Within-subject variability of measures of beta cell function derived from a 2 h OGTT: implications for research studies

Aims/hypothesis Knowledge of the within-subject variability of a parameter is required to properly design and calculate sample sizes for longitudinal studies. We sought to determine the day-to-day variability of measures of beta cell function derived from an OGTT. Methods Thirty-seven adults (13 with normal glucose tolerance, ten with impaired glucose tolerance, 14 with type 2 diabetes) underwent a standard 2 h 75 g OGTT on two separate days (median time between tests, 7 days; range, 5-14). From these data, the reproducibility of several indices of beta cell function were determined: insulinogenic index (ΔI₀-₃₀/ΔG₀-₃₀), early C-peptide response (ΔCP₀-₃₀/ΔG₀-₃₀), incremental AUC insulin to glucose response (incAUCins/incAUCglu), integrated insulin secretion response from 0 to 120 min (IS/Glu₀-₁₂₀) and indices of beta cell function derived from a mathematical model. Results Within-subject variability for ΔI₀-₃₀/ΔG₀-₃₀ (CV 57.1%) was higher than ΔCP₀-₃₀/ΔG₀-₃₀ (CV 34.7%). Measures integrated over the full 120 min of the OGTT, incAUCins/incAUCglu (CV 24.9%) and IS/Glu₀-₁₂₀ (CV 17.4%), demonstrated less variability. The mathematical model-derived measures of beta cell glucose sensitivity (CV 20.3%) and potentiation (CV 33.0%) showed moderate variability. The impact of the different measures' variability on sample size (30% change from baseline) is demonstrated by calculated sample sizes of 89 for ΔI₀-₃₀/ΔG₀-₃₀, 37 for ΔCP₀-₃₀/ΔG₀-₃₀, 21 for incAUCins/incAUCglu and 11 for IS/Glu₀-₁₂₀. Conclusions/interpretation Some OGTT-derived indices of beta cell function, in particular the insulinogenic index, demonstrate high within-subject variability. Integrated measures that utilise multiple time points and measures that use C-peptide show less variability and may lead to a reduced sample size requirement.