Computational techniques are now being used to simulate virtually every facet of drug discovery and development. Attempts are being made to carry out a full computational evaluation at each step to avoid the failure of a large number of molecules at various stages of the drug development process. Use of computational approaches for a better understanding of the behavior of drugs in the human body has resulted in models that scale molecular events to intact organ systems and from events that occur in micro-seconds to an entire human lifetime. Mathematical models that relate the pharmacology of drug(s) in biological system(s) are known as systems pharmacology models and are usually very complex as well as complicated and are often described by hundreds of equations. Applications of existing systems pharmacology models have been limited to answer “what-if” style questions using simulations.
The overarching aim of the work carried out in this thesis was to apply pharmacometric methodologies to an existing blood coagulation systems pharmacology model for use in two different aspects of drug development. It focused on identification and evaluation of a biomarker for monitoring enoxaparin therapy and on simplification of the systems model for use in development of a population pharmacokinetic-pharmacodynamic (PKPD) model for fibrinogen.
In the first part of this thesis, simulations using the systems model identified that a clotting time test based on factor Xa (TenaCT) had the potential to be used to assess the effect of enoxaparin on the clotting system. This Xa based test produced clotting times that are likely to be manageable in the clinic and an in vitro evaluation using plasma from a single healthy volunteer demonstrated proof-of-mechanism of the test. A pilot study, designed using adaptive DP-optimality, was then carried out to evaluate the TenaCT test. The pilot study has provided a range of sets of values of the activating agent factor Xa and the co-factor Actin FS; where using any of these values would give a high probability of a successful proof-of-concept study. These sets of values were identified using a response surface modelling technique that utilized the clotting time surfaces at 4 different enoxaparin concentrations to obtain various Xa-Actin FS concentrations that determined the overall success of the TenaCT test.
In the second part of this thesis, simulations using the systems model provided a reasonable prediction of the observed concentrations of clotting factors over time in patients bitten by Australian elapid snakes. The model predictions suggested a higher consumption of factors (fibrinogen, II and IX in particular) in patients with complete venom induced consumption coagulopathy (VICC) compared to those with partial VICC. The model also predicted that snakes with “Xa-like” venoms may produce a less severe VICC than snakes with “Xa:Va like” venoms.
Time for recovery from snakebite is often based on recovery of fibrinogen. A simplification of the systems model was necessary to provide the mechanistic structure to predict recovery of fibrinogen profiles. The systems coagulation model consists of 62-states and 178 parameters. This was simplified to a 5-state model with 11 parameters using proper lumping as a model order reduction technique. This simplified 5-state model described the time course of changes in fibrinogen concentrations in response to brown snake envenoming which was comparable to that with the original 62-state model. Carrying out a structural identifiability analysis before attempting to estimate the parameters showed that 9 out of 11 parameters of the model were identifiable. Use of a population approach to estimate the parameters of the simplified model resulted in precise estimates of all the parameters. Prothrombin (factor II) and thrombin (factor IIa) seemed to play the most important role in the brown snake venom-fibrinogen relationship.
In conclusion, pharmacometric methodologies were applied to an existing blood coagulation systems pharmacology model. The methods used in this thesis can be applied to other systems pharmacology models, though further work would be required to optimize these techniques.
