*2.7. Validation*

Method A was used for automated real-time quantitation of docetaxel and validated in terms of selectivity, carry-over, lower limit of quantitation (LLOQ), calibration function, accuracy, and precision following the recommendations of the European Medicines Agency's (EMA) guideline on bioanalytical method validation [15]. All calibration (CAL) and quality control (QC) samples were freshly prepared in serum-free model differentiation medium as sample diluents.

Selectivity and carry-over: The guidelines require the analysis of matrix from four different lots. Since the matrix was artificial, no remarkable differences had to be considered. Thus, only one batch was used for assessing selectivity. Blank samples (serum-free model differentiation medium) were

analyzed and compared with samples spiked with docetaxel at the LLOQ. Less than 20% detector response of the LLOQ is required for the blank samples. Carry-over was determined by analyzing blank samples after the injection of a high concentration QC (HQC) sample (7,500 ng/mL). Again, less than 20% detector response of the LLOQ is required to comply with the EMA guidelines.

Lower limit of quantitation and calibration: The LLOQ needs to be determined with sufficient accuracy and precision and with at least 5 times higher detector response than a blank sample. For evaluation, matrix-matched samples of 0.1, 0.25, 0.5, 1.0, 5.0 ng/mL were investigated. Additionally, the limit of detection (LOD) was determined based on calculations according to ICH guidelines [16]. Calibration samples in medium ranged from the LLOQ of 0.001 µg/mL to the upper limit of quantitation (ULOQ) of 10 µg/mL. In addition to an analyte free matrix sample, eight levels of calibration samples were prepared in triplicate and analyzed on two consecutive days.

Accuracy and precision: Accuracy and precision were assessed on serum-free medium samples spiked with docetaxel at 4 different QC levels with 5 replicates per level in a concentration range from the LLOQ to the ULOQ covering the calibration range. Samples were analyzed on two different days. Mean concentrations and the coefficient of variation (CV) of QC samples were required to be within ±15% in general, or ±20% at the LLOQ of the nominal concentrations, respectively. Within-run and between-run accuracy and precision were determined.

### *2.8. Sample Preparation for the Identification of Degradation Products*

The degradation products of docetaxel were analyzed in the differentiation medium cultivated with the models (Table 2). To handle these samples, a protein precipitation procedure was performed. Aliquots of 100 µL of the samples were added to 400 µL of cold acetonitrile and centrifuged at 3328× *g* for 10 min. The serum-free supernatant was then transferred into LC-MS/MS vials for further analysis, according to method B.


**Table 2.** LC-MS data of docetaxel (highlighted gray) and postulated degradation products, acquired using method B.
