CoChat
Sign Up
Cited by 56Genetic Associations and EpidemiologyCerebrospinal fluid and hydrocephalusAlzheimer's disease research and treatments

The genetic regulation of protein expression in cerebrospinal fluid

Open access
Hansson, Oskar, Kumar, Atul, Janelidze, Shorena, Stomrud, Erik, Insel, Philip S, Blennow, Kaj, Zetterberg, Henrik, Fauman, Eric +6 more·2023·EMBO Molecular Medicine
DOI 10.15252/emmm.202216359PMC PMC9832827
Open this paper in CoChat — free accountOpen PDF

Abstract

Studies of the genetic regulation of cerebrospinal fluid (CSF) proteins may reveal pathways for treatment of neurological diseases. 398 proteins in CSF were measured in 1,591 participants from the BioFINDER study. Protein quantitative trait loci (pQTL) were identified as associations between genetic variants and proteins, with 176 pQTLs for 145 CSF proteins ( P < 1.25 × 10 −10 , 117 cis ‐pQTLs and 59 trans ‐pQTLs). Ventricular volume (measured with brain magnetic resonance imaging) was a confounder for several pQTLs. pQTLs for CSF and plasma proteins were overall correlated, but CSF‐specific pQTLs were also observed. Mendelian randomization analyses suggested causal roles for several proteins, for example, ApoE, CD33, and GRN in Alzheimer's disease, MMP‐10 in preclinical Alzheimer's disease, SIGLEC9 in amyotrophic lateral sclerosis, and CD38, GPNMB, and ADAM15 in Parkinson's disease. CSF levels of GRN, MMP‐10, and GPNMB were altered in Alzheimer's disease, preclinical Alzheimer's disease, and Parkinson's disease, respectively. These findings point to pathways to be explored for novel therapies. The novel finding that ventricular volume confounded pQTLs has implications for design of future studies of the genetic regulation of the CSF proteome.

AI snapshot

Written from the paper’s full text.

TL;DR

Genome-wide analysis of 398 CSF proteins in 1,591 BioFINDER participants identified 176 pQTLs for 145 proteins and nominated proteins with shared genetic effects in Alzheimer’s disease, Parkinson’s disease, ALS, and other brain-related traits. CSF and plasma pQTLs were correlated but not identical, and ventricular volume was identified as a confounder for several CSF pQTLs.

Key findings

  • In CSF, 176 pQTLs were associated with 145 proteins at Bonferroni significance (P < 1.253 × 10−10), including 117 cis-pQTLs and 59 trans-pQTLs; plasma analyses found 197 pQTLs among 153 proteins.
  • CSF and plasma pQTLs were highly correlated overall (Spearman Rho = 0.69, P < 0.001), but 74 of 162 (46%) Bonferroni-significant CSF pQTLs were non-significant in plasma, indicating CSF-specific regulation.
  • Mendelian randomization found 18 pQTL–protein pairs with shared genetic effects across 13 outcomes, including APOE, CD33, and GRN in AD; MMP10 in preclinical AD; SIGLEC9 in ALS; and CD38, GPNMB, and ADAM15 in PD. Observational CSF levels were altered for GRN in AD, MMP-10 in preclinical AD, and CD38 and GPNMB in PD.
  • Ventricular volume was associated with 64 CSF proteins after FDR correction; adjusting for ventricular volume changed pQTL β-coefficients by up to 14%, and rs71635338 was associated with 1.51 SD larger ventricle volume (P = 0.007).

Methods

  • Prospective BioFINDER cohort study of 1,591 participants (53% female; mean age 71.3 years) with CSF and plasma proteomics, mostly proximity extension assays (PEA), plus orthogonal assays; 398 CSF proteins and 368 plasma proteins were analyzed.
  • Genome-wide pQTL analysis used 10.53 million common variants, linear regression adjusted for age, sex, diagnostic group, collection site, and sampling-to-assay time, with rank-based inverse-normal transformed residuals and Bonferroni correction (P < 1.253 × 10−10 for CSF).
  • Mendelian randomization used cis-pQTL instruments with external GWAS summary data for 32 outcomes, and colocalization was tested with the coloc package; ventricular volume analyses used MRI data from a subset of 720 participants.

Limitations

  • The paper states the cohort was elderly, limiting generalizability to younger populations; PEA antibody binding could be affected by missense variants, potentially producing erroneous pQTLs, especially for APOE, SIGLEC9, and FLRT2; and the study covered only ~400 proteins, about 2% of the proteome.
  • The authors also state that lack of post-mortem brain tissue limited protein quantification to CSF and plasma, that genetic association does not necessarily reveal functional consequence, and that whole-exome or whole-genome sequencing is recommended for future work.
  • Evident from the text: ventricular volume confounded several CSF pQTLs, and although most findings remained significant after adjustment, some were attenuated, so residual confounding or dilution effects may affect other CSF pQTL studies.

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

Integrity signals

Registry

Crossref · 10.15252/emmm.202216359

Cites

58 of 58 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

35 of 43 cited sources checked · 24 supported · 0 flagged · accuracy not yet measured

  • Findingown data

    Eight cis pQTL-protein pairs had a proxy match (up to 10 kb) with Yang et al (2021) pQTL-protein pairs.

    “Further, we identified 8 cis pQTL‐protein pairs with a proxy match (up to 10 kb) with Yang et al (2021) pQTL‐protein pairs (with the same protein and same directionality in the two studies).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Yang et al, 2021 source
  • Premisecitation

    Only a few genetic studies of the CSF proteome have been performed to date on larger sets of proteins.

    “only a few such studies have been performed to date on larger sets of proteins (Kunkle et al, 2019; Sasayama et al, 2019; Yang et al, 2021)”
    Quote in the paperSubstantial claim2 of 3 sources checked
    • cites Kunkle et al, 2019 source

      The Results section on candidate gene prioritization discusses Alzheimer’s disease risk loci and brain gene-expression data, not genetic studies of the CSF proteome. Searches for “CSF,” “cerebrospinal fluid proteins,” and larger protein sets found no passage establishing either that this paper is such a study or that only a few such studies existed. The cited paper therefore does not verify this claim in our copy. (LLM, conf=0.74)

      Experimental — accuracy not yet measured
    • cites Sasayama et al, 2019 source
    • cites Yang et al, 2021 source
  • Findingown data

    53 of the cis-pQTLs identified in this study showed previous evidence of association with gene expression in brain tissues.

    “When comparing with the eQTL meta‐analysis study, 53 of the cis‐pQTLs identified in the current study showed previous evidence of association with gene expression in brain tissues.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Sieberts et al, 2020 source
  • Findingcitation

    The overall genetic architecture of the CSF pQTLs was similar to findings reported for plasma pQTLs, gene expression QTLs, and a recent study on CSF pQTLs.

    “The overall genetic architecture of the CSF pQTLs was similar to findings reported for plasma pQTLs (Sun et al, 2018), gene expression QTLs (Stranger et al, 2012), and a recent study on CSF pQTLs (Yang et al, 2021).”
    Quote in the paperNot a substantial claim3 of 3 sources checked
    • cites Sun et al, 2018 source
    • cites Stranger et al, 2012 source
    • cites Yang et al, 2021 source
  • Findingcitation

    The causal association between CD33 (also known as Siglec-3) and AD was replicated between this study and Yang et al (2021).

    “Among our gene‐level findings, the causal association between CD33 (also known as Siglec‐3) and AD was replicated between the studies.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Yang et al, 2021 source
    • Rests on an assumption: The genetic variants used as instruments in Mendelian randomization are valid instruments: they are associated with the protein, not associated with confounders, and affect the outcome only through the protein.
  • May combine two claims. 47
  • Classified as probably not a substantial claim. 38

  • May combine two claims.

    Findingcitation

    APOE genotype was associated with lower Aβ42 and higher T-tau and P-tau in CSF, validating known associations with Alzheimer's disease biomarkers.

    “This genetic variant defines the APOE genotype, which has well‐known associations with CSF biomarkers related to Alzheimer's disease (AD), which were validated here (lower Aβ42, higher T‐tau and P‐tau (Chung et al, 2018)).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Chung et al, 2018 source

      The Results (§3.1) and Discussion (§4; page unavailable) report APOE ε4 associations with CSF Aβ42, total tau, and phosphorylated tau in people with mild cognitive impairment. They do not state the claimed directions—lower Aβ42 and higher tau—and the Discussion says APOE ε4 was not associated with CSF tau biomarkers in cognitively normal subjects. (LLM, conf=0.86)

      Experimental — accuracy not yet measured

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

    Study design and conductcitation

    CSF and plasma samples were collected and analyzed following a standardized protocol.

    “CSF and plasma samples were collected and analyzed following a standardized protocol (Palmqvist et al, 2017).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Palmqvist et al, 2017 source

      The Methods sections on CSF biomarkers and the BioFINDER study describe CSF sampling and specify the assays used to analyze CSF biomarkers (page numbers unavailable). They do not describe plasma sample collection or analysis, or establish that both sample types followed a standardized protocol. (LLM, conf=0.88)

      Experimental — accuracy not yet measured

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

    Premisecitation

    The specificity of aptamer-based multiplex methods has been discussed before.

    “The specificity of aptamer‐based multiplex methods has also been discussed before (Christiansson et al, 2014; Joshi & Mayr, 2018).”
    Quote in the paperNot a substantial claim2 of 2 sources checked
    • cites Christiansson et al, 2014 source
    • cites Joshi & Mayr, 2018 source

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

    Premisecitation

    Genetic variants in the GMNC-OSTN region on chromosome 3 were associated both with smaller ventricles and with increased CSF levels of the AD biomarker P-tau.

    “Recently, it was reported that genetic variants in the GMNC‐OSTN region on chromosome 3 (described in detail in this study) were associated both with smaller ventricles and with increased CSF levels of the AD biomarker P‐tau (Jansen et al, 2022).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Jansen et al, 2022 source

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

    Premisecitation

    Another recent CSF pQTL study also implicated CD33 for AD by MR analysis.

    “We note that another recent CSF pQTL study also implicated CD33 for AD by MR analysis (Yang et al, 2021), while several other proteins were uniquely implicated in this study versus in Yang et al (2021).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Yang et al, 2021 source

      The Results section (“Mendelian randomization and drug repositioning”; page unavailable) says MR implicated CD33 in AD, and the study did analyze CSF pQTLs. However, it identifies the CD33–AD result specifically as a plasma-protein finding, not a CSF CD33 finding. (LLM, conf=0.86)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    OSTN encodes osteocrin, a hormone that regulates dendritic growth in cerebral cortex, and has enhanced expression in the cerebral cortex.

    “OSTN encodes a hormone (osteocrin) that regulates dendritic growth in cerebral cortex (Ataman et al, 2016), with enhanced expression in the cerebral cortex (according to The Human Protein Atlas).”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites Ataman et al, 2016 source
    • cites The Human Protein Atlas

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

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

    Study design and conductcitation

    The use of covariate adjustment in the pQTL analyses follows standard procedures.

    “This use of covariate adjustment follows standard procedures (Folkersen et al, 2020) in pQTL analyses.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Folkersen et al, 2020 source

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

    Study design and conductcitation

    Colocalization analysis used a strict Bayesian model with the coloc package in R.

    “To estimate the prior likelihood of each genomic locus having a single variant affecting both the protein and the phenotype, we used a strict Bayesian model (colocalization analysis) using the coloc package in R (Giambartolomei et al, 2014).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Giambartolomei et al, 2014 source

  • May combine two claims.

    Premisecitation

    Yang et al (2021) described 226 CSF cis-pQTLs and 48 trans-pQTLs.

    “Yang et al (2021) described 226 CSF cis‐pQTLs (at P < 5 × 10−8) and 48 trans‐pQTLs (at Bonferroni corrected P) compared to 158 CSF cis‐pQTLs (at P < 5 × 10−8) and 59 CSF trans‐pQTLs (at Bonferroni corrected P) in our study.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Yang et al, 2021 source

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

    Premisecitation

    Several members of the SIGLEC protein family have been implicated for roles in neurodegenerative diseases, especially due to neuroinflammatory properties.

    “Regarding the MR findings of CD33 (SIGLEC‐3) in AD and SIGLEC‐9 in ALS, we note that several members of the SIGLEC protein family have been implicated for roles in neurodegenerative diseases, especially due to neuroinflammatory properties (Siddiqui et al, 2019).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Siddiqui et al, 2019 source

  • May combine two claims.

    Findingown data

    Two pQTLs were common in both studies but associated with different proteins: rs76904798 was trans-acting and associated with GPNMB in this study and GRN in Yang et al (2021); rs12126142 was trans-acting in this study (PSME1) but cis-acting in Yang et al (2021) (IL6R).

    “We also identified two additional pQTLs that were common in both studies but associated with different proteins. One of these (rs76904798) was trans‐acting and was associated with GPNMB in our study and GRN in Yang et al (2021). The second pQTL (rs12126142) was trans‐acting in our study (PSME1) but cis‐acting in the Yang et al (2021) study (IL6R).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Yang et al, 2021 source

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

      Experimental — accuracy not yet measured

  • Classified as probably not a substantial claim.

    Premisecitation

    Previous studies have demonstrated associations between individual CSF biomarkers (Aβ, T-tau, and P-tau) and ventricle volume, likely dilution effects.

    “Previous studies have demonstrated associations between individual CSF biomarkers (for the AD biomarkers Aβ, T‐tau, and P‐tau) and ventricle volume (likely dilution effects; Edsbagge et al, 2017; van Waalwijk van Doorn et al, 2017), but to our knowledge, this is the first large‐scale study of such associations, and how they affect associations between CSF biomarker and genetic variants.”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites Edsbagge et al, 2017 source

      The Abstract and Aβ-related markers section report negative associations between ventricular CSF volume and Aβ38 and Aβ40. But the Discussion says T-tau and P-tau were not associated with measured CSF volumes and explicitly rejects mere dilution as the explanation for the Aβ associations. (LLM, conf=0.88)

      Experimental — accuracy not yet measured
    • cites van Waalwijk van Doorn et al, 2017 source

      The retrieved copy provides bibliographic details and keywords (no section or page), but no findings about associations of Aβ, T-tau, or P-tau with ventricular volume. Searches for those associations and for dilution effects returned no passage that addresses the claim. (LLM, conf=0.70)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation

    Proximity extension assays have been shown to be sensitive and specific.

    “We mainly used proximity extension assays (PEA), which have been shown to be sensitive and specific (Assarsson et al, 2014; Katz et al, 2022)”
    Quote in the paperSubstantial claim2 of 2 sources checked
    • cites Assarsson et al, 2014 source
    • cites Katz et al, 2022 source

      The Discussion (page unavailable) reports evidence supporting the specificity of Olink’s proximity-extension-based measurements, including cis-pQTL associations and agreement with ELISA for four proteins. It does not establish generally high sensitivity: the authors say they could not independently determine detection limits, although the reagents in their ELISA validation quantified at least as low as the levels measured by ELISA. (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

    GPNMB is a transmembrane glycoprotein that has been studied before in cancer and inflammation.

    “GPNMB (Glycoprotein nonmetastatic melanoma protein B) is a transmembrane glycoprotein, which has been studied before in cancer and inflammation (Saade et al, 2021).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Saade et al, 2021 source

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

    Premisecitation

    Several studies in blood plasma have shown that pQTLs using aptamer-based and antibody-based assays are common and can explain up to 30% of protein variance.

    “several studies in blood plasma have consistently shown that genetic variants associated with protein levels, known as protein quantitative trait loci (pQTL), using both aptamer‐based (Sun et al, 2018; Ferkingstad et al, 2021) and antibody‐based assays (Folkersen et al, 2020) are common and can explain up to 30% of the protein variance.”
    Quote in the paperNot a substantial claim2 of 3 sources checked
    • cites Sun et al, 2018 source

      The Methods (Protein measurements) describe the aptamer-based SOMAscan assay, and Genetic architecture of the plasma proteome reports 1,927 pQTLs, supporting the claim that such associations are common in plasma. That section reports a median of 5.8% of protein-level variation explained and more than 20% for 193 proteins, but does not establish the cited figure of up to 30%. The antibody-based studies are attributed to other citations in the sentence, not to Sun et al. (LLM, conf=0.83)

      Experimental — accuracy not yet measured
    • cites Ferkingstad et al, 2021 source
    • cites Folkersen et al, 2020 source

      The Results section on genome-wide meta-analysis reports pQTLs for 85 of 90 plasma proteins, and Online Methods describes the antibody-based Olink assay. The Results section on heritability reports substantial variation explained by genetic scores, but our copy does not substantiate the specific claim that pQTLs explain up to 30% of protein variance; it gives different summary figures. (LLM, conf=0.86)

      Experimental — accuracy not yet measured

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

    Premisecitation

    FLRT2 has not been described extensively in relation to epilepsy or seizures, but there is evidence that this cell adhesion molecule is involved in neuronal development, including development of inhibitory cortical circuits.

    “To our knowledge, FLRT2 (fibronectin leucine‐rich transmembrane protein 2) has not been described extensively in relation to epilepsy or seizures, but there is evidence that this cell adhesion molecule is involved in neuronal development, including development of inhibitory cortical circuits (Fleitas et al, 2021).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Fleitas et al, 2021 source

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

    Premisecitation

    GPNMB's potential role in PD pathogenesis and links to aggregation of α-synuclein has recently been highlighted.

    “Its potential role in PD pathogenesis and links to aggregation of α‐synuclein has recently been highlighted (Diaz‐Ortiz et al, 2022).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Diaz‐Ortiz et al, 2022 source

  • May combine two claims.

    Premisecitation

    Katz et al (2022) found evidence supporting more reliable protein target specificity and a higher number of phenotypic associations for the PEA assays used in the current study.

    “Importantly, Katz et al (2022) found evidence supporting more reliable protein target specificity and a higher number of phenotypic associations for the PEA assays, used in the current study.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Katz et al, 2022 source

  • May combine two claims.

    Findingown data

    Fifteen pQTL-protein pairs were replicated between this study and Yang et al (2021) with significant effects and similar direction; 12 were cis-acting and 3 were trans-acting.

    “Fifteen pQTL‐protein pairs were replicated (significant effects with similar direction of effect with respect to the risk allele) between the studies (see Fig EV4; Dataset EV7). Out of these 15, 12 pQTL‐protein pairs were cis‐acting and three were trans‐acting.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Yang et al, 2021 source

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

      Experimental — accuracy not yet measured

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

    Premisecitation

    Previous studies found increased levels of CSF MMP-10 in AD, as well as increased rates of disease progression in those with high CSF MMP-10 levels at baseline.

    “Regarding the role of MMP‐10 (matrix metalloproteinase 10) in preclinical AD, we note that previous studies found increased levels of CSF MMP‐10 in AD, as well as increased rates of disease progression in those with high CSF MMP‐10 levels at baseline (Martino Adami et al, 2022).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Martino Adami et al, 2022 source

  • May combine two claims.

    Findingown data

    CTSF was associated with one pQTL (rs572846) and had shared effects with bipolar disorder and depression.

    “CTSF was associated with one pQTL (rs572846) and had shared effects with bipolar disorder and depression.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The genetic variants used as instruments in Mendelian randomization are valid instruments: they are associated with the protein, not associated with confounders, and affect the outcome only through the protein.

  • May combine two claims.

    Findingown data

    The APOE pQTL (rs429358) had a colocalization posterior probability of 1 with AD, 0.81 with preclinical AD, 0.99 with amygdala volume, 0.87 with hippocampus volume and 0.95 with nucleus accumbens volume.

    “The APOE pQTL (rs429358) had a colocalization posterior probability of 1 with AD, 0.81 with preclinical AD, 0.99 with amygdala volume, 0.87 with hippocampus volume and 0.95 with nucleus accumbens volume.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    When adjusting for ventricle volume and intracranial volume, pQTL β-coefficients changed by up to 14%, generally reduced, while P-values were generally increased.

    “When adjusting for ventricle volume (and ICV), the β‐coefficients were changed by up to 14% (β‐coefficients were generally reduced, while P‐values were generally increased) (Fig EV5).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The pQTL-protein pairs CTSS-rs79671334 and RARRES2-rs3735172 had shared effects with schizophrenia and year of schooling, respectively.

    “The pQTL‐protein pairs CTSS‐rs79671334 and RARRES2‐rs3735172 had shared effects with schizophrenia and year of schooling, respectively.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The genetic variants used as instruments in Mendelian randomization are valid instruments: they are associated with the protein, not associated with confounders, and affect the outcome only through the protein.

  • May combine two claims.

    Findingown data

    In a comparison between CSF pQTLs and brain eQTLs, there were both concordant associations, supporting direct links between gene transcription and translation, as well as discordant associations.

    “In a comparison between CSF pQTLs and brain eQTLs, we noted that there were both concordant associations, supporting direct links between gene transcription and translation, as well as discordant associations.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Concordant QTLs tended to have higher co-localization probability (18/35 had co-localization probability of at least 0.85), while discordant QTLs tended to have lower co-localization probability (only 4/18 had co-localization probability of at least 0.85).

    “Co‐localization experiments showed that concordant QTLs tended to have higher co‐localization probability (18/35 had co‐localization probability of at least 0.85), while discordant QTLs tended to have lower co‐localization probability (only 4/18 had co‐localization probability of at least 0.85).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Most CSF proteins had only one locus, predominantly acting in cis.

    “Most CSF proteins had only one locus, predominantly acting in cis (Fig 4A, see Appendix Fig S3 for plasma pQTLs)”
    Quote in the paperSubstantial claim

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

    Findingown data

    Although the study studied approximately 400 proteins, this is only approximately 2% of the proteome.

    “Another limitation is that although we studied ~ 400 proteins, this is only ~ 2% of the proteome.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Twenty out of 29 unique reported traits were brain-related, mainly comprising volumetric brain measures, including ventricle volume or white matter lesions.

    “Twenty out of 29 unique reported traits were brain‐related, mainly comprising volumetric brain measures, including of ventricle volume or white matter lesions, supporting the relevance of this genomic region for CNS‐related molecular pathways.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    There were no plasma pQTLs in the GMNC-OSTN region, emphasizing its specificity for the CSF proteome in this study.

    “There were no plasma pQTLs in this region, emphasizing its specificity for the CSF proteome in this study.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    No altered levels were found for CD33 or APOE for AD, or ADAM15 for PD.

    “We did not find altered levels for CD33 or APOE (for AD) or ADAM15 (for PD).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    CD33 pQTL (rs273634) and GRN pQTL (rs5848) had a posterior probability of 0.78 and 0.99, respectively, with AD.

    “CD33 pQTL (rs273634) and GRN pQTL (rs5848) had a posterior probability of 0.78 and 0.99, respectively, with AD.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Significant observational differences between diagnostic groups were found for GRN in AD, MMP-10 in preclinical AD, and CD38 and GPNMB in PD, supporting the MR findings.

    “We found significant effects for GRN in AD, for MMP‐10 in preclinical AD, and for CD38 and GPNMB in PD, supporting the MR findings (Table EV2).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Our study described 158 CSF cis-pQTLs and 59 CSF trans-pQTLs.

    “Yang et al (2021) described 226 CSF cis‐pQTLs (at P < 5 × 10−8) and 48 trans‐pQTLs (at Bonferroni corrected P) compared to 158 CSF cis‐pQTLs (at P < 5 × 10−8) and 59 CSF trans‐pQTLs (at Bonferroni corrected P) in our study.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The GTEx eQTL database query identified 7 additional cis-pQTLs, with concordant direction for 5 and discordant for 2.

    “The GTEx eQTL database query resulted in identification of 7 additional cis‐pQTLs (not found in the meta‐analysis study), where the effect direction was concordant for 5 and discordant for 2 pQTLs (Appendix Fig S4 and Dataset EV9).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Functional annotation of CSF variants found enrichment for intronic, exonic, and downstream variants, while intergenic and ncRNA-intronic variants were significantly less common.

    “We performed functional annotation of the CSF variants (Fig 3C) and found enrichment (after Bonferroni correction over 11 categories) for intronic (49% for CSF variants vs. 36% in a reference panel), exonic (1.8% vs. 1.0%), and downstream (1.7% vs. 1.1%) variants, while intergenic (35% vs. 47%) and ncRNA‐intronic (8.9% vs. 11.5%) were significantly less common.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The effect direction was concordant for 35 and discordant for 18 of the pQTLs.

    “The effect direction was concordant for 35 and discordant for 18 of the pQTLs (Fig 5; Dataset EV8).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The seven trans-pQTLs with interaction score > 0.7 included CSF Aβ42, P-tau, and T-tau (for APOE), CCL4 (for CCR2 and CCR9), IL12A (for IL12B), and TNFSF13B (for TNFRSF13C).

    “These included CSF Aβ42, P‐tau, and T‐tau (for APOE), CCL4 (for CCR2 and CCR9), IL12A (for IL12B), and TNFSF13B (for TNFRSF13C; Appendix Fig S5).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Genes for proteins that were associated with ventricle volume showed strong enrichment for the cerebral cortex, while the remaining proteins did not show enrichment for any brain region.

    “Genes for proteins that were associated with ventricle volume showed strong enrichment for the cerebral cortex (Fig 6B), while the remaining proteins did not show enrichment for any brain region (Fig 6C).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    A few CSF proteins had several trans-acting loci, including six loci for CCL4, four for RBKS, and three each for CCL24, CTSS, and LRPAP1.

    “a few proteins had several trans‐acting loci (for all identified loci, the strongest pQTL was nominated and carried on further in all analyses), including six loci for CSF CCL4 (two cis‐pQTLs and four trans‐pQTLs), four for CSF RBKS (all cis‐pQTLs), and three each for CSF CCL24 (all cis‐pQTLs), CTSS (all cis‐pQTLs), and LRPAP1 (all trans‐pQTLs).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    Among all significant CSF pQTLs, there were 140 unique genetic loci and 145 unique proteins.

    “Among all significant (after Bonferroni correction) CSF pQTLs, there were 140 unique genetic loci and 145 unique proteins.”
    Quote in the paperNot a substantial claim

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

    Study design and conductnone given

    There was no significant variance in findings related to assay batches.

    “There was no significant variance in findings related to assay batches.”
    Quote in the paperNot a substantial claim

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

    Study design and conductown data

    Subject-level QC was conducted, including removal of sexual incompatibility between chip-inferred sex and self-reported sex, low call rates, and extreme heterozygosity.

    “Subject‐level QC was conducted, including the removal of sexual incompatibility between chip‐inferred sex and self‐reported sex, low call rates (5% cut‐off), and extreme heterozygosity.”
    Quote in the paperSubstantial claim

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

    Study design and conductnone given

    The laboratory staff analyzing samples were blinded to all data.

    “The laboratory staff analyzing samples were blinded to all data.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data

    18F-florbetapir PET SUVRs were regressed on each genetic variant adjusting for age, sex, and the first five principal components of genetic background to account for population stratification.

    “18F‐florbetapir PET SUVRs were regressed on each genetic variant separately, adjusting for age, sex, and the first five principal components of the genetic background to account for unmeasured population stratification.”
    Quote in the paperSubstantial claim

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

    Study design and conductreasoning

    The MR analysis may be affected by pseudo pQTLs or pQTLs driven by epitope effects, especially for APOE, SIGLEC9, and FLRT2, which were missense variants.

    “We cannot exclude that the MR analysis may be affected by “pseudo pQTLs” or pQTLs driven by epitope effects, i.e., that the assay detects protein with missense variant differently (especially for APOE, SIGLEC9, and FLRT2, which were missense variants).”
    Quote in the paperNot a substantial claim

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

    Study design and conductreasoning

    The genetic association to a protein does not necessarily give information about the actual functional consequence of a genetic variant.

    “Furthermore, the genetic association to a protein does not necessarily give information about the actual functional consequence of a genetic variant (i.e., the variant might alter total expression, or expression of one particular isoform of the protein that is detected, but not others).”
    Quote in the paperNot a substantial claim

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

    Study design and conductreasoning

    The rs71635338-ventricle volume test was a single test based on an a priori hypothesis, so correction for multiple comparisons was not applicable.

    “Note that this was a single test, based on an a priori hypothesis, whereby correction for multiple comparisons was not applicable.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data

    Although some pQTLs can be robustly replicated across distinct proteomics methods, aptamer- and antibody-based technologies largely provide non-overlapping information.

    “This suggests that although some pQTLs can be robustly replicated across distinct proteomics methods, aptamer‐ and antibody‐based technologies largely provide non‐overlapping information.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    CSF is produced within the brain and is a rich source for biomarkers that correlate with brain pathologies across different diseases.

    “Cerebrospinal fluid (CSF) is produced within the brain and is a rich source for biomarkers that correlate with brain pathologies across different diseases (Hansson, 2021).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Hansson, 2021 source

  • Classified as probably not a substantial claim.

    Study design and conductcitation

    The genome-wide association analysis to identify pQTLs adjusted for genetic principal components to adjust for population stratification.

    “The genome‐wide association analysis to identify pQTLs were performed using PLINK1.9 with adjustment for genetic principal components to adjust for population stratification.”
    Quote in the paperSubstantial claim

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

    Premisecitation

    Discordant associations between pQTLs and eQTLs have been reported to varying degrees before.

    “Discordant associations between pQTLs and eQTLs have been reported to varying degrees before (preprint: Sun et al, 2022) and may hypothetically reflect biological pathways with negative feedback loops that regulate protein levels.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites preprint: Sun et al, 2022

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Study design and conductown data

    All proteins passed quality control in all measured samples, except for BDNF and CCL22, which were removed from all analyses.

    “All proteins passed quality control (QC) (with warnings indicated in < 20% of samples) in all measured samples, except for BDNF and CCL22, which were therefore removed from all analyses (note that these proteins were later highlighted by Olink Proteomics as having technical issues and were replaced with other proteins in subsequent versions of the assays).”
    Quote in the paperSubstantial claim

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

    Study design and conductnone given

    Standard QC steps were performed for SNP-level filter to ensure conformity with the reference panel used for imputation.

    “Standard QC steps were performed for SNP‐level filter to ensure conformity with the reference panel used for imputation (strand continuity, names of the alleles, position and assignments for Ref/Alt).”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data

    The pQTL analysis adjusted for age, gender, diagnostic group, collection site, and time from sampling to assay execution.

    “For each protein, a linear regression analysis was performed with adjustments for age, gender, diagnostic group, collection site, and time from sampling to assay execution.”
    Quote in the paperSubstantial claim

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

    Study design and conductnone given

    Multi-allelic variants and SNPs with a data imputation score < 0.2 were excluded, genotype calls with a posterior likelihood < 0.9 were set to fail, SNPs with a genotyping rate > 0.9 were retained, and SNPs with MAF ≥ 1% were used for the analyses.

    “Multi‐allelic variants and SNPs with a data imputation score < 0.2 were excluded as part of post‐imputation QC and genotype calls with a posterior likelihood < 0.9 were set to fail (i.e., hard‐called). SNPs with a genotyping rate > 0.9 were retained. SNPs with MAF ≥ 1% were used for the analyses.”
    Quote in the paperSubstantial claim

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

    Study design and conductnone given

    All participants provided informed consent and the study conformed to the principles set out in the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report.

    “All participants provided informed consent and the study conformed to the principles set out in the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report.”
    Quote in the paperNot a substantial claim

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

    Premisecitation

    RARRES2 has chemotactic properties during inflammatory responses, and serum RARRES2 levels were associated with severity of seizures in children with idiopathic epilepsy.

    “RARRES2 (retinoic acid receptor responder protein 2, also known as chemerin) is described as having chemotactic properties during inflammatory responses and serum RARRES2 levels were associated with severity of seizures in children with idiopathic epilepsy (Elhady et al, 2018).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Elhady et al, 2018 source

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    The genome-wide analyses were adjusted for diagnostic group.

    “The genome‐wide analyses were adjusted for diagnostic group”
    Quote in the paperSubstantial claim

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

    Premisecitation

    CTSF is a lysosomal protein involved in the pathogenesis of some types of neuronal ceroid lipofuscinosis.

    “CTSF (cathepsin F), implicated here for depression and bipolar disease, is a lysosomal protein which is involved in the pathogensis of some types of neuronal ceroid lipofuscinosis (Smith et al, 2013).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Smith et al, 2013 source

  • May combine two claims.

    Interpretationown data

    Ventricle volume is a possible confounder for many CSF pQTL studies, although most findings remain significant (but sometimes attenuated) after adjustment for ventricle volume.

    “The results point to ventricle volume as a possible confounder for many CSF pQTL studies, although we demonstrate that most findings remain significant (but sometimes attenuated) after adjustment for ventricle volume.”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The MRI subset is representative of the full cohort for analyses of ventricle volume and CSF pQTL confounding.

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    Assay batches were randomized with respect to covariates.

    “Assay batches were randomized with respect to covariates.”
    Quote in the paperSubstantial claim

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

    Study design and conductnone given

    The BioFINDER study was approved by the regional Ethical Review Board in Lund.

    “The BioFINDER study was approved by the regional Ethical Review Board in Lund (2008/289, 2008/695, 2010/156, 2010/401, 2013/494).”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    Prior to downstream analyses, the study checked and verified that there was no allelic heterogeneity or locus heterogeneity in the available GWAS summary statistics files.

    “Prior to downstream analyses, we checked and verified that there was no allelic heterogeneity or locus heterogeneity in the available GWAS summary statistics files.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisenone given

    Identification of pQTLs makes it possible to test if proteins are likely to be causal for development of human diseases.

    “Identification of pQTLs makes it possible to test if proteins are likely to be causal for development of human diseases (hence nominating them as candidate drug targets).”
    Quote in the paperSubstantial claim

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

    Study design and conductown data

    Genetic data from the A4 study underwent quality control limiting to variants with call rate > 95%, minor allele frequency > 1%, and Hardy-Weinberg equilibrium P > 10−6, and excluding cryptic relatedness.

    “Genetic data from A4 underwent a quality control process limiting to variants with a call rate > 95%, minor allele frequency > 1%, if cryptic relatedness was identified, or if out of Hardy–Weinberg equilibrium (P > 10−6).”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductown data

    Relatedness among the samples was eliminated by removing one participant from each pair of close relatives (first or second degree).

    “Relatedness among the samples was eliminated by removing one participant from each pair of close relatives (first or second degree) identified as π^ ≥ 0.0625.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductnone given

    Multi-dimensional scaling was done using PLINK2 to create principal components in genetic analyses to account for ancestry.

    “Multi‐dimensional scaling was done using PLINK2 to create principal components in genetic analyses to account for ancestry.”
    Quote in the paperSubstantial claim

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

    Premisecitation

    eQTL and disease-associated loci may have fundamental differences, contributing to different results with these approaches.

    “As recently pointed out, eQTL and disease‐associated loci may have fundamental differences, contributing to different results with these approaches (preprint: Mostafavi et al, 2022).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites preprint: Mostafavi et al, 2022

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Study design and conductnone given

    The non-Olink proteins all passed thresholds of being quantifiable and passing QC in at least 80% of measured samples.

    “The non‐Olink proteins all passed thresholds of being quantifiable and passing QC in at least 80% of measured samples.”
    Quote in the paperNot a substantial claim
70 of 70

  • The paper relies on this without support.

    Ventricular volume had not been evaluated as a confounder for CSF pQTLs before this study.

    “We also evaluated ventricular volume measured with magnetic resonance imaging (MRI) as a potential confounder (due to dilution effects) for CSF pQTLs, which to our knowledge had not been done before in CSF pQTL studies.”

  • The paper relies on this without support.

    The genetic variants used as instruments in Mendelian randomization are valid instruments: they are associated with the protein, not associated with confounders, and affect the outcome only through the protein.

    “For the MR, we performed Wald ratio (WR) tests based on the independent cis‐pQTL instruments available for each protein.”

  • The paper relies on this without support.

    CSF protein levels reflect proteins expressed in the brain and subsequently secreted to CSF.

    “The overarching aim of the study was to map and decipher clinically relevant genetic regulation of proteins expressed in the brain and subsequently secreted to the CSF, with a focus on brain diseases in general and neurological diseases in particular.”

  • The paper relies on this without support.

    GMNC has only weak expression in brain.

    “GMNC (geminin coiled‐coil domain‐containing protein‐1) is a regulator of DNA replication, with only weak expression in brain.”

  • The paper relies on this without support.

    The MRI subset is representative of the full cohort for analyses of ventricle volume and CSF pQTL confounding.

    “Note that these analyses were done on a subset of participants with MRI data.”

  • The paper relies on this without support.

    Many brain diseases lack efficient disease-modifying treatments.

    “Many brain diseases lack efficient disease‐modifying treatments.”

  • The paper relies on this without support.

    Studies of the genetic regulation of CSF proteins can give clues to novel targetable biological pathways of brain diseases.

    “Studies of the genetic regulation of cerebrospinal fluid (CSF) proteins can give clues to novel targetable biological pathways of brain diseases.”

  • The paper relies on this without support.

    Only a few studies have combined genotyping data and CSF data on large sets of proteins, which is needed for this type of analysis.

    “However, only a few studies have combined genotyping data and CSF data on large sets of proteins, which is needed for this type of analysis.”

  • The paper relies on this without support.

    There is a lack of studies using sensitive and specific antibody-based methods, such as proximity extension assays, for a large number of proteins.

    “There is especially a lack of studies using sensitive and specific antibody‐based methods, such as proximity extension assays, for a large number of proteins.”

  • The paper relies on this without support.

    Most published pQTLs represent protein abundance from plasma or serum samples, and only a small number represent prior CSF pQTL studies.

    “While most published pQTLs represent protein abundance from plasma or serum samples, a small number represent prior CSF pQTL studies.”

  • The authors state this as an assumption.

    SUVRs have a linear association with allele frequency.

    “SUVRs were modeled using ordinary least squares regression assuming a linear association with allele frequency.”
11 of 11

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

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.

  • CSF-specific pQTL subtypes exceed the reported total

    Major

    The reported 44 cis-pQTLs and 43 trans-pQTLs sum to 87, not 74. This changes the stated count of CSF pQTLs without a genome-wide-significant plasma association.

    Reported 74 · recomputed 87

  • Discussion names GMBP instead of GPNMB for Parkinson's disease

    Minor

    The discussion describes an observationally validated PD protein as GMBP, while the results identify that protein as GPNMB.

  • Drug-candidate table names CTCF instead of the MR protein CTSF

    Minor

    Table 2 lists CTCF for depression and bipolar disorder, but Table 1 identifies CTSF as the protein associated with both outcomes.

3 of 3

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Hansson, O., Kumar, A., Janelidze, S., Stomrud, E., Insel, P. S., Blennow, K., Zetterberg, H., Fauman, E., Hedman, Å. K., Nagle, M. W., Whelan, C. D., Baird, D., Mälarstig, A., & Mattsson‐Carlgren, N.
Affiliations
  • Lund University (SE)
  • University of California, San Francisco (US)
  • Sahlgrenska University Hospital (SE)
  • Pfizer (United States) (US)
  • Karolinska Institutet (SE)
  • Eisai (United States) (US)
  • Biogen (United States) (US)
Corresponding author
Hansson, Oskar
ORCID
Venue
EMBO Molecular Medicine
Publisher
Springer Science and Business Media LLC
Published
2022-12-12
Volume
15
Issue
1
Pages
e16359-e16359
ISSN
  • 1757-4676 (ISSN-L)
  • 1757-4684
DOI
10.15252/emmm.202216359
PMC
PMC9832827
OpenAlex
W4311205850
Cited by
56
References
58
Citations by year
2026: 13 · 2025: 16 · 2024: 20 · 2023: 6 · 2022: 1
FWCI
6.49
Type
journal article
Language
en
Open access
Licence
Locations
  • EMBO Molecular Medicine (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion)
  • UCL Discovery (University College London) (submittedVersion)
  • Lund University Publications (Lund University) (submittedVersion)
Funders
  • Alzheimer Drug Discovery Foundation (RDAPB‐201809‐2016615)
  • Erling‐Persson Family Foundation
  • European Research Council (681712)
  • European Union Joint Program for Neurodegenerative Disorders (JPND2021‐00694)
  • European Union's Horizon 2020 research and innovation programme under the Marie Skłodowska‐Curie grant agreement (860197)
  • Hjärnfonden (FO2017‐0243, FO2019‐0228)
  • Knut and Alice Wallenberg Foundation (2017‐0383, WCMM 2019)
  • Olav Thon Foundation
  • Region Skåne (WCMM 2019)
  • Regionalt Forskningsstöd (2020‐0314)
  • Stiftelsen för Gamla Tjänarinnor
  • Stiftelsen Gamla Tjänarinnor (2019‐00845)
  • Strategic Research Area MultiPark
  • Swedish Alzheimer Foundation (AF‐939932, AF‐930655, AF‐968453, AF‐742881)
  • Swedish Research Council (2018‐02532)
  • Swedish State Support for Clinical Research (ALFGBG‐720931)
  • Swedish state under the agreement between the Swedish government and the County Councils, the ALF‐agreement (ALFGBG‐715986)
  • the AD Strategic Fund and the Alzheimer's Association (ADSF‐21‐831376‐C)
  • the Alzheimer Drug Discovery Foundation (201809‐2016862)
  • the Alzheimer's Association 2021 Zenith Award (ZEN‐21‐848495)
  • The Bundy Academy
  • the European Union Joint Program for Neurodegenerative Disorders (JPND2019‐466‐236)
  • the Konung Gustaf V:s och Drottning Victorias Frimurarestiftelse
  • the Marianne and Marcus Wallenberg foundation (2015.0125)
  • The Medical Faculty at Lund University (WCMM 2019)
  • the National Institute of Health (1R01AG068398‐01)
  • The Parkinson foundation of Sweden (1280/20)
  • the Skåne University Hospital Foundation (2020‐O000028)
  • the Swedish Brain Foundation (FO2021‐0293, FO2019‐0029, FO2020‐0275)
  • the Swedish federal government under the ALF agreement (2018‐Projekt0279, 2018‐Projekt0054)
  • UK Dementia Research Institute at UCL
  • Vetenskapsrådet (2021‐02219, 2016‐00906, 2017‐00915)
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Genetic Associations and Epidemiology
  • Life Sciences › Neuroscience › Cellular and Molecular Neuroscience › Cerebrospinal fluid and hydrocephalus
  • Health Sciences › Medicine › Physiology › Alzheimer's disease research and treatments
Keywords
cerebrospinal fluid protein, protein quantitative trait loci, Alzheimer's disease, preclinical Alzheimer's disease, amyotrophic lateral sclerosis, Parkinson's disease, apolipoprotein E, CD33, gene regulatory networks, MMP-10, CD38, GPNMB, ADAM15, CSF proteome, plasma proteins, Mendelian randomization
MeSH terms
Alzheimer Disease, Humans, Membrane Glycoproteins, Membrane Proteins, Parkinson Disease, Biomarkers, Antigens, CD, Quantitative Trait Loci, Proteomics, ADAM Proteins, Matrix Metalloproteinase 10, Sialic Acid Binding Immunoglobulin-like Lectins

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 2026-11-01.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-11-01.

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