Table 1.

In Vitro to In Vivo translational model.

In Silico		In Vitro Assay	Examples of In Vitro to In Vivo Translation
			Technologies Used in Examples	Parameter	In Vitro to In Vivo Translation Result	Application in Drug Development
PBPK	Absorption	Caco-2 [81], MDCK [82], gut MPS [83]	MDCK-MDR1 and Caco-2 [84]	Obtaining half maximal inhibitory concentration (IC50) for P-gp and integrating it to models	Demonstrating non-interaction between Axitinib and P-gp substrate	Prediction of drug-drug interaction and exemption of related clinical trials
	Distribution	MDCK [85], hiPSC- brain endothelial cells [86], co-culture [87]	MDCK Ⅱ [88]	Using apparent permeability coefficient (Papp) to obtain in vitro efflux transporter-mediated clearance and scaling it to the whole-brain in vivo efflux transporter-mediated clearance	Exploring the penetration of AZD1775 across BBB	Prediction of drug distribution and target concentration
	Metabolism	Recombinant enzymes [89], microsomes [90], primary hepatocytes [91], HepG2 [92], HepaRG [93], hESC or hiPSC-hepatocytes [94], liver MPS [95]	Primary hepatocytes [96]	Inputting the intrinsic clearance (CLint) to Simcyp software to establish PBPK model	Predicting the difference of AUC in patients with different liver damage after a single oral administration of sirolimus	Prediction of drug metabolism and inter-population extrapolations
	Excretion	MDCK, CHO, HEK-293, HeLa [97], primary cultured renal tubule cells [98], renal MPS [99]	renal MPS [99]	Scaling renal clearance (CLR) based on surface area	Predicting human renal excretion for cisplatin and nicotine	Prediction of excretion
PBPK	Integrate ADME	MPS [99]	MPS [99]	Scaling intestinal permeability (Papp) based on absorptive surface, liver clearance (CLint, in vivo) based on the number of hepatocytes, renal clearance (CLR) based on surface area	Reproducing the clinical PK profiles for both nicotine and cisplatin at different doses and different routes of administration	Simulation of clinical PK profiles
PK/PD		Disease-related cell [100], 2D [80,101], 3D [102], MPS [103], organoids [104]	Six human epithelial cancer cell lines [100]	Directly combining maximal killing rate (Kmax), drug concentrations yielding 50% of Kmax (KC50) and hill index (γ) into in vivo model	Demonstrating that low doses and high dosing frequency for paclitaxel is prior to maximum tolerated doses	Dose and schedule selection
			L540cy cells, Karpas cells [99]	Integrating association and dissociation rate constants (Kon and Koff) to describe the interaction between ADC and target	Predicting therapy in clinical trials employing different dosing regimens	Clinical response prediction
			primary liver cells, red blood cells and brain homogenates [101]	Based on the total enzyme content, scaling metabolic capacity (Vmax) and clearance (CLint); Correcting bimolecular inhibition constant (Ki) considering different states of targets in vitro and in vivo	Evaluating the biotoxicity of carbaryl and other carbamates with an anticholinesterase mode of action	Toxicity prediction
			MPS [105]	Based on the number of nephrons in human kidney, scaling maximal injury rate (Emax) and drug concentrations yielding 50% of Emax (EC50) into in vivo model	Assessing renal proximal tubule injury caused by three nephrotoxic drugs	Toxicity prediction
QSP (QST)		Disease-related cell [106], 2D [106,107], 3D [107], MPS [108], organoids [109]	Primary hepatocytes [106]	Applying directly the IC50 values for the bile acid transporters to DILIsym, fitting the mitochondrial toxicity parameters (Vmax, Km) in MITOsym, and converting them to DILIsym	Explaining the liver toxicity mechanism of PF-04895162 and expound the differences of species	Characterization of target mechanism
			JIMT-1 cells in 2D or 3D and dynamic cell Culture [107]	Integrating drug inhibition or stimulation coefficient (S1p, S2p, Kp etc.) to describe signal pathway molecules perturbation	Optimizing the sequence and inter-dose interval of the three agents (paclitaxel, dasatinib, and everolimus) in the combination	Design of drug administration protocol and evaluation of drug combination
			effector T cells (Teffs), EL4 and E.G7-OVA thymoma cells [110,111,112]	Integrating rate constants defining the half-life of engagement or dissociation between cancer cells and effector T cells (CancerTEng, CancerTInt) directly into the QSP model; scaling number of CD28 receptors expressed on each T cell during priming (CD28_receptors-per-Tcell) by the number of T cell in vivo	Predicting the checkpoint inhibitors’ therapies administered as mono-, combo- and sequential therapies	Clinical response prediction

Abbreviations: Caco-2: human colon adenocarcinoma cells; MDCK: Madin–Darby canine kidney epithelial cells; MPS: microphysiological systems; hiPSC: human-induced pluripotent stem cell; PBPK: physiologically based pharmacokinetic model; PK/PD: pharmacokinetic/pharmacodynamic model; QSP: quantitative systems pharmacology model; QST: quantitative systems toxicology model; hESG: human embryonic stem cell lines; CHO: Chinese hamster ovary cells; HEK-293: human embryonic kidney cells.