Alzheimer’s disease (AD) is an irreversible, progressive brain disorder that destroys memory and cognitive function. There are currently no FDA approved disease-modifying treatments, and the reasons for the high failure rate of clinical trials in this area (99.6%: Cummings et al. 2014) are not well understood. Incomplete understanding of disease molecular mechanisms, the timing of treatment, insufficient drug exposure to and lack of brain penetration have shown to be the contributing factor.



  1. A quantitative systems pharmacology (QSP) model was developed describing the molecular mechanisms of plaque formation in AD as well as the mechanism-of-action for four anti-Aβ drugs, two b-site amyloid precursor protein cleaving enzyme (BACE) inhibitor drugs, and one g-secretase inhibitor.


  2. The single QSP model captures the pharmacokinetic & pharmacodynamic data obtained with all the drugs involved in this example and is able to predict changes in  different Aβ species in blood and brain, including capturing plaque reductions observed in reported clinical trials.

Alzheimer's Disease QSP Model Diagram

The Model

  • A four-compartment model was developed: (1) brain interstitial fluid (ISF), (2) cerebrospinal fluid (CSF), (3) plasma, and (4) peripheral compartment. The model includes three Aβ species: monomer, oligomer, and plaque.
  • The model simulates monomers aggregating to oligomer and oligomers aggregating to plaque. Intercompartment transport of soluble Aβ species (ie. monomer and oligomer) and drug between brain ISF, CSF, plasma, and peripheral tissues is incorporated.
  • For each drug, drug-specific parameters were set for:  PK properties, binding affinity to its target(s), and ability to induce antibody dependent cellular phagocytosis (ADCP) if needed. All biological parameters were kept identical between molecules.

Model Fits

Simulations shown to the right show the model fits calibrated to selected therapies.



  • We developed a semi-mechanistic mathematical model to describe the pharmacokinetics and their effects on different Aβ species, including plaque of four clinical anti-Aβ antibodies, two BACE inhibitors, and one g-secretase inhibitor.
  • Importantly, the model is able to describe the available long-term plaque reduction data.
  • For the two BACE inhibitors, the model considered identical mechanism of action. Doing so, it accurately captured plaque reduction for Elenbecestat but not for Verubecestat.
  • Sensitivity analysis results showed that plaque reduction in brain ISF was most sensitive to drugs that bind to plaque and triggers ADCP.
  • Future direction of the work can include: description of disease progression, simulating the effects of combination therapy, addition of tau formation and spreading, and to understand patient variability.

Model fits

Download the Case Study