**1. Introduction**

Selecting clinically relevant doses for the evaluation of anticancer drugs remains challenging in preclinical drug development and contributes to the low translatability of effects in vitro to efficacy in patients. While the understanding of cancer biology advances as the complexity of tumor models and analytical techniques increases, the success rate of drug development in oncology remains the lowest among all therapeutic areas.

Historically, anticancer drug doses for clinical trials have been determined by extrapolating the maximum tolerated dose (MTD) in animals to the human patient. Taking the MTD as the starting point, the effective and safe dose for humans was anticipated in the range of −3 to +3, with three concentrations below and three concentrations above. The revision of this concept is urgently needed, since many nonoptimal doses were taken into late stages of drug development. Especially the testing of high-risk drugs requires a more conservative approach, using the minimum anticipated biological effect level (MABEL) in first-in-human trials [1,2].

Up to now, new concepts focused on the improved extrapolation from animal studies to clinical trials, e.g., by introducing drug metabolism and pharmacokinetic studies in early drug development [3]. In particular, model-based, adaptive/Bayesian approaches already helped to better find effective and safe dosage [4]. Nevertheless, animal models are affected by differences in the human pathophysiology and even xenograft models do not fully recapitulate the barriers of drug uptake into human solid tumors [5].

In fact, drug exposure of tumor cells depends on the architecture of solid tumors with cell density, the spatial arrangement of cells and extracellular matrix proteins, interstitial fluid pressure, and vascular supply [6–8]. While 2D monolayer cell culture cannot provide meaningful insights into the pharmacokinetic profiles of solid tumors, sophisticated 3D tumor models such as spheroids or multilayered tumor models could do this, eventually even in a patient-specific manner [9,10].

The introduction of in vitro tumor models into the dose selection for a particular patient requires adapting the protocols to high-content, high-throughput approaches to handle high numbers of tests, e.g., with different drugs and several combinations. However, analytical approaches to quantify drug amounts in tissues comprise imaging- and microdialysis-based methods. While imaging techniques and microdialysis closely map the drug distribution within (tumor) tissues, all methods share the high effort needed in sample preparation [11,12], restricting their use for personalized medicine.

Herein, the development of an in vitro approach for real-time pharmacokinetic investigations in human cell-based models of head-and-neck squamous cell carcinoma is reported. It aims for an automated measurement of docetaxel concentrations within the tumor tissue to quantify the drug absorption. Therefore, an UHPLC-MS/MS method was adapted from clinical practice and optimized for a maximum number of online measurements per time.
