Associations between perfluoroalkyl substances and multiple sclerosis in U.S. cohorts compared with NMOSD, MOGAD, and healthy controls.

Associations between perfluoroalkyl substances and multiple sclerosis in U.S. cohorts compared with NMOSD, MOGAD, and healthy controls - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Search Log in Primary site navigation Search Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Environ Res . Author manuscript; available in PMC: 2026 Jun 5. Published in final edited form as: Environ Res. 2026 May 15;303(Pt 1):124763. doi: 10.1016/j.envres.2026.124763 Associations between perfluoroalkyl substances and multiple sclerosis in U.S. cohorts compared with NMOSD, MOGAD, and healthy controls Farren B S Briggs 1, ¥, * , Federico Montini 2, ¥ , Charles F Spurlock III 3 , Christine Ryan 4 , Jessica N Cooke Bailey 5 , Erika S Trapl 6 , Guzel I Shaginurova 7 , Lukasz S Wylezinski 8 , Cheryl L Sesler 9 , Mahboobeh Fereidan Esfahani 10 , Jessica Sagen 11 , Hannah E Gardener 12 , Jacob L McCauley 13 , Alberto J Caban-Martinez 14 , Roberta Brambilla 15 , Lilyana Amezcua 16 , W Oliver Tobin 17 Author information Article notes Copyright and License information 1 Department of Public Health Sciences, University of Miami Miller School of Medicine, Miami, FL, USA 2 Center for Multiple Sclerosis and Autoimmune Neurology, Mayo Clinic, Rochester, MN, USA 3 Decode Health, Inc., Department of Medicine, Vanderbilt University School of Medicine, Thomas F. Frist, Jr. College of Medicine, Belmont University, Nashville, TN, USA; New York University, New York, NY, USA 4 University of Miami Miller School of Medicine, Miami, FL, USA 5 Department of Pharmacology and Toxicology, East Carolina University, Greenville, NC, USA 6 Department of Population and Quantitative Health Sciences, Case Western Reserve University School of Medicine, Cleveland, OH, USA 7 Decode Health, Inc., Nashville, TN, USA 8 Decode Health, Inc., Department of Medicine, Vanderbilt University School of Medicine, Nashville, TN, USA 9 Decode Health, Inc., Nashville, TN, USA 10 Department of Neurology, The University of Texas at Austin Dell Medical School, Austin TX, USA 11 Center for Multiple Sclerosis and Autoimmune Neurology, Mayo Clinic, Rochester, MN, USA 12 Department of Neurology, University of Miami Miller School of Medicine, Miami, FL, USA 13 John P. Hussman Institute for Human Genomics, University of Miami Miller School of Medicine, Miami, FL, USA 14 Department of Public Health Sciences, University of Miami Miller School of Medicine, Miami, FL, USA 15 Department of Neurological Surgery, University of Miami Miller School of Medicine, Miami, FL, USA 16 Department of Neurology, University of Southern California Keck School of Medicine, Los Angeles, CA, USA 17 Center for Multiple Sclerosis and Autoimmune Neurology, Mayo Clinic, Rochester, MN, USA ¥ Co-first authors Author Contributions FB was responsible for conception and design of the study, and FM, CFS, JCB, ET, HG, CR, ACM, RB, JM, LA, and WOT provided guidance. FB, JS, ME were responsible for participant data curation. FB was responsible for metabolomic data curation, statistical analyses, and figures. CFS, GIS, LSW, CLS were responsible for RNA-seq data curation. FB and FM drafted the manuscript. All authors guided interpretations, and edited and approved the final draft. * Corresponding Author: Farren Briggs, PhD ScM, Department of Public Health Sciences, University of Miami Miller School of Medicine, 1120 NW 14 th Street, CRB 912, Miami, FL 33136, fbb31@miami.edu Issue date 2026 Aug 15. PMC Copyright notice PMCID: PMC13235680  NIHMSID: NIHMS2179662  PMID: 42142568 The publisher's version of this article is available at Environ Res Abstract Background and Objectives. Perfluoroalkyl substances (PFAS) are persistent environmental contaminants with endocrine and neuroimmune properties, yet relevance to multiple sclerosis (MS) remains unclear. Recent European data suggested MS-PFAS associations may vary by sex. We evaluated sex-stratified and disease-specific PFAS-MS associations in U.S. cohorts and explored endocrine and transcriptomic correlates. Methods. Liquid chromatography–mass spectrometry-derived serum PFHxS, PFOA, and PFOS were quantified in 439 participants (226 MS, 106 healthy controls [HC], 59 neuromyelitis optica spectrum disorder [NMOSD], 48 myelin oligodendrocyte glycoprotein antibody-associated disease [MOGAD]) across two U.S. cohorts. PFAS levels were compared across diagnostic groups with sex stratification. Exposure-risk models estimated MS odds. Among MS-HC females, associations between PFAS and sex-steroid metabolites were evaluated, and PFAS-correlated gene-expression signatures were examined in a subset of MS females. Results. PFHxS was consistently elevated in MS females compared with HC, NMOSD, and MOGAD females, while PFOA showed a similar but less uniform pattern; male associations were imprecise. Exposure-risk models demonstrated dose-response and non-linear associations for PFHxS, with directionally similar patterns for PFOA; PFHxS was the primary contributor in mixture analyses. PFOS associations were weaker and less consistent. Exploratory analyses observed significant MS×PFOA interactions for several steroid metabolites, and PFAS-correlated genes showed tentative enrichment of neuronal membrane pathways. Conclusions. Across two U.S. cohorts, PFHxS demonstrated consistent association with MS, particularly among females, while PFOA showed a similar but less uniform pattern. Exploratory findings suggest plausible pathways linking PFAS exposure to MS. Multi-ethnic and longitudinal studies are needed to expand generalizability and clarify temporality and causality. Keywords: multiple sclerosis, per- and polyfluoroalkyl substances, PFHxS, PFOA, PFOS, PFAS, NMOSD, MOGAD Graphical Abstract 1. Background Multiple sclerosis (MS) is a chronic autoimmune and neurodegenerative disease characterized by myelin and axonal loss. Established MS risk factors include female sex, genetic susceptibility variants such as HLA-DRB1*15:01 , and a limited number of common environmental and lifestyle exposures, including Epstein-Barr virus seropositivity, smoking, low vitamin D or limited sun exposure, and adolescent obesity.( Olsson et al., 2017 ) Perfluoroalkyl substances (PFAS) are synthetic, persistent “forever chemicals” that bioaccumulate in humans and the environment, contaminating air, soil, and water supplies across the U.S. Industrial-scale production of long-chain PFAS such as perfluorohexane sulfonic acid (PFHxS), perfluorooctanoic acid (PFOA), and perfluorooctanesulfonic acid (PFOS) began in the 1940s,( Gluge et al., 2020 ) and continued ongoing use in consumer and industrial products has resulted in near-universal exposure in the U.S.( Cousins et al., 2020 ) Ninety-six percent of Americans have measurable levels of four or more legacy PFAS (including PFHxS, PFOA and PFOS),( Botelho et al., 2025 ) with higher concentrations in older individuals, males, and rural, Black, Hispanic, and low-income communities.( Botelho et al., 2025 ; Chukwuonye et al., 2024 ; Jordan et al., 2024 ; Liddie et al., 2023 ; Maruzzo et al., 2025 ; Sukhram et al., 2025 ) Although legacy PFAS levels are declining, concentrations remain higher in Americans than Europeans, including children and adolescents, likely reflecting differences in exposure sources and limited U.S. regulatory oversight.( ATSDR, 2024 ; Engelhardt et al., 2023 ; Engelhardt et al., 2025 ; Richterova et al., 2023 ) Long human half-lives (up to 35 years for PFHxS) result in cumulative exposure,(2021) and >200 million Americans are exposed to drinking water containing PFOA or PFOS above 1 ng/L.( Andrews and Naidenko, 2020 ) The U.S. Environmental Protection Agency warns PFAS may impair vaccine response, disrupt endocrine function, and increase risk of several chronic health outcomes.( Agency, 2024 ) Population-based studies suggest sex-specific endocrine effects, including altered testosterone and sex hormone profiles in adolescent girls and women.( Rodriguez-Carrillo et al., 2023a ; Yang et al., 2025 ) Additionally, while neurological outcomes remain understudied, perinatal PFOS exposure disrupts oligodendrogenesis and impairs remyelination in animal models,( Butruille et al., 2023 ) and human studies link PFAS to altered B-cell development and BCR signaling.( Rudzanova et al., 2024 ) Together, disruptions in these core MS mechanisms (endocrine disruption, oligodendrogenesis, and B-cell–mediated immunity) provide biologically plausible pathways linking PFAS exposure to MS risk.( Gharibi et al., 2020 ; Murgia et al., 2022 ; Yeung et al., 2019 ) Consistent with this mechanistic framework, a recent Swedish MS case-control study reported a sex-differentiated association, with elevated PFHxS and PFOA in MS females but not in males compared to sex-matched healthy controls (HC); however, logistic models for continuous PFAS measures were only suggestive.( Vaivade et al., 2025 ) In a follow-up analysis incorporating non-linear modeling, higher quartiles of PFOA and PFOS were associated with increased MS odds, with a similar trend for PFHxS.( Vaivade et al., 2026 ) A 14-PFAS mixture model (including PFHxS, PFOA and PFOS) also demonstrated a significant association with MS. However, sex-stratified analyses were not reported for these follow-up models. Whether PFAS associations are specific to MS or reflect broader neuroimmune mechanisms remains unknown. Although, MS, neuromyelitis optica spectrum disorder (NMOSD), and myelin oligodendrocyte glycoprotein antibody-associated disease (MOGAD) are inflammatory demyelinating diseases, their immunopathologies differ: MS involves complex T- and B-cell-mediated autoimmunity with neurodegenerative features, whereas NMOSD and MOGAD are defined by pathogenic antigen-specific autoantibodies (aquaporin-4 [APQ4] and MOG IgG).( Lopez et al., 2021 ) While PFAS exposure has been associated with immune modulation, these effects do not directly parallel antigen-specific mechanisms in NMOSD and MOGAD. Inclusion of mechanistically distinct comparator diseases would enable evaluation of MS specificity. Since PFAS concentrations are higher in Americans than in Europeans, it is important to determine if similar or stronger PFAS-MS associations exist in the U.S. In the current study, we evaluated sex-stratified and disease-specific associations between LC-MS-derived serum PFAS levels and early-stage MS across two U.S. cohorts. Our objectives were to: 1a) compare PFHxS, PFOA, and PFOS levels between MS and HC, and by sex, and 1b) complete complementary exposure-risk models (dose-response, non-linear, and mixture effects); 2) determine whether PFAS levels observed in MS differ from those in NMOSD and MOGAD, and by sex; and 3) explore potential mechanisms by evaluating: a) interactions with sex-related hormone metabolites, and b) PFAS-associated gene-expression correlates in MS. Together, these aims assess whether PFAS may represent environmental contributors to MS susceptibility in U.S. populations, clarify disease specificity, and provide hypothesis-generating mechanistic insights – extending recent Swedish findings to: U.S.-based cohorts, sex-stratified risk-modeling, comparator disease analyses, and mechanistic exploration. 2. Methods 2.1. Ethical approval This study was approved by the institutional review boards at Case Western Reserve University (STUDY20210045), Mayo Clinic (21–003944), and the University of Miami (#20240098). All participants provided written informed consent at sample collection. 2.2. Study design In this case-control study (MS vs HC, NMOSD, MOGAD), we quantified LC-MS-derived serum levels for three PFAS (PFHxS, PFOA, and PFOS) in non-Hispanic White individuals from two independent U.S. biobanks: Accelerated Cure Project (ACP) and Mayo Clinic Center for Multiple Sclerosis and Autoimmune Neurology (CMSAN). Practical sample sizes for other ethnoracial groups were not readily available. ACP recruited persons with MS (PwMS), persons with NMOSD (PwNMOSD), and HC from 10 U.S. MS tertiary care centers between 2006–2013, who completed epidemiologic questionnaires (i.e., smoking history, menopausal status, days since menses, height, weight, time of last meal, etc.), and provided biological samples for use in future research projects ( www.acceleratedcure.org ). Recruiting neurologists confirmed PwMS met 2001 or 2010 McDonald diagnostic criteria, or that PwNMOSD met Wingerchuk criteria (2007) and subsequently abstracted AQP4 antibody status from their medical records.( McDonald et al., 2001 ; Polman et al., 2011 ; Wingerchuk et al., 2007 ) ACP excluded participants if they had histories of blood borne pathogens, an allogenic bone marrow transplantation, or known to be pregnant or currently breastfeeding. For this study, we restricted PwMS and PwNMOSD to those within five years of symptom onset at blood draw to reduce the potential for reverse confounding due to altered behaviors because of disease chronicity. PwMS cases were restricted to those aged 18 or older at symptom onset, who were disease-modifying therapy naïve or free for at least 3 months, and who had a relapsing remitting MS (RRMS) disease course. HCs were individuals with no features suggestive of MS, NMOSD, transverse myelitis, acute disseminated encephalomyelitis, optic neuritis, or other demyelinating or neurologic diagnoses. HCs were age (+/− 3 years) and sex matched to PwMS and PwNMOSD. The CMSAN cohort is comprised of CMSAN MS patients evaluated between 2017–2021, who met 2017 McDonald criteria, were 18 years or older at symptom onset, and DMT naïve or free for at least 3 months.( Thompson et al., 2018 ) In re-review, one MS patient was on a DMT within the prior 90 days. CMSAN NMOSD patients met the 2015 consensus criteria and tested positive for AQP4 antibodies, and MOGAD patients met the 2023 consensus criteria and tested positive for MOG antibodies.( Banwell et al., 2023 ; Thompson et al., 2018 ; Wingerchuk et al., 2015 ) Those included had no history of cancer, or other autoimmune or rheumatologic disease. Clinical and demographic data (e.g., age, sex, smoking history, BMI, medication status) were abstracted from medical records by a trained neurologist (MFE). 2.3. Sample Collection In ACP, blood was drawn into serum-separating tubes, serum isolated by centrifugation, transferred in 0.5mL aliquots into 1.0mL cryovials, stored at −80°C at the collection site, and shipped monthly on dry ice to SeraCare (Milford, MA) for long-term storage at −80°C. In CMSAN, blood was collected into non-additive tubes at a clinical encounter, and serum was separated within six hours at the Mayo Clinic Biospecimen Accessioning and Processing Core, aliquoted into 1.0 mL cryovials, and stored at −80°C. Non-thawed serum samples collected by ACP and CMSAN were analyzed. 2.4. Metabolomic Profiling Serum metabolomic profiles were generated using an untargeted LC-MS platform (Precision Metabolomics ™ , Metabolon Inc., Durham, NC), which integrates multiple methods.( Ryals et al., 2007 ) Data were processed against a library of authenticated standards encompassing >12,000 biochemicals. ACP and CMSAN samples were profiled separately. Technical reproducibility was high, with duplicate samples from ACP (n=6 pairs; 3 split across the two profiling runs) and CMSAN (n=1 pair) yielding r 2 >0.996. Three legacy PFAS were captured by this LC-MS platform: PFHxS, PFOA, and PFOS; in addition to 51 sex-related hormone or sterol metabolites ( eTable 1 ). PFHxS and PFOA had detectable peak intensities in all samples, while PFOS was missing in one MS sample. Missing metabolomic values were imputed to 1/5 the lowest observed value for that biochemical trait using MetaboAnalyst 6.0 (default imputation setting).( Pang et al., 2024 ) As internal validity checks, PFHxS, PFOA, and PFOS levels increased with age and were higher in males, consistent with expected patterns in U.S. populations ( eFigures 1 – 2 ).( Jordan et al., 2024 ) 2.5. RNA-sequencing Peripheral whole-blood RNA-seq data were available for a subset of ACP RRMS patients.( Wylezinski et al., 2024 ) Specimen handling, library preparation, sequencing, and end-to-end QC followed an established framework.( Sesler et al., 2025 ) Briefly, blood was collected in PAXgene Blood RNA tubes, and RNA was isolated with on-column DNase treatment. Strand-specific total RNA libraries with rRNA and globin depletion were sequenced on an Illumina NovaSeq instrument (paired-end 150 bp). Reads were evaluated with standard tools, aligned to GRCh38 using STAR, and summarized to gene-level counts. Samples that did not meet criteria for integrity, depth, base quality, or unique mapping, or were flagged as outliers by distance metrics or PCA, were excluded, as previously described.( Sesler et al., 2025 ) 2.6. Statistical Analyses Descriptive statistics were summarized by cohort (ACP, CMSAN); continuous variables were compared using Wilcoxon rank-sum tests and categorical variables using Fisher’s exact tests. There were no missing values for primary confounders considered in analyses. PFAS peak area intensities were log-transformed and standardized (mean=0, SD=1) within each cohort. Analyses used Stata v17 and Python v3.12. 2.6.1. Objective 1a (PFAS levels in MS vs HC). In ACP, multivariable linear regression modeled each LC-MS-derived PFAS (PFHxS, PFOA, PFOS) as dependent variables with MS case-control status as the primary independent variable, adjusted for age, sex, body mass index (BMI), and smoking status (ever/never) (hypothesized causal relationships illustrated in eFigure 3 ). Both birth year and calendar time of sample collection were separately explored as tentative covariates to account for possible temporal trends in PFAS exposures and MS incidence, but due to multicollinearity with age they were not retained. BMI and smoking were included as potential confounders given their established associations with both PFAS levels and MS risk.( Pomazal et al., 2024 ; Qi et al., 2020 ) Cluster-robust standard errors accounted for recruiting site-level dependencies. Sex-stratified models were estimated. Among females, analyses were additionally stratified by menopausal status (ACP females: self-reported status and/or ≥365 days from last self-reported menstrual cycle; in CMSAN females menopausal status was defined by age: ≤45 years [premenopausal] vs >45 years [postmenopausal]). 2.6.2. Objective 1b (Exposure-risk models). For exposure-risk, logistic regression models with MS case-control status as the outcome were conducted: i) a dose-response model with each standardized PFAS as a linear predictor; ii) a monotonic trend model with a quartile-categorized predictor for each PFAS; iii) a non-linear model including indicator variables for quartiles (Q) 2–4 compared with Q1 for each PFAS; and iv) a weighted quantile sum (WQS) regression model to evaluate mixture effects of PFHxS, PFOA, and PFOS (Stata package wqsreg ) given their moderately strong correlations ( eFigure 4 ). The positive-direction WQS specification was empirically based on the directionally consistent individual PFAS associations. Models adjusted for age, sex, BMI, and smoking status, with cluster-robust standard errors to account for recruiting site-level dependencies. Sex-stratified models were estimated. 2.6.3. Objective 2 (PFAS levels in MS vs other neuroinflammatory diseases). Analyses parallel to Objective 1a were conducted. To compare MS with NMOSD, mixed-effects models were fit pooling ACP and CMSAN data, with each PFAS as the dependent variable and MS vs NMOSD status as the primary fixed effect, a random intercept for site (ACP recruiting site and CMSAN) to account for between-site heterogeneity, with covariate adjustment described above. Multivariable linear regression model described for Objective 1a were conducted for MS vs NMOSD within each cohort, and for MS vs MOGAD in CMSAN alone. Sex-stratified models and female models stratified by menopausal status were estimated. Sensitivity analyses including disease duration as a covariate were explored. 2.6.4. Pooled analysis (PFAS levels in MS vs non-MS (HC + NMOSD + MOGAD)). To evaluate cross-cohort and cross-phenotype consistency, mixed-effects models were fit pooling ACP and CMSAN data, with LC-MS–derived PFAS levels as the dependent variable and MS vs non-MS (HC, NMOSD, or MOGAD) status as the primary fixed effect (binary variable; MS=1), adjusting for age, sex, BMI, and smoking status, with site as a random intercept. Intraclass correlation coefficients (ICC) quantified the proportion of variance attributable to site-level differences. The inclusion of both early untreated RRMS (ACP) and a broader clinical spectrum of MS patients (CMSAN) enabled assessment across disease stages and phenotypes, done to increase external validity. Cohort-specific estimates were examined to confirm consistency in direction and magnitude of associations, in addition to sensitivity analyses restricting to MS, NMOSD, and MOGAD cases who were within five years from onset, to support the appropriateness of pooled inference. 2.6.5. Exploratory Objective 3a (PFAS×MS status interactions in sex-related hormones). Given the results of Objectives 1–2, together with prior evidence that PFAS have female-specific endocrine effects,( Rodriguez-Carrillo et al., 2023a ; Yang et al., 2025 ) analyses were restricted to ACP females who reported days from last menses at sample collection (1 RRMS female had missing data and excluded). Fifty-one sex-related hormone and sterol metabolites ( eTable 1 ) were modeled as outcomes with PFAS level, MS-HC status, and their interaction (PFAS×MS) as predictors, adjusting for age, BMI, smoking, time of blood draw, hours fasted, and menstrual-phase (days from last menses was categorized as: 0–7d early follicular, 8–14d late follicular, 15–28d luteal phase, 29–59d likely anovulatory, 60–364d perimenopausal, and ≥365d postmenopausal). Exploratory interaction terms tested whether associations between PFAS and hormonal metabolite levels differed by MS status, consistent with a biological hypothesis that PFAS may differentially influence endocrine pathways in women with MS. As internal validity checks, both estrone-3-sulfate peak-intensities and age varied across self-reported menstrual-phase categories in the expected physiological patterns supporting the validity of self-report and the biological fidelity of the metabolomic hormone measurements ( eFigure 5 ). 2.6.6. Exploratory Objective 3b (PFAS-associated gene-expression correlates). Given the results of Objectives 1–2, Spearman correlations were computed between PFAS levels and gene expression (RNA-seq) data in ACP MS females with available data (n=23). For each PFAS, the top 2.5% of genes (by absolute correlation) were submitted to STRING for functional analysis.( Szklarczyk et al., 2023 ) It evaluates protein-protein interaction enrichment (whether observed connectivity amongst submitted genes exceeds random expectation given network size and degree distribution) and overrepresentation using strength (log 10 [observed/expected]). Significance criteria. Benjamini-Hochberg method controlled the false discovery rate (FDR) for all comparisons within Objective 1, 2, 3a, and 3b (q<0.05 was considered statistically significant; Stata package qqvalue ). 3. Results A total of 214 participants from ACP (73 MS, 106 HC, 35 NMOSD) and 225 participants from CMSAN (153 MS, 24 NMOSD, 48 MOGAD) were included ( Table 1 ). Across cohorts, individuals with MS and controls were similar with respect to age, BMI, and smoking prevalence, although MS participants were modestly older in CMSAN. In ACP, RRMS cases had a mean disease duration of 1.2 years (SD 1.1) and no recent immunomodulatory treatment, reflecting an early, treatment-naïve RRMS population. In CMSAN, MS participants had a broader range of disease duration and clinical course (67% RRMS, 14% SPMS, 18% PPMS), and 99.3% were untreated in the prior 90 days. As expected, NMOSD and MOGAD cases differed from MS in age, sex, and treatment status. Unadjusted comparisons suggest elevated PFHxS in MS versus HCs and NMOSD in ACP, and PFOA and PFOS in MS versus NMOSD, while all three PFAS were higher in MS versus NMOSD, and trending higher versus MOGAD in CMSAN. Once adjusting for age and sex, progressive MS cases did not differ in PFAS levels from RRMS cases in CMSAN ( eTable 2 ). Table 1. Characteristics of ACP (MS, HC, NMOSD) and CMSAN (MS, NMOSD, MOGAD) study participants. ACP cohort Attribute MS (mean [SD]; % [n]) HC (mean [SD]; % [n]) MS vs HC P NMOSD (mean [SD]; % [n]) MS vs NMOSD P N 73 106 35 Age (years) 38.8 (10.0) 41.0 (11.8) 0.26 45.5 (14) 0.014 Birth year 1970 (10) 1967 (12) 0.12 1965 (14) 0.063 Female 78% (n=57) 71.7% (n=76) 0.39 60% (n=21) 0.066 BMI (kg/m 2 ) 27.0 (5.7) 27.0 (5.9) 0.92 28.7 (5.5) 0.11 Ever smoker 47.9% (n=35) 34.0% (n=36) 0.064 40% (n=14) 0.54 Disease duration (years) 1.2 (1.1) - - 1.2 (1.3) 0.88 Disease duration ≤5 years 100% (n=73) - - 100% (n=35) - Immunomodulatory treatment in prior 90 days 0% - - 0% - Clinical course RRMS 100% (n=73) - - - - SPMS 0% - - - - PPMS 0% - - - - Aquaporin-4 antibody positive - - - 51.4% (n=18) - Log-transformed, standardized intensities PFHxS 0.25 (0.84) −0.09 (1.12) 0.018 −0.25 (0.83) 0.0055 PFOA 0.19 (0.81) 0.07 (1.04) 0.18 −0.59 (1.04) 0.00012 PFOS 0.15 (0.93) 0.09 (1.01) 0.81 −0.57 (0.92) 0.00056 CMSAN cohort Attribute MS (mean [SD]; % [n]) NMOSD (mean [SD]; % [n]) MS vs NMOSD P MOGAD (mean [SD]; % [n]) MS vs MOGAD P N 153 24 48 Age (years) 46.6 (14.0) 54.2 (15.6) 0.0073 35.8 (8.2) <0.001 Birth year 1972 (14) 1964 (16) 0.0031 1983 (8) <0.001 Female 60.0% (n=92) 87.5% (n=21) 0.011 47.9% (n=23) 0.17 BMI (kg/m 2 ) 28.6 (6.6) 27.7 (7.0) 0.48 28.2 (8.3) 0.33 Ever smoker 26.8% (n=41) 25% (n=6) 1 27.1% (n=13) 1 Disease duration (years) 5.9 (9.4) 3.4 (4.4) 0.69 2.2 (3.0) 0.95 Disease duration ≤5 years 70.6% (n=108) 70.8% (n=17) 85.4% (n=41) Immunomodulatory treatment in prior 90 days 0.7% (n=1) 33.3% (n=8) <0.001 41.7% (n=20) <0.001 Clinical course RRMS 67.3% (n=103) - - - - SPMS 14.4% (n=22) - - - - PPMS 18.3% (n=28) - - - - Aquaporfin-4 antibody positive - 100% - - - Myelin oligodendrocyte glycoprotein antibody positive - - - 100% - Log-transformed, standardized intensities PFHxS 0.14 (0.99) −0.45 (0.76) 0.0015 −0.13 (0.90) 0.055 PFOA 0.17 (0.95) −0.44 (0.78) 0.0012 −0.25 (0.94) 0.0048 PFOS 0.12 (0.97) −0.26 (0.79) 0.046 −0.15 (0.95) 0.070 Open in a new tab Abbreviations: ACP: Accelerated Cure Project for MS; BMI: body mass index; CMSAN: Center for Multiple Sclerosis and Autoimmune Neurology (Mayo Clinic); HC: healthy controls; MOGAD: myelin oligodendrocyte glycoprotein antibody disease; MS: multiple sclerosis; NMOSD: neuromyelitis optica spectrum disorder; PFHxS, perfluorohexanesulfonate; PFOA, perfluorooctanoate; PFOS, perfluorooctanesulfonate; PPMS: primary progressive MS; RRMS: relapsing remitting MS; SPMS: secondary progressive MS 3.1. Objective 1a: PFAS levels in MS vs HC ( Table 2 , Figure 1 ). Table 2. Associations for PFAS levels with MS versus HC, NMOSD, and MOGAD, overall and stratified by sex. Comparison Cohort Chemical All Females Males β (95% CI) P FDR q d β (95% CI) P FDR q d β (95% CI) P FDR q d MS vs HC (Objective 1a) ACP a N 73 RRMS vs 106 HC 57 RRMS vs 76 HC 16 RRMS vs 30 HC PFHxS 0.38 (0.17, 0.59) 0.0025 0.0069 0.44 (0.16, 0.72) 0.0058 0.014 0.06 (−0.30, 0.41) 0.71 0.74 PFOA 0.16 (−0.00, 0.33) 0.054 0.097 0.24 (−0.03, 0.50) 0.075 0.12 −0.15 (−0.80, 0.50) 0.60 0.68 PFOS 0.14 (−0.16, 0.45) 0.32 0.43 0.19 (−0.24, 0.61) 0.34 0.44 −0.08 (−0.60, 0.44) 0.72 0.74 MS vs NMOSD (Objective 2) ACP + CMSAN b , c N 226 MS vs 59 NMOSD 149 MS vs 42 NMOSD 77 MS vs 17 NMOSD PFHxS 0.52 (0.26, 0.77) 6.3×10 −5 0.00031 0.53 (0.23, 0.83) 0.00047 0.0015 0.53 (0.05, 1.00) 0.031 0.059 PFOA 0.76 (0.50, 1.02) 6.8×10 −9 2.4×10 −7 0.80 (0.49, 1.10) 2.9×10 −7 4.0×10 −6 0.67 (0.19, 1.16) 0.0064 0.014 PFOS 0.64 (0.40, 0.89) 3.3×10 −7 4.0×10 −6 0.54 (0.24, 0.84) 0.00045 0.0015 0.89 (0.47, 1.32) 3.7×10 −5 .00022 MS vs MOGAD (Objective 2) CMSAN a N 153 MS vs 46 MOGAD 92 MS vs 22 MOGAD 61 MS vs 24 MOGAD PFHxS 0.31 (−0.01, 0.63) 0.058 0.099 0.42 (−0.02, 0.87) 0.061 0.10 0.20 (−0.27, 0.67) 0.40 0.50 PFOA 0.33 (0.00, 0.66) 0.010 0.021 0.60 (0.13, 1.08) 0.013 0.026 0.07 (−0.39, 0.52) 0.77 0.77 PFOS 0.08 (−0.22, 0.37) 0.60 0.68 0.23 (−0.19, 0.65) 0.27 0.37 −0.09 (−0.52, 0.34) 0.67 0.73 MS vs non-MS (HC, NMOSD, MOGAD) (Pooled Analysis) ACP + CMSAN b N 226 MS vs 211 non-MS 149 MS vs 140 non-MS 77 MS vs 71 non-MS PFHxS 0.35 (0.18, 0.53) 6.9×10 −5 0.00031 0.42 (0.20, 0.65) 0.00018 0.00072 0.19 (−0.08,0.47) 0.17 0.24 PFOA 0.39 (0.21, 0.57) 3.4×10 −5 0.00022 0.50 (0.28, 0.72) 8.2×10 −6 7.4×10 −5 0.09 (−0.22, 0.41) 0.55 0.66 PFOS 0.31 (0.13, 0.48) 0.00057 0.0017 0.31 (0.09, 0.52) 0.0052 0.013 0.22 (−0.07, 0.50) 0.14 0.21 Open in a new tab Abbreviations: ACP, Accelerated Cure Project; CMSAN, Center for Multiple Sclerosis and Autoimmune Neurology (Mayo Clinic); HC, healthy controls; MOGAD, myelin oligodendrocyte glycoprotein antibody-associated disease; MS, multiple sclerosis; NMOSD, neuromyelitis optica spectrum disorder; PFHxS, perfluorohexanesulfonate; PFOA, perfluorooctanoate; PFOS, perfluorooctanesulfonate; RRMS: relapsing remitting multiple sclerosis a. Results from multivariable models that adjusted for age, sex, BMI, and ever smoker. The standard errors for ACP models accounted for recruiting site. b. Results from mixed-effect models that adjusted for age, sex, BMI, and ever smoker, with a random intercept for recruiting site. c. Results stratified by cohort are presented in eTable 4 . d. Q-values reflect Benjamini–Hochberg false discovery rate correction across all tests presented in this table. Figure 1. Open in a new tab Box plots of standardized PFAS levels unadjusted for additional covariates stratified by phenotype, cohort, and sex. Asterisks denoted p-values from multivariable regression models presented in Table 2 : * p<0.05; ** p<0.01; *** p<0.001 In ACP, PFHxS was significantly higher in MS than HC overall (β=0.38, 95% CI 0.17, 0.59, q=0.0069), and PFOA showed a similar but attenuated pattern that did not reach significance after multiple-testing correction (β=0.16, 95% CI −0.00, 0.33, q=0.097); PFOS did not differ (β=0.14, q=0.43). The PFHxS association remained evident among females with MS compared with female HCs (β=0.44, q=0.014), while PFOA showed a similar trend (β=0.24, q=0.12); male estimates were imprecise and not statistically significant (q>0.6). Female results were generally consistent when stratified by menopausal status, given the reduced sample sizes ( eTable 3 ). Irrespective of disease status, males had higher PFAS levels than females ( Figure 1 ). Models adjusting for collection year or birth year yielded similar results but are not presented because of multicollinearity with age. 3.2. Objective 1b: Exposure-risk models for MS vs HC ( Table 3 ). Table 3. Adjusted odds ratios for MS case-control status according to individual, quartile-based, and weighted quantile sum PFAS measures (Objective 1b). Model Chemical All Females Males 73 RRMS vs 106 HC 57 RRMS vs 76 HC 16 RRMS vs 30 HC OR (95% CI) P FDR q e OR (95% CI) P FDR q e OR (95% CI) P FDR q e Continuously PFHxS 1.53 (1.20, 1.96) 0.00066 0.0043 1.56 (1.20, 2.23) 0.0010 0.0051 1.12 (0.61, 2.04) 0.71 0.78 PFOA 1.21 (0.99, 1.49) 0.066 0.16 1.35 (0.94, 1.94) 0.11 0.24 0.88 (0.56, 1.39) 0.58 0.68 PFOS 1.21 (0.85, 1.71) 0.29 0.43 1.25 (0.82, 1.89) 0.30 0.43 0.84 (0.35, 2.02) 0.70 0.78 Monotonic trend (quartiles) PFHxS 1.54 (1.28, 1.86) 4.5×10 −6 0.00015 1.56 (1.25, 1.94) 6.6×10 −5 0.00015 1.21 (0.75, 1.94) 0.43 0.53 PFOA 1.35 (1.15, 1.58) 0.00018 0.0021 1.38 (1.06, 1.80) 0.017 0.065 1.21 (0.68, 2.16) 0.52 0.63 PFOS 1.15 (0.86, 1.54) 0.33 0.43 1.23 (0.86, 1.77) 0.25 0.43 0.68 (0.31, 1.47) 0.32 0.43 Non-linear trend (quartiles) PFHxS Q1 Ref Ref Ref Q2 1.95 (0.99, 3.84) 0.052 0.14 1.73 (0.98, 3.03) 0.057 0.15 0.83 (0.26, 2.64) 0.76 0.79 Q3 3.67 (1.69, 7.96) 0.00098 0.0051 3.08 (1.05, 9.05) 0.041 0.13 1.57 (0.73,3.35) 0.25 0.43 Q4 3.67 (2.06, 6.56) 1.1×10 −5 0.00017 3.59 (1.81, 7.10) 0.00024 0.0021 - - - PFOA Q1 Ref Ref Ref Q2 1.78 (0.65, 4.86) 0.26 0.43 2.22 (0.49, 10.1) 0.30 0.43 0.51 (0.02, 11.8) 0.68 0.78 Q3 2.29 (0.80, 6.55) 0.12 0.24 2.86 (0.67, 12.2) 0.16 0.29 0.80 (0.20, 3.21) 0.75 0.79 Q4 2.62 (1.56, 4.40) 0.00028 0.0021 2.77 (1.00, 7.73) 0.051 0.14 1.17 (0.27, 5.0) 0.84 0.86 PFOS Q1 Ref Ref Ref Q2 1.74 (0.84, 3.60) 0.14 0.27 1.43 (0.71, 2.90) 0.32 0.43 6.09 (0.64, 58.1) 0.12 0.24 Q3 2.32 (0.99, 5.44) 0.052 0.14 1.73 (0.90, 3.34) 0.10 0.23 4.84 (1.17, 20.1) 0.030 0.11 Q4 1.56 (0.54, 4.53) 0.41 0.52 1.88 (0.55, 6.48) 0.32 0.43 - - - Weighted Quantile Sum (Mixture) PFHxS, PFOA & PFOS 1.49 (1.08, 2.05) 0.016 0.065 1.63 (1.14, 2.33) 0.0070 0.032 1.45 (0.51, 2.15) 0.91 0.92 Open in a new tab Abbreviations: HC, healthy controls; PFHxS, perfluorohexanesulfonate; PFOA, perfluorooctanoate; PFOS, perfluorooctanesulfonate; RRMS: relapsing remitting multiple sclerosis; Q, quartile a. Continuous model: Odds ratios represent the adjusted odds of MS per 1 standard deviation increase in PFAS. Models adjusted for age, sex, smoker status, and BMI, with standard errors clustered by site. b. Monotonic trend (quartile score model): PFAS quartiles were modeled as an ordinal variable (coded 1–4) to estimate the adjusted odds of MS per quartile increase, assuming a linear monotonic dose–response relationship. Models adjusted for age, sex, smoker status, and BMI, with standard errors clustered by site. c. Nonlinear (indicator quartile model): PFAS quartiles were modeled as indicator variables with the lowest quartile (Q1) as the reference category, allowing for non-monotonic associations. Models adjusted for age, sex, smoker status, and BMI, with standard errors clustered by site. d. Weighted Quantile Sum (WQS) model: A weighted quantile sum index was constructed from PFHxS, PFOA, and PFOS quartiles to evaluate the joint association of the PFAS mixture with MS. Odds ratios represent the adjusted odds of MS per unit increase in the WQS index. Models adjusted for age, sex, smoker status, BMI, and site. e. Q-values reflect Benjamini–Hochberg false discovery rate correction across all tests presented Exposure-risk models demonstrated several positive associations with MS, with more precise and statistically significant estimates among females than males. In continuous models, higher PFHxS was associated with increased odds of MS overall (OR per SD=1.53, 95% CI: 1.20, 1.96, q=0.0043) and among females (OR=1.56, 95% CI: 1.20, 2.23), but male estimates were attenuated and imprecise (OR=1.12, 95% CI: 0.61, 2.04). Continuous PFOA showed a directionally similar but non-significant pattern overall (OR=1.21, q=0.16) and among females (OR=1.35, q=0.24), but not among males (OR=0.88, q=0.68), whereas PFOS associations were not significant (q>0.4). Monotonic quartile models showed dose-response relationships for PFHxS overall and among females (OR>1.5, q<0.0005), for PFOA overall (OR=1.35, q<0.005) with a similar trend among females (OR=1.38, q=0.065), while male estimates were less precise (q>0.5). In non-linear models, individuals in the highest PFHxS quartile had markedly elevated odds of MS compared with the lowest quartile (Q4 vs Q1: OR=3.67, 95% CI: 2.06, 6.56, q=0.00017; Q3 vs Q1: OR=3.67, 95% CI: 1.69, 7.96, q=0.0051); a similar pattern was observed for the highest PFOA quartile (Q4 vs Q1: OR=2.62, 95% CI: 1.56, 4.40, q=0.0021) – and these trends were also observed in females with MS. Lastly, WQS regression showed a suggestive overall mixture association with MS (OR=1.49, 95% CI: 1.08, 2.05, q=0.065) and a significant association among females (OR=1.63, 95% CI: 1.14, 2.33, q=0.032), with directionally similar but imprecise male estimates (OR=1.43, q=0.92). PFHxS received the largest weight in the overall and female models, exceeding the equal-weight threshold (1/3), indicating that PFHxS was the primary contributor to the mixture association ( eFigure 6 ). 3.3. Objective 2: PFAS levels in MS vs other neuroinflammatory diseases ( Table 2 ; Figure 1 ). In ACP+CMSAN combined, MS cases had higher PFHxS, PFOA, and PFOS than NMOSD cases (q<0.0005; Table 2 ), and these associations were consistent in ACP and CMSAN separately ( eTable 4 ). In sex-stratified comparisons, PFHxS, PFOA, and PFOS were elevated in both females and males with MS relative to their NMOSD counterparts, although within-cohort male estimates were less consistent, particularly in CMSAN, reflecting limited sample size ( eTable 4 ). In CMSAN, compared with MOGAD, MS cases showed higher PFOA (q=0.021), a similar trend for PFHxS (q=0.099), and no difference for PFOS (q=0.68). Similar female patterns were observed in sex stratified analyses (PFOA q=0.026, PFHxS q=0.10), whereas male estimates were near the null and imprecise (q≥0.5). These MS vs NMOSD and MS vs MOGAD results were directionally similar in models additionally adjusting for collection year, birth year, or disease duration, but those models are not presented because of multicollinearity with age. Associations were also consistent when stratifying female models by menopausal status, although sample sizes were reduced ( eTable 3 ), and when restricting analyses to cases within five years of onset ( eTable 5 ), thereby aligning CMSAN case definitions more closely with ACP. 3.4. Pooled analysis: MS vs non-MS ( Table 2 ; Figure 2 ). Figure 2. Open in a new tab Box plots of standardized PFAS levels unadjusted for additional covariates for the pooled MS versus non-MS (HC, NMOSD and MOGAD) observations stratified by sex. Asterisks denoted p-values from multivariable mixed-effect regressions presented in Table 3 : * p<0.05; ** p<0.01; *** p<0.001, **** p<0.00001 Pooling ACP and CMSAN data across phenotypes, MS cases had significantly higher PFHxS (β=0.35, 95% CI 0.18, 0.53, q=0.00031), PFOA (β=0.39, 95% CI 0.21, 0.57, q=0.00022), and PFOS (β=0.31, 95% CI 0.13, 0.48, q=0.0017) than the non-MS group (HC+NMOSD+MOGAD combined). These differences were significant in MS females (q<0.05) but not in males where effect estimates were closer to the null (q>0.2). Although PFOS showed a significant pooled association, individual comparisons from Objectives 1–2 did not demonstrate uniformly consistent MS–non-MS differences, suggesting that the pooled PFOS result may reflect a weaker, more heterogeneous signal detected with greater statistical power. Between-site variance was minimal (ICC≤0.04 across models), and sensitivity analyses restricted to MS, NMOSD, and MOGAD participants who were five years of onset were consistent ( eTable 5 ), supporting the pooled approach and generalizability of findings. 3.5. Objective 3a: Exploratory PFAS-MS interactions in sex-related hormones in females ( eTable 6 , Figure 3 ). Figure 3. Open in a new tab PFOA × MS interaction plots for the sex hormone metabolites identified in Objective 3a (FDR q<0.05). Slopes represent model-based linear fits. Given prior Swedish findings of elevated PFHxS and PFOA in females with MS, the suggestive female biased associations for PFHxS and PFOA observed in Objective 1, and prior evidence that PFAS may exert female-specific endocrine effects, these exploratory analyses were restricted to ACP MS and HC females with available menstrual-timing data ( eTable 7 ). Eleven sex hormone metabolites showed nominally significant associations (p<0.05), and three MS×PFOA interactions remained significant after multiple-testing correction (FDR q<0.05; eTable 6 ; Figure 3 ): 5α-pregnan-3β,20α-diol disulfate, pregnenolone/21-hydroxypregnenolone disulfate, and androstenediol disulfate. For each, higher PFOA was associated with greater hormone abundance among MS females but not HCs. Stratified models confirmed positive PFOA–hormone associations within MS females and null associations within HC females. 3.6. Objective 3b: Exploratory PFAS-associated gene-expression correlates ( Figure 4 ). Figure 4. Open in a new tab STRING functional enrichment of the top 2.5% of genes correlated with PFOA (A) and PFHxS (B) (FDR q<0.05). Given the findings from Objectives 1–2, these hypothesis-generating analyses were restricted to 23 ACP females with MS and available transcriptomic data ( eTable 7 ). Among the top 2.5% of genes correlated with PFOA, there was greater protein-protein interaction than expected by chance (p<5×10 −8 ), and overrepresentation of UniProt categories related to secreted, signaling-associated, glycoprotein-containing, and disulfide-bond-forming; synapse-related terms were also observed in the primary analysis ( Figure 4A ). For PFHxS-correlated genes, there was also significant protein-protein interaction enrichment (p=0.00064). GO Cellular Component terms were significantly enriched for dendritic spine and synaptic membrane components ( Figure 4B ). Noteworthy, only 17% of PFHxS- and PFOA gene-sets overlapped ( eTable 8 ). Post hoc sensitivity analyses using 1% and 5% gene-selection thresholds showed broadly similar enrichment patterns overall, although the synapse-related annotation observed for PFOA at the 2.5% threshold was not retained at the alternative cutoffs. 4. Discussion Our findings indicate that PFHxS was elevated in females with MS compared with HC females and females with NMOSD or MOGAD across two independent U.S. cohorts, whereas corresponding male estimates were not consistent and often imprecise. Several comparisons suggested higher PFOA in females with MS, including quartile-based exposure-risk models and comparisons with females with NMOSD and MOGAD. These results corroborate recent European findings and extend them to a U.S. population with likely higher baseline PFAS exposures, supporting the hypothesis that PFAS may contribute to MS susceptibility, with associations appearing more evident among females in our data. The less consistent male findings may in part reflect limited precision and warrant further investigation. Our first set of exploratory findings suggested that PFOA may have MS-specific associations with several sex-related hormone metabolites, including progestin-, pregnenolone-, and androgen-related pathways. Finally, our second set of exploratory findings identified PFAS-related gene-expression signatures with enrichment of synaptic and neuronal membrane components, providing additional, though still exploratory, biological plausibility for CNS-relevant effects in females with MS. 4.1. Comparison with prior epidemiologic evidence Our findings both replicate and extend recent Swedish MS case-control studies (Objective 1 findings). The original Swedish report observed a sex-differentiated association, with elevated PFHxS and PFOA in MS females but not males, although continuous exposure-risk models were only suggestive.( Vaivade et al., 2025 ) A subsequent analysis incorporating non-linear exposure modeling identified increased MS odds in higher quartiles of PFOA and PFOS, along with a significant 14-PFAS mixture association; however, sex-stratified analyses were not reported.( Vaivade et al., 2026 ) In our U.S. cohorts, complementary exposure-risk modeling approaches demonstrated the most consistent associations for PFHxS, particularly among females. Continuous, monotonic quartile, and non-linear models revealed strong dose-response relationships, with individuals in the highest PFHxS quartiles exhibiting markedly elevated odds of MS. Similar but less uniform patterns were observed for PFOA, whereas PFOS associations were generally weaker. WQS regression further supported a mixture association, particularly in females, with PFHxS receiving the greatest weight and contributing most strongly to the overall signal. While male-specific estimates were imprecise due to smaller sample size, some effect estimates were directionally similar. Together, these findings strengthen evidence that PFAS-MS associations may vary by sex, with more consistent associations observed among females in our cohorts, and extend prior European observations by demonstrating robust dose-response and mixture effects in U.S. populations, especially for PFHxS. Our study adds new insights, as it is the first to compare PFAS levels in MS with those in NMOSD and MOGAD. Across two cohorts for NMOSD and one cohort for MOGAD, higher PFHxS and PFOA levels were consistently observed among MS females than females with these other inflammatory demyelinating diseases, whereas male estimates were less consistent. This cross-disease pattern, particularly among females, supports the hypothesis that PFHxS and PFOA are not generalized biomarkers of neuroinflammation or demyelination but may be more specifically associated with MS, with this pattern more evident among women. Given the long biological half-lives of PFHxS and PFOA, measured serum levels likely reflect cumulative exposure over years rather than short-term fluctuations related to disease activity or treatment. This reduces the likelihood that the observed differences are driven by reverse causation and is more consistent with potential etiologic mechanisms. PFAS exposure may differentially influence immune pathways central to MS pathogenesis, such as B-cell signaling and oligodendrocyte biology,( Butruille et al., 2023 ; Rudzanova et al., 2024 ) rather than the antigen-specific autoantibody mechanisms that define NMOSD and MOGAD. Nonetheless, longitudinal and mechanistic studies are needed to clarify the biological basis of these MS-specific associations. Although PFOS did not show consistent pairwise associations, it reached significance in pooled analyses, suggesting a weaker or more heterogeneous signal. PFAS are not biologically interchangeable, and differences in toxicokinetics and tissue partitioning may be relevant. For example, experimental data indicate that PFOS neurotoxicity is more strongly linked to NMDA/AMPA receptor–mediated calcium influx, whereas PFOA and PFHxS appear less NMDA-dependent and may act through broader receptor and signaling pathways.( Brown-Leung and Cannon, 2022 ) Post-mortem data have also demonstrated high variability in PFAS accumulation and composition in the human brain, which may reflect differences in environmental PFAS mixtures. ( Brown-Leung and Cannon, 2022 ) These compound-specific differences in bioavailability and signaling pathways may partly explain why PFHxS and PFOA showed more consistent associations with MS than PFOS in our cohorts. 4.2. Endocrine disruption as a sex-specific vulnerability in MS Our exploratory Objective 3a findings suggest that PFOA may interact with steroid hormone metabolism among women with early-stage MS. PFOA was positively associated with several sulfated progestin, pregnenolone, and androgen metabolites among early-stage MS females, with no corresponding associations in healthy women. These findings are directionally consistent with a recent steroidomic analysis reporting altered pregnenolone-, progestin-, androgen-, and estrogen-related pathways in women with MS, including shifts toward conjugated steroid metabolites.( Kancheva et al., 2024 ) Notably, the metabolites associated with PFOA in our data fell within sulfate-conjugated steroid pathways. However, given the modest sample size and multiple-testing burden (even with FDR correction), these findings should be considered exploratory. In addition, these associations may reflect differences in steroid conjugation or sulfotransferase-related activity rather than a direct disruption of steroidogenesis. Experimental work has shown that PFOA can modulate progesterone receptor activity and downstream transcriptional responses in human endometrial cells,( Di Nisio et al., 2020b ) supporting endocrine plausibility. Overall, these results raise the possibility that PFAS exposure may interact with steroid hormone metabolism in women with MS, although larger studies with mechanistic follow-up are needed. Beyond sex-steroid metabolism, PFAS have also been linked to endocrine pathways relevant to MS biology, including vitamin D and thyroid hormone signaling. PFOA can bind the vitamin D receptor (VDR) and alter VDR-dependent transcription, ( Di Nisio et al., 2020a ) and PFAS exposure has been associated with altered thyroid hormone profiles in adolescents. ( Rodriguez-Carrillo et al., 2023b ) While these pathways were not evaluated here and do not appear to show clear sex-specific associations with MS, they provide additional context for the broader endocrine framework through which PFAS may influence neuroimmune function. 4.3. PFAS-Linked Neuroimmune Mechanisms Our exploratory transcriptomic enrichment results provide tentative support for PFAS-related mechanisms relevant to MS. Experimental studies demonstrate that PFHxS, PFOA, and PFOS cross central nervous system barriers, particularly under inflammatory conditions, and can disrupt intracellular calcium signaling and synaptic function – though their neurotoxic properties differ.( Brown-Leung and Cannon, 2022 ; Running et al., 2024 ; Starnes et al., 2022 ; Wang et al., 2018 ) Consistent with this, we observed enrichment of dendritic spine and synaptic membrane gene sets – notably the PFHxS- and PFOA-gene sets modestly overlapped. PFAS exposure have also been linked to immune dysregulation relevant to MS pathogenesis. Experimental and human studies demonstrate perturbations in B-cell maturation, germinal center activity, and T-cell subset balance following PFOA exposure, with evidence of sex-dependent effects on myelin integrity and immune-cell infiltration.( Neisiani et al., 2023 ; Rudzanova et al., 2024 ; Taylor et al., 2023 ; Tursi et al., 2024 ) Together, these findings provide biologically plausible context for the more consistent PFAS associations observed among females in our study, although these mechanistic interpretations remain exploratory and require targeted validation. 4.4. Public health context PFAS exposures disproportionately affect marginalized U.S. communities, including rural, Black, Hispanic, and low-income populations, reflecting longstanding environmental injustices.( Chukwuonye et al., 2024 ; Liddie et al., 2023 ; Maruzzo et al., 2025 ; Sukhram et al., 2025 ) Although our study was restricted to non-Hispanic White participants and therefore cannot directly evaluate these disparities, the broader public health context underscores the importance of incorporating neurological and neuroimmune endpoints into PFAS risk assessment frameworks. Certain populations disproportionately burdened by PFAS exposure, including Black Americans and individuals from lower socioeconomic backgrounds, also experience elevated MS risk.( Briggs et al., 2014 ; Langer-Gould et al., 2022 ) While our data do not address these population-level inequities directly, they highlight the need for future longitudinal studies in more diverse populations to determine whether chronic PFAS exposure may contribute to demographic or geographic disparities in MS incidence. Continued regulatory efforts and longitudinal investigations are needed to clarify these potential population-level implications.( O’Sullivan Bakshi et al., 2025 ) 4.5. Limitations and future directions This study has limitations. The case-control study design and single-time-point PFAS measurements preclude inference regarding temporality and cumulative exposure, although the long biological half-lives of PFAS reduce concern for acute reverse causation. In addition, PFAS were quantified using an untargeted LC-MS platform that yields semi-quantitative peak intensity measures rather than calibrated serum concentrations, limiting direct comparison with clinically defined PFAS thresholds (e.g., those proposed by the National Academies of Sciences, Engineering, and Medicine) and estimation of total PFAS burden based on absolute concentration cutoffs. Sample sizes for some contrasts were modest, particularly in the NMOSD and MOGAD groups, and RNA-seq analyses were conducted in a small whole-blood subset. Although covariates were carefully harmonized and pooled models incorporated site-level random effects, residual confounding by diet, geography, health behavior, and water-source exposures remains possible ( eFigure 3 ). Other unmeasured determinants of PFAS burden, including parity and breastfeeding history, may also have contributed to variability in PFAS levels but were unavailable across both biobanks. Participants were non-Hispanic White Americans, limiting generalizability. In addition, menopausal status was self-reported in the ACP cohort but unavailable in CMSAN, where an age-based cutoff was used instead. Finally, while the pooled non-MS comparator group was intentionally heterogeneous, combining healthy and inflammatory neurologic controls to test broader disease specificity, it may obscure subtype-specific differences. Future work should include longitudinal sampling, formal evaluation of additive and multiplicative interactions to clarify potential sex-specific relationships, multi-ethnic cohorts, more comprehensive assessment of short and long-chain PFAS and other persistent chemicals (i.e., hydroxylated polychlorinated biphenyls), detailed exposure-route assessment (i.e., diet, occupation, geography) and toxicokinetic studies in persons with MS, considerations for social determinants of health, and detailed mechanistic analyses. While we were able to explore relationships in females by menopausal status, future work needs to confirm and expand these findings in large sample sizes. Also meriting closer investigation are the relationships among PFAS, menstruation, pregnancy, and lactation, which represent important elimination routes for PFAS in females.( Rickard et al., 2022 ) 5. Conclusions In summary, across two independent U.S. cohorts we observed more consistent, elevations of PFHxS, and to a lesser extent PFOA, among females with MS, less consistent and often imprecise estimates among males, and clear separation from NMOSD and MOGAD, supporting disease-specific associations and possible heterogeneity by sex. Complementary exposure-risk models demonstrated robust dose-response relationships, particularly for PFHxS, which emerged as the primary contributor in mixture analyses. Exploratory analyses further suggested potential biological relevance: PFOA-sex hormone associations differed by MS status, and PFAS-correlated gene-expression signatures showed exploratory enrichment of neuronal membrane components. Although temporality cannot be established, findings were consistent when restricting to those early in their disease processes, extending prior European observations to U.S. populations. Given the ubiquity of PFAS exposure and its disproportionate burden in marginalized communities, clarifying PFAS-MS relationships may inform future MS prevention research and environmental health policy. Supplementary Material 1 NIHMS2179662-supplement-1.docx (1MB, docx) 2 NIHMS2179662-supplement-2.xlsx (996.6KB, xlsx) Highlights. Female-biased PFHxS and PFOA elevations in U.S. MS cases Robust dose-response and non-linear MS-PFAS associations, particularly in females PFHxS drove mixture effects in weighted quantile models Exploratory analyses link PFAS altered steroid metabolites and possibly synaptic genes in MS Acknowledgements We thank Dr. Kristy Miskimen and Dr. Jonathan Haines for their invaluable contributions to sample processing and management. Funding Statement FB, JS, ME, OT and metabolomic profiling of the samples were supported by the NIH/NINDS R01NS121928. CFS, GIS, LSW, CLS and the RNA-seq profiling were supported by NIH/NIAID R43AI157674 and R44AI195280. Footnotes Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. Conflicts of Interest and Disclosure The authors have no financial or other conflicts of interest that might have biased their contributions to the current study. Declaration of interests The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Farren Briggs reports financial support was provided by National Institute of Neurological Disorders and Stroke. W. Oliver Tobin reports financial support was provided by National Institute of Neurological Disorders and Stroke. Jessica Sagen reports financial support was provided by National Institute of Neurological Disorders and Stroke. Mahboobeh Fereidan Esfahani reports financial support was provided by National Institute of Neurological Disorders and Stroke. Chase Spurlock reports financial support was provided by National Institute of Allergy and Infectious Diseases. Guzel Shaginurova reports financial support was provided by National Institute of Allergy and Infectious Diseases. Lukasz Wylezinski reports was provided by National Institute of Allergy and Infectious Diseases. Cheryl Sesler reports financial support was provided by National Institute of Allergy and Infectious Diseases. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Contributor Information Farren B. S. Briggs, Department of Public Health Sciences, University of Miami Miller School of Medicine, Miami, FL, USA. Federico Montini, Center for Multiple Sclerosis and Autoimmune Neurology, Mayo Clinic, Rochester, MN, USA. Charles F. Spurlock, III, Decode Health, Inc., Department of Medicine, Vanderbilt University School of Medicine, Thomas F. Frist, Jr. College of Medicine, Belmont University, Nashville, TN, USA; New York University, New York, NY, USA. Christine Ryan, University of Miami Miller School of Medicine, Miami, FL, USA. Jessica N. Cooke Bailey, Department of Pharmacology and Toxicology, East Carolina University, Greenville, NC, USA. Erika S. Trapl, Department of Population and Quantitative Health Sciences, Case Western Reserve University School of Medicine, Cleveland, OH, USA. Guzel I. Shaginurova, Decode Health, Inc., Nashville, TN, USA. Lukasz S. Wylezinski, Decode Health, Inc., Department of Medicine, Vanderbilt University School of Medicine, Nashville, TN, USA. Cheryl L. Sesler, Decode Health, Inc., Nashville, TN, USA. Mahboobeh Fereidan Esfahani, Department of Neurology, The University of Texas at Austin Dell Medical School, Austin TX, USA. Jessica Sagen, Center for Multiple Sclerosis and Autoimmune Neurology, Mayo Clinic, Rochester, MN, USA. Hannah E. Gardener, Department of Neurology, University of Miami Miller School of Medicine, Miami, FL, USA. Jacob L McCauley, John P. Hussman Institute for Human Genomics, University of Miami Miller School of Medicine, Miami, FL, USA. Alberto J. Caban-Martinez, Department of Public Health Sciences, University of Miami Miller School of Medicine, Miami, FL, USA. Roberta Brambilla, Department of Neurological Surgery, University of Miami Miller School of Medicine, Miami, FL, USA. Lilyana Amezcua, Department of Neurology, University of Southern California Keck School of Medicine, Los Angeles, CA, USA. W. Oliver Tobin, Center for Multiple Sclerosis and Autoimmune Neurology, Mayo Clinic, Rochester, MN, USA. Data Availability and Access The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. FB has full access to the data used in analyses presented. References . Toxicological Profile for Perfluoroalkyls, Atlanta (GA), 2021. [ PubMed ] [ Google Scholar ] Agency, U. E. P., 2024. PFAS Explained. Andrews DQ, Naidenko OV, 2020. Population-Wide Exposure to Per- and Polyfluoroalkyl Substances from Drinking Water in the United States. Environmental Science & Technology Letters. 7, 931–936. [ Google Scholar ] ATSDR, 2024. PFAS Information for Clinicians. Banwell B, et al. , 2023. Diagnosis of myelin oligodendrocyte glycoprotein antibody-associated disease: International MOGAD Panel proposed criteria. The Lancet Neurology. 22, 268–282. [ DOI ] [ PubMed ] [ Google Scholar ] Botelho JC, et al. , 2025. Per- and polyfluoroalkyl substances (PFAS) exposure in the U.S. population: NHANES 1999-March 2020. Environ Res. 270, 120916. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Briggs FB, et al. , 2014. Adverse socioeconomic position during the life course is associated with multiple sclerosis. J Epidemiol Community Health. 68, 622–9. [ DOI ] [ PubMed ] [ Google Scholar ] Brown-Leung JM, Cannon JR, 2022. Neurotransmission Targets of Per- and Polyfluoroalkyl Substance Neurotoxicity: Mechanisms and Potential Implications for Adverse Neurological Outcomes. Chem Res Toxicol. 35, 1312–1333. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Butruille L, et al. , 2023. Deleterious functional consequences of perfluoroalkyl substances accumulation into the myelin sheath. Environ Int. 180, 108211. [ DOI ] [ PubMed ] [ Google Scholar ] Chukwuonye GN, et al. , 2024. Toxic layering and compound extremes: Per- and polyfluoroalkyl substances (PFAS) exposure in rural, environmental justice copper mining communities. Sci Total Environ. 957, 177767. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Cousins IT, et al. , 2020. The high persistence of PFAS is sufficient for their management as a chemical class. Environ Sci Process Impacts. 22, 2307–2312. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Di Nisio A, et al. , 2020a. Endocrine disruption of vitamin D activity by perfluoro-octanoic acid (PFOA). Sci Rep. 10, 16789. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Di Nisio A, et al. , 2020b. Perfluorooctanoic acid alters progesterone activity in human endometrial cells and induces reproductive alterations in young women. Chemosphere. 242, 125208. [ DOI ] [ PubMed ] [ Google Scholar ] Engelhardt AJ, et al. , 2023. Anthropogenic Organic Contaminants Analysed in Human Blood and Combined Risk. Exposure and Health. 15, 551–565. [ Google Scholar ] Engelhardt JA, et al. , 2025. An extended PFAS profiling of a Swedish subpopulation and mixture risk assessments using multiple approaches. Environ Int. 195, 109214. [ DOI ] [ PubMed ] [ Google Scholar ] Gharibi T, et al. , 2020. The role of B cells in the immunopathogenesis of multiple sclerosis. Immunology. 160, 325–335. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Gluge J, et al. , 2020. An overview of the uses of per- and polyfluoroalkyl substances (PFAS). Environ Sci Process Impacts. 22, 2345–2373. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Jordan M, et al. , 2024. Exogenous factors associated with legacy PFAS concentrations in the general U.S. population: NHANES 1999–2018. Emerging Contaminants and Environmental Health. 3, 15. [ Google Scholar ] Kancheva R, et al. , 2024. Altered Steroidome in Women with Multiple Sclerosis. Int J Mol Sci. 25. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Langer-Gould AM, et al. , 2022. Racial and Ethnic Disparities in Multiple Sclerosis Prevalence. Neurology. 98, e1818–e1827. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Liddie JM, et al. , 2023. Sociodemographic Factors Are Associated with the Abundance of PFAS Sources and Detection in U.S. Community Water Systems. Environ Sci Technol. 57, 7902–7912. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Lopez JA, et al. , 2021. Pathogenesis of autoimmune demyelination: from multiple sclerosis to neuromyelitis optica spectrum disorders and myelin oligodendrocyte glycoprotein antibody-associated disease. Clin Transl Immunology. 10, e1316. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Maruzzo AJ, et al. , 2025. Socioeconomic Disparities in Exposures to PFAS and Other Unregulated Industrial Drinking Water Contaminants in US Public Water Systems. Environ Health Perspect. 133, 17002. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] McDonald WI, et al. , 2001. Recommended diagnostic criteria for multiple sclerosis: guidelines from the International Panel on the diagnosis of multiple sclerosis. Ann Neurol. 50, 121–7. [ DOI ] [ PubMed ] [ Google Scholar ] Murgia F, et al. , 2022. Sex Hormones as Key Modulators of the Immune Response in Multiple Sclerosis: A Review. Biomedicines. 10. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Neisiani AK, et al. , 2023. Perfluorooctanoic acid exposure and its neurodegenerative consequences in C57BL6/J mice. Naunyn Schmiedebergs Arch Pharmacol. 396, 2357–2367. [ DOI ] [ PubMed ] [ Google Scholar ] O’Sullivan Bakshi S, et al. , 2025. Per- and polyfluoroalkyl substances (PFAS) in the United States: Current knowledge and regulatory context. Sci Total Environ. 1003, 180711. [ DOI ] [ PubMed ] [ Google Scholar ] Olsson T, et al. , 2017. Interactions between genetic, lifestyle and environmental risk factors for multiple sclerosis. Nat Rev Neurol. 13, 25–36. [ DOI ] [ PubMed ] [ Google Scholar ] Pang Z, et al. , 2024. MetaboAnalyst 6.0: towards a unified platform for metabolomics data processing, analysis and interpretation. Nucleic Acids Res. 52, W398–W406. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Polman CH, et al. , 2011. Diagnostic criteria for multiple sclerosis: 2010 revisions to the McDonald criteria. Ann Neurol. 69, 292–302. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Pomazal R, et al. , 2024. Determinants of per- and polyfluoroalkyl substances (PFAS) exposure among Wisconsin residents. Environ Res. 254, 119131. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Qi W, et al. , 2020. Per- and Polyfluoroalkyl Substances and Obesity, Type 2 Diabetes and Non-alcoholic Fatty Liver Disease: A Review of Epidemiologic Findings. Toxicol Environ Chem. 102, 1–36. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Richterova D, et al. , 2023. PFAS levels and determinants of variability in exposure in European teenagers - Results from the HBM4EU aligned studies (2014–2021). Int J Hyg Environ Health. 247, 114057. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Rickard BP, et al. , 2022. Per- and poly-fluoroalkyl substances (PFAS) and female reproductive outcomes: PFAS elimination, endocrine-mediated effects, and disease. Toxicology. 465, 153031. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Rodriguez-Carrillo A, et al. , 2023a. PFAS association with kisspeptin and sex hormones in teenagers of the HBM4EU aligned studies. Environ Pollut. 335, 122214. [ DOI ] [ PubMed ] [ Google Scholar ] Rodriguez-Carrillo A, et al. , 2023b. Association of exposure to perfluoroalkyl substances (PFAS) and phthalates with thyroid hormones in adolescents from HBM4EU aligned studies. Environ Res. 237, 116897. [ DOI ] [ PubMed ] [ Google Scholar ] Rudzanova B, et al. , 2024. Altered Transcriptome Response in PBMCs of Czech Adults Linked to Multiple PFAS Exposure: B Cell Development as a Target of PFAS Immunotoxicity. Environ Sci Technol. 58, 90–98. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Running L, et al. , 2024. Investigating the Mechanism of Neurotoxic Effects of PFAS in Differentiated Neuronal Cells through Transcriptomics and Lipidomics Analysis. ACS Chem Neurosci. 15, 4568–4579. [ DOI ] [ PubMed ] [ Google Scholar ] Ryals J, et al. , 2007. Metabolon, Inc. Pharmacogenomics. 8, 863–6. [ DOI ] [ PubMed ] [ Google Scholar ] Sesler CL, et al. , 2025. Development of an End-to-End Total RNA Sequencing Quality Control Framework for Blood-Based Biomarker Discovery. J Mol Diagn. 27, 869–881. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Starnes HM, et al. , 2022. A Critical Review and Meta-Analysis of Impacts of Per- and Polyfluorinated Substances on the Brain and Behavior. Front Toxicol. 4, 881584. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Sukhram SD, et al. , 2025. PFAS Exposure, Mental Health, and Environmental Justice in the United States: Impacts on Marginalized Communities. Int J Environ Res Public Health. 22. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Szklarczyk D, et al. , 2023. The STRING database in 2023: protein-protein association networks and functional enrichment analyses for any sequenced genome of interest. Nucleic Acids Res. 51, D638–D646. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Taylor KD, et al. , 2023. Quantifying the impact of PFOA exposure on B-cell development and antibody production. Toxicol Sci. 194, 101–108. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Thompson AJ, et al. , 2018. Diagnosis of multiple sclerosis: 2017 revisions of the McDonald criteria. Lancet Neurol. 17, 162–173. [ DOI ] [ PubMed ] [ Google Scholar ] Tursi AR, et al. , 2024. Immune cell profiles associated with human exposure to perfluorinated compounds (PFAS) suggest changes in natural killer, T helper, and T cytotoxic cell subpopulations. Environ Res. 256, 119221. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Vaivade A, et al. , 2025. Associations of PFAS and OH-PCBs with risk of multiple sclerosis onset and disability worsening. Nat Commun. 16, 2014. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Vaivade A, et al. , 2026. Co-exposure to PFAS and hydroxylated PCBs is associated with increased odds of multiple sclerosis. Environ Int. 207, 109993. [ DOI ] [ PubMed ] [ Google Scholar ] Wang J, et al. , 2018. Penetration of PFASs Across the Blood Cerebrospinal Fluid Barrier and Its Determinants in Humans. Environ Sci Technol. 52, 13553–13561. [ DOI ] [ PubMed ] [ Google Scholar ] Wingerchuk DM, et al. , 2015. International consensus diagnostic criteria for neuromyelitis optica spectrum disorders. Neurology. 85, 177–89. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Wingerchuk DM, et al. , 2007. The spectrum of neuromyelitis optica. Lancet Neurol. 6, 805–15. [ DOI ] [ PubMed ] [ Google Scholar ] Wylezinski LS, et al. , 2024. Machine Learning Analysis Using RNA Sequencing to Distinguish Neuromyelitis Optica from Multiple Sclerosis and Identify Therapeutic Candidates. J Mol Diagn. 26, 520–529. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Yang LX, et al. , 2025. Female-specific associations of serum perfluoroalkyl and polyfluoroalkyl substances with sex hormonal/insulin dysregulation: An integrated population-based study. Ecotoxicol Environ Saf. 298, 118273. [ DOI ] [ PubMed ] [ Google Scholar ] Yeung MSY, et al. , 2019. Dynamics of oligodendrocyte generation in multiple sclerosis. Nature. 566, 538–542. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Supplementary Materials 1 NIHMS2179662-supplement-1.docx (1MB, docx) 2 NIHMS2179662-supplement-2.xlsx (996.6KB, xlsx) Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. FB has full access to the data used in analyses presented. ACTIONS View on publisher site PDF (1.0 MB) Cite Collections Permalink RESOURCES Similar articles Cited by other articles Links to NCBI Databases Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top

Related documents

• Association between perfluoroalkyl and polyfluoroalkyl substances and risk of obstructive sleep apnea#343176
• Kappa Index Versus Csf Oligoclonal Bands in Diagnosis of ms and Prediction of Disease Activity#857636
• Impact of cancer on multiple sclerosis–related healthcare and disease-modifying drug use: A multinational cohort study#337511
• A Study to Evaluate the Effect of SAR443820 on Serum Neurofilament Levels in Male and Female Adult Participants With Multiple Sclerosis#1060186
• MultiSCRIPT-Cycle 1: Personalized Medicine in Multiple Sclerosis - Pragmatic Platform Trial Embedded Within the SMSC#1173754