Anxiety disorder and cardiovascular disease: a two‐sample Mendelian randomization study

Abstract

Aims

Cardiovascular disease is the leading cause of death worldwide. Anxiety disorders are common psychiatric conditions associated with cardiovascular outcomes. This two‐sample Mendelian randomization (MR) study investigated the causal relationship between anxiety disorders and coronary heart disease (CHD), myocardial infarction (MI), heart failure (HF), and atrial fibrillation (AF).

Methods

Single nucleotide polymorphisms (SNPs) associated with anxiety disorders (16 730 cases; 101 021 controls) were obtained from the UK Biobank genome‐wide association study (GWAS). Cardiovascular outcome data were derived from the FinnGen study (CHD: 21 012 cases and 197 780 controls; MI: 12 801 cases and 187 840 controls; HF: 23 397 cases and 194 811 controls; and AF: 22 068 cases and 116 926 controls). Inverse variance weighted (IVW), MR–Egger, weighted median, simple mode, and weighted mode analyses examined causality.

Results

IVW analysis demonstrated significant causal relationships between anxiety disorders and increased risk of CHD [odds ratio (OR): 4.496; 95% confidence interval (CI): 1.777–11.378; P = 0.002], MI (OR: 5.042; 95% CI: 1.451–17.518; P = 0.011), and HF (OR: 3.255; 95% CI: 1.461–7.252; P = 0.004). No relationship was observed with AF (OR: 1.775; 95% CI: 0.612–5.146; P = 0.29). Other methods showed non‐significant associations. Two‐way analysis indicated no reverse causality.

Conclusions

Anxiety disorders were causally associated with greater risk of CHD, MI, and HF but not AF among individuals of European descent. Further research on mediating mechanisms and in diverse populations is warranted.

Introduction

According to the World Health Organization (WHO), cardiovascular disease has been the leading cause of death worldwide for the past 20 years, accounting for 16% of all causes of death. ¹ As the pace of life increases and psychological stress becomes more prevalent, the incidence of cardiovascular disease and psychological disorders, including anxiety disorders, continues to rise. Anxiety disorders are among the most common psychiatric disorders, characterized by severe symptoms, clinical burden, and treatment difficulties. ² Research has demonstrated a close association between anxiety disorders and cardiovascular disease. ³ Furthermore, anxiety disorder itself is an independent risk factor for cardiovascular morbidity and mortality. ⁴ , ⁵ The interplay between these two conditions can worsen the patient's overall health. ⁶ Therefore, it is crucial to explore the causality between anxiety disorders and cardiovascular disease to guide prevention and treatment strategies.

Although numerous studies have shown a correlation between anxiety disorders and cardiovascular disease, establishing causality is complex due to confounding factors and reverse causation. Currently, there are on randomized controlled trials with large sample sizes to investigate the causal relationship between these two disorders.

In this study, we employ Mendelian randomization (MR), which utilizes the principles of the second Mendelian law. The random assortment of genes on nonhomologous chromosomes during allele segregation is similar to the randomization process in randomized controlled trials. Furthermore, genotype determines phenotype, and the phenotype is linked to the disease. Genotype is not influenced by environmental or confounding factors that occur after disease onset. Therefore, genotype‐based MR can be used to infer associations between diseases. ⁷ Our study employs MR to analyse the causal relationship between anxiety disorders and several common cardiovascular diseases, including coronary heart disease (CHD), myocardial infarction (MI), atrial fibrillation (AF), and heart failure (HF).

Methods

Data sources

In this study, the exposure variable is anxiety disorders, and the relevant data were obtained from the UK Biobank genome‐wide association study (GWAS). ⁸ The exposure is defined by the UK Biobank as mental health problems ever diagnosed by a professional, specifically anxiety, nerves, or generalized anxiety disorder. The cohort used for the exposure consists of European Caucasians, with 16 730 cases and 101 021 controls.

Regarding the outcome variable, the study focuses on four common cardiovascular diseases: CHD, MI, AF, and HF. To minimize sample overlap and reduce research bias, outcome data were obtained from the FinnGen database. ⁹ These cardiovascular diseases were defined by FinnGen using the International Classification of Diseases, 9th Revision (ICD‐9) diagnostic codes and encompass a European Caucasian cohort. The cohort sizes for each disease are as follows: CHD (21 012 cases and 197 780 controls), MI (12 801 cases and 187 840 controls), AF (22 068 cases and 116 926 controls), and HF (23 397 cases and 194 811 controls).

By utilizing these datasets, the study aims to investigate the causal relationship between anxiety disorders and the aforementioned cardiovascular diseases through MR.

Selection of instrumental variable

In classical MR studies, instrumental variables (IVs) are required to satisfy the following three core assumptions (Figure 1): (1) associational assumption: single nucleotide polymorphisms (SNPs) are closely associated with the exposure. Because only one SNP meets the significance threshold (P < 5 × 10⁻⁸), the P‐value threshold will be expanded to 5 × 10⁻⁶. In addition, we also use the F statistic to assist in evaluating the association hypothesis, using the formula F statistic = R ² (N − 2)/(1 − R ²) to calculate the F statistic for an SNP. R ² was calculated using the following formula: R ² = (2 × EAF × (1 − EAF) × β ²) /[(2 × EAF × (1 − EAF) × β ²) + (2 × EAF × (1 − EAF) × N × SE². ¹⁰ , ¹¹ When F statistic >10, the correlation hypothesis is satisfied; that is, there is no weak tool bias in the study. ¹² (2) Independence assumption: SNPs are independent of confounding factors on the exposure–outcome pathway. To verify this, the SNPs obtained in the first step should be tested to ensure their lack of significant association (P > 0.05) with the outcome. (3) Exclusion restriction assumption: SNPs can only be related to the outcome through the exposure and should not influence the outcome through other pathways. ¹² , ¹³ The MR–Egger model is used to test the exclusion restriction assumption. When the intercept term of the MR–Egger model has no significant difference from zero, it is considered that the MR research satisfies the exclusion restriction assumption. Meanwhile, this assumption also can be assessed by checking the association (P > 5 × 10⁻⁸) between the SNPs and confounding factors using the PhenoScanner website (http://www.phenoscanner.medschl.cam.ac.uk/). ¹⁴ In addition, to ensure that there is no linkage disequilibrium (LD) between SNPs affecting the results of the analysis, we chose the LD clumping, R ² < 0.001, and 10 000 KB as the minimum window size. ⁷

Figure 1.

The overall design of the Mendelian randomization (MR) study. Instrumental variables must meet the following: (1) associational assumption: single nucleotide polymorphisms (SNPs) are closely associated with the exposure; (2) independence assumption: SNPs are independent of confounding factors on the exposure–outcome pathway; and (3) exclusion restriction assumption: SNPs can only be related to the outcome through the exposure and should not influence the outcome through other pathways.

Mendelian randomization analysis

For the MR analysis, the main method is the inverse variance weighted (IVW) analysis, ¹⁵ along with MR–Egger regression (MR–Egger) method, weighted median estimator (WME) method, simple mode (SM) method, and weighted mode (WM) method as supplementary analyses. These methods estimate the causal effects of anxiety disorders on cardiovascular diseases. ⁷ , ¹⁶ , ¹⁷ Sensitivity analysis is conducted to assess the presence of heterogeneity between IVs using Cochran's Q test. The test of MR–Egger intercept and leave‐one‐out analysis are used to detect potential horizontal pleiotropy between IVs. ¹⁸ The ‘leave‐one‐out method’ involves iteratively removing an SNP at a time to evaluate its impact on the correlation between the exposure and outcome variables. To address potential reverse causality, a two‐way MR analysis is performed. ¹⁴ All statistical analyses are conducted using R Version 4.2.1 (https://cran.r‐project.org/) and the R package TwoSampleMR Version 0.5.7 (https://mrcieu.github.io/TwoSampleMR). ¹⁹ The threshold for statistical significance is set at P < 0.05.

Results

Basic information of instrumental variable

According to Assumption 1, we initially identified 17 SNPs strongly associated with the anxiety disorders (P < 5 × 10⁻⁶). According to the formula given, all SNPs have F statistics >10, which also indicates that these SNPs are closely related to anxiety disorders. Please refer to Supporting Information, Table S1 for specific information.

Considering Assumption 2, all SNPs were found to have no significant correlation with the outcome factors (P > 0.05). However, two SNPs were not matched in the outcome dataset and were subsequently excluded. Please refer to Supporting Information, Table S2 .

To address Assumption 3, we subjected the remaining 15 SNPs to analysis on the PhenoScanner website, which confirmed their lack of association with any confounding factors.

The effect of anxiety disorder on cardiovascular disease

In the main IVW analysis, we observed a significant causal relationship between anxiety disorder and CHD [odds ratio (OR): 4.496; 95% confidence interval (CI): 1.777–11.378; P = 0.002], MI (OR: 5.042; 95% CI: 1.451–17.518; P = 0.011), and HF (OR: 3.255; 95% CI: 1.461–7.252; P = 0.004). However, there was no significant causal relationship found between anxiety disorder and AF (OR: 1.775; 95% CI: 0.612–5.146; P = 0.29). MR–Egger method, WME method, SM method, and WM method did not find significant causal relationship (P > 0.05) between anxiety disorders and CHD, MI, HF, and AF (refer to Table 1 ). The Q statistics for the heterogeneity test of each exposure factor and the MR regression intercept for the horizontal pleiotropy test were finally selected, as shown in Table 2 . According to Cochran's Q statistic and its P value, there is no heterogeneity among the exposed SNPs, so the IVW analysis results of the random effects model by default are reliable. Similarly, according to the intercept term and its P value in MR–Egger regression, this indicates that there is no significant difference between the intercept term and zero. Therefore, we believe that horizontal pleiotropy does not exist, and the MR study further confirms the exclusion restriction assumption. Sensitivity analysis using the ‘leave‐one‐out method’ demonstrated robust stability of the MR results (refer to Supporting Information, Figures S1–S4 ). Hence, anxiety disorder is positively causally related to CHD, MI, and HF (refer to Figure 2 ).

Table 1.

MR estimates from each method of assessing the causal effect of anxiety disorder on risk of cardiovascular disease

Disease	Method	NSNPs	β value	SE	OR (95% CI)	P value
CHD	MR–Egger	15	0.199	1.072	1.22 (0.149–9.978)	0.856
	WME	15	1.484	0.666	4.41 (1.195–16.268)	0.026
	IVW	15	1.503	0.474	4.496 (1.777–11.378)	0.002
	SM	15	2.302	1.242	9.998 (0.877–113.983)	0.085
	WM	15	2.077	1.229	7.978 (0.717–88.773)	0.113
MI	MR–Egger	15	0.419	1.45	1.521 (0.089–26.059)	0.777
	WME	15	1.197	0.874	3.31 (0.597–18.35)	0.171
	IVW	15	1.618	0.635	5.042 (1.451–17.518)	0.011
	SM	15	1.469	1.554	4.347 (0.207–91.43)	0.361
	WM	15	1.139	1.725	3.124 (0.106–91.871)	0.52
HF	MR–Egger	15	1.021	0.925	2.777 (0.453–17.007)	0.289
	WME	15	0.998	0.586	2.713 (0.86–8.557)	0.089
	IVW	15	1.18	0.409	3.255 (1.461–7.252)	0.004
	SM	15	0.742	1.008	2.101 (0.291–15.177)	0.474
	WM	15	0.789	0.925	2.201 (0.359–13.495)	0.408
AF	MR–Egger	15	1.828	1.228	6.219 (0.56–69.031)	0.161
	WME	15	0.746	0.728	2.109 (0.506–8.783)	0.305
	IVW	15	0.574	0.543	1.775 (0.612–5.146)	0.29
	SM	15	1.114	1.417	3.046 (0.189–49.018)	0.445
	WM	15	1.114	1.364	3.046 (0.21–44.157)	0.428

AF, atrial fibrillation; CHD, coronary heart disease; CI, confidence interval; HF, heart failure; IVW, inverse variance weighted; MI, myocardial infarction; MR, Mendelian randomization; MR–Egger, Mendelian randomization–Egger regression; NSNPs, number of single nucleotide polymorphisms; OR, odds ratio; SE, standard error; SM, simple mode; WM, weighted mode; WME, weighted median estimator.

Table 2.

The heterogeneity and pleiotropy tests of the instrumental variables

	Cochran's Q test		MR–Egger intercept
	Value	P value	Value	P value
CHD	12.965	0.529	0.017	0.198
MI	17.29	0.24	0.0152	0.374
HF	10.141	0.752	0.002	0.851
AF	13.11	0.518	−0.016	0.276

AF, atrial fibrillation; CHD, coronary heart disease; HF, heart failure; MI, myocardial infarction; MR–Egger, Mendelian randomization–Egger regression.

Figure 2.

Scatter plots of Mendelian randomization (MR) analysis of anxiety disorders and four cardiovascular diseases. On the basis of scatter, the effect of single nucleotide polymorphism (SNP) is fitted by five methods. The effect of SNP on exposure is placed on the horizontal axis of the image, and the effect on outcome is placed on the vertical axis. The slope in the figure is the effect value of several methods of MR analysis. (A) Anxiety disorder (AD) and coronary heart disease (CHD), (B) AD and myocardial infarction (MI), (C) AD and heart failure (HF), and (D) AD and atrial fibrillation (AF). IVW, inverse variance weighted; MR Egg, Mendelian randomization–Egger regression; WM, weighted mode; WME, weighted median estimator.

The effect of cardiovascular disease on anxiety disorder

To investigate the presence of a reverse causal relationship between the two diseases, we attempted a reverse MR analysis. However, the IVW results indicated no significant causal relationship between the four heart diseases and anxiety disorders (refer to Table 3 ). Therefore, further analysis in this direction was not pursued.

Table 3.

The bidirectional MR estimates from IVW of assessing the causal effect of cardiovascular disease on risk of anxiety disorder

Exposure	NSNPs	β value	SE	P value
CHD	24	0.0012	0.0037	0.745
MI	12	−0.00048	0.0037	0.897
HF	2	0.0078	0.0139	0.573
AF	29	−0.0015	0.0023	0.519

AF, atrial fibrillation; CHD, coronary heart disease; HF, heart failure; IVW, inverse variance weighted; MI, myocardial infarction; MR, Mendelian randomization; NSNPs, number of single nucleotide polymorphisms; SE, standard error.

Discussion

Cardiovascular disease, being a leading cause of mortality, poses a significant burden on society. ²⁰ Consequently, the prevention of cardiovascular disease is of utmost importance. Research indicates that anxiety and other psychiatric disorders serve as independent risk factors for cardiovascular disease, exhibiting a close association. ²¹ However, due to the lack of large‐scale randomized controlled trials, establishing a causal relationship between these two conditions remains challenging. Therefore, this study utilizes a two‐sample MR study with two distinct samples to investigate the causal relationship. The findings indicate that anxiety disorders increase the risk of CHD, MI, and HF, while providing limited evidence for an effect on AF. Importantly, the two‐sample MR study also rules out reverse causality between cardiovascular disease and anxiety disorder.

Numerous studies have consistently reported elevated levels of inflammatory markers, such as interleukin, C‐reactive protein, and homocysteine, in individuals with anxiety. ²² , ²³ Consequently, anxiety disorders may instigate chronic inflammation, which has the potential to harm the endothelial lining of blood vessels, a factor associated with heart attacks or strokes. ²⁴ , ²⁵ The impact of anxiety on the cardiovascular system involves a complex autonomic circuit, comprising anxiety‐related nuclei and the autonomic nervous system. ²⁶ Anxiety disorders may induce psychological stress, triggering the activation of the sympathetic nervous system and the hypothalamic–pituitary–adrenal axis, resulting in elevated levels of catecholamines, cortisol, and other hormones that can destabilize atherosclerotic plaques, ultimately leading to thrombosis, a primary cause of acute coronary syndromes. ²⁷ , ²⁸ , ²⁹ Furthermore, anxiety disorders may elevate the risk of cardiac dysfunction and adverse outcomes by potentially impeding the cardioprotective effects of vagus‐mediated B‐type natriuretic peptide (BNP). ³⁰ These findings collectively support a plausible connection between anxiety disorders and cardiovascular disease.

Observational studies provide compelling evidence for the association between cardiovascular disease and anxiety disorder. A meta‐analysis encompassing 37 studies confirmed that anxiety increases the risk of developing new cardiovascular disease. ³¹ Additionally, a separate meta‐analysis comprising 38 studies revealed an overall prevalence of anxiety in HF patients of ~32.0%. ³² In a study conducted in 43 general practice clinics in Australia, anxiety disorders were frequently observed among patients with HF. ³³ Furthermore, anxiety has been shown to elevate hospitalization rates and mortality, exacerbate symptoms, and worsen the New York Heart Association (NYHA) classification in patients with congestive HF. ³⁴ , ³⁵ These findings align with the results of our study.

Furthermore, our study demonstrates an increased risk of CHD and acute MI in individuals with anxiety disorders. A cross‐sectional study involving 15 105 individuals found that generalized anxiety disorder tripled the risk of CHD and MI. ³⁶ Meta‐analyses have consistently highlighted a close association between anxiety and CHD, estimating a 41% increased risk. ⁴ , ³⁷ Epidemiological studies have also identified that persistent anxiety and tension over four consecutive years substantially elevate the risk of cardiovascular death in patients with stable angina by 2–4 times. ³⁸ Additionally, an observational study revealed greater carotid intima‐media thickness in patients exhibiting anxiety symptoms and generalized anxiety, ³⁹ providing further evidence of the relationship between anxiety and atherosclerosis.

However, our study did not find a causal relationship between anxiety disorder and AF. Although previous studies generally suggest a positive association between anxiety and the risk of AF, ⁴⁰ subsequent findings have yielded inconsistent results. A meta‐analysis of 11 cohort studies reported no significant association between psychological factors such as anxiety, depression, anger, work stress, and the risk of AF. ⁴¹ Similarly, a large population‐based follow‐up study found no evidence supporting an association between anxiety or severe depression symptoms and the risk of AF. ⁴² These contradictory findings indicate a lack of substantial causal relationship between anxiety disorder and AF. Nevertheless, given the absence of causality, it is advisable to focus on the screening and intervention of AF in patients with anxiety disorder.

Limitations

MR methods are widely used to evaluate treatments, drug effects, biological pathways, and disease associations. It provides an alternative to deal with causal issues, especially when randomized controlled trials are not available, and can provide a degree of causal inference support. However, MR also has some limitation, including the selection of genetic variation, the diversity of genetic variation between observation and the complexity of the problem. The appropriate selection of appropriate instrumental variables and data sources in conjunction with other evidence and domain knowledge is therefore crucial for the interpretation and validation of results. Our study, which used an MR approach, also has limitations. Firstly, the limited availability of SNPs associated with anxiety disorder necessitated a relaxation of the most stringent significance threshold. Moreover, the study primarily focuses on participants of European descent, and although efforts were made to minimize bias through the utilization of samples from Britain and Finland, the generalizability of the conclusions to Asian or African populations may be limited. Therefore, conducting further GWAS in diverse ethnic populations, including Asian and other groups, is recommended to obtain additional SNPs and enhance the research quality of MR analysis. Our study only analysed the simple causal relationship between anxiety disorders and cardiovascular diseases by two‐sample MR method. The specific mechanisms mediating the relationship between anxiety disorders and cardiovascular diseases were not studied indetail. Therefore, future studies can be further explored by mediating MR method combined with GWAS data of potential risk factors.

Conclusions

In summary, our study indicates that anxiety disorders are associated with an increased risk of CHD, MI, and HF. However, we did not find a significant causal relationship between anxiety disorder and AF. These findings suggest that anxiety disorder may play a role in the development of certain cardiovascular diseases but may not have a direct impact on AF. Further research is needed to explore the underlying mechanisms and clarify the relationship between anxiety disorders and AF.

Conflict of interest

None of the article contents are under consideration for publication in any other journal or have been published in any journal. The authors declare that they have no conflict of interest.

Funding

This work was supported by grants from the National Natural Science Foundation of China (82070330 and 82200348).

Supporting information

Table S1. Basic information on instrumental variable.

Table S2. The P value of SNPs on the outcome.

Figure S1. The result of Leave‐one‐out method of SNPs associated with anxiety disorder and the risk of coronary heart disease.

Figure S2. The result of Leave‐one‐out method of SNPs associated with anxiety disorder and the risk of myocardial infarction.

Figure S3. The result of Leave‐one‐out method of SNPs associated with anxiety disorder and the risk of heart failure.

Figure S4. The result of Leave‐one‐out method of SNPs associated with anxiety disorder and the risk of atrial fibrillation.

Peng, B. , Meng, H. , Guo, L. , Zhu, J. , Kong, B. , Qu, Z. , Shuai, W. , and Huang, H. (2024) Anxiety disorder and cardiovascular disease: a two‐sample Mendelian randomization study. ESC Heart Failure, 11: 1174–1181. 10.1002/ehf2.14676.