Exploring Various Sensitivity Methods for the Analysis of Pregnancy and Non-Pregnancy PBPK Models

Authors:: Poster Author(s)

Erukulla, Nikith, FDA/NCTR (Student); Fairman, Kiara, FDA/NCTR (Mentor); Lumen, Annie, FDA/NCTR (Mentor); Li, Miao, FDA/NCTR (Mentor)

Center:: Contributing Office

National Center for Toxicological Research

Abstract

Poster Abstract

Background

Physiologically based pharmacokinetic (PBPK) models use mathematical equations to predict the absorption, distribution, metabolism, and excretion (ADME) of chemical substances, such as a drug of interest, in humans and other species. They are generally used in drug research and development and work by representing anatomy, chemical parameters, and biochemistry in mathematical terms. These models have multiple compartments with each one representing a part or system in the body such as the blood plasma, liver, kidneys, etc. The compartments are represented by a system of differential equations to track the amount of the desired substance within each compartment. Compartment defining parameters that represent filtration rate, organ volumes, blood flows, etc. (such information is sourced from available research) can be altered to see how each affects the ADME of the drug present. Sensitivity analysis allows for the increased understanding of a model by showing how a change in inputs changes the outputs. Local sensitivity analysis is characterized by changing one parameter at a time while global sensitivity analysis is characterized by changing multiple parameters at a time. Sensitivity analysis findings can assert the need for greater accuracy in model outputs, especially if a parameter is very sensitive but not much data is available. Such data scarcity exists in pregnancy pharmacokinetics. Labetalol is commonly prescribed to hypertensive pregnant women in addition to being used to treat hypertensive crises. It is a beta blocker that works by causing vasodilation and slowing the heart rate to decrease blood pressure. Therefore, our objective is to use various sensitivity analysis methods to determine which method is best suited for the model code. Furthermore, we will determine which parameters are the most sensitive to increase model confidence. Additionally, comparison of sensitive parameters between both pregnant women and non-pregnant women will be explored to further elucidate pregnancy specific parameter sensitivities.

Methods

A PBPK model of labetalol is translated to the R coding language in this project. The existing non-pregnancy PBPK model of labetalol will first be translated from Berkeley Madonna code to R using the mrgsolve or deSolve package. We will then extend the translation to the more complex pregnancy PBPK model. This will be done by learning the aforementioned packages and Berkeley Madonna and then translating the code line by line. The inputs and outputs will be tested and compared to ensure the code was translated correctly. Additionally, a parameter sensitivity package in R will be used to perform local and global sensitivity analysis on the labetalol PBPK models. Other packages may be explored to compare multiple methods of parameter sensitivity analysis. The results of each sensitivity analyses sensitivity coefficients parameters will be compared with one another.

Conclusion

The sensitivity analyses will allow the exploration of the various sensitive parameters identified. From previous results from a local sensitivity analysis, we are aware of sensitive parameters for labetalol such as body weight, albumin, and clearance. A global sensitivity analysis can be more beneficial than a local sensitivity analysis because the prior allows for the quantification of parameter interactions. The pertinent sensitive parameters will bring clarity to parameter contributions to model confidence.