Determination of Intrinsic Clearance and Fraction Unbound in Human Liver Microsomes and In Vitro-In Vivo Extrapolation of Human Hepatic Clearance for Marketed Central Nervous System Drugs.

Abstract

Objective: The objective of this study was to determine the apparent intrinsic clearance (Cl_{int, app}) and fraction unbound in human liver microsomes (f_{u, mic}) of 86 marketed central nervous system (CNS) drugs and to predict the in vivo hepatic blood clearance (CL_{h, b}).

Methods: Cl_{int, app} in human liver microsomes (HLM) was determined by substrate depletion, and f_{u, mic} was determined by equilibrium dialysis. The relationship between lipophilicity (logP) and unbound intrinsic clearance (Cl_{int, u}) was explored using the Biopharmaceutical Drug Disposition Classification System (BDDCS) and Extended Clearance Classification System (ECCS). The predicted hepatic blood clearance by direct scaling, conventional method and Poulin method using well-stirred (WS) and parallel-tube (PT) models were compared with observed values.

Results: The Cl_{int, app} in HLM ranged from < 5.8 to 477 µl/min/mg. The f_{u, mic} in HLM ranged from 0.02 to 1.0. The scaled Cl_int values ranged from < 5 to 4496 ml/min/kg. The metabolic rate increased with an increase in logP (logP ≥ 2.5) of the CNS compounds. The direct scaling and Poulin methods showed comparable results based on the percentage of clearance predictions within a two-fold error. The conventional method resulted in under-predictions of Cl_{int, in vivo} or CL_{h, b} using the WS or PT models. The Poulin method is favored over the other methods based on the statistical parameters.

Conclusions: Experimental Cl_{int, app} and f_{u, mic} for 86 CNS compounds were successfully determined, and the scaled clearance was used to predict the hepatic blood clearance of 34 drugs. The success of prospective clearance predictions using HLM is expected to be high for most of the lipophilic BDDCS class 1 and class 2 and ECCS class 2 CNS compounds. The Poulin method resulted in more accurate predictions falling within a two-fold error of the observed values using the WS or PT models.