A novel artificial intelligence (AI) tool has been proven to accurately stratify patients with early cognitive impairment according to their speed of progression toward Alzheimer’s disease. In a study published in Nature Communications, the AI model was used to analyze data from a past late-stage trial that originally failed to meet the efficacy endpoints, retroactively showing significant cognitive improvements in those patients with a slow progression profile.
“Promising new drugs fail when given to people too late, when they have no chance of benefiting from them,” said Zoe Kourtzi, PhD, professor of experimental psychology at the University of Cambridge and senior author of the study. “With our AI model, we can finally identify patients precisely and match the right patients to the right drugs. This makes trials more precise, so they can progress faster and cost less, turbocharging the search for a desperately needed precision medicine approach for dementia treatment.”
Despite decades of research, late-stage clinical trials in Alzheimer’s disease remain largely unsuccessful, with a 95% failure rate. One of the major challenges is the high variability among the patient population in regards to symptoms, disease progression, and treatment responses; a problem that is compounded by the lack of tools for precise patient stratification.
The AI model was trained on data from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database to make predictions about future cognitive decline. The machine learning algorithm achieved 91% classification accuracy by leveraging baseline data from PET and MRI scans, taken before treatment began, on three key biomarkers: beta-amyloid, apolipoprotein E (ApoE4), and medial temporal lobe grey matter density.
“Our AI model gives us a score to show how quickly each patient will progress toward Alzheimer’s disease,” said Kourtzi. “This allowed us to precisely split the patients on the clinical trial into two groups—slow and fast progressing, so we could look at the effects of the drug on each group.”
The AI tool was then used to analyze data from the AMARANTH trial, a randomized Phase II/III clinical trial sponsored by AstraZeneca and Eli Lilly that was discontinued in 2018. The trial investigated the effects of the experimental drug lanabecestat, an oral inhibitor of the beta-site amyloid precursor protein-cleaving enzyme 1 (BACE1), in a population of more than 2,200 participants. While the drug had been successful at reducing levels of beta amyloid throughout the entire patient population, it did not result in any significant improvements in cognitive symptoms.
A re-analysis of the data using the AI tool showed that among patients at earlier stages of neurodegeneration and predicted to have slow progression, the investigational drug achieved a 46% reduction in cognitive decline compared to placebo. Meanwhile, patients classified as undergoing rapid disease progression showed no significant differences in cognitive outcomes compared to placebo. Because the majority of the patients enrolled in the trial belonged to the rapid progression group, the effects of lanabecestat in the population with slow progression could not be observed in the initial analysis of the trial data.
The study also showed that precise patient stratification can substantially decrease the sample size necessary to identify significant clinical outcomes, highlighting that clinical trials can be run much faster and at a fraction of the cost by identifying the right patients at earlier stages of neurodegeneration. These findings further reinforce the importance of precise patient stratification, especially as recent research increasingly shows that the timing of beta-amyloid removal strategies is critical for the success of Alzheimer’s trials.
“AI can guide us to the patients who will benefit from dementia medicines, by treating them at the stage when the drugs will make a difference, so we can finally start fighting back against these cruel diseases,” said Kourtzi. “Making clinical trials faster, cheaper, and better, guided by AI, has strong potential to accelerate discovery of new precise treatments for individual patients, reducing side effects and costs for healthcare services.”
