A SEQUENTIAL BAYESIAN ALGORITHM FOR DOSE INDIVIDUALIZATION OF CARBOPLATIN

Carboplatin is associated with significantly less nephrotoxicity and n eurotoxicity than is cisplatin. The dose-limiting toxicity of carbopla tin is myelotoxicity. A number of dosing methods have been described t hat allow a value for the area under the concentration-time curve to b e targeted on the basis of the patient's renal function. Recently a fo rmalised analysis of the pharmacodynamic response to carboplatin revea led a therapeutic window in which the response rate was maximal and to xicity, tolerable. Optimal therapy would result from targeting this wi ndow in the individual patient. The aim of this study was to develop a Bayesian dose-individualisation method for carboplatin. The method in volved (1) development of a highperformance liquid chromatography (HPL C) method to measure serum concentrations of carboplatin; (2) a pharma cokinetic study in 12 women receiving carboplatin for ovarian cancer t o estimate the population pharmacokinetic values for this group of pat ients; (3) development of population models to describe the concentrat ion-time course of carboplatin in serum along with associated errors; and (4) development of an algorithm that uses a sequential Bayesian de sign, which enables estimation of future doses of carboplatin on the b asis of feedback from serum concentrations. The results of each of the stages were (1) the coefficient of variation of the assay was 6.3% wi thin day and 8.4% between days (r(2) = 0.9993), and the limit of detec tion was 0.25 mg/l; (2) Patients' ages ranged from 49 to 68 years, the ir weights varied from 46 to 85 kg, and their glomerular filtration ra te ranged from 3.2 to 7.4 l/h. A geometric mean clearance (Cl) of 6.8 L/h and a steady-state volume of distribution (Vss) of 221 were estima ted, which are similar to previously published data; (3) and a two-com partment model best described the data. Two error models were develope d, the first describing the error associated with the assay and the se cond, the error of the two-compartment model, i.e. error due to indivi dual variation in pharmacokinetics and error due to model mis-specific ation. Finally, (4) the development of a sequential Bayesian dose-indi vidualisation method for carboplatin is described. To our knowledge, t his is the first sequential design that has been used for dose individ ualisation of chemotherapy. The program is specific for carboplatin an d operates independently of commercially available Bayesian software. Doses predicted by this program are being tested prospectively against conventional dosing methods.