dim T cell %T cell (OR = 1.034, 95% CI = 0.937–1.161, P = 0.437), CD64 on monocyte (OR = 0.958, 95% CI = 0.872–1.052, P = 0.370), CD64 on CD14 + CD16- monocyte (OR = 0.977, 95% CI = 0.894–1.068, P = 0.605), RBC1-KCl (OR = 1.101, 95% CI = 0.830–1.462, P = 0.504), Platelet-ba (OR = 1.004, 95% CI = 0.668–1.508, P = 0.985), Eosinophil-colchicine (OR = 1.112, 95% CI = 0.8888–1.392, P = 0.357), Neutrophil-LPS (OR = 1.139, 95% CI = 0.937–1.384, P = 0.190), Reticulocyte-KCl (OR = 0.977, 95% CI = 0.798–1.196, P = 0.821).

Table 2 Reverse Mendelian randomization analysis results

Sensitivity analysis

The Cochran’s Q test provided strong evidence for the absence of heterogeneity (Table 3 Sensitive analysis results for causality from immune cells and blood cells perturbation on VTE). No directional pleiotropy was found in the MR-Egger regression and MR-PRESSO Global test. The leave-one-out analysis results support that a single SNP will not cause bias in MR estimation.

Table 3 Sensitive analysis results for causality from immune cells and blood cells perturbation on VTE

Discussion

The potential causal relationship of immune and blood cells phenotypes on VTE was explored in this study using GWAS data through two-sample MR analysis. We found that CD20 on naive-mature B cell, CD20 on IgD- CD38dim B cell and CD20 on unswitched memory B cell were suggestively associated with an increased risk of VTE, whereas CD28- CD8dim T cell %T cell, CD64 on monocyte and CD64 on CD14 + CD16- monocyte were suggestively associated with a decreased risk of VTE. We also discovered that platelet count in response to butyric acid perturbation measured by platelet-F dye and forward scatter standard deviation of RBC 1 in response to KCl perturbation measured by reticulocyte dye both are risk factors for VTE, however, side fluorescence coefficient of variation of neutrophil 4 in response to LPS perturbation measured by WDF dye, side scatter standard deviation of reticulocyte 1 in response to KCl perturbation measured by reticulocyte dye and side fluorescence median of eosinophil 2 in response to colchicine perturbation measured by WDF dye are all protective for VTE. These findings contribute to our comprehension of immune thrombosis’s role in VTE and suggest potential biomarkers for disease prediction.

The formation of thrombosis has been proven to be an integral part of innate immunity, namely immune thrombosis [41, 42]. CD20 is a phosphatidylinositol-anchored protein primarily expressed on the surface of B lymphocytes and serves as a hallmark differentiation antigen for B cells [43]. Moreover, CD20 consists of 297 amino acid residues with a molecular weight of approximately 33 to 37 kDa and has four transmembrane domains. Over 95% of B-cell lymphomas abnormally overexpress CD20, and CD20 is an established therapeutic target for B-cell malignancies [44,45,46,47]. The development of humoral immunity requires functional BCR signaling pathways. It has been reported that CD20 co-localizes with lipid rafts39 and directly physically interacts with BCR [48]. The BCR is crucial for B cells to correctly induce immune responses [49]. Naive-mature B cells, IgD- CD38dim B cells, and unswitched memory B cells, all expressing CD20, are destined to differentiate into plasma cells that secrete antibodies. Early clinical evidence suggests that antibodies, regardless of their origin, can lead to thrombosis [50]. This prothrombotic function of antibodies has also been confirmed in an experimental study [51]. Initially, IgM attaches to the endothelium and activates it via FcµR and pIgR, causing the exposure of P-selectin and vWF on its surface, which triggers the recruitment of platelets to veins. Then, the fibrinogen and CS-A on activated platelets cause IgG to deposit on the platelet surface, triggering the activation of pathway of components [52], thereby initiating a vicious cycle of platelet and neutrophil activation that promotes thrombosis [53, 54]. Besides, a previous cohort study found that the use of immunoglobulin therapy in patients with dermatomyositis increases the risk of venous thromboembolism [55]. The results of our Mendelian randomization study further support these findings, confirming that CD20 on naive-mature B cell, CD20 on IgD- CD38dim B cell and CD20 on unswitched memory B cell are potential risk factors for VTE. CD28 is a costimulatory molecule on the surface of T cells that plays a crucial role in the activation process of T cells [56, 57]. T cells recognize antigens through T-cell receptors (TCRs) and require the interaction between CD28 and CD80/CD86 on antigen-presenting cells [58]. Stimulation of CD28 promotes T-cell proliferation, survival and cytokine production [59, 60]. The CD28/B7 system is one of the primary co-stimulatory pathways [61]. Evidence suggests that the regulation of co-stimulatory and co-inhibitory pathways involving molecules of the B7-CD28 and TNF-TNFR families promotes T-cell responses in atherosclerotic diseases [62, 63] and hypertension [64]. Furthermore, the inhibition of CD28-dependent signaling leads to a significant reduction in tissue-infiltrating T cells and potent suppression of vasculitis [65]. This is consistent with our research findings, where the percentage of CD28- CD8dim T cells among total T cells serves as a protective factor against venous thromboembolism. CD64 is a high-affinity Fc gamma receptor type I (FcγRI) [66], primarily distributed on the surface of macrophages, monocytes, and dendritic cells, serving as a bridge between humoral and cellular immunity. In normal peripheral blood, CD64 is predominantly expressed on CD14 + CD16- monocytes and macrophages [67]. Although some studies have confirmed that monocytes promote thrombosis [53, 68], our Mendelian randomization study found them to be a protective factor. In terms of mechanism, monocytes/macrophages may key cell types involved in the resolution of thrombosis [69, 70].

Both human and mouse DVT originate from a large number of red blood cells and platelets, becoming entangled with fibrinogen and ultimately transforming into collagen guided by various types of white blood cells. In our study, environmental KCl-impacted red blood cells and butyric acid-impacted platelet both accelerated incidence of VTE. KCl primarily exert their influence by altering the osmolarity of the extracellular fluid, ultimately disrupting the homeostasis of the internal environment. These alterations can induce changes in cellular morphology, increase granularity and modify the count of red blood cells, leading to blood stasis or turbulence and favoring the formation of blood clots. Butyrate has been proven to stimulate bovine neutrophil function and enhances platelet-activating factor (PAF) [71]. Moreover, the PAF signaling system can trigger inflammatory and thrombotic cascades [72]. Colchicine, plant-based salt derived from saffron, is a classical medication used for gout attacks. Besides, colchicine as an NLRP3 inhibitor is increasingly being utilized in the treatment of cardiovascular diseases [73, 74]. Furthermore, a low eosinophil count is associated with an increased risk of DVT and PE [75]. LPS, the outer wall of the Gram-negative bacterial cell, is composed of lipids and polysaccharides. Studies have confirmed that LPS has the ability to promote thrombosis, acting on cells involved in the process of thrombosis such as platelets, white blood cells, and endothelial cells [76]. However, firstly at the genetic level, neutrophils perturbed by LPS play a protective role in both venous thromboembolism and pulmonary embolism. When neutrophils are stimulated by LPS, they may release some substances with anticoagulant effects, such as heparin-like molecules or antithrombin. They may also promote the activation of the fibrinolytic system by releasing certain factors, thereby accelerating the dissolution of blood clots. Additionally, neutrophils can interact with other immune cells to jointly maintain the immune homeostasis of the body. Furthermore, we also found that the more reticulocytes affected by KCl, the lower the possibility of thrombosis. KCl may interfere with the process of reticulocyte development into mature red blood cells, reducing the number of red blood cells and decreasing blood viscosity. Nevertheless, more research is necessary to delve deeper into the issue.

This research focuses on analyzing the genetic causality linking immune cells, blood cells, and conditions like VTE, DVT and PE through bi-directional MR analysis, leveraging extensive GWAS summary data. However, it does have its constraints. Primarily, our study only involves a European cohort, necessitating further validation for application to other demographics. Additionally, we did not stratify our analysis by gender, which might lead to some variations in single-sex evaluations. Thirdly, our reverse MR results did not indicate a statistically significant effect of the disease on immune or blood cells, although this lack of significance is confined to the statistical realm. Actually in vivo environment, however, thrombosis may lead to changes in the blood environment or affect the state and function of surrounding cells through signal transduction. This represents a limitation of our study. Lastly, our investigation solely delves into genetic causality, urging a comprehensive consideration of disease complexity when applying our conclusions to predictive assessments of VTE, DVT and PE.

Conclusions

We utilized MR analysis to investigate the causal relationships between various immune cells, blood cells and VTE, potentially offering new directions for research on immune cells and blood cells as biomarkers for VTE diagnosis and treatment. However, to fully understand the relationship between the behavior of immune cells, blood cells and VTE, it is crucial to conduct in vivo, in vitro and pre-clinical studies and further explore their underlying mechanisms and biological processes involved, thus providing a basis for innovative prevention or treatment strategies targeting immune cells and blood cells in VTE.