Mechanistic modeling of alpha-synuclein aggregation and brain pharmacokinetics for drug development in neurodegeneration

Mechanistic modeling plays an increasingly important role in central nervous system (CNS) drug development, but its impact depends on how well models reflect the underlying biology and account for the limitations of available data. This thesis addresses these challenges through two distinct yet complementary research areas, focused on foundational and methodological aspects of quantitative systems pharmacology (QSP) and physiologically based pharmacokinetic (PBPK) modeling. The first part of the work focuses on the mathematical modeling of alpha-synuclein (aSyn) aggregation dynamics under in vitro conditions—a molecular process relevant to Parkinson’s disease (PD) and other synucleinopathies. Building on in vitro aggregation studies and chemical kinetic models, we develop a mechanistic model of aSyn accumulation that incorporates both pure and lipid-mediated aggregation pathways. This unified framework captures key features of in vitro aggregation mechanisms, including toxic oligomer structural changes, fibril-mediated secondary nucleation, and aSyn–lipid interactions. It enables the evaluation of PD therapeutic strategies targeting different steps of the aggregation cascade, with applications in QSP. The second part tackles specific methodological limitations in evaluating a priori predictions of brain drug distribution against experimental homogenate measurements. Building on the Rodgers & Rowland method and related tissue composition-based models, we propose a homogenate-specific metric that accounts for artifacts induced by tissue homogenization. This metric serves as a mechanistically appropriate endpoint for homogenate assays, provided brain lipid composition is well characterized. The analysis contributes to improving model evaluation practices and enhancing the reliability of a priori predictions of drug partitioning in intact brain tissue, with implications for PBPK modeling of CNS-active small molecules. Although these two projects differ in biological scale and modeling approach, both target issues in advancing mechanistic models in the CNS field. Rather than delivering ready-to-use preclinical tools, the aim is to promote more biologically grounded and interpretable modeling practices—either by improving the representation of critical pathological processes or by clarifying how model predictions relate to the data used to evaluate them. By building on and complementing existing frameworks, the models developed here foster more transparent, robust, and data-aligned modeling in neurodegeneration. Ultimately, through enhanced mechanistic understanding at both the molecular and pharmacokinetic levels, this work can support therapeutic strategy design for Parkinson’s disease, with implications for target characterization and brain exposure optimization.