Amyloid beta (Aβ) peptides, and in particular Aβ42, are found in senile plaques associated with Alzheimer's disease. A compartmental model of Aβ production, exchange and irreversible loss was recently developed to explain the kinetics of isotope-labeling of Aβ peptides collected in cerebrospinal fluid (CSF) following infusion of stable isotope-labeled leucine in humans. The compartmental model allowed calculation of the rates of production, irreversible loss (or turnover) and short-term exchange of Aβ peptides. Exchange of Aβ42 was particularly pronounced in amyloid plaque-bearing participants. In the current work, we describe in much greater detail the characteristics of the compartmental model to two distinct audiences: physician-scientists and biokineticists. For physician-scientists, we describe through examples the types of questions the model can and cannot answer, as well as correct some misunderstandings of previous kinetic analyses applied to this type of isotope labeling data. For biokineticists, we perform a system identifiability analysis and a sensitivity analysis of the kinetic model to explore the global and local properties of the model. Combined, these analyses motivate simplifications from a more comprehensive physiological model to the final model that was previously presented. The analyses clearly demonstrate that the current dataset and compartmental model allow determination with confidence a single 'turnover' parameter, a single 'exchange' parameter and a single 'delay' parameter. When combined with CSF concentration data for the Aβ peptides, production rates may also be obtained.

Original languageEnglish
Pages (from-to)48-61
Number of pages14
JournalMathematical Biosciences
StatePublished - Mar 1 2015


  • Alzheimer's disease
  • Amyloid beta
  • Compartmental model
  • Identifiability
  • Kinetics
  • Sensitivity analysis


Dive into the research topics of 'Analysis of a compartmental model of amyloid beta production, irreversible loss and exchange in humans'. Together they form a unique fingerprint.

Cite this