Scientists identify three distinct paths of cognitive decline in early Alzheimer’s disease

Recent research published in Alzheimer’s & Dementia suggests that older adults with early biological signs of Alzheimer’s disease experience vastly different rates of cognitive decline. While many of these individuals remain mentally sharp for years, a smaller subset experiences a rapid loss of memory and thinking skills. The study, funded in part by the National Institutes of Health, indicates that this variation is not well predicted by existing medical tests and provides evidence that clinical trials testing new treatments may need to rethink participant selection.

Alzheimer’s disease develops in the brain long before memory problems become apparent to the patient or their family. This early stage is known as preclinical Alzheimer’s disease. During this extended period, abnormal proteins begin to build up slowly in the brain.

One of these primary proteins is called amyloid beta, which forms sticky plaques between brain cells. Another protein involved in the disease process is called tau. Tau typically helps stabilize the internal structure of healthy brain cells.

In Alzheimer’s disease, tau becomes abnormal and forms destructive tangles inside the cells. Scientists can measure the presence of these proteins using specialized brain scans or blood tests. These measurements help doctors identify patients in the preclinical stages of the disease.

Medical professionals run specific studies, known as secondary prevention trials, to test medications during this very early stage. The goal of these preventative trials is to stop or slow down the disease before it causes noticeable memory loss. In the past, these clinical trials have usually grouped older adults together based simply on the presence of amyloid plaques.

The underlying assumption in the medical field has been that everyone with these plaques is on a similar path toward memory loss. However, scientists noticed that the speed of cognitive decline varied greatly from person to person in real clinical settings. Some older adults with amyloid buildup seem to maintain their cognitive abilities for an extended period without issue, while others worsen quickly.

“Most studies look at the average across participants, which can make it seem like everyone is slowly getting worse at the same rate,” said Michael Donohue, a professor of neurology and associate director of biostatistics at the USC Epstein Family Alzheimer’s Therapeutic Research Institute at the Keck School of Medicine. Donohue is the corresponding author of the paper. “But we found that this approach masks major differences between people, suggesting that Alzheimer’s disease is more variable than often depicted.”

To better understand these different paths, the researchers investigated whether specific biological markers could predict a person’s trajectory. They also explored how this natural variation affects the underlying mathematics of running a successful clinical trial.

To explore these differing paths, the authors analyzed data from two large, parallel studies involving older adults aged 65 to 85. The first was the Anti-Amyloid Treatment in Asymptomatic Alzheimer’s Disease study, a clinical trial testing a specific treatment. They also included data from a companion study of people without elevated amyloid levels.

All participants were considered completely cognitively unimpaired at the beginning of the research. The total sample included exactly 1,629 individuals. The average age of the participants was about 71 years old, and roughly 60 percent of the group was female.

Of this group, 1,110 participants had elevated levels of amyloid beta in their brains, while the remaining 519 participants did not. The researchers tracked the participants’ memory and thinking skills for up to seven years. The median follow-up time for the entire group was exactly six years.

Before and during the trial, participants completed a series of cognitive tests that measured their memory, attention, and thinking. Scores on these standardized assessments were used to track their rate of cognitive decline.

The researchers also collected blood samples to measure a specific type of abnormal tau protein called p-tau217. In addition to blood tests, they used advanced brain imaging to look at exact amyloid levels across the brain. They also used magnetic resonance imaging scans to measure the exact volume of the hippocampus, a brain region responsible for forming new memories and one of the first affected by the disease.

A smaller subset of exactly 427 participants also underwent a specific type of positron emission tomography scan, or PET scan, to measure tau tangles directly. The authors applied complex statistical models to all of this collected data to identify unobserved subgroups of cognitive decline.

Rather than assuming an average rate of decline for everyone, the mathematical model allowed the data to dictate how many distinct paths existed. The data analysis revealed three distinct trajectories of cognitive decline. These included a stable group with no change or slight improvement, a slow decline group showing a gradual drop in test scores, and a fast decline group exhibiting a rapid and more pronounced drop in scores.

About 70 percent of all participants remained stable over the study period of approximately six years. When looking specifically at the participants who had elevated brain amyloid, the scientists noted a similar pattern. Almost 70 percent of these individuals remained entirely stable in their cognitive abilities during the observation period.

The scientists then examined whether biological markers at the beginning of the study could predict which group a person would fall into. They found that participants who showed a gradual or a fast decline had higher p-tau217 levels when the study began. These declining individuals also had higher levels of tau on brain scans than those who remained stable.

“P-tau217 was one of the strongest signs of which participants would decline, but we still cannot predict exactly how an individual person’s disease will progress,” Donohue said.

Individuals who entered the study with smaller hippocampi were also much more likely to experience faster cognitive decline. Interestingly, while amyloid levels played a role in predicting decline, the blood test for tau and the brain shrinkage measurements tended to be stronger predictors.

The genetic risk factor known as the APOE e4 allele also increased the chances of a person falling into a declining group. This specific gene variant is widely recognized as a major genetic predictor of Alzheimer’s disease. Using all of this biomarker data, the researchers could correctly predict whether participants were likely to stay stable or worsen about 70 percent of the time.

The authors discovered an interesting contrast regarding biological changes in the stable group. Even though the stable participants with elevated amyloid showed no signs of memory loss, their biological tests worsened over time. Their brain scans showed increasing amyloid and tau buildup, and their hippocampi continued to shrink. This suggests these stable individuals are simply in a much earlier phase of the disease process.

The researchers also performed mathematical simulations to understand how these different groups impact clinical trials. They modeled a hypothetical trial with 500 participants receiving a drug and 500 receiving a placebo. They found that a trial filled with stable individuals would have very little statistical power to detect a drug’s benefit after two or four years.

Because stable participants do not decline naturally, a medication cannot demonstrate a protective effect on their memory. The statistical power of a trial ends up relying almost entirely on the small minority of participants who are actually declining. The findings indicate that enrolling too many stable individuals dilutes the measurable impact of a new treatment.

“These results suggest we may need to rethink how we design clinical trials in preclinical Alzheimer’s disease,” said Runpeng Li, a postdoctoral scholar at the Keck School of Medicine and the study’s first author. “Many people with Alzheimer’s remain stable over the course of a study, which can make it hard to tell if a treatment is working. Identifying those who are more likely to decline could make trials more efficient and more informative.”

While the study provides robust data, there are a few potential misinterpretations to avoid. For example, the findings do not mean that stable individuals will never experience memory loss. Preclinical disease phases can last for over a decade before obvious symptoms surface. The six years of observation in this study might simply be too short to capture the eventual decline of the stable group.

The study also has certain limitations that affect how the findings can be applied. The statistical models used to group participants rely on probabilities rather than absolute categories. Additionally, the specific brain scans used to measure tau tangles were only available for a fraction of the participants, which limits the broader application of those specific tau findings.

The sample of participants also consisted of volunteers who met strict criteria for clinical trials. This specific group might not perfectly represent the general population of older adults living in the community. Unmeasured factors, such as diet, exercise, and other medical conditions, could also influence how quickly a person’s memory declines.

An important next step is to refine the model to more accurately predict which patients will decline quickly. Adding more blood tests, brain scans, or other biomarkers is one possible way to achieve that goal. This kind of prediction tool could someday give patients a more accurate prognosis when they are diagnosed with Alzheimer’s disease. Better predictive models could also support more effective clinical trials of potential treatments.

The scientists say future trials should focus less on average results and more on different patterns of decline. Next, Donohue and his colleagues plan to look at the participants who did not fit the model’s expectations. These include patients who were predicted to remain stable but worsened, as well as those predicted to decline who remained symptom-free.

“What is different about certain patients that makes them more resilient, and can these insights be leveraged to slow down Alzheimer’s disease in others?” Donohue said.

The study, “Divergent patterns of cognitive decline in preclinical Alzheimer’s disease: Implications for secondary prevention trials,” was authored by Runpeng Li, Oliver Langford, Philip S. Insel, Reisa A. Sperling, Rema Raman, Paul S. Aisen, and Michael C. Donohue.