The causal effect of serum 25‐hydroxyvitamin D levels on epilepsy: A two‐sample Mendelian randomization study

Abstract

Objective

Observational studies have shown an association between 25‐hydroxyvitamin D (25 (OH) D) and epilepsy, but it is unclear whether the association is causal. Therefore, we applied Mendelian randomization (MR) analysis to determine the causal relationship between serum 25 (OH) D levels and epilepsy.

Methods

We conducted a two‐sample Mendelian randomization (TSMR) study to investigate the association between serum 25 (OH) D levels and epilepsy using pooled statistics from genome‐wide association studies (GWAS). Data for 25 (OH) D came from a GWAS comprising 417,580 participants, and data for epilepsy were obtained from the International League Against Epilepsy (ILAE) consortium. Five methods were used to analyze TSMR, including the inverse variance weighting method, MR Egger method, weighted median method, simple model, and weighted model. In the sensitivity analysis, MR Egger and MR PRESSO methods were used to test for pleiotropy, inverse variance weighting and MR Egger in Cochran's Q statistics were used to test for heterogeneity.

Results

MR analyzed the relationship between 25 (OH) D and different types of epilepsy, and the results showed that a 1 standard deviation increase in natural log‐transformed serum 25 (OH) D levels was associated with reduced risk for juvenile absence epilepsy (IVW OR = 0.985; 95% confidence interval [CI]: 0.971–0.999; P‐value = 0.038). There was no apparent heterogeneity and horizontal gene pleiotropy.

Significance

Higher serum levels of 25 (OH) D were a protective factor for adolescent absence epilepsy, but had no effect on other types of epilepsy.

Key points.

• Mendelian randomization was used to investigate the association between serum 25 (OH) D levels and epilepsy.

• Serum 25 (OH) D level is associated with adolescent absence epilepsy, but not with other epilepsy.

• Higher serum levels of 25 (OH) D were a protective factor for adolescent absence epilepsy.

1. INTRODUCTION

Epilepsy is a common neurological disease characterized by sudden abnormal discharge of the brain, resulting in transient brain dysfunction. ¹ It is estimated that the incidence of epilepsy is about 50 per 100 000 people, and more than 70‐million people worldwide suffer from epilepsy. ² The treatment of epilepsy and seizures not only bring economic burden but also seriously affects the quality of life. ³

Vitamin D is a group of fat‐soluble compounds whose deficiency is associated with multiple health outcomes. ⁴ 25 (OH) D concentrations are routinely measured in humans to determine individual vitamin D status. ⁵ At present, observational studies and meta‐analyses have shown that taking antiseizure medication can lead to a reduction in 25 (OH) D levels, ⁶ , ⁷ but the relationship between serum 25 (OH) D levels and the risk of epilepsy is not clear. Observational studies have shown inconsistent results between 25 (OH) D supplementation and seizure risk. ⁸ , ⁹ The reason for the different results is that observational studies often have different confounding or reverse causality, and due to the limitations of traditional statistical methods, potential confounders or reverse causality cannot be ruled out, and the observed associations cannot be well established as causal. ¹⁰

These limitations can be addressed by applying MR methods. ¹¹ , ¹³ Because alleles are randomly assigned at conception according to Mendelian's second law, MR removes systematic bias by selecting genetic variants associated with exposure as instrumental variables (IVs), so its effect is similar to that of a randomized controlled trial (RCT). ¹² Confounding factors in the population can be overcome by the MR method. In this study, a TSMR analysis using GWAS summary statistics was performed to assess the causal relationship between 25 (OH) D and epilepsy.

2. MATERIALS AND METHODS

2.1. Study design and data sources

In this study, the TSMR method was used to explore the effects of serum 25 (OH) D levels on epilepsy.

Data on exposure variables for genetic variants associated with 25 (OH) D were obtained from the GWAS meta‐analysis that included 417 580 individuals of European ancestry. ¹⁴ The sources of data were obtained from the European UKB participants.

The outcome variable data for genetic variants associated with epilepsy were derived from the GWAS meta‐analysis that included 15,212 cases and 29,677 controls. ¹⁵ The sources of data were obtained from the International League Against Epilepsy Consortium on Complex Epilepsies (ILAE), including all epilepsy (15,212 cases and 29,677 controls), hereditary generalized epilepsy (3,769 cases), focal epilepsy (9,671 cases), focal epilepsy (documented lesion negative, 2,716 cases), juvenile absence epilepsy (415 cases), childhood absence epilepsy (793 cases), focal epilepsy (documented hippocampal sclerosis, 803 cases), focal epilepsy (documented lesion other than hippocampal sclerosis, 3,070 cases), generalized epilepsy with tonic–clonic seizures (228 cases), and juvenile myoclonic epilepsy (1,181 cases).

2.2. Instrumental variable (IV) selection

The extracted genetic variants were selected as IVs to estimate the causal effect of serum 25 (OH) D levels on epilepsy risk based on the following MR correlation hypotheses: (1) being predictive of 25 (OH) D, (2) being independent of confounders, (3) results were not altered by independent pathways other than25 (OH) D. ¹⁶ We first screened SNPs associated with 25 (OH) D at a genome‐wide significance (P < 5 × 10⁻⁸), and the linkage disequilibrium (LD) (r ² < 0.001, distance <1000 kb) is removed. The F‐statistic is used to exclude weak tool biases that violate MR's first hypothesis, SNPs with instrument strengths (F) larger than 10 were selected. ¹⁷ The remaining SNPs were aggregated into the outcomes GWAS database, palindromic SNPs with intermediate allele frequencies were removed and delete outliers SNPs by the MR pleiotropy residual and outlier (MR PRESSO) test. ¹⁸ We did not use proxy SNPs. Following the above principles, finally, multiple independent SNPs strongly associated with each exposure trait were selected as IVs.

2.3. Statistical analysis

In this TSMR study, inverse variance weighting (IVW), weighted median (WM), MR Egger regression, simple mode, and weighted mode methods were used to analyze the causal association of 25 (OH) D with epilepsy. ¹⁸ , ¹⁹ We used the IVW method as the primary analysis to assess the causal relationship. The IVW method combines Wald estimates for each SNP using a meta‐analysis approach to obtain an overall estimate and then performed a weighted linear regression with a forced intercept of zero. When IVs satisfy these three basic assumptions, it achieves higher estimation accuracy and test power to obtain an overall estimate of the effect. ²⁰ To test the stability of our results and avoid unmeasured confounding factors and unknown interference, we performed the MR Egger regression, WM, simple mode, and weighted mode methods for estimation. MR Egger regression was used to assess whether pleiotropy was present with an intercept term based on the assumption that instrument strength was not related to the direct effect. ²¹ If the intercept term is zero, it indicates that horizontal pleiotropy does not exist. WM provides consistent estimates of causality even when as many as 50% of variants in the gene are null. ²²

MR PRESSO methods and MR Egger intercept were used to test for pleiotropy of IVs in the outcomes GWAS dataset. ¹⁸ In these GWAS data, if the pleiotropy test results P‐values >0.05 indicates no significant pleiotropy of genetic IVs. MR Egger and IVW in Cochran's Q statistics were used to test the heterogeneity of IVs in the GWAS dataset of outcomes. ²³ The heterogeneity test result P‐value >0.05 represents no heterogeneity. Scatter plot was used to test the causal effect of the individual hypothesized.

The F‐statistic was calculated using the formula to assess the intensity of the IVs F = β²/SE², where β represents the β value of the exposed SNPs, and SE represents the SE value of the exposed SNPs. ²⁴ If the corresponding F‐statistic is P > 10, no significant weak instrument bias was considered.

All statistical analyses were performed using R version 4.2.1 (R Foundation for Statistical Computing). MR analyses were performed using the two‐sample MR (version 0.5.6) ²⁵ and MRPRESSO (version 1.0). ¹⁸

3. RESULTS

SNPs were used as IVs according to the selection criteria of IVs. No palindrome SNPs were found, and only juvenile myoclonic epilepsy had an outlier SNP. The F‐statistic for each IV was greater than 10, indicating the absence of bias from weak IVs. Details of selected instrumental variables and outcome data are provided in Table S1.

As shown in Figure 1, based on three MR analysis methods (IVW, MR Egger, and WM), We found that 25 (OH) D was associated with a lower juvenile absence epilepsy risk (IVW OR = 0.985; 95% CI: 0.971–0.999; P‐value = 0.038). Based on the IVW method, there was no causal relationship between 25 (OH) D and other types of epilepsy. In the sensitivity analysis, except for generalized epilepsy, MR Egger intercept and MR PRESSO methods suggested no significant pleiotropy, and the result of Cochran's IVW and MR Egger Q test showed no heterogeneity (Table S2).

FIGURE 1.

Forest plot of the genetic causal relationship between 25 (OH) D and epilepsy. (CI, confidence interval; No, number; OR, odds ratios; SNP, single‐nucleotide polymorphism).

In view of the causal relationship between 25 (OH) D and juvenile absence epilepsy, MR leave‐one‐out sensitivity analysis suggested that removing a specific SNP of the 25 (OH) D SNPs did not change the results, and no single SNP had a significant effect on the results of MR estimation (Figure S1). Scatter points in the Funnel plot indicate that using a single SNP as instrumental variable estimation is symmetric, vertical lines indicate that the population estimates obtained from the IVW estimation and MR Egger regression are both on the same side. The funnel plot shows robust MR results (Figure S2). Altogether, these results indicate that our data were robust without obvious bias.

4. DISCUSSION

This MR Study shows that genetic liability for higher levels of 25 (OH) D is causologically associated with reduced risk for juvenile absence epilepsy but not for other types of epilepsy. To the best of our knowledge, the present study is the first to illustrate the causal relationship between 25 (OH) D and different types of epilepsy using a Mendelian randomization study.

Observational studies have shown that the risk of childhood epilepsy increases with increasing neonatal 25 (OH) D levels, ²⁶ but there are animal studies have shown that 25 (OH) D can inhibit drug‐induced seizures in mice. ²⁷ Most observational studies have shown that 25 (OH) D level is decreased by taking antiepileptic drugs, and 25 (OH) D supplementation can reduce seizure frequency .  ⁶ , ²⁸ , ²⁹ However, some studies have shown that 25 (OH) D supplementation does not reduce seizure frequency in patients with drug‐resistant epilepsy. ⁸ As can be seen from the above, there is a contradiction in whether 25 (OH) D prevents epilepsy or controls seizure frequency in observational studies.

Now much of the research focuses on antiepileptic drugs caused by bone health, ³⁰ , ³¹ or considering the influence of 25 (OH) D supplementation for other health outcomes. ³² Little is known about whether 25 (OH) D can prevent epilepsy and seizures. Taking into account, the effect of taking antiepileptic drugs on 25 (OH) D level and the effect of 25 (OH) D on bone health, combined with the results of this MR Study, 25 (OH) D supplementation in adolescents is beneficial. However, more randomized controlled trials are needed to confirm this.

This study has several strengths. One strength of our study is that MR analysis was used to determine the causal relationship between 25 (OH) D and epilepsy, thus allowing for the exclusion of confounding factors. Secondly, the use of non‐overlapping exposure and outcome summary level data in two‐sample MR could avoid bias. ³³ Second, this is the first study to examine the relationship between vitamin D and different types of epilepsy. In addition, no significant pleiotropy or heterogeneity of overweight and obesity genetic IVs was shown by MR PRESSO, MR Egger, and IVW analysis methods.

However, this study has several limitations. First, we could not perform subgroup analyses because we used summary statistics rather than raw data in the analysis. Second, our findings were limited to individuals of European ancestry. Our findings need to be replicated to different races for verification in the future.

5. CONCLUSION

Higher serum levels of 25 (OH) D were a protective factor for adolescent absence epilepsy but had no effect on other types of epilepsy.

AUTHOR CONTRIBUTIONS

Ling Liu and Zhichao Ruan conceived, initiated, and supervised the project. Xinxin Luo collected and analyzed the data and wrote a draft of the manuscript. The authors read and approved the final manuscript.

FUNDING INFORMATION

This work was supported by the science and technology program of Jiangxi Provincial Administration of Traditional Chinese Medicine (rant No:2021A378).

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

ETHICS STATEMENT

This study was conducted using published studies and consortia that provided publicly available summary statistics. All original studies have been approved by the appropriate ethical review boards, and participants have provided informed consent. In addition, this study did not use individual‐level data. Therefore, no new ethics review board approval was required.

CONSENT FOR PUBLICATION

Not applicable.

Supporting information

Tables S1–S2:

Figures S1–S2:

Luo X, Ruan Z, Liu L. The causal effect of serum 25‐hydroxyvitamin D levels on epilepsy: A two‐sample Mendelian randomization study. Epilepsia Open. 2023;8:912–917. 10.1002/epi4.12758

DATA AVAILABILITY STATEMENT

The published article at https://www.nature.com/articles/s41467‐018‐07524‐z and https://www.nature.com/articles/s41467‐020‐15421‐7. These data were derived from the IEU OpenGWAS Project at https://gwas.mrcieu.ac.uk/.