Summary: A critical research letter warns that a recently promoted statistical method used to support a new generation of Alzheimer’s drugs can greatly overstate clinical benefit. The study evaluated a technique called quantile aggregation, which groups trial participants by outcome, averages those group results, and then looks for patterns across the grouped averages.
Applied to high-profile reanalyses — including a reanalysis of data for Eli Lilly’s donanemab — the investigators used computational simulations to demonstrate that quantile aggregation can conceal individual differences and dramatically amplify the apparent relationship between amyloid clearance and cognitive improvement, at times inflating that association by as much as 29-fold.
Key Facts
- What quantile aggregation does: The method reorganizes trial participants into ordered groups based on posttreatment biomarker levels, computes average outcomes for each group, and evaluates trends across those averaged blocks rather than examining individuals or randomized groups.
- Magnitude of distortion: In simulations designed to mirror recent Alzheimer’s trials, the approach exaggerated the link between lowering amyloid burden and slowing cognitive decline by up to 29 times the true effect size.
- Hidden patient variability: Averaging across large, heterogeneous groups smooths over the genuine range of individual responses, producing an artificial consistency that can mislead about clinical predictability.
- Undermining randomization: By regrouping participants based on posttreatment measures rather than initial random assignment, the technique mixes treated and placebo patients, making it impossible to determine whether amyloid reduction actually caused any observed cognitive change.
- Resurrecting failed trials: The researchers demonstrated the method’s weakness by applying it to data from a trial of solanezumab conducted from 2014–2023. Although that trial failed to slow cognitive decline, the quantile aggregation procedure falsely produced a strong association between amyloid reduction and better cognitive outcomes.
- Independent scrutiny: The senior author noted that independent academic investigators, free from industry financial incentives, were able to rigorously test and expose important methodological shortcomings in approaches used to evaluate consequential Alzheimer’s treatments.
Source: Brown University
Overview
A team led by researchers at the Brown University School of Public Health published a concise research letter in JAMA Neurology that raises methodological concerns about quantile aggregation, an analytic technique increasingly cited in analyses of anti-amyloid therapies. The letter explains why this statistical strategy can produce misleading conclusions about whether reductions in brain amyloid translate into meaningful cognitive benefits.
The letter specifically examined how quantile aggregation performs when applied to relationships between cognition and amyloid, the protein that accumulates in Alzheimer’s disease. That analytic approach was previously used in a reanalysis of randomized trial data for donanemab. Brown’s team ran simulations and secondary analyses to test the statistical properties and limitations of this method.
Sarah Ackley, the study’s senior author and an assistant professor of epidemiology at Brown who directs the Computational Epidemiology Lab, explained that while many researchers target amyloid because it is hypothesized to contribute to cognitive decline, the real-world association between amyloid change and cognitive outcomes is noisy and often weak. Quantile aggregation, she said, can make a weak or noncausal relationship appear robust.
In simulation scenarios designed to reflect recent clinical trials, the Brown team found the method could exaggerate the apparent link between amyloid reduction and cognition by a factor of 29. The principal mechanism is straightforward: grouping and averaging remove the variation across individuals, creating smoothed group-level trends that overstate predictability.
Another critical limitation is that quantile aggregation regroups participants by posttreatment biomarker levels rather than preserving the randomized treatment assignments. That practice collapses the trial’s protected comparison between drug and placebo, preventing a reliable inference that amyloid lowering caused any observed cognitive benefit.
To make the issue concrete, the researchers applied the same aggregation method to historical data from the Anti-Amyloid Treatment in Asymptomatic Alzheimer’s Disease Study (2014–2023) testing solanezumab. That trial failed to show cognitive benefit, yet when put through the quantile aggregation procedure it produced a misleading result suggesting a strong link between lower amyloid and improved cognition — a finding inconsistent with the original trial’s conclusions.
Ackley emphasized that these results do not resolve the broader scientific debate about whether and how anti-amyloid therapies provide clinical benefit. Instead, the study highlights an urgent need for rigorous, transparent statistical methods and for broader data sharing in Alzheimer’s research as new treatments are adopted and covered by public programs.
“Our analysis was simple in concept but powerful in demonstrating why independent academic evaluation of analytic methods matters,” Ackley noted. “Working outside industry incentives allowed us to test a methodological approach that could materially influence how major new drugs are interpreted.”
Key Questions Answered:
A: Not necessarily. While the hypothesis that amyloid removal will slow cognitive decline motivates many therapies, real-world cognitive measures are noisy. The true relationship between plaque reduction and sustained cognitive benefit may be weak or inconsistent, and certain statistical approaches can magnify a weak association into an apparently strong one.
A: Quantile aggregation groups and averages patient outcomes, smoothing over individual variation. That averaging can erase outliers and idiosyncratic responses, producing a smooth, deceptive trend line that overstates how predictive amyloid change is for cognitive outcomes.
A: The researchers tested the method on solanezumab trial data, which originally showed no cognitive benefit. After regrouping participants by posttreatment amyloid and averaging results, the quantile aggregation analysis produced a false positive signal implying both amyloid removal and cognitive improvement that did not occur in the underlying randomized trial.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- The journal paper was reviewed in full by the reporting team.
- Additional context and clarification were added by staff editors.
About this Alzheimer’s disease and neuropharmacology research news
Author: Juan Siliezar
Source: Brown University
Contact: Juan Siliezar – Brown University
Image: The image is credited to Neuroscience News
Original Research: Open access. “Methodological Considerations for Quantile Aggregation in Alzheimer Disease Trials” by Michael D. Flanders, Michelle Caunca, Renaud La Joie, Lon S. Schneider, and Sarah F. Ackley. JAMA Neurology. DOI: 10.1001/jamaneurol.2026.1240
Abstract
Methodological Considerations for Quantile Aggregation in Alzheimer Disease Trials
Is amyloid reduction an appropriate surrogate outcome for clinical benefit in anti-amyloid trials? This question is central to interpreting current results and designing future studies. Noisy cognitive measures complicate that assessment and have prompted alternative analytic strategies. Quantile-aggregation methods that regroup participants by posttreatment amyloid rather than by randomized assignment have recently produced apparent amyloid–cognition relationships in data from trials such as TRAILBLAZER-ALZ 2. The statistical properties and limitations of this approach for interpreting trial data require careful characterization.