Evaluating the effect of green tea intake on cardiovascular diseases: A Mendelian randomization study in European and East Asian populations

Abstract

Previous research shows that more than 70% of cardiovascular diseases (CVDs) are attributed to modifiable risk factors. Here, we investigated relationship between consumption of green tea in European and East Asian populations and risk of CVDs using Mendelian randomization (MR). Instrumental variables for green tea intake were obtained from genome-wide association studies (GWASs) of 64,949 Europeans and 152,653 East Asians. GWASs for CVDs were derived from UK BioBank and BioBank Japan projects. The main method selected for MR analysis was either the inverse variance weighted (IVW) or Wald ratio, depending on the quantity of single nucleotide polymorphisms. Furthermore, we performed sensitivity analyses to confirm the reliability of the findings. Based on the results of IVW, there is no causal relationship between consumption of green tea and risk of 4 CVDs among Europeans (atrial fibrillation: OR = 1.000, 95% CI: 0.995–1.005, P = .910; heart failure: OR = 1.003, 95% CI: 0.994–1.012, P = .542; ischemic stroke: OR = 1.002, 95% CI: 0.993–1.011, P = .690; coronary artery disease: OR = 1.001, 95% CI: 0.996–1.007, P = .677). Sensitivity analyses and supplementary MR analyses also verify the robustness of the findings. Likewise, there was no correlation between the consumption of green tea and the occurrence of CVDs in East Asians. The consumption of green tea is not associated with a reduced risk of CVDs in populations from Europe and East Asia. This means that those who are trying to reduce their risk of CVDs by drinking more green tea may not benefit from doing so.

1. Introduction

Cardiovascular diseases (CVDs), such as heart failure (HF), coronary artery disease (CAD), atrial fibrillation (AF), and ischemic stroke (IS), are currently one of the major diseases of global concern due to their high mortality and morbidity.^[1] Different regions of the world suffer from different levels of CVDs prevalence and mortality, with Eastern Europe and Central Asia having the highest mortality and North Africa and the Middle East having the highest prevalence.^[2] Recent statistics show that CVDs cause one-third of all global deaths in 2019.^[3] Males typically have lower mortality rates from CVDs compared to females, with 40% of deaths in males and 49% in females.^[4,5] Previous reports show that more than 70% of CVDs are caused by modifiable risk factors,^[6] such as diet,^[7] smoking,^[8] and education.^[9] It is therefore essential to improve our understanding of factors that lead to CVDs and to implement appropriate primary prevention approaches to manage the diseases.

Green tea is a beverage that is popular all over the world. Many Asian countries have traditionally used green tea for medicinal purposes.^[10] Approximately 21% of Americans drink green tea daily.^[11] Diet is the most modifiable factor closely linked to the prevention of many diseases. Researchers and the public are highly interested in the relationship between green tea consumption and various illnesses. Green tea was thought to have health benefits for diabetes,^[12] obesity,^[13] cancer,^[14] neurological disorders,^[15] and CVDs.^[16] Over the past decade, several studies have been performed to assess drinking green tea and the risk of CVDs.^[17–20] For instance, it has been noted that for every increase of 236.6 mL of green tea consumed, there is an associated reduction of approximately 4% in CVD mortality risk.^[21] Among older adults aged 65 and above, this reduction in CVD mortality risk increases to 11% for the same amount of green tea intake. Furthermore, a separate study indicated that an increase in green tea intake by 3 cups per day was linked to a 27% reduction in CVD events.^[22] However, these results have not been consistent. Making causal inferences from these observational studies, which are susceptible to reverse causation bias and residual confounding, is challenging. Randomized controlled trials (RCTs) are widely considered the gold standard for establishing causation. RCTs may not always be feasible due to their potential high cost and ethical concerns. Additionally, the dosage and duration of green tea intake may limit the conclusions drawn from RCTs.

Mendelian randomization (MR) is a powerful analytical method that uses genetic variation to study the causal impacts of risk factors on traits.^[23] MR utilizes genetic variants as instrumental variables to proxy green tea intake, thus controlling for confounding factors that often plague observational studies, such as lifestyle and socioeconomic status. This genetic randomization ensures that the direction of causality flows from the genetic variant to the disease outcome, effectively mitigating issues of reverse causation. Additionally, MR reduces measurement errors associated with self-reported dietary data, providing a more objective measure of exposure. Previous MR analysis analyzed association between consumption of green tea and risk of Parkinson disease,^[24] digestive tumors,^[25] and lung cancer.^[26] Based on the basic assumptions of MR, we have also previously identified a number of causal relationships.^[27–31] However, the impact of consuming green tea on CVDs has not been thoroughly investigated. This research utilized a 2-sample MR analysis to examine the effects of green tea intake on CVDs.

2. Materials and methods

2.1. Study design

With the hypothesis that green tea intake reduces CVD risk, we performed MR analysis on 2 independent datasets from European and East Asian populations. SNPs were chosen as instrumental variables (IVs) according to 3 key assumptions of the MR analysis: SNPs have a strong correlation with green tea intake; SNPs are not directly linked to the onset of CVDs; and SNPs impact the risk of CVDs solely through green tea intake, excluding any other intermediate factors (Fig. 1). Data for this study were sourced from publicly accessible databases. In this case, there was no need for ethical approval.

Figure 1.

Three main assumptions of MR analysis. The figure was created by Figdraw (www.figdraw.com). MR = Mendelian randomization.

2.2. Data sources

In European populations, genome-wide association analysis (GWASs) for phenotypes were obtained from UK Biobank^[32] and GWAS Catalog.^[33] GWASs information regarding consumption of green tea was acquired from UK Biobank, encompassing a total of 64,949 individuals. Participants were inquired about the quantity of green tea they consumed the previous day. Information regarding the quality control procedures, statistical techniques, and complete study layout can be found on the Neale Laboratory website.^[32] The GWAS Catalog was used to obtain GWASs data for 4 CVDs (AF, HF, IS, and CAD). Specifically, the GWAS Catalog included 970,216 controls and 60,620 cases for AF, 930,014 controls and 47,309 cases for HF, 406,111 controls and 34,217 cases for IS, and 424,528 controls and 122,733 cases for CAD.

For East Asian populations, GWASs data for phenotypes were acquired from National Bioscience Database Center (NBDC) and BioBank Japan (BBJ) project. Specifically, GWASs data for green tea intake was downloaded from the NBDC, which was uploaded by Matoba et al.^[34] They analyzed the frequency of green tea intake among 152,653 participants. The original literature contains details about the standard questionnaire, quality control procedures, and analytical techniques. GWAS data for CVDs were provided by the BBJ project. These data can be downloaded from IEU OpenGWAS platform.^[35] Specifically, the BBJ project included 8180 cases and 28,612 controls for AF, 9413 cases and 203,040 controls for HF, 17,671 cases and 192,383 controls for IS, and 29,319 cases and 183,134 controls for CAD. The detailed information on each trait. Detailed information on each trait is displayed in Table 1.

Table 1.

The detailed information on each trait.

Trait	Population	N cases	N controls	Sample size	Source
Green tea intake	European	-	-	64,949	UK Biobank
Atrial fibrillation	European	60,620	970,216	1030,836	GWAS Catalog
Heart failure	European	47,309	930,014	977,323	GWAS Catalog
Ischemic stroke	European	34,217	406,111	440,328	GWAS Catalog
Coronary artery disease	European	122,733	424,528	547,261	GWAS Catalog
Green tea intake	East Asian	-		152,653	NBDC
Atrial fibrillation	East Asian	8180	28,612	36,792	BBJ
Congestive heart failure	East Asian	9413	203,040	212,453	BBJ
Ischemic stroke	East Asian	17,671	192,383	210,054	BBJ
Coronary heart disease	East Asian	29,319	183,134	212,453	BBJ

“-” represents not applicable. BBJ = BioBank Japan, NBDC  = National Bioscience Database Center.

2.3. Instrumental variables selection

For European populations, we set the following criteria to screen instrumental variables: P < 5 × 10^-8, r² < 0.001, and kb = 10,000. Since the same criteria did not obtain enough SNPs in East Asian populations, a looser threshold was used (P < 5 × 10^-6, r² < 0.001, and kb = 10,000). The threshold has also recently been used to explore the causal relationship between green tea and digestive system cancer.^[26] The PhenoScanner database was utilized for phenotype scanning to examine potential confounders.^[36] The following criteria were set to be analyzed in the PhenoScanner database: P-value = 5 × 10^-8 or 5 × 10^-6, R² = 0.8, Build = 37 and Proxies: EUR/EAS. To ensure that the effects of SNPs were caused by the same allele, we discarded palindromic SNPs. In order to evaluate the efficacy of the instrumental variables, the F-statistic was computed.^[37]

2.4. MR analysis

Figure 2 displays the main approach used for MR analysis. First, the effector allele of the SNPs was used to harmonize the exposure and outcome information. If only one SNP was ultimately included, we used the Wald ratio for MR estimates. When more than one SNP was available, inverse variance weighting (IVW) was used as main method. According to the MR guidelines, IVW is deemed more effective than alternative methods in specific situations.^[38] IVW, on the other hand, assumes that each genetic variant is a reliable tool, which may not always be the case in reality. Therefore, weighted median and MR Egger were used as supplementary methods to enhance the assessment of IVW. The MR Egger approach permits pleiotropic effects in all genetic variations, demanding that such effects are independent from the variant-exposure correlation. When less than half of the instrumental variables used in MR analysis are effective, the weighted median method permits the inclusion of ineffective instruments in evaluating causal effects. To ensure the reliability of the findings, we performed various sensitivity analyses such as Cochran Q test, MR Egger intercept test, and MR-PRESSO. We employed Cochran Q test, a robust statistical tool, to assess the heterogeneity among the SNPs. Next, we employed MR Egger intercept approach to investigate pleiotropy. Finally, MR-PRESSO method was implemented for the identification and correction of outliers, ensuring the accuracy and reliability of our results.

Figure 2.

Schematic overview of the MR study design. IVW = inverse variance weighted, MR = Mendelian randomization, MR-PRESSO = MR Pleiotropy RESidual Sum and Outlier, SNPs = single-nucleotide polymorphisms.

To control for false discovery rate resulting from multiple tests, we set a significance threshold of 0.0125 (0.05/4 outcomes). The P values lower than .05 but not meeting the threshold for multiple-testing significance are considered as suggestive evidence. R software version 4.2.2 was used to perform the MR analysis using the TwoSampleMR (version 0.5.6) package.

3. Results

3.1. Instrumental variables

In European populations, 21 separate SNPs were discovered to have a connection with the intake of green tea (Supplementary Table 1). http://links.lww.com/MD/N243 Twenty-one SNPs remained after phenotypic scanning analysis using the PhenoScanner database. The range of F-statistics was 29.819 to 43.570, suggesting that the bias from weak instruments is minimal. The GWASs data from each of the 4 CVDs were then searched for these 21 SNPs. Specifically, 21 SNPs were obtained in the GWAS for AF, while only 8 were obtained in the GWAS for HF, 13 in the GWAS for IS, and 15 in the GWAS for CAD. In addition, no palindromic SNPs were found in AF and IS, whereas 2 palindromic SNPs were removed in HF and 3 palindromic SNPs were removed in CAD. Specifically, 21 SNPs for AF, 6 for HF, 13 for IS, and 12 for CAD were included as IVs for MR estimation in European populations. In East Asian populations, there were 10 separate SNPs that were linked to the intake of green tea (Supplementary Table 2) http://links.lww.com/MD/N243. Through phenotypic scanning, we found one SNP (rs79105258) that correlates with CAD and IS. Subsequently, we repeated the screening process described above. Specifically, 4 SNPs for AF, 8 for HF and 7 for both IS and CAD were included as IVs for MR estimation in the East Asian populations.

3.2. Green tea intake and CVDs

Causal association between the consumption of green tea and the risk of CVDs evaluated by the 3 MR approaches is presented in Figure 3. The IVW results indicated that there was no link between higher consumption of green tea and the risk of AF (P = .910), HF (P = .542), IS (P = .690), and CAD (P = .677) in European populations (Fig. 3A). In addition, the estimates of MR Egger and weighted median were consistent with IVW (P > .05). In East Asian populations, we repeated the MR analysis to increase the reliability of the results. Interestingly, we found the same results. Specifically, the IVW findings also indicated no link between increased consumption of green tea and the risk of AF (P = .869), HF (P = .410), IS (P = .840), and CAD (P = .125) in East Asian populations (Fig. 3B). The estimates of MR Egger and the weighted median also support the results of IVW (P > .05).

Figure 3.

MR estimates from green tea intake on CVD risk in (A) European and (B) East Asian populations. Inverse variance-weighted results have been bolded. AF = atrial fibrillation, CAD = coronary artery disease, HF = heart failure, IS = ischemic stroke.

MR-PRESSO, pleiotropy, and heterogeneity results are shown in Table 2. In European populations, there was no heterogeneity (P values: .244–.573) or pleiotropy (P values: .461–.837) in our MR estimates. In addition, MR-PRESSO method also supported the IVW results (P values: .517–.911) and detected no outliers, adding to the robustness of our results. In East Asian populations, there was heterogeneity in 2 traits, AF (P = .010) and IS (P = .003). However, the effect of heterogeneity appeared to be negligible, as the MR-Egger regression analysis showed no evidence of pleiotropy (P values: .272–.889). The MR-PRESSO method detected outliers that could be responsible for the heterogeneity. Interestingly, both raw estimates (P values: .163–.879) of the MR-PRESSO method and estimates (P for AF: 0.893; P for IS: 0.280) corrected for outliers supported the IVW results. Leave-one-out analysis plots, funnel plots and scatter plots are shown in Supplementary Figures 1–3 http://links.lww.com/MD/N243.

Table 2.

The results for heterogeneity, pleiotropy, and MR-PRESSO analysis.

Outcome	Heterogeneity*	Pleiotropy*	MR-PRESSO*		Heterogeneity†	Pleiotropy†	MR-PRESSO†
	IVW	MR Egger	Raw	Outlier-corrected	IVW	MR Egger	Raw	Outlier-corrected
Atrial fibrillation	0.244	0.837	0.911	-	0.010	0.821	0.879	0.893
Heart failure	0.573	0.461	0.517	-	0.786	0.526	0.309	-
Ischemic stroke	0.471	0.626	0.693	-	0.003	0.272	0.847	0.280
Coronary artery disease	0.477	0.786	0.679	-	0.471	0.889	0.163	-

“

-” represents no significant outliers.

“*” represents European populations

^“†”represents East Asian populations. IVW, inverse variance weighting.

4. Discussion

Our current research aimed to investigate the possible cause-and-effect connection between drinking green tea and the likelihood of developing 4 CVDs in European and East Asian groups. To our knowledge, this is the first MR analysis to investigate this causal estimate from a genetic perspective on a large scale and across populations. We systematically studied 4 common CVDs, including AF, HF, IS, and CAD. Consistently, our results revealed no association between the intake of green tea and the risk of CVDs using 3 methods of MR analysis and a series of sensitivity analyses.

The impact of green tea intake on CVDs risk is still inconsistent. A meta-analysis of East Asian groups found a notable link between drinking green tea and a lower risk of CAD (OR = 0.72).^[39] In addition, there was a 10% reduction in CAD risk (OR = 0.90) with increasing intake of 1 cup of green tea daily. Another meta-analysis that included 7 prospective cohort studies (9211 cases of CAD and 772,922 participants) also showed that consumption of green tea was linked to a lower risk of CAD, especially in East Asian populations with low to moderate consumption.^[40] One cup (300 mL) to 5 cups per day corresponded to OR values of 0.89, 0.84, 0.85, 0.88, and 0.92, respectively. Nevertheless, it is crucial to mention that many studies have indicated no link between drinking green tea and CAD risk,^[41–43] aligning with our results. Results from a meta-analysis conducted by Pang et al in 2014, which included 259,267 participants, showed that intake of green tea was negatively associated with IS (OR = 0.64).^[17] The link between consumption of green tea and the risk of AF was investigated by Liu et al who recruited 801 participants (400 controls and 401 patients with AF) using logistic regression analysis.^[44] They found that low-dose green tea intake was a protective factor against AF. Most of research into the link between green tea and HF has been done in animals.^[45,46] The current MR results showed that there was no association between the risk of HF and green tea intake in East Asian and European populations.

Despite the large body of experimental evidence suggesting a preventive effect of green tea and its constituents on CVDs, it is interesting to note that epidemiological studies have also shown an alternative result (absence of preventive effect), and this inconsistency deserves discussion. Firstly, incorrect categorization of green tea may be a key factor leading to inconsistent results, as there may be significant differences in the observed cohort, including factors such as concentration, brewing time, dose, and even the type of green tea. Secondly, the type of study may also be a contributing factor. Cohort studies usually show no association between exposure and outcome, whereas case-control studies show an association. Case-control studies are often affected by biases, like recall bias, which can result in hypotheses being invalidated. Similarly, patients from different ethnic backgrounds and geographical areas could be one of the reasons for the discrepancies observed in previous studies. Additionally, observational research can be impacted by confounding variables. For example, people who frequently drink green tea may develop other healthy lifestyle habits, such as a balanced diet and regular exercise, which independently lower the risk of CVDs.

While the current research did not demonstrate a preventative impact of consuming green tea on CVDs, we believe it is important to discuss the biological processes behind the possible health advantages of green tea. Catechins are the most abundant flavonoids found in extracts of green tea.^[47] By neutralizing reactive oxygen species, diminishing pro-oxidant enzymes, and promoting antioxidant enzymes, catechins display antioxidant activities.^[48] The antioxidant properties of catechins are thought to play a critical role in the cardioprotective benefits.^[49] Low-density lipoprotein (LDL) has been widely acknowledged and researched as a contributor to CVDs. Accumulation of LDL in the arterial wall causes lesions by oxidation when protective agents (such as antioxidants) are depleted.^[50] A number of studies have reported that catechins can be inhibitors of LDL oxidation.^[51,52] In addition, dyslipidemia, caused by abnormal lipid metabolism, is a major risk factor for CVDs. Catechins have an impact on cellular solubility, the absorption of lipids in the intestines, and the hydrolysis of luminal lipids.^[53] Additionally, catechins have the ability to increase the expression of LDL receptors in the liver, thus regulating lipid biosynthesis, intracellular processing, and excretion. Vascular inflammation is increasingly recognized as a key process in the pathogenesis of atherosclerosis, initiated by the adhesion of monocytes to vascular endothelial cells and their subsequent transmigration to sites of inflammation.^[54] This process is dependent on the expression of monocyte chemoattractant protein-1 (MCP-1) and interleukin-8 (IL-8), along with adhesion molecules such as intercellular cell adhesion molecule-1 (ICAM1), vascular cell adhesion molecule 1 (VCAM-1), and endothelial leukocyte adhesion molecule-1 (E-selectin).^[55] Recent studies have shown that green tea catechins, particularly epigallocatechin gallate (EGCG), can significantly reduce markers of inflammation.^[56] EGCG and other catechins dose-dependently inhibited cytokine-induced VCAM-1 expression and monocyte adhesion in human umbilical vein endothelial cells (HUVEC).^[57] In LDL receptor knock-out mice, catechin administration significantly suppressed VCAM-1 expression in atherosclerotic lesions.^[58] Green tea catechins also reduce the expression of pro-inflammatory cytokines, chemokines, and adhesion molecules regulated by the nuclear transcription factor NF-kappa B.^[56] Although this MR analysis suggests that green tea intake is not a protective factor against CVDs, the potential cardiovascular benefits of green tea extracts cannot be entirely dismissed.

The study of diet as a factor influencing a disease or trait is particularly susceptible to confounding variables, as diet is associated with various lifestyle and socio-economic factors. The findings of case-control studies frequently indicate a link between drinking green tea and CVDs, implying that green tea could potentially offer protection against these illnesses. However, it is important to consider the possibility of reverse causality influencing these conclusions. MR studies are designed to prevent the influence of confounding factors and reverse causality. Although there may be other factors that challenge the assumptions of the MR, the significance of these findings remains. In conclusion, we suggest that the influence of consuming green tea on the progression of CVDs may have been overestimated in previous observational studies, especially among East Asian groups. This study, utilizing Mendelian randomization, found no significant causal relationship between green tea intake and CVD risk in European and East Asian populations. These findings hold substantial implications for public health recommendations. Firstly, there is currently insufficient genetic evidence to support the protective effect of green tea intake against CVDs, suggesting that solely relying on green tea intake for CVD prevention may be impractical. Secondly, CVD prevention should emphasize overall dietary and lifestyle modifications rather than relying on a single dietary factor. Although this study did not demonstrate the protective effects of green tea, it does not entirely negate its potential health benefits; future research should continue to explore other components of green tea and their possible health effects. Health education should scientifically convey the potential benefits and limitations of green tea to prevent public misinformation and enable more informed health decisions. In summary, this study provides scientific evidence for public health recommendations, underscoring the importance of comprehensive health strategies.

While MR offers significant methodological advantages, it also has inherent limitations that may influence the findings. First, MR relies on the availability and validity of genetic instruments. If the selected genetic variants are weakly associated with green tea intake or influence the outcome through pathways other than the exposure of interest, this could lead to biased estimates. This issue, known as weak instrument bias or pleiotropy, can undermine the assumptions of MR and affect the reliability of the results. Second, MR studies assume a linear relationship between the genetic variants, the exposure, and the outcome. Non-linear relationships, where the effect of green tea intake on CVDs might vary at different levels of intake, are not well captured by this approach. This could result in an underestimation or overestimation of the true effect. Third, population stratification can introduce bias if the genetic variants used as instruments differ in frequency across populations due to ancestry, potentially confounding the results. Although our study includes both European and East Asian populations to mitigate this issue, there is still a possibility of residual confounding due to unaccounted-for population differences. Finally, MR assumes that the relationship between genetic variants and exposure remains constant over time. However, changes in dietary habits, availability of green tea, and other external factors can influence this relationship, potentially affecting the validity of the findings.

5. Conclusion

MR estimates do not provide evidence of a causal association between consumption of green tea and risk of CVDs among European and East Asian groups. Therefore, those trying to reduce risk of CVDs by increasing green tea intake may not benefit.

Author contributions

Conceptualization: Ziming Peng.

Data curation: Qiaoli Liang.

Formal analysis: Qiaoli Liang.

Methodology: Qiaoli Liang.

Writing – original draft: Qiaoli Liang.

Writing – review & editing: Ziming Peng.