Michael Monine (1), Daniel Norris (2), Yanfeng Wang (2), Ivan Nestorov (1)
(1) Clinical Pharmacology and Pharmacometrics, Biogen, Cambridge, MA, (2) PK/Clinical Pharmacology, Ionis Pharmaceuticals, Carlsbad, CA
Objectives: Antisense oligonucleotides (ASOs) [1] are promising therapeutic agents for a variety of neurodegenerative disorders caused by genetic abnormalities or increased protein synthesis (e.g., Alzheimer’s and Parkinson’s diseases, ALS). Effective treatment of these disorders requires achieving adequate exposures at the relevant sites of action in the central nervous system (CNS). The blood-brain barrier (BBB) represents a challenge to the delivery of ASOs following systemic administration. Intrathecal (IT) delivery, in which drugs are administered directly into the cerebrospinal fluid (CSF) space, enables to bypass the BBB. Due to narrow specialty and novelty of IT dosing for ASOs, very limited PK data is available and only a few modeling reports have been generated [2,3]. In this work, mechanisms driving the PK across the CNS are explored based on PK studies in non-human primates (NHPs).The objectives of this work were to (1) develop a physiologically-based pharmacokinetic (PBPK) modeling framework to describe distribution of IT-injected ASOs in the CNS and through the body, and (2) compare different model structures with respect to parameter identifiability to select the most practical model.
Methods: The PBPK modeling framework is built around 11 observable compartments including plasma, lumbar CSF (site of injection), 3 segments of the spinal cord (lumbar, thoracic and cervical), 4 brain tissues (pons, cerebellum, hippocampus and cortex), liver and kidneys, for which time-dependent ASO concentrations are available from the NHP study. The entire CSF space is divided into a series of compartments. Since stirring in the CSF is significantly constrained, the formation of concentration gradients along the spinal column towards cranial regions can be a critical factor determining access to the CNS tissues. The system is described by ordinary differential equations (ODEs) that are formulated in terms of ASO amounts in different physiological compartments connected by first-order transfer rates. The number of rate constants to be estimated exceeds 20 depending on the model structure. To elucidate the trade-off between parameter identifiability and physiological plausibility of such complex and overparameterized models, several alternative model structures were chosen and fitted to the NHP data. ADAPT5 [4] was used as a fitting tool. Due to sparsity of the data (serial sacrifice design), naïve pooled data fitting was performed. To evaluate models, the Akaike Information Criterion (AIC) and standard error of estimates were used as ranking criteria.
Results: We analysed various compartmental model structures and proposed the model that splits the CSF space into a physiologically-justifiable number of compartments while keeping the uncertainty in parameter estimates acceptable. Model-based analysis of the NHP data allows to identify key processes that determine complex PK of IT-administered ASOs. Shortly after the injection, a sharp decrease in the CSF exposure is observed. A plasma exposure peak appears almost instantaneously and is followed by ASO accumulation in the liver and kidneys. The maximum concentrations in the spinal cord and brain tissues are reached within 1-2 days after the IT injection and significantly exceed those obtained after systemic administration. Delays in CSF peak concentrations emerge towards the upper CSF regions and result in decrease in exposures up the spinal cord. Thus, concentration gradients emerging along the CSF canal can limit the initial access to the brain tissues as well, which may have potential implications in human because of the lengthier spinal column.
Conclusions: As human CNS tissues are impossible to analyse for ASO concentrations in vivo, and the CSF concentration is the only measurable source of PK data in human CNS, understanding of distribution mechanisms that are likely to be similar in humans and monkeys is critical to make clinically relevant predictions. The proposed PBPK modelling framework provides a hypothesis-generation tool enabling to guide preclinical and clinical study design for IT-administered ASOs.
References:
[1] Geary RS et al. Adv Drug Delivery Rev (2015) 87:46–51.
[2] Biliouris K et al. CPT Pharmacometrics Syst Pharmacol (2018) 9:581-592.
[3] Luu KT et al. J Clin Pharmacol (2017) 57:1031-1041.
[4] D’Argenio DZ et al. ADAPT 5 User’s Guide: PK/PD Systems Analysis Software. Biomedical Simulations Resource, Los Angeles, 2009.
Reference: PAGE () Abstr 9501 [www.page-meeting.org/?abstract=9501]
Poster: Drug/Disease Modelling - Absorption & PBPK