Genetic Variant Score May Flag Dementia Risk

Pauline Anderson

July 06, 2016

A genetic risk score may detect patients, even young and healthy ones, who are at higher risk for Alzheimer's disease (AD), new research shows.

"The development of Alzheimer's disease probably lasts decades before a patient actually has clinical symptoms of dementia," said lead study author Elizabeth C. Mormino, PhD, Department of Neurology, Massachusetts General Hospital, and an instructor at Harvard Medical School, Boston.

"We were actually able to see an impact of the genetic risk before symptoms are even present."

Dr Elizabeth C. Mormino

The study was published online July 6 in Neurology.

Potential Selection Tool for Clinical Trials

With some refinements (such as isolating the most important genetic variants and boosting the statistical signal), this polygenic risk score may eventually be used to select patients for clinical trials, said Dr Mormino.

"If it did have some good predictive value of who is going to be at risk within a few years, it could be used in the context of enrollment of clinical trials. And once there is a treatment, then it would be used for completely different reason — to decide who should get these drugs."

Dr Mormino and colleagues developed a polygenic risk score (PGRS) based on the results of the recent large meta-analysis of AD dementia from the International Genomics of Alzheimer's Project (IGAP).

Researchers for that project determined which of 7 million different genetic variants are more frequent in patients with AD than in those without the disease, and the amount of risk associated with each loci.

The investigators sought to determine whether these genetic risks were associated with AD markers during the stages preceding dementia.

"We were specifically interested in whether or not this aggregate measure of risk of AD is related to early changes, before you actually have dementia," said Dr Mormino.

From the IGAP study, they developed both a conservative and a liberal risk score. The conservative score took only the top 20 genetic loci, those most associated with AD.

"We actually didn't see effects when we only looked at the top 20," she added.

The researchers then lowered the threshold to create a more liberal score that included 16,000 of the loci.

They arrived at that number during a side analysis that started with the top 20 variants and added to it.

"Once we started to increase the number, the difference got stronger and stronger, and it kind of reached a plateau at around 16,000 and so we stopped there."

"Exciting" Findings

The study included older clinically normal patients, those with mild cognitive impairment (MCI), and those with AD dementia from the Alzheimer's Disease Neuroimaging Initiative (ADNI) study. The mean age of the older participants without dementia was 75 years, and those with dementia were slightly older. The study also included cognitively normal volunteer participants from the Brain Genomics Superstruct Project who were 18 to 35 years old.

All study participants had genotyping data available and were of European ancestry.

The investigators found that among 526 older participants without dementia — including those with normal cognition and those with MCI — the PGRS was associated with multiple AD markers. With use of the more liberal score, higher PGRS was significantly associated with worse memory (P = .002), and smaller hippocampus volume (P = .002).

The genetic score was also associated with AD-like levels of amyloid β (Aβ) at baseline. In a subset of 272 ADNI participants without dementia who had cerebral spinal fluid (CSF) data available, higher PGRS was associated with lower (or more AD-like) CSF Aβ levels.

Although this was not statistically significant, an analysis of a larger sample of 505 persons without dementia who underwent florbetapir positron emission tomography found that elevated PGRS was associated with significantly greater (more AD-like) levels of florbetapir Aβ (P = .03).

Over the course of 3 years, higher PGRS was associated with greater longitudinal change in both memory (P = .0005) and executive function (P = .01), as well as elevated risk for clinical progression to MCI or AD (P < .00001).

Of the 194 participants who were cognitively normal at baseline, 15 developed MCI or AD, and 143 of 332 with MCI at baseline developed AD after 3 years. Each SD increase in polygenic risk was associated with a 1.6 times increase in risk for clinical progression. 

"So in older individuals who are at elevated risk for AD, we do see that having higher genetic risk is associated with more abnormal levels of β amyloid, which is thought to be a really important pathology for AD, but also smaller hippocampus and a decline in cognition, which we know are markers of early AD," said Dr Mormino. "That's exciting."

Major Limitation

She acknowledged that combining the patients with MCI with those who did not have dementia, which was done to create statistical power, was a "major limitation" of the study.

"It would have been great to be able to look just at the 'normals,' and just at the MCI, to understand how early we can actually see this association, but we just didn't have the statistical power."

However, she noted that there was a diagnosis covariate in the analysis "to make sure it wasn't driven by the MCI compared to the normal."

Separating out the "normal" persons is "definitely a future direction," commented Dr Mormino. "We want to know if prediction is different in this population because ultimately, we would want to have something that tells us who is at risk when there's no symptoms, and in MCI there is already impairment."

In a sample of 1322 young, clinically normal young adults, higher PGRS was associated with smaller hippocampus volume (P = .05), suggesting that an effect of aggregate genetic risk is not specific to processes occurring in late life.

"This was kind of surprising because that's an age that is many, many decades before you would expect to see any symptoms of Alzheimer's disease," said Dr Mormino.

"Even though the effect was very small, it does provide hope that maybe at some point, these common genetic risk factors might be useful in tagging people early who are most at risk for this disease."

Dr Marmino said she is "really excited" by the implications of the study results. "It's definitely important to keep in mind that the effect sizes are really small, which is common in these genetic biomarker studies, but I think it's an initial step towards the direction of potentially using these common genetic variants to identify people who are at risk for AD."

This is "an optimal time" for prevention strategies to ultimately stop the disease, "and to do that, we really need to be able to estimate who is at greatest risk," added Dr. Mormino. "This new genetic score provides promise for that, although we definitely need more refinement in these metrics to get to that point."

Amyloid: Not the Be All, End All?

Reached for a comment, Ronald Petersen, MD, PhD, director, Mayo Clinic Alzheimer's Disease Research Center and Mayo Clinic Study of Aging, agreed the study is "a first step" to creating a genetic score to identify those destined to develop AD.

The authors, he said, "are sort of validating" certain polygenic features against the more typical measures, such as amyloid and hippocampal volume.

That the study found an increased score in participants who did not have the typical amyloid biomarkers suggests that there might be other pathways to AD, he said.

"It may imply that amyloid is not the end all, be all," said Dr Petersen. "Most people don't think that it is, but this study does raise the possibility of some other explanations for cognitive impairment later in life."

Dr Petersen believes this research is "a conceptual notion" about genetic tendencies early in life. "I don't think there is a polygenic risk formula coming out of this study that is going to be the answer."

He agreed with the authors that this approach needs to be replicated and that the polygenic risk score needs to be refined. "There can be other things that may contribute to it."

The study was funded by National Institute on Aging grants. Dr Mormino and Dr Petersen have disclosed no relevant financial relationships.

Neurology. Published online July 6, 2016. Abstract

For more Medscape Neurology news, join us on Facebook and Twitter.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: