A single 24-hour plasma sample does not predict the carboplatin AUC from carboplatin-paclitaxel combinations or from a high-dose carboplatin-thiotepa-cyclophosphamide regimen.

We studied the performance of a limited sampling model, based on a single 24-h sample (the Ghazal-Aswad model). in 52 patients who received carboplatin in two different chemotherapy regimens (a carboplatin-paclitaxel combination and a high-dose carboplatin-thiotepa-cyclophosphamide combination).

Although easy and attractive to use, the Ghazal-Aswad formula is not precise enough to predict the carboplatin AUC, and needs to be evaluated prospectively in other patient populations.

The measured mean AUC in our population was 4.1 min x mg/ml (median 3.9, range 1.9 6.3, SD 1.0 min x mg/ml). With the limited sampling model, the predicted mean AUC was 4.4 min x mg/ml (median 4.2, range 2.4-8.4, SD 1.2 min x mg/ml). Statistical analysis revealed that the model was slightly biased (MPE%, 6.5%), but imprecise (RMSE%, 20.6%) in our study population.

It has been observed that the area under the free carboplatin concentration in plasma ultrafiltrate versus time curve (AUC) is related to toxicity and tumour response. For this reason, it can be important to measure the carboplatin AUC and subsequently adjust the dose to achieve a predefined target AUC. The use of limited sampling strategies enables relatively simple measurement and calculation of actual carboplatin AUCs.