Association of plasma biomarkers with longitudinal change in age- and Alzheimer's disease-related brain atrophy patterns
Bilgel, M., Shah, I., Cooper, J., An, Y., Walker, K. A., Ho, S. G., Moghekar, A. R., Yang, Z., et al.
radiology and imaging
Quick verdict
Well-powered BLSA study: plasma p-tau217 and related biomarkers precede AD-like atrophy in unimpaired adults; but AD-specificity rests on few non-AD cases.
Summary
Using 818 cognitively unimpaired participants from the Baltimore Longitudinal Study of Aging (2,293 MRIs; median 3.9-year MRI span), the authors asked whether plasma biomarkers collected once at a baseline index visit preferentially precede Alzheimer's disease (AD)-related rather than age-related brain atrophy. Atrophy was summarized with machine-learned pattern scores: SPARE-BA (brain age), SPARE-AD (AD-like atrophy), and five R-indices (subcortical, medial temporal, parieto-temporal, diffuse cortical, perisylvian). Biomarkers were Quanterix Aβ42/40, GFAP, NfL, p-tau181, plus Lumipulse Aβ42/40 and p-tau217. In Cox models, p-tau181 (HR 1.7), Lumipulse Aβ42/40 (HR 0.6), and p-tau217 (HR 1.8) predicted incident MCI/dementia due to AD but not non-AD etiologies, whereas GFAP predicted both. In linear mixed models with Benjamini-Hochberg FDR across 42 tests, higher GFAP, p-tau181, and p-tau217 and lower Lumipulse Aβ42/40 predicted steeper SPARE-AD increases; only p-tau217 predicted medial temporal (R2) atrophy; p-tau181, p-tau217, and both Aβ42/40 assays predicted parieto-temporal (R3) atrophy. No biomarker tracked SPARE-BA. Limitations include only 24 non-AD converters, no multiplicity correction for the Cox analyses, a highly educated predominantly white cohort, and baseline-only plasma measurement.
Review complete
Assessment at a glance
4
risk flags identified
● high 0
● medium 2
● low 2
4 strengths · 4 weaknesses · 5 author questions
AI text signal = Binoculars zero-shot detector score over 20 sampled passages (Fast-DetectGPT co-reported) — absence does not prove human authorship, and a flag does not imply misconduct.
Levels reflect verification complexity — not a quality score. Evolve never rates your paper.
Key claims under review
- Plasma p-tau181, Lumipulse Aβ42/40, and p-tau217 measured in cognitively unimpaired adults predict incident MCI/dementia due to AD but not non-AD etiologies, while GFAP predicts both.
- Higher GFAP, p-tau181, and p-tau217 and lower Lumipulse Aβ42/40 are associated with steeper longitudinal increases in AD-like atrophy (SPARE-AD), FDR-corrected across 42 models.
- p-tau217 is the only biomarker associated with longitudinal medial temporal lobe atrophy (R2) and shows the most extensive associations with AD-related atrophy patterns.
- Plasma biomarkers show no association with brain-age atrophy (SPARE-BA) or with subcortical, diffuse cortical, or perisylvian atrophy dimensions (R1, R4, R5).
Strengths
Unusual longitudinal depth for a plasma biomarker study
2,293 MRIs in 818 cognitively unimpaired participants (median 2 scans; median 3.9-year span, 4.8 years among the 568 multi-scan participants, §3.1), with MRIs up to 5 years before plasma included to stabilize slope estimation (§2.1).
Pattern-score design mitigates the multiplicity problem inherent to regional MRI analyses
Seven machine-learned scores (SPARE-BA, SPARE-AD, R1-R5, §2.4) replace dozens of regional volumes; the 42 main models (6 biomarkers × 7 scores) are BH-FDR corrected (§2.5.2), and several survive (e.g., p-tau217→SPARE-AD, estimate 0.097/decade, pFDR=1.3e-4, Table 2).
Assay triangulation with transparent QC and orthogonal validation
Honest specificity framing and open analysis code
Weaknesses
AD-specificity claims rest on a very small non-AD comparator
Asymmetric multiplicity control
All positive biomarker findings are interaction-term-only
Platform contrasts are confounded by subsample composition
Full Review Report
$1.99 · one-timeUnlock the complete preliminary analysis Evolve produced for this paper.
Report preview
Scope & Claims
This medRxiv preprint (posted 2026-09-17, NIH/NIA intramural BLSA) asks whether plasma biomarkers measured once, while participants are cognitively unimpaired, preferentially precede **AD-related rather than age-related** brain atrophy. The cohort: 818 CU participants (Table 1: median age 73, 45% male, 67% White, median 18 years of education), 2,293 longitudinal 3T MRIs (median 2 scans/person; 3.9-year median span, 4.8 years among multi-scan participants). Atrophy is compressed into seven machine-learned pattern scores: SPARE-BA (brain age), SPARE-AD (AD-like), and five Surreal-GAN R-indices (R1 subcortical, R2 medial temporal, R3 parieto-temporal, R4 diffuse cortical, R5 perisylvian). Central claims: (1) p-tau181, Lumipulse Aβ42/40, and p-tau217 predict incident MCI/dementia due to AD but not non-AD, while GFAP predicts both (Supplementary Table 4); (2) these four track steeper SPARE-AD increase (Table 2); (3) p-tau217 alone tracks medial temporal atrophy; (4) no biomarker tracks SPARE-BA.
The remaining sections cover Methodology Assessment, Statistical Evidence, Results & Reproducibility, Limitations & Risks, and Verdict & Recommendations.
- All 4 strengths & 4 weaknesses, analyzed in detail
- 4 risk flags with severity, section & reviewer questions
- Methods & statistical evidence assessment
- 5 questions a peer reviewer would ask
- Full six-dimension evaluation report & recommendations
One-time purchase · this paper only
One-time purchase, this paper only. AI-generated preliminary analysis — not peer review.
Frequently asked questions
Does plasma p-tau217 predict Alzheimer's disease before symptoms appear?
In this BLSA preprint of 818 cognitively unimpaired older adults, a single higher baseline plasma p-tau217 measurement was associated with roughly 1.8 times the hazard of later developing MCI or dementia due to Alzheimer's disease, and with faster progression of AD-like brain atrophy on MRI over a median 3.9-year follow-up. It is a not-yet-peer-reviewed study and not clinical guidance.
What are SPARE-AD and SPARE-BA brain atrophy scores?
They are machine-learned MRI pattern scores. SPARE-BA summarizes brain atrophy typical of aging (trained to predict chronological age from brain structure), while SPARE-AD summarizes an Alzheimer's-like distributed atrophy pattern (trained to separate cognitively normal older adults from AD patients). This paper also uses five 'R-indices' capturing subcortical, medial temporal, parieto-temporal, diffuse cortical, and perisylvian atrophy.
Is this paper peer reviewed?
No. It is a medRxiv preprint posted September 17, 2026 (DOI 10.64898/2026.09.11.26362869), explicitly not certified by peer review, and should not be used to guide clinical practice.
Which plasma biomarker best tracks Alzheimer's-related brain atrophy?
In this study, p-tau217 showed the most extensive associations: it predicted incident MCI/dementia due to AD and was the only biomarker linked to faster medial temporal lobe atrophy (R2), while also tracking SPARE-AD and parieto-temporal (R3) atrophy. GFAP was associated with both AD and non-AD outcomes, and NfL and Quanterix Aβ42/40 showed the fewest AD-specific links.
About this review
This is an AI-generated preliminary review by Evolve.Pub, produced from the openly available text of the paper. It is not peer review and does not represent the authors' views. The canonical record of the paper is at the publisher link above.
- Figures 1-4 are described in captions but not visually analyzable from the extracted text; all quantitative judgments rely on in-text numbers and Tables 1-2 plus Supplementary Tables 1-4.
- PDF extraction garbled special characters (e.g., Aβ subscripts rendered as '!/', '#', p-values as '3.7 × 10$%'); values were recoverable from context but occasional digit-level transcription risk remains.
- The review covers the preprint as posted; peer-review-stage revisions could alter results.
- Supplementary Figure 2 (converter-stratified trajectories) is available only as a caption summary.
Go deeper on this paper
Wrote a paper? Stress-test it before reviewers do.
Upload your manuscript for the same structured scrutiny — reasoning, statistics, evidence, figures, and reproducibility.