Pharmacokinetic analyses with RUGFIT: an interactive pharmacokinetic computer program

RUGFIT is an interactive computer program to fit polyexponential equations to experimental points without the need of assumptions about the number of exponential terms and without the need of initial parameter estimates. The fitting is performed by iterative stripping minimizing the sum of squared residuals. This procedure leads to multiple model equations describing the profile of the y-value in time, that are compared with the test of Boxenbaum et al. The examples analysed in this paper are: a plasma decay curve after an intravenous bolus injection and one after an oral gift, a dA/dt versus time curve with a lag time and a curve possibly reflecting an entero-hepatic recirculation. The outcomes are, where possible, compared with the outcomes obtained with the computer programs CFT3, NONLIN, and NONLINEAR (SPSS). The latter three programs sometimes run into a local minimum, as the outcomes obtained with RUGFIT were close to the best ones obtained with the other programs. The analyses with RUGFIT show that a choice of an inadequate model equation can lead to a marked error in the value of the area under the curve that may result in a marked error of, for instance, the bioavailability. The procedures of RUGFIT are illustrated in the Appendix. RUGFIT also contains options to calculate microparameters, to simulate dosage regimes, to calculate exponents and intercepts from rate constants and to analyse time effect relationships.     

Automated pharmacokinetic analysis: Experiences of a user

Two curve-stripping and three nonlinear regression computer programs currently in use for automated pharmacokinetic analysis were tested alone and in combination on their suitability to solve a number of pharmacokinetic problems. The five programs, NONLIN, NONLINEAR, CFT3, ESTRIP, and RUGFIT (all non-commercial and readily available from the particular authors), were compared on these features: reliability, robustness, user convenience, and availability on macro or micro computers. The results show that if reliability and robustness are considered only, a combination of the curve-stripping program ESTRIP with the nonlinear regression program CFT3 operates better than the other combinations. This combination is even more attractive if, in addition, user convenience and availability on micro computers are taken into account.     

Nonlinear least-squares regression analysis by a simplex method using differential equations containing Michaelis-Menten type rate constants

Computer curve fittings were carried out to observed data as well as theoretically generated plasma concentrations of several drugs, using differential equations which contained nonlinear Michaelis-Menten type rate constants to discuss problems of initial parameter estimation in pharmacokinetic analysis. Calculation based on two different algorithms, each carried out by using SIMP (simplex method) and NONLIN (modified Gauss-Newton method) produced similar results. However, occasional divergence or unreasonable solutions occurred in a later case, when assumed values of Km and Vmax were used as initial parameters. A combined use of SIMP and NONLIN in which calculated values by SIMP were used as initial values for NONLIN, was shown to be effective to analyse plasma concentration data of indocyanine green bearing difficulty in estimating initial values. It is suggested that the successive method is useful for the curve fitting of plasma concentration with nonlinear pharmacokinetic rate processes.