Dissecting causal relationships between dietary factors and intracerebral hemorrhage: A Mendelian randomization study

Abstract

Intracerebral hemorrhage (ICH) is a severe and often fatal brain disorder. Despite the recognition of dietary adjustments as a preventive measure, there is a need for well-designed studies to investigate the dietary factors of ICH patients. We employed Mendelian randomization to explore the relationship between 35 dietary factors (exposures) and ICH (outcome). We applied several statistical methods, including inverse variance weighted, weighted median, Mendelian randomization-Egger regression, simple mode, and weighted mode. Furthermore, sensitivity analyses were conducted to ensure the robustness of our findings. Our study suggested that hot drink temperature (OR: 0.39, P = .037) and fresh fruit intake (OR: 0.389, P = .048) could potentially reduce the risk of ICH. Additionally, although a suggestively increased risk of ICH was observed with cooked vegetable intake (OR: 4.493, P = .036), the wide confidence interval (1.107–18.246) reflects imprecision in the estimate, and further research is needed to clarify this association. These findings provide novel dietary intervention strategies and insights for preventing ICH, holding the potential to inform more targeted health guidance recommendations.

1. Introduction

Intracerebral hemorrhage (ICH), which accounts for approximately 10% to 20% of all stroke cases, is a prominent cause of stroke and significantly contributes to mortality and long-term disability worldwide.^[1,2] While the exact causes of ICH remain elusive, some studies have revealed certain risk factors that play a role in the development of ICH, such as hypertension, diabetes and chronic kidney disease.^[3–7] Addressing these risk factors, dietary habits play a crucial role in maintaining stable blood pressure, controlling blood sugar, and improving kidney function.^[8–10] Through adjusting our dietary habits, it might decrease the likelihood of developing ICH. A systematic review and meta-analysis, which synthesized data from 20 prospective cohort studies across Europe, the United States, and Asia, found that the intake of fruits and vegetables has a protective effect against stroke.^[11] Additionally, a meta-analysis has indicated that in Japanese, but not in non-Japanese, a diet high in saturated fat is associated with a low risk of ICH.^[12] However, despite some studies involving a small amount of dietary information, there remains a lack of systematic and well-designed research into the dietary habits of ICH patients, which is crucial for developing dietary interventions to prevent ICH.

Mendelian randomization (MR) analysis, an innovative statistical technique, employs genetic variants like single nucleotide polymorphisms (SNPs) as instrumental variables (IVs) to estimate the causal relationship between exposure and outcome with high precision.^[13] By leveraging the random assortment of genetic variations during meiosis, MR analysis can minimize the impact of confounding factors and reverse causality, thus controlling for biases often present in observational studies.^[14,15] In this study, we applied MR analysis to explore the genetic causal relationship between 35 dietary factors and ICH, aiming to shed light on potential preventive measures for ICH.

2. Materials and methods

2.1. Study design

This study investigates the relationship between dietary factors and ICH. The design and data sources of the study are presented in Figure 1. To more accurately estimate the causal effects, the application of SNPs as IVs in MR analysis is predicated on 3 critical assumptions: first, the IV must be closely associated with the exposure; second, the IV should be independent of confounding factors; and third, the IV should exert its influence on the outcome exclusively through the exposure and not through any other pathways.

Figure 1.

A directed acyclic graph depicting the hypothesized impact of dietary factors on ICH. ICH = intracerebral hemorrhage.

2.2. Data sources

We conducted a comprehensive analysis of 35 dietary factors, encompassing a diverse range of food intakes. These included various types of staple food intakes such as cereal, plain cereal, whole-wheat cereal, bran cereal, white pasta, whole meal pasta, bread, and white rice; meat intakes such as pork, poultry, beef, bacon, processed meat, oil fish, and nonoily fish; beverage intakes such as water, tea, green tea, herbal tea, milk, yogurt, coffee, fizzy drink, red wine, beer plus cider, white wine, rose wine and spirits; fruit and vegetable intakes such as fresh fruit, stewed fruit, dried fruit, salad/raw vegetable, and cooked vegetable intake; alcohol intake frequency; and hot drink temperature. The relevant genome-wide association study (GWAS) datasets were sourced from the IEU OpenGWAS project (https://gwas.mrcieu.ac.uk), and dataset details are presented in Table 1.

Table 1.

Detailed information about data sources of dietary factors and ICH.

Traits	GWAS ID	Sample size	SNPs	Population	P value
Cereal intake	ukb-b-15926	441,640	9851,867	European	5.00E‐08
Plain cereal intake	ukb-b-20023	64,949	9851,867	European	5.00E‐06
Whole-wheat cereal intake	ukb-b-2375	64,949	9851,867	European	5.00E‐06
Bran cereal intake	ukb-b-19498	64,949	9851,867	European	5.00E‐06
White pasta intake	ukb-b-2512	64,949	9851,867	European	5.00E‐06
Wholemeal pasta intake	ukb-b-4150	64,949	9851,867	European	5.00E‐06
Bread intake	ukb-b-11348	452,236	9851,867	European	5.00E‐08
White rice intake	ukb-b-14690	64,949	9851,867	European	5.00E‐06
Pork intake	ukb-b-5640	460,162	9851,867	European	5.00E‐08
Poultry intake	ukb-b-8006	461,900	9851,867	European	5.00E‐08
Beef intake	ukb-b-2862	461,053	9851,867	European	5.00E‐08
Bacon intake	ukb-b-4414	64,949	9851,867	European	5.00E‐06
Processed meat intake	ukb-b-6324	461,981	9851,867	European	5.00E‐08
Oily fish intake	ukb-b-2209	460,443	9851,867	European	5.00E‐08
Nonoily fish intake	ukb-b-17627	460,880	9851,867	European	5.00E‐08
Water intake	ukb-b-14898	427,588	9851,867	European	5.00E‐08
Tea intake	ukb-b-6066	447,485	9851,867	European	5.00E‐08
Green tea intake	ukb-b-4078	64,949	9851,867	European	5.00E‐08
Milk intake	ukb-b-2966	64,943	9851,867	European	5.00E‐06
Yogurt intake	ukb-b-7753	64,949	9851,867	European	5.00E‐06
Coffee intake	ukb-b-5237	428,860	9851,867	European	5.00E‐08
Fizzy drink intake	ukb-b-2832	64,949	9851,867	European	5.00E‐06
Average weekly red wine intake	ukb-b-5239	327,026	9851,867	European	5.00E‐08
Average weekly beer plus cider intake	ukb-b-5174	327,634	9851,867	European	5.00E‐08
White wine intake	ukb-b-311	64,949	9851,867	European	5.00E‐06
Rose wine intake	ukb-b-16124	64,949	9851,867	European	5.00E‐06
Average weekly spirits intake	ukb-b-1707	326,565	9851,867	European	5.00E‐08
Alcohol intake frequency	ukb-b-5779	462,346	9851,867	European	5.00E‐08
Herbal tea intake	ukb-b-13344	64,949	9851,867	European	5.00E‐08
Hot drink temperature	ukb-b-14203	457,873	9851,867	European	5.00E‐08
Fresh fruit intake	ukb-b-3881	446,462	9851,867	European	5.00E‐08
Stewed fruit intake	ukb-b-8676	64,942	9851,867	European	5.00E‐06
Dried fruit intake	ukb-b-16576	421,764	9851,867	European	5.00E‐08
Salad/raw vegetable intake	ukb-b-1996	435,435	9851,867	European	5.00E‐08
Cooked vegetable intake	ukb-b-8089	448,651	9851,867	European	5.00E‐08
Intracerebral hemorrhage	ebi-a-GCST90018870	473,513	24,191,284	European	NA

ICH = intracerebral hemorrhage.

2.3. The selection of IVs

We selected SNPs as IVs, adhering to a set of criteria to ensure the accuracy and reliability of our findings. We adopted a primary P-value threshold of P < 5.0 × 10^‐8 to ensure the robustness of the IVs. However, given the limited sample sizes and lower heritability of some dietary factors (Table 1), we implemented a secondary threshold of P < 5.0 × 10^‐6. This approach balances 2 risks: the risk of losing informative instruments at overly stringent thresholds, against the risk of introducing non-informative instruments and increased pleiotropy at higher thresholds.^[16] To avoid linkage disequilibrium bias, we required that the significant SNPs associated with exposure factors must have r² < 0.001 with any other SNP, and be genetically distant by 10,000 kb. We extracted the significant SNPs related to exposure factors from the GWAS dataset of the outcome variable, documenting comprehensive details of these IVs. This information included the effect allele, the effect size of the allele (β), SE, and P-value. To test the strength of each IV, we calculated the F-statistic with the formula F = (β_exposure/SE_exposure)². We selected all IVs with an F-statistic > 10 to address potential weak instrument bias.^[17] Additionally, we searched for each selected SNP in databases such as Phenoscanner^[18] and the GWAS catalog^[19] to ensure that the selected SNPs are not associated with potential confounders such as age, sex, and lifestyle factors (P < 5.0 × 10⁻⁸).

2.4. Statistical analysis

In the MR analysis, we utilized a variety of methods to infer causal relationships, including inverse variance weighted (IVW), weighted median, MR-Egger regression, simple mode, and weighted mode. Among these, we adopted the IVW method as the primary approach for MR analysis. When each genetic variant meets the IV assumptions, the IVW method integrates the Wald ratio estimates of causal effects from different SNPs, providing a consistent estimate of the causal effect of exposure on the outcome.^[20] To enhance the reliability of our results, we employed the weighted median and MR-Egger methods as supplements to the IVW method.^[21,22] We used Cochran Q test to assess heterogeneity, considering significant heterogeneity to be present when the P < .05.^[23] Furthermore, we assessed horizontal pleiotropy using the intercept of the MR-Egger regression.^[24] To detect and correct for horizontal pleiotropy, we applied the MR-PRESSO test, which corrects by removing outliers.^[25] We also conducted a “leave-one-out” analysis, where we sequentially excluded 1 SNP at a time and calculated the MR effect of the remaining SNPs, to evaluate the impact of each SNP on the outcome. This was visually presented through a forest plot, allowing us to assess the stability of the MR analysis results. Statistical analyses were conducted using R software (version 4.4.0) and R packages (TwoSampleMR^[26] and MR-PRESSO^[27]). We applied the Bonferroni correction to adjust for multiple comparisons, considering P < .0014 (0.05/35) as statistically significant, and 0.0014 < P < .05 as suggestive of potential associations.^[28]

3. Results

In our study, we screened the SNPs using a set of criteria, and excluded palindromic SNPs (i.e., A/T or G/C) and those not available in the outcome dataset. Ultimately, the number of SNPs used in our study ranged from 4 to 90, and all IVs had F-statistics > 10, indicating no weak IV bias.

We identified 3 dietary factors that have a potential causal relationship with ICH using the IVW method, all with P < .05. Specifically, hot drink temperature (OR: 0.39; 95% CI: 0.161–0.947; P = .037) and fresh fruit intake (OR: 0.389; 95% CI: 0.153–0.99; P = .048) were both recognized as potential protective factors for ICH (Figs. 2A, B and 3). These results suggest that increasing the intake of fresh fruits and hot drinks may help reduce the risk of ICH. Additionally, a suggestively increased risk of ICH was observed with cooked vegetable intake (OR: 4.493; 95% CI: 1.107–18.246; P = .036) (Figs. 2C and 3). However, the wide confidence interval reflects the imprecision in the estimate.

Figure 2.

Scatter plots of the causal relationship between ICH and dietary factors. (A) Hot drink temperature. (B) Fresh fruit intake. (C) Cooked vegetable intake. ICH = intracerebral hemorrhage.

Figure 3.

Forest plot from IVW method showing links between 35 dietary factors and ICH. CI = confidence interval, ICH = intracerebral hemorrhage, OR = odds ratio, SNP = single nucleotide polymorphism.

Cochran Q test revealed no statistically significant heterogeneity in the MR studies examining the relationship between hot drink temperature, fresh fruit intake, and cooked vegetable intake with ICH (Table S1, Supplemental Digital Content, https://links.lww.com/MD/P860). This lack of heterogeneity, along with the absence of significant pleiotropy in the correlation analysis between dietary factors and ICH (all P > .05, Table S2, Supplemental Digital Content, https://links.lww.com/MD/P860), enhances the credibility of our findings. Furthermore, the leave-one-out analysis demonstrated that excluding each SNP individually had no significant impact on the robustness of our results (Fig. 4). All sensitivity analyses supported our findings.

Figure 4.

The results of leave-one-out sensitivity analysis on ICH. (A) Hot drink temperature. (B) Fresh fruit intake. (C) Cooked vegetable intake. ICH = intracerebral hemorrhage.

4. Discussion

ICH is a severe form of brain disease, associated with a relatively low survival rate.^[29] Survivors of ICH often face long-term functional decline and have a higher risk of developing dementia.^[30,31] As is widely recognized, dietary therapy plays a crucial role in disease prevention.^[32,33] Understanding the causal relationship between dietary factors and ICH, and subsequently adjusting dietary habits, may effectively prevent and intervene in the occurrence of ICH. MR analysis has the advantage of being able to effectively exclude the effects of reverse causality and confounding factors.^[34] Therefore, in this study, we employed MR analysis to investigate the causal relationships between 35 dietary factors and ICH. After obtaining our research results, we conducted sensitivity analyses to validate their reliability. Our findings suggested that hot drink temperature and fresh fruit intake may be potential dietary factors associated with a reduced risk of ICH, while cooked vegetable intake may be linked to an increased risk. Additionally, no significant associations were found between other dietary factors and ICH.

Quinlan et al have demonstrated that hot drink intake can rapidly increase skin conductance and temperature, potentially promoting vasodilation and enhancing mood.^[35] This mood enhancement and anxiety reduction may help mitigate stress-induced blood pressure fluctuations, thereby indirectly reducing the risk of ICH. Similarly, fresh fruits also play a unique role in health promotion. He et al have confirmed that increasing the intake of fresh fruits can reduce the risk of stroke.^[36] Our results further indicate that fresh fruit intake can decrease the likelihood of ICH. The dietary fiber abundant in fresh fruits supports intestinal health and guards against constipation, thereby avoiding the potential risk of ICH from straining during bowel movements.^[37,38] Moreover, regular intake of fresh fruits promotes long-term weight control and diminishes the incidence of cardiovascular diseases, type 2 diabetes, and metabolic syndrome, all of which are potential risk factors for ICH.^[37] In addition, fresh fruits are a rich source of vitamins and minerals. These micronutrients are essential for critical physiological processes including energy-yielding metabolism, DNA synthesis, and oxygen transport, underscoring their importance for brain health.^[39] As time passes, the nutritional value of fruits, such as vitamins and minerals, may gradually decrease, and prolonged storage can also degrade the taste, and even potentially foster mold growth.^[40,41] Therefore, prioritizing the intake of fresh fruits is essential for preventing ICH.

It is well-known that vegetables are rich in a variety of nutrients and greatly beneficial to both physical and mental health.^[42,43] Nevertheless, in our study, a suggestive increase in ICH risk was observed for cooked vegetable intake. This could be attributed to the changes in the nutritional content and chemical composition of vegetables during the cooking process.^[44,45] Previous studies also support this viewpoint. For instance, Feng et al have indicated that a higher intake of raw rather than cooked vegetables is associated with a lower risk of cardiovascular diseases.^[46] Our study results serve as a reminder that not all vegetable intake is beneficial to health. However, the association between cooked vegetable intake and higher ICH risk should be interpreted with caution due to the wide confidence interval (1.107–18.246). Further studies with larger sample sizes are needed to confirm this potential link.

Moreover, it is important to note that our study sample was predominantly European, and previous research has indicated that the incidence of ICH varies across different regions.^[2] This implies that our findings may not be fully applicable to other areas, thus our research results have certain limitations in terms of universality. Additionally, the use of a secondary, less stringent threshold (P < 5.0 × 10^‐6) for selecting IVs may have increased the risk of including weak or false-positive instruments. In our study, we mitigated this risk by requiring all IVs to have F-statistics > 10 and performing a series of sensitivity analyses. However, residual bias cannot be completely ruled out.

5. Conclusion

Our study indicates that hot drink temperature and fresh fruit intake may act as potential protective factors against ICH risk. Additionally, a suggestively increased risk of ICH was observed in association with cooked vegetable intake. However, the wide confidence interval reflects imprecision in the estimate, highlighting the need for further investigation to clarify this association.

Author contributions

Conceptualization: Peng Wang, Dashuai Qiao.

Formal analysis: Fangwen Liu, Jinfeng Li.

Investigation: Dashuai Qiao.

Methodology: Peng Wang, Dashuai Qiao.

Project administration: Gang Xue.

Resources: Fanzhen Dong.

Writing – original draft: Peng Wang, Dashuai Qiao, Gang Xue.

Writing – review & editing: Peng Wang, Dashuai Qiao, Fangwen Liu, Fanzhen Dong, Jinfeng Li, Gang Xue.