Can Diabetes Classification Be Fine-Tuned?
Researchers identified 5 subgroups of diabetes using genomic data.
NEW ORLEANS — Researchers may have identified the first step toward precision medicine in the field of diabetes. According to data presented at the American Diabetes Association (ADA) 76th Scientific Sessions, it may now be possible to combine genetic and non-genetic markers and classify patients with diabetes into 5 subgroups, stratifying disease progression.
“This is by far the largest and most comprehensive study trying to reclassify diabetes and move away from only type 1 diabetes and type 2 diabetes. In total, close to 20,000 patients were included in either the primary or replication cohort,” said study investigator Petter Storm, PhD, researcher at Lund University Diabetes Centre, Malmö, Sweden.
He and his colleague contend that the current diagnosis and classification of diabetes are inaccurate, outdated, and are not useful in predicting disease outcome or guiding therapy. However, genomic studies may be game changers.
Dr Storm and his colleagues have set up the ANDIS project (All New Diabetics in Scania) in Southern Sweden, which combines information on genetic and non-genetic markers. The database includes patient records from 10,785 newly diagnosed diabetes patients aged 0 to 97.
The researchers performed a cluster analysis to identify new subgroups of diabetes. To validate their findings, a similar registry from Western Finland, DIREVA (Diabetes Registry Vaasa), with 5107 patients aged 0 to 94 at diagnosis, was used for replication. In addition, a second Swedish cohort of approximately 4000 with longer follow-up time (Scania Diabetes Registry) was used to further validate association with complications of diabetes.
The cluster analysis included information on age at onset, BMI, HbA1c, insulin secretion and action, and GADA autoantibodies. It was able to identify 5 subgroups of diabetes, which predicted change in HbA1c over time as well as development of early complications of diabetes.
“Every endocrinologist knows that the terms type 1 and type 2 do not capture the full spectrum of diabetes,” Dr Storm told Endocrinology Advisor. “We present the first attempt to more fine-grained diabetes classification, predicting disease progression. By reclassifying diabetes into 5 new subgroups we can better guide clinicians in treatment and risk for complications.”
He said the 5 categories had distinct genetic profiles and some showed changes in metabolites. They were presented as: cluster 1 (11%) comprised mostly of type 1 diabetes; cluster 2 (20%) included patients who showed impaired beta-cell function of non-autoimmune background; cluster 3 (6%) included the most insulin-resistant patients with the highest risk for kidney disease; cluster 4 (20%) included the most obese patients; and cluster 5 (43%) was age-related and most type 2 diabetes-like with similar frequencies seen in the replication cohort.
“Actually, the findings made a lot of sense when we got the first result and we were very happy to achieve them using a data-driven approach,” said Dr Storm. “Taken together, the study enables the development of a road map for the diabetes patient, paving the way for early intensified treatment and thereby a way to possibly prevent late diabetic complications ascribed to the ‘metabolic memory.' It may also help the pharmaceutical industry to better stratify patients for drug trials and thereby reduce costs for development of new drugs.”
Disclosures: The researchers report no conflicts of interest.
- Storm P, Rosengren A, Groop L. Abstract 359-OR. A Novel Fine-Tuned Classification of Diabetes with Prognostic Value: Steps towards Precision Medicine. Presented at: ADA 76th Scientific Sessions; June 10-14, 2016; New Orleans, LA.