Assessing the Causality Between Frailty and Epilepsy: Bidirectional 2-Sample Mendelian Randomization

Epilepsy, a complex chronic neurologic condition, is marked by the sporadic interruption of cerebral function due to the abnormal synchronization of neuronal discharges, which severely impacts an individual’s overall well-being and daily functioning.¹ Recognized as one of the prevalent severe neurologic afflictions, it is estimated that, annually, nearly 5 million individuals receive an epilepsy diagnosis with a current global prevalence affecting approximately 50 million people, representing 0.5% of the global disease burden. This condition exerts considerable demands on health care systems, contributes to premature deaths, and handicaps work productivity.² The intricate pathophysiologic mechanisms underlying epilepsy remain elusive with a spectrum of diseases and habits, including cerebral infections, neoplasms, stress, alcohol abuse, and smoking, identified as potential risk factors for the onset of epilepsy.^3–5

Frailty also stands as a public health challenge. It is identified by heightened vulnerability to stress and undermined system functionality.⁶ Frailty leads to a host of negative outcomes, such as lower life quality and increased morbidity and mortality, a widespread condition that naturally comes with aging.⁷ Evidence from epidemiologic studies suggests that frailty is a precursor to chronic ailments and precipitates a cascade of adverse health ramifications, encompassing cardiovascular disorders, delirium, hospital admissions, and fatalities.^8–11 Research on the nexus between frailty and epilepsy has pointed to a pronounced relationship between the frailty status of geriatric epilepsy sufferers and their antiepileptic drug responsiveness.¹² Another study utilizing a United Kingdom primary care database has correlated the exacerbation of dementia and frailty to an elevated mortality rate among patients with late-onset epilepsy.¹³ Nonetheless, the intrinsic causality between frailty and epilepsy has yet to be explored on a broad population scale. Pinpointing this link is vital. Early identification of risk factors for epilepsy can better prevent the development of the disease, guide intervention strategies, and mitigate the escalating health crisis.

The frailty index (FI), a composite metric for gauging the impact of aging on health, compares health deficits to a comprehensive set of measured variables.¹³ It operates independently of any particular variables, amalgamating a multitude of health indicators into a singular index via a basic mathematic model. This index bespeaks unpredictability of environmental factors as well as the ability to recuperate from illnesses, thereby emerging as an acutely sensitive gauge of frailty.¹⁴ Mendelian randomization (MR), recognized for its strength in epidemiologic research, serves as a potent instrument to determine the causal links between exposures and health outcomes. MR minimizes the confounding effects of environmental factors by relying on the random distribution of genetic variations at the moment of conception, and it precludes the issue of reverse causation because an individual’s genotype is not dictated by diseases.¹⁵ The purpose of our study was to use MR analysis to probe into the causality between FI and epilepsy and to investigate, through reverse causation analysis, the potential influence of epilepsy on the incidence and development of frailty. Although our findings may not bring any groundbreaking changes to the existing knowledge of epilepsy’s etiology, they pave the way for future inquiries in this domain.

Methods

Research design

A 2-sample MR analysis was performed to assess the causality between FI and epilepsy with the schematic flow of the study design outlined in Fig. 1. FI was set as the exposure and epilepsy as the outcome. In a reverse scenario, epilepsy was chosen as the exposure and FI as the outcome. Utilizing independent genetic variants as instrumental variables (IVs), the MR analyses were grounded in 3 basic postulates: (1) a tight correlation between IVs and the exposure is necessary, (2) IVs must remain uncorrelated with confounders, (3) the influence of IVs on the outcome should be mediated exclusively through the exposure.¹⁶ To avoid bias due to population heterogeneity, the main body of the genome-wide association studies (GWAS) data we selected comes from individuals of European descent with a minor representation from other ancestries. These populations exhibit a high degree of independence.

View Figure

• Download as Powerpoint
• Download Figure

The data for this study were all derived from research that has been published or from GWAS summary statistics. These summary statistics, open to the public, have been ethically approved and are recorded in the initial GWAS publications. Approval from the institutional review boards was secured for each study we analyzed, and all participants have supplied written informed consent.

Data source

FI data

FI is widely accepted as a standard for frailty assessment. Our MR analysis included GWAS data related to FI from the United Kingdom Biobank and Twin Gene.¹⁷ All data for this study are publicly accessible and can be retrieved from the website http://ftp.ebi.ac.uk/pub/databases/gwas/summary_statistics/GCST90020053/. A total of 175,226 subjects with European ancestry were included with 164,610 from the United Kingdom Biobank and 10,616 from Twin Gene.

Epilepsy data

The International League Against Epilepsy provided the GWAS summary data for epilepsy, covering various epilepsy phenotypes, including all epilepsy, generalized seizures, focal seizures, childhood absence epilepsy, juvenile absence epilepsy, juvenile myoclonic epilepsy, tonic-clonic seizures in generalized epilepsy, hippocampal sclerosis in focal epilepsy, and lesional negative focal epilepsy.¹⁸ The data set comprises 15,212 epilepsy cases and 29,677 healthy controls with the majority being of European descent and a smaller number of African and Asian individuals (Caucasian: 38,752, Asian: 3,406, African: 2,731). Additional information is detailed in Table 1.

Table 1 Detailed information of the genome-wide association studies in our analysis

View Fullscreen

Results

Causal effect of FI on the risk of epilepsy

In Table 2, the data show a pronounced link between the FI and epilepsy. IVW analysis indicates a positive correlation between genetically forecasted FI and epilepsy (OR = 1.2126, 95%CI: 1.0143-1.4497, P = 0.0343), and the Weighted Median method detected a congruent causal effect [odds ratio (OR) = 1.2619, 95% confidence interval (CI): 1.0387–1.5330, P = 0.0192]. The detailed MR forest plot is provided in Appendix 1.

Table 2 Results of causal association of 5 methods MR regression

View Fullscreen

Sensitivity analyses

A suite of sensitivity analyses was executed, including the MR Egger intercept, funnel plot examination, MR-PRESSO detection, Cochran Q test, and exclusion analyses, to verify the precision of the positive outcomes. The results (Table 3) showed P < 0.05 for both the IVW and the MR Egger intercept, suggesting heterogeneity between the SNPs. Nevertheless, the combined results of the MR Egger intercept and the MR-PRESSO global tests indicated that the MR analysis was unaffected by any potential horizontal pleiotropy (P > 0.05) (Appendix 1). Considering the presence of outliers and pleiotropy in the MR-PRESSO test (Table 3), we iterated these analyses after the removal of the outlier SNPs identified by the MR-PRESSO test (rs11245450 and rs755249). More information can be found in Appendix 1. The leave-one-out test revealed SNPs that potentially affected the causal association (Supplemental Figure S1). Yet, as no other statistical methods showed any abnormalities, the impact of the outlier SNPs on the results was not significant. These analyses affirmed the robustness and reliability of our MR findings.

Table 3 Tests for heterogeneity and pleiotropy

View Fullscreen

Results for iterated MR

After the elimination of outlier SNPs, the aforementioned analyses were reiterated. The IVW method was utilized for causal analysis, which supported a positive correlation between FI and epilepsy, identifying that the epilepsy risk in frail populations is 1.1862 times greater than in those who are not frail (95% CI: 1.0236–1.3746, P = 0.0232) (Table 4). The closeness of the IVW estimate to that of the weighted median (OR = 1.2622, 95% CI: 1.0309–1.5455, P = 0.0242) demonstrates the reliability of the FI-epilepsy effect estimation. The MR forest plot is presented in Appendix 2.

Table 4 Outlier SNPs were removed to replicate Mendelian randomization results

View Fullscreen

The Cochran Q test showed no significant heterogeneity (P > 0.05), and both the MR Egger regression intercept test and the MR-PRESSO test did not suggest horizontal pleiotropy (P > 0.05) (Table 5). Scatterplots and funnel plots illustrate a symmetric distribution of all the SNPs included, implying that the causal association is not likely to be biased (Appendix 2). Sensitivity analysis confirmed the reliability of the MR results in this study (Figure S2).

Table 5 Tests of MR heterogeneity and pleiotropy after removal of outlier SNPs

View Fullscreen

Results for reverse MR

IVs

Eighteen SNPs were ascertained as the ultimate set of IVs, all demonstrating remarkable relevance to the exposure. With epilepsy as the exposure and FI as the outcome, all F-statistics surpassed 10. Further details are provided in Supplemental Table S2.

Causal effect of epilepsy on the risk of FI

Table 6 demonstrates evidence of a reverse causality. The IVW model indicates that epilepsy raised the risk of FI (OR = 1.0896, 95% CI: 1.0242–1.1592, P = 0.0066), a finding that was supported by the weighted median model (OR = 1.0955, 95% CI: 1.0326–1.1621, P = 0.0025). The Cochran Q test pointed to heterogeneity among the included SNPs in the outcomes (P < 0.05) (Appendix 3). The MR Egger intercept test confirmed that the MR analysis is not biased by any potential horizontal pleiotropy (P > 0.05), notwithstanding the presence of outliers and pleiotropy in the MR-PRESSO test for MR (Appendix 3). The leave-one-out test found no evident impact of SNPs on the causal estimate, suggesting robustness in the MR outcomes (Supplemental Figure S3).

Table 6 Causal association results from 5 methods of reverse MR regression

View Fullscreen

To ensure the veracity of the results, the analysis was repeated after excluding the outlier SNPs identified by the MR-PRESSO test (rs4638568 and rs61779328). Table 7 shows the causality between epilepsy and FI following the exclusion of the outlier SNPs. The IVW model pointed to a statistically significant difference (OR = 1.0975, 95% CI: 1.0532–1.1436, P = 9.69e-06), which was confirmed in the weighted median and MR Egger models (OR = 1.0970, 95% CI: 1.0363–1.1612, P = 0.0014; OR = 1.1670, 95% CI: 1.0153–1.3414, P = 0.0473). Epilepsy is recognized as a risk factor for frailty. Sensitivity analysis showed no evidence of heterogeneity or horizontal pleiotropy (Table 8), and no outlier SNPs were noted in other assessments (Appendix 4, Supplemental Figure S4), demonstrating the robustness of the MR results.

Table 7 Causal association results of reverse MR regression by 5 methods after removing outlier SNPs

View Fullscreen

Table 8 Tests of reverse MR heterogeneity and pleiotropy after removal of abnormal SNPs

View Fullscreen

Discussion

For this study, we leveraged GWAS summary data to determine the causality between frailty and epilepsy. The bidirectional MR analysis provided preliminary evidence of a positive causal relationship between FI and epilepsy, suggesting that frailty could be a contributor to the risk of epilepsy. Inverse MR outcomes suggested that epilepsy patients suffer an elevated risk of FI, rendering epilepsy a risk factor for frailty. Our research is the first MR study to explore the bidirectional causality between epilepsy and frailty, aiding in the genetic contextualization of epilepsy risk and frailty.

Epilepsy, a common and formidable chronic neurologic disorder with heterogeneous causes, has become a focal point of public concern. Research has established that oxidative stress is a key characteristic of the pathophysiologic mechanisms behind epilepsy, integral to the harmful effects of epileptic activity in the brain and associated with the advancement of neuronal damage.²² In the epilepsy patient population, oxidative stress can inflict functional cellular damage, creating a surge of reactive oxygen species (ROS) that directly catalyze the transition of mitochondrial permeability, resulting in compromised mitochondrial membrane integrity, increased permeability, reduced membrane potential, mitochondrial dysfunction, and neuronal cell death.^23,24 Frailty, a dynamic clinical syndrome, is defined by a diminished capacity for physiologic reserve to sustain homeostasis.⁷ Individuals identified as frail are facing greater risk of adverse health outcomes compared with the nonfrail population.²⁵ Frailty is marked by pronounced deficiencies in the innate immune function, T-cell activity, and antibody production in conjunction with an upregulation of oxidative stress biomarkers and a diminished antioxidant capacity.²⁶ This complex interplay culminates in the buildup of proinflammatory cytokines and the creation of byproducts associated with oxidative stress.^27–29 A study examining the effects of frailty and dementia on elderly epilepsy patients’ mortality rates has shown that increased frailty is significantly linked to a higher risk of death.¹³ A plethora of animal model studies on epilepsy now support the role of antioxidants in easing the condition by combating the adverse effects of an overabundance of ROS in mitochondria and preventing cell death.^30–32 Antioxidants, therefore, are emerging as a potential neuroprotective treatment for epilepsy.³³

Individuals with epilepsy may undergo accelerated frailty when compared with their peers. A study focusing on the incidence of hospitalization for chronic conditions in young people has demonstrated that individuals with epilepsy are at an elevated risk of hospital admission relative to their peers (absolute risk reduction: 10.95, 95% CI: 9.98–12.02).³⁴ Our research has likewise identified a notable positive association between epilepsy and FI, marking epilepsy as a determinant of frailty. As previously stated, we conjectured that oxidative stress is pivotal in the pathophysiologic mechanisms linking frailty to epilepsy, prompting the emergence of great amounts of proinflammatory cytokines and ROS. It is documented that the continuous state of epilepsy, fueled by heightened ROS production, may lead to the accumulation of peroxidation products.²⁴ Epileptic seizures elicit an inflammatory response that activates inducible nitric oxide synthase and NADPH oxidase 2, leading to the production of reactive nitrogen species and ROS. These reactive molecules contribute to the severity of oxidative stress during seizures and, in conjunction with elevated levels of proinflammatory cytokines, facilitate the progression of the disease.^35,36 The rise in pro-inflammatory cytokines can affect the frail state either directly by enhancing protein breakdown or indirectly by influencing important metabolic pathways in the brain.^37–39 We speculate that this may be the reason for the increased frailty associated with epilepsy. Nonetheless, the esoteric mechanisms linking frailty and epilepsy necessitate further experimental confirmation.

This investigation is fortified by several key advantages. MR is adept at discerning the causality between FI and epilepsy while eliminating the impact of confounding factors. Furthermore, the use of nonoverlapping summary data for exposure and outcome in 2-sample MR effectively avoids bias. The study is enhanced by the deployment of large-scale GWAS summary data and sensitivity analyses, which substantiate the robustness of our results. To sum up, the employment of genetic risk factors derived from GWAS, combined with the MR design and IVW statistical methodology, has reinforced the validity of our research findings, delivering valuable perspectives on the link between FI and epilepsy.

Despite its strengths, our study encounters some limitations. We could not perform subgroup analyses as we relied on summary data rather than the original data. The populations screened for SNPs associated with FI and epilepsy were primarily of European descent, indicating a need for further research in other ethnic groups to confirm the generalizability of our results. The impact of gender on epilepsy is notable with males showing a higher rate of prevalence.⁴⁰ Yet the 2-sample design of our study did not differentiate by gender; thus, the dissection of the relationship between FI and epilepsy in different genders is not achievable. Finally, whereas the positive bidirectional MR causal relationships between frailty and epilepsy suggest shared elements or mediators, a comorbidity analysis was beyond the scope of this study.

Conclusion

The MR analysis we conducted has brought to light a pronounced bidirectional causal relationship between frailty and epilepsy with both conditions influencing each other as risk factors. This revelation holds value for the early identification of risk, intervention, and therapeutic intervention for mental health issues, such as frailty and epilepsy. Future studies should investigate the underlying mechanisms of this causal relationship, aimed at crafting preventive and therapeutic strategies that may reshape the dynamic between frailty and epilepsy.