Marion Bouillon-Pichault, Bernard Sebastien, Clemence Rigaux
Sanofi R&D
Objectives: The link between fasting plasma glucose (FPG) and glycosylated haemoglobin (HbA1c) has previously been described using empirical correlations at steady state and semi-mechanistic modeling for few anti-diabetic compounds. In particular, the correlation between HbA1c and FPG has been established in several studies (1)(2). Although these correlations or models have been calculated or developed on large populations of subjects, they have generally not been assessed and compared across different types of treatments. To that end, a model based meta-analysis was performed to assess the link between FPG and Hba1c over a large set of anti-diabetic drug class.
Methods: The analysis database was made of publicly available summary results from Type 2 Diabetes clinical trials published in the medical literature. Only monotherapy treatment arms from randomized controlled studies with at least 12 weeks duration, with documented HbA1bc with fasting plasma glucose data, were included in the analysis database. Only drug classes including more than 5 trials were kept, to ensure a sufficient representation of each drug class. The data were analyzed in the non-linear mixed effect model framework, using Monolix 4.3.3, considering trial arm as individual, with mean change (from study baseline) HbA1c as dependent variable and mean change in FPG as explanatory variable (3)(4).
Results: A general empirical model was established linking FPG to HbA1c. Trial arms were used as ID and the residual variability was weighted by the square root of the number of measurements at each time point, to account for summary data precision. Several covariates were integrated in the model: treatment drug class, washout, gender and age. The model was qualified, using Goodness of fit plots, Visual Predictive Checks and other classical diagnosis criteria generated from an external database made of 100 studies previously removed from the model development database for model qualification.
Conclusion: By predicting long term HbA1c response given shorter terms glycemia, this work will help phase 2 decision making and designing phase 3 in T2DM.
References:
[1] Winter W de, DeJongh J, Post T, Ploeger B, Urquhart R, Moules I, et al. A Mechanism-based Disease Progression Model for Comparison of Long-term Effects of Pioglitazone, Metformin and Gliclazide on Disease Processes Underlying Type 2 Diabetes Mellitus. J Pharmacokinet Pharmacodyn. 2006 Jun;33(3):313–43.
[2] Samtani MN. Simple pharmacometric tools for oral anti-diabetic drug development: competitive landscape for oral non-insulin therapies in type 2 diabetes. Biopharm Drug Dispos. 2010;n/a – n/a.
[3] UsersGuide_Mnx_4.3.2.pdf.
[4] Ahn JE, French JL. Longitudinal aggregate data model-based meta-analysis with NONMEM: approaches to handling within treatment arm correlation. J Pharmacokinet Pharmacodyn. 2010 Apr;37(2):179–201.
Reference: PAGE 24 (2015) Abstr 3349 [www.page-meeting.org/?abstract=3349]
Poster: Drug/Disease modeling - Other topics