Journal of Multiple Sclerosis

Pleiotropic Genetic Effects between Multiple Sclerosis and Musculoskeletal Traits

Author info »

Introduction

Multiple sclerosis (MS) is a demyelinating inflammatory autoimmune disorder presenting continuous and diffuse changes in the white and grey matter, breakdown of myelin, and damage to axons [1]. Despite recent progress characterized by the advent of new disease-modifying therapies, patients with MS still show an increased risk of musculoskeletal disorders such asfractures [2, 3], accidents and falls [4], and frailty [5] compared to controls.

A meta-analysis comprising 9 cohort studies proposed significantly increased fracture risk in MS by 1.58 times compared to controls [6]. Falls are a debilitating consequence of MS and 56% of MS patients experience falls in any 3-month period [4]. Frailty is a marker of poor prognosis in patients suffering from systemic sclerosis [5] and a significantly higher percentage of MS patients were frail compared to controls (28% vs 8%) [5].

Multiple attributable risk factors of musculoskeletal disorders of MS have been proposed, including secondary causes delineating systemic bone loss by chronic inflammatory status [7], muscle weakness due to immobility [8], and ataxia from disruption of neuronal impulses [9], as well as bone and muscle loss due to MS treatment such as corticosteroids use [10], and psychotropic drugs (antidepressants, anxiolytics, and anti-epileptics) use [11, 12], which make MS patients vulnerable to falls. Beyond the secondary causes, intrinsic factors of MS etiology might be associated with the increased musculoskeletal disorders such that early-stage comparatively young MS patients with minimal or no physical disability also presented low BMD [13].

However, most of the findings were typical observational studies not precluding potential biases from undefined residual confounding and reverse causation. Therefore, the elucidation of the effects of genetically unmodifiable predisposing factors will give further insights into preventative strategies for MS patients’ care.

In this context, LDSR is a reliable and efficient method of using GWAS summary-level data to estimate the genetic correlation between different phenotypes [14]. In addition, MR is a widely used method using measured variation in genes (genotypes, SNPs) of exposure to examine the causal effect of exposures on disease outcomes in observational studies [15, 16]. We implemented LD score regression as well as univariable, multivariable and bidirectional MR, and estimated genetic pleiotropic effect (correlation) and direct MS effects to 3 musculoskeletal outcomes (fracture, falls, and frailty) adjusting for the confounding effects from well-established bone and muscle disorder-related risk factors such as bone mineral density (BMD), lean body mass (whole body fat-free mass, appendicular lean mass), grip strengths and vitamin D.

We performed the following analyses:

(1) LDSR to assess genetic heritability and genetic pleiotropy (correlations) between MS and musculoskeletal traits;

(2) univariable MR to assess the causal effect of genetic determinants of MS on fracture, falls, frailty, and multiple risk factors;

(3) multivariable MR to evaluate independent specific effects of MS to fracture, falls, frailty adjusting for other contributing risk factors;

(4) bidirectional MR to assess the directionality among fracture, falls, frailty.

Methods

Linkage Disequilibrium Score Regression (LDSR)

We used the LD Hub interface which provides an automatic LD score regression analysis pipeline for users [17]. To standardize the input file, quality control is automatically performed on the uploaded file in LD Hub. In order to restrict the analysis to well-imputed SNPs, LD Hub filters the uploaded SNPs to HapMap3 SNPs [18] with 1000 Genomes EUR MAF above 5%, which tend to be well-imputed in most studies. In addition, to make the estimates of the genetic correlation to be reliable, the recent MS GWAS metaanalysis of the International Multiple Sclerosis Genetics Consortium (IMSGC) comprising 47,429 cases ad 68,374 controls [19] needs to meet the following criteria;

1) Z score is at least > 1.5 (optimal > 4),

2) Mean Chi-square of the test statistics > 1.02, and

3) the intercept estimated from the SNP heritability analysis is between 0.9 and 1.1 [17].

We used UK Biobank GWAS datasets provided by LD hub pipeline estimating genetic heritability and correlation with MS.

Mendelian Randomization (MR)

Supplementary Table 1 describes the detailed data source used in MR study. The largest publicly available GWAS summary statistics were used. Most  GWAS summary statistics data were downloaded from the “MRC IEU OpenGWAS database” repository (https://gwas.mrcieu.ac.uk/) or each GWAS consortia. The overall study design for MR is displayed in Figure 1

Supplementary Table 1. GWAS summary statistics used in Mendelian Randomization

Figure 1: Conceptual model of univariable and multivariable mendelian randomization

Exposure data

Instrumental variables (IVs) for MS are obtained from the recent GWAS metaanalysis of the IMSGC [19]. Among the 233 genome-wide statistically independent associations with MS susceptibility, 200 effects (SNPs) located in the autosomal non-MHC genome were selected as IVs Supplementary Table 2. The detailed analysis and results description is provided in the previous IMSGC publication [19]. SNPs of genome-wide and suggestive effects jointly explain ~48% of the estimated heritability for MS [19].

Supplementary Table 2. 200 MS SNPs located in the autosomal non-MHC genome from International Multiple Sclerosis Genetics Consortium

Outcome data

Fracture and falls GWAS data used in the analysis were selected from the UK Biobank data, fractured/broken in last 5 years (ID: ukb-b-13346, n=44,502 cases and 415,887 controls) and falls last year (ID: ukb-b-2535, n=461,725). Frailty GWAS data was from Atkins et al, 2019 (n=175,226) [20,21].

Risk factors of musculoskeletal outcomes

Known bone and muscle strength-related factors were used as covariates in MVMR.They are eBMD (heel)[22], appendicular lean mass, whole body fat-free mass, right handgrip strength, left grip strength [23] vitamin D [24], frailty [21], and falls [25]

Statistical analysis

Univariable MR between MS and three musculoskeletal traits as well as risk factors:

Univariable MR was applied to assess the causal relation between MS and 3 musculoskeletal outcomes (fracture, falls, and frailty) and 5 risk factors. Furthermore, MR effects of musculoskeletal risk factors on 3 musculoskeletal outcomes were assessed to examine the potential genetically determined risk paths of 3 musculoskeletal outcomes. Multi- instrument MR approaches were applied, which were to perform a single instrumental variable MR first; and then combined MR results from multiple instrumental variables via fixed-effect meta-analysis to estimate the overall causal relationships

MVMR between MS and three musculoskeletal outcomes adjusting for risk factors:

We also applied regression-based MVMR to account for many variants having pleiotropic effects that are associated with multiple risk factors in order to provide coefficients (effect sizes) representing the direct causal effects of MS to three musculoskeletal outcomes in turn with the other risk factors being fixed. MVMR is an extension of MR to deal with genetic variants that are associated with multiple risk factors. We selected risk factors that are known to be biologically related to musculoskeletal traits and where there is a potential network of causal effects (mediation) from one risk factor to another. The causal estimate is obtained by regression of the associations with the outcome on the associations with the risk factors, with the intercept set to zero and weights being the inverse-variances of the associations with the outcome [26]. Risk factors presenting significant univariable MR association with each musculoskeletal trait were included in MVMR.

Bidirectional MR among the three musculoskeletal outcomes

Bidirectional MR was performed among fracture, fall and frailty to examine their mutual effects and directionality considering their highly correlative nature. In bidirectional MR, instruments for both exposure and outcome are used to evaluate whether the “exposure” variable causes the “outcome” or whether the “outcome” variable causes the “exposure”[27].

Heterogeneity analysis

We used Cochran’s Q statistic to check the heterogeneity of IVs. Heterogeneity indicates a possible violation of the necessary IVs or modeling assumptions of which pleiotropy is a likely major cause. Cochran’s Q statistic derived from the inverse variance weighted (IVW) estimate shoud follow a χ2 distribution with degrees of freedom equal to the number of SNPs minus 1. Excessive heterogeneity is an indication that either the modeling assumptions have been violated, or that some of the genetic variants violate the IV assumptions—e.g. by exerting a direct effect on the outcome, not through the exposure [28]. This is termed ’horizontal pleiotropy’ [29]

When Q statistics showed significant heterogeneity, we applied MR-PRESSO method to reassess the MR effects after filtering out outlier SNPs and compared the fixed MR results with IVW-based results [29].

Sensitivity analysis

The sensitivity analysis including IVW, weighted median, weighted mode, and MR-Egger tests were performed in univariable MR analysis. For MVMR, IVW and MR-Egger tests were evaluated. To exclude instrumental variables that may have horizontal pleiotropy and cause false positive findings of the causal inference, we also applied the MR-Egger regression test [30] to verify horizontal pleiotropy in each MR. Among the multiple methods, we set the results from IVW and MR-PRESSO outlier removal results as primary findings.

All the MR analyses were conducted in R (version 4.0.0) using TwoSampleMR, MendelianRandomizatio, MRPRESSO packages, and the MR-Base platform [31-33]. The causal effect size (beta) with standard error and p-values were presented as appropriately. We applied a false discovery rate (FDR) to correct for multiple testing. The criteria used to select final instrument variables in each MR were GWAS significance: p<5e-8, LD clumping: r2>0.001 within a 10MB window, and proxy SNPs in 1000 Genomes EUR r2=0.8 [33].

Results

Genetic heritability and correlations by LDSR

Regression weight LD Scores for 1,293,150 SNPs of MS GWAS data were read. After merging with reference panel LD, a total of 1,109,876 SNPs remained for the final analysis. LDSR demonstrated that MS has statistically significant moderate genetic correlation with falls (rG: 0.105, p=0.010) but not with fracture (rG: -0.017, p=0.711) or frailty (rG: 0.151, p=0.082). Both handgrip strengths and both leg fat-free masses also showed a genetically significant correlation with MS. However, the whole-body fat free mass didn’t have a significant correlation with MS. The heritability of these traits estimated from GWAS summary statistics was low or considered not high; fracture (h2:0.019), falls (h2: 0.033), and frailty (h2: 0.015) Table 1. We found genetic variant rs13191659 mapped to LINC00240 (Long Intergenic NonProtein Coding RNA 240) gene is commonly associated with MS and fall (MS GWAS p <5e-8 and fall GWAS<5e-5). LINC00240 is known for total iron binding capacity [37], folic acid deficiency anemia and mean corpuscular volume [38].

Table 1. Genetic correlations between MS and musculoskeletal phenotypes and heritability by LD score regression

Mendelian Randomization

The univariable MR analysis showed a minimal causal effect of the total 109 MS SNPs to fracture, falls, and frailty in the IVW method. After excluding SNP outliers by MR-PRESSO and applying multiple testing corrections, without meaningful ORs, MS doesn’t seem to causally increase risks of falls (OR:1.004, 95% CI: 1.001-1.006, q=0.018); fracture (OR:1.002, 95% CI:1.000-1.003, q=0.056) and frailty score (beta:0.009, 95% CI: 0.001- 0.017, q=0.147) and none of the univariable MR effects between MS and musculoskeletal risk factors were significant Table 2 and Supplementary Table 3. Sensitivity analyses with different MR methods such as MR Egger,Weighted median, and Weighted mode methods didn’t replicate the significant results in all the MR analyses. All pleiotropy tests (MR Egger intercept test) were insignificant Supplementary Table 3. Single SNP MR effects between MS and falls are presented in Supplementary Table 4 and 5. In MVMR analysis, except for adjustment of frailty, adjustment of each risk factor didn’t modify the overall MR results between MS and fall Table 3. MVMR between MS and fracture as well as MS and frailty presented similar results that adjustment for each risk factor had no meaningful modification in overall MR results Supplementary Table 6 and 7.

Table 2. Univariable MR effects between exposure (MS) and outcomes (musculoskeletal outcomes and risk factors)

Supplementary Table 3. Sensitivity analsys: MR results between MS and musculoskeletal outcomes and risk factors

Supplementary Table 4. Harmonized MR results between MS(exposure) and falls (outcomes)

Supplementary Table 5. Single SNP effects between MS and falls

Supplementary Table 6. MVMR between MS and fracture adjusting for risk factors

Supplementary Table 7. MVMR between MS and frailty adjusting for musculskeletal risk factors

Table 3. Multivariable MR between MS and fall adjusting for musculoskeletal risk factor

To account for the directional effects among the 3 musculoskeletal outcomes; fracture, fall and frailty, we conducted bidirectional MR. As expected, Fall causally increased fracture risk (OR:1.215, 95% CI: 1.098- 1.345, p=1.63E-04) and frailty risk (OR: 4.870, 95% CI: 2.412-9.832, p=1.01E-05).

Discussion

In this study, we found genetic correlation with shared pleotropic effects between MS and fall proven by LDSR and MR with minimal effect size. However, we don’t find meaningful causal relation to support epidemiological observation with increased musculoskeletal traits in MS patients. The MS doesn’t causally increase risks of musculoskeletal traits. Based on the univariable MR between MS and musculoskeletal risk factors, none of the well-known muscle and bone-related risk factors (whole-body and appendicular fat-free masses, both grip strengths, heel BMD and vitamin D) were causally affected by MS. In bidirectional MR among the three musculoskeletal outcomes, fall causally increased fracture and frailty risk. Taking into all these findings together, we didn’t provide evidence to support that MS causally increased risks of neither musculoskeletal traits nor other musculoskeletal risk factors. The widely observed musculoskeletal disorders in MS patients in clinical studies might be more likely due to secondary factors that are associated with MS disease progression and treatment. When we further explored SNPs commonly found in MS and fall with the criteria of GWAS/suggestive GWA significant variants, genetic variants mapped to a gene associated with iron biomarkers was found. Therefore, we could hypothesize MS and falls share pleiotropic genetic effects in iron biomarker pathways. It is well known that there is a correlation between anemia and falls, and specifically patients with lower hemoglobin levels are more likely to fall [36]. MRI and histological studies also suggest that iron levels are dysregulated in MS: iron accumulates in grey matter and is depleted in normal-appearing white matter [37].

Most of the previous observational studies predominantly reported high fracture risk in MS in comparison to age-matched controls. Dyn one study, Bazelier et al, 2011 [38] reported 3-fold increased hip fracture risk in patients with MS during 5 years of follow-up. However, their baseline fractures were higher in MS than age, and sex-matched controls, furthermore the baseline falls history was higher in MS. Therefore, increased fractures observed in MS patients in this study might have been via increased fall risk in this population. Falls are common in people with MS, with a large international data set demonstrating that 56% MS individuals fall at least once within 3 months’ period with 37% categorized as frequent fallers [4]. Notably, MS patients fall more frequently and experience more injurious falls, and are more likely to attribute their falls to tripping and distraction [39].

Medication use is a well-established risk factor for falls, either directly (affecting balance, attention, or muscle tone) or indirectly (as proxies of underlying conditions influencing the risk of falling; joint pain, arthrosis, and cardiovascular diseases among many others). A recent falls GWAS study examined the genetic correlation of falls with medication use and found medications such as opioids, anti-inflammatory and anti-rheumatic drugs, anilides, and drugs for peptic ulcer and gastro-esophageal reflux disease were in positive genetic correlations with falls [40]. Medication use associated with MS treatments and complications should be one of the strong risk factors for falls in this population. However, we are not able to evaluate such hypothesis based on our current study design.

Based on our study, genetic factors of MS might predispose MS patients to increased risk of falls by shared genetic pleiotropy but with minimal effects. Therefore, taking into account the secondary environmental or treatmentrelated factors such as careful medication use and incorporating stringent fall prevention strategy in patient care will mitigate the undesirable muscle and bone effects in MS patients

We have a few novelties and strengths to mention. Given the highly increased availability of public GWAS data and identification of thousands of traitassociated loci in each trait, we could utilize LDSR and a multifaceted MR approach together and could elucidate the stepwise causal relations, and draw new perspectives on musculoskeletal traits in MS patients. We also illuminated genetically predisposed iron biomarker alteration might have a role in fall risk in MS patients. However, we have some limitations. Due to the low heritability of musculoskeletal traits and risk factors Table 1, incorporating MR analysis might induce limited implications as well as GWAS data from these phenotypes might not highly reliable to elucidate any causal inference

In conclusion, genetic association between MS and falls still stands, suggesting there are shared pleiotropic genetic factors between these two phenotypes. However, we didn’t establish a causal relationship between MS and multiple musculoskeletal traits which were reported by observational studies.

References

1. Lassmann, Hans. "Pathogenic mechanisms associated with different clinical courses of multiple sclerosis." Front Immunol. 9 (2019): 3116. [Google Scholar] [Crossref]
2. Hearn, Andrew P., and Eli Silber. "Osteoporosis in multiple sclerosis." Mult Scler J. 16.9 (2010): 1031-1043.[Google Scholar] [Crossref]
3. Marrie, Ruth Ann, et al. "A cross-sectional study of bone health in multiple sclerosis." Neurology 73.17 (2009): 1394-1398.[Google Scholar] [Crossref]
4. Nilsagård, Ylva, et al. "Falls in people with MS—an individual data meta-analysis from studies from Australia, Sweden, United Kingdom and the United States." Mult Scler J 21.1 (2015): 92-100.[Google Scholar] [Crossref]
5. Ayrignac, Xavier, et al. "Frailty in ageing persons with multiple sclerosis." Mult Scler J 27.4 (2021): 613-620.[Google Scholar] [Crossref]
6. Dong, Guixian, et al. "Multiple sclerosis increases fracture risk: a meta-analysis." BioMed Res Int. 2015 (2015).[Google Scholar] [Crossref]
7. Schett, Georg, and Jean-Pierre David. "The multiple faces of autoimmune-mediated bone loss." Nat Rev Endocrinol. 6.12 (2010): 698-706. [Google Scholar] [Crossref]
8. Formica, C. A., et al. "Reduced bone mass and fat-free mass in women with multiple sclerosis: effects of ambulatory status and glucocorticoid use." Calcif Tissue Int. 61 (1997): 129-133. [Google Scholar] [Crossref]
9. Josyula, Sowmya, et al. The nervous system's potential role in multiple sclerosis associated bone loss. J Neurol Sci 319.1-2 (2012): 8-14. [Google Scholar] [Crossref]
10. Dovio, Andrea, et al. "Immediate fall of bone formation and transient increase of bone resorption in the course of high-dose, short-term glucocorticoid therapy in young patients with multiple sclerosis." J Clin Endocrinol Metab. 89.10 (2004): 4923-4928. [Google Scholar] [Crossref]
11. Bazelier, Marloes T., et al. "Risk of fractures in patients with multiple sclerosis: a population-based cohort study." Neurology 78.24 (2012): 1967-1973. [Google Scholar] [Crossref]
12. Bazelier, Marloes T., et al. "The risk of fracture in patients with multiple sclerosis: the UK general practice research." J Bone Miner Res. 26.9 (2011): 2271-2279. [Google Scholar] [Crossref]
13. Moen, S. M., et al. "Low bone mass in newly diagnosed multiple sclerosis and clinically isolated syndrome." Neurology 77.2 (2011): 151-157. [Google Scholar] [Crossref]
14. Bulik-Sullivan, Brendan K., et al. "LD Score regression distinguishes confounding from polygenicity in genome-wide association studies." Nat Genet 47.3 (2015): 291-295. [Google Scholar] [Crossref]
15. Davey Smith, George, and Shah Ebrahim. "‘Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease?." Int J Epidemiol. 32.1 (2003): 1-22. [Google Scholar] [Crossref]
16. Evans, David M., and George Davey Smith. "Mendelian randomization: new applications in the coming age of hypothesis-free causality." Annu Rev Genom Hum Genet. 16 (2015): 327-350. [Google Scholar] [Crossref]
17. Zheng, Jie, et al. "LD Hub: a centralized database and web interface to perform LD score regression that maximizes the potential of summary level GWAS data for SNP heritability and genetic correlation analysis." Bioinformatics 33.2 (2017): 272-279. [Google Scholar] [Crossref]
18. Altshuler, DM, et al. “(2010) Integrating common and rare genetic variation in diverse human populations.” Nature 467.7311 (2010): 52-58. [Google Scholar]

    [Crossref]

19. International Multiple Sclerosis Genetics Consortium*†, et al. "Multiple sclerosis genomic map implicates peripheral immune cells and microglia in susceptibility." Science 365.6460 (2019): eaav7188. [Google Scholar] [Crossref]
20. Li, Guowei, et al. "An overview of osteoporosis and frailty in the elderly." BMC Musculoskelet Disord. 18 (2017): 1-5. [Google Scholar] [Crossref]
21. Kemp, John P., et al. "Identification of 153 new loci associated with heel bone mineral density and functional involvement of GPC6 in osteoporosis." Nat Genet. 49.10 (2017): 1468-1475.[Google Scholar] [Crossref]
22. Taekema, Diana G., et al. "Handgrip strength as a predictor of functional, psychological and social health. A prospective population-based study among the oldest old." Age Ageing 39.3 (2010): 331-337.[Google Scholar] [Crossref]
23. Hanwell, Heather EC, and Brenda Banwell. "Assessment of evidence for a protective role of vitamin D in multiple sclerosis." Biochim Biophys Acta (BBA)-Mol Basis Dis. 1812.2 (2011): 202-212.[Google Scholar] [Crossref]
24. Panel on Prevention of Falls in Older Persons, American Geriatrics Society and British Geriatrics Society. "Summary of the updated American Geriatrics Society/British Geriatrics Society clinical practice guideline for prevention of falls in older persons." J Am Geriatr Soc. 59.1 (2011): 148-157.[Google Scholar] [Crossref]
25. Burgess, Stephen, and Simon G. Thompson. "Multivariable Mendelian randomization: the use of pleiotropic genetic variants to estimate causal effects." Am J Epidemiol. 181.4 (2015): 251-260. [Google Scholar] [Crossref]
26. Evans, David M., and George Davey Smith. "Mendelian randomization: new applications in the coming age of hypothesis-free causality." Annu Rev Genom Hum Genet. 16 (2015): 327-350. [Google Scholar] [Crossref]
27. Greco M, Fabiola Del, et al. "Detecting pleiotropy in Mendelian randomisation studies with summary data and a continuous outcome." Stat Med. 34.21 (2015): 2926-2940.[Google Scholar] [Crossref]
28. Davey Smith, George, and Gibran Hemani. "Mendelian randomization: genetic anchors for causal inference in epidemiological studies." Hum Mol Genet. 23.R1 (2014): R89-R98.[Google Scholar] [Crossref]
29. Verbanck, Marie, et al. "Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases." Nat Genet. 50.5 (2018): 693-698.[Google Scholar] [Crossref]
30. Burgess, Stephen, and Simon G. Thompson. "Interpreting findings from Mendelian randomization using the MR-Egger method." Eur J Epidemiol. 32 (2017): 377-389.[Google Scholar] [Crossref]
31. Hemani, G., et al. "Mendelian randomization with GWAS summary data." (2020). [Google Scholar]
32. Verbanck, Marie, et al. "Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases." Nat. Genet. 50.5 (2018): 693-698. [Google Scholar] [Crossref]
33. Hemani, Gibran, et al. "The MR-Base platform supports systematic causal inference across the human phenome." eLife 7 (2018): e34408. [Google Scholar] [Crossref]
34. D’Agostino Sr, Ralph B., et al. "General cardiovascular risk profile for use in primary care: the Framingham Heart Study." Circulation 117.6 (2008): 743-753. [Google Scholar] [Crossref]
35. Raffield, Laura M., et al. "Genome-wide association study of iron traits and relation to diabetes in the Hispanic Community Health Study/Study of Latinos (HCHS/SOL): potential genomic intersection of iron and glucose regulation?" Hum Mol Genet. 26.10 (2017): 1966-1978. [Google Scholar] [Crossref]
36. Dharmarajan, T. S., Sai Avula, and Edward P. Norkus. "Anemia increases risk for falls in hospitalized older adults: an evaluation of falls in 362 hospitalized, ambulatory, long-term care, and community patients." J Am Med Dir Assoc. 7.5 (2006): 287-293. [Google Scholar] [Crossref]
37. Stephenson, Erin, et al. "Iron in multiple sclerosis: roles in neurodegeneration and repair." Nat Rev Neurol. 10.8 (2014): 459-468. [Google Scholar] [Crossref]
38. Bazelier, Marloes T., et al. "The risk of fracture in patients with multiple sclerosis: the UK general practice research." J Bone Miner Res. 26.9 (2011): 2271-2279. [Google Scholar] [Crossref]
39. Mazumder, Rajarshi, et al. "Falls in people with multiple sclerosis compared with falls in healthy controls." PloS One 9.9 (2014): e107620. [Google Scholar] [Crossref]
40. Trajanoska, Katerina, et al. "Genetic basis of falling risk susceptibility in the UK Biobank Study." Commun Biol. 3.1 (2020): 543. [Google Scholar] [Crossref]