Determination of enzyme or binding constants using generalized linear models, with particular reference to Michaelis-Menten models.
The estimation of Michaelis-Menten pharmacokinetic parameters in patients with epilepsy receiving phenytoin continues to be a vexing problem. The various approximate methods suggested in the literature have serious shortcomings, primarily due to the role of the error term in the statistical model. In this report we present an accurate statistical approach using the Generalized Linear Interactive Modeling (GLIM) computer package developed by the Numerical Algorithms Group, Oxford, U.K. There are several advantages to this model: a meaningful error term can be maintained by means of a link function, the model can incorporate within-subject and between-subject variables, and additional potential explanatory variables can be added to the model. The method is applied to predicting serum phenytoin levels of pregnant women monitored at monthly intervals during pregnancy and for two to five months after pregnancy. Michaelis-Menten parameters are estimated for each women and compared.