Vatalanib is a novel, potent, orally bioavailable angiogenesis inhibitor which can block all known vascular endothelial growth factor (VEGF) receptor tyrosine kinases. It has been shown to have potential in treating hematological malignancies, including myelodysplastic syndromes (MDS). Previous studies have revealed substantial pharmacokinetic variability that is further increased by auto-induction of its metabolism. And safety and tolerability are of great importance for such anti-angiogenic treatment. Therefore, the objectives of thesis was to characterize the pharmacokinetics and auto-induction process of vatalanib using population pharmacokinetic approaches, and to evaluate the relationship between systemic exposure to vatalanib and its toxicity in order to achieve optimal therapy in the treatment of MDS with standard fixed vatalanib doses.
In this thesis, we characterized the time dependent pharmacokinetics of vatalanib employing both the standard two-stage approach with maximum a posteriori Bayesian estimation method in Chapter 2 and the nonlinear mixed effects modeling approach in Chapter 3 with sparse sampling data from a phase II clinical study, in which adult MDS patients received vatalanib orally once daily on a 28-day cycle of therapy. A one compartment model with a time dependent oral clearance term and lagged first order absorption was determined to be the best structural model for both approaches. The mean parameter estimates from both approaches were similar, probably due to similar model structure and the same underlying model assumptions. The mean apparent oral clearance was estimated to increase to approximately 2-fold after the enzyme auto-induction. The relationship between pharmacokinetic parameters and subject specific covariates was assessed using simple linear regression analysis in Chapter 2. In Chapter 3, graphical exploration, generalized additive models, and stepwise forward addition and backward elimination approaches were utilized for covariate modeling. Results from both chapters revealed that none of the available covariates was found to significantly influence the pharmacokinetics of vatalanib.
Finally, exposure toxicity analysis was carried out to characterize the relationship between the systemic exposure to vatalanib and its toxicity. Both simple logistic regression and multinomial logistic regression approaches were utilized. Results from both analyses showed no correlation between the vatlanib exposure and the magnitude of its toxicity.