nabu.science
MethodologyAboutContact

 

  • Snapshot
  • Key Claims
  • Strengths & Limitations
  • Commentary
  • From the Authors
  • Author Note / Field Feedback
  • Dimension breakdown
  • Impact Profile
Free · About 5 minutes
Report #NBU-ABA9D1D3·Framework v5.9·Pipeline v0.4·Evaluated 08 Oct 2026·OECD: 3.2 Clinical medicine·Subfield: Genetics·Basis: Full text·Type: Computational or machine-learning study

Shared proteomic landscape between arteriosclerosis and cardiovascular endpoints: a Mendelian randomization and observational study integrating AlphaFold3 for structural prediction.

Huang et al.Huang, Meena, Achtari et al.·2026·Cardiovascular research·10.1093/cvr/cvag095 ↗

Primary claim

We prioritized 10 proteins potentially causally associated with both the arteriosclerotic/atherosclerotic markers and cardiovascular events.

Supported · staged colocalization and cis-MR support prioritization, though many associations rely on single instruments
In brief

A competent cross-vascular proteomic prioritization study that adds useful mediation hypotheses, but single-instrument evidence and overextended causal framing limit confidence.

ReliabilityCaveats4 of 4 checks completeRead the 2 entries ↓
Abstract

AIMS: Atherosclerosis and arteriosclerosis are major contributors to cardiovascular disease (CVD), yet their shared and distinct molecular underpinnings remain incompletely understood. This study integrates proteomics, Bayesian co-localization, Mendelian randomization (MR), and structural modelling to explore the shared and distinct plasma proteome associated with arteriosclerosis and atherosclerosis across different vascular beds.

METHODS AND RESULTS: We leveraged cis-protein quantitative trait loci (pQTLs) for 5813 unique proteins from the UK Biobank (UKB) Pharma Proteomics Project (N = 54 219) and deCODE genetics (N = 35 559) and assessed the association with five arteriosclerotic/atherosclerotic markers, as well as eight cardiovascular events, using Bayesian co-localization and bidirectional MR. We validated and replicated the findings through independent proteomics datasets, tissue-specific transcriptomics, observational data from UKB, and AlphaFold3 for structural prediction. Finally, mediation analysis evaluated the role of vascular traits in linking proteins to CVD risk. We prioritized 10 proteins potentially causally associated with both the arteriosclerotic/atherosclerotic markers and cardiovascular events. Five of them (ANGPTL4, apolipoprotein B [APOB], BRAP, lipoprotein(a) [LPA], and ZPR1) were associated with increased levels of arteriosclerosis/atherosclerosis and risk of CVD, whereas four (DUSP13, FN1, IL6R, and matrix metalloproteinase 12 [MMP12]) were associated with reduced levels of arteriosclerosis/atherosclerosis and risk of CVD. ABO was associated with increased risk of peripheral artery disease (PAD) and CVD but inversely related to arterial stiffness index (ASI). Of these, seven were replicated in an independent pQTLs data source from the Fenland study. Mediation analyses estimated that LPA's effect on stroke was primarily mediated through carotid plaque (92.4%). Observational analyses and transcriptomic validation corroborated these associations. Structural modelling using AlphaFold3 identified key functional variants in several proteins, including ANGPTL4 and FN1, potentially underlying the pathogenic mechanists.

CONCLUSION: The present study elucidates the shared and distinct proteomic signatures across arteriosclerosis, atherosclerosis, and CVD, underscoring the importance of vascular-bed-specific mechanisms. These identified proteins offer promising avenues for biomarker-driven risk stratification and therapeutic interventions, with potential for dual-purpose interventions across vascular territories.

Key Claims

Primary finding · Supported

We prioritized 10 proteins potentially causally associated with both the arteriosclerotic/atherosclerotic markers and cardiovascular events.

staged colocalization and cis-MR support prioritization, though many associations rely on single instruments

Abstract
Quote verified
Secondary · Tentative

LPA's effect on stroke was primarily mediated through carotid plaque (92.4%).

network-MR mediation lacks reported headline uncertainty and depends on unstated sequential causal assumptions

Abstract; Results, Figure 6B
Quote verified
Secondary · Supported

limited overlap of potential causal proteins across the studied markers in different vascular beds

the reported screen found little overlap, although differing phenotype power may contribute

Discussion, paragraph 1
Quote verified
Secondary · Tentative

Structural modelling using AlphaFold3 identified key functional variants in several proteins, including ANGPTL4 and FN1

structural visualizations are hypothesis-generating and lack functional validation or reported confidence metrics

Abstract; Results, “Structural alterations of 7 proteins predicted by AlphaFold3”
Quote verified

Derived from the full evaluation — not a separate score.

Strengths · Limitations

Strengths

#methodological rigour

Layered causal-prioritization pipeline

The study combines colocalization, cis-pQTL MR, directionality checks, transcriptomic analyses, observational triangulation, and mediation across multiple vascular phenotypes.

↳ Methods, Figure 1; Results, Figures 2–6

#methodological rigour

Detailed assay-artefact sensitivity work

Missense-linked instruments are cross-referenced against eQTL evidence, removed in sensitivity analyses, and examined in a gene-dense-region analysis that leads to exclusion of ERP29, LTBP4, and FRZB where appropriate.

↳ Methods, “Proteome-wide Mendelian randomization analysis”; Results, missense-IV and gene-dense-region sections

#contribution

Broad vascular-bed comparison

Five markers and eight cardiovascular outcomes are analyzed to distinguish shared from vascular-bed-specific protein associations, yielding a focused ten-protein set for downstream study.

↳ Abstract; Results, Figures 2–4

Limitations

#methodological rigour

Many estimates rely on one instrument

Only 33.3% of candidates had at least three instruments, leaving most estimates without weighted-median or MR-Egger sensitivity analyses and making residual pleiotropy difficult to assess.

↳ Results, “Potential causal proteins prioritized”; Discussion, “Strengths and limitations”

#methodological rigour

Mediation evidence lacks robust qualification

The headline LPA mediation estimate differs between the Abstract and Results, lacks a reported confidence interval in the main text, and rests on assumptions not explicitly developed in the mediation methods.

↳ Abstract; Methods, “Network MR and mediation analysis”; Results, Figure 6B

#reporting

Validation and therapeutic language overreaches

Transcriptomic results include missing instruments and opposite-direction findings, while observational support covers only part of the protein set. The Abstract and Conclusions nevertheless describe broad corroboration, causal evidence, and therapeutic promise.

↳ Abstract; Results, SMR/HEIDI and observational triangulation sections; Conclusions

Does this match your read? Add your assessment ↓

Nabu Commentary

The paper's strongest feature is its staged cis-pQTL framework, including colocalization, directionality checks, and unusually detailed handling of missense-linked assay artefacts. The cross-vascular comparison and mediation layer add a useful, delimited advance, but many candidate estimates rely on a single instrument and cannot support the strongest causal language. The LPA mediation result is potentially informative yet is reported with inconsistent percentages and insufficient uncertainty and assumption reporting. The extensive limitations section improves interpretability, although the Abstract and Conclusions still overstate corroboration, structural mechanism, and therapeutic readiness.

Evaluate another paper

Free · About 5 minutes · Research preview

Author Note / Field Feedback

Nabu’s assessment, alongside the field’s view.

Are you an author of this paper?

Dimension breakdown

Evidence Score

Out of 5
012345
3.3±0.2

Contribution
3.3

The study adds a cross-vascular comparison, a ten-protein prioritization, and network-MR mediation to an established proteome-wide MR literature. The advance is meaningful but incremental because several highlighted proteins are already established cardiovascular actors and the structural layer is hypothesis-generating.

Introduction; Discussion, paragraphs 1 and 3; Results, Figure 6

We prioritized 10 proteins potentially causally associated with both the arteriosclerotic/atherosclerotic markers and cardiovascular events.

Methodological Rigour
3.3

The staged use of cis-pQTLs, colocalization, directionality checks, missense-instrument analyses, and gene-dense-region sensitivity testing supports competent execution. Confidence is reduced because most candidates lack multi-instrument sensitivity testing and the mediation and structural claims omit important assumptions or uncertainty metrics.

Methods, “Proteome-wide Mendelian randomization analysis” and “Network MR and mediation analysis”; Results, “Sensitivity test in gene-dense regions”

33.3% of the protein candidates were assessed using three or more IVs.

Reporting
3.2

The progression from colocalization through MR, validation, mediation, and structural prediction is readily followable. Causal, corroborative, and therapeutic language in the Abstract and Conclusions is stronger than the partial transcriptomic, observational, and single-instrument evidence supports.

Abstract; Conclusions; Results, SMR/HEIDI and observational triangulation sections

Observational analyses and transcriptomic validation corroborated these associations.

Positioning
3.5

The discussion addresses ancestry, horizontal pleiotropy, single-instrument evidence, tissue discordance, lifelong MR effects, and sample overlap. Those caveats are not fully carried into statements describing the proteins as causal drivers and potential intervention targets.

Discussion, “Strengths and limitations”; Discussion, therapeutic implications; Conclusions

several associations were only supported by the Wald ratio method using a single IV

Reliability

Caveats4 of 4 checks


Results checkNoted
  • LPA's effect on stroke was primarily mediated through carotid plaque (92.4%): "Mediation analyses estimated that LPA's effect on stroke was primarily mediated through carotid plaque (92.4%)." (Abstract, Methods and Results) vs "The most prominent mediation effect was the LPA effect on any stroke via carotid plaque (proportion mediated: 92.7%; Figure 6B), followed by its effect on large-artery stroke via carotid plaque (proportion mediated: 42.7%)." (Results, Mediation effect of protein on cardiovascular events via arteriosclerosis and atherosclerosis). The Results estimate exceeds the Abstract estimate by 92.7 − 92.4 = 0.3 percentage points, and both estimates indicate predominantly mediated effects. → Conclusion holds
Reporting and transparencyNoted

The stated mediation formula does not define a proportion, and the SMR FDR inequality leaves its selection threshold unclear; neither establishes that an incorrect analysis was actually performed.

Preregistration: not stated · Data availability: declared · Code availability: not stated

Reference IntegrityClean

57 references in manuscript 56 of 56 checkable references found in an index 1 are books, websites or datasets — counted, but not index-checkable

References not verified automatically are resolved manually within 24 hours.

Post-Publication RecordClean

No retraction or expression of concern found in Crossref (includes Retraction Watch).

Last checked: 17 Oct 2026

Sources: Crossref (incl. Retraction Watch) ✓

Impact Profile

Impact areaHealth · Cardiovascular · Ischemic heart disease
Type of advanceConceptual
Route to useSuggested
AdvanceExtends
BeneficiariesLarge
CitationsToo early
Policy · patentNot yet traced
ReachedNot reachedNot assessable or unverified
ReadinessValidated
Foundational
Controlled
Piloted
Validated
Implementable
FoundationalImplementable

The prioritized protein associations were replicated in independent cohorts (Fenland, deCODE) and triangulated with transcriptomic and observational UK Biobank data, though no intervention has been tested.

From the paper
“these proteins may enable biomarker-driven early risk stratification and targeted therapeutic strategies”

The paper names real-world uses of its findings, namely biomarker-based risk stratification and therapeutic targeting of the prioritized proteins, but does not specify who would act on them or in what decision setting.

Pathway

Drug developers and clinical researchers would pursue the prioritized proteins, such as LPA and ANGPTL4, as therapeutic targets and biomarkers for cardiovascular risk stratification.

Barrier

The causal evidence is genetic and observational, so experimental validation and clinical trials are still needed before these proteins can guide therapy or risk stratification.

Evaluated 13 Nov 2026 · GBD 2023 · glm_v1_3_3 · impact_profile_v1_3_3

Report an issue →

AI-generated, human-governed. Something look off? Contact us to request a review.

Keep reading

Related evaluations

Other papers Nabu has evaluated blind, starting with the same field.

  1. Genome-wide meta-analysis of quantitatively measured generalized anxiety symptoms in individuals of European ancestry

    Skelton, Mitchell, Assary, et al. · 2026 · Clinical medicine

    Primary claim

    We identified 80 independent genome-wide significant variants within 74 loci, 39 of which were newly associated with anxiety.

    Well-supported
    Evidence Score
    4.0±0.2
    Reliability
    Caveats
  2. Co-expression-based models improve eQTL predictions for transcriptome-wide association studies and highlight new schizophrenia-associated genes

    Rossi, Sportelli, Kikidis, et al. · 2026 · Clinical medicine

    Primary claim

    Applying this framework to Psychiatric Genomics Consortium wave 3 genotypes identified 766 genes associated with schizophrenia

    Tentative
    Evidence Score
    3.0±0.2
    Reliability
    Caveats
  3. GWAS of extended prescription analgesic use identifies genetic loci in chronic pain

    Harlow, Uzochukwu, Fernando, et al. · 2026 · Clinical medicine

    Primary claim

    Integrating these genetic associations with functional data identifies potential mechanisms involved in chronic pain, implicating oligodendrocyte differentiation, neuronal guidance, endolysosomal function and post-synaptic endosome recycling.

    Tentative
    Evidence Score
    3.4±0.2
    Reliability
    Caveats

How the rubric works →