Mechanistic Physiologically Based Pharmacokinetic (PBPK) Modeling of Renal and Systemic Disposition of Drugs and Metabolites

Abstract

Physiologically-based pharmacokinetic (PBPK) models integrate system specific anatomy and physiology information with drug specific physicochemical and pharmacokinetic properties to predict drug disposition. Such integration permits items, events, processes, and pathways to communicate and influence each other interactively. By taking advantage of such mechanistic nature of PBPK modeling, drug dispositions under untested scenarios could be predicted by extrapolation from observed data in known conditions. Renal clearance is one of the major pathways governing drug dispositions, which has three main mechanisms: unbound filtration, passive reabsorption, and active secretion. In comparison to intestinal absorption and hepatic metabolism, renal clearance has been relatively underappreciated. Controlled clinical experiments that test renal clearance changes under altered conditions and mechanisms have been primarily focusing on drug-drug interaction on active secretion. However, huge gaps in understanding renal clearance still exist in other areas such as altered urine pH and impaired renal function. Further, passive reabsorption has not been paid significant attention by the pharmaceutical field. Therefore, the overarching goal of this thesis is to leverage mechanistic PBPK modeling technique to understand and predict renal clearance of drugs and metabolites under altered urine pH and impaired renal function, with a special focus on compounds undergoing significant renal passive reabsorption. In Chapter 2, to predict the spatiodynamic process of renal passive reabsorption in human, we developed a dynamic physiologically-based mechanistic kidney model based on human data that can integrate drug permeability, tubular surface area, ionization status, and drug concentration gradient between lumen and system to estimate renal passive reabsorption and predict renal clearance of drugs. Using 46 test compounds with a variety of physicochemical properties, the model successfully predicted the renal clearances of 87% compounds within 2-fold and 98% compounds within 3-fold. Further, by incorporating active secretion, the model also successfully predicted the renal clearances of para-aminohippuric acid (PAH), cimetidine, salicylic acid, and memantine. In Chapter 3, to ensure the simulation output from PBPK models can be meaningfully compared to the arm vein plasma drug concentrations collected in clinical studies, we developed a forearm model that captures the tissue distribution at the peripheral sampling site using human arm physiology data, allowing for a better prediction of plasma drug concentrations that are comparable to observed data. The model was successfully verified using arterial and venous concentrations of nicotine, ketamine, lidocaine, and fentanyl simultaneously. Further, I demonstrated that use of a discrepant sampling site in PBPK modeling than observed clinical studies may lead to biased model evaluation, erroneous model parameterization, and misleading prediction in unstudied clinical scenarios. In Chapter 4, to predict the altered renal excretion and systemic AUC of drug and metabolite when urine pH is changed, the mechanistic kidney model developed and verified from Chapter 2 was integrated with the peripheral arm sampling and full body PBPK model developed from Chapter 3. The model was successfully verified with methamphetamine and amphetamine under varying urine pH statuses, and showed feasibility to predict quantitatively and clinically significant changes in drug and metabolite disposition under comedications and diseases that can alter urine pH. In Chapter 5, to predict renal clearance in patients with impaired renal function such as chronic kidney diseases, physiological changes in tubular flow and urine flow observed in chronic kidney disease patients were incorporated into the mechanistic kidney model developed and verified from Chapter 2. The model accounts for the adaptive renal tubular filtrate flows that decrease disproportionately with glomerular filtration rate, and was successfully verified using three parent-metabolite pairs, six non-permeable drugs, six permeable drugs, and two secreted drugs. In conclusion, in this thesis, I developed and verified a physiologically-based mechanistic kidney model to translate drug properties such as plasma protein binding, transcellular permeability, and active transport into renal clearance of drugs and metabolites. This mechanistic kidney model allows prediction of alterations in renal clearance of drugs and metabolites upon changes in urine pH and renal functions, and can be incorporated into a full-body PBPK model to predict alterations in systemic disposition of drugs and metabolites.