Open access journal of the Ferrata-Storti Foundation, a non-profit organization
Articles

Two-sample Mendelian randomization analysis reveals causal relationships between blood lipids and venous thromboembolism

Key Laboratory of Biomedical Information Engineering of Ministry of Education, Biomedical Informatics and Genomics Center, School of Life Science and Technology, Xi'an Jiaotong University, Xi'an, Shaanxi, P. R. China, 710049; Department of Trauma Surgery, Honghui Hospital, College of Medicine, Xi'an Jiaotong University, Xi'an, Shaanxi, P. R. China, 710054
Key Laboratory of Biomedical Information Engineering of Ministry of Education, Biomedical Informatics and Genomics Center, School of Life Science and Technology, Xi'an Jiaotong University, Xi'an, Shaanxi, P. R. China, 710049
Key Laboratory of Biomedical Information Engineering of Ministry of Education, Biomedical Informatics and Genomics Center, School of Life Science and Technology, Xi'an Jiaotong University, Xi'an, Shaanxi, P. R. China, 710049
National and Local Joint Engineering Research Center of Biodiagnosis and Biotherapy, The Second Affiliated Hospital, Xi'an Jiaotong University, Xi'an, Shaanxi, P. R. China, 710004
Key Laboratory of Biomedical Information Engineering of Ministry of Education, Biomedical Informatics and Genomics Center, School of Life Science and Technology, Xi'an Jiaotong University, Xi'an, Shaanxi, P. R. China, 710049
Key Laboratory of Biomedical Information Engineering of Ministry of Education, Biomedical Informatics and Genomics Center, School of Life Science and Technology, Xi'an Jiaotong University, Xi'an, Shaanxi, P. R. China, 710049
Department of Trauma Surgery, Honghui Hospital, College of Medicine, Xi'an Jiaotong University, Xi'an, Shaanxi, P. R. China, 710054
Department of Trauma Surgery, Honghui Hospital, College of Medicine, Xi'an Jiaotong University, Xi'an, Shaanxi, P. R. China, 710054
Key Laboratory of Biomedical Information Engineering of Ministry of Education, Biomedical Informatics and Genomics Center, School of Life Science and Technology, Xi'an Jiaotong University, Xi'an, Shaanxi, P. R. China, 710049
Department of Trauma Surgery, Honghui Hospital, College of Medicine, Xi'an Jiaotong University, Xi'an, Shaanxi, P. R. China, 710054
Key Laboratory of Biomedical Information Engineering of Ministry of Education, Biomedical Informatics and Genomics Center, School of Life Science and Technology, Xi'an Jiaotong University, Xi'an, Shaanxi, P. R. China, 710049
Vol. 109 No. 9 (2024): September, 2024 https://doi.org/10.3324/haematol.2023.284566

Abstract

Venous thromboembolism (VTE) is a complex disease that can be classified into two subtypes: deep vein thrombosis (DVT) and pulmonary embolism (PE). Previous observational studies have shown associations between lipids and VTE, but causality remains unclear. Hence, by utilizing 241 lipid-related traits as exposures and data from the FinnGen consortium on VTE, DVT, and PE as outcomes, we conducted two-sample Mendelian randomization (MR) analysis to investigate causal relationships between lipids and VTE, DVT and PE. The MR results identified that fatty acid (FA) unsaturation traits (ratio of bis-allylic bonds to double bonds in lipids, and ratio of bis-allylic bonds to total fatty acids in lipids) were associated with VTE (odds ratio [OR]=1.21, 95% confidence interval [CI]: 1.15-1.27; OR=1.21, 95% CI: 1.13-1.30), DVT (OR=1.24, 95% CI: 1.16-1.33; OR= 1.26, 95% CI: 1.16-1.36) and PE (OR=1.18, 95% CI: 1.08-1.29; OR=1.18, 95% CI: 1.09-1.27). Phosphatidylcholines (PC) exhibit potential causal effects on VTE and PE. PC acyl-alkyl C40:4 (PC ae C40:4) was negatively associated with VTE (OR=0.79, 95% CI: 0.73-0.86), while PC diacyl C42:6 (PC aa C42:6) and PC acyl-alkyl C36:4 (PC ae C36:4) were positively associated with PE (OR=1.44, 95% CI: 1.20-1.72; OR=1.22, 95% CI: 1.10-1.35). Additionally, we found that medium LDL had a protective effect on VTE. Our study indicates that higher FA unsaturation may increase the risk of VTE, DVT, and PE. Different types of PC have either promotive or inhibitory effects on VTE and PE, contributing to a better understanding of the risk factors for VTE.

Introduction

Venous thromboembolism (VTE) is a highly prevalent and complex chronic disease, which can manifest specifically as deep vein thrombosis (DVT) and pulmonary embolism (PE) based on the location of the blood clot.1 DVT occurs when a blood clot forms in the deep veins of the leg or pelvis. Once part of the clot detaches and enters the pulmonary artery through the circulation, it may lead to PE. PE can be life-threatening due to oxygen deprivation and circulatory failure.2 VTE is influenced by environmental and genetic factors, and various factors such as aging, major surgery, prolonged immobility, malignancies, and obesity may affect the risk of VTE.3,4 Approximately 10 million people are affected by VTE annually, making it the third largest vascular disease after acute myocardial infarction and stroke, contributing significantly to the global disease burden.1

Lipids are a common and diverse class of compounds that play important biological functions in various aspects, including serving as structural components of cell membranes, energy storage sources, and participating in signaling pathways.5 The different blood lipids and lipoproteins have procoagulant and anticoagulant functions, implying that blood lipids may be linked to venous thrombosis.6 Previous observational studies7-9 have reported that the fluctuation of lipids was associated with VTE. For example, a population-based case-control study found that elevated triglyceride (TG) was associated with an increased risk of venous thrombosis in postmenopausal women, while higher level of high-density lipoprotein cholesterol (HDL-C) was associated with a decreased risk.7 Another study compared plasma lipid profiles between patients with post-VTE and those without VTE, and revealed phosphatidylcholine (PC) and TG were higher in most of the historical patients with VTE.8 Jiang et al. investigated 240 cases of VTE (including 125 cases of PE) and 6,963 controls. They found a significant association between C5 carnitine and VTE events, while confirming elevated levels of diacylglycerol in VTE and PE patients.9 Although a large amount of evidence confirming the association between blood lipids and VTE, most studies are observational and subject to sample size limitations and confounding factors. The causal link between blood lipids and VTE remains unclear.

Two-sample Mendelian randomization (MR) utilizes genetic variants of the exposure as instruments to estimate the potential causal association between exposures and outcomes. MR provides a reliable method to assess the causal relationship between genetic risk factors and phenotypic outcomes from a genetic perspective, which ensures that the estimation is less likely to be influenced by environmental confounding.10 Based on this robust method, the genome-wide association studies (GWAS) with large sample size have identified multiple genetic variations on lipids and lipid-related traits, which may provide a great deal of genetic instrumental variables for causality estimation.11 Several MR studies have been conducted to evaluate the causal relationship between lipids and VTE.10,12-14 Lin et al. utilized bidirectional MR analysis to investigate the relationship between three classical lipids (low-density lipoprotein [LDL], HDL, and TG) and VTE, and found no significant causal association.12 Another two-sample MR study exploring the causal relationship between five circulating lipids (apolipoprotein A1, apolipoprotein B, LDL, HDL, and TG) and DVT also yielded similar conclusions.14 MR studies investigating the causality between fatty acids (FA) and VTE suggest that different types of FA have different inhibitory or protective effects on VTE.10,13 However, most of MR studies focus on only a subset of lipids, and currently, causal relationship between lipids and the risk of VTE still need to be confirmed in larger samples.

In this study, we conducted two-sample MR analyses to investigate the causal effects between blood lipids and VTE, DVT and PE, respectively. The lipids and lipid-related traits including PC, sphingomyelin, acylcarnitine, FA, lipoproteins, and others. Our findings provide new insights into the relationship between endogenous lipid metabolism and VTE, and contribute to a better understanding of the risk factor for VTE from a genetic perspective.

Methods

Study design

We performed a two-sample MR study to explore the possible causal effects of 241 blood lipids and lipid-related phenotypes on VTE, DVT and PE, respectively. The outline of the study design is shown in Figure 1.

Genome-wide association study data sources

A total of 241 blood lipids and lipid-related phenotypes were used as exposures (Online Supplementary Table S1), with 139 lipids derived from proof-of-concept cross-platform GWAS study.15 This study provided GWAS summary data for each metabolite with sample sizes ranging from 8,569 to 86,507 individuals. In addition, 98 lipids and lipid-related traits were derived from a study published in 2016.16 The sample sizes for different metabolites in this study ranged from 13,000 to 19,000. Finally, the GWAS summary statistics of LDL cholesterol (LDL-C), HDL-C, total cholesterol (TC) and TG were obtained from the Million Veteran Program (MVP), which involved up to 215,551 European individuals. The “aa” and “ae” denote that FA are bound to the glycerol backbone via ester or ether bonds. The x and y in “x:y” indicate the number of carbon atoms and double bonds in the FA chain of the lipids. The outcome data for VTE, DVT and PE were available from the FinnGen study, comprising 377,277 individuals (19,372 cases and 357,905 controls), 333,230 individuals (9,109 cases and 324,121 controls) and 376,351 individuals (9,243 cases and 367,108 controls), respectively. All data sets used in this study had been approved by a relevant ethical review board.

Data filtration and genetic instruments selection

We filtered all GWAS summary datasets according to the following steps: i) removing single nucelotide variants (SNP) located in the major histocompatibility complex (MHC) region; ii) removing SNP with minor allele frequency (MAF) less than 0.01; iii) removing palindrome SNP with alleles A/T or G/C and MAF close to 0.5. According to the three assumptions that MR genetic instrumental variables (IV) must satisfy, we selected SNP that are genome-wide significance (P value threshold <5×10-8), not in linkage equilibrium (r2 threshold >0.001, window size =1,000kb), and free of weak instrument bias (F-statistic >10). The RadialMR package was used to remove outlier pleiotropic SNP.17 After IV selection, we harmonized the effect alleles and adjusted |3 values in the outcome data to make it consistent with the exposure data.18

Statistical analyses

Consistent with our previous studies,19,20 we conducted two-sample MR analyses using five methods, including inverse-variance weighted (IVW), robust adjusted profile score (RAPS), MR-Egger, weighted median and weighted mode, with IVW as the primary method. All MR analyses were implemented by the TwoSampleMR R package.21 The intercept of MR-Egger regression can be used to detect the pleiotropy in MR estimates.22 Cochran’s Q statistic (Q) and Rucker’s Q statistic (Q’) are used to assess heterogeneity in IVW and MR-Egger estimates, respectively.23 The difference between the two Q statistics (Q-Q’) can be used to assess the horizontal pleiotropy of the MR estimates, while a P value of less than 0.05 for the Q statistics and Q-Q’ indicates the presence of directional pleiotropy.24 Leave-one-out (LOO) sensitivity analysis is conducted to detect the presence of potential dominant SNP. We also applied the Bonferroni correction to adjust for multiple testing. Complete details of genetic instruments selection and MR analysis are available in the Online Supplementary Appendix.

Results

The Mendelian randomization estimates of blood lipids and lipid-related phenotypes on venous thromboembolism

The complete results of MR analyses and pleiotropy assessment for 187 blood lipids and lipid-related traits on VTE are shown in Online Supplementary Table S2. We identified 20 blood lipids and related lipid-related traits that were causally associated with VTE according to the IVW method, of which 11 were PC, one was lysophosphatidylcholine, five traits related to FA saturation, and three traits related to medium LDL (P<2.67×10-4; Online Supplementary Table S2). The pleiotropy assessment showed that no significant evidence of pleiotropy was detected by the Cochran’s Q test and the intercept of the MR-Egger method (P>0.05). However, there were six PC (PC aa C36:4, PC ae C36:3, PC ae C36:2, PC ae C38:2, PC ae C42:3, PC aa C34:4) with Q-Q’ differences that were extreme enough to suggest the presence of directional pleiotropy (P<0.05; Online Supplementary Table S2). Sensitivity analysis for remaining 14 blood lipids and lipid-related traits showed that eight of them had major influential SNP driving causal estimates, suggesting that the significant MR estimates for these blood lipids and lipid-related traits are not robust in terms of causal effects (Online Supplementary Figure S1).

After excluding 14 exposures that exhibit directional PC or sensitivity, we ultimately identified one PC (PC ae C40:4) and three traits related to medium LDL (total lipids in medium LDL, concentration of medium LDL particles, total cholesterol in medium LDL) had protective effects on VTE, two FA saturation-related traits (ratio of bis-allylic bonds to double bonds in lipids, ratio of bis-allylic bonds to total fatty acids in lipids) showed pathogenic effects (Figure 2). These results were also validated in MR-RAPS (5/6) and weighted median (3/6) methods at the threshold of P value <2.67×10-4, and partially validated in weighted mode (6/6) and MR-Egger (3/6) methods with the threshold of P value <0.05. The F-statistics for the genetic instruments are all over the common threshold of ten, indicating that there is no weak instrumental bias (Online Supplementary Table S2).

Figure 1.Study design. GWAS: genome-wide association study; VTE: venous thromboembolism; DVT: deep vein thrombosis; PE: pulmonary embolism; IV: instrumental variable; MR: Mendelian randomization; IVW: inverse-variance weighted; MR-RAPS: Mendelian randomization robust adjusted profile score; SNP: single nucleotide variant; MAF: minor allele frequency.

The Mendelian randomization estimates of blood lipids and lipid-related phenotypes on deep vein thrombosis

There were 17 blood lipids and lipid-related traits that showed genetic causal relationships with DVT according to the IVW results (P<2.64×10-4; Online Supplementary Table S3). The heterogeneity and pleiotropy test found that all P values of MR-Egger intercepts and Q statistics were greater than 0.05, while the P values for the Q-Q’ of four PC (PC aa C34:2, PC ae C36:2, PC ae C38:2, PC aa C36:3) were less than 0.05, indicating the presence of pleiotropy (Online Supplementary Table S3). The LOO analysis found that eight PC and one FA-related trait had main effect SNP, contributing to the instability of the corresponding MR estimates (Online Supplementary Figure S2). Additionally, the MR analysis for sphingomyelin ceramide 16:1 to DVT included only two IV, allowing for IVW model analysis exclusively. It was impossible to evaluate heterogeneity and sensitivity, thus subsequent analyses were not included (Online Supplementary Table S3).

After removing 14 unstable exposures, we ultimately identified three FA saturation-related traits (ratio of bis-allylic bonds to double bonds in lipids, other polyunsaturated FA than 18:2, ratio of bis-allylic bonds to total FA in lipids) that showed a positive correlation with the risk for DVT (Figure 3). The results of MR-RAPS and weighted median methods for these three traits also met the strict threshold of significance (P<2.64×10-4), and all of them showed a suggestive causality to DVT in weighted mode and MR-Egger methods (P<0.05). The F-statistics for the genetic instruments are all greater than ten, indicating that there is no weak instrumental bias (Online Supplementary Table S3).

Figure 2.Causal effects of one phosphatidylcholine, two traits related to double bond composition and three traits related medium low-density lipoprotein on venous thromboembolism. Summary Mendelian randomization (MR) estimates derived from the inverse-variance weighted (IVW), MR robust adjusted profile score (MR-RAPS), weighted median, weighted mode, and MR-Egger methods. The error bars represent 95% confidence intervals (CI). VTE: venous thromboembolism, PC: phosphatidylcholine; LDL: low-density lipoprotein; OR: odds ratio.

Figure 3.Causal effects of three traits related to fatty acid unsaturation on deep vein thrombosis. Summary Mendelian randomization (MR) estimates derived from the inverse-variance weighted (IVW), MR robust adjusted profile score (MR-RAPS), weighted median, weighted mode, and MR-Egger methods. The error bars represent 95% confidence intervals (CI). DVT: deep vein thrombosis; OR: odds ratio.

Figure 4.Causal effects of two phosphatidylcholines and two traits related to double bond composition on pulmonary embolism. Summary Mendelian randomization (MR) estimates derived from the inverse-variance weighted (IVW), MR robust adjusted profile score (MR-RAPS), weighted median, weighted mode, and MR-Egger methods. The error bars represent 95% confidence intervals (CI). PC: phosphatidylcholine; PE: pulmonary embolism; OR: odds ratio.

The Mendelian randomization estimates of blood lipids and lipid-related phenotypes on pulmonary embolism

The complete results of MR analysis and pleiotropy evaluation for 189 blood lipids and lipid-related traits on PE can be found in Online Supplementary Table S4. After Bonferroni correction (P<2.64×10-4), we found that two PC (PC aa C42:6, PC ae C36:4) and two FA saturation-related traits (ratio of bis-allylic bonds to double bonds in lipids, ratio of bis-allylic bonds to total fatty acids in lipids) were positively correlated with PE using IVW, MR-RAPS, and weighted median methods (Figure 4). The MR-Egger intercepts, Q statistics and the difference Q-Q’ showed that no significant pleiotropy was detected in these MR results (Online Supplementary Table S4). The leave-one-out permutation did not identify any IV with major effects in MR estimation (Online Supplementary Figure S3). The F-statistics for the genetic instruments are all over the common threshold of 10, indicating that there is no weak instrumental bias (Online Supplementary Table S4). These results confirmed the reliability of putative causal effects in our MR analyses.

Discussion

In our study, we employed a two-sample MR approach to investigate the potential causal relationships between 241 blood lipids and lipid-related traits on VTE, DVT and PE. Our findings suggested that higher lipid unsaturation was linked to an increased risk of VTE, DVT and PE. Furthermore, we have revealed a causal relationship of PC on VTE and PE. MR estimates of medium LDL also demonstrate a protective effect to VTE. The current study provides a foundation to explore the metabolic risk factors of VTE, DVT and PE from the perspective of genetic mechanisms, which is helpful to guide future hypothesis-driven analyses. We have also summarized the biological insights or observational studies related to the causal relationship outcomes in Table 1 for reference.

Two traits associated with FA saturation (ratio of bis-allylic bonds to double bonds in lipids, ratio of bis-allylic bonds to total FA in lipids) showed a significant pathogenic effect on VTE, and another FA saturation-related trait (other polyunsaturated FA than 18:2) also had an impact on increasing the risk of DVT, indicating that the degree of unsaturation in lipids may be a risk factor for VTE, DVT and PE. The number of double bonds is related to the degree of unsaturation of FA, and the bis-allylic bonds refer to the presence of adjacent double bonds in a molecule. There is little research directly exploring the relationship between lipid unsaturation characteristics and the risk of VTE, but previous studies on the relationship between polyunsaturated FA (PUFA) and VTE seem to support our MR estimates. Maria et al. explored the involvement of PUFA biosynthesis in cardiovascular diseases in Europeans and East Asians and found that higher PUFA biosynthesis rates were associated with a higher risk of VTE.25 Arachidonic acid (AA) is the major PUFA that undergoes enzymatic oxidation, with cyclooxygenase and lipoxygenase enzymes extracting hydrogen atoms from its bis-allylic carbons to initiate oxidation, generating lipid radicals that then react with molecular oxygen.26 Higher levels of AA in the serum have been reported for association with a higher risk of VTE.27,28 The mechanism by which lipid unsaturation affects the risk of VTE may be related to oxidative stress. The presence of bis-allylic methylene between double bonds weakens the carbon-hydrogen bonds, forms carbon-centered radicals and/or hydroperoxides of unsaturated FA, which initiate radical-mediated chain reactions leading to a greater susceptibility of FA to oxidation.29 The oxidation of certain lipids produces substances with platelet-stimulating properties, such as the oxidation of LDL, which generates lysophosphatidylcholine, some oxidized phosphatidylcholine molecules, and lysophosphatidic acid (LPA). These lipoproteins or lipids activate platelets by stimulating G protein-coupled LPA receptors and the Rho/Rho kinase signaling pathway, resulting in platelet shape change and subsequent aggregation.30 The more unsaturated fatty acid chains in lipid, the more likely it is to be oxidized to produce reactive substances that promote platelet aggregation. Platelets are essential in hemostasis and are involved in thrombus formation through various mechanisms, including collagen-mediated activation occurring when collagen is exposed beneath the endothelium, adhesion mediated by ultra-large von Willebrand factor multimers, and platelet thrombus formation facilitated by neutrophil extracellular traps.31 In addition, the formation of certain lipid oxidation products can generate an excess of reactive oxygen species. These free radicals may damage vascular function, increase endothelial permeability, alter responsiveness to vasodilators, and promote the development of focal endothelial cell membrane lesions at very low levels through increased vascular relaxation and platelet aggregation.32 These events contribute to the progression of VTE by facilitating a series of events that support the formation of venous thrombosis.

In this study, we have identified eight lipid-VTE pairs and nine lipid-DVT pairs containing main SNP by using leave-one-out analysis. The MR results of these pairs may be driven by the pleiotropic effects of the specific variants rather than the causal effects of the risk factors. We annotated the located genes of these main SNP in Online Supplementary Table S5 and found that some SNP are located within specific genes related to lipid unsaturation. For example, rs174546 serves as the influential SNP driving the causal relationship between the trait of double bonds in fatty acids and VTE, while rs174547 drives the causal relationship between traits related to lipid unsaturation (other polyunsaturated FA than 18:2 and CH2 groups to double bonds) and VTE. Both SNP are located in FA desaturase 1 (FADS1) gene. This gene encodes a protein belonging to the FA desaturase gene family, which regulates FA unsaturation by introducing double bonds at specific carbons of the fatty acyl chain.33,34 Therefore, for exposures related to FA unsaturation, SNP located in this gene might be used as suitable IV. Two PC (PC aa C42:6, PC ae C36:4) had a positive causal relationship with PE, and PC ae C40:4 showed a negative correlation with VTE. Our results showed that the causality of PC with different carbon chain lengths and double bond numbers on VTE and PE are vary in both positive and negative directions. There is limited direct observational evidence suggesting an association between PC and PE. However, previous metabolomics studies have revealed a correlation between PC and the risk of venous thrombus formation. Sung et al.35 performed metabolomics study using serum and vascular wall extracted from DVT mice, and found that multiple PC were higher in DVT mice. The specific mechanism by which PC increase the risk of PE may be related to their stimulation of platelet aggregation. The PUFA chains of PC are oxidized to produce highly reactive decomposition products such as malondialdehyde and 4-hydroxynonenal, which potentiate platelet aggregation and thromboxane A2 formation in low concentration ranges.36,37 The alkyl-phosphatidylcholine and acyl-phosphatidylcholine oxidation products oxidize platelet activating factor (PAF) receptors or induce alterations in human platelet shape, which subsequently stimulate platelet aggregation and thus inducing thrombosis.38 Moreover, several distinct clinical studies and cohorts have shown that PC and choline are metabolized by the intestinal microbiota to form the gas trimethylamine, which is absorbed into the blood and converted to trimethylamine N-oxide (TMAO) by hepatic flavin monooxygenases.39 TMAO promotes thrombosis in vivo by stimulating Ca2+ release from intracellular stores and regulating platelet hyperreactivity and clot formation rate.40 Whether PC affects thrombosis through TMAO remains to be considered. It should be noticed that the effects of endogenous PC may not be identical to dietary intake. Additionally, the underlying mechanisms explaining inhibitory associations between PC and VTE are far from clear. Previous studies have found that the PC fraction of various yoghurts, such as PC (18:0/16:0) and PC (18:0/18:1), were inversely correlated with PAF and thrombin inhibition,41 and the same findings have been found for Salmon PC.42 The negative correlation between PC and VTE requires further experimental and clinical verification.

Table 1.Summary of the literature supporting findings of causality between specific lipids and venous thromboembolism, deep vein thrombosis and pulmonary embolism.

Our findings also identified negative causality of three phenotypes associated with medium LDL on VTE, including total lipids in medium LDL, concentration of medium LDL particles, and total cholesterol in medium LDL. LDL in plasma is a heterogeneous collection of particles, with differences in size, density, and composition among different subgroups of LDL particles.43 Subfractions of LDL characterized by particle size, particle number, and lipid composition have different effects on disease. Although the mechanisms underlying the association between medium LDL and VTE are less directly elucidated, numerous studies have shown a significant association between medium LDL and cardiovascular events,44,45 likely through their effects on endothelial cell function and lipid metabolism influencing thrombosis formation.46 Previous research has shown that medium-sized and medium-density LDL particles exhibit stronger binding affinity to LDL receptors compared to large, buoyant and small, dense LDL particles.47,4 8 LDL particles are highly sensitive to oxidative damage, and oxidized LDL is the primary modified form of native LDL.49 Ox-LDL has been proven to induce changes in platelet shape and aggregation, as well as to transform endothelial cells from an anticoagulant phenotype to a procoagulant phenotype, directly or indirectly promoting coagulation and thrombus formation.30,37 Therefore, the stronger binding affinity of medium LDL to LDL receptors may help reduce the generation of oxidized LDL, thereby lowering the risk of VTE.

Several limitations should be addressed in current study. Firstly, only 241 blood lipids and related traits were included in our study. More metabolites should be included in future studies, which is important for a more comprehensive understanding of the risk factors and underlying mechanisms of VTE. Secondly, MR studies can help determine whether the observed correlation has a causal relationship based on genetic evidence, which is considered as a causal hypothesis. To confirm the exact causal relationship between specific lipid forms and VTE, more laboratory research and clinical studies are often needed to reveal potential biological mechanisms. Although our study explored the potential mechanisms behind the causal relationship between specific lipid forms and VTE, further clinical trials and mechanistic studies are needed for validation.

In conclusion, our study provides MR evidence supporting a causal role of PC and lipid unsaturation in VTE, DVT, and PE. We hope to get a better understanding of the metabolic mechanisms underlying VTE and predict potential risk factors.

Footnotes

  • Received October 30, 2023
  • Accepted April 22, 2024

Correspondence

*WH, YZ and KZ contributed equally as co-first authors

Y. Guo guoyan253@xjtu.edu.cn

K. Zhang hhyyzk@126.com

Disclosures

No conflicts of interest to disclose.

Funding

This study was supported by grants from the National Natural Science Foundation of China (32170616, 82170896, 31970569, and 31871264), Key Research and Development Project of Shaanxi Province (2022SF-394), Science Fund for Distinguished Young Scholars of Shaanxi Province (2021JC02), Innovation Capability Support Program of Shaanxi Province (2022TD-44), the General project in the field of social development in Shaanxi Province (2024SF-YBXM-368) and the Fundamental Research Funds for the Central Universities.

Acknowledgments

The authors thank the participants of the contributing biobanks and consortia. We thank the Million Veteran Program (MVP) and the FinnGen study for sharing data. We also thank the authors of the blood metabolite GWAS as curated by the GWAS catalog (https://www.ebi.ac.uk/gwas/), and the authors of the NMR-GWAS (

References

  1. Khan FZ, Tritschler T, Kahn SR, Rodger MA. Venous thromboembolism. Lancet. 2021; 398(10294):64-77. Google Scholar
  2. Mukhopadhyay S, Johnson TA, Duru N. Fibrinolysis and inflammation in venous thrombus resolution. Front Immunol. 2019; 10:1348. Google Scholar
  3. Goldhaber SZ. Risk factors for venous thromboembolism. J Am Coll Cardiol. 2010; 56(1):1-7. Google Scholar
  4. Yuan S, Bruzelius M, Xiong Y, Håkansson N, Åkesson A, Larsson SC. Overall and abdominal obesity in relation to venous thromboembolism. J Thromb Haemost. 2021; 19(2):460-469. Google Scholar
  5. Fahy E, Cotter D, Sud M, Subramaniam S. Lipid classification, structures and tools. Biochim Biophys Acta. 2011; 1811(11):637-647. Google Scholar
  6. Deguchi H, Elias DJ, Griffin JH. Minor plasma lipids modulate clotting factor activities and may affect thrombosis risk. Res Pract Thromb Haemost. 2017; 1(1):93-102. Google Scholar
  7. Doggen CJM, Smith NL, Lemaitre RN, Heckbert SR, Rosendaal FR, Psaty BM. Serum lipid levels and the risk of venous thrombosis. Arterioscler Thromb Vas Biol. 2004; 24(10):1970-1975. Google Scholar
  8. Fraser K, Roy NC, Goumidi L. Plasma biomarkers and identification of resilient metabolic disruptions in patients with venous thromboembolism using a metabolic systems approach. Arterioscler Thromb Vasc Biol. 2020; 40(10):2527-2538. Google Scholar
  9. Jiang X, Zeleznik OA, Lindström S. Metabolites associated with the risk of incident venous thromboembolism: a metabolomic analysis. J Am Heart Assoc. 2018; 7(22):e010317. Google Scholar
  10. Sekula P, Del Greco MF, Pattaro C, Köttgen A. Mendelian randomization as an approach to assess causality using observational data. J Am Soc Nephrol. 2016; 27(11):3253-3265. Google Scholar
  11. Kastenmüller G, Raffler J, Gieger C, Suhre K. Genetics of human metabolism: an update. Hum Mol Genet. 2015; 24(R1):R93-R101. Google Scholar
  12. Lin L, Luo P, Yang MY, Wang JC, Hou WK, Xu P. A bidirectional Mendelian randomized study of classical blood lipids and venous thrombosis. Sci Rep. 2023; 13(1):3904. Google Scholar
  13. Liu ZY, Mi JR. Serum albumin and circulating metabolites and risk of venous thromboembolism: a two-sample Mendelian randomization study. Front Nutr. 2021; 8:712600. Google Scholar
  14. Luo P, Yuan QL, Wan XJ, Yang MY, Xu P. A two-sample Mendelian randomization study of circulating lipids and deep venous thrombosis. Sci Rep. 2023; 13(1):7432. Google Scholar
  15. Lotta LA, Pietzner M, Stewart ID. A cross-platform approach identifies genetic regulators of human metabolism and health. Nat Genet. 2021; 53(1):54-64. Google Scholar
  16. Kettunen J, Demirkan A, Würtz P. Genome-wide study for circulating metabolites identifies 62 loci and reveals novel systemic effects of. Nat Commun. 2016; 7:11122. Google Scholar
  17. Bowden J, Spiller W, Del Greco MF. Improving the visualization, interpretation and analysis of two-sample summary data Mendelian randomization via the Radial plot and Radial regression. Int J Epidemiol. 2018; 47(4):1264-1278. Google Scholar
  18. Hartwig FP, Davies NM, Hemani G, Davey Smith G. Two-sample Mendelian randomization: avoiding the downsides of a powerful, widely applicable but potentially fallible technique. Int J Epidemiol. 2016; 45(6):1717-1726. Google Scholar
  19. Guo J, Yu K, Dong SS. Mendelian randomization analyses support causal relationships between brain imaging-derived phenotypes and risk of psychiatric disorders. Nat Neurosci. 2022; 25(11):1519-1527. Google Scholar
  20. Yao S, Zhang M, Dong SS. Bidirectional two-sample Mendelian randomization analysis identifies causal associations between relative carbohydrate intake and depression. Nat Hum Behav. 2022; 6(11):1569-1576. Google Scholar
  21. Hemani G, Zhengn J, Elsworth B. The MR-Base platform supports systematic causal inference across the human phenome. Elife. 2018; 7:e34408. Google Scholar
  22. Bowden J, Smith GD, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol. 2015; 44(2):512-525. Google Scholar
  23. Egger M, Smith GD, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ. 1997; 315(7109):629-634. Google Scholar
  24. Rücker G, Schwarzer G, Carpenter JR, Binder H, Schumacher M. Treatment-effect estimates adjusted for small-study effects via a limit meta-analysis. Biostatistics. 2011; 12(1):122-142. Google Scholar
  25. Borges MC, Haycock P, Zheng J. The impact of fatty acids biosynthesis on the risk of cardiovascular diseases in Europeans and East Asians: a Mendelian randomization study. Hum Mol Genet. 2022; 31(23):4034-4054. Google Scholar
  26. Navratil AR, Shchepinov MS, Dennis EA. Lipidomics reveals dramatic physiological kinetic isotope effects during the enzymatic oxygenation of polyunsaturated fatty acids ex vivo. J Am Chem Soc. 2018; 140(1):235-243. Google Scholar
  27. Hiki M, Miyazaki T, Shimada K. Significance of serum polyunsaturated fatty acid level imbalance in patients with acute venous thromboembolism. J Atheroscler Thromb. 2017; 24(10):1016-1022. Google Scholar
  28. Yuan S, Bäck M, Bruzelius M, Mason AM, Burgess S, Larsson S. Plasma phospholipid fatty acids, and risk of 15 cardiovascular diseases: a Mendelian randomisation study. Nutrients. 2019; 11(12):3001. Google Scholar
  29. Bochkov VN, Oskolkova OV, Birukov KG, Levonen AL, Binder CJ, Stöckl J. Generation and biological activities of oxidized phospholipids. Antioxid Redox Signal. 2010; 12(8):1009-1059. Google Scholar
  30. Siess W. Platelet interaction with bioactive lipids formed by mild oxidation of low-density lipoprotein. Pathophysiol Haemost Thromb. 2006; 35(3-4):292-304. Google Scholar
  31. Koupenova M, Kehrel BE, Corkrey HA, Freedman JE. Thrombosis and platelets: an update. Eur Heart J. 2017; 38(11):785-791. Google Scholar
  32. Kim F, Tysseling KA, Rice J. Free fatty acid impairment of nitric oxide production in endothelial cells is mediated by IKKβ. Arterioscler Thromb Vas Biol. 2005; 25(5):989-994. Google Scholar
  33. Park WJ, Kothapalli KS, Lawrence P, Brenna JT. FADS2 function loss at the cancer hotspot 11q13 locus diverts lipid signaling precursor synthesis to unusual eicosanoid fatty acids. PLoS One. 2011; 6(11):e28186. Google Scholar
  34. Los DA, Murata N. Structure and expression of fatty acid desaturases. Biochim Biophys Acta. 1998; 1394(1):3-15. Google Scholar
  35. Sung YJ, Spagou K, Kafeza M. Deep vein thrombosis exhibits characteristic serum and vein wall metabolic phenotypes in the inferior vena cava ligation mouse model. Eur J Vasc Endovasc Surg. 2018; 55(5):703-713. Google Scholar
  36. Esterbauer H, Schaur RJ, Zollner H. Chemistry and biochemistry of 4-hydroxynonenal, malonaldehyde and related aldehydes. Free Radic Biol Med. 1991; 11(1):81-128. Google Scholar
  37. Obermayer G, Afonyushkin T, Binder CJ. Oxidized low-density lipoprotein in inflammation-driven thrombosis. J Thromb Haemost. 2018; 16(3):418-428. Google Scholar
  38. Murohara T, Scalia R, Lefer AM. Lysophosphatidylcholine promotes P-selectin expression in platelets and endothelial cells - possible involvement of protein kinase C activation and its inhibition by nitric oxide donors. Circ Res. 1996; 78(5):780-789. Google Scholar
  39. Wang ZN, Klipfell E, Bennett BJ. Gut flora metabolism of phosphatidylcholine promotes cardiovascular disease. Nature. 2011; 472(7341):57-63. Google Scholar
  40. Zhu WF, Gregory JC, Org E. Gut microbial metabolite TMAO enhances platelet hyperreactivity and thrombosis risk. Cell. 2016; 165(1):111-124. Google Scholar
  41. Lordan R, Vidal NP, Pham TH, Tsoupras A, Thomas RH, Zabetakis I. Yoghurt fermentation alters the composition and antiplatelet properties of milk polar lipids. Food Chem. 2020; 332:127384. Google Scholar
  42. Tsoupras A, Lordan R, Demuru M. Structural elucidation of Irish organic farmed Salmon - polar lipids with antithrombotic activities. Mar Drugs. 2018; 16(6):176. Google Scholar
  43. Lakshmy R, Dorairaj P, Tarik M, Gupta R, Reddy KS. LDL particle heterogeneity, and its association with other established cardiovascular risk factors in a young Indian industrial population. Heart Asia. 2012; 4(1):141-145. Google Scholar
  44. Pichler G, Amigo N, Tellez-Plaza M. LDL particle size and composition and incident cardiovascular disease in a South-European population: the Hortega-Liposcale follow-up study. Int J Cardiol. 2018; 264:172-178. Google Scholar
  45. Musunuru K, Orho-Melander M, Caulfield MP. Ion mobility analysis of lipoprotein subfractions identifies three independent axes of cardiovascular risk. Arterioscler Thromb Vas Biol. 2009; 29(11):1975-1980. Google Scholar
  46. Mora S, Szklo M, Otvos JD. LDL particle subclasses, LDL particle size, and carotid atherosclerosis in the Multi-Ethnic Study of Atherosclerosis (MESA). Atherosclerosis. 2007; 192(1):211-217. Google Scholar
  47. Swinkels DW, Hendriks JC, Demacker PN, Stalenhoef AF. Differences in metabolism of three low density lipoprotein subfractions in Hep G2 cells. Biochim Biophys Acta. 1990; 1047(3):212-22. Google Scholar
  48. Campos H, Arnold KS, Balestra ME, Innerarity TL, Krauss RM. Differences in receptor binding of LDL subfractions. Arterioscler Thromb Vas Biol. 1996; 16(6):794-801. Google Scholar
  49. Gao S, Zhao D, Wang M. Association between circulating oxidized low-density lipoprotein and atherosclerotic cardiovascular disease: a meta-analysis of prospective observational studies. Can J Cardiol. 2017; 33(12):1624-1632. Google Scholar
  50. Gu Y, Zang P, Li JX, Yan YY, Wang J. Plasma metabolomics in a deep vein thrombosis rat model based on ultra-high performance liquid chromatography-electrostatic field orbitrap high resolution mass spectrometry. Se Pu. 2022; 40(8):736-745. Google Scholar

Data Supplements

Figures & Tables

Article Information

Vol. 109 No. 9 (2024): September, 2024 : Articles
 
DOI
https://doi.org/10.3324/haematol.2023.284566
Pubmed
38695130
 
Published
2024-05-02
Published By
Ferrata Storti Foundation, Pavia, Italy
Print ISSN
0390-6078
Online ISSN
1592-8721
 
Copyright & Usage
Copyright (c) 2024 Ferrata Storti Foundation
Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Article Usage

Online Views
3302
PDF Downloads
1394

Statistics from Altmetric.com

No Data
Navigate
For Authors
For Reviewers
For Advertisers
Education
Privacy
More
Copyright © 2024 by the Ferrata Storti Foundation | Web design ￫ | Development ￫
ISSN 0390-6078 print | ISSN 1592-8721 online