University of Edinburgh scientists have found that analyzing changes to DNA in a blood sample can help predict a person’s risk of developing type 2 diabetes within a decade.
Their study, published Thursday in the journal Nature Aging, could lead to earlier preventative measures, reducing diabetes’ economic and health burden, the researchers said.
Type 2 diabetes happens when a person’s pancreas cannot make enough insulin, or the insulin doesn’t function properly. This can lead to high blood sugar levels and, in turn, serious conditions including heart disease, stroke, nerve damage, and foot issues. Given the risk of diabetes leading to such diseases, delaying its onset is important.
Current risk prediction tools for type 2 diabetes consist of information that includes age, sex, body mass index, and family history of the disease.
The researchers found that measuring certain changes to the DNA in a blood sample, alongside these other risk factors, provided a more accurate prediction of the likelihood of developing the condition years in advance of any symptoms.
The DNA changes, also called DNA methylation, reflect a chemical process in the body in which a small molecule called a methyl group is added to DNA. The addition or removal of methyl groups can affect how molecules act in the body, and methylation patterns can help to track aging processes and disease development.
The scientists estimated predictive performance using a hypothetical screening scenario of 10,000 people, where one in three individuals develop type 2 diabetes over a 10-year period. The model that used DNA methylation correctly classed an extra 449 individuals compared with traditional risk factors alone.
Real-world data came from 14,613 participants in a large study called Generation Scotland. The team repeated the analyses with 1,451 individuals from the German-based study KORA (Cooperative Health Research in the Region Augsburg). Their approach was based on a range of linear and tree-ensemble models that incorporated time-to-event data for prediction. In the Generation Scotland cohort, their best-performing model showed notable improvement in 10-year onset prediction beyond standard risk factors. Replication was observed in the KORA study, ensuring that the findings could be replicated in people from different backgrounds.
The researchers concluded that their findings demonstrate the potential for DNA data to provide notable improvement in predictive performance for type 2 diabetes above and beyond commonly used risk factors. After systematically evaluating different models, they presented a framework with the ability to generalize to other traits and datasets for future training and testing predictors.
“Similar approaches could be taken for other common diseases to generate broad health predictors from a single blood or saliva sample,” University of Edinburgh principal investigator Riccardo Marioni said in a statement. “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.”
Generation Scotland is currently recruiting volunteers. This study is designed to help scientists investigate the causes of disease and inform healthcare priorities and policies. Anyone residing in Scotland can sign up at www.generationscotland.org.
