Mohammed Saleh (1), Chi Loo (1), Jeroen Elassaiss-Schaap (1,2), Elizabeth de Lange (1)
(1) Leiden Academic Center for Drug Research, The Netherlands. (2) PD-value B.V., The Netherlands
Introduction: Predicting drug exposure of brain, though challenging, remains critical for drug development. The challenges of predicting brain pharmacokinetic (PK) profiles arise mainly from the ethical restrictions of human brain sampling and the complex physiology of the central nervous system (CNS). Furthermore, in patients with CNS diseases, brain PK profiles might be altered due to the disease-specific pathophysiology. Lumbar cerebrospinal fluid (CSF) drug concentration has typically been assumed as an adequate surrogate of unbound drug concentration of brain; the value of that assumption remains questioned.
We previously published a comprehensive CNS physiology-based PK (PBPK) model that predicts PK profiles of brain cells, brain extracellular fluid, and cranial and spinal CSF including lumbar CSF (1,2). The physiologic buildup of this model enables the interspecies and interpopulation translation of PK profiles and provides a framework to explore the impact of altering a single or multiple physiological parameters on CNS PK profiles. Such a model can therefore be used to mechanistically predict the effect of disease-altered CNS physiology on unbound drug exposure of brain.
Objectives: The aim of this study is to predict the effect of altered cerebrospinal fluid (CSF) dynamics on PK profiles of brain and lumbar CSF and on the brain-to-lumbar CSF drug concentration ratio. Changes in CSF dynamics, i.e. CSF volume and flow, are common in CNS disease pathophysiology and often alter drug exposure of CSF; their effect on the drug exposure of brain remains unexplored.Text regarding objectives.
Methods: Simulations were performed using an improved version of the previously published comprehensive CNS PBPK model [1, 2]. The improvements included aspects related to non-specific binding of the brain tissue, effect of the brain tissue pH on drug ionization, and assumptions of neutral and charged molecules passive transport across the blood brain barrier and blood-CSF barrier. We refer to this new model as “Leiden CNS physiology-based PK predictor V3.0” or LeiCNS-PK3 (Saleh MAA et al., manuscript in preparation). Simulations were carried out with seven small drug molecules with different physicochemical properties. PK profiles of brain ECF and lumbar CSF were compared at different CSF volume and flow values: physiological, two-, and five-fold change. Simulations were executed in R version 3.6.1 using RxODE package version 0.9.1-0.
Results: The improved LeiCNS-PK3 can predict the in vivo-measured unbound drug concentrations in brain and CSF compartments of rats and humans with less than 3-fold error. LeiCNS-PK3 simulations show that altered CSF dynamics change lumbar CSF PK but not brain PK profiles for all test drugs. Thus, the drug concentration ratio of brain-to-lumbar CSF is changed.
Conclusions: LeiCNS-PK3 can predict brain PK profiles of drugs with different physicochemical properties. Furthermore, we demonstrate with LeiCNS-PK3 that drug concentration of lumbar CSF is not an accurate surrogate of unbound drug concentration of brain, particularly in diseases involving altered CSF dynamics. Brain PK can rather be adequately predicted using a systems approach, e.g. PBPK modeling, to account for the diseased-induced, multi-level changes in CNS physiology and their effect in turn on brain PK. LeiCNS-PK3 is therefore useful at early stage drug development to support (pre-) clinical study design.
References:
[1] Yamamoto Y. et al. CPT-PSP (2017). 6 (11). 765-777
[2] Yamamoto Y. et al. Eur. J. Pharm. Sci (2018). 112 (September). 168-179
Reference: PAGE () Abstr 9254 [www.page-meeting.org/?abstract=9254]
Poster: Drug/Disease Modelling - CNS