medwireNews: Research published in Nature Medicine shows how the specific characteristics of people with type 2 diabetes could one day guide disease management.
“Current clinical guidelines for the management of [type 2 diabetes] generally do not consider individual patient phenotype when considering what is the optimal treatment or what are the risks of progression to insulin or microvascular disease,” say the researchers.
They believe their findings “support the concept” that people’s individual phenotype should inform management decisions such as frequency of screening for complications and which medications to prescribe.
“The incorporation of individual phenotypic variation into clinical practice has clear potential to make a substantial contribution to a precision approach to the management of [type 2 diabetes],” they add.
For the study, Ewan Pearson (University of Dundee, UK) and colleagues used healthcare data of 23,137 people from Scotland, diagnosed with type 2 diabetes between 1993 and 2017. They validated their findings using additional data from the UKBB primary care database (n=7332) and from the randomized ADOPT trial (n=4150).
The team analyzed the data using a statistical algorithm called DDRTree, to group people into clusters based on nine phenotypic characteristics while still preserving the continuum seen in type 2 diabetes. The data are visualized as a tree-like structure with roots and branches – people toward the end of branches have more extreme characteristics with those toward or on the trunk having milder/mixed characteristics. All analyses were adjusted for age and sex.
Of the characteristics assessed, high-density lipoprotein (HDL) cholesterol, and systolic and diastolic blood pressure were very widely distributed across the tree at the point of diabetes diagnosis, while total cholesterol and triglyceride levels were a little less variable, followed by glycated hemoglobin (HbA1c) and BMI. Creatinine and alanine aminotransferase, on the other hand, had a narrow distribution.
People who had obesity and hyperglycemia with high triglyceride levels and low HDL cholesterol were clustered in one section of the tree, and these individuals were also those most at risk for the outcomes of insulin initiation, major adverse cardiovascular events, and chronic kidney disease, with death included in the analysis as a competing risk.
Moreover, the addition of medication data from ADOPT revealed that people in this cluster were also at the highest risk for treatment failure on metformin or sulfonylureas, and genetic data from 10,657 people revealed an increased genetic risk for obesity within this cluster.
In contrast with the cardiorenal outcomes, retinopathy outcomes were concentrated in a cluster of people who had high blood pressure, “moderately elevated” HbA1c levels, and dyslipidemia.
Finally, the researchers identified a cluster of individuals at increased risk for failure of thiazolidinedione (TZD) treatment. These people also had an increased genetic risk for lipodystrophy but did not have high genetic obesity risk, and the researchers suggest their increased risk for TZD failure “might be explained by the mechanism of action of TZDs to increase subcutaneous fat mass.”
Pearson and team applied the genetic data in the form of process-specific partitioned polygenic scores (pPSs), which are based on clusters of genetic risk variants linked to underlying etiologic processes. This also uncovered a group of people with increased genetic risk for beta-cell dysfunction and an increased risk for diabetes “mediated via liver and lipid-mediated insulin resistance.”
The researchers believe the associations of the pPSs with phenotypic characteristics and outcomes indicate that causal etiological processes mediate some of the variability between individuals with type 2 diabetes.
However, they stress: “We are not advocating the use of DDRTree to improve prediction but, rather, to reduce a complex multi-dimensional disease into a simpler, understandable two-dimensional model that can be readily visualized and used to enhance the therapeutic process between clinicians and individual patients, to see how their personal [type 2 diabetes] profile compares to others of similar age and sex.”
medwireNews is an independent medical news service provided by Springer Healthcare Ltd. © 2022 Springer Healthcare Ltd, part of the Springer Nature Group