PI-097 - MECHANISTIC MODELING OF SPATIAL HETEROGENEITY OF DRUG PENETRATION AND EXPOSURE IN THE HUMAN CENTRAL NERVOUS SYSTEM AND BRAIN TUMORS

J. Li¹, C. Wickramasinghe², J. Jiang¹, a. Wu³, y. Jiang¹, A. Tovmasyan⁴, S. Kim², N. Sanai⁴; ¹Karmanos Cancer Institute, Detroit, Michigan, United States, ²Karmanos Cancer Institute, Detroit, United States, ³Northville High School, Northville, Michigan, United States, ⁴Barrow Neurological Institute, Phoenix, Arizona, United States.

Background: The spatial pharmacokinetics of many systemically administered drugs in the human central nervous system (CNS) and brain tumors remain poorly understood because direct measurement of spatial-temporal drug penetration and exposure in the CNS is difficult or infeasible given the challenge of sampling and limitation of currently available imaging and analytical technologies. The lack of quantitative knowledge on spatial CNS and tumor pharmacokinetics makes development of new drugs and use of existing drugs for treating brain cancer a challenging and often unsuccessful task. This study aimed to develop an innovative mechanistic modeling platform for quantitative prediction of spatial pharmacokinetics of systemically administered drugs in the human CNS and brain tumors.

Methods: A novel 9-compartment CNS (9-CNS) physiologically based pharmacokinetic model was developed to account for general anatomical structure and pathophysiological heterogeneity of the human CNS and brain tumors. Drug distribution into and within the CNS and tumors is driven by plasma concentration – time profiles and governed by drug properties and CNS pathophysiology. The model was validated by comparisons of model predictions and clinically observed data of 6 drugs (abemaciclib, ribociclib, pamiparib, olaparib, temuterkib, and ceritinib) in glioblastoma patients.

Results: As rigorously validated, the 9-CNS model allows reliable prediction of spatial pharmacokinetics of systemically administered drugs in different regions of the brain parenchyma (i.e., parenchyma adjacent to CSF and deep parenchyma), tumors (i.e., tumor rim, bulk tumor, and tumor core), and CSF (i.e., ventricular CSF, cranial and spinal subarachnoid CSF). By considering inter-individual plasma pharmacokinetic variability and CNS/tumor heterogeneity, the model well predicts the interindividual variability and spatial heterogeneity of drug exposure in the CNS and tumors as observed for all 6 drugs in glioblastoma patients.

Conclusion: The 9-CNS model is a first-of-its kind, mechanism-based computational modeling platform that enables early reliable prediction of spatial CNS and tumor pharmacokinetics based on plasma concentration – time profiles. It provides a valuable tool to assist rational drug development and treatment for brain cancer.