To assess efficacy and toxicity of a drug in humans, it is important to measure the tissue concentration of a drug at the target site. For a drug that is transported into or out of the tissue, the tissue unbound steady-state concentration can be dramatically different from its corresponding unbound steady-state plasma concentration. Because routine measurement of drug tissue concentrations is not possible, using rosuvastatin as a model transporter substrate drug, we compared the ability of the proteomics-informed relative expression factor (REF) approach and sandwich-cultured human hepatocytes (SCH) to accurately predict rosuvastatin human hepatobiliary clearances and hepatic concentrations. REF-predicted rosuvastatin biliary clearance (CL_bile ), estimated using BCRP-overexpressing, MDR1-overexpressing, and MRP2-overexpressing vesicles, together with our previously published REF-predicted rosuvastatin hepatic sinusoidal uptake clearance (CL_uptake ) and physiologically scaled sinusoidal passive uptake and efflux clearance (CL_s,efflux ), were used to predict rosuvastatin hepatic concentrations. For SCH, the estimated rosuvastatin CL_bile , CL_uptake , and CL_s,efflux were scaled using physiological scaling. The REF-predicted CL_bile (6.39 ± 1.56 mL/minute) and hepatic rosuvastatin area under the concentration-time curve (AUC) fell within our a priori defined success criterion, i.e., within twofold of the observed positron emission tomography-imaged values. In contrast, as expected, SCH dramatically overpredicted (predicted/observed ratio P/O = 8.38-10.41) rosuvastatin CL_bile , and underpredicted hepatic AUC (P/O = 0.08-0.14). For both approaches, predictions were improved by using the parallel tube model vs. well-stirred model. Overall, using rosuvastatin as a model drug, this study demonstrates the success of the REF approach in predicting in vivo CL_bile and hepatic concentration of drugs, and highlights the shortcomings of the SCH approach in making such predictions.

	Year	Citations

About