NEW YORK (GenomeWeb) – Patients with type 2 diabetes, based on their genetic variants, can be grouped into five clusters that reflect different disease mechanisms and could help personalize treatments in the future, according to a new study.
About 30.3 million people in the US have diabetes, and type 2 diabetes accounts for between 90 and 95 percent of those cases, according to the Centers for Disease Control and Prevention. Diabetes is a heterogeneous disease and can be broadly grouped into whether the condition stems from insulin deficiency or insulin resistance.
An international team of researchers sought to further classify type 2 diabetes patients based on the risk loci they inherited. As they reported today in PLOS Medicine, Massachusetts General Hospital endocrinologist Jose Florez and his colleagues uncovered five clusters from known diabetes risk loci and related traits. Two clusters were linked to insulin deficiency-related mechanisms and three were linked to insulin resistance mechanisms. These groupings, the researchers said, could help tailor patients' treatments.
"When treating type 2 diabetes, we have a dozen or so medications we can use, but after you start someone on the standard algorithm, it's primarily trial and error," said Florez, a professor at Harvard Medical School, in a statement. "We need a more granular approach that addresses the many different molecular processes leading to high blood sugar."
He and his colleagues applied a soft-clustering method — one that allows a variant to fall into more than one cluster — to 94 independent diabetes-link genetic variants and 47 traits derived from genome-wide association studies. This approach, first author and MGH endocrinologist Miriam Udler noted in a statement, is better suited for complex diseases as gene variants may influence multiple genes and processes.
This analysis identified five clusters: two marked by mechanisms of pancreatic beta cell dysfunction and three by mechanisms of insulin resistance.
One cluster — the beta cell cluster — was associated with traits like decreased corrected insulin response and decreased insulin secretion following a glucose tolerance test, but increased proinsulin levels. It encompassed known beta cell-related loci like MTNR1B, HHEX, TCF7L2, SLC30A8, HNF1A, and HNF1B.
Cluster 2 — the proinsulin cluster — was also associated with decreased insulin secretion following a glucose tolerance test, but also with decreased proinsulin levels, among other traits, and included loci at ARAP1 and SPRY2.
Meanwhile, cluster 3 — the obesity cluster — was linked to increased waist circumference, BMI, and body fat percentage and included the obesity-linked FTO and MC4R loci. Cluster 4 — the lipodystrophy cluster — was associated with a decreased score on the insulin sensitivity index and decreased HDL cholesterol, but with increased triglycerides. And cluster 5 — the liver/lipid cluster — was linked to decreased triglyceride, palmitoleic acid, and urate levels, among others, and with increased gamma-linolenic acid, and encompassed loci previously linked to nonalcoholic fatty liver disease.
To support these findings, the researchers used data from the US National Institutes of Health's Roadmap Epigenomics Project and found the genes implicated by these clusters were active in the expected tissues.
Florez and his colleagues also generated genetic risk scores for each of these clusters based on the top-weighted loci associated with each group.
These risk scores, the researchers found, were associated with a number of clinical outcomes. For instance, high beta cell cluster risk scores were linked to increased risk for ischemic stroke, and increased lipodystrophy risk scores were linked to increased blood pressure, but decreased BMI.
They applied these risk scores to individuals with type 2 diabetes from four cohorts, totaling more than 17,000 people. As expected, increased risk scores for each of the clusters was associated with traits that defined those clusters. Almost a third of patients scored highly for just one cluster.
This suggested to the researchers that these genetically determined clusters could identify patients with particular disease characteristics.
"The clusters from our study seem to recapitulate what we observe in clinical practice," Florez said. "Now we need to determine whether these clusters translate to differences in disease progression, complications, and response to treatment."