Validation of the limited-sampling models for carboplatin AUC in combination chemotherapy with taxanes.

We previously developed limited-sampling models (LSM) for estimation of the area under the concentration curve (AUC) of free carboplatin (Anticancer Res 17: 4571-4576, 1997). The one-point model was: AUC (mg/ml x min)=0.93 x C3h + 0.47 and the two-point model was: AUC (mg/ml x min)=0.16 x C1h + 2.26 x C8h + 0.75, in which C3h, C1h and C8h represented the plasma concentration, sampled at 3 h, 1 h and 8 h after the start of a 1-h carboplatin infusion, respectively. In this study, we tried to confirm the utility of our LSMs in combination chemotherapy with taxanes. We obtained 29 data series from 29 patients who received carboplatin in combination with docetaxel (n=16) or paclitaxel (n=13). The actual AUC was strongly correlated with the AUC calculated from both one- and two-point models (one-point model; the mean prediction error (MPE)=-4.44±2.15%, the root mean squared error (RMSE)=12.22±0.84%, two-point model; MPE=- 1.89±1.66%, RMSE=8.99±0.81%). Our LSMs proved to be unbiased and precise and are also useful for calculating carboplatin AUC in combination with taxanes. Carboplatin is a second-generation platinum analog that is less nephrotoxic, neurotoxic and emetogenic than the parent compound cisplatin. The main hematological toxicity for carboplatin is thrombocytopenia. The area under the concentration curve (AUC) of the free carboplatin is a major determinant for thrombocytopenia and the likelihood of response (1). However, a conventional pharmacokinetic study to examine the carboplatin AUC necessitates frequent blood sampling. To overcome this problem, we previously constructed a limited-sampling model (LSM) of carboplatin from lung cancer patients who received carboplatin in combination with oral etoposide (2). The one-point model gave an excellent estimation of the AUC: AUC (mg/ml x min)=0.93 x C3h + 0.47 (the mean prediction error (MPE)=4.4%, the root mean square error (RMSE)=8.9%), while the two-point model was shown to improve the bias and precision of the AUC estimation: AUC (mg/ml x min)=0.16 x C1h + 2.26 x C8h + 0.75 (MPE=0.9%, RMSE=5.3%), in which C3h, C1h and C8h represented the plasma concentration, sampled at 3 h, 1 h and 8 h after the start of a 1-h carboplatin infusion, respectively. Taxanes, docetaxel and paclitaxel are promising drugs for non-small cell lung cancer chemotherapy (3). Taxanes shift the equilibrium toward microtubule assembly and stabilize microtubules by preventing depolymerization. The main hematological toxicity for taxanes is neutropenia. The non- overlapping toxicities of carboplatin and taxanes and their different mechanisms of action have made the combination of both drugs attractive for further clinical exploration. There are some limitations to the use of LSMs, in that a model can be used only to the dose regimen which is an exact copy of the original regimen for which the model was developed; furthermore, concomitant administration of other drugs and infusions may affect the predictive value of the model (4), so that, in those cases the models should be re-evaluated and tested using a new validation data set. The purpose of this study was to confirm the utility of our LSMs for carboplatin AUC in combination chemotherapy with taxanes.