Genetic test can predict Alzheimer’s disease risk in teens, but how accurate is it?

Screen Shot at PM

If you could find out in your 20’s or early 30’s that you were at risk of developing Alzheimer’s disease—the devastating, progressive disease that affects one in nine Americans age 65 and older and for which there is no cure—would you want to know?

The results of a new study take a small step toward identifying otherwise healthy individuals who may be at high risk of Alzheimer’s disease in the future. The test purportedly works at a very young age: 18 and 35.

In a study, published in the journal Neurology, researchers surveyed entire genomes using data from the International Genomics of Alzheimer’s Project to calculate risk scores based on many genes in individuals for Alzheimer’s disease. They looked to see if higher risk scores would be associated with known indicators of Alzheimer’s. They found that, even in healthy, young people (though just barely), there was a link—finding significant associations between the risk scores and known indicators of Alzheimer’s disease. For young people, a higher risk score indicated a likely decreased volume of the hippocampus, the area of the brain associated with memory. The participants were otherwise symptomless, showing no signs of dementia.

Would you want to know?


It’s a very personal question, one which some people have already confronted. Those with early-onset familial Alzheimer’s struggle with their decision whether to take a blood test for the disease, less common but more predictable than late-onset. Many others may have faced it when using the genetic testing services like 23andMe.

Before the FDA regulatory crackdown, 23andMe supplied its U.S. customers with a risk rating for Alzheimer’s disease. But did people really want that information when there isn’t yet a cure for Alzheimer’s? 23andMe didn’t know the answer, so they required an extra confirmation step to reveal the sensitive risk status.

The methods of predicting risk by 23andMe, however, differed from this latest study. The risk 23andMe’s test was primarily based on just two known variants in one gene, the strongly-linked Alzheimer’s gene apolipoprotein E (APOE). In this study, researchers used thousands of genetic variants to calculate the risk of Alzheimer’s disease and were able to show significant associations of that risk with known indicators of the disease.

“The stage of Alzheimer’s before symptoms show up is thought to last over a decade,” said corresponding author Elizabeth Mormino to the American Academy of Neurology in a press release. She suggested that identifying early indicators of the disease before symptoms appear will be critical for treatment. 


So while identifying early indicators of Alzheimer’s disease may not be comforting, it may become important for early prevention and treatment before symptoms develop.

How significant is significance?

Two scientists not involved in the study spoke with the Genetic Experts News Service (GENeS) about the significance of the paper. Dimitrios Avramopoulos, a professor at the Johns Hopkins Institute of Genetic Medicine, thought that the findings were significant in a statistical sense, but their impact may not yet be clinically significant:

The statistical support is strong however, the polygenic score only explains a small fraction of the variation in these pre-Alzheimer’s disease marker measurements, meaning it is still far from being useful for prediction. Nevertheless it is an important first attempt.

In older populations, the polygenic risk scores were significantly associated with memory decline over time and a higher risk for clinical progression of the disease. Yet it only explained 3.2 percent of the difference in memory performance between those with high and low risk scores. And in young people, the risk scores were significant by a hair and only explained 0.2 percent of the difference in hippocampal volume.


Dr. Sudha Seshadri, of Boston University School of Medicine, agreed with Avramopoulos. 

“The effect of polygenic risk scores was statistically significant but small,” Dr. Seshadri said. “Hence such polygenic risk scores will help in ‘risk stratification’- for example identifying a higher risk sample for enrollment in a clinical trial. It will not be particularly useful for risk prediction.”

This idea of risk stratification is an important concept for the complex genetic diseases like Alzheimer’s disease. While, genetic risk scores may not be definitive, they are likely to identify people at higher risk for Alzheimer’s disease.

Seshadri also highlighted the importance of using more than just the known variants to assess risk, pointing out an interesting finding in the paper:


Interestingly, a genetic risk score restricted to only 18 variants known to be associated with dementia risk was not associated with the Alzheimer’s markers. However, it was found that the polygenic risk scores, based on thousands of variants, do affect Alzheimer’s markers. This result suggests that there is still a lot of undetected genetic variation underlying the risk of Alzheimer’s disease.

Bigger is better

Part of the limitations of a study like this are the relatively small sample size. The study used only 166 participants with Alzheimer’s disease to compare with just 1,322 in the young group. This fact isn’t lost on the authors. Mormino admitted that the study was small and that larger sample sizes would be needed to verify the findings.

Avramopoulos agreed, adding that larger studies will improve its predictive power. Despite the size limitations, the current results manage to paint a new picture of Alzheimer’s disease and the importance of early identification:

These data support the notion that the processes leading to Alzheimer’s disease start early and progress for a long time before leading to a clinically detectable disease. This underlines the importance of accurate predictions, which can allow early lifestyle modifications or treatments. Genetic risk scores will certainly have a place in these predictions.

So while the predictive power of these risk scores may be limited at the moment, they will could improve as studies get larger. This improvement may allow for more accurately answering the question of whether someone is at high risk of developing Alzheimer’s. And identifying high risk candidates may open to door to early treatments or prevention methods, making it an answer worth knowing.


Mikel Shybut is a CLEAR Fellow at the Genetic Expert News Service. He has a Ph.D. in Plant Biology from the University of California, Berkeley. Follow him on Twitter @MShybut

Outbreak Daily Digest
Biotech Facts & Fallacies
GLP Podcasts
Infographic: Deaths from COVID-19 are far higher than reported estimates

Infographic: Deaths from COVID-19 are far higher than reported estimates

More than 2.8 million people have lost their lives due to the pandemic, according to a Wall Street Journal analysis ...
News on human & agricultural genetics and biotechnology delivered to your inbox.
glp menu logo outlined

Newsletter Subscription

Optional. Mail on special occasions.
Send this to a friend