Large multi-cohort longitudinal evidence base
The analysis covers 60,542 individuals across GNPC, UKB and NSHD, with prospective disease and mortality follow-up in UKB and repeated measurements in NSHD.
↳ Main, study overview; Results; Figs. 1, 3–5
Abstract
Aging is asynchronous across cells and organs. Here we tested whether plasma proteomics can be used to analyze cell type-specific aging. From analyses of over 7,000 plasma proteins measured in 60,542 individuals, we developed machine learning models to estimate the biological age of over 40 cell types spanning neuronal, immune, glial, endocrine, epithelial and musculoskeletal origins. We observed that 20-25% of individuals exhibited accelerated aging in a single cell type and 1-3% in 10 or more cell types. Cellular aging signatures were associated with disease status and predicted incident disease and mortality over 15 years of follow-up. Individuals with the APOE4 genotype showed older astrocytes but younger macrophages compared to APOE3 carriers, whereas the APOE2 genotype had inverse associations. Moreover, extreme astrocyte aging tripled the risk of incident Alzheimer's Disease in individuals with two APOE4 alleles, while youthful astrocytes reduced risk. Individuals with extremely aged compared to youthful skeletal myocytes exhibited a 12.7-fold higher risk of developing amyotrophic lateral sclerosis. In individuals who smoked, extreme respiratory epithelial cell aging was associated with a 58% higher lung cancer risk compared to smoking alone. Specific cellular vulnerabilities and cumulative cellular aging burden influenced survival, with youthful immune and neuronal cell types conferring protective effects. Finally, we developed a polycellular aging risk score that stratified mortality risk across cohorts and proteomics platforms. These findings establish a framework for quantifying human physiology at cellular resolution, revealing heterogeneous aging trajectories and their impact on disease susceptibility and resilience.
Key Claims
Primary findingTentative
Cellular aging signatures were associated with disease status and predicted incident disease and mortality over 15 years of follow-up.
Possible — the claim's weakest requirement is that the effect exists at all. Experimental. Every point below is traceable to a quoted passage, but this forecast has not yet been validated against known replication outcomes.
hazard associations support stratification, but predictive discrimination, calibration and incremental value were not reported
Abstract
SecondaryTentative
Individuals with extremely aged compared to youthful skeletal myocytes exhibited a 12.7-fold higher risk of developing amyotrophic lateral sclerosis.
the reported association uses a small youthful reference group with uncertain event counts and estimate stability
Abstract; Results, incident ALS; Fig. 3c
SecondarySupported
Extreme astrocyte aging tripled the risk of incident Alzheimer's Disease in individuals with two APOE4 alleles, while youthful astrocytes reduced risk.
cumulative incidence supports the threefold contrast, though the youthful APOE4/4 group contained only 23 participants
Abstract; Results, APOE-stratified AD analysis; Fig. 3e
SecondarySupported
In individuals who smoked, extreme respiratory epithelial cell aging was associated with a 58% higher lung cancer risk compared to smoking alone.
the reported hazards support the contrast, with observational adjustment rather than clinical prediction testing
Abstract; Results, lung cancer; Fig. 4a
SecondaryTentative
We developed a polycellular aging risk score that stratified mortality risk across cohorts and proteomics platforms.
one external cohort and cohort-specific percentile thresholds support stratification but not platform-agnostic deployment
Abstract; Results, PARS; Fig. 5e–f
Derived from the full evaluation — not a separate score.
Strengths · Limitations
Strengths
Held-out and cross-platform evaluation
Olink clocks and PARS were developed in the UKB training partition and evaluated in a held-out test partition, while PARS was also applied to SomaScan measurements in NSHD.
↳ Methods, Cellular Age Estimation and PARS Development; Fig. 5e–f
Targeted adjustments and sensitivity analyses
The study benchmarks astrocyte aging against APOE4, PRS and age, adjusts respiratory signatures for smoking exposure, and repeats the ALS analysis after a three-year diagnostic lag.
↳ Fig. 3f; Extended Data Figs. 6 and 9; Methods, Prognostic Analysis
Limitations
Prediction lacks essential performance benchmarks
The disease and mortality analyses emphasize hazard ratios and survival separation without reporting discrimination, calibration or broad comparisons against established clinical and general proteomic prognostic models.
↳ Results, Figs. 3–5; Methods, Prognostic Analysis and PARS Development
Cell specificity remains unisolated
The analysis does not compare the assigned panels with matched random protein sets, whole-proteome models or organ clocks, so the endpoint value attributable to cell-type mapping is unresolved.
↳ Methods, Identification of Cell Type-Enriched Plasma Proteins and Cellular Age Estimation
Cellular interpretation exceeds source evidence
The Discussion acknowledges that plasma proteins may be produced or released more broadly, yet the conclusions retain cellular-resolution and therapeutic interpretations.
↳ Discussion, limitations and conclusion
Nabu Commentary
The study's strongest features are its scale, center-based UKB train-test design, prospective follow-up and application of PARS across Olink and SomaScan. Its methodological score is constrained because hazard ratios and Kaplan–Meier curves do not establish prediction quality without discrimination, calibration and fair prognostic comparators. The absence of matched non-cell-specific or organ-clock analyses also leaves the added value of cell-type assignment unresolved. The paper acknowledges plasma-source ambiguity, but its cellular-resolution and therapeutic framing remains broader than that limitation supports.
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 mediumContribution
3.9
The study meaningfully extends plasma aging-clock work to putative cell-type signatures across 60,542 participants, three cohorts and two assay platforms. The advance is tempered by the absence of direct endpoint comparisons with organ-level or general proteomic clocks.
Main, study overview; Results, cellular aging model development; Figs. 3–5
We map plasma proteins to their putative cellular origin and construct computational models that measure the biological age of over 40 cell types.
Methodological Rigour
2.4
The UKB center-based train-test split, bootstrap aggregation, disease-specific adjustments and NSHD evaluation support the analysis. The central prediction framing lacks discrimination, calibration and broad prognostic baselines, while no matched ablation establishes that cell-type assignment adds endpoint-relevant information.
Methods, Cellular Age Estimation; Prognostic Analysis of Incident Disease; PARS Development; Figs. 3–5
Participants were stratified into high, medium, and low risk groups based on cohort-specific score distributions.
Reporting
3.3
The manuscript follows a coherent progression from protein mapping and clock construction to disease, mortality and PARS analyses. However, statements that the signatures predict or strongly prognosticate disease imply validated predictive performance beyond the reported hazard and survival separation analyses.
Abstract; Results, cancer and chronic disease prognosis; Discussion
Our findings reveal cellular aging signatures strongly prognosticate future disease.
Positioning
3.0
The paper positions the work against organ-level and tissue-based aging clocks and acknowledges transcript-to-protein and population limitations. Its conclusions nevertheless retain cellular-resolution and therapeutic interpretations that the plasma-source ambiguity materially narrows.
Main, prior-work framing; Discussion, limitations and conclusion
Plasma proteins arise from multiple cellular processes not explicitly disentangled here.
Reliability
Caveats3 of 4 checks
Internal CoherenceNoted
Several numerical and cohort-accounting issues warrant caution but do not reverse the main direction of the prospective findings. The most consequential is the unclear NSHD mortality event count used in longitudinal and cross-platform validation analyses.
Research ConductClean
Ethics approval and consent are declared for the contributing cohorts, with specific identifiers for KADRC and NSHD. Online data, code, competing-interest and reporting materials were not supplied for inspection, which is treated as an assessability gap.
Flags: 2 declared / 5 total
Reference IntegrityNot checked
Reference integrity check could not run: references could not be reliably parsed from the manuscript; verification skipped rather than risk false verdicts.
0 references in manuscript
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.
Demonstrated
Prospective follow-up, a held-out UKB set and NSHD validation move the framework beyond initial association. Deployment remains premature without calibration, discrimination, comparator performance or clinical-utility evidence.
Methods, Prognostic Analysis and PARS Development; Fig. 5f
The PARS model was validated in the UKB test set and the independent NSHD cohort.
AI-generated, human-governed. Something look off? Contact us to request a review.