d i a b e t e s r e s e a r c h a n d c l i n i c a l p r a c t i c e 208 (2024) 111094

Contents lists available at ScienceDirect

Diabetes Research and Clinical Practice

journal homepage: www.journals.elsevier.com/diabetes-research-and-clinical-practice

Cardiovascular diseases consequences of type 1, type 2 diabetes mellitus

and glycemic traits: A Mendelian randomization study
Pan Zhang a, 1, Zihang Zhang b, 1, Jinghui Zhong a, Xueying Zheng c, Junling Zhou d, *, Wen Sun a, *
a
Department of Neurology, Centre for Leading Medicine and Advanced Technologies of IHM, The First Affiliated Hospital of USTC, Division of Life Sciences and
Medicine, University of Science and Technology of China, Hefei, Anhui 230001, China
b
Department of Cardiovascular Surgery ICU, The First Affiliated Hospital of Anhui Medical University, Hefei, Anhui 230001, China
c
Department of Endocrinology, Institute of Endocrine and Metabolic Diseases, The First Affiliated Hospital of USTC, Division of Life Sciences and Medicine, Clinical
Research Hospital of Chinese Academy of Sciences (Hefei), University of Science and Technology of China, Hefei, China
d
Department of Cardiology, The First Affiliated Hospital of USTC, Division of Life Science and Medicine, University of Science and Technology of China, Hefei, Anhui,
China

A B S T R A C T

Objective: This Mendelian randomization (MR) study aimed to investigate the relationships between type 1 diabetes (T1D), type 2 diabetes (T2D), and glycemic traits,
including fasting insulin, fasting glucose, and HbA1c, with cardiovascular diseases (CVDs).
Methods: We selected genetic instruments for predisposition to T1D, T2D, fasting insulin, fasting glucose, and HbA1c based on published genome-wide association
studies. Using a 2-Sample MR approach, we assessed associations with 12 common CVDs sourced from the FinnGen and UK Biobank studies, along with stroke
subtypes obtained from the GIGASTROKE and MEGASTROKE Consortium.
Results: T1D was associated with SVS. T2D showed associations with AIS, LAA, CES, SVS, coronary heart disease, myocardial infarction, pulmonary embolism, DVT of
lower extremities, peripheral vascular diseases. Genetically predicted higher HbA1c levels were associated with eight CVDs. The results of MVMR aligned with the
primary findings for T1D and T2D.
Conclusions: T1D and T2D exhibit different genetic predisposition to CVDs. BMI, LDL, and HDL play intermediary roles in connecting TID and T2D to specific types of
CVDs, providing insights into the potential underlying pathways and mechanisms involved in these relationships. Strategies aimed at achieving sustained reductions
in HbA1c levels may offer potential for reducing the risk of various CVDs.

1. Introduction secretion, has consistently been associated with a higher risk of CVDs,
including CAD, MI, stroke, and heart failure [10–14]. The mechanisms
Epidemiological studies indicate that cardiovascular diseases (CVDs) underlying the increased risk of CVDs in T2D are complex and involve a
are highly prevalent among individuals over the age of 50, making them combination of traditional cardiovascular risk factors, such as dyslipi­
the leading cause of mortality [1,2]. Diabetes mellitus (DM) is a chronic demia, hypertension, obesity, and a pro-inflammatory state [15,16].
condition that can seriously harm a variety of target organs and body However, due to uncontrolled confounding factors and the potential bias
organs, which include the brain vasculature [3]. of reverse causation in observational studies, the causal links between
DM is generally divided into four categories, while type 1 diabetes DM and CVDs, as well as whether there are other factors mediating these
(T1D) and type 2 diabetes (T2D) are two major types but pathophysio­ relationships, are still uncertain.
logically different diseases [4]. T1D, an autoimmune disease charac­ Mendelian randomization (MR) analysis is an epidemiological
terized by the destruction of pancreatic beta cells, has been linked to an approach that enhances causal inference by employing genetic variants
increased risk of coronary artery disease (CAD), myocardial infarction as instrumental variables for an exposure [17]. MR analysis can reduce
(MI), stroke, and peripheral arterial disease (PAD) [5–9]. The underly­ confounding bias as genetic alleles are randomly distributed at
ing mechanisms linking T1D to these CVDs are multifactorial. Similarly, conception, eliminating correlations with environmental and self-
T2D, which is characterized by insulin resistance and impaired insulin selected factors. The MR methodology additionally guards against

* Corresponding authors at: Department of Cardiology, The First Affiliated Hospital of USTC, Division of Life Science and Medicine, University of Science and
Technology of China, Hefei, Anhui 230001, China (Junling Zhou). Department of Neurology, Centre for Leading Medicine and Advanced Technologies of IHM, The
First Affiliated Hospital of USTC, Division of Life Sciences and Medicine, University of Science and Technology of China, Hefei, Anhui 230001, China (Wen Sun).
E-mail addresses: [email protected] (J. Zhou), [email protected] (W. Sun).
1
These authors have contributed equally to this work.

https://doi.org/10.1016/j.diabres.2024.111094
Received 12 November 2023; Received in revised form 29 December 2023; Accepted 9 January 2024
Available online 13 January 2024
0168-8227/© 2024 Elsevier B.V. All rights reserved.
P. Zhang et al. Diabetes Research and Clinical Practice 208 (2024) 111094

reverse causality, as the germline genotype remains impervious to of exposure were at the genome-wide significance level (P < 5 × 10-8).
modification by the disease [18]. We evaluated linkage disequilibrium among SNPs for each exposure
Therefore, this study aims to fill this gap by investigating the genetic using the 1000 Genomes European panel and applied the PLINK
associations of both T1D and T2D with 12 common CVDs and to seek the clumping method, excluding SNPs with linkage disequilibrium (defined
genetic susceptibility differences between T1D and T2D in relation to as R2 > 0.001 and a clump distance of < 10,000 kb). Conditional F-
various CVDs. Furthermore, we evaluated the causal relationships be­ statistics were used to assess instrument strength, F > 10 indicated the
tween DM and CVDs, taking into account potiential factors adjustments absence of weak instrument bias.
using a multivariable Mendelian randomization analysis (MVMR) and Comprehensive information regarding the SNPs employed in this
carried out a mediation analysis to explore whether these traits medi­ study can be found in Supplementary Tables 2. Following a rigorous
ated the causal effects of DT1D, T2D on CVDs. We also explored the screening process, we retained 89 SNPs associated with T1D, 187 SNPs
connections between genetically predicted glycemic traits (fasting in­ related to T2D, 38 SNPs for fasting insulin, 71 SNPs for fasting glucose,
sulin, fasting glucose, and HbA1c) and CVDs. and 404 SNPs for HbA1c (Table S2).

2. Methods 2.3. Associations with outcomes

2.1. Study design and data sources This study encompassed 12 CVDs, specifically any ischemic stroke
(AIS), its subtypes, atrial fibrillation and flutter (AF), coronary heart
This two-sample MR study was structured to investigate the genetic disease (CHD), myocardial infarction (MI), intracerebral hemorrhage
predisposition’s impact on T1D, T2D, and glycemic traits concerning the (ICH), subarachnoid hemorrhage (SAH), heart failure (HF), aortic
risk of 12 CVDs. The study design was shown in Fig. 1. We utilized aneurysm (AA), lower extremity deep vein thrombosis (DVT), pulmo­
publicly accessible summary statistics obtained from GWAS consortia. nary embolism (PB), peripheral vascular diseases (PVD), and hyper­
For a comprehensive list of data sources, please refer to Table S1. tension (HP). We utilized summary data from the MEGASTROKE [23]
and GIGASTROKE [24] studies, which exclusively comprised European
2.2. Genetic instrument selection individuals, to extract summary statistics for subtypes of ischemic
stroke: large artery stroke (LAA), cardioembolic stroke (CES), and small
We selected single-nucleotide polymorphisms (SNPs) associated with vessel stroke (SVS). Summary-level genetic associations for other out­
comes were obtained from the FinnGen R9 consortium and the UK
T1D from a meta-analysis of a GWAS dataset comprising 520,580 Eu­
ropean individuals [19]. SNPs related to T2D were obtained from the Biobank [25] study. All the outcomes were derived from individuals of
European ancestry.
Diabetes Genetics Replication and Meta-analysis (DIAGRAM) con­
sortium [20], and data on glycemic traits, including fasting insulin and
glucose, were sourced from the Meta-Analyses of Glucose and Insulin 2.4. Statistical analysis
Related Traits Consortium, which encompassed up to 196,991 partici­
pants of European ancestry without diabetes [21]. Hemoglobin A1c A two-sample Mendelian randomization approach was employed to
(HbA1c) data were extracted from the UK Biobank study [22]. All SNPs analyze association estimates derived from the aforementioned sources.

Fig. 1.

2
P. Zhang et al. Diabetes Research and Clinical Practice 208 (2024) 111094

The primary MR method used was the random-effects multiplicative in Fig. 2. Genetic predisposition to T1D showed an association with an
inverse variance-weighted (IVW) method. To assess the robustness of elevated risk of AIS, SVS, and MI in the comprehensive meta-analysis of
the results and identify potential horizontal pleiotropy, several sensi­ all outcome sources. The association with SVS remained statistically
tivity analyses were conducted, including the weighted median, MR- significant even after multiple testing correction (Table S3). For every 1-
Egger, and MR Pleiotropy Residual Sum and Outlier (MR-PRESSO) unit increase in log-transformed odds of T1D, the combined OR was
methods. The weighted median method can produce valid MR estimates 1.0221 (95 % CI, 1.0012–1.0434, p = 0.038, padjusted = 0.192) for AIS,
when more than 50 % of the weight is attributed to valid SNPs. MR- 1.0824 (95 % CI, 1.0306–1.1367, p = 0.002, padjusted = 0.023) for SVS,
Egger provides less precise estimates and relies on the assumption that and 1.0027 (95 % CI, 1.0007–1.0047, p = 0.008, padjusted = 0.06) for MI.
the strengths of potential pleiotropic instruments are independent of (Fig. 3 and Table S3) The results concerning the genetic predisposition
their direct associations with the outcome. The MR-PRESSO method is to T1D and its association with the risk of CVDs remained directionally
capable of detecting and correcting for potential outliers, and the MR- consistent in sensitivity analyses (Table S4). In MR-Egger intercept
PRESSO global test can be employed to evaluate horizontal pleiotropy analysis, no horizontal pleiotropy was evident, except for CHD, MI, and
arising from heterogeneity among SNP estimates. HP in the FinnGen dataset and CHD in the UK Biobank dataset
To assess the heterogeneity of SNP estimates in each Mendelian (Table S4). MR-PRESSO analysis identified 1 to 5 outliers; however, the
randomization association, we applied both the Cochran Q test. For associations remained stable even after the removal of these outliers
controlling multiple testing across the 12 CVDs, we employed the false (Table S4).
discovery rate correction based on the Benjamini-Hochberg method. The
procedure involves ordering the p-values from smallest to largest, and 3.2. T2D and CVDs
then comparing each p-value to a critical value that depends on the
desired FDR control level. The critical value is chosen such that the After multiple testing correction, a one-unit increase in log-
proportion of false rejections, or the FDR, is below the desired level. The transformed odds of T2D was associated with a significantly higher
adjusted p-value is then calculated based on the rank of the observed p- risk of AIS (OR 1.092, 95 %CI 1.066–1.118, p < 0.001, padjusted < 0.001),
value and the critical value. An association with a nominal P value below LAA (OR 1.203, 95 %CI 1.115–1.297, p < 0.001, padjusted < 0.001), CES
0.05 but a Benjamini-Hochberg adjusted P value exceeding 0.05 was (OR 1.076, 95 %CI 1.018–1.138, p = 0.01, padjusted = 0.022), SVS (OR
considered suggestive, while an association with a Benjamini-Hochberg 1.167, 95 %CI 1.102–1.236, p < 0.001, padjusted < 0.001), CHD (OR
adjusted P value less than 0.05 was regarded as significant. Additionally, 1.021, 95 %CI 1.003–1.039, p = 0.022, padjusted = 0.036), MI (OR 1.165,
we conducted a MVMR analysis to explore the associations between 95 %CI 1.106–1.228, p < 0.001, padjusted < 0.001), PB (OR 1.013, 95 %
T1D, T2D, and CVDs while adjusting for genetically predicted BMI, CI 1.007–1.019, p < 0.001, padjusted < 0.001), DVT (OR 1.011, 95 %CI
hypertention, smoking, C-reactive protein (CRP), high-density lipopro­ 1.005–1.018, p = 0.001, padjusted = 0.002), PVD (OR 1.003, 95 %CI
tein (HDL), low-density lipoprotein (LDL), triglyceride (TG) and total 1.001–1.006, p = 0.013, padjusted = 0.025). (Fig. 4 and Table S3）The
cholesterol (TC). Furthermore, we conducted a mediation analysis to results for genetic liability to T2D and its association with the risk of
examine whether theses traits mediated the causal effects of T1D and CVDs were directionally consistent in sensitivity analyses (Table S5).
T2D on CVDs.We additionally conducted sensitivity analyses: utilized MR-Egger analysis did not reveal horizontal pleiotropy except for hy­
T1D data from a meta-analysis of a GWAS dataset consisting of 24,840 pertension in the UK BiobankB data (Table S5). MR-PRESSO analysis
European individuals [26] and T2D data from a meta-analysis of a GWAS identified 1 to 13 outliers; however, the associations remained stable
dataset comprising 490,089 European individuals [27] to replicate our after the removal of these outliers (Table S5).
findings. All analyses were two-sided and performed using R packages
TwoSampleMR [28] in R 4.1.3. 3.3. MVMR and mediation MR analysis of T1D and T2D

3. Results The MVMR analysis was conducted to assess the direct effect of TIDM
and T2D on CVDs, while adjusting for multiple other risk factors for
3.1. T1D and CVDs CVD. After adjusting for BMI, smoking, HDL, LDL, triglycerides, and
total cholesterol, T1D remained associated with SVS (Table S6).
An overview of the main results of the primary analyses was shown Furthermore, we performed a mediation MR analysis to examine

Fig. 2.

3
P. Zhang et al. Diabetes Research and Clinical Practice 208 (2024) 111094

Fig. 3.

mediation effects. The results showed that BMI mediated the relation­ 3.5. Glycemic traits and CVDs
ship between T1D and SVS, with a proportion of mediation estimated at
− 0.3 % (95 % CI: − 0.65 % to − 0.06 %) (Table S7). The results of the Genetically predicted levels of fasting insulin, fasting glucose, and
MVMR analysis for the effect of T2D on CVDs are presented in Table S8, HbA1c were associated with risk of 0, 2, and 8, respectively, of the 12
which were generally consistent with the main analysis. Additionally, CVDs outcomes (Table S3 and 10). After multiple testing correction,
we found that LDL mediated the relationship between T2D and LAA there were suggestive associations between genetically predicted 1
(-1.23 %, 95 % CI: − 2.9 % to − 0.02 %), while HDL mediated the rela­ mmol/L increase in fasting glucose levels and increased risk of AIS (OR
tionship between T2D and SVS (0.76 %, 95 % CI: 0.08 % to 1.73 %) as 1.177, 95 %CI 1.011–1.371, p = 0.036, padjusted = 0.285), LAA (OR
well as DVT (-0.041 %, 95 % CI: − 0.095 % to − 0.003 %).(Table S9). 1.876, 95 %CI 1.232–2.858, p = 0.003, padjusted = 0.054) and significant
associations between genetically predicted HBA1c and increased risk of
3.4. Sensitivity analysis AIS (OR 1.105, 95 %CI 1.059–1.154, p < 0.001, padjusted < 0.001), LAA
(OR 1.387, 95 %CI 1.203–1.599, p < 0.001, padjusted < 0.001), CES (OR
The replication of results for T1D and T2D is presented in Table S14. 1.145, 95 %CI 1.037–1.263, p = 0.007, padjusted = 0.015), SVS (OR
The results are generally consistent with the primary findings. 1.175, 95 %CI 1.056–1.308, p = 0.003, padjusted = 0.009), CHD (OR
1.023, 95 %CI 1.009–1.041, p = 0.003, padjusted = 0.009), MI (OR 1.053,

4
P. Zhang et al. Diabetes Research and Clinical Practice 208 (2024) 111094

Fig. 4.

95 %CI 1.014–1.093, p = 0.007, padjusted = 0.015), PB (OR 1.012, 95 % of CVDs.

CI 1.006–1.018, p < 0.001, padjusted = 0.001), PVD (OR 1.004, 95 %CI Our investigation revealed a significant association between genetic
1.001–1.006, p = 0.01, padjusted = 0.018) (Table S10). The associations liability to T1D and an elevated risk of SVS, with a suggestive association
remained stable in sensitivity analyses and no indication of heteroge­ observed for AIS and MI. Moreover, we observed that T2D was associ­
neity and horizontal pleiotropy was detected in most outcomes ated with an elevated risk of various ischemic stroke subtypes, including
(Table S11-13). LAA, CES, and SVS. After adjusting for potiential factors, the associa­
tions remained significant. Additionally, we identified several novel
4. Discussion associations between genetic susceptibility to T2D and an increased risk
of CHD, MI, PB, DVT, and PVD. We also found suggestive associations
Our Mendelian randomization results offer genetic evidence sup­ between predicted levels of fasting glucose and AIS, particularly LAA, as
porting a causal relationship between T1D, T2D, and glycemic traits and well as significant associations between predicted levels of HbA1c and
the risk of CVDs. AIS, LAA, CES, SVS, CHD, MI, PB and PVD. Moreover, we found that BMI
This study has revealed two interesting findings: (1) T1D and T2D played a mediating role in the relationship between T1D and SVS, LDL
exhibit different genetic predisposition to CVDs, and (2) BMI, LDL, and mediated the relationship between T2D and LAA, HDL was identified as
HDL play intermediary roles in connecting TID and T2D to specific types a mediator between T2D and both SVS and DVT.

5
P. Zhang et al. Diabetes Research and Clinical Practice 208 (2024) 111094

Focused management of CVDs risk factors is the most effective relationship between T1D, BMI, and SVS. Further research is needed to
strategy for preventing and minimizing CVDs occurrences [29]. In our elucidate the precise mechanisms and pathways involved in this medi­
study, we observed that T1D is associated with an increased risk of SVS ation effect.
and showed a suggestive connection with MI. However, T2D was asso­ Furthermore, we found that LDL mediated the relationship between
ciated with various CVDs. Interestingly, it reflects the disparate genetic T2D and LAA, HDL was identified as a mediator between T2D and both
susceptibility of T1D and T2D in relation to CVDs. This phenomenon SVS and DVT.
may be attributed to several factors: (1) T1D is a chronic condition The identification of LDL as a mediator between T2D and LAA sug­
stemming from the autoimmune destruction of pancreatic β cells [5]. It gests that LDL cholesterol plays a role in connecting T2D to the devel­
is primarily regarded as a disease closely linked to genetics, with he­ opment of LAA. This finding implies several potential mechanisms: (1)
redity accounting for approximately 50 % of the susceptibility to T1D LDL cholesterol is a well-established risk factor for atherosclerosis, a
[30]. In contrast, T2D is characterized by a relative insufficiency of in­ condition characterized by the buildup of plaque within the arteries
sulin due to both pancreatic β-cell dysfunction and insulin resistance in [45]. T2D is known to be associated with dyslipidemia, including
target organs [31]. It is conceivable that these differing mechanisms are elevated levels of LDL cholesterol. Therefore, the mediation effect of LDL
responsible for their distinct impacts on CVDs. (2) The immune re­ suggests that the increased LDL cholesterol levels in individuals with
sponses in T1D are complex and encompass the coordinated interactions T2D may contribute to the development of LAA through the promotion
of various cell types. Apart from B cells and T cells, there is evidence of atherosclerotic plaque formation in the large arteries. (2) Elevated
suggesting potential roles for neutrophils, natural killer cells, macro­ LDL cholesterol levels can lead to endothelial dysfunction, impairing the
phages, and dendritic cells in the context of T1D [32–34]. (3) Glycemic normal function of the blood vessel lining. Endothelial dysfunction
variability could potentially serve as a contributing factor. It has been contributes to the initiation and progression of atherosclerosis and may
observed that individuals with T1D experience greater fluctuations in be involved in the development of LAA. T2D, along with associated
blood sugar levels compared to those with T2D. Certain tomographic metabolic abnormalities, can further exacerbate endothelial dysfunc­
data suggest a plausible association between glycemic variability and tion. Hence, LDL mediation may reflect the impact of T2D-induced
factors like plaque instability and subclinical coronary atherosclerosis endothelial dysfunction on LAA development [46].
[35]. (4) T2D is often accompanied by obesity and insulin resistance, Regarding HDL as a mediator between T2D and both SVS and DVT
both of which further trigger systemic inflammatory responses, dysli­ suggests a different set of potential mechanisms: (1) HDL cholesterol is
pidemia, hypertension, endothelial dysfunction, oxidative stress, and known for its protective effects on the cardiovascular system, including
alterations in cardiac metabolism [36]. These factors play a pivotal role its role in maintaining the health and function of small blood vessels
in the occurrence of CVDs. Even after adjusting for potiential risk fac­ [45,47]. T2D is associated with impaired microvascular function and the
tors, T2D remains associated with multiple CVDs, indicating that the risk development of SVS. HDL cholesterol levels in individuals with T2D may
of CVDs in individuals with T2D is influenced by a combination of contribute to microvascular dysfunction and the development of SVS.
various mechanisms. In conclusion, the mechanisms underlying the (2) T2D is characterized by a prothrombotic state, increasing the risk of
differences in genetic susceptibility to cardiovascular diseases between blood clot formation. HDL cholesterol has antithrombotic properties,
T1D and T2D require further investigation. helping to maintain a balance between prothrombotic and antith­
In this study, we found that T1D was associated with SVS, and BMI rombotic factors [47]. The levels of HDL cholesterol in individuals with
played a mediating role in the relationship between T1D and SVS. While T2D may contribute to an imbalance in the coagulation system, leading
the specific mechanisms underlying this mediation are complex and can to an increased risk of DVT.
vary, there are several potential explanations: (1) Both T1D and obesity Moreover, we also identified an increased risk of PB and PVD in in­
(reflected by higher BMI) are associated with insulin resistance. Insulin dividuals with T2D. Patients with diabetes may be prone to these con­
resistance can lead to endothelial dysfunction, inflammation, and ditions due to abnormalities in blood coagulation, heightened viscosity,
oxidative stress, which are implicated in the development of vascular and systemic inflammatory responses [48]. These factors can lead to
diseases, including SVS [36,37]. Thus, BMI may mediate the T1D-SVS damage to microvascular endothelial cells and thickening of the base­
relationship through its impact on insulin resistance and related physi­ ment membrane, resulting in the narrowing and blockage of blood vessel
ological processes. (2) Obesity is characterized by chronic low-grade lumens. Pulmonary embolism, on the other hand, often occurs when
inflammation. Adipose tissue releases various pro-inflammatory mole­ thrombi dislodge from the leg or pelvic veins, causing a cessation of
cules, such as cytokines and adipokines, which can promote vascular blood flow to the lungs.
inflammation and endothelial dysfunction [38]. Inflammation is a key In our study, we observed that HbA1C was associated with various
contributor to the development of atherosclerosis and other vascular CVDs, which aligns with the results related to T2D. This finding suggests
diseases [39]. Therefore, the inflammatory pathways activated by that preventive strategies targeting long-term HbA1c reduction can
higher BMI may mediate the association between T1D and SVS. (3) potentially reduce the risk of most CVDs.
Obesity, particularly excess visceral adiposity, is often associated with Furthermore, our study revealed a potential connection between
dyslipidemia, characterized by elevated levels of LDL cholesterol and fasting glucose levels and ischemic stroke, particularly the LAA subtype.
triglycerides, and reduced levels of HDL cholesterol [40]. Dyslipidemia This association may be linked to the oxidative stress induced by
is a well-established risk factor for atherosclerosis and cardiovascular elevated blood sugar levels, which could contribute to the formation of
diseases [40]. (4) Increased BMI is associated with alterations in he­ atherosclerotic plaques [35]. However, the precise underlying mecha­
modynamics, such as higher blood pressure and increased cardiac nisms involved in this relationship warrant further investigation.
output [41]. These hemodynamic changes can contribute to endothelial This study has several strengths. First of all, the MR study itself,
dysfunction, vascular remodeling, and increased shear stress, which are which effectively alleviated concerns related to confounding factors and
implicated in the development of SVS [42]. (5) Obesity and T1D are both reverse causality bias. Second, the use of summary-level data from large-
metabolic disorders that can influence each other’s pathophysiology. scale genetic studies within European populations further bolsters the
Higher BMI in individuals with T1D may worsen metabolic control, such credibility of our findings, substantially minimizing the potential for
as glycemic control and lipid metabolism, which can further contribute bias stemming from population structure. Third, the large sample size
to the development of SVS [43]. Poor metabolic control can lead to and the inherent characteristics of our datasets empowered us to
hyperglycemia, dyslipidemia, and other metabolic abnormalities, all of perform multiple sensitivity analyses, allowing us to rigorously test the
which are linked to vascular complications [44]. validity of the Mendelian randomization assumptions and thereby
It’s important to note that these potential mechanisms are not reducing the likelihood of obtaining biased results.
mutually exclusive, and multiple factors likely interact to mediate the Our study also has limitations. Firstly, the participants involved in

6
P. Zhang et al. Diabetes Research and Clinical Practice 208 (2024) 111094

the GWASs were exclusively of European ancestry, which limits the References
applicability of our findings to other population groups. Secondly, due
to the nature of the data we utilized, which was at a summary level, we [1] Zhu Y, Liu C, Zhang L, Fang Q, Zang S, Wang X. How to control the economic
burden of treating cardio-cerebrovascular diseases in China? Assessment based on
were unable to explore potential effects related to factors such as sex, system of health accounts 2011. J Glob Health 2020 Jun;10(1):010802.
age, or specific types of exposure on the outcomes. Thirdly, we cannot [2] Dalen JE, Alpert JS, Goldberg RJ, Weinstein RS. The epidemic of the 20(th)
completely rule out the possibility that SNPs associated with exposure to century: coronary heart disease. Am J Med 2014;127(9):807–12.
[3] Woldeteklie AA, Kebede HB, Abdela AA, Woldeamanuel Y. Prevalence of extended-
the disease may influence the outcome through alternative causal Spectrum β-lactamase and carbapenemase producers of gram-negative bacteria,
pathways, but we have adjusted for common confounding factors. and methicillin-resistant Staphylococcus aureus in isolates from diabetic foot ulcer
In conclusion, our study revealed the differences in genetic suscep­ patients in Ethiopia. Infect Drug Resist 2022;15:4435–41.
[4] ElSayed NA, Aleppo G, Aroda VR, et al. 2. classification and diagnosis of diabetes:
tibility to CVDs between T1D and T2D. Additionally, glycemic regula­ Standards of Care in Diabetes-2023. Diabetes Care 2023;46(Suppl 1):S19–40.
tion characterized by elevated glucose levels and HbA1c, are associated [5] Quattrin T, Mastrandrea LD, Walker LSK. Type 1 diabetes. Lancet 2023;401
with an increased risk of several CVDs. These findings underscore the (10394):2149–62.
[6] Keshawarz A, Pyle L, Alman A, et al. Type 1 diabetes accelerates progression of
importance of early screening and preventive measures for CVDs in in­
coronary artery calcium over the menopausal transition: The CACTI study.
dividuals with diabetes. Furthermore, interventions aimed at long-term Diabetes Care 2019;42(12):2315–21.
reductions in HbA1c levels may hold promise for lowering the risk of [7] de Ferranti SD, de Boer IH, Fonseca V, et al. Type 1 diabetes mellitus and
numerous CVDs. Factors such as BMI, LDL, and HDL play intermediary cardiovascular disease: a scientific statement from the American Heart Association
and American Diabetes Association. Diabetes Care 2014;37(10):2843–63.
roles in connecting TIDM and T2D to specific types of CVDs, providing [8] Harjutsalo V, Pongrac Barlovic D, Groop PH. Long-term population-based trends in
insights into the potential underlying pathways and mechanisms the incidence of cardiovascular disease in individuals with type 1 diabetes from
involved in these relationships. Finland: a retrospective, nationwide, cohort study. Lancet Diabetes Endocrinol
2021;9(9):575–85.
[9] Lee YB, Han K, Kim B, et al. Risk of early mortality and cardiovascular disease in
Source of funding type 1 diabetes: a comparison with type 2 diabetes, a nationwide study. Cardiovasc
Diabetol 2019;18(1):157.
[10] Elliott MD, Heitner JF, Kim H, et al. Prevalence and prognosis of unrecognized
The study was supported by grants from National Natural Science myocardial infarction in asymptomatic patients with diabetes: a two-center study
Foundation of China (Grant No. 82100822), Natural Science Foundation with up to 5 years of follow-up. Diabetes Care 2019;42(7):1290–6.
[11] Rezende PC, Hlatky MA, Hueb W, et al. Association of Longitudinal Values of
of Anhui Province (Grant No. 2108085MH271) and Leading Medicine
glycated hemoglobin with cardiovascular events in patients with type 2 diabetes
and Advanced Technologies of IHM (Grant No. 2023IHM01054). and multivessel coronary artery disease. JAMA Netw Open 2020;3(1):e1919666.
[12] Polovina M, Lund LH, Đikić D, et al. Type 2 diabetes increases the long-term risk of
heart failure and mortality in patients with atrial fibrillation. Eur J Heart Fail 2020;
Disclosure 22(1):113–25.
[13] Bonaca MP, Gutierrez JA, Cannon C, et al. Polyvascular disease, type 2 diabetes,
None. and long-term vascular risk: a secondary analysis of the IMPROVE-IT trial. Lancet
Diabetes Endocrinol 2018;6(12):934–43.
[14] Moran GM, Bakhai C, Song SH, Agwu JC. Guideline committee. Type 2 diabetes:
CRediT authorship contribution statement summary of updated NICE guidance. BMJ 2022;377:o775.
[15] Collaboration ERF, Sarwar N, Gao P, et al. Diabetes mellitus, fasting blood glucose
concentration, and risk of vascular disease: a collaborative meta-analysis of 102
Pan Zhang: Conceptualization, Data curation, Formal analysis, prospective studies. Lancet 2010;375(9733):2215–22.
Investigation, Methodology, Validation, Visualization, Writing – orig­ [16] Joseph JJ, Donner TW. Long-term insulin glargine therapy in type 2 diabetes
mellitus: a focus on cardiovascular outcomes. Vasc Health Risk Manag 2015;11:
inal draft, Writing – review & editing. Zihang Zhang: Conceptualiza­ 107–16.
tion, Data curation, Formal analysis, Visualization, Writing – original [17] Davies NM, Holmes MV, Davey SG. Reading mendelian randomisation studies: a
draft, Writing – review & editing. Jinghui Zhong: Conceptualization, guide, glossary, and checklist for clinicians. BMJ 2018;362:k601.
[18] Pingault JB, O’Reilly PF, Schoeler T, Ploubidis GB, Rijsdijk F, Dudbridge F. Using
Data curation, Supervision, Validation, Visualization, Writing – review
genetic data to strengthen causal inference in observational research. Nat Rev
& editing. Xueying Zheng: Conceptualization, Data curation, Formal Genet 2018;19(9):566–80.
analysis, Funding acquisition, Supervision, Writing – original draft, [19] Chiou J, Geusz RJ, Okino ML, et al. Interpreting type 1 diabetes risk with genetics
Writing – review & editing. Junling Zhou: Conceptualization, Data and single-cell epigenomics. Nature 2021;594(7863):398–402.
[20] Mahajan A, Spracklen CN, Zhang W, et al. Multi-ancestry genetic study of type 2
curation, Investigation, Methodology, Supervision, Visualization, diabetes highlights the power of diverse populations for discovery and translation.
Writing – review & editing. Wen Sun: Conceptualization, Data curation, Nat Genet 2022;54(5):560–72.
Formal analysis, Funding acquisition, Investigation, Project adminis­ [21] Chen J, Spracklen CN, Marenne G, et al. The trans-ancestral genomic architecture
of glycemic traits. Nat Genet 2021;53(6):840–60.
tration, Resources, Software, Validation, Visualization, Writing – orig­ [22] Mbatchou J, Barnard L, Backman J, et al. Computationally efficient whole-genome
inal draft, Writing – review & editing. regression for quantitative and binary traits. Nat Genet 2021;53(7):1097–103.
[23] Malik R, Chauhan G, Traylor M, et al. Multiancestry genome-wide association
study of 520,000 subjects identifies 32 loci associated with stroke and stroke
subtypes. Nat Genet 2018;50(4):524–37.
Declaration of Competing Interest [24] Mishra A, Malik R, Hachiya T, et al. Stroke genetics informs drug discovery and risk
prediction across ancestries. Nature 2022;611(7934):115–23.
[25] Sudlow C, Gallacher J, Allen N, et al. UK biobank: an open access resource for
The authors declare that they have no known competing financial
identifying the causes of a wide range of complex diseases of middle and old age.
interests or personal relationships that could have appeared to influence PLoS Med 2015;12(3):e1001779.
the work reported in this paper. [26] Forgetta V, Manousaki D, Istomine R, et al. Rare genetic variants of large effect
influence risk of type 1 diabetes. Diabetes 2020;69(4):784–95.
[27] Sakaue S, Kanai M, Tanigawa Y, et al. A cross-population atlas of genetic
Acknowledgments associations for 220 human phenotypes. Nat Genet 2021;53(10):1415–24.
[28] Hemani G, Zheng J, Elsworth B, et al. The MR-base platform supports systematic
causal inference across the human phenome. Elife 2018;7:e34408.
The authors thank the participants and staff of the UK Biobank for [29] Meng L, Xu J, Li J, et al. Self-reported prevalence and potential factors influencing
their valuable contributions, as well as the participants and investigators cardio-cerebral vascular disease among the Chinese elderly: a national cross-
of FinnGen study. sectional study. Front Cardiovasc Med 2022;9:979015.
[30] Redondo MJ, Evans-Molina C, Steck AK, Atkinson MA, Sosenko J. The influence of
type 2 diabetes-associated factors on type 1 diabetes. Diabetes Care 2019;42(8):
Appendix A. Supplementary data 1357–64.
[31] Chatterjee S, Khunti K, Davies MJ. Type 2 diabetes. Lancet 2017;389(10085):
2239–51.
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.diabres.2024.111094.

7
P. Zhang et al. Diabetes Research and Clinical Practice 208 (2024) 111094

[32] Vecchio F, Lo Buono N, Stabilini A, et al. Abnormal neutrophil signature in the [40] Khalil WJ, Akeblersane M, Khan AS, Moin ASM, Butler AE. Environmental
blood and pancreas of presymptomatic and symptomatic type 1 diabetes. JCI pollution and the risk of developing metabolic disorders: obesity and diabetes. Int J
Insight 2018;3(18). Mol Sci 2023;24(10):8870.
[33] Carrero JA, McCarthy DP, Ferris ST, et al. Resident macrophages of pancreatic [41] Coroyannakis C, Khalil A. Management of hypertension in the obese pregnant
islets have a seminal role in the initiation of autoimmune diabetes of NOD mice. patient. Curr Hypertens Rep 2019;21(3):24.
Proc Natl Acad Sci U S A 2017;114(48):E10418–27. [42] Desai L, Stefek H, Berhane H, Robinson J, Rigsby C, Markl M. Four-dimensional
[34] Wållberg M, Cooke A. Immune mechanisms in type 1 diabetes. Trends Immunol flow magnetic resonance imaging for assessment of pediatric coarctation of the
2013;34(12):583–591.39. aorta. J Magn Reson Imaging 2022;55(1):200–8.
[35] Ducluzeau PH, Fauchier G, Herbert J, et al. Prevalence and incidence of [43] American Diabetes Association. 10. Cardiovascular disease and risk management:
cardiovascular and renal diseases in type 1 compared with type 2 diabetes: a standards of medical care in diabetes-2021. Diabetes Care 2021;44(Suppl 1):
nationwide French observational study of hospitalized patients. Diabetes Metab S125–50.
2023;49(3):101429. [44] Ménégaut L, Laubriet A, Crespy V, et al. Inflammation and oxidative stress markers
[36] Ormazabal V, Nair S, Elfeky O, Aguayo C, Salomon C, Zuñiga FA. Association in type 2 diabetes patients with advanced carotid atherosclerosis. Cardiovasc
between insulin resistance and the development of cardiovascular disease. Diabetol 2023;22(1):248.
Cardiovasc Diabetol 2018;17(1):122. [45] Cersosimo E, DeFronzo RA. Insulin resistance and endothelial dysfunction: the road
[37] Shah AS, Black S, Wadwa RP, et al. Insulin sensitivity and arterial stiffness in youth map to cardiovascular diseases. Diabetes Metab Res Rev 2006;22(6):423–36.
with type 1 diabetes: the SEARCH CVD study. J Diabetes Complications 2015;29 [46] Holmannova D, Borsky P, Andrys C, et al. Chromosomal aberrations and oxidative
(4):512–6. stress in psoriatic patients with and without metabolic syndrome. Metabolites
[38] Lumeng CN, Saltiel AR. Inflammatory links between obesity and metabolic disease. 2022;12(8):688.
J Clin Invest 2011;121(6):2111–7. [47] Yao X, Levine SJ. Scavenger hunt: SR-B1, adrenal insufficiency, IL-17A, and
[39] Ebrahimi F, Urwyler SA, Schuetz P, et al. Effects of interleukin-1 antagonism on neutrophilic airway inflammation in asthma. Am J Respir Cell Mol Biol 2021;64
cortisol levels in individuals with obesity: a randomized clinical trial. Endocr (6):650–1.
Connect 2019;8(6):701–8. [48] Li X, Leng J, Yu L, et al. Efficacy of the biocontrol agent Trichoderma hamatum
against lasiodiplodia theobromae on macadamia. Front Microbiol 2022;13:
994422.