The use of statistical approaches to support a new class of Alzheimer’s disease drugs may lead to exaggerated claims about the drugs’ effects, according to a new study led by researchers at Brown University School of Public Health.
Published in JAMA Neurologythe research letter focuses on quantile aggregation, a new statistical method that divides people into groups and averages the results to look for patterns across those groups.
In this letter, we examined how this approach works when applied to cognition and amyloid, a protein that accumulates in the brains of people with Alzheimer’s disease. This approach was originally published in Eli Lilly and Company’s analysis of the Alzheimer’s disease drug donanemab.
“Many researchers believe that reducing amyloid accumulation could slow the memory loss and cognitive decline associated with the disease, and could be a key target for new Alzheimer’s drugs,” said Sarah Ackley, lead author of the study. He is an assistant professor of epidemiology at the Brown School of Public Health and runs the Computational Epidemiology Laboratory.
The problem is that using this method to assess the impact of amyloid removal on cognition can yield misleading results. ”
Sarah Ackley, Brown University School of Public Health
According to the analysis, the researchers’ concern is that this approach could make the link between amyloid reduction and improved cognitive function appear much stronger than it actually is. The study the researchers focused on was a reanalysis of original data from a randomized controlled trial of donamab. It was led by scientists affiliated with pharmaceutical companies.
“When we ran the simulations, we found that we could basically take a very weak relationship between amyloid and cognition and make it seem like something very strong and important,” Ackley said.
The research team had expected there might be problems with this method, but were shocked by the magnitude of the effect.
In simulations designed to reflect conditions in recent trials, the researchers found that the method showed a link between amyloid and cognition that was 29 times higher than it would be in real life.
The researchers said this happens because combining large groups of patients and averaging their results masks variations in cognitive changes between patients. Therefore, amyloid reduction may appear to be more predictive of cognitive benefit than it actually is.
This method also combines patients who received the drug with those who received a placebo. Without that randomization, the study says, the analysis cannot reliably determine whether the amyloid reduction is actually causing the cognitive benefit or whether other factors are at play.
To illustrate this, the research team also tested the method using data from an anti-amyloid treatment study in asymptomatic Alzheimer’s disease conducted from 2014 to 2023. The trial tested whether the drug solanezumab could slow cognitive decline in older people who have elevated amyloid levels in the brain, an early sign associated with Alzheimer’s disease.
Although this trial showed that solanezumab did not slow cognitive decline, when the team ran data from that trial into an analysis of donanemab using quantile aggregation, they found a strong association between lower amyloid and improved cognitive outcomes.
“We’ve basically built a case that this method produces misleading results,” Ackley said. “This drug made a failed trial look like it succeeded in clearing amyloid, and made it look like clearing amyloid reduced cognitive decline. In reality, the drug had neither effect.”
Ackley stressed that the study results do not solve the broader question of how new treatments for Alzheimer’s disease work. If anything, she says, the study highlights the need for more rigorous statistical methods. She also emphasized the need for more data sharing in Alzheimer’s disease research, especially as new treatments become more widely used and covered by public programs such as Medicare.
“Our study was simple but a great demonstration of the value of academic research,” she said. “Working outside of industry incentives has given me the freedom to closely examine methodological issues that impact how some of the most important new medicines are understood.”
sauce:
Reference magazines:
Flanders, Maryland; others. (2026) Methodological considerations for quantile aggregation in Alzheimer’s disease clinical trials. JAMA Neurology. DOI: 10.1001/jamaneurol.2026.1240. https://jamanetwork.com/journals/jamaneurology/fullarticle/2849325.
