Quantifying drug effects in phase 2a anti-diabetic studies: Power and accuracy of four HbA1c models
Gustaf J. Wellhagen, Mats O. Karlsson, Maria C. Kjellsson
Department of Pharmaceutical Biosciences, Uppsala University
Background and Objectives: Several dynamic models of HbA1c have been suggested for the analysis of anti-diabetic study data: FPG-FSI-HbA1c (FFH) [1], FPG-Hb-HbA1c (FHH) [2], Integrated Glucose-RBC-HbA1c (IGRH) [3] and ADOPT [4]. HbA1c formation is in these models driven by fasting plasma glucose (FPG) or mean plasma glucose (MPG), with or without fasting serum insulin (FSI). The aim with this project was to investigate the power to detect a drug effect and the accuracy of the estimated drug effect on HbA1c for the four models.
Methods: Data was simulated to mimic a 12-week parallel group phase 2a clinical trial with type 2 diabetic patients. Glucose and insulin were simulated using the Integrated Glucose-Insulin (IGI) model [5] with one of 5 drug effects acting on: basal insulin secretion (BASI), incretin response (INCR), insulin dependent glucose elimination (CLGI), insulin independent glucose elimination (CLG) or endogenous glucose production (EGP). The IGRH model was used to simulate HbA1c as a mechanistic function of MPG. The simulated data was analyzed with and without the drug effects using each of the four models. The power to detect a drug effect was assessed using Monte Carlo Mapped Power (MCMP) calculations. Using the difference in Objective Function Value (OFV) between full and reduced runs gave rise to the ∆OFV displayed in Table 1. Accuracy was assessed by calculating the Relative Estimation Error (REE) of the ∆∆HbA1c for each model.
Results: 1. As seen in Table 1, the FFH model needed the least number of individuals to identify a drug effect compared to other models except in one case. The FHH and ADOPT models were the most quick and stable to run. The accuracy of the parameter estimates was better for the MPG driven models (IGRH and ADOPT), with lower REE.
Table 1. Ratio of ∆OFV
|
BASI |
CLG |
CLGI |
EGP |
INCR |
|
|
IGRH/FFH |
0.87 |
0.62 |
0.88 |
0.71 |
1.07 |
|
IGRH/FHH |
1.00 |
0.87 |
1.02 |
0.84 |
1.26 |
|
IGRH/ADOPT |
0.98 |
0.98 |
0.98 |
0.98 |
0.98 |
Conclusions: The FFH model displayed a higher power compared to the other models except for one drug effect where the ADOPT model needed the fewest number of individuals. This is probably due to additional information given by the FSI records. The IGRH and ADOPT models produce the most accurate predictions of HbA1c at the end of the study. The relative merit of models depends on which mechanism of action the studied drug has.
Acknowledgements: This work was supported through the DDMoRe project (ddmore.eu) and the FP7-HEALTH-2013-602552.
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] Hamrén B, Björk E, Sunzel M, Karlsson MO. Models for Plasma Glucose, HbA1c, and Hemoglobin Interrelationships in Patients with Type 2 Diabetes Following Tesaglitazar Treatment. Clin Pharmacol Ther. 2008 Mar;84(2):228–35.
[3] Lledó-García R, Mazer NA, Karlsson MO. A semi-mechanistic model of the relationship between average glucose and HbA1c in healthy and diabetic subjects. J Pharmacokinet Pharmacodyn. 2013 Apr;40(2):129–42.
[4] Møller JB, Overgaard RV, Kjellsson MC, Kristensen NR, Klim S, Ingwersen SH, et al. Longitudinal Modeling of the Relationship Between Mean Plasma Glucose and HbA1c Following Antidiabetic Treatments. CPT Pharmacomet Syst Pharmacol. 2013 Oct;2(10):e82.
[5] Jauslin PM, Silber HE, Frey N, Gieschke R, Simonsson USH, Jorga K, et al. An Integrated Glucose-Insulin Model to Describe Oral Glucose Tolerance Test Data in Type 2 Diabetics. J Clin Pharmacol. 2007;47(10):1244–55.
