Reviews/Alzheimers Disease Treatment/Amyloid positron emission tomography reduction predicts benefit
Biomarkers · AD-BIOM-02 · v2
This corpus asks whether lowering amyloid on a brain scan reliably forecasts a slowing of cognitive decline when an anti-amyloid drug is used. The strongest supportive evidence comes from lecanemab: a modelling study [7] and the Clarity AD trial [23], where a large amyloid drop and a slower decline appeared together. But the one study built specifically to test whether amyloid change explains the cognitive benefit [37] found it accounted for only a small, statistically unreliable share, and a trial of a different compound [24] lowered amyloid with no cognitive or functional improvement at all. The remaining supportive papers [17][28] are about how much amyloid someone starts with, or a non-standard scan signal, rather than the exact thing claimed. The panel landed between support and contradiction because the direct trial evidence genuinely points both ways.
Within high-efficacy amyloid removal, the signal is present: [7] reports amyloid PET was a better predictor of drug effect than drug exposure and frames amyloid reduction as a surrogate of efficacy, and [23] shows a −59.1 centiloid amyloid reduction and a −0.45 CDR-SB difference arising in the same lecanemab trial. Against this, the purpose-built surrogacy evaluation [37] found amyloid change mediated 14.6% of the cognitive effect with a 95% CI from −122% to 208%, and states the question of validity cannot be conclusively answered. [24] is a direct within-trial dissociation: a 3% PiB-PET SUVR decline with no accompanying cognitive or functional benefit. The cohort evidence [17][28] concerns white-matter retention or baseline amyloid burden predicting decline, not cortical amyloid reduction predicting trial benefit. The contested point: The division is whether a predictive relationship seen within one strong-removal drug programme ([7], [23]) is a general rule or a context-bound co-occurrence. The corpus's only formal mediation test [37] is near-zero and wildly imprecise, and [24] shows reduction without benefit, so the same body of primary evidence reads as supportive, null, and contradictory depending on the trial.
All 39 papers behind this claim with their DOIs and what each does to it, the panel’s arbitration, the evidence gap map and the open research questions. Free, and sent to your inbox.
We’ll also send you other claims from Alzheimers Disease Treatment, one at a time, and tell you when a release or a pivotal trial changes a grade. Unsubscribe by replying.
One reviewer's only job is to attack the finding. The challenges below were weighed before the grade was set.
Clarity AD [23] demonstrates only that amyloid reduction and cognitive benefit co-occur in one trial; co-occurrence is not a predictive/surrogate relationship and does not generalise across anti-amyloid programmes.
The single study designed to test surrogacy [37] found amyloid change mediated only 14.6% of the cognitive effect with a CI from -122% to 208%, and states validity cannot be conclusively answered — the corpus's most direct test is null/inconclusive.
Where it remains uncertain. The purpose-built surrogacy analysis [37] found amyloid change mediated only 14.6% of the cognitive effect (95% CI −122% to 208%), and PBT2 [24] cut amyloid by 3% with no cognitive benefit.
Aleatory
The [24] dissociation rests on 40 completers (placebo=15, PBT2=25) with large individual SUVR variation the authors themselves flag, so chance cannot be excluded as driving the null.
Epistemic
The claim's directness depends on [37] alone, which was underpowered by minimal disease progression and minimal amyloid removal and states validity cannot be conclusively answered.
Inferential
[23] shows amyloid reduction and benefit arising together in one trial, which does not by itself establish that the former predicts the latter across programmes.
The load-bearing papers from a corpus of 39, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
Direct supportive evidence that amyloid-PET reduction predicts treatment effect within lecanemab trial data.
“Amyloid PET was a better predictor of drug effect than lecanemab exposure, demonstrating amyloid reduction as a surrogate marker of efficacy.”
Panel auditInternal validity: medium · scope of inference: Model-based surrogacy of amyloid PET for CDR-SB change in lecanemab-treated early AD.Key limitations: modeling inference · single drug · no effect estimate in abstract
Shows substantial amyloid reduction co-occurring with slowed CDR-SB decline (difference -0.45) in the pivotal anti-amyloid RCT.
“there were greater reductions in brain amyloid burden with lecanemab than with placebo (difference, -59.1 centiloids; 95% CI, -62.6 to -55.6)”
Panel auditInternal validity: high · scope of inference: Early AD treated with lecanemab over 18 months.Key limitations: co-occurrence not demonstrated mediation · single drug · modest absolute effect
Within-trial dissociation: a 3% SUVR decline occurred without cognitive/functional benefit, counter-evidence to the claim.
“There were no significant differences between PBT2 and controls in fluorodeoxyglucose-positron emission tomography, magnetic resonance imaging volumetrics, blood Aβ biomarkers, or cognition/function over the course of the double-blind phase.”
Panel auditInternal validity: low · scope of inference: Prodromal/mild AD treated with the 8-OH quinoline PBT2.Key limitations: small n · exploratory · underpowered for cognition
Indirect support linking a florbetapir PET signal to cognitive decline and to lecanemab-induced reduction.
“Elevated NAWM ¹⁸F-florbetapir retention independently predicted cognitive decline, correlated with plasma p-tau217 and CSF p-tau/Aβ42 ratio, and showed significant reductions following lecanemab therapy.”
Panel auditInternal validity: medium · scope of inference: White-matter florbetapir signal as an AD progression biomarker.Key limitations: white-matter not cortical · observational · benefit link not tested
4 of 39 papers shown. The full review lists all of them, with the panel’s arbitration →
Claims are graded at each quarterly release, and re-graded between releases only when a pivotal trial lands. Every change is a new version of the claim.
Directly conflicting primary trial evidence — supportive in lecanemab, null in the formal surrogacy test and dissociated in PBT2 — leaves the predictive claim genuinely unresolved.
Release 2026-q4 re-graded every claim from its corpus under protocol v0.19. An earlier release's grade is shown for the record; it came from an earlier protocol and is not a movement.
Assumptions that carry weight but are not themselves established.
The panel records the gap that would most move the claim, and the study that would close it.
Individual-level surrogacy test in a high-efficacy remover. The only formal mediation test [37] used a weak remover in a low-progression sample, exactly the conditions its authors say limit conclusions; the claim's directness depends on repeating it where amyloid removal is large. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: Apply the [37] instrumental-variable and mediation methods to individual-level data from a strong-removal anti-amyloid trial.
Open question: Does amyloid reduction mediate the cognitive benefit in high-efficacy anti-amyloid trials, not just the low-progression A4 setting? Why it is still open: [37], the only formal mediation test, returned 14.6% mediation with a CI from −122% to 208% and states validity cannot be conclusively answered in that sample.
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 21 Oct 2026 · protocol v0.19