Obstructive Sleep Apnea, Glucose Tolerance, and β-Cell Function in Adults With Prediabetes or Untreated Type 2 Diabetes in the Restoring Insulin Secretion (RISE) Study

Abstract

Objective: Obstructive sleep apnea (OSA) is associated with insulin resistance and has been described as a risk factor for type 2 diabetes. Whether OSA adversely impacts pancreatic islet β-cell function remains unclear. We aimed to investigate the association of OSA and short sleep duration with β-cell function in overweight/obese adults with prediabetes or recently diagnosed, treatment-naive type 2 diabetes.

Research design and methods: Two hundred twenty-one adults (57.5% men, age 54.5 ± 8.7 years, BMI 35.1 ± 5.5 kg/m²) completed 1 week of wrist actigraphy and 1 night of polysomnography before undergoing a 3-h oral glucose tolerance test (OGTT) and a two-step hyperglycemic clamp. Associations of measures of OSA and actigraphy-derived sleep duration with HbA_1c, OGTT-derived outcomes, and clamp-derived outcomes were evaluated with adjusted regression models.

Results: Mean ± SD objective sleep duration by actigraphy was 6.6 ± 1.0 h/night. OSA, defined as an apnea-hypopnea index (AHI) of five or more events per hour, was present in 89% of the participants (20% mild, 28% moderate, 41% severe). Higher AHI was associated with higher HbA_1c (P = 0.007). However, OSA severity, measured either by AHI as a continuous variable or by categories of OSA severity, and sleep duration (continuous or <6 vs. ≥6 h) were not associated with fasting glucose, 2-h glucose, insulin sensitivity, or β-cell responses.

Conclusions: In this baseline cross-sectional analysis of the RISE clinical trial of adults with prediabetes or recently diagnosed, untreated type 2 diabetes, the prevalence of OSA was high. Although some measures of OSA severity were associated with HbA_1c, OSA severity and sleep duration were not associated with measures of insulin sensitivity or β-cell responses.

Figure 1

Effect of OSA on insulin sensitivity and β-cell function as assessed by the hyperglycemic clamp. Adjusted means from multiple linear regression models for the relationship of log-transformed dependent variables (insulin sensitivity and β-cell function as assessed by the hyperglycemic clamp) with categories of OSA severity and treatment adherence. Data are least square means and 95% CIs adjusted for age, sex, race/ethnicity, BMI, WHR, and actigraphically measured sleep duration and physical activity. Steady-state C-peptide, ACPR_g, and ACPR_max are also adjusted for log insulin sensitivity (M/I).

Figure 2

Effect of OSA on HbA_1c and measures of glycemia as assessed by the OGTT. Adjusted means from multiple linear regression models for the relationship of dependent variables (glycemia as assessed by the 3-h OGTT) with categories of OSA severity and treatment adherence. Data are least square means and 95% CIs adjusted for age, sex, race/ethnicity, BMI, WHR, and actigraphically measured sleep duration and physical activity. Insulinogenic index, C-peptide index, and HOMA-IR were log-transformed for analyses.