2019 - Stockholm - Sweden

PAGE 2019: Drug/Disease modelling - CNS
Tatiana Karelina

Amyloid β in vitro prion properties and in vivo pathology exacerbation mechanisms studied by translation quantitative systems pharmacology model for prediction of anti-protofibril immunotherapy effects

Tatiana Karelina

InSysBio LLC, Moscow

Objectives: Clinical trials for Alzheimer’s disease (AD) are focused at present time mostly at clearing different forms of accumulated amyloid (Ab). Complicated relationship between different markers of Ab pathology (soluble vs insoluble forms) requires mechanistic understanding of their formation and accumulation. Translational quantitative systems pharmacology (QSP) model of Ab distribution and aggregation has been developed and published earlier [1]. However, it has some limitations: it did not consider HWM oligomers (protofibrils) as intermediate step in aggregation; disease progression in mouse and humans was described mostly through empirical time-dependent functions independently and simultaneously influencing multiple processes (Ab degradation, production, nucleation). While some external age-dependent mechanisms driving Ab pathology may exist in human, in transgenic mouse Ab mutation and overexpression are the only cause of appearance of Ab plaques and thus it should be described without additional drivers. The objective of the study was to describe the behavior of the diverse amyloid species, by analysis of in vitro information on aggregation and antibody (mAb) effect; to capture the accumulation of Ab during pathology in AD subjects and preclinical models (Tg2576 mice) in the model using similar mechanisms; to study the biomarker efficacy of different types of therapies: beta-secretase inhibitors (BACEi) and immunotherapy in preclinical and clinical models.

Methods: The model describes Ab production, clearance and distribution in brain, CSF, plasma and other tissues, as well as different ways of Ab aggregation in the brain [1]. Reactions describing amyloid precursor (APP) processing were added. Different intermediates in the amyloid polymerization process are considered in the model (dimers, protofibrils) to capture available information about amyloid aggregation. The novel approach for hetero-polymerization description was elaborated to describe information about properties of Ab40 and Ab42 during aggregation and mutual influence on formation of specific toxic species. Secondary nucleation process is taken into account, assuming that new fibril formation is favored in presence preexisting fibrillary oligomers. Multiple feedback regulations from amyloid species on the amyloid related processes (production, degradation) were considered for description of pathology in Ab overexpressing mice (Tg2576). Analogous mechanisms were tested in the model for human. Immunotherapy was assumed to influence both aggregation process and degradation of amyloid by glial cells. Mouse, transgenic mouse, healthy and AD human data on Ab species concentration in different compartments as well as available PD data have been used for verification and validation. 

Results: Model correctly reproduces the specific properties of Ab40 and Ab42 observed in vitro: difference in rate of ThT fluorescence increase at different portions of Ab40 and Ab42 with minimal fluorescence (no mature fibrils) at the equal Ab40 and Ab42 amounts. Several hypothetic feedback mechanisms were tested to explain the exacerbation of pathology in Tg2576 mouse after 5 months, and only combination of the degradation inhibition and amyloid production increase have allowed for correct description of soluble and insoluble species accumulation. The Tg2576 mouse model was validated across BACEi (verubecestat) data [2], and it correctly describes the difference between vehicle and chronic treatment: about 60% decrease for soluble Ab in the brain and 20% difference for total Ab after 12 weeks of treatment. Model satisfactorily describes the accumulation of insoluble and soluble Ab species during AD and results of published clinical trials: gamma- and beta-secretase inhibitors. Based on vitro data, mechanistic description of immunotherapy by mAb158 was elaborated. It was introduced into the integrated model and verified on the SUVR data from BAN2401 study [3] and was shown to describe concentration dependent effect of mAb on SUVR (decline by up to 0.3) for early AD population. Simulations for such a therapy predict a significant decline of protofibrils in the brain (by 50%). For the moderate AD population, no effect on SUVR is predicted, but the decrease of protofibrils is still close to 50%.

Conclusions: The QSP model allows for prediction of the dynamics of the different Ab species during pathology progression and treatment. It can serve as a tool for better understanding of the specific immunotherapy mechanism of action, based on in vitro data, and for prediction of clinical trial results. 

[1] Karelina, T. et al. (2017) CPT: Pharmacometrics & Systems Pharmacology, 6(10), 676–685.
[2] Villarreal, S. et al. (2017). Journal of Alzheimer’s Disease, 59(4), 1393–1413. 

[3] Swanson, C. J. et al. (2018). Alzheimer’s & Dementia, 14(7), P1668. 

Reference: PAGE 28 (2019) Abstr 8992 [www.page-meeting.org/?abstract=8992]
Poster: Drug/Disease modelling - CNS