Model choice for teicoplanin kinetics in man.

Teicoplanin is a new long half life antibiotic, characterized by a polyexponential profile. A pharmacokinetic study was made in healthy volunteers receiving 15, 20 and 25 mg/Kg iv doses of Teicoplanin. Plasma concentration-time data were fitted by a polyexponential equation, consisting of two, three, four and five terms. The software was PCNONLIN, using the Gauss-Newton method with Harley's and Levenberg's modification and 1/Y2 as a weighting factor, where Y is the predicted concentration. The increase of the number of exponential terms from two to five resulted in a continuous minimization of the sum of the weighted squared residuals. To test whether or not this sum had been sufficiently reduced to justify the fitting with additional exponential terms, Akaike, Gallant and F-ratio test criteria were used. The four exponential equation best fit most of the data sets. However, the estimates of the main parameters (AUC, V, Vss, CL, Css min, Css max, number of doses to reach 95%ss) calculated according to tri and four-exponential models did not significantly differ. In the dose range studied teicoplanin pharmacokinetics are linear. In conclusion, an additional exponential term in the equation can significantly improve the best fit of teicoplanin plasma curves according to the above criteria, but it may not lead to a significant variation of the main pharmacokinetic parameters and derived values. This indicated that for clinical purposes a tri exponential model suffices for describing the kinetics of Teicoplanin in man.