Bidirectional Mendelian Randomization analysis of the genetic association between neuromyelitis optica spectrum disorder and cortical structure

doi:10.21203/rs.3.rs-4850074/v1

Download PDF

Research Article

Bidirectional Mendelian Randomization analysis of the genetic association between neuromyelitis optica spectrum disorder and cortical structure

https://doi.org/10.21203/rs.3.rs-4850074/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background

Observational studies have suggested an association between neuromyelitis optica spectrum disorder (NMOSD) and cortical structure, but the results have been inconsistent.

Objective

We used two-sample Mendelian randomization (MR) to assess the bidirectional causal relationship between NMOSD and cortical structure.

Methods

Publicly available research by Karol Estrada et al. provided the NMOSD data, which included 1244 control patients, 132 cases of AQP4-IgG seropositive NMOSD, and 83 cases of AQP4-IgG seronegative NMOSD. ENIGMA Consortium provided genome-wide association study (GWAS) data for cortical surface area (SAw/nw) and thickness (THw/nw) in 51,665 people with European ancestry. For MR, the primary analysis approach employed was the inverse-variance weighted (IVW) method. Sensitivity analyses were used to assess pleiotropy and heterogeneity.

Results

Significant associations were identified between specific cortical regions and NMOSD subtypes. For NMOSD as an outcome, significant results included associations with pericalcarine THw (p = 0.0047,beta =-0.003), pericalcarine THnw (p = 0.0070,beta=-0.002), and superior temporal THw (p = 0.0252,beta = 0.002). For NMOSD as an exposure, significant associations included rostral middle frontal SAw (p = 0.0126,beta = 6.907), rostral middle frontal THw (p = 0.0288, beta =-0.001), and inferior parietal SAw (p = 0.0186, beta = 4.572).

Conclusion

Our findings support a reciprocal causal link between cortical anatomy and NMOSD.Confirming these relationships and clarifying the underlying mechanisms will require more investigation.

Neuromyelitis optica spectrum disorder

Cortical structure

Mendelian randomization

Bidirectional analysis

Neuromyelitis optica spectrum disorder (NMOSD) is a rare autoimmune inflammatory demyelinating disorder of the central nervous system (CNS). Although considered a rare disorder, epidemiological investigations reveal that the worldwide prevalence of NMOSD is low, with estimates ranging from 0.3 to 4.4 per 100,000 population ^{[1, 2]}. NMOSD is also recognized as a severely disabling condition that can lead to devastating sequelae, including permanent blindness, paralysis, and even death ^{[3, 4]}. NMOSD encompasses a group of inflammatory disorders of the CNS that primarily affect the optic nerves and spinal cord, with less frequent involvement of the brainstem, resulting in acute episodes of optic neuritis, myelitis, and brainstem encephalitis ^[5].

Recent studies have highlighted the presence of longitudinal changes in brain and spinal cord atrophy during the subclinical pathological process of NMOSD. These findings suggest an ongoing underlying pathology that progresses even when the disease appears clinically inactive. Further research has indicated that these changes in atrophy are associated with immune-inflammatory responses, providing insight into the chronic nature of NMOSD beyond its acute manifestations ^[6].

With the advent of new diagnostic testing methods, the identification and classification of NMOSD have significantly improved. Serological testing for anti-aquaporin-4 antibodies (AQP4-IgG) has become a critical tool in distinguishing between seropositive and seronegative NMOSD cases^[7]. This serological distinction aids in the accurate diagnosis and management of the disorder, as it helps to differentiate NMOSD from other similar neuroinflammatory conditions. However, diagnosing NMOSD remains challenging, particularly in cases where patients are negative for AQP4-IgG and myelin oligodendrocyte glycoprotein antibodies (MOG-IgG)^[8]. In such cases, clinicians must rely heavily on clinical and imaging manifestations, which can sometimes be ambiguous.

Distinguishing NMOSD from multiple sclerosis (MS) is especially difficult due to the intricate and overlapping clinical and neuroimaging features of the two disorders^{[3, 9]}. MS, like NMOSD, is a demyelinating disease of the CNS, but it has different underlying pathophysiological mechanisms and treatment responses. Recent advancements in imaging and biomarker research are gradually improving the ability to differentiate between these two conditions, but significant challenges remain.

Excitingly, recent studies have utilized Mendelian randomization (MR) to explore the causal relationships between neuroinflammatory diseases and structural changes in the brain^[10]. MR has emerged as a powerful tool in epidemiology and genetics, allowing researchers to infer causal relationships between exposures and outcomes by using genetic variants as instrumental variables ^[11]. This method leverages the random assortment of genes from parents to offspring, akin to a natural randomized controlled trial, thus minimizing confounding and reverse causation.

The field of MR has also seen significant methodological advancements that enhance its robustness and applicability. Techniques such as multivariable MR, which can account for pleiotropy by simultaneously analyzing multiple genetic variants, and MR-Egger regression, which provides a way to detect and adjust for horizontal pleiotropy, have further strengthened the reliability of MR findings^[12].These methodological improvements have made MR a valuable tool for investigating complex causal relationships in neuroinflammatory diseases.

In this study, we conducted a two-sample MR analysis using publicly available genome-wide association study (GWAS) data to determine the causal relationship between NMOSD and brain structure. By employing this advanced genetic epidemiology approach, we aim to uncover potential causal pathways linking NMOSD to structural changes in the brain, thereby enhancing our understanding of the disease's pathophysiology and informing future therapeutic strategies.

2.1 Data sources and Participants

We obtained NMOSD GWAS data from the publicly available study by Karol Estrada et al (https://doi.org/10.1038/s41467-018-04332-3).The dataset comprises 132 cases of AQP4-IgG seropositive NMOSD, 83 cases of AQP4-IgG seronegative NMOSD, and 1244 control subjects. In our study, they are respectively referred to as NMO-IgG+, NMO-IgG-, and NMO.

The results from the MRI-derived brain morphometry GWAS conducted by the ENIGMA consortium were used as our outcome data for brain cortical structure, adjusted for covariates such as age, sex, four multidimensional scaling, intracranial volume, and site. The GWAS data included MRI data of cortical surface area (SA) and thickness (TH) from 51,665 individuals, categorized into 34 functionally specialized regions according to the Desikan-Killiany atlas. More details can be found in the original study (https://doi.org/10.1126/science.aay6690). We utilized the aforementioned data to analyze the bidirectional causal effects of different types of NMO on the cortical SA and TH of the whole brain and the 34 functional regions (categorized by whether they are whole-brain weighted), conducting a total of 138*3*2 analyses.

2.2 Selection of instrumental variables

Based on the basic assumptions of MR^[13], we first extracted the genetic variants associated with NMO-IgG +, NMO-IgG-, and NMO.A threshold of 510^-6 was used for exposure-related instrumental variable selection and linkage disequilibrium pruning (r2 < 0.001 within a 5000kb range) due to the small sample size of NMOSD. The remaining SNPs were then retrieved from the outcome dataset. Moreover, we manually eliminated SNPs linked to outcomes (5*10 − 6) or confounders using the OpenGWAS web platform (https://gwas.mrcieu.ac.uk/). Moreover, we used F-tests (F-statistic > 10) to confirm that our study was free of weak instrument bias.The next steps of the reverse MR approach were the same as the forward process, with the exception of selecting genetic variants highly associated with the entire genome (5*10 − 8) and removing linkage disequilibrium (r2 < 0.001 within a 10000kb range).

2.3 MR analysis

We conducted MR tests using inverse variance weighted (IVW), MR-Egger, weighted-median, weighted-mode, and Simple mode methods, each relying on different assumptions^{[14, 15]}. For sensitivity analyses, we employed the MR-PRESSO global test and MR-Egger regression test to account for potential pleiotropy, and used Cochran’s Q statistic to assess heterogeneity among genetic variants^[16]. In the reverse MR process, we applied the same analytical methods. The specific procedure is illustrated in Fig. 1.

MR analyses were performed using R (version 4.2.2) and the packages TwoSampleMR and MR-PRESSO. The results were reported following the MR-STROBE guidelines to enhance the reporting of observational studies. All hypothesis tests were two-sided. Multiple comparisons were corrected using the Bonferroni method, setting the thresholds for significance and potential significance at P < 0.05/138 = 0.0004 and P < 0.05.

3.1 Primary MR

In the primary MR process, we preliminarily identified potential causal associations using the IVW method between NMO and the cortical structure of 3 brain regions, NMO-IgG- and the cortical structure of 4 brain regions, and NMO-IgG + and the cortical structure of 8 brain regions (Figs. 2 and 3). We conducted sensitivity tests on these results, and due to the detection of directional pleiotropy, we removed 2 of the potential causal associations between NMO-IgG + and the cortical structure. The specific results are shown in Table 1.

Table1. Heterogeneity and pleiotropy tests of the causal effects of NMO on brain cortical structure.

Therefore, after a series of sensitivity analyses, we found three MR results suggesting a causal relationship between NMO and cortical structure (with IVW results as the primary analysis), including pericalcarine THw (P = 0.0047, beta(se) = -0.0026(0.0009)), pericalcarine THnw (P = 0.0070, beta(se) = -0.0025(0.0009)), and superior temporal THw (P = 0.0252, beta(se) = 0.0019(0.0008))(Fig. 4). Additionally, four MR results indicated a causal association between IgG- NMO and cortical structure, as follows: rostral middle frontal SAw (P = 0.0126, beta(se) = 6.9071(2.7696)), rostral middle frontal SAnw (P = 0.0431, beta(se) = 9.7888(4.8395)), rostral middle frontal THw (P = 0.0288, beta(se) = -0.0013(0.0006)), and parahippocampal SAw (P = 0.0409, beta(se) = -1.5389(0.7526)). Furthermore, after removing the two results affected by directional pleiotropy, only six MR results showed a causal relationship between IgG + NMO and cortical structure: cuneus THw (P = 0.0345, beta(se) = 0.0010(0.0005)), frontal pole SAnw (P = 0.0295, beta(se) = -0.3245(0.1491)), inferior parietal SAw (P = 0.0186, beta(se) = 4.5725(1.9433)), inferior temporal SAnw (P = 0.0457, beta(se) = -3.4768(1.7400)), posterior cingulate SAw (P = 0.0307, beta(se) = 1.0269(0.4750)), and postcentral SAnw (P = 0.0396, beta(se) = -3.8958(1.8931)) (Fig. 4). Notably, these MR estimates are limited to the regional functional level, with no evidence supporting a potential causal relationship between NMO (IgG+/IgG-) and overall cortical TH or SA.

3.2 Reverse MR

The reverse MR analysis suggests several significant and borderline significant associations between cortical structures and NMOSD subtypes. These findings indicate that NMOSD might influence specific cortical regions, potentially contributing to changes in brain morphology. However, the observed associations warrant further investigation to confirm the causal relationships and to understand the underlying biological mechanisms.

Our study utilized a comprehensive two-sample MR approach to explore the potential causal relationships between NMOSD and brain cortical structures.To our knowledge, this is the first MR study investigating the causal relationship between NMOSD and brain structure. We discovered significant associations suggesting that NMOSD, both seropositive and seronegative, may influence specific cortical regions, and vice versa. These findings are crucial for understanding the intricate interplay between NMOSD and cortical morphology, providing new insights into the disease's pathophysiology.

We found that NMOSD is associated with cortical atrophy in specific brain regions, including the pericalcarine cortex and the superior temporal cortex. These associations indicate that NMOSD may contribute to localized cortical thinning, which could be linked to the inflammatory and demyelinating processes characteristic of the disorder.For instance, Kim et al^[17] identified cortical thinning in regions associated with sensory and motor functions in NMOSD patients, using advanced MRI techniques.A study demonstrate that the correlation between changes in the cerebral cortex and clinical features in NMOSD patients with normal-appearing brain tissue. The findings reveal that cortical thinning in regions such as the bilateral rostral middle frontal gyrus and left superior frontal gyrus is associated with clinical disability and cognitive function^[18]. It is well known that inflammatory responses are closely related to NMOSD^[19–21], where antigen-antibody mediated immune responses can impact the structure of the cerebral cortex, resulting in corresponding clinical symptoms. Inflammation plays a critical role in cortical thinning, as evidenced by studies on MS^[23], another demyelinating disorder. In MS, chronic inflammation leads to cortical damage and atrophy, which are associated with neurological and cognitive impairments^{[24, 25]}. Similarly, in NMOSD, persistent inflammatory activity likely contributes to the observed cortical changes, suggesting a shared pathological mechanism involving inflammation-driven cortical atrophy in these demyelinating diseases^[26].

Our bidirectional MR analysis also revealed that certain cortical regions, such as the rostral middle frontal cortex and the inferior parietal cortex, may influence the risk of developing NMOSD. This bidirectional relationship underscores the complexity of NMOSD and suggests that structural brain abnormalities could play a role in disease susceptibility.The research conducted by Kim et al ^[17] demonstrated widespread cortical thinning in NMOSD patients, including in the rostral middle frontal cortex, which is associated with sensory and motor functions. This suggests that structural changes in these areas might influence the risk of developing NMOSD by affecting critical brain functions related to disease symptoms.Another study highlights that NMOSD patients experience decreased cortical thickness and increased ventricle enlargement compared to healthy controls^{[27, 28]}. This indicates that brain structural changes, particularly in the frontal and parietal lobes, are not only a consequence of the disease but may also contribute to its development and progression.These findings underscore the complexity of NMOSD and suggest that structural abnormalities in specific cortical regions could predispose individuals to the disorder, reinforcing the bidirectional relationship observed in your MR analysis. Further research is needed to elucidate the precise mechanisms by which these structural changes influence NMOSD susceptibility and progression.

Mendelian randomization (MR) has emerged as a powerful tool in epidemiological research for assessing causal relationships between exposures and diseases. By utilizing instrument variants, MR can mimic a randomized controlled trial setting, providing more robust evidence for causal inference compared to observational studies alone^[29].A notable advantage of MR is its ability to elucidate causal relationships that might not be feasible through traditional observational studies. For instance, MR has been instrumental in studying the impact of various exposures on disease outcomes across different fields^[30–32]. Furthermore, MR studies can serve as a foundation for subsequent clinical research. By identifying potential causal pathways, MR findings can inform the design of targeted interventions and therapeutic strategies. For example, a recent MR study investigating the relationship between vitamin D levels and multiple sclerosis risk highlighted promising avenues for further clinical trials on vitamin D supplementation as a preventive measure ^[31].Moreover, MR has been pivotal in advancing our understanding of complex neurological disorders. In Alzheimer's disease research, MR studies have explored the causal links between genetic variants associated with lipid metabolism and Alzheimer's risk, suggesting novel therapeutic targets ^[33,34]. To strengthen the robustness of our findings, ensuring that the observed associations are less likely to be due to confounding factors or biases, sensitivity analyses, including MR-Egger and MR-PRESSO, were conducted to account for potential pleiotropy and heterogeneity.

Our results have several important implications for future research and clinical practice. Firstly, the identified associations between NMOSD and specific cortical regions provide new targets for further investigation into the mechanisms underlying the disease. Understanding how NMOSD affects brain structure and vice versa could lead to the development of novel therapeutic strategies aimed at protecting or restoring cortical integrity in affected individuals.Secondly, the bidirectional nature of the associations suggests that monitoring cortical structure could potentially serve as a biomarker for NMOSD progression or risk assessment. This could be particularly useful in clinical settings, where early detection and intervention are crucial for improving patient outcomes.Despite the strengths of our study, including the use of large-scale GWAS data and robust MR methodologies, there are several limitations to consider. The sample size for NMOSD cases, particularly for the seronegative subgroup, was relatively small, which may limit the generalizability of our findings. Additionally, our study focused on individuals of European ancestry, and further research is needed to confirm these associations in diverse populations.

In conclusion, our MR analysis provides evidence for a bidirectional causal relationship between NMOSD and brain cortical structure. These findings highlight the importance of considering both the neurological and immunological aspects of NMOSD in understanding its pathogenesis and developing effective treatments. Further research is warranted to elucidate the underlying mechanisms and to explore the potential for cortical structure-based biomarkers in NMOSD.

6.Ethics approval and consent to participate

The study did not require new ethical approval or informed permission because all the data were obtained from publicly available sources and prior approval and informed consent had been obtained.

7.Funding

This work was supported by the Xinjiang Uygur Autonomous Region Natural Science Foundation(grant number:2022D01C759).

8.Consent for publication

Since all of the data for this study came from publicly available sources and informed consent and ethical approval had already been obtained, there was no need to obt-ain them again.

9.Author contribution

The study was designed by Rena Abudusalamu,who also conducted the methodology.Rena Abudusalamu and Aierpati Maimaiti, co-authors, completed the software implement, validation, and contributed to the writing, reviewing, and editing of the original document.Jianhua Ma provided oversight throughout the project, contributed to the conceptualization, supervised the research process, and provided critical revisions to the manuscript. They ensured the integrity and accuracy of the work and provided final approval for submission. Additionally,Dengfeng Han, Paziliya Akram and Chenguang Hao participated in coordinating data collection and analysis, and also evaluated the text. Each author contributed to and accepted the submitted version of the paper.

10.Conflicts of interests disclosure

The authors declare that no financial or commercial relationship that would be seen as having a conflict of interest existed during the conduct of the study.

Author Contribution

Ma Jianhua, Han Dengfeng this two authors contributed equally as joint corresponding authors.Rena.Abudusalamu and Aierpati.Maimaiti contributed equally to this work and and are considered co-first authors

Pandit L. Neuromyelitis optica spectrum disorders: an update. Ann Indian Acad Neurol. 2015;18(Suppl 1):S11–S15.
Wu Y, Zhong L, Geng J. Neuromyelitis optica spectrum disorder: Pathogenesis, treatment, and experimental models. Mult Scler Relat Disord. 2019;27:412–418.
Wingerchuk DM, Banwell B, Bennett JL, et al. International consensus diagnostic criteria for neuromyelitis optica spectrum disorders. Neurology. 2015;85(2):177–189.
Cree BAC, Bennett JL, Kim HJ, et al. Inebilizumab for the treatment of neuromyelitis optica spectrum disorder (N- MOmentum): a double- blind, randomised placebo- controlled phase 2/3 trial. Lancet (London, England). 2019;394(10206):1352–1363.
Jarius, Sven et al.Update on the diagnosis and treatment of neuromyelits optica spectrum disorders (NMOSD) revised recommendations of the Neuromyelitis Optica Study Group (NEMOS). Part I: Diagnosis and differential diagnosis.Journal of neurology vol. 270,7 (2023): 3341–3368.
Lorefice, L, and R Cortese. Brain and spinal cord atrophy in NMOSD and MOGAD: Current evidence and future perspectives. Multiple sclerosis and related disorders vol. 85 (2024): 105559.
Gospe SM 3rd, Chen JJ, Bhatti MT. Neuromyelitis optica spec-trum disorder and myelin oligodendrocyte glycoprotein associated disorder- optic neuritis: a comprehensive review of diagnosis and treatment. Eye (London, England). 2021;35(3):753–768.
Liu C, Shi M, Zhu M, Chu F, Jin T, Zhu J. Comparisons of clinical phenotype, radiological and laboratory features, and therapy of neuromyelitis optica spectrum disorder by regions: update and challenges. Autoimmun Rev. 2021;21(1):102921.
Shi, Mingchao et al.Progress in treatment of neuromyelitis optica spectrum disorders (NMOSD): Novel insights into therapeutic possibilities in NMOSD. CNS neuroscience & therapeutics vol. 28,7 (2022): 981–991.
Sun, Dongren et al.Causal relationship between multiple sclerosis and cortical structure: a Mendelian randomization study.” Journal of translational medicine vol. 22,1 83. 20 Jan. 2024.
Davey Smith, George, and Gibran Hemani.Mendelian randomization: genetic anchors for causal inference in epidemiological studies.Human molecular genetics vol. 23,R1 (2014): R89-98.
Bowden, Jack et al.Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genetic epidemiology vol. 40,4 (2016): 304–14.
Burgess S, Butterworth A, Thompson SG. Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol. 2013;37(7):658–65.
Burgess S, Davey Smith G, Davies NM, Dudbridge F, Gill D, Glymour MM, et al. Guidelines for performing Mendelian randomization investigations. Wellcome open research. 2019;4:186.
Choi KW, Chen CY, Stein MB, Klimentidis YC, Wang MJ, Koenen KC, et al. Assessment of bidirectional relationships between physical activity and depression among adults: a 2-sample Mendelian randomization study. JAMA Psychiat. 2019;76(4):399–408.
Ong, Jue-Sheng, and Stuart MacGregor. “Implementing MR-PRESSO and GCTA-GSMR for pleiotropy assessment in Mendelian randomization studies from a practitioner's perspective.” Genetic epidemiology vol. 43,6 (2019): 609–616.
Kim, S-H et al.Widespread cortical thinning in patients with neuromyelitis optica spectrum disorder. European journal of neurology vol. 23,7 (2016): 1165–73.
Huang, Chuxin et al. Correlation between cerebral cortex changes and clinical features in patients with neuromyelitis optica spectrum disorder with normal-appearing brain tissue: a case-control study.Neural regeneration research vol. 18,11 (2023): 2520–2525.
Shahmohammadi, Sareh et al. Autoimmune diseases associated with Neuromyelitis Optica Spectrum Disorders: A literature review.Multiple sclerosis and related disorders vol. 27 (2019): 350–363.
Kim, Woojun, and Ho Jin Kim. An update on biologic treatments for neuromyelitis optica spectrum disorder. Expert review of clinical immunology vol. 19,1 (2023): 111–121.
Li, Wenqun et al. The role and mechanisms of Microglia in Neuromyelitis Optica Spectrum Disorders. International journal of medical sciences vol. 18,14 3059–3065. 16 Jun. 2021.
Lee, Chi-Yan et al. Differential brainstem atrophy patterns in multiple sclerosis and neuromyelitis optica spectrum disorders. Journal of magnetic resonance imaging: JMRI vol. 47,6 (2018): 1601–1609.
Musella, Alessandra et al. Interplay Between Age and Neuroinflammation in Multiple Sclerosis: Effects on Motor and Cognitive Functions. Frontiers in aging neuroscience vol. 10 238. 8 Aug. 2018.
Hämäläinen, P, and E Rosti-Otajärvi. Cognitive impairment in MS: rehabilitation approaches.” Acta neurologica Scandinavica vol. 134 Suppl 200 (2016): 8–13.
Huang, Chuxin et al. Correlation between cerebral cortex changes and clinical features in patients with neuromyelitis optica spectrum disorder with normal-appearing brain tissue: a case-control study. Neural regeneration research vol. 18,11 (2023): 2520–2525.
Tian, De-Cai et al. Cortical Thinning and Ventricle Enlargement in Neuromyelitis Optica Spectrum Disorders. Frontiers in neurology vol. 11 872. 27 Aug. 2020.
Zhang, Ying et al. Brain structural and functional connectivity alterations are associated with fatigue in neuromyelitis optica spectrum disorder. BMC neurology vol. 22,1 235. 27 Jun. 2022.
Boef, Anna G C et al. Mendelian randomization studies: a review of the approaches used and the quality of reporting. International journal of epidemiology vol. 44,2 (2015): 496–511.
Carreras-Torres, Robert et al. Obesity, metabolic factors and risk of different histological types of lung cancer: A Mendelian randomization study. PloS one vol. 12,6 e0177875. 8 Jun. 2017.
Mokry, Lauren E et al.Vitamin D and Risk of Multiple Sclerosis: A Mendelian Randomization Study. PLoS medicine vol. 12,8 e1001866. 25 Aug. 2015.
Larsson, Susanna C et al.Serum magnesium and calcium levels in relation to ischemic stroke: Mendelian randomization study. Neurology vol. 92,9 (2019): e944-e950.
Dunk, Michelle M et al.Associations of dietary cholesterol and fat, blood lipids, and risk for dementia in older women vary by APOE genotype. Alzheimer's & dementia: the journal of the Alzheimer's Association vol. 19,12 (2023): 5742–5754.
Storm, Catherine S et al. Finding genetically-supported drug targets for Parkinson's disease using Mendelian randomization of the druggable genome. Nature communications vol. 12,1 7342. 20 Dec. 2021.

No competing interests reported.

Download PDF

Editor assigned by journal
10 Aug, 2024
Submission checks completed at journal
10 Aug, 2024
First submitted to journal
02 Aug, 2024

You are reading this latest preprint version

Bidirectional Mendelian Randomization analysis of the genetic association between neuromyelitis optica spectrum disorder and cortical structure

Status:

Version 1

Abstract

Background

Objective

Methods

Results

Conclusion

Figures

1. Introduction

2. Material and Method

2.1 Data sources and Participants

2.2 Selection of instrumental variables

2.3 MR analysis

3. Results

3.1 Primary MR

3.2 Reverse MR

4. Discussion

5. Conclusion

Declarations

6.Ethics approval and consent to participate

Author Contribution

References

Additional Declarations

Status:

Version 1