NEW YORK (GenomeWeb) – Researchers have developed a polygenic hazard score that they said does a better job than existing methods of identifying older adults with normal cognition who are at the highest risk of developing Alzheimer’s dementia.
The genetic prediction technique, developed by researchers at the University of California, San Francisco, and the University of California, San Diego, combines the effects of more than 31 genetic variants, most of whom, by themselves, are associated with only a small risk of Alzheimer’s disease, the researchers said.
Their polygenic hazard score, or PHS, does a better job than testing with just APOE E4, a genetic variant long considered the strongest genetic predictor of whether someone is likely to develop Alzheimer’s, Chin Hong Tan, a postdoctoral scholar at UCSF and one of the test's developers, said in an interview.
Last week, the researchers published the results of a study that describes the polygenic hazard score in the Annals of Neurology.
They noted that the trouble with testing for Alzheimer's using APOE E4 alone is that only 10 to 15 percent of the population carries the variant. The team's polygenic hazard score provides risk estimates for the remaining 85 to 90 percent of people who do not carry at least one copy of APOE E4, the researchers said. These people still have some combination of other genetic variants that put them at risk of developing Alzheimer’s disease, Tan said.
He said that the team's polygenic hazard score can identify cognitively normal and mildly impaired older people who are at greatest risk for developing Alzheimer’s-associated clinical decline over time.
As a basis for their analysis, the researchers used genetic data from more than 70,000 people in the National Alzheimer's Coordinating Center database, the International Genomics of Alzheimer’s Disease Project, and the Alzheimer’s Disease Genetics Consortium.
To test out their technique, they analyzed five years of data on 1,081 people who did not have dementia. The data included genetic test results for all the population and autopsy results for a fraction of the population. They found that the PHS test could predict how long it would take for them to progress to Alzheimer’s dementia, and how steep their cognitive decline would be.
Autopsies of people whose data were recorded in the NACC database and who developed Alzheimer’s showed that a higher polygenic hazard score was associated with a higher level in the brain of amyloid plaque — a protein aggregate that is a hallmark of Alzheimer’s. The higher levels were present even among those who did not carry a copy of the APOE E4 variant, the researchers said.
Patients with higher polygenic hazard scores also showed steeper declines on cognitive tests during their lifetimes, the researchers said.
Using several genetic variants in an analysis takes on even great importance considering recent research that suggests the effects of APOE E4 have been overstated, Tan said.
For example, the risk of dementia in those with APOE E4/E4 is somewhat lower than previously estimated, according to a study published in March in PLOS Medicine by researchers from several organizations and countries.
"The risk that an individual with APOE E4/E4 will develop Alzheimer disease dementia has been reported to be as high as 50 percent to 67 percent, but these estimates come from statistical modeling, not direct observation," the researchers said in the paper.
In their study, they found that the five-year risk for APOE E4/E4 individuals ranged from 0 percent to 23 percent in those entering the study at age 60 to 64 years; 9 percent to 35 percent in those entering at 65 to 69 years; and 19 percent to 38 percent in those entering at 70 to 75 years.
They reported that lifetime risk did not vary as much with age and ranged from 31 percent to 40 percent for those with APOE E4/E4.
Tan noted that he is aware that direct-to-consumer genetic test companies are providing APOE E4 information with their services, and added "We also know from other studies and research, that this information does not fully inform individuals about their risk of contracting Alzheimer's disease."
“Our findings have strong implications for…direct-to-consumer genetic tests, some of which have recently received FDA clearance,” Anders Dale, a professor of neurosciences and radiology at UC San Diego and coauthor of the study, said in a statement.
In the future, their PHS analysis could be made available in primary care settings and it could be offered as a direct-to-consumer test, Tan said, but added that achieving US Food and Drug Administration clearance and launching PHS in that manner are "far off and much further down the road."
In marketing a direct-to-consumer risk test, Tan said, it would be important to help individuals understand what the PHS analysis means. People would need to understand that it is not a diagnostic test, but a risk score, he said. It would be important, he added, to not interpret a high-risk score as having already developed Alzheimer's.
"On the other hand, if you know that you have a high score for PHS, it can help you better prepare for the future, such as doing financial planning or changing your lifestyle, diet, and exercise" to slow its potential onset and progression, he noted.
The harms and benefits of giving people access to their genetic tests results is an ongoing debate.
23andMe began selling genetic health risk tests directly to consumers in 2007 and then stopped doing so in 2013, after receiving a warning letter from the FDA. However, in April, the FDA granted premarket authorization for 10 genetic health risk reports, including one for Alzheimer's, from 23andMe.
The agency touted these as the first genetic tests for gauging disease risk that it has authorized for direct-to-consumer access — meaning that the consumer doesn't need a doctor's prescription to receive testing.
Scott Hadly, a spokesperson for 23andMe, said that its late-onset Alzheimer's risk report "looks at just the E4 variant in the APOE gene." The firm is familiar with the UCSF-UCSD study, he noted, and added that 23andMe has "done a lot of work with polygenic risk modeling," but not for Alzheimer's risk.
Last year, the firm collaborated on a study published in PLOS Medicine that used polygenic modeling to test the contribution of genetic factors and other influences on the risk of chronic pain and its correlation with major depressive disorder.
Separately, in a poster presented at ASHG 2016 in Vancouver, 23andMe scientist Nick Furlotte showed how he used polygenic risk models as the basis of a study that calculated a body-mass-index genetic risk score and its application to a 23andMe report that provides customers with information about their genetic weight dispositions.
The genetic models used more than 300 single nucleotide polymorphisms to predict BMI based on data from more than 600,000 research participants.
In polygenic risk scores that have been developed by other groups, researchers have compared test results for patients with Alzheimer's disease against results for healthy controls to ascertain genetic variants that are represented in Alzheimer's disease cases, Tan said. However, these tests have not considered the age of onset of Alzheimer's disease, which is the highest risk factor, he said.
In the study published in PLOS Medicine, the researchers noted that with the onset of prevention trials for individuals at high risk for Alzheimer's disease, "there is increasing need for accurate risk prediction to inform study design and enrollment, but available risk estimates are limited."
UCSD's Dale noted that aside from direct-to-consumer testing, their findings also have important implications for disease stratification and secondary prevention Alzheimer’s trials.