NEW YORK (GenomeWeb News) – In a study appearing online last night in PLoS ONE, a University of California at Los Angeles research team demonstrated that it's possible to predict a person's age from saliva samples using DNA methylation patterns.
Those involved in the study say the findings could have both forensic applications and medical implications, particularly for finding individuals whose biological age differs from their chronological age.
The researchers used microarrays to look at DNA methylation patterns in spit samples from nearly three-dozen male twin pairs between the ages of 21 and 55 years old, tracking down nearly 90 sites where methylation coincides with age. From follow-up experiments involving another group of men and women between the ages of 18 and 70 years old, the team was able to come up with a model that predicts an individual's age to within an average of five years based on methylation status at two sites in the genome.
"Our approach supplies one answer to the enduring quest for reliable markers of aging," senior author Eric Vilain, a professor of human genetics at the University of California at Los Angeles who is also affiliated with the school's Center for Society and Genetics, said in a statement. "With just a saliva sample, we can accurately predict a person's age without knowing anything else about them."
DNA modifications change with tissue development, differentiation, and age, Vilain and his co-authors explained. For the current study, they looked at whether it was possible to exploit these shifts to find age markers, focusing on methylation at cytosine residues — first in identical twins and then in unrelated individuals from the general population.
"While certain methylation changes are genetically controlled, environmental exposure and stochastic processes can lead to a change in methylation patterns," they explained. "In this context, identical twins can be considered replicates of the same developmental and aging experiment."
Using Illumina HumanMethylation27 arrays, the researchers assessed methylation patterns at nearly 16,100 CpG sites in the genomes of 34 pairs of identical male twins between the ages of 21 and 55 years old.
When they sifted through the data for the twins using an analytical approach known as weighted correlation network analysis, the team found five methylation modules containing loci with comparable methylation patterns.
Within the modules, researchers narrowed in on 88 loci at which cytosine methylation status depended on an individual's age, including 69 showing positive correlation and 19 showing negative correlation. The sites fell in and around 80 different genes, they noted, including several genes that have been implicated in cardiovascular, neurological, and other conditions.
Because methylation at three sites in the promoter regions of the EDARADD, TOM1L1, and NPTX2 genes were particularly well correlated with age, the team used either targeted bisulfite sequencing or Sequenom MassArrays to evaluate CpG methylation profiles for these three genes in DNA from saliva samples from 22 of the twins and from another 31 men and 29 women who were between 18 and 70 years old.
Although methylation status for all three genes corresponded to age in the DNA from the male saliva samples, the researchers reported, methylation profiles for just two of these — EDARADD and TOM1L1 — coincided with age in the females.
Meanwhile, the team's predictive model, based on methylation profiles for cytosine residues near the EDARADD and NPTX2 genes, explained some 73 percent of the age variance and could predict age to within a 5.2 year window based on average.
"[O]ur ability to predict an individual's age to an average accuracy of 5.2 years could be used by forensic scientists to estimate a person's age based on a biological sample alone, once the model has been tested in various biological tissues," the study authors wrote.
And, they say, such predictive models may find favor for diagnosing and treating some age-related diseases and for finding situations in which an individual's biological age differs from their age in years.
"Doctors could predict your medical risk for a particular disease and customize treatment based on your DNA's true biological age, as opposed to how old you are," Vilain said in a statement.