Glucose Complexity Estimates Insulin Resistance in Either Nondiabetic Individuals or in Type 1 Diabetes

Context: The clinical significance of glucose complexity is not fully understood. Objective: We investigated the relationship between glucose complexity, glucose variability (GV), and insulin resistance. Design and Setting: This was an observational study of 37 nondiabetic volunteers age 12-65 years and 49 adults with longstanding type 1 diabetes (T1D) who wore a blinded continuous glucose monitoring (CGM) device under real-life conditions. Sample entropy (SpEn) and detrended fluctuation analysis (DFA) were used to determine complexity either from the unaltered CGM or after removing fast glucose oscillations by using digital filters. Results: SpEn was inversely correlated with insulin resistance (ie, homeostasis model assessment of insulin resistance), the body mass index (BMI) z score (BMI-Z), GV, and DFA in nondiabetic subjects (NDs), and with BMI-Z, GV, and DFA in patients with T1D. T1D was characterized by a decomplexification of CGM profiles. In multivariate analysis, SpEn of NDs but not GV correlated inversely with markers of insulin resistance and SpEn correlated inversely with BMI-Z across both groups (r = -0.46; P < .0001). Low-pass filtering of the CGM data showed that inverse correlations of SpEn with insulin resistance and BMI-Z were dependent on the fast glucose oscillations. Conclusions: Low complexity of CGM profiles is associated with insulin resistance in both NDs and patients with T1D. In NDs, low complexity could be an earlier marker of glucose regulation failure than GV. The relationship between glucose complexity and insulin sensitivity is determined by the richness of fast oscillations in the glucose profiles.