CoChat
Sign Up
Cited by 89Multiple Sclerosis Research StudiesGenetic Associations and EpidemiologyPsoriasis: Treatment and Pathogenesis

Genome‐wide association studies of multiple sclerosis

Open access
Cotsapas, Chris, Mitrovic, Mitja·2018·Clinical & Translational Immunology
DOI 10.1002/cti2.1018PMC PMC5983059
Open this paper in CoChat — free accountOpen PDF

Abstract

Large‐scale genetic studies of multiple sclerosis have identified over 230 risk effects across the human genome, making it a prototypical common disease with complex genetic architecture. Here, after a brief historical background on the discovery and definition of the disease, we summarise the last fifteen years of genetic discoveries and map out the challenges that remain to translate these findings into an aetiological framework and actionable clinical understanding.

AI snapshot

Written from the paper’s full text.

TL;DR

This review summarizes 15 years of multiple sclerosis (MS) genome-wide association studies, which have identified over 230 risk effects and strongly implicate immune, especially T-cell, biology. It argues that functional interpretation and clinical translation remain difficult because of fine-mapping, regulatory-variant, and data limitations.

Key findings

  • The latest IMSGC GWAS brought the total to 233 MS risk associations: 200 autosomal variants outside the MHC, 1 on the X chromosome, and 32 independent MHC effects; all combined explain 19.2% of total heritability, with MHC effects accounting for 4% and approximately 2% residing in regions not initially associated or not replicated.
  • Milestones include the 2007 GWAS of 1540 parent-affected offspring trios identifying IL-2RA and IL-7RA; a 2011 GWAS of 9772 cases and 17,376 controls replicating 23/26 prior associations and finding 29 novel loci; and an ImmunoChip study of 29,300 cases and 50,794 controls identifying 48 new variants, bringing the total to 110 variants at 103 non-MHC loci.
  • MS risk variants localize mainly to regulatory regions, especially enhancers active in stimulated immune cells; in the latest IMSGC GWAS, 104/200 non-MHC loci overlapped eQTLs active in prefrontal cortex or immune cells, yielding 212 potentially relevant eQTLs, 45 of which were present only in prefrontal cortex.
  • Colocalization analysis of 59 densely genotyped loci found eQTLs within 100 kb of the most associated MS variant in 56 loci, but only 14/56 showed evidence that eQTL and MS risk were driven by the same underlying signal; 11/14 matches were in CD4+ T cells. A separate DHS analysis found significant enrichment in 25/48 MS risk loci, due to 177 DHS sites, and identified 49 genes in 17/25 loci.

Methods

  • Narrative review of MS genetic studies, covering the linkage and candidate-gene era, GWAS, MHC dissection, fine-mapping, eQTL and DNase I hypersensitive site integration, autoimmune disease overlap, and future directions.
  • Synthesizes published GWAS and meta-analyses and consortium datasets, including IMSGC GWAS with over 115,000 cases and controls, ImmunoChip data in 29,300 cases and 50,794 controls, over 48,000 MHC samples, and NIH Roadmap Epigenome Mapping Consortium data.
  • No new primary cohort is described; the review reports prior findings and the authors’ integrative analyses, including a joint likelihood approach for eQTL colocalization and a statistical framework for identifying risk-driving accessible chromatin regions.

Limitations

  • The paper states that fine mapping remains difficult because linkage disequilibrium means many variants show association but only one is likely causal; in one analysis, only 8/68 loci resolved to fewer than five candidate variants, and a relevant candidate gene was identified in three.
  • The paper states that eQTL colocalization is challenging and that simply finding an eQTL in a GWAS locus can produce spurious causal inferences; it also notes that CNS data are absent or limited in most datasets, potentially causing ascertainment bias, and that secondary phenotypes such as sex, disease course, progression, and therapy response lack genome-wide significant associations and detailed cross-center data.
  • Evident from the text: as a narrative review, it does not describe a systematic search strategy or formal quality assessment, so its conclusions depend on the selected cited studies and authors’ synthesis.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1002/cti2.1018

Cites

80 of 80 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

56 of 91 cited sources checked · 9 supported · 1 flagged · accuracy not yet measured

  • Premisecitation

    By 2011, common variants in 26 genomic loci had been associated with multiple sclerosis risk and independently replicated.

    “by 2011, common variants in 26 genomic loci had been associated with MS risk and independently replicated”
    Quote in the paperNot a substantial claim7 of 10 sources checked
    • cites [33] Wellcome Trust Case Control Consortium. Association scan of 14,500 nonsynonymous SNPs in four diseases identifies autoimmunity variants. Nat Genet 2007; 39: 1329– 1337. 17952073 PMC2680141 source

      The Results (Association with nsSNPs; page unavailable) identify IL7R among genes with variants previously associated with MS, supporting the general point that genetic variants had been linked to MS risk. The Abstract (page unavailable) reports independent replication of two loci for ankylosing spondylitis, not MS. This 2007 paper does not support the claim that 26 MS loci had been independently replicated by 2011. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
    • cites [34] Comabella M, Craig DW, Camiña‐Tato M et al Identification of a novel risk locus for multiple sclerosis at 13q31.3 by a pooled genome‐wide scan of 500,000 single nucleotide polymorphisms. PLoS ONE 2008; 3: e3490. 18941528 PMC2566815 source

      The Results report replication of an MS-associated variant on chromosome 13 in a separate US cohort, supporting the general point that risk associations were independently replicated. But this 2008 paper does not report that 26 loci had been associated with MS risk and independently replicated by 2011 (searched for the count, date, and replication claim). (LLM, conf=0.82)

      Experimental — accuracy not yet measured
    • cites [35] de Jager PL, Jia X, Wang J et al Meta‐analysis of genome scans and replication identify CD6, IRF8 and TNFRSF1A as new multiple sclerosis susceptibility loci. Nat Genet 2009; 41: 776– 782. 19525953 PMC2757648 source

      The Results (GWAS meta-analysis and replication; page not provided) report MS-risk associations tested in replication samples, and the Discussion lists validated and strongly suggested susceptibility loci. But this 2009 paper does not establish that 26 loci had been independently replicated by 2011; its Discussion also distinguishes strongly suggested loci from validated ones. (LLM, conf=0.87)

      Experimental — accuracy not yet measured
    • cites [36] Baranzini SE, Wang J, Gibson RA et al Genome‐wide association analysis of susceptibility and clinical phenotype in multiple sclerosis. Hum Mol Genet 2009; 18: 767– 778. 19010793 PMC4334814 source

      The Introduction (page unavailable) identifies several MS susceptibility genes, including IL2RA, CD58, CLEC16A and EVI5, as independently replicated. It does not report 26 replicated genomic loci or establish a count as of 2011; the Results (page unavailable) also caution that this study lacked full-scale independent replication. (LLM, conf=0.88)

      Experimental — accuracy not yet measured
    • cites [37] Australia and New Zealand Multiple Sclerosis Genetics Consortium (ANZgene). Genome‐wide association study identifies new multiple sclerosis susceptibility loci on chromosomes 12 and 20. Nat Genet 2009; 41: 824– 828. 19525955 source
    • cites [38] Jakkula E, Leppä V, Sulonen A‐M et al Genome‐wide association study in a high‐risk isolate for multiple sclerosis reveals associated variants in STAT3 gene. Am J Hum Genet 2010; 86: 285– 291. 20159113 PMC2820168 source
    • cites [39] Nischwitz S, Cepok S, Kroner A et al Evidence for VAV2 and ZNF433 as susceptibility genes for multiple sclerosis. J Neuroimmunol 2010; 227: 162– 166. 20598377 source
    • cites [40] Patsopoulos NA; the Bayer Pharma MS Genetics Working Group; the Steering Committees of Studies Evaluating IFNβ‐1b and a CCR1‐Antagonist et al Genome‐wide meta‐analysis identifies novel multiple sclerosis susceptibility loci. Ann Neurol 2011; 70: 897– 912. 22190364 PMC3247076 source

      The Introduction reports over a dozen MS susceptibility loci, and the Results discuss additional known loci. But the paper does not establish a count of 26 independently replicated loci: the Results say its findings are not independent replication evidence for IL2RA, CD58, or STAT3, and the Discussion says the new loci require replication. No page numbers are available in our copy. (LLM, conf=0.88)

      Experimental — accuracy not yet measured
    • cites [41] International Multiple Sclerosis Genetics Consortium. The genetic association of variants in CD6, TNFRSF1A and IRF8 to multiple sclerosis: a multicenter case‐control study. PLoS ONE 2011; 6: e18813. 21552549 PMC3084233 source

      The Introduction lists multiple MS-associated loci, and the Discussion reports independent replication of variants at TNFRSF1A and near IRF8. Neither section states that common variants at 26 loci had been independently replicated by 2011; the count and its replication qualifier are unsupported. (LLM, conf=0.86)

      Experimental — accuracy not yet measured
    • cites [42] Sanna S, Pitzalis M, Zoledziewska M et al Variants within the immunoregulatory CBLB gene are associated with multiple sclerosis. Nat Genet 2010; 42: 495– 497. 20453840 PMC3786343 source

      The paper’s Abstract describes multiple MS susceptibility loci and reports confirmation of previously described markers at known non-HLA loci. It does not state that common variants at 26 loci had been associated with MS and independently replicated by 2011; its detailed follow-up concerns a CBLB variant. (LLM, conf=0.82)

      Experimental — accuracy not yet measured
  • Premisecitation

    Candidate gene studies are underpowered to detect risk alleles for common complex disease.

    “such studies are underpowered to detect risk alleles for common complex disease”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [22] Altshuler D, Daly MJ, Lander ES. Genetic mapping in human disease. Science 2008; 322: 881– 888. 18988837 PMC2694957 source

      In “Statistical analysis” (page unavailable), the paper says four modest-sized replication studies lacked adequate power to detect a type 2 diabetes risk variant. But “Genetic association in populations” (page unavailable) describes the broader problem with candidate-gene studies as their limited choice of variants and weak statistical support; it does not say candidate-gene studies generally were underpowered to detect risk alleles for common complex disease. (LLM, conf=0.84)

      Experimental — accuracy not yet measured
  • Premisecitation

    Candidate gene studies often assess genes that have broad relevance to the immune system but do not drive disease risk per se.

    “often assess genes that have broad relevance to the immune system but do not drive disease risk per se.22”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [22] Altshuler D, Daly MJ, Lander ES. Genetic mapping in human disease. Science 2008; 322: 881– 888. 18988837 PMC2694957 source

      In “Genetic association in populations” (page unavailable), the paper says candidate-gene studies tested biological candidates with a low prior probability of being disease-causing, and that few reported associations were replicated. It does not say those candidates were often genes with broad immune-system relevance, or establish that the genes themselves did not affect disease risk. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
  • Premisecitation

    Large-scale genetic studies of multiple sclerosis have identified over 230 risk effects across the human genome.

    “Large‐scale genetic studies of multiple sclerosis have identified over 230 risk effects across the human genome”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [46] International Multiple Sclerosis Genetics Consortium. The multiple sclerosis genomic map: role of peripheral immune cells and resident microglia in susceptibility. bioRxiv 2017: 143933. PMC7241648 31604244 source

      The indexed copy returned under this DOI is headed “NR1H3 p.Arg415Gln is not associated to multiple sclerosis risk,” rather than the named genomic-map paper. Its Abstract discusses particular MS-risk variants, but neither it nor the passages returned by searches for “over 230 risk effects” and “233 independent effects” gives a genome-wide count of more than 230. (LLM, conf=0.76)

      Experimental — accuracy not yet measured
  • Premisecitation

    Compilation of large reference populations with both serological and genotyping data on MHC variation has made imputation of classical alleles possible from standard SNP array data to single amino acid resolution.

    “Over the last several years, the compilation of large reference populations with both serological and genotyping data on MHC variation has made imputation of classical alleles possible from standard SNP array data to single amino acid resolution.48, 49”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites [48] de Bakker PIW, McVean G, Sabeti PC et al A high‐resolution HLA and SNP haplotype map for disease association studies in the extended human MHC. Nat Genet 2006; 38: 1166– 1172. 16998491 PMC2670196 source

      The Abstract (page not specified) reports that SNP tags captured many common classical HLA alleles, and Methods > Empirical validation (page not specified) describes testing how well those tags predicted patients’ HLA alleles. The paper does not establish imputation from standard SNP-array data to single-amino-acid resolution; Methods > HLA typing describes PCR-based typing rather than serological typing. (LLM, conf=0.88)

      Experimental — accuracy not yet measured
    • cites [49] Leslie S, Donnelly P, McVean G. A statistical method for predicting classical HLA alleles from SNP data. Am J Hum Genet 2008; 82: 48– 56. 18179884 PMC2253983 source
  • May combine two claims. 32
  • Classified as probably not a substantial claim. 8

  • May combine two claims.

    Premisecitation

    The largest single risk factor for MS is biological sex, with >75% of patients being female.

    “The largest single risk factor for MS is biological sex, with > 75% of patients being female – but the causes for this discrepancy in incidence are unknown.73”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [73] Confavreux C, Compston A. The natural history of multiple sclerosis In: Compston A ed. McAlpine's Multiple Sclerosis. Philadelphia, PA: Elsevier, 2006: 183– 272. source

      The section “The sex ratio in multiple sclerosis” (page unavailable) confirms female predominance, but reports a usual ratio of two females to one male and says the highest reported female proportion is 71%. That directly contradicts the claim that more than 75% of patients are female; the passage also does not rank sex as the largest single risk factor. (LLM, conf=0.88)

      Checked by a reasoning model — may be wrong.

      Experimental — accuracy not yet measured

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Multiple sclerosis is a neurological disorder of the central nervous system resulting from an autoimmune attack on CNS white matter.

    “Multiple sclerosis [MS, (MIM 126200)] is a neurological disorder of the central nervous system (CNS), resulting from an autoimmune attack on CNS white matter.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [4] Kabat EA, Moore DH, Landow H. An electrophoretic study of the protein components in cerebrospinal fluid and their relationship to the serum proteins. J Clin Invest 1942; 21: 571– 577. 16694947 PMC435175 source

      The Results (page unavailable) examine cerebrospinal fluid from five people with multiple sclerosis, supporting the paper’s relevance to the disease. Neither those results nor the Discussion passage on antibody production—which concerns neurosyphilis—states that MS results from an autoimmune attack on CNS white matter. (LLM, conf=0.76)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    Linkage analyses in extended multiple sclerosis affected families validated the HLA association but showed no significant linkage to loci outside the MHC.

    “These validated the HLA association but showed no significant linkage to loci outside the MHC.”
    Quote in the paperSubstantial claim2 of 10 sources checked
    • cites [23] Barcellos LF, Thomson G. Genetic analysis of multiple sclerosis in Europeans. J Neuroimmunol 2003; 143: 1– 6. 14575906 source
    • cites [24] Sawcer S, Ban M, Maranian M et al A high‐density screen for linkage in multiple sclerosis. Am J Hum Genet 2005; 77: 454– 467. 16080120 PMC1226210 source
    • cites [25] International HapMap Consortium. A haplotype map of the human genome. Nature 2005; 437: 1299– 1320. 16255080 PMC1880871 source

      The Abstract lists HLA as a validated example in autoimmunity generally, but does not discuss multiple sclerosis affected families or their linkage results. Searches for multiple sclerosis linkage and linkage outside the MHC found no passage addressing whether those families showed significant linkage at other loci. (LLM, conf=0.76)

      Experimental — accuracy not yet measured
    • cites [26] International HapMap Consortium. Integrating common and rare genetic variation in diverse human populations. Nature 2010; 467: 52– 58. 20811451 PMC3173859 source

      Our copy’s Regional sequencing and Haplotype sharing sections describe population samples and genetic variation, not linkage analyses in multiple-sclerosis-affected families. Searches for multiple sclerosis, HLA, and linkage outside the MHC found no passage addressing the claimed result; the paper’s use of “linkage” concerns linkage disequilibrium. (LLM, conf=0.76)

      Experimental — accuracy not yet measured
    • cites [27] Risch N, Merikangas K. The future of genetic studies of complex human diseases. Science 1996; 273: 1516– 1517. 8801636 source
    • cites [28] Pe'er I, Yelensky R, Altshuler D, Daly MJ. Estimation of the multiple testing burden for genomewide association studies of nearly all common variants. Genet Epidemiol 2008; 32: 381– 385. 18348202 source
    • cites [29] Lander ES. The new genomics: global views of biology. Science 1996; 274: 536– 539. 8928008 source
    • cites [30] International Multiple Sclerosis Genetics Consortium. Risk alleles for multiple sclerosis identified by a genomewide study. N Engl J Med 2007; 357: 851– 862. 17660530 source
    • cites [31] Gregory SG, Schmidt S, Seth P et al Interleukin 7 receptor alpha chain (IL7R) shows allelic and functional association with multiple sclerosis. Nat Genet 2007; 39: 1083– 1091. 17660817 source
    • cites [32] Lundmark F, Duvefelt K, Iacobaeus E et al Variation in interleukin 7 receptor alpha chain (IL7R) influences risk of multiple sclerosis. Nat Genet 2007; 39: 1108– 1113. 17660816 source

  • May combine two claims.

    Premisecitation

    Candidate gene studies suffer from population stratification and other artefacts.

    “suffer from population stratification and other artefacts”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [22] Altshuler D, Daly MJ, Lander ES. Genetic mapping in human disease. Science 2008; 322: 881– 888. 18988837 PMC2694957 source

  • May combine two claims.

    Findingcitation

    When a fine-mapping approach was applied to 6356 multiple sclerosis cases and 9617 controls from the United Kingdom, it only meaningfully resolved a small subset of associations, with 8/68 loci resolving to fewer than five candidate variants, and from these, a relevant candidate gene was identified in three.

    “When applied to 6356 MS cases and 9617 controls from the United Kingdom, this approach only meaningfully resolved a small subset of associations, with 8/68 loci we analysed resolving to fewer than five candidate variants, and from these, we have been able to identify a relevant candidate gene in three.45”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [45] International Multiple Sclerosis Genetics Consortium. Analysis of immune‐related loci identifies 48 new susceptibility variants for multiple sclerosis. Nat Genet 2013; 45: 1353– 1360. 24076602 PMC3832895 source

      Split into 2 atoms — 1 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 0 unverifiable. Expand the claim to see per-atom verdicts.

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    Pathway, transcriptomic and epigenetic enrichment analyses suggest T-cell biology is a major feature of multiple sclerosis, but also highlight the involvement of many other components of both adaptive and innate immunity in pathogenesis.

    “Again, careful pathway, transcriptomic and epigenetic enrichment analyses suggest T‐cell biology is a major feature of the disease, but also highlight the involvement of many other components of both adaptive and innate immunity in pathogenesis.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [46] International Multiple Sclerosis Genetics Consortium. The multiple sclerosis genomic map: role of peripheral immune cells and resident microglia in susceptibility. bioRxiv 2017: 143933. PMC7241648 31604244 source

      The supplied passages and additional searches return text headed “NR1H3 p.Arg415Gln is not associated to multiple sclerosis risk,” rather than the named genomic-map paper. Those passages address an NR1H3 variant, not pathway, transcriptomic or epigenetic enrichment of adaptive and innate immune cells. The claim cannot be assessed from the available copy. (LLM, conf=0.75)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Interpretationcitation

    Approximately 2% of overall heritability resides in regions that either did not show suggestive association in the initial genome-wide association study or failed to replicate in independent samples, suggesting that additional loci remain to be found.

    “However, a small portion – approximately 2% of the overall heritability – resides in regions that either did not show suggestive association in the initial GWAS or that failed to replicate in independent samples, suggesting that there remain additional loci to be found”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [46] International Multiple Sclerosis Genetics Consortium. The multiple sclerosis genomic map: role of peripheral immune cells and resident microglia in susceptibility. bioRxiv 2017: 143933. PMC7241648 31604244 source

      Split into 2 atoms — 0 confirmed, 0 partially supported, 0 disputed, 0 misattributed, 2 unverifiable. Expand the claim to see per-atom verdicts.

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    The most recent IMSGC genome-wide association study including over 115,000 cases and controls brings the total number of multiple sclerosis risk associations to 233, including 200 autosomal variants outside the MHC, one on the X chromosome and 32 independent effects in the broader MHC locus.

    “This latest report brings the total number of MS risk associations to 233, including 200 autosomal variants outside the MHC, one on the X chromosome and 32 independent effects in the broader MHC locus, covering both classical and nonclassical gene regions.46”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [46] International Multiple Sclerosis Genetics Consortium. The multiple sclerosis genomic map: role of peripheral immune cells and resident microglia in susceptibility. bioRxiv 2017: 143933. PMC7241648 31604244 source

      Split into 2 atoms — 0 confirmed, 0 partially supported, 0 disputed, 0 misattributed, 2 unverifiable. Expand the claim to see per-atom verdicts.

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    Multiple sclerosis genome-wide association study loci are enriched for expression quantitative trait loci in multiple tissues, supporting the idea that much risk is due to changes to gene regulation.

    “MS GWAS loci are also enriched for expression quantitative trait loci (eQTL) in multiple tissues,46, 57 supporting the idea that much risk is due to changes to gene regulation.”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites [46] International Multiple Sclerosis Genetics Consortium. The multiple sclerosis genomic map: role of peripheral immune cells and resident microglia in susceptibility. bioRxiv 2017: 143933. PMC7241648 31604244 source

      The indexed copy is labelled as an NR1H3 association paper, not the genomic-map paper named in the citation. Its Abstract discusses MS susceptibility at the NR1H3 locus but does not address eQTL enrichment across tissues or regulatory changes as an explanation for MS risk. Searches for those terms found no relevant passage in the available copy. (LLM, conf=0.78)

      Experimental — accuracy not yet measured
    • cites [57] Raj T, Rothamel K, Mostafavi S et al Polarization of the effects of autoimmune and neurodegenerative risk alleles in leukocytes. Science 2014; 344: 519– 523. 24786080 PMC4910825 source

      The paper reports significant cis-eQTL enrichment among autoimmune-disease GWAS variants in monocytes and/or T cells, and identifies multiple sclerosis susceptibility alleles as having predominantly T-cell-specific regulatory effects (Abstract; page unavailable). It does not establish that MS GWAS loci specifically are enriched for eQTLs in multiple tissues, or quantify how much MS risk is due to regulatory changes. (LLM, conf=0.84)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Interpretationcitation

    The vast majority of genome-wide association study loci encode genes obviously active in the immune system, and particularly in the lymphocyte lineage, placing beyond a doubt the nature of multiple sclerosis as autoimmune.

    “The vast majority of GWAS loci encode genes obviously active in the immune system,45, 63 and particularly in the lymphocyte lineage,64 placing beyond a doubt the nature of the disease as autoimmune.”
    Quote in the paperSubstantial claim2 of 3 sources checked
    • cites [45] International Multiple Sclerosis Genetics Consortium. Analysis of immune‐related loci identifies 48 new susceptibility variants for multiple sclerosis. Nat Genet 2013; 45: 1353– 1360. 24076602 PMC3832895 source

      The Gene Ontology passage (page unavailable) reports that a majority of 97 associated variants lie within 50 kb of genes with immunological functions, supporting an immune-system association. It does not establish that a vast majority of GWAS loci encode those genes; the paper also cautions that proximity does not necessarily indicate functional importance. The lymphocyte-lineage clause has a separate citation in the citing sentence. (LLM, conf=0.88)

      Experimental — accuracy not yet measured
    • cites [63] Chun S, Casparino A, Patsopoulos NA et al Limited statistical evidence for shared genetic effects of eQTLs and autoimmune‐disease‐associated loci in three major immune‐cell types. Nat Genet 2017; 45: 1353. PMC5374036 28218759 source

      The paper’s main text (page unavailable) links autoimmune-disease GWAS risk to regulatory regions, including enhancers active in stimulated T cells, and examines multiple sclerosis among the diseases studied. It does not establish that the vast majority of multiple-sclerosis GWAS loci encode genes obviously active in the immune system; it says the associations are primarily non-coding and that specific disease-relevant regulatory effects are difficult to identify. The separate citation following the lymphocyte-lineage phrase is not assessed here. (LLM, conf=0.84)

      Experimental — accuracy not yet measured
    • cites [64] Shooshtari P, Huang H, Cotsapas C. Integrative genetic and epigenetic analysis uncovers regulatory mechanisms of autoimmune disease. Am J Hum Genet 2017; 101: 75– 86. 28686857 PMC5501874 source

  • May combine two claims.

    Premisecitation

    The majority of multiple sclerosis risk variants appear to localise to gene regulatory regions, rather than coding sequence, and specifically to enhancer elements active in stimulated immune cell subsets.

    “The second challenge is that the majority of MS risk variants appear to localise to gene regulatory regions, rather than coding sequence,55 and specifically to enhancer elements active in stimulated immune cell subsets.56”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites [55] Maurano MT, Humbert R, Rynes E et al Systematic localization of common disease‐associated variation in regulatory DNA. Science 2012; 337: 1190– 1195. 22955828 PMC3771521 source

      The Abstract states that about 93% of disease- and trait-associated variants overall lie in noncoding sequence. The ‘De novo identification of pathogenic cell types’ section reports enrichment of MS-associated variants in regulatory DNA of particular immune cell types, but does not establish that a majority of MS risk variants lie in regulatory regions rather than coding sequence. The separate claim about enhancers in stimulated immune cells carries citation 56, not this citation. (LLM, conf=0.84)

      Experimental — accuracy not yet measured
    • cites [56] Farh KK‐H, Marson A, Zhu J et al Genetic and epigenetic fine mapping of causal autoimmune disease variants. Nature 2015; 518: 337– 343. 25363779 PMC4336207 source

  • May combine two claims.

    Premisecitation

    A 2011 collaborative genome-wide association study of 9772 cases and 17,376 controls of European descent replicated 23 of 26 previously identified associations and identified 29 novel risk loci.

    “A further expansion of the IMSGC resulted in a collaborative GWAS of 9772 cases and 17 376 controls, again of European descent, in 2011.43 This study replicated 23 of 26 previously identified associations and identified 29 novel risk loci.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [43] International Multiple Sclerosis Genetics Consortium; Wellcome Trust Case Control Consortium 2, Sawcer S et al Genetic risk and a primary role for cell‐mediated immune mechanisms in multiple sclerosis. Nature 2011; 476: 214– 219. 21833088 PMC3182531 source

  • May combine two claims.

    Premisecitation

    A targeted ImmunoChip study in 29,300 multiple sclerosis cases and 50,794 unrelated healthy individuals identified 48 new susceptibility variants, bringing the total number of MS risk variants to 110 at 103 discrete loci outside the MHC.

    “A further study, this time on a targeted array (the ImmunoChip44) in 29 300 MS cases and 50 794 unrelated healthy individuals, identified 48 new susceptibility variants, bringing the total number of MS risk variants to 110 at 103 discrete loci outside the MHC.45”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites [44] Cortes A, Brown MA. Promise and pitfalls of the Immunochip. Arthritis Res Ther 2011; 13: 101. 21345260 PMC3157635 source
    • cites [45] International Multiple Sclerosis Genetics Consortium. Analysis of immune‐related loci identifies 48 new susceptibility variants for multiple sclerosis. Nat Genet 2013; 45: 1353– 1360. 24076602 PMC3832895 source

      The Abstract reports 48 new susceptibility variants and a total of 110 established variants at 103 non-MHC loci. The case and control totals also match, but only 14,498 cases and 24,091 controls were analysed on the ImmunoChip; the remaining participants contributed previous GWAS data in the replication phase. Thus, the sentence overstates the size of the targeted-array sample. (LLM, conf=0.88)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    These loci sometimes involve more than one eQTL per locus, for a total of 212 eQTLs potentially relevant to pathogenesis.

    “These sometimes involve more than one eQTL per locus, for a total of 212 eQTLs potentially being relevant to pathogenesis.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [46] International Multiple Sclerosis Genetics Consortium. The multiple sclerosis genomic map: role of peripheral immune cells and resident microglia in susceptibility. bioRxiv 2017: 143933. PMC7241648 31604244 source

      The indexed passages are headed “NR1H3 p.Arg415Gln is not associated to multiple sclerosis risk,” rather than the named genomic-map paper. Its Abstract discusses an NR1H3 association, but neither it nor searches for “212 eQTLs” and multiple eQTLs per locus address the claimed count or its relevance to pathogenesis. (LLM, conf=0.76)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    The 32 MHC effects accounted for 4% of the overall heritability, with the bulk of the remaining signal resident in other regions of the genome associated with multiple sclerosis risk.

    “The 32 MHC effects accounted for 4% of the overall heritability, with the bulk of the remaining signal resident in the other regions of the genome associated with MS risk.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [46] International Multiple Sclerosis Genetics Consortium. The multiple sclerosis genomic map: role of peripheral immune cells and resident microglia in susceptibility. bioRxiv 2017: 143933. PMC7241648 31604244 source

      The indexed passages are headed “NR1H3 p.Arg415Gln is not associated to multiple sclerosis risk” (Abstract and References; no page numbers supplied), despite the supplied source title. Searches for the 32 MHC effects, 4% heritability, and remaining genomic signal found no passage addressing those figures or their comparison. The claim cannot be verified from our copy. (LLM, conf=0.76)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    The nature of the interaction may be more complicated than expected, with multiple different amino acid-level alleles demonstrating consistent interaction with HLA-DRB1*15:01, in addition to the HLA-DQA1*01:01 allele.

    “However, the nature of the interaction may be more complicated than expected, with multiple different amino acid‐level alleles demonstrating consistent interaction with HLA‐DRB1*15:01, in addition to the HLA‐DQA1*01:01 allele.”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites [50] Moutsianas L, Jostins L, Beecham AH et al Class II HLA interactions modulate genetic risk for multiple sclerosis. Nat Genet 2015; 47: 1107– 1113. 26343388 PMC4874245 source

      The Results section, “Interactions among classical HLA alleles” (page not available), reports an interaction involving HLA-DRB1*15:01 and HLA-DQA1*01:01. It says the same interaction can alternatively be explained by HLA-DQB1 alleles encoding glycine at position 70. It does not establish multiple distinct amino-acid-level interactions in addition to the HLA-DQA1*01:01 interaction. (LLM, conf=0.87)

      Experimental — accuracy not yet measured
    • cites [51] Patsopoulos NA, Barcellos LF, Hintzen RQ et al Fine‐mapping the genetic association of the major histocompatibility complex in multiple sclerosis: HLA and non‐HLA effects. PLoS Genet 2013; 9: e1003926. 24278027 PMC3836799 source

      The Results section on amino acid residues in DRβ1 identifies several associated amino acid positions, and the Discussion says the DRB1 effect is more complex than a single amino acid model can explain. But these passages do not demonstrate consistent interactions between amino acid-level alleles and DRB1*15:01. The Results section on multi-allelic association discusses DQA1*01:02, not the claimed DQA1*01:01, and says its effect could be explained by DRB1*15:01 (page numbers unavailable). (LLM, conf=0.86)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    Several variants outside the classical regions of the MHC, both class I and class II, were shown to be independently associated, suggesting biological functions beyond antigen display underlie the MHC risk effects.

    “Several variants outside the classical regions of the MHC (both class I and class II) were also shown to be independently associated, suggesting biological functions beyond antigen display underlie the MHC risk effects.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites [50] Moutsianas L, Jostins L, Beecham AH et al Class II HLA interactions modulate genetic risk for multiple sclerosis. Nat Genet 2015; 47: 1107– 1113. 26343388 PMC4874245 source

      The Results section, “A high-resolution map of HLA risk for multiple sclerosis” (page unavailable), reports associated variants near HLA-DPB1 and in LTA, supporting part of the claim about variants beyond classical HLA alleles. It does not establish independently associated nonclassical variants in both class I and class II regions. The Discussion (page unavailable) says the disease mechanisms require functional study, rather than concluding that functions beyond antigen display underlie the risk effects. (LLM, conf=0.84)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    MS is a remarkably heterogeneous disease, with some patients declining rapidly and others showing few or no symptoms for decades.

    “MS is also a remarkably heterogeneous disease, with some patients declining rapidly and others showing few or no symptoms for decades.73”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [73] Confavreux C, Compston A. The natural history of multiple sclerosis In: Compston A ed. McAlpine's Multiple Sclerosis. Philadelphia, PA: Elsevier, 2006: 183– 272. source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationcitation

    The causes for the female predominance in MS incidence are unknown.

    “The largest single risk factor for MS is biological sex, with > 75% of patients being female – but the causes for this discrepancy in incidence are unknown.73”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites [73] Confavreux C, Compston A. The natural history of multiple sclerosis In: Compston A ed. McAlpine's Multiple Sclerosis. Philadelphia, PA: Elsevier, 2006: 183– 272. source

      The section ‘The sex ratio in multiple sclerosis’ (page unavailable) confirms female predominance, but reports a usual ratio of two females to one male and a highest reported female proportion of 71%, not >75%. It does not say that the causes of the predominance are unknown. (LLM, conf=0.85)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    Many loci were associated with multiple diseases and the genes encoded in those loci fall into distinct immune pathways.

    “many loci were associated with multiple diseases and that the genes encoded in those loci fall into distinct immune pathways.69, 70”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites [69] Cotsapas C, Hafler DA. Immune‐mediated disease genetics: the shared basis of pathogenesis. Trends Immunol 2013; 34: 22– 26. 23031829 source
    • cites [70] Cotsapas C, Voight BF, Rossin E et al Pervasive sharing of genetic effects in autoimmune disease. PLoS Genet 2011; 7: e1002254. 21852963 PMC3154137 source

      The Introduction reports that nearly half the loci identified for one disease influence at least two diseases. The Results find significant protein connectivity in three of four locus clusters and suggest that they represent distinct molecular mechanisms; two groups are preferentially expressed in immune-cell subtypes. The claim overstates this evidence by presenting distinct immune pathways for the genes as established rather than inferred. (LLM, conf=0.84)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    Several studies have explored the genetic basis of clinical course, age of onset and severity, although no genome-wide significant associations have been discovered.

    “Several studies have explored the genetic basis of clinical course, age of onset and severity, although no genome‐wide significant associations have been discovered.36, 37, 43, 74, 75, 76, 77”
    Quote in the paperNot a substantial claim2 of 3 sources checked
    • cites [36] Baranzini SE, Wang J, Gibson RA et al Genome‐wide association analysis of susceptibility and clinical phenotype in multiple sclerosis. Hum Mol Genet 2009; 18: 767– 778. 19010793 PMC4334814 source

      The Introduction describes earlier studies of genetic influences on MS severity and clinical course, and the genotype–phenotype analysis section says this study assessed age of onset and the severity metric MSSS. The cited passages do not establish the broader claim that no genome-wide significant associations had been discovered; the Discussion reports significant phenotype associations without stating whether they met a genome-wide threshold. Page numbers are unavailable in our copy. (LLM, conf=0.78)

      Experimental — accuracy not yet measured
    • cites [37] Australia and New Zealand Multiple Sclerosis Genetics Consortium (ANZgene). Genome‐wide association study identifies new multiple sclerosis susceptibility loci on chromosomes 12 and 20. Nat Genet 2009; 41: 824– 828. 19525955 source
    • cites [43] International Multiple Sclerosis Genetics Consortium; Wellcome Trust Case Control Consortium 2, Sawcer S et al Genetic risk and a primary role for cell‐mediated immune mechanisms in multiple sclerosis. Nature 2011; 476: 214– 219. 21833088 PMC3182531 source

      The paper’s results passage (indexed under Abstract; page unavailable) reports no genetic associations with clinical course or severity and no other individual SNPs strongly associated with age at onset. But it also says the analysis replicated a previously suggested association between age at onset and DRB1*15:01. The blanket claim that no associations were discovered omits that exception; the passage does not state whether it met a genome-wide significance threshold. (LLM, conf=0.84)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingown data

    The DNase I hypersensitive sites are preferentially accessible in immune cell subsets, particularly T cells and their precursors, and the 49 genes are strongly expressed in these tissues.

    “As expected, the DHS are preferentially accessible in immune cell subsets, particularly T cells and their precursors, and the 49 genes are strongly expressed in these tissues.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [64] Shooshtari P, Huang H, Cotsapas C. Integrative genetic and epigenetic analysis uncovers regulatory mechanisms of autoimmune disease. Am J Hum Genet 2017; 101: 75– 86. 28686857 PMC5501874 source

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    The currently acceptable standard for genome-wide association study significance is P < 5 × 10−8, equivalent to P < 0.05 after Bonferroni correction for the number of independent tests in the genome given linkage disequilibrium between common variants.

    “The currently acceptable standard is a significance level of P < 5 × 10−8, which is equivalent to P < 0.05 after Bonferroni correction for the number of independent tests in the genome given linkage disequilibrium between common variants.28”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [28] Pe'er I, Yelensky R, Altshuler D, Daly MJ. Estimation of the multiple testing burden for genomewide association studies of nearly all common variants. Genet Epidemiol 2008; 32: 381– 385. 18348202 source

  • May combine two claims.

    Premisecitation

    Multiple sclerosis aggregates in families, with siblings and offspring of people with MS being far more likely to develop the disease themselves.

    “aggregates in certain families and that siblings and offspring of people with MS are far more likely to develop the disease themselves.7”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [7] Compston A, Wekerle H. The genetics of multiple sclerosis In: Compston A, ed. McAlpine's Multiple Sclerosis. Philadelphia, PA: Elsevier, 2006: 113– 181. source

  • May combine two claims.

    Interpretationreasoning

    Multiple sclerosis is a prototypical common disease with complex genetic architecture.

    “making it a prototypical common disease with complex genetic architecture.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    Linkage studies strongly supported the notion that multiple sclerosis is not caused by a small number of mutations of large effect, but is likely due to many small risk effects spread across the genome.

    “Although largely negative, these studies strongly supported the notion that MS is not caused by a small amount of mutations of large effect, but is likely to be due to many small risk effects spread across the genome.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    The success of α4 integrin blockade by natalizumab, which prevents T cells from crossing the blood-brain barrier and forming new lesions, reinforces the view that multiple sclerosis risk is mediated by CD4+ T cells.

    “a view reinforced by the success of α4 integrin blockade by natalizumab, which prevents T cells from crossing the blood–brain barrier and forming new lesions.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    The limited resolution of fine mapping is likely a limitation in power, and larger sample sizes may help the resolution of these approaches.

    “This is likely a limitation in power, and larger sample sizes may help the resolution of these approaches.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationreasoning

    The landscape of antigen presentation is very dynamic in the population and risk-relevant phenotypes may be more complex than changes to diversity or binding strength of individual epitopes.

    “This suggests that the landscape of antigen presentation is very dynamic in the population and risk‐relevant phenotypes may be more complex than changes to diversity or binding strength of individual epitopes.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationnone given

    The clinical course of MS is unpredictable, and no tools for prognosis currently exist.

    “This clinical course is unpredictable, and no tools for prognosis currently exist.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given

    Patient response to therapy is largely unpredictable.

    “Similarly, patient response to therapy is largely unpredictable;”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    The G allele of rs744166 in STAT3 is associated with increased risk in MS and decreased risk in Crohn's disease.

    “One example is rs744166 (located in an intron of the STAT3 gene on chromosome 17): the G allele is associated with increased risk in MS38 and decreased risk in Crohn's disease.71”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [38] Jakkula E, Leppä V, Sulonen A‐M et al Genome‐wide association study in a high‐risk isolate for multiple sclerosis reveals associated variants in STAT3 gene. Am J Hum Genet 2010; 86: 285– 291. 20159113 PMC2820168 source
    • cites [71] Barrett JC, Hansoul S, Nicolae DL et al Genome‐wide association defines more than 30 distinct susceptibility loci for Crohn's disease. Nat Genet 2008; 40: 955– 962. 18587394 PMC2574810 source

  • May combine two claims.

    Premisecitation

    Autoimmune and inflammatory diseases as a group have proven remarkably tractable to genetic dissection in large cohorts, with several hundred risk loci now known in each disease.

    “As a group, the autoimmune and inflammatory diseases have proven remarkably tractable to genetic dissection in large cohorts, with several hundred risk loci now known in each disease.68”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [68] Zhernakova A, van Diemen CC, Wijmenga C. Detecting shared pathogenesis from the shared genetics of immune‐related diseases. Nat Rev Genet 2009; 10: 43– 55. 19092835 source

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given

    There is no evidence to date that different patients have slightly different pathologies and would thus respond to distinct modes of therapy targeting those specific pathways.

    “there is no evidence to date, for example, that different patients have slightly different pathologies and would thus respond to distinct modes of therapy targeting those specific pathways, although efforts to dissect this issue are underway.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisenone given

    Genome-wide association studies have proven remarkably successful in MS, with >200 risk loci now identified.

    “Genome‐wide association studies have proven remarkably successful in MS, with > 200 risk loci now identified.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Interpretationnone given

    The determinants of PPMS and the risk factors for conversion from RRMS to SPMS are not understood.

    “We still do not understand the determinants of either PPMS or the risk factors for conversion from RRMS to SPMS.”
    Quote in the paperSubstantial claim
36 of 36

  • The paper relies on this without support.

    The MHC region is complex, with multiple alleles under positive and balancing selection in different populations, complex long-range haplotype structures, and high sequence homology among many classical genes, making genotyping and sequencing technically challenging.

    “One of the main challenges to interrogating the MHC is the complexity of the region: multiple alleles in the region are under both positive and balancing natural selection in different populations, leading to complex long‐range haplotype structures, and many of the genes in the classical regions also show high‐sequence homology.”

  • The paper relies on this without support.

    There has been little evidence for causal roles for CNS-resident cells from GWAS analyses.

    “In contrast, there has been little evidence for causal roles for CNS‐resident cells from GWAS analyses.”

  • The paper relies on this without support.

    In most other common complex autoimmune and inflammatory diseases, gene regulation in target tissues is not a major feature of GWAS loci.

    “This is in common with most other common complex autoimmune and inflammatory diseases, where gene regulation in target tissues appears to not be a major feature of GWAS loci.”

  • The paper relies on this without support.

    Circulating immune cells are overrepresented in most available transcriptional, epigenetic and pathway data sets, and CNS is either totally absent or represented only as gross anatomical regions.

    “as circulating immune cells are overrepresented in most available transcriptional, epigenetic and pathway data sets, and CNS is either totally absent or represented only as gross anatomical regions”
4 of 4

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Two different papers have the same journal volume and starting page

    Minor

    References 45 and 63 assign different papers the same Nature Genetics volume and starting page, 45:1353, despite giving different publication years.

1 of 1

Cites

Flagged first, then by citations

1 of the 81 references couldn’t be matched to a known paper.

Cited by

Sorted by citations

Related

Authors
Cotsapas, C., & Mitrovic, M.
Affiliations
  • Broad Institute (US)
  • Departments of Neurology and Genetics Yale School of Medicine New Haven CT USA
Corresponding author
Cotsapas, Chris
ORCID
Cotsapas, Chris
Venue
Clinical & Translational Immunology
Publisher
Wiley
Published
2018-01
Volume
7
Issue
6
Pages
e1018-e1018
ISSN
2050-0068 (ISSN-L)
DOI
10.1002/cti2.1018
PMC
PMC5983059
OpenAlex
W2805130113
Cited by
89
References
81
Citations by year
2026: 5 · 2025: 10 · 2024: 11 · 2023: 7 · 2022: 19 · 2021: 16 · 2020: 14 · 2019: 4 · 2018: 3
FWCI
5.95
Type
journal article
Language
en
Update notices
correction (2018-08-16)
Open access
Licence
Locations
  • Clinical & Translational Immunology (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Europe PMC (PubMed Central) (submittedVersion, other-oa)
  • PubMed Central (submittedVersion, cc-by)
Topics
  • Health Sciences › Medicine › Pathology and Forensic Medicine › Multiple Sclerosis Research Studies
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Genetic Associations and Epidemiology
  • Life Sciences › Immunology and Microbiology › Immunology › Psoriasis: Treatment and Pathogenesis
Keywords
multiple sclerosis, genome-wide association study, genetic risk variants, etiology, genetic architecture, complex genetic disorders

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-13.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-13.

Cited sources are checked by gpt-6-sol.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review