Introduction
New medicines are approved and manufactured following basic research, nonclinical tests, and clinical tests. The efficacy and toxicity of the drug candidate compounds are evaluated in both nonclinical and clinical tests. However, toxicities not observed in clinical trials may appear after the medicines are sold. For example, serious side effects of the combination of sorivudine and 5-fluorouracil were reported in the 1990s after sorivudine was made available for medical consumption. These unexpected toxicities are due to drug–drug interactions (DDI) caused by the concomitant administration of multiple drugs for effective healing.(Diasio, (1998)) All of the countless combinations of the new drugs with other drugs are impossible to be evaluated during pre-production testing. Therefore, a detailed understanding of pharmacokinetics with DDI is necessary to reduce the risks associated with drug combinations.(Beijnen and Schellens, (2004)) However, due to the species difference between experimental animals and human, experimental animals cannot adequately reproduce the reactions on humans. Therefore, it becomes difficult to accurately predict and understand the pharmacokinetics of drug–drug interactions in the human body through animal testing.(Dehne et al., (2017))
To address this, pharmacokinetic (PK) models have been actively utilized. The prediction of pharmacokinetics in vivo is made possible by mathematical models of variability over time of drug concentration through absorption, distribution, metabolism, excretion, and accumulation in tissues and cohesion with drug proteins. Furthermore, efficacy and toxicity can be predicted effectively from the drug dose by combining PK and pharmacodynamic (PD) models (PK–PD), thereby representing the relationship between drug concentration and physiological effects.(Abaci and Shuler, (2015)) A PK–PD model is established based on drug-specific parameters obtained from in vitro tests using cultured cells and microsomes and then evaluated by comparing with results of animal and clinical tests.(Prantil-Baun et al., (2018)) However, data from clinical tests are limited to concentrations of the compounds, metabolites in blood or excreta, drug efficacy to disease, and side effects. Additionally, the drug accumulation in organs and tissues and physiological efficacy such as activation and deactivation could be evaluated by animal tests. However, the detailed evaluation of pharmacokinetics remains challenging because data from animal tests are discrete, and it is difficult to continuously observe the efficacy and drug concentration. Regarding in vitrotests using cultured cells, in vivo environments are not sufficiently reproduced, and the co-culture of multiple organ model cells is challenging. Therefore, novel cell-based assay systems, which supplement conventional in vitro tests and animal tests, are essential for the accurate prediction and increased understanding of pharmacokinetics of DDI.(Ishida, (2018))
Microfluidics-based in vitro culture models such as organs-on-a-chip (OsoC) and microphysiological systems (MPS) have recently been considered as novel cell-based assay systems for pharmacokinetic research.(Bhatia and Ingber, (2014); Esch et al., (2015); Marx et al., (2017)) To evaluate organ interactions in vitro , several organ model cells are cultured in different compartments connected by microchannels on MPS. Parameters of MPS, such as flow ratio between organ models, residence time of circulating medium in the organ parts, and ratio of the cell number to medium volume, accommodate the PK model by the design of microchannels.(Abaci and Shuler, (2015)) Therefore, MPS may be useful to evaluate the PK model and examine drug-specific physiological effects.(Lee et al., (2017)) The usefulness of combining the PK model and MPS has been demonstrated by the consistency between the drug concentration calculations obtained from the PK model and the experimental results obtained using MPS.(Sung et al., (2010)) In addition, the calculated parameters using MPS experimental results were more similar to those of in vivo than to those of conventional in vitro tests.(Lee et al., (2019)) The combination of the PK model and MPS has also been useful for research involving not only drugs but also substances, such as glucose, in the body.(Lee et al., (2019)) However, previous studies have not undertaken the study of concomitant administration with multi drugs using MPS; thus, the usefulness for DDI studies using PK models and MPS has yet to be shown. Therefore, we estimated drug-specific parameters such as extraction ratios and subsequently evaluated the changes in drug efficacy due to DDI. We demonstrate the usability of the combination of the PK model and MPS in this DDI study. We further propose a multi-organ microphysiological system (MO–MPS), with a liver part as the metabolic model and a cancer part as the drug target model, and a PK–PD model describing the MO–MPS. A prodrug, CPT-11, was used to evaluate the drug efficacy of the metabolite in the liver part of the MPS. Drug-specific parameters were estimated by using the proposed PK–PD model from the results of drug efficacy tests by varying the flow ratio to liver part and lung cancer part in the MPS. The DDI were evaluated by comparing the results of the concomitant administration experiment using the MPS and the results of simulation using the proposed PK–PD model with the estimated parameters. We found that it was possible to evaluate and understand DDI by comparing the results of the PK-PD model and the MO-MPS.
Materials and Methods