Comparison of Visual Methods to Reanalyze Provocative Test Data from Randomized Controlled Clinical Trials in New- or Recent-Onset Type 1 Diabetes.

INTRODUCTION: Over the last 13 years, the Immune Tolerance Network (ITN), has conducted trials of agents to abrogate the autoimmunity underlying type 1 diabetes. Primary endpoints center on the change of C-peptide production during mixed meal tolerance tests (MMTT), measured as the area under the curve (AUC) or AUC mean over 2-3 years. Studies permit rapid-acting insulin until a few hours before the MMTT, and thus do not exclude overnight hyperglycemia prior to testing. We hypothesize that overnight or fasting hyperglycemia will deplete pre-formed insulin and impact measurements of first-phase insulin secretion and C-peptide AUC.

METHODS: Publicly available, deidentified, subject-level data were obtained from ITN TrialShare. We developed several graphical analyses to reexamine results from each MMTT including combined glucose and C-peptide response curves, the centroids of polygons of MMTT timepoints, and ratios comparing extents of excursions of glucose and c-peptide production.

RESULTS: We have applied these graphical analyses to 1161 MMTT from 245 subjects in 8 studies. Graphical analyses of MMTT results for individuals over the course of the follow-up period reflect the expected loss of c-peptide and higher blood glucose during MMTT; centroids move accordingly, upwards and leftwards.

CONCLUSIONS: We were able to analyze MMTT data from ITN studies with several graphical analyses. We are poised to apply these approaches to test our central hypothesis by comparing how deviations from modeled rates of predicted changes for an individual over time correlate with blood glucose levels in the hours before a MMTT. This may lead to refinement of future trial protocols to ensure tighter regulation of glycemic excursions ahead of provocative testing.