Development and Application of a Semi-Mechanistic Model for Modulation of Amyloid-beta in Cerebrospinal Fluid after Inhibition of γ-secretase.
Huub J Kleijn(1), Jeremy Hing(2), Jing Su(3), Rik de Greef(1), Greg Winchell(4), Laura B Rosen(3), Julie A Stone (4)
(1) Merck Research Laboratories, MSD, Oss, The Netherlands; (2) Merck Research Laboratories, Rahway, NJ, USA; (3) Merck Research Laboratories, Upper Gwynedd, PA, USA; (4) Merck Research Laboratories, West Point, PA, USA
Objectives: Establish the relation between drug exposure and amyloid-beta modulation in cerebrospinal fluid (CSF) to enable optimization of drug development strategies and benchmarking across drugs targeted for Alzheimer’s disease.
Methods: Data on amyloid-beta concentrations in CSF (sampled from the lumbar region) in healthy volunteers (n=47) upon single dose treatment with placebo or the γ-secretase inhibitor MK-0752 were available [1]. The relation between plasma PK and CSF PK was explored. MK-0752 brain concentration profiles, derived from individual plasma kinetics scaled by the individual CNS/plasma partitioning constant from the CSF PK model, were used as the driver for amyloid-beta modulation. Direct and indirect effect models were explored to describe amyloid-beta modulation, and transit compartments were included to address the delay between brain and lumbar CSF amyloid-beta. Linear and non-linear models were explored to account for the substantial baseline drift in CSF amyloid-beta. Data were fitted using the FOCE method in NONMEM VII.
Results: CSF exposure was best described with an extended link model connecting a CSF compartment via a transit compartment to the central plasma compartment. Amyloid-beta CSF concentrations could not be fitted with an indirect effect model; amyloid-beta turn-over and CSF flow were indistinguishable. A direct effect model combined with a set of transit compartments best described the amyloid-beta modulation. Population parameter estimates for Emax and IC50 for inhibition of γ-secretase were 0.86 and 14.1 μM, respectively. The delay between brain amyloid-beta and lumbar CSF amyloid-beta was addressed by five consecutive transit compartments and a transit rate constant of 0.48 h-1. A Hill function with time accounted for baseline drift effects.
Conclusions: A model structure and approach for model-based interpretation of CSF amyloid-beta data was developed. The time-courses for amyloid-beta in CSF following placebo and MK-0752 treatment were well characterized. Description of the baseline drift with an empirical time-varying model facilitated a more precise estimation of Emax and IC50 as well as treatment effect size.
References:
[1] L.B. Rosen, J.A. Stone, et al. The Gamma Secretase Inhibitor MK-0752 Acutely and Significantly Reduces CSF AB40 Concentrations in Humans. Oral presentation at 10th International Conference on Alzheimer's Disease and Related Disorders (July 16-20, 2006, Madrid, Spain).
