Largest CSF biomarker GWAS resource
The analysis combines 18,948 participants from 30 studies, substantially extending the sample sizes described for earlier CSF Aβ42, t-tau, and p-tau181 GWAS.
↳ Introduction; Methods, “Study design”; Table 1
Abstract
Cerebrospinal fluid amyloid beta 42, total tau, and phosphorylated tau 181 are well accepted markers of Alzheimer's disease. These biomarkers better reflect disease pathogenesis compared to clinical diagnosis. Here, we perform a genome wide association study meta-analysis including 18,948 individuals of European ancestry and identify 12 genome-wide significant loci across all three biomarkers, eight of them novel. We replicate the association of biomarkers with APOE, CR1, GMNC/CCDC50 and C16orf95/MAP1LC3B. Novel loci include BIN1 for amyloid beta and GNA12, MS4A6A, SLCO1A2 with both total tau and phosphorylated tau 181, as well as additional loci on chr. 8, near ANGPT1 and chr. 9 near SMARCA2. We also demonstrate that these variants have significant association with Alzheimer's disease risk, disease progression and/or brain amyloidosis. The associated genes are implicated in lipid metabolism independent of APOE, coupled with autophagy and brain volume regulation driven by total tau and phosphorylated tau 181 dysregulation.
Key Claims
Primary findingTentative
The associated genes are implicated in lipid metabolism independent of APOE, coupled with autophagy and brain volume regulation driven by total tau and phosphorylated tau 181 dysregulation.
mechanistic interpretation relies on exploratory enrichment, gene nomination, and non-significant brain-volume covariance
Abstract
SecondaryTentative
higher levels of Aβ42 are causally associated with a reduced risk of Alzheimer’s disease
causal framing lacks reported sensitivity analyses addressing instrument dominance and possible sample overlap
Results, “CSF Aβ42 is causally associated with increased risk of AD”
SecondaryTentative
We also demonstrate that these variants have significant association with Alzheimer’s disease risk, disease progression and/or brain amyloidosis.
several lookups are nominal and some lead variants show no association with the evaluated phenotypes
Abstract
Derived from the full evaluation — not a separate score.
Strengths · Limitations
Strengths
Comprehensive primary GWAS quality controls
The study applies cohort-level biomarker normalization, genotype and imputation QC, relatedness and ancestry filtering, inverse-variance meta-analysis, LDSC checks, and per-locus heterogeneity reporting.
↳ Methods, “CSF measurement and quality control” through “Meta-analysis and conditional analysis”; Table 2
Strong cumulative research trajectory
The study benchmarks against earlier CSF GWAS, replicates established signals, applies MTAG and pleioFDR, and makes phenotype-specific summary statistics available for follow-up.
↳ Introduction; Results, “Shared genetic influences across traits”; Figure 2 caption
Limitations
Causal and mechanistic claims overreach
The MR analysis is presented causally without reported leave-one-out or APOE-excluded sensitivity analyses, while pathway and brain-volume interpretations are stronger than the nominal or non-significant evidence supports.
↳ Results, MR, genetic covariance, and pathway sections; Figures 3C and 4; Discussion
Central result labels are inconsistent
The Abstract’s novelty count and BIN1 designation conflict with Table 2 and the Results, and the C16orf95 lead-variant identifier differs between the text and table.
↳ Abstract; Results, “Meta-analysis replicates previously reported associations”; Table 2
Limited replication and population coverage
The novel loci lack independent replication, and the principal analysis is restricted to European ancestry; the non-European comparison includes only 416 participants and is exploratory.
↳ Methods, “Study design”; Results, “Cross-ancestry comparisons”; Discussion, limitations paragraph
Nabu Commentary
The large multi-cohort GWAS and its primary quality-control framework support a meaningful contribution to CSF Alzheimer’s biomarker genetics. The main downward pressure comes from the inferential layer: causal MR language is insufficiently stress-tested, pathway results use unadjusted thresholds, and the claimed brain-volume mechanism exceeds the reported covariance evidence. The Abstract and Table 2 also disagree on novelty classifications and overgeneralize the external AD-related associations. The paper is therefore most reliable as a locus-discovery and prioritization resource rather than as confirmation of causal or mechanistic pathways.
Check another paper
Author Note / Field Feedback
Nabu’s assessment, alongside the field’s view.
Are you an author of this paper?
Dimension breakdown
Study Quality
Sound3.0
Confidence highContribution
3.9
The study meaningfully enlarges the CSF-biomarker GWAS evidence base through 18,948 participants across 30 studies and reports replicated and newly associated signals. Its contribution remains below the highest band because the new loci lack independent replication and their novelty labels are inconsistent.
Introduction; Results, “Meta-analysis replicates previously reported associations”; Table 2; Discussion, paragraph 1
making this study the largest to date for these endophenotypes
Methodological Rigour
2.5
The primary GWAS uses standardized cohort processing, genotype QC, inverse-variance meta-analysis, heterogeneity reporting, conditional analysis, and sensitivity analyses. Methodological Rigour is limited by causal and mechanistic conclusions that are firmer than the MR, nominal pathway analysis, and non-significant brain-volume covariance warrant.
Methods, “Study design” through “Biological inference from pathway analysis”; Results, MR and pathway sections; Figure 3C; Figure 4
MS4A6A is directly responsible for mediating both protective and aggravating microglial response
Reporting
3.0
The manuscript follows a coherent progression from association discovery to downstream interpretation, with principal estimates displayed in tables and figures. Central Abstract language conflicts with Table 2 and the Results over novelty and whether all lead variants show AD-related associations.
Abstract; Table 2; Results, “Significant genetic loci show association with AD risk and progression”
Lead variants on SLCO1A2 and chromosomes 8 loci did not show significant associations
Positioning
2.8
The paper compares its findings with prior CSF GWAS and acknowledges ancestry, replication, functional-validation, and QTL-resource limitations. Positioning is constrained because the Discussion states that mediation cannot be inferred without functional analysis but elsewhere presents specific genes and pathways as directly responsible.
Introduction; Discussion, mechanistic interpretation and limitations paragraphs
without additional functional analysis we cannot comment on the directionality of these changes
Lower confidence on Methodological Rigour, Positioning — domain match limited.
Reliability
Caveats4 of 4 checks
Internal CoherenceNoted
The paper contains material inconsistencies in how it describes novelty, the universality of AD-related associations, and the direction of the MR finding. These affect interpretation of central claims but do not establish that the primary GWAS estimates are invalid.
Research ConductClean
Ethics approval and informed consent are declared, and summary-statistic links are provided. No affirmative research-conduct concern is established in the supplied text.
Flags: 2 declared / 5 total
Reference IntegrityNoted
90 references in manuscript 90 of 90 checkable references found in an index
Post-Publication RecordClean
No retraction notice found in Retraction Watch.
Last checked: 08 Oct 2026
Sources: Retraction Watch ✓
Readiness
Where this paper’s evidence sits on the path from initial observation to real-world use.
Mechanistic
The findings provide association-based hypotheses and candidate-gene prioritization rather than validated targets or clinical tools. Independent replication, functional experiments, and intervention evidence remain necessary before practical use.
Results, “Gene prioritization for significant loci”; Discussion, limitations paragraph
future replication of the novel associations in larger studies will be needed
AI-generated, human-governed. Something look off? Contact us to request a review.