Менделианская случайность не охватывает потенциальные лекарственные средства, которые могут быть восстановлены при нейропсихических расстройствах.

INTRODUCTION

Neuropsychiatric disorders involve multiple pathological mechanisms, including neuronal damage, loss of function, loss of synaptic connectivity, and cell death. With the increasing complexity of environmental factors, rising social pressures, and a growing aging population, the prevalence of neuropsychiatric disorders, such as Alzheimer's disease (AD), Parkinson's disease (PD), and autism spectrum disorder, has been steadily increasing. As of 2021, approximately 3.4 billion individuals have been affected by neurological issues, attributed to 11.1 million deaths. These disorders are the leading causes of disability-adjusted life years (DALYs) and years of life lost (YLL), imposing a significant burden on global health [ 1 ]. In the United States, over one in nine individuals aged 65 years (10.9%) were diagnosed with AD. It is predicted that the number of cases will reach 13.8 million by 2060, representing over 48% of the elderly population aged 65 and above, which will not only bring a huge burden to the patient’s family but also have a serious impact on the economy [ 2 ]. In 2019, the global economic cost of AD was estimated at $1.3 trillion, and it is expected to be $2.8 trillion by 2030 [ 3 ].

Pharmacotherapy is effective in treating neuropsychiatric disorders, but it faces several challenges. Symptom overlap increases the complexity of diagnosis and treatment, leading to difficulties in drug use. Some medications can cause severe side effects resulting in poor patient tolerance. Additionally, developing new drugs is a lengthy process that requires substantial financial investments. Prescribing medications that are already on the market allows healthcare providers to take advantage of their well-established safety profiles and known side effects, which have been thoroughly assessed through preclinical studies, clinical trials, and long-term follow-up. This practice, known as drug repurposing, involves exploring new uses for these established drugs and has the potential to lead to effective treatments with predictable side effects, ultimately helping to reduce healthcare costs [ 4 - 6 ]. For instance, colchicine, traditionally used for gout, is currently being studied for its ability to reduce inflammation related to COVID-19 [ 7 , 8 ]. Similarly, metformin, which is primarily prescribed for diabetes, is now being investigated for its possible anti-cancer effects [ 9 ]. Thalidomide, once infamous for causing birth defects, has been repurposed for the treatment of multiple myeloma [ 10 ]. Additionally, auranofin, originally developed as an anti-rheumatic drug, is being researched for its effectiveness against gastrointestinal stromal tumors [ 11 ]. These examples highlight the significant benefits of drug repurposing for broadening treatment options and improving the efficiency of healthcare systems.

Mendelian Randomization (MR) is a statistical approach based on Mendel’s laws of random inheritance used to infer causal associations between exposure and outcome [ 12 ]. This approach effectively avoids reverse causation and provides evidence comparable to that of randomized controlled trials [ 13 ]. Previously, Zhang et al. utilized this methodology to identify cannabidiol, doxorubicin, genistein, and propylthiouracil as potential treatments for manic episodes and type 1 diabetes [ 14 ]. Duan et al. found that fostamatinib, amlexanox, BIIB-023, RG-7212, and glutathione may be effective in treating Amyotrophic Lateral Sclerosis (ALS), and cell experiments confirmed that fostamatinib and amlexanox can reduce neuroinflammation in ALS models [ 15 ]. These examples reveal the significant potential of the MR methodology in drug screening and repurposing. Linkage Disequilibrium Score Correlation (LDSC), used to estimate genetic correlations between phenotypes, and Bayesian Colocalization (COLOC), employed to determine whether phenotypes share one genetic basis, both of which provide crucial support to MR results, ensuring that discovered causal associations are more likely to be genuine rather than confounded by other factors [ 16 , 17 ].

542 treatments/medications were collected as exposures from the UK Biobank Neale Lab and 15 neuropsychiatric diseases from the FinnGen database as outcomes for the MR analysis [ 18 , 19 ]. The robustness of the results was ensured through sensitivity tests, and further exploration of genetic associations was conducted using LDSC and COLOC analyses. Ultimately, we identified medications that may exert negative effects on neuropsychiatric diseases, thereby offering new insights into drug repurposing.

METHODS

Study Design

This study employed an MR analysis utilizing exposure data from the UK Biobank Neale Lab Round 2 ( http://www.nealelab.is/uk-biobank/ ), encompassing a variety of conventional treatments and medications. The outcomes of interest consisted of 15 neurological and psychiatric disorders sourced from the FinnGen database. The flowchart of the study design is shown in Fig. ( 1 ). In brief, our MR analysis was based on three key assumptions: (1) Instrumental variants (IVs) were strongly associated with exposure; (2) IVs were independent of confounding factors; and (3) IVs affected the outcome only through exposure, that is, IVs were not directly associated with the outcome. This study focused on the results with statistical significance, which showed negative associations indicative of a therapeutic effect. Sensitivity analyses were used to assess pleiotropy and heterogeneity using leave-one-out tests to confirm potential treatments. The validated candidates were further scrutinized using LDSC and COLOC for genetic association exploration.

The overall design of Mendelian randomization analysis in the present study.

Data Source for Exposure/Outcome

The exposure data were sourced from a comprehensive genome-wide association study (GWAS) within the UK Biobank, relying on interview-derived prescription information. Specifically, the data on medication use included 542 GWAS datasets for treatment and medication across both sexes, encompassing 361,141 unrelated individuals of European ancestry. These datasets cover a diverse range of categories, including cardiovascular, respiratory, digestive, neurological, endocrine, metabolic, skin, immunological, and rheumatic disorders, as well as infections, inflammation, supplements, adjuvant therapies, various symptoms, and nonspecific health issues (Table S1 ). Neuropsychiatric disorder data, serving as the outcome, were derived from 15 GWAS datasets obtained from the FinnGen database. These diseases included attention deficit hyperactivity disorder (ADHD), all anxiety disorders, anorexia nervosa, bipolar affective disorders, dementia, obsessive-compulsive disorder, schizophrenia, AD, epilepsy, migraine, multiple sclerosis (MS), PD, stroke (excluding subarachnoid hemorrhage, SAH), autism, and substance abuse. Detailed information on both datasets is provided in Table 1 .

Table 1 Data sources for treatment/medication as exposure and neuropsychiatric disorders as outcomes.
Traits

Label

Consortium

Participants

Web Source

Genome-wide Significance

Ncase

Ncontrol

Total
Treatment/medication code - UKBB Neale lab 361141 in total, ncase can be found in the source data
https://www.nealelab.is/uk-biobank
5e-8; 5e-6 when IVs<3 ADHD Finngen_R10_F5_ADHD FinnGen 2788 405153 407941
https://www.finngen.fi/en
- All anxiety disorders Finngen_R10_F5_ALLANXIOUS 27664 368054 395718 Anorexia nervosa Finngen_R10_F5_ANOREX 2115 400510 402625 Bipolar affective disorders Finngen_R10_F5_BIPO 7569 359290 366859 Dementia Finngen_R10_F5_DEMENTIA 20338 391843 412181 Obsessive-compulsive disorder Finngen_R10_F5_OCD 2175 368054 370229 Schizophrenia Finngen_R10_F5_SCHZPHR 6708 398386 405094 Alzheimer disease Finngen_R10_G6_ALZHEIMER 10520 401661 412181 Epilepsy Finngen_R10_G6_EPLEPSY 12891 312803 325694 Migraine Finngen_R10_G6_MIGRAINE 5229 367565 372794 Multiple Sclerosis Finngen_R10_G6_MS 2409 408561 410970 Parkinson's disease Finngen_R10_G6_PARKINSON 4681 407500 412181 Stroke, excluding SAH Finngen_R10_I9_STR 27497 371723 399220 Autism Finngen_R10_KRA_PSY_ AUTISM_EXMORE 646 301879 302525 Substance abuse Finngen_R10_KRA_PSY_ SUBSTANCE_EXMORE 25459 301879 327338

Data sources for treatment/medication as exposure and neuropsychiatric disorders as outcomes.

Selection of IVs

A two-sample MR was used to explore the causal relationship between treatments/medications and neuropsychiatric disorders. Following the STROBE-MR checklist and the guidelines of Haycock et al. , we identified IVs that met essential MR assumptions [ 20 , 21 ]. Genome-wide significant SNPs (Table S2 ) were assessed within a 10,000 kb window and an r 2 < 0.001 threshold to handle linkage disequilibrium. Palindromic SNPs, those linked to outcomes at p < 0.01, and SNPs absent in the outcomes were excluded. Confounding factors, specifically hypertension, diabetes, smoking, alcohol consumption, and sleep disorders, were controlled by excluding the related SNPs identified in PhenoScanner V2 [ 22 ]. All the selected SNPs had F -statistics > 10, guaranteeing strong instrumental validity [ 23 ].

MR Analysis and Sensitivity Test

Five MR methods were used in this study [ 24 ]. The primary method employed is the Inverse Variance Weighted (IVW) method. It combines estimates from genetic variants weighted by inverse variances, assuming that they meet the key assumptions for causal effect estimation. To ensure analytical robustness, this study used four other methods as supplementary approaches. The weighted median method is robust when some variants do not fully meet these assumptions. The MR-Egger regression detects horizontal pleiotropy. The simple mode determines the causal effect direction based on the majority of variant effects. The weighted mode considers the effect magnitudes to obtain a more accurate combined effect. Using these methods, we aimed to obtain reliable and comprehensive insights.

A series of sensitivity analyses were conducted to ensure the robustness of the MR results [ 24 ]. For the heterogeneity analysis, Cochran's Q test was utilized to identify SNP effect heterogeneity [ 25 ]. The pleiotropy test, commonly associated with the MR-Egger intercept, was employed to ascertain whether the IVs exerted a direct influence on the outcome through pathways distinct from the exposure [ 26 ]. The threshold for both tests was set at p > 0.05. The leave-one-out method typically involves systematically removing one variable at a time to determine whether a specific SNP has an independent effect on the MR outcomes. In this study, SNPs that were found to influence the results independently were excluded.

LDSC and COLOC Analyses

To complement the MR results, LDSC was used to estimate the genetic correlation between exposure and outcome [ 17 ]. This method helps identify a shared genetic architecture and supports the presence of a causal relationship. The genetic correlation coefficient (Rg) from LDSC indicates the nature of this correlation: Rg < 0 suggests a negative correlation and vice versa . Additionally, COLOC was employed to assess the probability that the same causal variant underlies both traits [ 27 ]. This approach enhances the robustness of our findings by reducing the likelihood of horizontal pleiotropy and provides additional evidence for the proposed causal pathway. COLOC presents five hypotheses: H0 states that there are no shared causal variants; H1 claims that one phenotype has a causal variant while the other does not; H2 asserts that each phenotype has its causal variants; H3 suggests that the causal variants of each phenotype affect one another; and H4 proposes that both phenotypes share a common causal variant. A non-significant posterior probability of H4 (PPH4) supports the exclusive effect of IVs on the outcome via exposure, aligning with MR's exclusion restriction.

Statistical Analysis

Given the exploratory nature of this study, a p -value threshold of < 0.05 was adopted to determine the statistical significance of our MR analyses. For the LDSC and COLOC analyses, a p -value threshold of < 0.05 was used in sensitivity analysis to indicate that heterogeneity and horizontal pleiotropy were not statistically significant, thereby supporting the robustness of the results.

All MR analyses were performed by the TwoSampleMR R package (version 4.3.1, https://mrcieu.github.io/TwoSampleMR/ ). LDSC analysis was performed using the ldscr R package (version 0.1.0, https://github.com/mglev1n/ldscr/ ). COLOC analysis was performed using the Coloc R package (version. 5.2.3; https://github.com/chr1swallace/coloc/ ).

RESULTS

Selection of IVs

To satisfy the relevance assumption, this study screened the IVs using a threshold of 5e-8. For cases with two or fewer IVs, the threshold was adjusted to 5e-6. Among the 542 treatment/medication codes, 450 utilized the 5e-8 threshold, whereas 92 employed the 5e-6 threshold (Table S1 ). Upon validation, the selected IVs exhibited F -statistics exceeding 10, indicating that these instruments possessed adequate strength.

Identification of Treatment/Medication with Negative Causal Associations to Neuropsychiatric Disorders

Using the IVW method, the treatments and medications were identified as associated with a reduced risk of neuropsychiatric disorders. Notably, we observed negative causal associations between pregabalin, bumetanide, and prednisolone and anxiety disorders (beta = -7.28, p = 4.00e-03; beta = -2.24, p = 6.00e-03; beta = -1.74, p = 2.84e-03, respectively). Additionally, vitamin B1 preparations were negatively associated with dementia (beta = -2.47, p = 1.51e-03), creon E/C granules in capsules with epilepsy (beta = -4.99, p = 3.91e-03), pentasa SR 250 mg modified-release tablets with multiple sclerosis (beta = -3.95, p = 3.83e-03), and zolmitriptan with stroke, excluding subarachnoid hemorrhage (beta = -1.61, p = 6.00e-03). Results from other MR methods, including MR-Egger, weighted median, simple mode, and weighted mode, consistently indicated a negative correlation between exposure and outcome. The forest plot is shown in Fig. ( 2 ), and the scatter plot is shown in Fig. ( S1 ).

The forest plot illustrates negative causal associations between medications and neuropsychiatric disorders.

All eligible results underwent sensitivity analysis, which showed that heterogeneity and horizontal pleiotropy were not statistically significant (Table 2 ). Additionally, leave-one-out analysis indicated that no single SNP significantly influenced the results (Fig. S1 ). Ultimately, we identified seven significant pairs.

Table 2 Sensitivity tests for the mendelian randomization analyses.
Id.exposure

Id.outcome

Method

Heterogeneity Test

Pleiotropy Test

Cochran's Q Test

MR_PRESSO Global Test

Q

Q_df

Q_pval

pval

Egger_ Intercept

se

pval
Pregabalin All anxiety disorders MR Egger 15.229 20 0.763 0.782 -0.014 0.046 0.767 IVW 15.319 21 0.807 Bumetanide All anxiety disorders MR Egger 77.865 64 0.114 0.147 -0.005 0.012 0.650 IVW 78.117 65 0.127 Prednisolone All anxiety disorders MR Egger 73.684 75 0.521 0.567 0.001 0.005 0.845 IVW 73.722 76 0.553 Vitamin B 1 preparation Dementia MR Egger 158.703 154 0.381 0.456 0.001 0.009 0.924 IVW 158.712 155 0.402 Creon e/c granules in capsule Eplepsy MR Egger 90.964 109 0.894 0.902 -0.011 0.015 0.452 IVW 91.533 110 0.899 Pentasa sr 250 mg m/r tablet Multiple Sclerosis MR Egger 94.096 95 0.507 0.564 -0.030 0.027 0.281 IVW 95.272 96 0.502 Zolmitriptan Stroke, excluding SAH MR Egger 52.242 67 0.907 0.936 -0.006 0.011 0.572 IVW 52.564 68 0.916

Sensitivity tests for the mendelian randomization analyses.

Genetic Correlation and Colocalization Results

To further assess the genetic associations between treatment/medication and neuropsychiatric disorders, the LDSC method was employed. Except for the genetic correlation between zolmitriptan and stroke (excluding SAH) (Rg = -0.549, p = 0.049), all other pairs showed no statistically significant results. Notably, the genetic correlations between bumetanide and all anxiety disorders, as well as between creon e/c granules and epilepsy, could not be calculated because of negative heritability estimates (Table S3 ). COLOC analyses were conducted to explore whether the selected medications and their corresponding neuropsychiatric disorders shared common genetic components. None of the results reached the threshold for PPH4, suggesting limited confounding factors and reinforcing the exclusionary assumption of MR (Table S3 , Fig. S2 ).

DISCUSSION

In this study, the causal associations were examined between 542 treatments/medications and 15 neuropsychiatric disorders and identified seven medications (pregabalin, bumetanide, prednisolone, vitamin B1 preparation, creon e/c granules, pentasa SR 250 mg m/r tablet, and zolmitriptan) with significant causal relationships to neuropsychiatric disorders. Of these, the first three drugs showed causality with all anxiety disorders, whereas the remaining four drugs were causally linked to dementia, epilepsy, multiple sclerosis, and stroke (excluding SAH). These findings provide evidence for the potential application of these drugs for the prevention and treatment of neuropsychiatric disorders.

Pregabalin, a potential drug for treating neuropsychiatric disorders, has been approved by the Food and Drug Administration for the treatment of diabetic peripheral neuropathy, spinal cord injury, post-herpetic neuralgia, and fibromyalgia [ 28 - 31 ]. Selective Serotonin Reuptake Inhibitors (SSRIs) and Serotonin-Norepinephrine Reuptake Inhibitors (SNRIs) are the first-line medications for anxiety, while pregabalin has a stronger sedative effect than them [ 32 ]. Using this drug may improve sleep disorders in anxiety patients earlier than using SSRIs or SNRIs [ 33 ]. Additionally, pregabalin is not affected by liver metabolism, has less impact on combination medications, and poses a lower risk of adverse reactions, effectively enhancing the safety of drug use [ 34 ]. In a randomized, double-blind, placebo-controlled trial for Generalized Anxiety Disorder (GAD), pregabalin 200 mg three times daily (600 mg/day) showed significant efficacy versus placebo by week 1 with good tolerability [ 35 ]. In a sequential open-label/double-blind study with 153 generalized social anxiety disorder (GSAD) patients, responders to 10-week open-label pregabalin (450 mg/day) were randomized to 26-week continuation (same dose) or placebo. The results further suggest the therapeutic effect of pregabalin on GSAD [ 35 , 36 ]. Pregabalin significantly delayed relapse, maintained symptom improvement, and was well-tolerated, supporting its effectiveness for both short- and long-term treatment of anxiety disorders [ 37 ]. Our MR analysis further extended the potential application of pregabalin to all anxiety disorders. The rationale of this study is further supported by the mechanism of action of pregabalin, which involves binding to the α2δ subunit of voltage-gated calcium channels, leading to reduced calcium influx, decreased neurotransmitter release, and attenuated excitability of the central nervous system [ 38 , 39 ].

Bumetanide, a loop diuretic, has been shown in previous studies to alleviate anxiety symptoms by modulating GABA(A) receptor signaling [ 40 ]. Since loop diuretics are antagonists of NKCC1 and KCC2 cotransporters, Krystal et al. indicated that the cation-chloride cotransport system may be a potential mechanism underlying the anxiolytic effect of bumetanide [ 41 ]. In the Fmr1 knockout mouse model, which exhibits sensory hypersensitivity akin to fragile X syndrome, bumetanide treatment normalizes cortical circuit dysfunction and alleviates anxiety-related behaviors by modulating the activity of layer 2/3 pyramidal neurons and enhancing feedforward inhibition in the somatosensory cortex [ 42 ]. In agreement with previous studies, our MR analysis corroborates the anxiolytic properties of bumetanide, offering further validation and highlighting the utility of our method in drug repurposing.

Prednisolone, a glucocorticoid with an extensive medical history, is primarily utilized to combat inflammation and autoimmune diseases. However, it carries a spectrum of side effects, necessitating a careful assessment of potential risks and benefits under the guidance of a physician. MR analysis suggests that prednisolone may mitigate anxiety disorders. In a study focusing on post-COVID-19 symptoms, the combination of prednisolone with B vitamins was found to alleviate anxiety and fatigue in patients [ 43 ]. Another possibility is that inflammation and autoimmune diseases may trigger anxiety symptoms, and since prednisolone can control inflammation, it might theoretically help alleviate these symptoms. However, considering the potential side effects associated with corticosteroids, the use of prednisolone to treat anxiety may not be ideal [ 44 ]. An exception is the treatment of autoimmune diseases that may co-occur with mental disorders, such as anxiety, where the use of prednisolone to address the primary condition and indirectly alleviate anxiety, as in the case of neuropsychiatric systemic lupus erythematosus, is advantageous [ 45 ].

Bumetanide and prednisolone may serve as adjunctive therapies for specific subgroups, such as treatment-resistant patients or those with comorbidities. However, further clinical studies are required to establish optimal dosing regimens and treatment durations to ensure safety and reduce adverse effects.

Thiamine or vitamin B 1 has been implicated in the mitigation of dementia, particularly in nutritional or chronic alcoholic encephalopathy [ 46 ]. So far, it hasn't exhibited comparable impacts in other dementias such as AD. However, MR study of dementia outcomes from the FinnGen database, which mainly includes Alzheimer's, vascular, and Parkinson's dementias, shows no association with nutritional or chronic alcoholic brain disorders. Vitamin B 1 can counteract oxidative stress, modulate apoptosis and inflammation, sustain energy metabolism, and hinder the accumulation of β-amyloid protein [ 47 - 50 ]. Recent studies suggest that thiamine may exert its protective effects by preventing brain iron overload, activating the Nrf2/ARE pathway, and reversing the hypometabolism of brain glucose [ 51 ]. Additionally, thiamine supplementation and its derivatives have been shown to improve dementia symptoms significantly [ 52 ]. This contradiction means that despite inconsistent clinical research findings, vitamin B 1 might still be related to non-nutritional brain diseases.

Creon e/c granules in capsules is a digestive system medication, while epilepsy is a neurological condition, and there is no current report of a causal relationship between them. The main components of Creon e/c granules in the capsule include pancreatic enzymes, such as protease, lipase, and amylase, which are used to treat pancreatic insufficiency for various reasons [ 53 ]. It serves as a pancreatic enzyme replacement therapy, primarily aiding digestion and absorption. Although the mechanism by which Creon affects epilepsy is unclear, studies suggest that it may involve the gut-brain axis and metabolic regulation [ 54 , 55 ]. In epilepsy patients, disrupted gut microbiota allows abnormal metabolites to cross the intestinal barrier into the bloodstream [ 56 ]. Creon's pancreatic lipase breaks down dietary fiber to produce short-chain fatty acids (SCFAs), which can inhibit abnormal discharges in the temporal lobe via the vagus nerve [ 57 ].

Additionally, its pancreatic protease enhances protein digestion and may indirectly reduce the pro-epileptic factor IL-6. The lipase also improves lipid absorption, increasing vitamin D levels (epilepsy control associated) and regulating branched-chain amino acid metabolism [ 58 , 59 ]. These biological processes provide a theoretical basis for Creon's potential cross-indication effects, but further clinical research is needed to confirm these mechanisms.

Pentasa SR 250 mg m/r tablet, on the other hand, is a slow-release formulation of mesalazine, a 5-aminosalicylic acid drug mainly used for the treatment of non-specific Inflammatory Bowel Disease (IBD). Neurological complications of IBDs are common yet often underdiagnosed, with studies showing a higher incidence of demyelinating diseases such as multiple sclerosis in patients with IBD [ 60 ]. Both IBD and MS are autoimmune diseases, possibly sharing a genetic basis related to immunity. Mesalazine might alleviate multiple sclerosis by inhibiting inflammation, reducing T-cell activation, regulating immune responses, affecting gut microbiota, and providing neuroprotection [ 61 - 63 ]. The effect of mesalazine is also based on the theory of the gut-brain axis, wherein it acts indirectly. Its efficacy, however, requires further research to be fully validated.

Zolmitriptan typically treats acute migraine attacks by activating the 5-HT1B/1D receptors, constricting intracranial blood vessels, and inhibiting inflammatory neuropeptides [ 64 ]. The MR results suggest that zolmitriptan reduces the risk of stroke, and LDSC indicates a negative genetic correlation between the two. However, no clinical research supports these findings. Conversely, zolmitriptan is contraindicated in patients with uncontrolled blood pressure, and because of the presence of 5-HT1B receptors in the coronary arteries and platelets, zolmitriptan may even cause coronary artery spasm and platelet aggregation, increasing the risk of stroke [ 65 , 66 ]. Nevertheless, pain, as a stress response, can significantly increase the risk of cardiovascular events, and zolmitriptan's rapid relief of acute migraine attacks could positively impact the overall cardiovascular health of some high-risk patients by effectively controlling migraines [ 67 ]. Notably, not a single research has confirmed zolmitriptan worsens stroke outcomes. Dosage and timing may be critical in its impact, and these could be explored in future studies.

In this study, the MR analysis has several limitations. The data derived from a European population suggests the need for further research with diverse racial groups to broaden the applicability of our conclusions. The reliance on verbal interviews for medication usage introduces potential subjectivity in quantification, and the limited number of cases results in a low sample size, which may skew the effect estimates. These issues could be addressed by using clinician-verified clinical records and integrating multiple datasets in future research. Additionally, while we controlled for known confounders, unmeasured factors such as medication adherence and environmental influences were not fully addressed; thus, meta-analyses and covariate adjustments could enhance the accuracy of our results [ 13 , 24 ]. Most of our results are corroborated by existing literature, supporting the utility of MR analyses for drug repurposing. However, caution is warranted for results lacking supporting evidence or those that contradict biological principles, such as the purported causal link between zolmitriptan and stroke risk, which necessitates clinical and experimental validation. Furthermore, the potential nonlinearity in medication dosage underscores the need for a stratified analysis with additional data to ensure the validity of our conclusions.

Drug repurposing offers cost-saving potential by bypassing early-stage development, though challenges such as regulatory barriers, post-market monitoring costs, and dose optimization must be addressed. Streamlined approval processes and rigorous clinical validation can mitigate delays caused by strict reapproval requirements. Patients, especially in mental health, often favor repurposed drugs due to established safety profiles. This strategy enhances access to affordable treatments in resource-limited settings, advancing health equity. Prioritizing cost-effective repurposing is critical for reducing global disease burdens.

CONCLUSION

This study utilizes the MR method to conduct an in-depth exploration into the repurposing of currently available medications. A significant portion of our research findings demonstrate alignment with the existing body of literature, thereby emphasizing that MR emerges as a highly promising strategy for unearthing novel therapeutic applications for well-established drugs. Nevertheless, it is important to note that due to the comparatively limited sample size of the medication data, the complexity of potential confounding factors, as well as the possibility of nonlinear drug effects, the obtained results of this study should merely be regarded as initial and provisional indications. To guarantee the precision of these findings and successfully transform this research into concrete therapeutic advantages that can be tangibly realized, further extensive clinical and experimental validations are unquestionably essential. Moreover, to achieve clinical implementation, it is crucial to address regulatory hurdles, cost-effectiveness, and patient acceptance. By addressing these challenges, we can unlock the full potential of MR to contribute significantly to the development of new therapeutic strategies and improve public health outcomes.

ACKNOWLEDGEMENTS

We want to acknowledge the participants and investigators of the UK biobank and the FinnGen study.

LIST OF ABBREVIATIONS

Alzheimer's Disease

Attention Deficit Hyperactivity Disorder

Amyotrophic Lateral Sclerosis

Bayesian Colocalization

Genome-wide Association Study

Inflammatory Bowel Disease

Inverse Variance Weighted

Instrumental Variants

Linkage Disequilibrium Score Correlation

Mendelian Randomization

Multiple Sclerosis

Parkinson's Disease

Posterior Probability of H4

Genetic Correlation Coefficient

Subarachnoid Hemorrhage

AUTHORS' CONTRIBUTIONS

The authors confirm their contributions to the paper as follows: study conception, design, and revision: L.G., S.L., L.Z.; Drafting of the manuscript: XX; Revision assistance: T.L., Q.W. All authors reviewed the results and approved the final version of the manuscript.

ETHICS APPROVAL AND CONSENT TO PARTICIPATE

Not applicable.

HUMAN AND ANIMAL RIGHTS

Not applicable.

CONSENT FOR PUBLICATION

Not applicable.

STANDARDS OF REPORTING

STROBE-MR guidelines were followed.

AVAILABILITY OF DATA AND MATERIALS

All genomic and clinical data utilized in this paper, specifically the GWAS data, can be retrieved from public repositories in the UK Biobank Neale Lab database and the FinnGen database.

FUNDING

The study was funded by the “National Natural Science Foundation of China” (Grant no. 82204368, 32371202, 32071182), the “Fundamental Research Funds for the Central Universities” (Grant no. SCU2022F4080), and supported by “Sichuan Science and Technology Program” (Grant no. 2023NSFSC0675).

CONFLICT OF INTEREST

The authors declare no conflict of interest, financial or otherwise.

SUPPLEMENTARY MATERIAL

Supplementary material is available on the publisher’s website along with the published article.

REFERENCES

Associated Data

Supplementary Materials

Supplementary material is available on the publisher’s website along with the published article.

Data Availability Statement

All genomic and clinical data utilized in this paper, specifically the GWAS data, can be retrieved from public repositories in the UK Biobank Neale Lab database and the FinnGen database.