Current diabetes risk assessments—including the American Diabetes Association’s 60-second risk test for Type 2 diabetes—typically take only a handful of basic, surface-level information into account: age, gender, race, family history and activity levels, among others.
But a new take on these tests that delves a little deeper under the skin—literally—could have a significant improvement on predictions that test-takers will develop Type 2 diabetes within the next 10 years.
The test was developed by researchers at the University of Edinburgh, who published a study Thursday in the journal Nature Aging that describes how taking DNA methylation into account can augment current diabetes risk-scoring systems.
Currently being explored in cancer testing, DNA methylation is a process where molecules called methyl groups are added to the structure of DNA molecules. Once a methyl group has fused to a DNA sequence, it may alter the function of the associated genes, often by stopping the gene from carrying out its usual task of creating RNA copies that are used to build new proteins.
In developing their risk prediction tool, the Scottish researchers factored in a methylation analysis alongside the basic health data used in standard diabetes risk scores to create a handful of potential models. They started by training the models on a cohort of more than 9,800 people already enrolled in the national Generation Scotland study, 374 of whom had Type 2 diabetes.
The tools were then put to the test on another group of Generation Scotland participants: this time, a total of 4,778 individuals, including 252 with diabetes. The best-performing of the researchers’ models achieved an area-under-the-curve of 0.872, which they described as a “notable improvement” over the 10-year risk prediction abilities of standard, methylation-free assessments, which registered an AUC of 0.839.
The model achieved similar results when applied to a separate cohort of nearly 1,600 participants in a German study, 142 of whom had Type 2 diabetes.
The researchers went on to extrapolate out their model’s results. When applied to a hypothetical group of 10,000 people in which about a third go on to develop diabetes within the ensuing decade, they concluded, the risk test would be able to correctly identify 449 more high-risk individuals than the standard assessments.
In addition to giving high-risk individuals the chance to take preventive measures before the onset of diabetes, catching early warning signs of the condition can also improve outcomes on a broader scale, according to Yipeng Cheng, an author of the study and a Ph.D. student from the University of Edinburgh’s Centre for Genomic and Experimental Medicine, who noted in a university release, “Delaying onset is important as diabetes is a risk factor for other common diseases, including dementias.”
Additionally, the model could ultimately be used as a risk-scoring tool for other conditions beyond Type 2 diabetes, according to the researchers.
“Similar approaches could be taken for other common diseases to generate broad health predictors from a single blood or saliva sample,” Riccardo Marioni, Ph.D., principal investigator of the study, said in the release. “We are incredibly grateful for our study volunteers who make this research possible—the more people that join our study, the more precisely we can identify signals that will help delay or reduce the onset of diseases as we age.”