POPULATION PHARMACOKINETIC ANALYSIS OF NEW AMINOGLYCOSIDES, ASTROMICIN AND ISEPAMICIN, AND EVALUATION OF BAYESIAN PREDICTION METHOD FOR APPROXIMATION OF INDIVIDUAL CLEARANCE OF DRUG

Pharmacokinetic data obtained from healthy subjects after a single i.v. infusion over 30 min of either astromicin (AST) or isepamicin (ISP), the newly developed aminoglycoside antibiotics, were analyzed by a computer program, NONMEM, together with those in patients having impaired renal functions of various degrees, which were cited from the literature. A two-compartment open model was utilized for the analysis, assuming that the total body clearance of drug (Cl-B) is linearly correlated with endogeneous creatinine clearance (Cl-cr). By the analysis, it was found that the body weight explains some part of interindividual variability in Cl-B of ISP, although it did not hold true in the case of AST. For each drug, the means and variances of Cl-B and the distribution volume of central compartment, and only the means of two intercompartmental constants (K-12 and K-21), all of which were obtained by the NONMEM analysis, were implemented in a Bayesian prediction program for a microcomputer. This, thus, clarified a point as to when blood sample should be collected in order to get the best prediction of individual Cl-B with the use of the above Bayesian program and the measurement of drug concentration in the sampled blood as a feedback information. For this purpose, the drug concentrations in plasma obtained in the multiple-dose study of each drug, in which the drug was administered in healthy subjects as i.v. infusion over 1 h every 12 h for 4.5 days, were used. The measurement of drug concentration in plasma at 5 h after the end of infusion and the measurement at 0 or 3 h combined with that at 5, 7 or 9 h were optimal as Bayesian feedback informations, respectively, when only one measurement and two combined measurements were allowed. The present study shows that individual pharmacokinetic parameter can be efficiently predicted by the Bayesian approach with one or two feedback data. This approach may be useful to promote an efficient and safe use of these drugs.