J. Rogers(1), D. Polhamus(1), W. Gillespie(1), C. Friedrich(2), A. Staab(2), S. Retlich(2)
(1)Metrum Research Group, Tariffville, CT, USA; (2)Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach an der Riss, Germany
Objectives: Linagliptin is a dipeptidyl peptidase 4 (DPP-4) inhibitor developed for treatment of Type 2 diabetes mellitus. Sitagliptin is another available DPP-4 inhibitor and serves as a relevant comparator. Our objective was to estimate the magnitude of the HbA1c lowering effects of linagliptin and sitagliptin, based on a comprehensive analysis of available clinical trial data. Specifically, we sought to provide the comparison by means of a longitudinal dose-response meta-analysis based on indirect comparisons. Given appropriate covariate adjustment to account for differences in study designs and patient population, one may infer the efficacies of linagliptin and sitagliptin relative to placebo when administered to comparable patients under comparable conditions.
Methods: An analysis data set was assembled based on a systematic review of available clinical trials for sitagliptin and summary statistics computed from Boehringer Ingelheim internal data sources for linagliptin. A Bayesian hierarchical model was developed to describe HbA1c levels as a function of dose, time, and selected covariates. Covariates related to demographics and study design were evaluated and incorporated in the model where appropriate. Standard model diagnostics were applied to ensure adequate model convergence and model fit. Population simulations based on the selected model were used to evaluate the average effects of linagliptin and sitagliptin in a reference population over 24 weeks of treatment.
Results: The final model described HbA1c levels for placebo treated individuals as a nonlinear function of time. Drug effects were incorporated as multiplicative adjustments to the placebo time course, and additional multiplicative covariate adjustments were made for baseline HbA1c, washout duration and race. Population simulations assuming a study design with no washout and a mean baseline HbA1c of 8% resulted in expected HbA1c differences from placebo at 24 weeks of -0.810 percentage points for linagliptin 5 mg (90% credible interval from -0.881 to -0.740) and -0.807 percentage points for sitagliptin 100 mg (90% credible interval from -0.878 to -0.737).
Conclusion: Consistent with the common mechanism of action, this model-based meta-analysis showed that the new DPP-4 inhibitor linagliptin (5 mg qd) results in a comparable efficacy as seen with the DPP-4 inhibitor sitagliptin (100 mg qd).
Reference: PAGE 21 (2012) Abstr 2448 [www.page-meeting.org/?abstract=2448]
Poster: New Modelling Approaches