First inter-laboratory comparison exercise for the determination of anticancer drugs in aqueous samples.

The results of an inter-laboratory comparison exercise to determine cytostatic anticancer drug residues in surface water, hospital wastewater and wastewater treatment plant effluent are reported. To obtain a critical number of participants, an invitation was sent out to potential laboratories identified to have the necessary knowledge and instrumentation. Nine laboratories worldwide confirmed their participation in the exercise. The compounds selected (based on the extent of use and laboratories capabilities) included cyclophosphamide, ifosfamide, 5-fluorouracil, gemcitabine, etoposide, methotrexate and cisplatinum. Samples of spiked waste (hospital and wastewater treatment plant effluent) and surface water, and additional non-spiked hospital wastewater, were prepared by the organising laboratory (Jožef Stefan Institute) and sent out to each participant partner for analysis. All analytical methods included solid phase extraction (SPE) and the use of surrogate/internal standards for quantification. Chemical analysis was performed using either liquid or gas chromatography mass (MS) or tandem mass (MS/MS) spectrometry. Cisplatinum was determined using inductively coupled plasma mass spectrometry (ICP-MS). A required minimum contribution of five laboratories meant that only cyclophosphamide, ifosfamide, methotrexate and etoposide could be included in the statistical evaluation. z-score and Q test revealed 3 and 4 outliers using classical and robust approach, respectively. The smallest absolute differences between the spiked values and the measured values were observed in the surface water matrix. The highest within-laboratory repeatability was observed for methotrexate in all three matrices (CV ≤ 12 %). Overall, inter-laboratory reproducibility was poor for all compounds and matrices (CV 27-143 %) with the only exception being methotrexate measured in the spiked hospital wastewater (CV = 8 %). Random and total errors were identified by means of Youden plots.