Curtis K. Johnston (1), Matthew M. Riggs (1), Jan Marquard (2), Nima Soleymanlou (3), Valerie Nock (2), Karl-Heinz Liesenfeld (2)
(1) Metrum Research Group, Tariffville, CT, USA, (2) Boehringer Ingelheim International GmbH, Ingelheim, Germany (3) Boehringer Ingelheim Canada Ltd./Ltée, Burlington, Canada
Objectives: The objective of this modelling and simulation study was to further characterize the empagliflozin (EMPA) 2.5 mg qd dose independent of data from EASE-3 [1], a phase 3 study investigating this dose. Specifically, this exposure-response modelling study (M‑EASE‑2) was performed to simulate the placebo-corrected change from baseline in the study population of a second phase 3 trial (EASE-2) which did not investigate this dose [1]. Moreover, covariate effects for the exposure-response relation were assessed during model building.
Methods: M-EASE-2 model development was informed by data from EASE-2 (52 weeks, EMPA 10 and 25 mg qd) and EASE-1 (4-week phase 2 study [2], EMPA 2.5, 10, 25 mg qd). Furthermore, prior information for AUC50 from an analysis in patients with T2DM [3], was used during parameter estimation. Predictions of EMPA exposure, based on individual parameter estimates from a previous population PK analysis, were used as input for this exposure-response analysis. The analysis was conducted in NONMEM Version 7.4, applying Markov chain Monte Carlo Bayesian estimation. The effect of EMPA exposure on HbA1c was modelled as a direct Emax model including a placebo effect. Investigation of the covariate effects was undertaken using a full covariate modelling approach. The primary covariates of interest were pre-defined based on findings in previously conducted analyses including patient sex and baseline weight, eGFR and HbA1c. Indication-specific factors were also evaluated including daily insulin dose at baseline and insulin dose type (INSDT): multiple daily injections vs. continuous subcutaneous insulin infusion. For internal and external model evaluation via visual posterior predictive checks, 500 Monte Carlo simulation replicates each using 500 samples from the posterior distribution of model parameters were generated. External model qualification focused on an out of sample prediction using data from EASE-3. For trial simulations investigating the effect of EMPA 2.5 mg in the study population of EASE‑2, 500 Monte Carlo simulation trial replicates with 239 patients were created. Each simulation utilized a random sample from the posterior distribution of model parameters and variability terms (inter-subject and intra-subject variability). The impact of prior information on HbA1c lowering was investigated via sensitivity analysis (varied informativeness and point estimate).
Results: A direct response model best described the effect of EMPA exposure on changes in HbA1c. Drug effect was characterized by a Emax model driven by AUCτ,ss, with a time-dependent linear placebo effect. Typical population PD parameters (95% credible interval) were: Baseline HbA1c: 8.14% (8.07%, 8.22%); AUC50: 498 (296, 819) nmol·h/L; Emax: 0.579% (0.491%, 0.678%); Placebo Effect: 2.61 x 10-5 (1.96 x 10-5, 3.29 x 10-5) %/h. Inter-individual variance (CV %) for baseline HbA1c and Emax were 7.2% and 38%, respectively. The proportional and additive residual variability estimates (CV% and SD) were 4.6% and 0.11, respectively. Sensitivity analyses for the AUC50 estimate demonstrated that the informative prior on AUC50 resulted in conservative estimates of the HbA1c lowering for EMPA 2.5 mg. The simulations performed for external qualification were consistent with EASE-3 results. The simulated median (95% CI) placebo-corrected HbA1c change for EMPA 2.5 mg was -0.29% (‑0.39%, ‑0.19%) at Week 26 and -0.29% (-0.40%, -0.19%) at Week 52. Moderate effects influencing the placebo-corrected change of HbA1c at Week 26 relative to baseline were observed for INSDT, baseline HbA1c and baseline eGFR. Simulations to illustrate the impact of baseline HbA1c on change in HbA1c were performed. For a 2.5 mg dose, a median placebo adjusted change in HbA1c at 26 weeks relative to baseline of -0.28% versus -0.32% was predicted for a baseline HbA1c of 8.0% and 9.0%, respectively.
Conclusions: The exposure-response model successfully predicted the time-course and dose-related changes of HbA1c in EASE-3, a study not included in the model development. This external model qualification demonstrated the utility of the developed model to predict hypothetical outcomes in populations similar to the EASE-2 population reliably. M-EASE-2, a descriptive modelling and simulation approach, illustrated how pharmacometric analyses can be utilized to create further evidence of efficacy and substantiate clinical findings.
References:
[1] Rosenstock J, et al. Diabetes Care. 2018; 41: 2560-2569.
[2] Pieber T R, et al. Diabetes Obes Metab. 2015;17:928-35.
[3] Baron KT, et al Diabetes Ther. 2016;7:455-71.
Reference: PAGE 28 (2019) Abstr 9140 [www.page-meeting.org/?abstract=9140]
Poster: Drug/Disease Modelling - Endocrine