Presenter of 1 Presentation
PRINCIPAL DIMENSIONS OF GLYCEMIC VARIABILITY AND QUALITY OF GLYCEMIC CONTROL IN DIABETES
Abstract
Background and Aims
Many of the available metrics to quantify glycemic variability (GV) and quality of glycemic control (QGC) derived from continuous glucose monitoring (CGM) data, are highly correlated. The aim of this work is to identify the principal uncorrelated dimensions of GV and QGC, to be considered in the assessment of diabetes management.
Methods
Six widely-used metrics were evaluated on CGM traces generated by 782 participants in 6 studies in type 1 and type 2 diabetes (T1D, T2D): mean blood glucose (MBG); percent time >180 mg/dL (T180), >250 mg/dL (T250), <70 mg/dL (T70), <54 mg/dL (T54); coefficient of variation (CV). Principal component analysis (PCA) was used to identify two principal uncorrelated dimensions of GV and QGC. These principal dimensions were first identified in a training set (550 subjects) and then fixed and validated in an independent test set (232 subjects).
Results
PCA identified two principal dimensions explaining >90% of the original variance in the testing data, irrespective of treatment modality, age range, and diabetes type. These dimensions represent exposure to hyperglycemia, or therapy efficacy, as indicated by a combination of MBG, T180, and T250 (Dimension 1), and risk for hypoglycemia, or therapy safety, as indicated by T70, T54, and CV (Dimension 2). A graphical representation of the two dimensions is shown in the figure.
Conclusions
Two uncorrelated dimensions are sufficient to characterize GV and QGC in diabetes, and to explain over 90% of the variance carried by common metrics. Thus, quantitatively, treatment optimization is reduced to a 2-dimensional problem.