K. Melissa Hallow[1], David A James[2], Wenping Wang[2]
[1] University of Georgia, Athens, GA, USA, [2] Novartis Pharmaceuticals, East Hanover, NJ
Objective: RxODE is a new R package that facilitates straightforward PKPD simulation within R, providing an efficient and versatile way to specify dosing scenarios and sampling schedules and to perform simulation with variability with minimal custom coding. Because simulations are performed entirely within R, it also takes advantage of the R ecosystem, including powerful graphics packages and the Shiny package for developing user interfaces. Here we show how we have used RxODE to develop an interactive tool for exploring different dose regimens and expected efficacy for a novel anti-diabetic agent.
Dual sodium glucose cotransporter (SGLT) 1/2 inhibitors reduce glucose levels by inhibiting both gut absorption and renal reabsorption of glucose. However, EC50s for these two mechanisms can be quite different, and drug concentrations in the gut decline much faster than in plasma. In this complex pharmacodynamic setting, PKPD simulation provides a means for determining an appropriate dosing regimen and for evaluating efficacy potential.
Methods: An existing model of glucose-insulin dynamics[1] was adapted to incorporate the dual mechanism of action of an SGLT 1/2 inhibitor, and K-PD parameters were estimated in NONMEM using phase II biomarker data. The model was then implemented in R using RxODE. RxODE compiles the model in C, and thus run times are extremely fast. It also provides a function for generating R shiny apps to interface with the model, which can then be further customized, shared online with clinical teams, and used to facilitate interactive simulation.
Results: The generated shiny app utilizes RxODE to allow users to vary simulation parameters, including dose, regimen (QD, BID, TID), and dose timing relative to meals, and to visualize in real time the effects on glucose profiles and HbA1c. It also allows real-time evaluation of the impact of parameters for which there is large uncertainty, such as the time constant for gut drug concentrations.
Conclusions: RxODE provides a tool for straightforward specification and simulation of a wide range of dosing scenarios. It takes advantage of the R ecosystem, including powerful graphics packages and Shiny. Thus, it can facilitate real-time, interactive engagement with clinical teams, reducing collaboration lag time. We are currently developing functionality to facilitate parameter estimation of RxODE models using the nlme package[2].
References:
[1] Jauslin PM, Frey N, Karlsson MO. Modeling of 24-hour glucose and insulin profiles of patients with type 2 diabetes. J Clin Pharmacol 2011, 51(2):153-64 2.
[2] Pinheiro, Jose C. and Douglas M. Bates. “Mixed Effects Models in S and S-Plus” Springer-Verlag, New York, 2000
Reference: PAGE 24 () Abstr 3542 [www.page-meeting.org/?abstract=3542]
Poster: Methodology - Other topics