Algorithm uses genetic markers to predict diabetic kidney disease

Kidney disease is a common, irreversible complication of diabetes. Researchers have developed an algorithm that uses genetic markers to predict whether a type 2 diabetic will develop kidney disease years in advance, which may provide a way of diagnosing and treating this preventable condition early.

According to the World Health Organization, the number of people globally with type 2 diabetes rose from 108 million in 1980 to 422 million in 2014. A common complication of diabetes is kidney disease, also known as diabetic nephropathy.

Diabetes is characterized by elevated blood sugar levels. Over time, high blood sugars damage the kidney’s tiny filtering units, which causes them not to work as effectively to remove waste from the blood and return the cleaned blood back into circulation. The damage is irreversible and eventually leads to kidney failure, requiring treatment with dialysis or a kidney transplant. In the US, 44% of end-stage kidney disease and dialysis cases are caused by diabetes.

Researchers from the Chinese University of Hong Kong have paired up with Sanford Burnham Prebys to develop a computational model that can predict whether a person with type 2 diabetes will develop kidney disease.

“There has been significant progress developing treatments for kidney disease in people with diabetes,” said Ronald Ma, corresponding author of the study. “However, it can be difficult to assess an individual patient’s risk for developing kidney disease based on clinical factors alone, so determining who is at greatest risk of developing diabetic kidney disease is an important clinical need.”

To address this clinical need, the researchers turned to DNA methylation, a biological process where methyl groups are added to the DNA molecule. The process is one of the ways cells can control which genes are active and at what time and is easily measured via a blood test.

DNA methylation is an inheritable (epigenetic) alteration associated with cancer and other diseases, such as cardiovascular disease. There have been previous attempts to identify a biomarker that can predict diabetic kidney disease. While genome-wide association studies (GWAS) have had some success in identifying the genetic markers for type 2 diabetes, it’s thought that epigenetic markers like methylation provide a way of capturing the interaction between genetics and environmental factors.

The researchers used DNA methylation as a marker to teach their computational model to predict kidney function, using data from 1,271 patients with type 2 diabetes in the Hong Kong Diabetes Register. They also tested the model on a separate group of 326 Native Americans with type 2 diabetes, ensuring that the model could predict kidney disease in different populations.

The computational model could predict kidney function well, achieving a mean area under the receiver-operator characteristic (AUROC) of 0.76 that wasn’t influenced by factors such as sex or age. AUROC is a performance metric that evaluates how well a machine learning model discriminates between cases (positive examples) and non-cases (negative examples). An AUROC of 1.0 corresponds to a perfect classifier, whereas an AUROC of 0.5 is the equivalent of a coin toss. An AUROC of 0.7 to 0.8 indicates good performance.

“Our computational model can use methylation markers from a blood sample to predict both current kidney function and how the kidneys will function years in the future, which means it could be easily implemented alongside current methods for evaluating a patient’s risk for kidney disease,” said Kevin Yip, co-corresponding author of the study.

Given the worldwide prevalence of type 2 diabetes and that diabetic kidney disease is preventable, this study’s findings are important. They may lead to a method of diagnosing and treating the disease early.

The researchers are refining their model and plan to expand it to incorporate other data that might enhance its ability to predict other diabetes-related health outcomes.

“We are delighted that the findings of this study could improve future care and make it easier to determine who will benefit most from these new treatments to prevent kidney damage from diabetes,” Ma said. “The science is still evolving, but we are working on incorporating additional information into our model to further empower precision diabetes medicine.”

The study was published in the journal Nature Communications.

Source: Chinese University of Hong Kong and Sanford Burnham Prebys

Source of Article