Genotyped versus phenotyped dosing to account for UGT polymorphism of the novel PPAR agonist sipoglitazar
F.Stringer (1), B.Ploeger (2,3), J. de Jongh (2,3), G.Scott (1), R Urquart (5), A.Karim (4) and M. Danhof (2,3)
(1)Takeda Global Research and Development Ltd. Europe; (2) LAP&P Consultants BV, Leiden, The Netherlands; (3) Leiden-Amsterdam Center for Drug Research, Division of Pharmacology, Leiden, The Netherlands; (4)Takeda Global Research and Development Ltd, US; (5)Takeda Pharmaceutical Company Ltd, Japan.
Objectives: Sipoglitazar, a novel orally available PPAR agonist with activities for PPAR α, δ and γ is metabolised through UGT. A polymorphism of the UGT enzyme was observed to result in a wide distribution of clearance for the same dose. The compound has a narrow therapeutic window therefore limiting exposure. The relative merits of prospective initial dose selection by genotype assignment versus retrospective dose selection by a TDM approach were assessed by population analysis of Phase I and II data.
Methods: A population PK model was developed using 3 small Phase I studies. The model was first used to predict the exposure in an additional large Phase I trial in 524 patients to characterise the genotype distribution. External model validation was performed using two Phase II trials with trough sample collection. The model was implemented in NONMEM and the distribution of the exposure resulting from the genotype was explored using the $MIX routine to evaluate the proportion in each genotype.
Results: The model developed using the 3 smaller Phase I trials was not able to accurately predict the exposure distribution of the genotype in the larger Phase I study. Therefore, the model was updated using the latter data and validation was confirmed using the 2 Phase II studies. The 3 small Phase I studies were shown to underpredict the variability associated with the clearance of the 3 different genotypes through the validation provided by the larger study. Updating the model by combining the Phase I studies resulted in a more accurate prediction of the genotype distribution in the Phase II data.
Conclusion: Use of a genotype approach was considered as a viable method to dose selection. However, population PK analysis showed this approach would result in over exposure in the poor metaboliser (PM) patients. Exploration of a TDM approach through simulation indicated that a smaller number of PM would be over exposed. These results become critical and rate limiting when considering the correct dosing approach for the PM group. Modeling and simulation has demonstrated that following a genotype approach from a small population in this case was not able to accurately predict with sufficient precision the exact distribution of the clearance associated with PM group. This could have resulted in the over exposure of PM patients and resulted in a potential safety risk.