Uncovering Potential Genetic Targets in the Mediating Role of Demyelinating Disease for Multiple Sclerosis-Induced Neuropathic Bladder: A Mendelian Randomization Analysis

doi:10.21203/rs.3.rs-4339401/v1

Download PDF

Research Article

Uncovering Potential Genetic Targets in the Mediating Role of Demyelinating Disease for Multiple Sclerosis-Induced Neuropathic Bladder: A Mendelian Randomization Analysis

https://doi.org/10.21203/rs.3.rs-4339401/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Objective

Despite the lack of a genetic explanation for the causal link between multiple sclerosis (MS) and neuropathic bladder (NPB), our study aims to explore this causality and underlying mechanisms using Mendelian Randomization (MR), aiming to identify novel protein targets for future therapeutic interventions.

Methods

Data pertaining to MS, demyelinating diseases (DD), neurogenic bladder, and plasma proteins were sourced from the IEU Open GWAS Project and encompassed a diverse population. After stringent screening, a bidirectional two-sample MR analysis was first conducted to establish the causality between MS and NPB. DD was then introduced as a mediator for further testing via the product of coefficients approach. Subsequently, plasma proteins were analyzed as exposures against the aforementioned phenotypes to screen for potential therapeutic targets.

Results

Our analysis substantiates that MS is associated with an increased risk of developing NPB (P-value = 0.001), with no evidence of reverse causality, reinforcing the unidirectional impact of MS on NPB. The mediation analysis revealed a clear causal pathway, supporting the hypothesis that DD serves as a crucial intermediary in the progression from MS to NPB (P-value = 0.005, mediation proportion = 70.29%). Notable proteins such as ADAM11, GRIA4, CXCL13, and PRKCG were identified, and by relaxing the FDR constraints, GSR and UBA2 were also pinpointed as potential risk factors for both MS and NPB.

Conclusions

Our MR analysis elucidated the causal connections among MS, DD, and NPB from a genetic perspective, identifying potential protein targets that facilitate future drug development and therapeutic strategies.

Multiple sclerosis

Demyelinating diseases

Neuropathic bladder

Plasma protein

Mendelian Randomization

Neuropathic bladder (NPB), characterized by a spectrum of urinary dysfunctions, emerges prominently in the context of neurological conditions, notably multiple sclerosis (MS), a chronic demyelinating disease(DD) of the central nervous system^[1]. The intersection of MS and NPB highlights a complex interplay between neurological damage and urinary system regulation, affecting millions globally^[2]. The epidemiological linkage between MS and the onset of NPB underscores an urgent need for understanding the underlying causative mechanisms^{[3, 4]}.

The risk landscape for NPB encompasses various neurological insults, with MS standing out due to its direct impact on neural pathways critical for bladder control. Despite extensive observational research delineating the association between MS and NPB, these studies often grapple with confounding variables and the challenge of reverse causation, leaving the causality question partially addressed^{[3, 5]}.

In the quest for clarity, Mendelian randomization (MR) emerges as a powerful tool, leveraging genetic variants as instrumental variables to unearth causal relationships between exposures and outcomes^[6]. This approach inherently minimizes the confounding biases that plague conventional observational studies, providing a more robust framework for causal inference^{[7, 8]}. Notably, the study currently lacks an exploration of the genetic dimensions connecting MS to NPB through DD, representing a significant gap in our understanding.

By harnessing the power of MR, our investigation aims to dissect the genetic architecture underlying MS and its potential causal influence on NPB, and simultaneously seeks new potential therapeutic targets for MS and NPB. This novel approach not only promises to enrich our comprehension of these conditions but also to pave the way for genetic-based interventions, offering new horizons for treatment modalities in affected populations.

3.1 Study overview

In the present study, we utilized a two-sample MR approach to investigate the potential causal relationship between MS and NPB. This analysis was underpinned by summary-level data from genome-wide association studies (GWAS)^[9], aiming to elucidate the genetic underpinnings influencing this association. Figure 1 offers a schematic overview of the study's methodology, underscoring three pivotal criteria essential for the validity of MR findings: (i) a robust genetic association with the exposure of interest; (ii) independence of the genetic instruments from confounders; and (iii) the instrumental variable’s effect on the outcome is mediated solely through the exposure in question. This research was guided by the STROBE-MR guidelines to ensure rigorous methodological standards^[10], with a detailed checklist available in the Supplementary Material for in-depth verification.

3.2 Data source

Details pertaining to our data sources are presented in Table 1. In our investigation of the genetic underpinnings of multiple sclerosis 1(MS1), we utilized exposure data from two independent sources. The primary dataset was derived from the FinnGen project's R10 release as detailed by Kurki et al^[11]. FinnGen amalgamates genome-wide genotyping data with longitudinal health records from Finnish biobanks. This resource capitalizes on Finland’s distinctive population attributes—genetic homogeneity and extensive healthcare documentation—to construct a well-defined cohort. Specifically, we included a subset of 410,970 individuals, comprising 2,409 cases and 408,561 controls. This cohort was subjected to rigorous quality control procedures to ensure diverse and representative genetic data.

To bolster the generalizability of our findings, we incorporated additional exposure data multiple sclerosis(MS2) from an international cohort sourced by Sawcer S. et al.^[12] from the MRC Integrative Epidemiology Unit (IEU) OpenGWAS database^{[9, 13]}. This dataset included 27,098 participants, with 9,722 MS cases and 17,376 controls, all genotyped using genome-wide SNP arrays. This inclusion aimed to expand the dataset’s autosomal SNP coverage, adhering to stringent quality control standards to enhance the robustness and sensitivity of our analyses.

Utilizing the Finn biobank platform, we obtained mediator data, which offered robust summary statistics from a cohort of 1,284 individuals with demyelinating diseases of the central nervous system, as well as 217,508 controls, ensuring the dataset’s integrity through meticulous screening and quality control processes.

For outcome data, we relied on the EBI database, accessed via the IEU OpenGWAS platform, which provided summary statistics for 2,237 individuals with a NPB and 464,793 controls, as compiled by Sakaue S. et al^[14]. This data underwent stringent screening and quality control to ensure a comprehensive and valid dataset.

Plasma protein samples were collected from 35,559 individuals by Kurki et al.^[15]from the Icelandic population^[16]. These samples were part of extensive genomic projects involving both whole-genome sequencing of 49,708 individuals and chip typing of an additional 166,281 individuals^[17]. Imputation techniques based on these data enhanced the genetic information for detailed analyses. Proteomic measurements were conducted using the SomaScan multiplex aptamer assay (version 4), which quantifies 4,907 unique plasma proteins. This cutting-edge technology allowed for the detailed detection and quantification of a broad spectrum of proteins, thereby providing a comprehensive overview of the plasma proteome landscape in Iceland. Rigorous quality control measures, including multiple testing corrections, were applied to maintain data integrity and statistical rigor. Validation efforts included cross-validation with established biomarkers and replication of significant findings in independent cohorts such as the INTERVAL study^[18] and SCALLOP consortium^[19].

Table 1. Characteristics of the genome-wide association study utilized in the research

Our manuscript offers a comprehensive summary of the datasets utilized, detailing ID numbers, trait, patient and control counts, and PMID, ensuring that our methodology is transparent and meticulously documented.

3.3 Selection of genetic instruments

Prior to conducting the analysis, the dataset was subjected to normalization techniques to mitigate the impact of potential confounders for NPB, such as spinal trauma, diabetes-induced, Parkinson's disease, and primary genetic elements that contribute to population stratification. Following the acquisition and normalization of SNP data relevant to the study's exposure and outcome measures, efforts were made to harmonize this data, thereby enhancing its suitability for subsequent analytical procedures.

For the identification of specific SNPs, a multifaceted filtering approach was adopted. Initially, a relaxed genome-wide significance threshold of p < 5×10^− 8 was utilized. Subsequently, the linkage disequilibrium (LD) test was executed, where an LD r² < 0.001 within a 1,0000 kb window was applied to ensure the selection of independent IVs^[20]. The F-statistic for each SNP was calculated, leading to the exclusion of SNPs with F < 10 to mitigate the risk of weak instrument bias^{[21, 22]}. If it is necessary, use F = (beta/se)² as a substitute for the F-value in filtering processes^[23]. Next, we introduced the steiger test to filter out instrumental variables that are strongly related to the outcome and enhance the stability of the results^[24]. Finally, SNPs that were either incompatible or palindromic with intermediate allele frequency were systematically removed during the harmonization phase to ensure the integrity of the instrumental variable analysis.

3.4 Mendelian randomization analysis and sensitivity analysis

In this study, we employed a sophisticated statistical framework, combining both traditional and novel methodologies to bolster the robustness and validity of our MR analyses^[25]. At the core of our approach was the IVW method, selected for its capacity to provide conservative estimates amidst heterogeneity, aligning with practices in current research^{[7, 26]}. We complemented this with several targeted strategies: MR-PRESSO^[27] for identifying outliers, the WM^[28] for its resilience in the face of pleiotropy, and MR-Egger regression^[7] to address unbalanced pleiotropy and heterogeneity, the latter offering more reliable estimates under specific conditions.

To address biases introduced by pleiotropy in causal inference frameworks, our study implemented the BWMR approach^[29]. This advanced methodology strategically applies Bayesian weights to mitigate violations of instrumental variable assumptions attributable to pleiotropic effects, the model incorporates considerations of the uncertainty associated with polygenic weak effects. Additionally, we have integrated Con_ML^[30] into our analysis to control both correlated and uncorrelated pleiotropic effects. Study heterogeneity and horizontal pleiotropy were quantified using Cochran's Q statistics(P > 0.05 indicates no heterogeneity)^[31] and the MR-Egger intercept test(P > 0.05 indicates no horizontal pleiotropy)^[7]. A leave-one-out sensitivity analysis facilitated the exclusion of outliers that might distort the causal inference, further refined by the MR-PRESSO test for outlier adjustment^[27]. Following outlier identification and heterogeneity among instrumental variables, we conducted revised analyses to ensure precision.

To authenticate our results, we scrutinized potential confounders through the LDlink database^{[32, 33]}, evaluating connections between significant MR estimates and neuropathic bladder risk factors. After removing SNPs associated with confounding variables, we re-evaluated the causal impacts, reaffirming their statistical importance.

3.5 Statistics

For our analyses, we utilized a suite of specialized R packages: TwoSampleMR(v0.5.10), Mendelian Randomization (v0.9.0)^[34], and MRPRESSO (v1.0), operating within the R Software environment (version 4.3.3) accessible (available at https://www.R-project.org). In addition to standard significance thresholds, we applied False Discovery Rate (FDR) corrections to adjust for multiple testing^[35]. We set the threshold for statistical significance at a two-tailed P-value of less than 0.05. The outcomes of our investigation are presented as odds ratios (ORs), quantifying the effect of an additional standard deviation increase in the exposure of interest.

4.1 Association of Genetically Predicted MS with NPB

Our comprehensive analysis identified 22 SNPs significantly associated with MS1 (P < 5×10⁻⁸). Utilizing linkage disequilibrium criteria with a reference panel and a threshold of r² = 0.001, we retained these 22 SNPs for the principal analysis. The F statistics (Mean = 13,663) confirmed the reliability of these genetic markers as robust instruments for MS1, devoid of statistical bias.

Employing a suite of MR methodologies(Figure 2), the IVW approach revealed a significant causal link between genetically predicted MS1 and NPB, evidenced by an OR of 1.116 (95% Confidence Interval [CI], 1.043-1.193; beta = 0.110, SE = 0.034, P-value = 0.001). This finding was corroborated by additional MR analyses: MR Egger (OR = 1.167; 95% CI, 1.038-1.312; beta = 0.155, SE = 0.060, P-value = 0.027), the WM method (OR = 1.153; 95% CI, 1.055-1.261; beta = 0.143, SE = 0.046, P-value = 0.002), the BWMR method (OR = 1.118; 95% CI, 1.044-1.197; beta = 0.111, SE = 0.035, P-value = 0.001), and the cML method (OR = 1.116; 95% CI, 1.034-1.206; beta = 0.110, SE = 0.039, P-value = 0.005). The MR-PRESSO outlier correction validated the robustness of the IVW results (OR = 1.116; 95% CI, 1.045-1.192, P-value = 0.008). Also, our scatter plot and leave-one-out sensitivity analysis further confirmed the causal relationship between the exposure and the outcome(Figure 2).

Tests for horizontal pleiotropy through MR-Egger regression intercept (P-value = 0.379) and MR-PRESSO global tests (P-value = 0.357) indicated no association bias from genetically predicted factors across analyzed subgroups. Furthermore, Cochran’s Q statistics (P-value = 0.474) underscored a lack of SNP heterogeneity in MR analyses, reinforcing the consistency of our findings (refer to supplementary file 1).

4.2 Validating the Association Between Genetically Inferred MS and NPB Risk Across Diverse Cohorts

To validate our findings and mitigate bias from overlapping samples, we sourced data from diverse cohorts. After rigorous screening and exclusion of rs3118470, linked to diabetes, 25 SNPs were employed in the MR analysis. Both IVW and MR-PRESSO revealed a significant association between genetically inferred MS2 and NPB risk, with consistent results across various MR methods (Figure 2). The F statistics (Mean = 352) underscored the robustness of these genetic markers concerning MS2.

The IVW reported an OR of 1.083; 95% CI, 1.008-1.163; beta = 0.080, SE = 0.037, P-value = 0.030, and the MR-PRESSO results mirrored this (OR = 1.083; 95% CI, 1.013-1.158, P-value = 0.028). Tests for horizontal pleiotropy (MR-Egger regression intercept P-value = 0.105, MR-PRESSO global tests P-value = 0.547) and Cochran’s Q statistics (P-value = 0.646) confirmed the absence of SNP heterogeneity and association bias for genetically predicted factors across analyzed subgroups (refer to supplementary file 1).

In the fixed effect meta-analysis conducted across two distinct populations, minimal heterogeneity was observed (P-value = 0.55). The test for overall effect size yielded a significant(P-value < 0.001), underscoring a statistically significant overall effect across the combined studies (refer to Figure 3).Our results robustly confirm that genetically predicted MS is a significant risk factor for NPB, supported by various analytical methods and diverse populations(supplementary file 2).

4.3 Potential Reverse Causality between NPB and MS

In an effort to elucidate potential reverse causality dynamics between NPB and MS, we employed MR with NPB posited as the exposure variable and MS as the outcome variable. Initial analyses focusing on NPB versus MS1 failed to yield any SNPs when applying a stringent significance threshold of 5×10^-8. In pursuit of a more comprehensive SNP profile, we adjusted the threshold to 5×10^-6, which facilitated the identification of 20 SNPs. Subsequent IVW analysis generated a p-value of 0.505, indicating no statistically significant reverse causal relationship. A parallel analysis involving NPB versus MS2 similarly resulted in no significant SNP detection, reinforcing the likelihood of an absence of reverse causality between MS and NPB. This suggests a nuanced interplay where NPB does not appear to exert a causative influence on MS under the conditions and thresholds specified in our study.

4.4 Mediation Analysis of the Causal Pathway from MS to NPB via DD

To explore the pathogenesis between MS and NPB, we hypothesized that MS contributes to the development of NPB through its demyelinating effects on neural cells. DD were therefore posited as mediators in this relationship, facilitating a mediation analysis via MR to ascertain the causal links among these diseases(Figure 4). The two-sample MR analysis demonstrated a significant causal relationship between MS and DD, with the IVW method yielding an OR of 2.02 (95% CI, 1.69-2.41; beta = 0.70, SE = 0.09, p-value = 1.02x10^-14). Additionally, a significant causal link was identified between DD and NPB, with the IVW method indicating an OR of 1.12 (95% CI, 1.04-1.20; beta = 0.11, SE = 0.04, p-value = 0.002)(Figure 3). Our scatter plot and leave-one-out sensitivity analysis further substantiate the causal link between the exposure and the outcome(Figure 4). All results were corroborated by robust checks for pleiotropy and heterogeneity, as detailed in the supplementary materials (supplementary file 3).

OR(95% CI) adjusted for 1-SD increase in MS for NPB. Analysis was conducted using IVW as the primary method, enhanced by MR-PRESSO, cML, BWMR, and WM for improved robustness. TE:Total effect, the effect of the exposure on the NPB. DE:Direct effect: the effect of the exposure on the NPB, not explained by the mediator. IE:intermediary effect, the effect of the exposure on the NPB acting through the mediator. An asterisk (*) denotes significance at P<0.001.

From Interactive Mediation Tests, the direct effect (DE) from the IVW method was calculated with an OR of 1.03 (95% CI, 0.95-1.13; beta = 0.03, SE = 0.04), indicating a significant direct influence of MS on NPB. The intermediary effect (IE), calculated by the IVW method, revealed an OR of 1.08 (95% CI, 1.02-1.14; beta = 0.08, SE = 0.03), suggesting mediation through DD. The total effect (TE) was determined to be an OR of 1.12 (95% CI, 1.04-1.19; beta = 0.11, SE = 0.03). The proportion mediated (IE_div_TE) was 70.29%, with a confidence interval ranging from 0.21 to 1.19. The mediation effect was further validated using the Sobel test, which returned a p-value of 0.005. These findings underscore a clear causal relationship between MS, DD, and NPB.

4.5 Exploring the Causal Impact of Plasma Proteins on MS, DD, and NPB

To investigate potential therapeutic targets, our research utilized multivariable MR to assess the causal relationship between 4,907 plasma proteins (as exposures) and MS as outcomes. Applying a stringent threshold, we initially identified 161 genes after excluding SNPs below the threshold of three effective instruments and where the IVW method p-value exceeded 0.05. Notably, upon applying FDR corrections to minimize false positives, only two genes, ADAM Metallopeptidase Domain 11(ADAM11) and Glutamate Ionotropic Receptor AMPA Type Subunit 4 (GRIA4), met the criteria for inclusion. For ADAM11, the IVW method yielded an OR of 1.470, with a 95% CI of 1.227-1.761, a beta of 0.385, SE of 0.092, and an FDR-adjusted p-value of 0.040. Horizontal pleiotropy assessments through MR-Egger regression and Cochran’s Q statistics did not indicate significant pleiotropic effects (intercept p-value = 0.243; Q p-value = 0.654). Similarly, for GRIA4, the IVW method reported an OR of 1.97, 95% CI of 1.43-2.72, beta of 0.68, SE of 0.16, and an FDR-adjusted p-value of 0.040, with MR-Egger and Cochran’s Q statistics suggesting no significant pleiotropy (intercept p-value = 0.814; Q p-value = 0.585).

In the analysis of plasma proteins versus DD, after rigorous screening, 247 proteins were initially identified. Following FDR corrections, two proteins were significantly associated: C-X-C Motif Chemokine Ligand 13(CXCL13), with an IVW OR of 0.07 (95% CI: 0.03-0.19; beta = -2.63, SE = 0.49, P-value = 6.94x10^-8, P_FDR = 1.31x10^-4), and Protein Kinase C Gamma(PRKCG), with an IVW OR of 1.23 (95% CI: 1.15-1.32; beta = 0.21, SE = 0.04, P-value = 6.32x10^-9, P_FDR = 2.44x10^-5).

Stringent filtering initially identified 91 genes in the plasma protein vs. NPB analysis, However, none remained after applying FDR corrections. Interestingly, relaxing the FDR thresholds highlighted the presence of Glutathione-Disulfide Reductase(GSR) and Ubiquitin Like Modifier Activating Enzyme 2(UBA2) in both the NPB and MS analyses. For UBA2 in the MS group, the IVW method indicated an OR of 1.33 (95% CI, 1.06-1.67; beta = 0.29, SE = 0.15, p-value = 0.014). For GSR, an OR of 0.66 (95% CI, 0.44-1.00; beta = -0.41, SE = 0.21, p-value = 0.049) was noted. In the NPB group, similar results for UBA2 and GSR were observed with corresponding ORs and high FDR values, detailed results are provided in supplementary file 4.

Patients with MS commonly experience a variety of symptoms, among which bladder dysfunction is prevalent^[4]. The autonomic and somatic nervous systems, which govern the lower urinary tract, are influenced respectively by the parasympathetic system facilitating detrusor muscle contraction for urination, and the sympathetic system which mediates urine storage by relaxing the detrusor and contracting the bladder neck and urethra^[36]. MS, an immune-mediated central nervous system disorder, is associated with a high incidence of lower urinary tract symptoms, affecting up to 70% of patients^{[37, 4]}. Additionally, about 10% of patients may exhibit voiding dysfunction at onset^[38]. Our findings reaffirm MS as a risk factor for NPB. Within the scope of our MR analysis, we primarily utilized the IVW method. However, the statistical significance was also confirmed by additional methods. Approximately 80% of individuals diagnosed with MS exhibit neurological symptoms within ten years^{[38, 39]}, likely due to immune cells targeting oligodendrocytes that produce myelin in the central nervous system. This immune attack progressively impairs neurological control over bladder function, leading to NPB^{[2, 40]}. Current clinical research continuously enhances the prevention and treatment strategies concerning MS, DD, and NPB^[41]. Genetic evidence from our studies confirms that neurological damage is a significant contributor to NPB in MS patients, with a mediation effect as high as 70%. This substantial effect size not only reinforces the validity of observational studies but also highlights the stability and superiority of MR in examining causality in two-sample analyses.

Current treatments for NPB due to MS include bladder training, physical therapy, desmopressin sprays, bladder stimulants, and percutaneous tibial nerve stimulation, with anticholinergic drugs as the main pharmacological treatment. Despite these, restoration of normal bladder function remains challenging^[39]. Therefore, exploring new therapeutic options is crucial. Advances in high-throughput genotyping and global research collaboration through GWAS have uncovered minor genetic variations linked to MS susceptibility^{[42, 2]}. Our multivariable MR analysis, stringently filtered through FDR, has revealed significant associations between MS and two genes: ADAM11 and GRIA4.

The gene ADAM11 belongs to the ADAM family of proteins, which are involved in various cellular processes including cell adhesion and proteolysis. Research on mice deficient in ADAM11 indicates impairments in spatial learning and motor coordination, as evident from their performance in tasks such as the Morris water maze and the rotating rod test^[43]. Further studies utilizing octopus knockout and gain-of-function experiments have suggested that ADAM11 acts as a novel regulator of Wnt and BMP4 signaling in cranial neural crest cells^[44], highlighting its crucial role in normal neural functioning and potentially in neurodevelopmental processes^[45]. GRIA4 is a subunit of AMPA receptors, which are ionotropic glutamate receptors that mediate rapid synaptic transmission in the central nervous system. Studies on GRIA4-deficient mouse models have observed significant downregulation of GRIA4 expression due to an insertion of an IAP retrotransposon, which notably affects epilepsy symptoms in these mice^[46]. Moreover, a female patient carrying a GRIA4 mutation exhibited severe developmental delays, spasticity, generalized epileptic seizures, and incomplete retinal and choroidoretinal pigmentation, underscoring the importance of GRIA4 in neural development^[47].

A research identifies CXCL13 as a critical biomarker for diagnosing neurogenic diseases, notably Lyme neuroborreliosis, which represents a neurological manifestation of Lyme disease^[48]. CXCL13 is believed to facilitate the formation and maintenance of ectopic lymphoid follicles found in the meninges of patients with progressive MS^[49]. Meanwhile, PRKCG, a protein kinase with a pronounced impact on neuronal function, is known to affect neurotransmission. Studies indicate that increased activity and altered localization of PRKCG can lead to the neurodegeneration observed in spinocerebellar ataxia type 14^{[50, 51]}. Further clinical research is necessary to determine whether targeted therapies can ameliorate conditions such as MS, DD, and NPB.
Although strict adherence to FDR corrections is standard to avoid type I errors, relaxing these criteria has allowed us to uncover potentially overlooked genetic associations with disease phenotypes. Our findings suggest differential roles for GSR and UBA2 in these conditions, where GSR appears to confer a protective effect, likely due to its central role in maintaining cellular redox homeostasis^[52]. Conversely, UBA2's involvement in the SUMOylation process might exacerbate disease pathologies through its regulatory impact on protein functions essential for neuronal health^[53]. This understanding highlights the need for precise statistical methods to identify the genetic determinants of complex diseases comprehensively. Future research should focus on these genetic pathways to devise treatments that may halt or reverse the progression of neurogenic disorders.

We utilized genetic data robustly to clarify the causal relationships involved, maintaining stringent validation thresholds for exposure and outcome settings (P < 1×10^-8). Our approach included both forward and reverse MR analyses across heterogeneous populations, effectively navigating the complexities typically encountered in genetic epidemiology. We selected cohorts exclusively of European descent from reputable sources, ensuring a homogenized study base that minimizes confounders from environmental and lifestyle factors, thereby enhancing the generalizability of our findings. Our comprehensive strategy, which also involved MR inverse-variance analysis and dual validation across diverse cohorts, ensured consistent and robust results. Addressing SNP-pleiotropy, a common challenge in MR studies, we utilized multiple methods including MR Egger regression, MR-PRESSO, cML, and BWMR. These techniques were chosen for their effectiveness in reducing biases from indirect associations between instrumental variables and external exposures. By carefully analyzing and excluding instrumental variables linked to potential confounders, we strengthened the integrity of our conclusions.

We must acknowledge several inherent limitations that could influence the interpretation of our findings. A primary concern arises from the heterogeneous genetic databases utilized, including the EBI database, FinnGen biobank, and data from Iceland. This diversity in sources introduces significant variability in how diseases are defined across cohorts, potentially affecting the uniformity and reliability of our causal inferences. Moreover, our study population is exclusively European, and the absence of data from other ethnic groups severely limits the generalizability and applicability of our results to broader populations. We have explored the hypothesis of a direct linear correlation between MS and the susceptibility to NPB, prompted by initial analyses suggesting a straightforward connection where slight variations in MS appear to directly influence the odds of developing NPB from DD. However, we remain cognizant of the potential for non-linear dynamics between MS and NPB, which could complicate the interpretation of our findings. This awareness underscores the necessity for future research to dissect more nuanced aspects of this relationship.

Our MR study confirms a direct causal link between MS and NPB, with DD mediating this relationship. We also identified significant associations with ADAM11 and GRIA4 genes related to MS, CXCL13 and PRKCG links to DD, enhancing our understanding of its etiology. Relaxed FDR constraints helped pinpoint GSR and UBA2 as potential contributors to both MS and NPB, underscoring their importance in disease development.

NPB Neuropathic bladder

MS Multiple sclerosis

DD Demyelinating diseases of the central nervous system

MR Mendelian randomization

GWAS Genome-wide association studies

IEU Integrative Epidemiology Unit

LD linkage disequilibrium

IVW Inverse Variance Weighting method

MR-PRESSO Mendelian Randomization Pleiotropy RESidual Sum and Outlier

WM Weighted median method

BWMR Bayesian Weighted Mendelian Randomization

cML Constrained maximum likelihood method

FDR False Discovery Rate

OR Odds ratio

CI Confidence Interval

ADAM11 ADAM Metallopeptidase Domain 11

GRIA4 Glutamate Ionotropic Receptor AMPA Type Subunit 4

CXCL13 C-X-C Motif Chemokine Ligand 13

PRKCG Protein Kinase C Gamma

GSR Glutathione-Disulfide Reductase

UBA2 Ubiquitin Like Modifier Activating Enzyme 2

Acknowledgments

We thank the IEU Open GWAS Project for providing summary results and data essential for our analyses, available at https://gwas.mrcieu.ac.uk/datasets/. We want to acknowledge the participants and investigators of the FinnGen study. Furthermore, we are grateful for the access to GWAS data on MS as reported by Sawcer S et al., NPB as reported by Sakaue S et al., plasma protein data provided by Egil F et al..

Authors' contributions

Y.X. designed the study and, along with H.S., completed the statistical analyses. Y.W. also contributed to these analyses. Y.X., J.X., and X.W. authored the manuscript's first draft, while H.S. and S.X. were responsible for the visualization. Every author played a role in interpreting the data and provided feedback on earlier manuscript drafts. Furthermore, each author reviewed and approved the manuscript's final version.

Funding

This research received funding support from the National Natural Science Foundation of China under grant number 82160145, the 2021 National Natural Science Foundation Post Subsidy Individual Fund of China with reference GPPH-NSFC-2021-10, and the Science and Technology Fund of the Guizhou Health Commission, identified by the grant number gzwkj2021-212.

Availability of data and materials

The datasets analysed during the current study are available in the IEU Open GWAS repository(https://gwas. mrcieu.ac.uk/).

Competing interests

The authors herein affirm that there has been no receipt of financial support from any entity for the work submitted. Furthermore, there have been no financial ties with organizations potentially interested in the submitted work over the past three years. Additionally, no other affiliations or engagements are perceived to influence the integrity of the submitted work.

Ethics approval

Not applicable.

Consent to participate

Not applicable.

Consent for publication

Not Applicable.

Erden E et al (2022) The neurogenic bladder characteristics and treatment approaches in the patients with multiple sclerosis. Mult Scler Relat Disord 58:103439. .https://DOI:10.1016/j.msard.2021.103439
Thompson AJ et al (2018) Multiple sclerosis. Lancet 391(10130):1622–1636. .https://DOI:10.1016/s0140-6736(18)30481-1
Ginsberg D (2013) The epidemiology and pathophysiology of neurogenic bladder. Am J Manag Care 19(10 Suppl):s191–s196. .https://
Phé V, Chartier–Kastler E, Panicker JN (2016) Management of neurogenic bladder in patients with multiple sclerosis. Nat Reviews Urol 13(5):275–288. .https://DOI:10.1038/nrurol.2016.53
Hamid R et al (2018) Epidemiology and pathophysiology of neurogenic bladder after spinal cord injury. World J Urol 36(10):1517–1527. https://DOI:10.1007/s00345-018-2301-z
Yarmolinsky J et al (2018) Causal Inference in Cancer Epidemiology: What Is the Role of Mendelian Randomization? Cancer Epidemiol Biomarkers Prev 27(9):995–1010. https://DOI:10.1158/1055-9965.Epi-17-1177
Bowden J, Davey Smith G, Burgess S (2015) Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol 44(2):512–525. .https://DOI 10.1093/ije/dyv080
Sanderson E et al (2022) Mendelian randomization. Nat Rev Methods Primers 2. https://DOI:10.1038/s43586-021-00092-5
Ben E et al (2020) The MRC IEU OpenGWAS data infrastructure. bioRxiv: p. 2020.08.10.244293.https://DOI:10.1101/2020.08.10.244293
Skrivankova VW et al (2021) Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization: The STROBE-MR Statement. JAMA 326(16):1614–1621. .https://DOI:10.1001/jama.2021.18236
Kurki MI et al (2023) FinnGen provides genetic insights from a well-phenotyped isolated population. Nature 613(7944):508–518. https://DOI:10.1038/s41586-022-05473-8
Sawcer S et al (2011) Genetic risk and a primary role for cell-mediated immune mechanisms in multiple sclerosis. Nature 476(7359):214–219. .https://DOI:10.1038/nature10251
Hemani G et al (2018) The MR-Base platform supports systematic causal inference across the human phenome. Elife 7. https://DOI:10.7554/eLife.34408
Sakaue S et al (2021) A cross-population atlas of genetic associations for 220 human phenotypes. Nat Genet 53(10):1415–1424. https://DOI:10.1038/s41588-021-00931-x
Ferkingstad E et al (2021) Large-scale integration of the plasma proteome with genetics and disease. Nat Genet 53(12):1712–1721. https://DOI:10.1038/s41588-021-00978-w
Rafnar T et al (2004) The Icelandic Cancer Project–a population-wide approach to studying cancer. Nat Rev Cancer 4(6):488–492. .https://DOI 10.1038/nrc1371
Gudbjartsson DF et al (2015) Large-scale whole-genome sequencing of the Icelandic population. Nat Genet 47(5):435–444. .https://DOI 10.1038/ng.3247
Sun BB et al (2018) Genomic atlas of the human plasma proteome. Nature 558(7708):73–79. https://DOI:10.1038/s41586-018-0175-2
Folkersen L et al (2020) Genomic and drug target evaluation of 90 cardiovascular proteins in 30,931 individuals. Nat Metab 2(10):1135–1148. https://DOI:10.1038/s42255-020-00287-2
Li L et al (2022) Mendelian randomization study of the genetic interaction between psoriasis and celiac disease. Sci Rep 12(1):21508. https://DOI:10.1038/s41598-022-25217-y
Burgess S, Thompson SG (2011) Avoiding bias from weak instruments in Mendelian randomization studies. Int J Epidemiol 40(3):755–764. .https://DOI 10.1093/ije/dyr036
Levin MG et al (2020) Genetics of height and risk of atrial fibrillation: A Mendelian randomization study. PLoS Med 17(10). p. e1003288.https://DOI:10.1371/journal.pmed.1003288
Huang W et al (2021a) Association of lipid-lowering drugs with COVID-19 outcomes from a Mendelian randomization study. Elife. 10.https://DOI:10.7554/eLife.73873
Hemani G, Tilling K, Davey G (2017) Smith Orienting the causal relationship between imprecisely measured traits using GWAS summary data. PLoS Genet. 13(11): p. e1007081.https://DOI:10.1371/journal.pgen.1007081
Yavorska OO, Burgess S (2017) MendelianRandomization: an R package for performing Mendelian randomization analyses using summarized data. Int J Epidemiol 46(6):1734–1739. .https://DOI:10.1093/ije/dyx034
Burgess S et al (2015) Using published data in Mendelian randomization: a blueprint for efficient identification of causal risk factors. Eur J Epidemiol 30(7):543–552. .https://DOI 10.1007/s10654-015-0011-z
Verbanck M et al (2018) Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 50(5):693–698. .https://DOI:10.1038/s41588-018-0099-7
Bowden J et al (2016) Consistent Estimation in Mendelian Randomization with Some Invalid Instruments Using a Weighted Median Estimator. Genet Epidemiol 40(4):304–314. .https://DOI 10.1002/gepi.21965
Zhao J et al (2020) Bayesian weighted Mendelian randomization for causal inference based on summary statistics. Bioinformatics 36(5):1501–1508. .https://DOI:10.1093/bioinformatics/btz749
Yin Q, Zhu L (2024) Does co-localization analysis reinforce the results of Mendelian randomization? Brain 147(1):e7–e8. .https://DOI 10.1093/brain/awad295
Burgess S, Butterworth A, Thompson SG (2013) Mendelian randomization analysis with multiple genetic variants using summarized data. Genet Epidemiol 37(7):658–665. .https://DOI 10.1002/gepi.21758
Lin SH, Brown DW, Machiela MJ (2020) LDtrait: An Online Tool for Identifying Published Phenotype Associations in Linkage Disequilibrium. Cancer Res 80(16):3443–3446. .https://DOI:10.1158/0008-5472.Can-20-0985
Machiela MJ, Chanock SJ (2015) LDlink: a web-based application for exploring population-specific haplotype structure and linking correlated alleles of possible functional variants. Bioinformatics 31(21):3555–3557. https://DOI:10.1093/bioinformatics/btv402
Patel A et al (2023) MendelianRandomization v0.9.0: updates to an R package for performing Mendelian randomization analyses using summarized data. Wellcome Open Res 8:449. .https://DOI:10.12688/wellcomeopenres.19995.1
Glickman ME, Rao SR, Schultz MR (2014) False discovery rate control is a recommended alternative to Bonferroni-type adjustments in health studies. J Clin Epidemiol 67(8):850–857. .https://DOI:10.1016/j.jclinepi.2014.03.012
Sakakibara R (2019) Neurogenic lower urinary tract dysfunction in multiple sclerosis, neuromyelitis optica, and related disorders. Clin Auton Res 29(3):313–320. https://DOI:10.1007/s10286-018-0551-x
Panicker JN, Fowler CJ (2015) Lower urinary tract dysfunction in patients with multiple sclerosis. Handb Clin Neurol. 130: pp. 371 – 81.https://10.1016/b978-0-444-63247-0.00021-3
Motta R, de Carvalho ML (2008) Management of bladder dysfunction in multiple sclerosis patients: the nurse's point of view. Neurol Sci 29:S356. -9.https://DOI:10.1007/s10072-008-1043-x
Rahnama'i MS (2020) Neuromodulation for functional bladder disorders in patients with multiple sclerosis. Mult Scler 26(11):1274–1280. https://DOI:10.1177/1352458519894714
Trapp BD, Nave KA (2008) Multiple sclerosis: an immune or neurodegenerative disorder? Annu Rev Neurosci 31:247–269. .https://DOI 10.1146/annurev.neuro.30.051606.094313
Panicker JN, Fowler CJ, Kessler TM (2015) Lower urinary tract dysfunction in the neurological patient: clinical assessment and management. Lancet Neurol 14(7):720–732. .https://DOI 10.1016/s1474-4422(15)00070-8
Hafler DA et al (2007) Risk alleles for multiple sclerosis identified by a genomewide study. N Engl J Med 357(9):851–862. .https://DOI 10.1056/NEJMoa073493
Takahashi E et al (2006) Deficits in spatial learning and motor coordination in ADAM11-deficient mice. BMC Neurosci 7:19. https://DOI:10.1186/1471-2202-7-19
Pandey A et al (2023) ADAM11 a novel regulator of Wnt and BMP4 signaling in neural crest and cancer. Front Cell Dev Biol 11:1271178. .https://DOI:10.3389/fcell.2023.1271178
Zhong S, Khalil RA (2019) A Disintegrin and Metalloproteinase (ADAM) and ADAM with thrombospondin motifs (ADAMTS) family in vascular biology and disease. Biochem Pharmacol 164:188–204. .https://DOI:10.1016/j.bcp.2019.03.033
Frankel WN et al (2014) Unraveling genetic modifiers in the gria4 mouse model of absence epilepsy. PLoS Genet 10(7):e1004454. .https://DOI 10.1371/journal.pgen.1004454
Wang H et al (2022) Novel Heterozygous Missense Variant in GRIA4 Gene Associated With Neurodevelopmental Disorder With or Without Seizures and Gait Abnormalities. Front Genet 13:859140. .https://DOI:10.3389/fgene.2022.859140
Kowarik MC et al (2012) CXCL13 is the major determinant for B cell recruitment to the CSF during neuroinflammation. J Neuroinflammation 9:93. .https://DOI:10.1186/1742-2094-9-93
Huang MW, Stock AD, Putterman C (2021b) CXCL13 Neutralization Attenuates Neuropsychiatric Manifestations in Lupus-Prone Mice. Front Immunol 12:763065. .https://DOI:10.3389/fimmu.2021.763065
Tada Y et al (2022) Comparison of two families with and without ataxia harboring novel variants in PRKCG. J Hum Genet 67(10):595–599. .https://DOI:10.1038/s10038-022-01057-6
Wong MMK et al (2018) Neurodegeneration in SCA14 is associated with increased PKCγ kinase activity, mislocalization and aggregation. Acta Neuropathol Commun 6(1):99. https://DOI:10.1186/s40478-018-0600-7
Domanskyi A, Parlato R (2022) Oxidative Stress in Neurodegenerative Diseases. Antioxid (Basel) 11(3). https://DOI:10.3390/antiox11030504
Mandel N, Agarwal N (2022) Role of SUMOylation in Neurodegenerative Diseases. Cells 11(21). https://DOI:10.3390/cells11213395

No competing interests reported.

SupplementaryfilesVersion3.xlsx

Download PDF

Editorial decision: Revision requested
12 Aug, 2024
Reviews received at journal
11 Aug, 2024
Reviewers agreed at journal
03 Aug, 2024
Reviews received at journal
09 Jun, 2024
Reviewers agreed at journal
02 Jun, 2024
Reviewers invited by journal
24 May, 2024
Submission checks completed at journal
15 May, 2024
Editor assigned by journal
15 May, 2024
First submitted to journal
28 Apr, 2024

You are reading this latest preprint version

Uncovering Potential Genetic Targets in the Mediating Role of Demyelinating Disease for Multiple Sclerosis-Induced Neuropathic Bladder: A Mendelian Randomization Analysis

Status:

Version 1

Abstract

Figures

Introduction

Methods

3.1 Study overview

3.2 Data source

3.3 Selection of genetic instruments

3.4 Mendelian randomization analysis and sensitivity analysis

3.5 Statistics

Results

Discussion

Conclusion

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1