An Evaluation of Three Semi-mechanistic PK/PD Models for Predictive Performance of Long-term HbA1c via Short-term Glycemic Changes for a Glucagon Receptor Antagonist in Type 2 Diabetes Patients

Glycated hemoglobin (HbA1c) is the gold standard for assessing long-term glucose control in diabetes. Ability to predict long-term HbA1c using glucose response from clinical trials of shorter duration is much desired, but uncertainty exists in using short-term glucose biomarker changes to predict long-term HbA1c changes in Type 2 diabetes (T2D) trials. Three common semi-mechanistic PK/PD models from the literature (ADOPT1, FFH2, and FHH3) are evaluated for their ability to predict longer-term HbA1c for a glucagon receptor antagonist (GRA) after 24 weeks of treatment.

GRA PK, glucose and HbA1c data from a 4-week Phase 1 trial were used to develop the PK/PD models. Models were assessed for predictive performance via glucose and HbA1c data from a 24-week Phase 2 trial. Mean change from baseline in HbA1c (ΔHbA1c) at Week 24 between observed and simulated values was utilized to evaluate the models’ predictive performance. Mean prediction error (MPE) for bias and root mean square error (RMSE) for precision were also computed using observations and post-hoc model estimates of mean ΔHbA1c.

The FHH model predicted closely the mean ΔHbA1c at Week 24 across all dose groups in the
Phase 2 trial (Table 1). Conversely, the ADOPT and FFH models had over-predicted the mean reduction in HbA1c. All three models had similar MPE and RMSE estimates.

The FHH model consists of a transit compartment structure, which is useful in modeling long delays between glucose and HbA1c. It also has a non-linear relationship between glucose and HbA1c. Thus, for GRA, the FHH model could be useful in predicting long-term HbA1c at 24 weeks from short-term glucose and HbA1c data in a 4-week trial.

Dose groupMean ΔHbA1c at Week 24 (%) ([95% CI of Mean], n)
ObservationsFHH SimulationsADOPT SimulationsFFH Simulations
Placebo-0.446 ([-0.720,
-0.173], 26)
-0.324 ([-0.352,
-0.296], 1000)
-0.765 ([-0.794,
-0.736], 1000)
-0.195 ([-0.214,
-0.177], 1000)
2.5 mg QD-0.566 ([-0.768,
-0.363], 32)
-0.554 ([-0.583,
-0.525], 1000)
-0.910 ([-0.940,
-0.881], 1000)
-0.636 ([-0.660,
-0.613], 1000)
10 mg QD-0.897 ([-1.16,
-0.639], 35)
-0.997 ([-1.03,
-0.964], 1000)
-1.17 ([-1.20,
-1.14], 1000)
-1.13 ([-1.16,
-1.10], 1000)
20 mg QD-0.928 ([-1.15,
-0.705], 40)
-1.23 ([-1.27,
-1.20], 1000)
-1.31 ([-1.34,
-1.28], 1000)
-1.50 ([-1.54,
-1.46], 1000)

Table 1. Summary statistics for observations vs model simulations for mean ΔHbA1c at Week 24.
Abbreviations: ΔHbA1c = change in HbA1c. CI = confidence interval. n = number of patients. QD = once daily.

1. Møller JB, Overgaard RV, Kjellsson MC, Kristensen NR, Klim S, Ingwersen SH, Karlsson MO. Longitudinal Modeling of the Relationship Between Mean Plasma Glucose and HbA1c Following Antidiabetic Treatments. CPT Pharmacometrics Syst Pharmacol. 2013;2(10):e82.
2. de Winter W, DeJongh J, Post T, 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;33(3):313-343.
3. 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;84(2):228-235.


  • NUS / Eli Lilly