Abstract
While the genetics of MS risk susceptibility are well-described, and recent progress has been made on the genetics of disease severity, the genetics of disease progression remain elusive. We therefore investigated the genetic determinants of MS progression on longitudinal brain MRI: change in brain volume (BV) and change in T2 lesion volume (T2LV), reflecting progressive tissue loss and increasing disease burden, respectively. We performed genome-wide association studies of change in BV (N = 3401) and change in T2LV (N = 3513) across six randomized clinical trials from Biogen and Roche/Genentech: ADVANCE, ASCEND, DECIDE, OPERA I & II, and ORATORIO. Analyses were adjusted for randomized treatment arm, age, sex, and ancestry. Results were pooled in a meta-analysis, and were evaluated for enrichment of MS risk variants. Variant colocalization and cell-specific expression analyses were performed using published cohorts. The strongest peaks were in PTPRD (rs77321193-C/A, p = 3.9 × 10–7) for BV change, and NEDD4L (rs11398377-GC/G, p = 9.3 × 10–8) for T2LV change. Evidence of colocalization was observed for NEDD4L, and both genes showed increased expression in neuronal and/or glial populations. No association between MS risk variants and MRI outcomes was observed. In this unique, precompetitive industry partnership, we report putative regions of interest in the neurodevelopmental gene PTPRD, and the ubiquitin ligase gene NEDD4L. These findings are distinct from known MS risk genetics, indicating an added role for genetic progression analyses and informing drug discovery.
Similar content being viewed by others
Introduction
Multiple sclerosis (MS) is an immune-mediated disorder of the central nervous system (CNS) characterized by clinical relapses with associated acute focal brain inflammation, and progressive worsening of disability with accompanying neurodegeneration. Disease outcomes are highly variable in persons with MS (PwMS)1 and injury contributing to disease occurs across the disease course, whether in the presence or absence of relapses2. While current MS disease-modifying therapies (DMTs) effectively suppress acute disease activity, development of novel treatments that effectively reduce disability progression remains an unmet need.
There is a strong genetic component for MS risk, with > 200 confirmed genome-wide significant variants which, along with variants in the extended major histocompatibility complex (MHC) region, account for approximately 38.2% of the overall heritability of MS3. These loci primarily affect the immune system, supporting the autoimmune hypothesis of MS pathogenesis and the use of available immune-modulating DMTs. Recently, a large MS study highlighted the importance of CNS resilience for severity4, and furthermore, did not observe overlap between severity and susceptibility loci. However, little is known about the genetics of MS progression, which may also differ from genetics of MS susceptibility. The paucity of progression genetics knowledge could possibly be due to the lack of meaningful longitudinal data at scale. Previous studies have used cross-sectional imaging traits and non-linear MS disability and severity scores to investigate MS severity5,6,7,8. Additionally, measuring MS disease progression using clinical outcomes is challenging. The Expanded Disability Severity Score (EDSS) is commonly used to measure worsening, but only partially captures underlying insidious injury9. Similarly, clinical relapses in the early stages of MS do not predict long-term outcomes in natural history studies10.
Therefore, we leveraged longitudinal imaging data from a unique dataset of randomized controlled trials (RCTs) conducted by Biogen and Roche/Genentech in a precompetitive partnership to examine the genetics of MS progression. We investigated genetic correlates of two longitudinal, quantitative brain MRI outcomes: change in brain volume (BV), and change in T2 lesion volume (T2LV). BV change measures the overall impact of MS injury, including demyelination and neurodegeneration11. The rate of brain atrophy over time is a quantitative measure that correlates with worsening clinical disability12,13. T2LV increases with accumulating injury and early measures of T2LV may be predictive of long-term disability14,15. A deeper understanding of the genetic factors that impact the rate of MS progression could reveal novel candidate targets and biomarkers relevant to MS progressive biology.
Materials and methods
Standard protocol approvals, registrations, and patient consents
We performed GWAS of BV change (N = 3401) and T2LV change (N = 3513) across six RCTs: ADVANCE (trial registry number: NCT00906399), ASCEND (trial registry number: NCT01416181), DECIDE (trial registry number: NCT01064401), OPERA I & II (trial registry number: NCT01247324, NCT01412333), and ORATORIO (trial registry number: NCT01194570) (eTables 1, 2). All participants provided written informed consent that covered the scope of this research. Ethical approval was provided as per the original RCTs. This work adheres to the Declaration of Helsinki.
Image acquisition and trait estimation
T1-weighted, proton density-weighted and T2-weighted brain MRIs were collected in accordance with standardized imaging protocols for each trial. All trials used the same centralized MRI reading center (NeuroRx, Montreal, Canada).
BV change was measured between week 24 and the last time point available for each participant using the automated SIENA (Structural Image Evaluation, using Normalization, of Atrophy) method16. We used week 24 instead of baseline to account for potential rapid volume changes related to resolution of acute inflammation following treatment initiation (also known as pseudoatrophy)12. For placebo arm patients in ADVANCE, BV change was calculated from baseline to week 48 (when participants switched to active treatment).
To calculate T2LV, T2 lesions were segmented using a fully-automated method, followed by visual verification and adjustments, if needed. T2LV change was then estimated as the change in total T2LV between baseline and last time point available for each participant. For the placebo arm in ADVANCE, we used week 48 instead of the last time point to be consistent with the BV change estimation.
Within each RCT, changes in BV and T2LV were annualized to account for varying lengths of follow-up time between the trials. This was accomplished by dividing each individual change by the number of years between the first and last time point used in estimation. Annualized traits were transformed using rank-based inverse normal transformation to mitigate departures from normality (eFigures 1, 2).
Genetic data generation and quality control (QC)
ADVANCE, ASCEND and DECIDE (Biogen trials)
DNA was extracted from blood and genotyped with the Affymetrix UK Biobank Axiom array by Thermo Fisher Scientific (Santa Clara, CA) in two batches: ASCEND and DECIDE (2017), and ADVANCE (2018). QC (PLINK v1.9) and imputation (Michigan Imputation server) was run on the batches separately (hg19). We excluded single nucleotide polymorphisms (SNPs) with missingness > 1%, minor allele frequency (MAF) < 0.01, and Hardy–Weinberg equilibrium (HWE) p < 1 × 10–50. As these are case-only cohorts, we applied a less stringent HWE threshold to avoid the inadvertent exclusion of disease-associated variants. We excluded samples with missingness > 2%, sex discrepancies (females: F < 0.2, males: F > 0.8), excess heterozygosity (> 6 standard deviations from the mean), and relatedness. Independent SNPs (PLINK option --indep-pairwise 50 5 0.5) were used to estimate sex, heterozygosity, relatedness, and ancestry.
Identity by descent (IBD) analysis was used to identify related individuals. Pairs of individuals with k0 > 0.4 were considered related, and one member from each pair was selected randomly for exclusion. We estimated ancestry separately in each RCT, and used principal component analysis (PCA; SmartPCA, Eigensoft v7.2.1) to identify outliers and calculate PCs. We excluded individuals who were 6 standard deviations from the top 10 PCs with a maximum of 10 outlier removal iterations (default). We estimated Tracy-Widom (TW)17 statistics to determine the number of PCs to include as covariates, which were defined as PCs with TW p < 0.05 that accounted for ≥ 1% of variance (calculated by subtracting the largest eigenvalue with p ≥ 0.05, from the sum of the eigenvalues with p < 0.05, and then estimating relative contributing variance).
Genotypes were imputed to 1000G phase3v5 (phasing: ShapeIT v2.r790). After imputation, we excluded SNPs with R2 < 0.30, MAF < 0.01 or in the pseudo-autosomal region of chromosome X. SNPs and samples at each QC step are tabulated in eTable 3.
OPERA I, OPERA II, and ORATORIO (Roche trials)
DNA was extracted from blood and whole-genome sequenced (WGS, mean read depth 30x) (TruSeq Nano library prep, Illumina HiSeq) to generate 150 bp paired-end reads (Illumina, Inc [Foster City, CA]). Burrows-Wheeler Aligner (BWA) was used to map reads (hg38). Resulting alignments (bam files) were analyzed using GATK for base quality score recalibration (BQSR), indel realignment, duplicate removal, SNP/INDEL discovery and joint genotyping across samples according to GATK Best Practices18. Sites that did not pass GATK variant quality score recalibration (VQSR) filter were removed. Genotypes with quality score (GQ) < 20 were set to missing, followed by removal of sites with missing rate > 10%. To further improve the coverage of common variants, genotype imputation was performed using a haplotype reference panel of 55,929 individuals constructed using 27,166 individuals from Haplotype Reference Consortium, 2548 samples from 1000 Genomes project and 26,215 samples sequenced at Genentech. Imputation was performed using Beagle version 5.2 and sites with imputation score < 0.3 or HWE p < 5 × 1–10 or MAF < 1% were removed.
Ancestry was estimated using ADMIXTURE (supervised mode; reference 1000G phase3). Samples with European ancestry coefficient < 0.7 were excluded. Analyses for heterozygosity outlier detection, relatedness (IBD, where pairs of individuals with k0 < 0.4 were considered related), TW statistic, and PCA were performed similarly to Biogen.
GWAS statistical analysis
We used linear regression models to test the association between SNPs with BV change and T2LV change (additive model), controlling for randomized treatment arm, age, sex, and genetic ancestry (3 PCs in ADVANCE and DECIDE, 4 PCs in ASCEND, and 11 PCs in OPERA I & II and ORATORIO). Biogen analyses were performed using imputed dosages (PLINK v2.0 option --glm --maf 0.01). Genentech analyses were performed using WGS (PLINK v1.9 option --linear --maf 0.01). Summary statistics from Roche RCTs were mapped from hg38 to hg19 using liftOver in Hail.
Summary statistics from RCTs were meta-analyzed using a fixed-effects model (PLINK v1.9). Regional plots were generated with LocusZoom v1.4 (1000G phase3 EUR).
To increase power beyond single variant analysis, we used MAGMA19 to evaluate associations on the gene level using the meta-analyzed summary statistics. We used the most recent version of MAGMA (1.10) with the reference data provided by the authors of MAGMA, which was generated from Phase 3 of the 1000 Genomes Project. We used a Bonferroni correction to determine the significance threshold (0.05/17,837 genes = 2.6 × 10–6).
Colocalization analysis
To better understand which genes that our top hits for BV change and T2LV change may impact, we performed colocalization (adapted from coloc package) with human expression data from 48 GTEx tissues (https://gtexportal.org/) and the Database of Immune Cell eQTLs (https://dice-database.org/). We also examined evidence for shared causal variants between selected MS risk loci and our top hits (p < 1 × 10–6). For each analysis, we included all SNPs within a 250 kb window (500 kb region) of the query study SNP (static default prior probabilities: p1 = p2 = 1 × 10–4; p12 = 1 × 10–5).
CNS cell-type specific expression
To visualize single-nucleus gene expression of putative genes identified through GWAS meta-analysis by CNS cell type, we used CellxGene VIP (https://cellxgenevip-ms.bxgenomics.com/). This is based on brain tissue from 12 MS patients20.
Comparison with MS risk SNPs
We examined all autosomal MS risk GWAS loci3, including peak variants in the MHC, for potential association with change in BV and T2LV. We selected available overlapping variants from WGS and genotype data, and report association statistics for variants whose effects on BV change and T2LV change were consistent across the six RCTs (heterogeneity I2 < 40). Additionally, we generated a polygenic risk score (PRS) using the non-MHC variants and weights from the IMSGC MS risk discovery GWAS, and tested the score for association with change in T2LV and change in BV3.
Results
GWAS meta-analysis included N = 3401 participants for BV change and N = 3513 participants for T2LV change from RCTs in relapsing remitting MS (RRMS) [ADVANCE (N = 540; 505, for BV and T2LV change data, respectively), DECIDE (N = 821; 886), OPERA I (N = 581; 581), OPERA II (N = 577; 577)], secondary progressive MS (SPMS) (ASCEND; N = 353; 435), and primary progressive MS (PPMS) (ORATORIO, N = 529; 529) (Table 1). Participants were mainly female (51–74%), mean age was higher in the PMS RCTs (45–47 vs 36–38 in RRMS), and mean study duration was ~ 1.8 years. In total, we meta-analyzed 10,382,375 and 10,608,740 SNPs for association with BV change and T2LV change, respectively. The genomic inflation factor did not show evidence of bias from ancestry in either GWAS (λBV = 1.008, λT2LV = 1.019) (eFigure 3).
The most significant association with BV change was an intronic SNP in PTPRD (top SNP: rs77321193-C/A, p = 5.3 × 10–7), which showed no evidence for heterogeneity across the trials (heterogeneity I2 = 0, heterogeneity p = 0.96) (Fig. 1A, Table 2). The minor allele (C; MAF = 0.18) of rs77321193 was associated with a greater reduction in BV over time. Multiple SNPs in this region showed association with BV change with similar effect sizes and no evidence for heterogeneity in the meta-analysis (eTable 4, Figs. 2A, 3A). We did not observe colocalization between the PTPRD peak and expression of PTPRD in the datasets we evaluated. However, analysis of cell-type specific gene expression in MS brain (Fig. 4) revealed that PTPRD was expressed in neurons, oligodendrocytes and oligodendrocyte precursor cells. The gene-based tests did not reveal significant association with PTPRD (p = 0.066), or other genes (top hit: TSPAN8, p = 8.08 × 10–5) (eTable 5).
The strongest association with T2LV change was in a regulatory region near NEDD4L (top SNP: rs11398377-GC/G, p = 9.5 × 10–8), which was consistent across the trials (heterogeneity I2 = 51, heterogeneity p = 0.07) (Fig. 1B, Table 3). The C deletion in rs11398377 (MAF = 0.16) was associated with greater T2LV accumulation over time. Several SNPs in this region associated with T2LV change had similar effect sizes and did not show significant heterogeneity in the meta-analysis (eTable 6, Figs. 2B, 3B). Interestingly, we found evidence for colocalization between the T2LV change peak (rs11398377) and NEDD4L expression in whole blood (colocalization pp4 = 0.78), suggestive of a common causal variant between the traits. Moreover, analysis of cell-specific gene expression in the MS brain showed that NEDD4L is expressed mainly in neurons, including inter-neurons and excitatory neurons (Fig. 4). Gene-based tests did not show significant associations with change in T2LV (top hit: WDR34, p = 2.0 × 10–5), however NEDD4L was among the top hits (p = 6.94 × 10–5) (eTable 7).
Comparison with the known MS risk GWAS3 SNPs (197 non-MHC and autosomal; 161 and 162 SNPs met our inclusion criteria for comparison with BV change and T2LV change meta-analyses results, respectively) did not show evidence of overlap with BV change nor T2LV change GWAS hits (eTables 8, 9). We did not observe any significant associations (p < 1 × 10–6) between MS susceptibility loci with BV change or T2LV change, and found no overlap with MS risk loci in the MHC region3. Conversely, neither the BV change peak in PTPRD nor the T2LV change peak in NEDD4L were associated with MS risk3, and no putative BV change or T2LV change peaks (p < 1 × 10–6) colocalized with MS risk. Additionally, the MS risk PRS was not significantly associated with change in BV (p = 0.54) nor with change in T2LV (p = 0.89).
Discussion
To our knowledge, this is the first GWAS meta-analysis of longitudinal MRI measures across multiple RCTs to investigate the genetics of MS progression. We achieved this by leveraging a precompetitive industry partnership aimed at utilizing well-characterized MS cohorts to interrogate the genetics of disease progression in MS.
The top BV change SNP occurs in an intron of PTPRD, a protein tyrosine phosphatase receptor involved in cellular signaling, growth and differentiation21. Previous studies of cross-sectional BV and Multiple Sclerosis Severity Score (MSSS) have also reported sub-significant associations with variants in the PTPRD region independent from our BV change peak (cross sectional BV top SNP: top SNP: rs1953594, adjusted trend − logP = 4.3; MSSS top SNP: rs10977017, p = 1.02 × 10–5)5,6. Additionally, variants mapped to PTPRD are associated with two sleep disorders that are enriched in PwMS: restless leg syndrome (top SNP rs1836229)22,23 and insomnia (top SNP rs10761240)24,25. While we did not observe evidence for colocalization between the top BV change association and PTPRD expression, it is possible that this study is insufficiently powered to detect colocalization, or that the locus impacts mechanisms that are time-dependent, cell lineage-specific, or independent of PTPRD expression.
The top T2LV association is an indel SNP that lies in the regulatory region upstream of NEDD4L. NEDD4L (neural precursor cell expressed developmentally down-regulated 4-like) plays a role in axon guidance, neurite growth, synaptic transmission, and pain sensitivity26,27. The NEDD4L peak showed evidence for colocalization with expression of NEDD4L in blood (GTEx). The minor allele of this top SNP (rs11398377*C- deletion) was associated with greater T2LV accumulation and increased gene expression, indicating that downregulation of the gene may have therapeutic potential. Missense SNPs in NEDD4L cause abnormal fetal neurodevelopment and brain malformations such as periventricular nodular heterotopia28. NEDD4L encodes an E3 ubiquitin-protein ligase that regulates several membrane channels and transporters including epithelial (ENaC)29 and voltage-gated (NaV)30 sodium channels. Both ENaCs and NaVs have been implicated in demyelination and MS pathophysiology, and research is ongoing to investigate the role of sodium channel blockers in preventing axonal damage31,32. This link is supported by use of the channel blocker dalfampridine for improvement of walking in PwMS33.
While BV and T2LV change traits may be correlated15 (eTable 10), they reflect complementary measures of MS progressive biology—BV reduction represents a global measure of brain tissue loss, while T2LV increase reflects immune-mediated injury in the CNS white matter. Both PTPRD and NEDD4L are strongly expressed in the CNS, particularly in neurons. PTPRD is also abundant in oligodendrocytes and oligodendrocyte precursor cells. Oligodendrocytes play an important role in MS because they myelinate neurons. NEDD4L is a paralog of NEDD4, which promotes ubiquitination and degradation of RAPGEF2, one of the few putative MS risk genes with enhanced single cell expression in CNS cells34,35.
The IMSGC recently published a large study of MS severity using the age-related MS severity score (ARMSS), and reported association in the DYSF–ZNF638 locus4. Additionally, a smaller study (N = 1813 individuals) examining median longitudinal ARMSS (l-ARMSS) and longitudinal MSSS (l-MSSS) did not find significant association36. The IMSGC DYSF–ZNF638 locus index variant (rs10191329) was not associated with change in BV (beta = − 0.004, p = 0.91) nor change in T2LV in our analysis (beta = 0.07, p = 0.04). This may be due to the greater power in the IMSGC analysis, but it is also possible that the variables we analyzed capture different aspects of the MS disease course. We chose to focus on objective imaging measures rather than clinical measures such as EDSS (on which the ARMSS variable is based) due to certain limitations in these measures. For instance, the EDSS score is numerical and nonlinear (ranging from 0 to 10) and is largely based on mobility, making it a poor proxy for progression.
Similar to the IMSGC MS severity study, we did not find overlap between MS risk loci and our peaks, nor did we observe association with an MS PRS4. This suggests that mechanisms driving MS susceptibility may differ from those modulating outcomes influencing disease progression. A prior study in ~ 500 MS patients found that the strongest genetic risk factor for MS (HLA- DRB1*15:01) was associated with reduction in brain parenchymal volume and higher T2 lesion load37, however these findings were not replicated in our study.
We acknowledge the following limitations in our study. All but two of our RCTs (ASCEND, ORATORIO) examined RRMS participants, and it is possible that the biological mechanisms underlying MS progression differ between early-stage and late-stage MS38. Moreover, the clinical trial participants may not be representative of the general MS population owing to strict inclusion and exclusion criteria used in participant selection of RCTs. Neither peak identified in this study reached genome-wide significance (p < 5 × 10–8) or a stricter threshold of p < 2.5 × 10–8 to account for multiple testing. Though powered to detect moderate associations with common SNPs, our ability to detect genome-wide associations would be improved with a larger sample size. Additional, well-characterized cohorts are needed for independent replication. Longitudinal MRI data was available for < 2 years, however this reflects a small proportion of the average MS patient’s disease course, and longer follow-up would provide more accurate outcomes on BV and T2LV changes. Several of the trials had additional longitudinal data available from open label extension studies, however including these data would have introduced complexities such as variable lengths of follow-up, and differential handling of placebo and treatment arms when accounting for pseudoatrophy at treatment initiation. We therefore limited our analyses to data from the placebo-controlled period. While we endeavored to account for pseudoatrophy by re-baselining at 24 weeks, it is possible this did not fully capture all pseudoatrophy. We chose to re-baseline using week 24 measures to have a reasonable duration of MRI follow-up for GWAS.
This work highlights the utility of using clinical trial data for genetic analyses. RCTs have assigned treatment arms and systematic imaging collection, and genetic analyses of RCTs are uniquely positioned to leverage objective, quantitative measurements to advance our understanding of MS disease course. Furthermore, the MRI analysis methods used to quantify atrophy and T2LV were standardized within studies, and to the extent possible, across the studies used in the meta-analysis, because all studies utilized the same centralized MRI reading center (NeuroRx). Finally, our findings complement existing risk studies by identifying additional genetic factors related to different aspects of the disease.
In conclusion, this study identified two novel putative loci (PTPRD, NEDD4L) that may impact MS disease progression. Investigation in additional cohorts are warranted to validate these findings.
Data availability
Individual level data is not available due to the use of clinical trial data. Summary statistics may be available upon request. Requests may be sent to Stephanie Loomis, stephanie.loomis@biogen.com.
References
Rotstein, D. & Montalban, X. Reaching an evidence-based prognosis for personalized treatment of multiple sclerosis. Nat. Rev. Neurol. 15, 287–300. https://doi.org/10.1038/s41582-019-0170-8 (2019).
University of California, S. F. M. S. E. T. et al. Silent progression in disease activity-free relapsing multiple sclerosis. Ann. Neurol. 85, 653–666. https://doi.org/10.1002/ana.25463 (2019).
International Multiple Sclerosis Genetics, C. Multiple sclerosis genomic map implicates peripheral immune cells and microglia in susceptibility. Science 365, 58. https://doi.org/10.1126/science.aav7188 (2019).
International Multiple Sclerosis Genetics, C. & Multiple, M. S. C. Locus for severity implicates CNS resilience in progression of multiple sclerosis. Nature 619, 323–331. https://doi.org/10.1038/s41586-023-06250-x (2023).
International Multiple Sclerosis Genetics, C. Genome-wide association study of severity in multiple sclerosis. Genes Immun. 12, 615–625. https://doi.org/10.1038/gene.2011.34 (2011).
Baranzini, S. E. et al. Genome-wide association analysis of susceptibility and clinical phenotype in multiple sclerosis. Hum. Mol. Genet. 18, 767–778. https://doi.org/10.1093/hmg/ddn388 (2009).
Brynedal, B. et al. MGAT5 alters the severity of multiple sclerosis. J. Neuroimmunol. 220, 120–124. https://doi.org/10.1016/j.jneuroim.2010.01.003 (2010).
George, M. F. et al. Multiple sclerosis risk loci and disease severity in 7,125 individuals from 10 studies. Neurol. Genet. 2, e87. https://doi.org/10.1212/NXG.0000000000000087 (2016).
Healy, B. C., Engler, D., Glanz, B., Musallam, A. & Chitnis, T. Assessment of definitions of sustained disease progression in relapsing-remitting multiple sclerosis. Mult. Scler. Int. 2013, 189624. https://doi.org/10.1155/2013/189624 (2013).
Scalfari, A. et al. The natural history of multiple sclerosis: A geographically based study 10: Relapses and long-term disability. Brain 133, 1914–1929. https://doi.org/10.1093/brain/awq118 (2010).
Bermel, R. A. & Bakshi, R. The measurement and clinical relevance of brain atrophy in multiple sclerosis. Lancet Neurol. 5, 158–170. https://doi.org/10.1016/S1474-4422(06)70349-0 (2006).
De Stefano, N. et al. Clinical relevance of brain volume measures in multiple sclerosis. CNS Drugs 28, 147–156. https://doi.org/10.1007/s40263-014-0140-z (2014).
Rocca, M. A. et al. Brain MRI atrophy quantification in MS: From methods to clinical application. Neurology 88, 403–413. https://doi.org/10.1212/WNL.0000000000003542 (2017).
Fisniku, L. K. et al. Disability and T2 MRI lesions: A 20-year follow-up of patients with relapse onset of multiple sclerosis. Brain 131, 808–817. https://doi.org/10.1093/brain/awm329 (2008).
Radue, E. W. et al. Correlation between brain volume loss and clinical and MRI outcomes in multiple sclerosis. Neurology 84, 784–793. https://doi.org/10.1212/WNL.0000000000001281 (2015).
Smith, S. M., De Stefano, N., Jenkinson, M. & Matthews, P. M. Normalized accurate measurement of longitudinal brain change. J. Comput. Assist. Tomogr. 25, 466–475. https://doi.org/10.1097/00004728-200105000-00022 (2001).
Patterson, N., Price, A. L. & Reich, D. Population structure and eigenanalysis. PLoS Genet. 2, e190. https://doi.org/10.1371/journal.pgen.0020190 (2006).
Van der Auwera, G. A. et al. From FastQ data to high confidence variant calls: The Genome Analysis Toolkit best practices pipeline. Curr. Protoc. Bioinform. 43, 10–11. https://doi.org/10.1002/0471250953.bi1110s43 (2013).
de Leeuw, C. A., Mooij, J. M., Heskes, T. & Posthuma, D. MAGMA: Generalized gene-set analysis of GWAS data. PLoS Comput. Biol. 11, e1004219. https://doi.org/10.1371/journal.pcbi.1004219 (2015).
Schirmer, L. et al. Neuronal vulnerability and multilineage diversity in multiple sclerosis. Nature 573, 75–82. https://doi.org/10.1038/s41586-019-1404-z (2019).
PTPRD protein tyrosine phosphatase receptor type D [Homo sapiens (human)] Gene ID: 5789. https://www.ncbi.nlm.nih.gov/gene/5789#gene-expression (2019).
Schormair, B. et al. Identification of novel risk loci for restless legs syndrome in genome-wide association studies in individuals of European ancestry: A meta-analysis. Lancet Neurol. 16, 898–907. https://doi.org/10.1016/S1474-4422(17)30327-7 (2017).
Didriksen, M. et al. Large genome-wide association study identifies three novel risk variants for restless legs syndrome. Commun. Biol. 3, 703. https://doi.org/10.1038/s42003-020-01430-1 (2020).
Jansen, P. R. et al. Genome-wide analysis of insomnia in 1,331,010 individuals identifies new risk loci and functional pathways. Nat. Genet. 51, 394–403. https://doi.org/10.1038/s41588-018-0333-3 (2019).
Brass, S. D., Li, C. S. & Auerbach, S. The underdiagnosis of sleep disorders in patients with multiple sclerosis. J. Clin. Sleep Med. 10, 1025–1031. https://doi.org/10.5664/jcsm.4044 (2014).
Donovan, P. & Poronnik, P. Nedd4 and Nedd4-2: Ubiquitin ligases at work in the neuron. Int. J. Biochem. Cell Biol. 45, 706–710. https://doi.org/10.1016/j.biocel.2012.12.006 (2013).
Yanpallewar, S. et al. Nedd4-2 haploinsufficiency causes hyperactivity and increased sensitivity to inflammatory stimuli. Sci. Rep. 6, 32957. https://doi.org/10.1038/srep32957 (2016).
Broix, L. et al. Mutations in the HECT domain of NEDD4L lead to AKT-mTOR pathway deregulation and cause periventricular nodular heterotopia. Nat. Genet. 48, 1349–1358. https://doi.org/10.1038/ng.3676 (2016).
Rotin, D. & Staub, O. Nedd4-2 and the regulation of epithelial sodium transport. Front. Physiol. 3, 212. https://doi.org/10.3389/fphys.2012.00212 (2012).
Ekberg, J. A. et al. Nedd4-2 (NEDD4L) controls intracellular Na(+)-mediated activity of voltage-gated sodium channels in primary cortical neurons. Biochem. J. 457, 27–31. https://doi.org/10.1042/BJ20131275 (2014).
Craner, M. J. et al. Molecular changes in neurons in multiple sclerosis: Altered axonal expression of Nav1.2 and Nav1.6 sodium channels and Na+/Ca2+ exchanger. Proc. Natl. Acad. Sci. U. S. A. 101, 8168–8173. https://doi.org/10.1073/pnas.0402765101 (2004).
Waxman, S. G. Axonal conduction and injury in multiple sclerosis: The role of sodium channels. Nat. Rev. Neurosci. 7, 932–941. https://doi.org/10.1038/nrn2023 (2006).
Boiko, N., Kucher, V., Eaton, B. A. & Stockand, J. D. Inhibition of neuronal degenerin/epithelial Na+ channels by the multiple sclerosis drug 4-aminopyridine. J. Biol. Chem. 288, 9418–9427. https://doi.org/10.1074/jbc.M112.449413 (2013).
Pham, N. & Rotin, D. Nedd4 regulates ubiquitination and stability of the guanine-nucleotide exchange factor CNrasGEF. J. Biol. Chem. 276, 46995–47003. https://doi.org/10.1074/jbc.M108373200 (2001).
Zhang, H. et al. NEDD4-1 regulates migration and invasion of glioma cells through CNrasGEF ubiquitination in vitro. PLoS ONE 8, e82789. https://doi.org/10.1371/journal.pone.0082789 (2013).
Jokubaitis, V. G. et al. Not all roads lead to the immune system: The genetic basis of multiple sclerosis severity implicates central nervous system and mitochondrial involvement. MedRxiv https://doi.org/10.1101/2022.02.04.22270362 (2022).
Okuda, D. T. et al. Genotype-Phenotype correlations in multiple sclerosis: HLA genes influence disease severity inferred by 1HMR spectroscopy and MRI measures. Brain 132, 250–259. https://doi.org/10.1093/brain/awn301 (2009).
Lassmann, H. Pathology and disease mechanisms in different stages of multiple sclerosis. J. Neurol. Sci. 333, 1–4. https://doi.org/10.1016/j.jns.2013.05.010 (2013).
Acknowledgements
The Biogen authors gratefully acknowledge the contributions of Jun Ke, Katherine Riester, Zheng Ren, Chunlei Ke (biostatistics); David Sexton (scientific computing); Janet Krause, Lukasz Kniola, Joanna Koft, Kendra Buzzell, Marija Jankovich, Phil Liston-Kraft, Anna Langhorne, Susan Wise (data sharing, legal, data privacy); Chao Sun, Carol Singh, Julie Ranuio, Eric David (samples); and Jimmy Z. Liu, Samira Asgari, Andi Dhroso, Allison Pine, Denis Baird, Helen McLaughlin (human genetics). Special thanks to Mark McCarthy, Nathalie Franchimont, Bing Zhu, Aaron Deykin, Pei-Ran Ho, Giorgio Giannattasio, Jonathan Smith, James Williams and Shibeshih Belachew for their support, and to Tatiana Plavina, David Brassat, Dipen Sangurdekar, Michelle Penny, Karol Estrada, Will Chen, and Richard Ransohoff for their contributions during earlier stages of this work. We thank all of our Genentech colleagues in the Human Genetics Initiative involved in generating the sequence data including Natalie Bowers, Julie Hunkapiller, Jens Reeder, and Suresh Selvaraj. We would also like to thank the trial participants, without whom this research would not be possible. The Genotype-Tissue Expression (GTEx) Project was supported by the Common Fund of the Office of the Director of the NIH, and by NCI, NHGRI, NHLBI, NIDA, NIMH, and NINDS. The data used for the analyses described in this manuscript were obtained from: the GTEx Portal on 08/26/2019.
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.
Author information
Authors and Affiliations
Contributions
S.J.L. and N.S. conceptualized and designed the study, analyzed the data and wrote the manuscript. P.B. conceptualized and designed the study, oversaw the work and helped write the manuscript. T.B. analyzed data, oversaw the work and helped write the manuscript. CY and YH analyzed data and critically reviewed the manuscript. E.M. assisted in data curation and critically reviewed the manuscript. E.H. helped prepare figures. A.G., E.F., A.H., H.R., S.J. and X.J. provided clinical and scientific expertise and critically reviewed the manuscript.
Corresponding author
Ethics declarations
Competing interests
SJL, NS, CY, EH, EF, EM, AG, HR, SJ, PGB are/were employees of Biogen and may hold stocks/stock options in Biogen. AH, XJ and TB are employees of Genentech (a subsidiary of Roche) and may hold stocks/stock options in F. Hoffmann-La Roche Ltd.
Additional information
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Loomis, S.J., Sadhu, N., Fisher, E. et al. Genome-wide study of longitudinal brain imaging measures of multiple sclerosis progression across six clinical trials. Sci Rep 13, 14313 (2023). https://doi.org/10.1038/s41598-023-41099-0
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-023-41099-0
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.