Metabolic tests to determine risk for type 1 diabetes in clinical trials.

AIMS: Evaluate the reproducibility and relationship of various metabolic tests conducted as part of the Diabetes Prevention Trial-type 1 diabetes.
METHODS: Coefficients of variation, intraclass correlation coefficients, and Pearson correlations between the same metabolic tests performed at different times as well as the different tests were determined.
RESULTS: Fasting samples on the same day had a coefficient of variation of < 10 for C-peptide, 11 for insulin, and 2 for glucose. Testing on separate days approximately doubled the variance. Stimulated insulin values had less variance than fasting values and there was only a moderate correlation between fasting and stimulated values on each test. While highly correlated, C-peptide values from mixed meal tolerance tests are significantly lower than that obtained during oral glucose tolerance tests (OGTTs). Neither peak nor area under the curve C-peptide on the oral glucose tolerance test was different between those with abnormal and normal glucose tolerance. Those with abnormal as compared with normal glucose tolerance had lower 30-min C-peptide and a longer time to peak C-peptide.
CONCLUSIONS: A large, multi-centre trial, with tests performed over a decade-long period, can provide robust data. C-peptide data from oral glucose tolerance tests and mixed meal tolerance tests differ; therefore, the same stimulation test should be used to evaluate changes in beta cell function over time. Worsening glucose tolerance is associated with lower C-peptide at 30 min and a delay in peak secretion on the oral glucose tolerance test. This Diabetes Prevention Trial-type 1 diabetes data can be used in planning parameters for future studies, including evaluation of new algorithms to determine risk of disease.