Identifying Clinical Phenotypes of Type 1 Diabetes for the Co-Optimization of Weight and Glycemic Control

Public Deposited
Analytics

Downloadable Content

Download PDF
Request Version for Screen Reader
Last Modified
  • July 24, 2019
Creator
  • Kahkoska, Anna Rachel
    • Gillings School of Global Public Health, Department of Nutrition
Abstract
  • Obesity is an increasing concern in the clinical care of youth with type 1 diabetes (T1D). Standard approaches to co-optimize weight and glycemic control are challenged by profound population-level heterogeneity. Therefore, the goal of the dissertation was to apply novel analytic methods to understand heterogeneity in the co-occurrence of weight, glycemia, and underlying patterns of minute-to-minute dysglycemia among youth with T1D. Data from the SEARCH for Diabetes in Youth study were used to characterize subgroups of youth with T1D showing similar weight status and level of glycemic control as distinct ‘weight-glycemia phenotypes’ of T1D. Cross-sectional weight-glycemia phenotypes were identified at the 5+ year follow-up visit (n=1,817) using hierarchical clustering on five measures summarizing the joint distribution of body mass index z-score (BMIz) and hemoglobin A1c (HbA1c), generated by reinforcement learning tree predictions. Longitudinal weight-glycemia phenotypes spanning eight years were identified with longitudinal k-means clustering using baseline and follow-up BMIz and HbA1c measures (n=570). Logistic regression modeling tested for differences in the emergence of early/subclinical diabetes complications across subgroups. Seven-day blinded continuous glucose monitoring (CGM) data from baseline of the Flexible Lifestyles Empowering Change randomized trial (n=234, 13-16 years, HbA1c 8-13%) was clustered with a neural network approach to identify subgroups of adolescents with T1D and elevated HbA1c sharing patterns in their CGM data as ‘dysglycemia phenotypes.’ We identified six cross-sectional weight-glycemia phenotypes, including four normal-weight, one overweight, and one subgroup with obesity. Subgroups showed striking differences in other sociodemographic and clinical characteristics suggesting underlying health inequity. We identified four longitudinal weight-glycemia phenotypes associated with different patterns of early/subclinical complications, providing evidence that exposure to co-occurring obesity and worsening glycemic control may accelerate the development and increase the burden of co-morbid complications. We identified three dysglycemia phenotypes with significantly different patterns in hypoglycemia, hyperglycemia, glycemic variability, and 18-month changes in HbA1c. Patient-level drivers of the dysglycemia phenotypes appear to be different from risk factors for poor glycemic control as measured by HbA1c. These studies provide pragmatic, clinically-relevant examples of how novel statistics may be applied to data from T1D to derive patient subgroups for tailored interventions to improve weight alongside glycemic control.
Date of publication
Keyword
DOI
Resource type
Advisor
  • Aiello, Allison E
  • Mayer-Davis, Elizabeth J
  • Adair, Linda S
  • Kosorok, Michael R
  • Buse, John B
  • Burger, Kyle S
Degree
  • Doctor of Philosophy
Degree granting institution
  • University of North Carolina at Chapel Hill Graduate School
Graduation year
  • 2019

Relations

Parents:

This work has no parents.

Items

Thumbnail Title Date Uploaded Visibility Actions
file details Kahkoska_unc_0153D_18315.pdf 2019-07-24 Public Download