Mendelian Randomization Provides No Evidence for the Bidirectional Relationship between Type 2 Diabetes and Venous Thromboembolism in East Asians and African Americans

We were very interested to read the recent paper by Hu et al, which presented a Mendelian randomization (MR) study examining the relationship between 13 cardiovascular risk factors and venous thromboembolism (VTE) in European individuals.[1] They found no evidence of an association between type 2 diabetes (T2D) and VTE. Recently, a bidirectional MR analysis showed similar results in individuals of European ancestry, indicating no association between T2D and VTE in either direction.[2] Nevertheless, their finding was limited to individuals of European ancestry, potentially restricting its applicability to other ethnic groups. Epidemiological studies in admixed American populations have suggested patients with type 1 diabetes or T2D may have an increased risk of VTE.[3] [4] However, there are few studies specifically investigating the relationship between T2D and VTE in African Americans. In East Asia, a large-scale prospective study among Chinese individuals indicated a higher risk of VTE in patients with T2D compared with the general population, but a Japanese study found no such association between diabetes and pulmonary thromboembolism.[5] [6] Given these gaps and inconsistencies, particularly regarding African Americans and East Asians, we were motivated to further explore the bidirectional association between T2D and VTE among these ethnicities. Thus, we employed a bidirectional two-sample MR method to investigate the presence and direction of the association between T2D and VTE in East Asians and African Americans.

Summary statistics of T2D were obtained from a multi-ancestry genome-wide association study (GWAS) from diabetes genetics replication and meta-analysis (DIAGRAM) consortium.[7] Specifically, the genetic associations of single nucleotide polymorphisms (SNPs) with T2D in individuals with African-American ancestry were derived from a meta-analysis of GWAS across 25 cohorts (50,251 T2D cases and 103,909 controls) and the summary statistics of T2D in East Asians were obtained from a meta-analysis of GWAS in 40 cohorts (88,109 T2D cases and 339,395 controls).[7] The genetic associations of SNPs with VTE in African Americans were derived from a GWAS of BioMe, BioVU, Michigan Genomics Initiative, University of California, Los Angeles (UCLA), and UK Biobank cohorts (1,466 VTE cases and 31,042 controls), while the summary statistics of VTE in East Asians were obtained from a GWAS of CKB and UCLA cohorts (193 VTE cases and 77,462 controls).[8] The definition of T2D and VTE in these cohorts was based on International Classification of Diseases codes or physician diagnoses.[7] [8] MR has three core assumptions: (1) the genetic variants were robustly associated with the exposure; (2) no confounders of the association between the genetic variants and the outcome; and (3) genetic variants affected the outcome only through the exposure.[9] To satisfy these assumptions, we selected commonly (minor allele frequency >1%), independently (linkage disequilibrium [LD] r ² < 0.001 within 10 Mb), and robustly (F-statistics >10) genome-wide significant (p < 5 × 10⁻⁸) SNPs as genetic instrumental variables to genetically proxy exposure. Since no SNPs reached genome-wide significance in GWASs of VTE for both East Asians and African Americans, we used a less stringent threshold of 5 × 10⁻⁶ to genetically proxy VTE. In cases where a SNP was absent in the outcome GWAS, we replaced it with a high LD proxy SNP (r ² > 0.8) and if a suitable proxy SNP was not available, the SNP was discarded.

To determine the causal direction of the relationship between T2D and VTE, we initially used the inverse-variance weighted (IVW) method as the primary method when more than two SNPs were available. If only one SNP was available, we only used the Wald ratio method to estimate causal effects, as described before.[10] Then, MR-Egger and weighted median methods were utilized as sensitivity methods. Finally, we used the MRlap method as another sensitivity method to correct for the potential bias of IVW-MR arising from overlapping samples between the GWAS of T2D and the GWAS of VTE.[11] To better interpret the results, we multiplied the causal estimate by 0.693 to reflect the increase in the risk of VTE associated with each doubling of the odds of genetic predisposition to T2D, and vice versa.[12] All analyses were two-sided and conducted using TwoSampleMR (version 0.5.10), and MRlap (version 0.0.3.0) packages in R software (version 4.2.2).

We selected a total of 45 and 193 SNPs to genetically proxy T2D for African Americans and East Asians, respectively. The IVW method suggested that genetic predisposition to T2D was not associated with the risk of VTE in East Asians (odds ratio [95% confidence interval]: 1.08 [0.88–1.33]; p = 0.463) and African Americans (1.10 [0.92–1.30]; p = 0.291; [Fig. 1A]). MR-Egger, weighted median, and MRlap methods also showed no evidence of the causal association between T2D and VTE across the two ethnic groups ([Fig. 1A]). To estimate the causal effect of VTE on T2D, we selected 1 and 27 SNPs to genetically proxy VTE for East Asians and African Americans, respectively. The Wald ratio method indicated no causal association between VTE and T2D in East Asians (0.99 [0.98–1.01]; p = 0.550; [Fig. 1B]). Similarly, the IVW method showed no association between genetic predisposition to VTE and the risk of T2D in African Americans (1.01 [0.99–1.02]; p = 0.128), and other sensitivity methods provided concordant results ([Fig. 1B]).

Fig. 1 Results of (A) MR estimation of T2D on VTE and (B) MR estimation of VTE on T2D in both East Asians and African Americans. [Fig. 1A] shows the estimation of causal effect of T2D on VTE in two ethnic groups, and [Fig. 1B] shows the estimation of causal effect of VTE on T2D in two ethnic groups. If there were fewer than two IVs, only the Wald ratio method was used. For more than two IVs, the IVW method was the primary method, with MR-Egger, weighted median, and MRlap methods used for sensitivity analysis. OR [95% CI] for VTE represented the odds ratio for VTE per doubling of the odds of genetic predisposition to T2D, and OR [95% CI] for T2D represented the odds ratio for T2D per doubling of the odds of genetic predisposition to VTE. CI, confidence interval; IVs, instrumental variables; IVW, inverse-variance weighted; MR, Mendelian randomization; OR, odds ratio; T2D, type 2 diabetes; VTE, venous thromboembolism.

Our MR study provides genetic evidence suggesting no evidence of causal associations between T2D and VTE in both directions among general individuals of East Asians and African Americans, which builds upon and strengthens the work of Hu et al by elucidating the generalizability of their earlier findings.[1] These results suggest that T2D may not be an ideal modifiable risk factor for VTE, and vice versa, within the populations studied. Although this study focused on T2D susceptibility rather than disease course, it remains crucial to identify and treat VTE in T2D patients, as those with both conditions have worse outcomes than those with only T2D. Overall, this study does not support an association between T2D genetic liability and VTE susceptibility among East Asians and African Americans, and vice versa.

Authors' Contributions

J.L. conceptualized the questions, implemented the analysis, and interpreted the issues and revised the manuscript. Z.W. curated the datasets, implemented the analysis, and wrote the original manuscript.

Data Availability Statement

Summary statistics of VTE of East Asian and African American were deposited on Global Biobank Meta-Analysis Initiative Web site (https://www.globalbiobankmeta.org/resources). Summary-level GWASs of T2D of East Asian and African American were deposited on the Web site (https://diagram-consortium.org/downloads.html). The data and R code used in this study are available from https://github.com/BarryJasmine/Bidirectional_T2D_VTE. Further information is available from the corresponding author upon request.

Publication History

Article published online:
19 July 2024

• References

• 1 Hu M, Li X, Yang Y. Causal associations between cardiovascular risk factors and venous thromboembolism. Semin Thromb Hemost 2023; 49 (07) 679-687
• 2 Hu S, Tan J-S, Hu M-J. et al. The causality between diabetes and venous thromboembolism: a bidirectional two-sample Mendelian randomization Study. Thromb Haemost 2023; 123 (09) 913-919
• 3 Piazza G, Goldhaber SZ, Kroll A, Goldberg RJ, Emery C, Spencer FA. Venous thromboembolism in patients with diabetes mellitus. Am J Med 2012; 125 (07) 709-716
• 4 Stein PD, Goldman J, Matta F, Yaekoub AY. Diabetes mellitus and risk of venous thromboembolism. Am J Med Sci 2009; 337 (04) 259-264
• 5 Chung W-S, Lin C-L, Kao C-H. Diabetes increases the risk of deep-vein thrombosis and pulmonary embolism. A population-based cohort study. Thromb Haemost 2015; 114 (04) 812-818
• 6 Sugimura K, Sakuma M, Shirato K. Potential risk factors and incidence of pulmonary thromboembolism in Japan: results from an overview of mailed questionnaires and matched case-control study. Circ J 2006; 70 (05) 542-547
• 7 Suzuki K, Hatzikotoulas K, Southam L. et al; VA Million Veteran Program. Genetic drivers of heterogeneity in type 2 diabetes pathophysiology. Nature 2024; 627 (8003): 347-357
• 8 Zhou W, Kanai M, Wu K-HH. et al; Biobank of the Americas, Biobank Japan Project, BioMe, BioVU, CanPath - Ontario Health Study, China Kadoorie Biobank Collaborative Group, Colorado Center for Personalized Medicine, deCODE Genetics, Estonian Biobank, FinnGen, Generation Scotland, Genes & Health Research Team, LifeLines, Mass General Brigham Biobank, Michigan Genomics Initiative, National Biobank of Korea, Penn Medicine BioBank, Qatar Biobank, QSkin Sun and Health Study, Taiwan Biobank, HUNT Study, UCLA ATLAS Community Health Initiative, Uganda Genome Resource, UK Biobank. Global Biobank Meta-analysis Initiative: powering genetic discovery across human disease. Cell Genomics 2022; 2 (10) 100192
• 9 Davey Smith G, Hemani G. Mendelian randomization: genetic anchors for causal inference in epidemiological studies. Hum Mol Genet 2014; 23 (R1): R89-R98
• 10 Wang Q, Shi Q, Wang Z, Lu J, Hou J. Integrating plasma proteomes with genome-wide association data for causal protein identification in multiple myeloma. BMC Med 2023; 21 (01) 377
• 11 Mounier N, Kutalik Z. Bias correction for inverse variance weighting Mendelian randomization. Genet Epidemiol 2023; 47 (04) 314-331
• 12 Burgess S, Labrecque JA. Mendelian randomization with a binary exposure variable: interpretation and presentation of causal estimates. Eur J Epidemiol 2018; 33 (10) 947-952