ANDROMEDA by Prosilico and log D outperform human hepatocytes for the prediction of intrinsic hepatic metabolic clearance of carboxylic acids

ABSTRACT Introduction Extrahepatic metabolism/conjugation, deconjugation of their metabolites, and low and varying unbound fraction in plasma (f u ), is characteristic for carboxylic drugs. Thus, it is comparably difficult to estimate their in vivo intrinsic hepatic metabolic clearance (CL int ) and hepatic CL (CL H ) and to predict their in vivo CL int , CL H and CL. One objective was to investigate the laboratory variability of f u and CL int for carboxylic acids. Another objective was to compare human hepatocytes, measured log D and the software ANDROMEDA with regards to prediction of human in vivo CL int of carboxylic acids. Materials and Methods Measured unbound hepatocyte CL int , non-renal CL (surrogate for CL H ), non-renal CL int (surrogate for hepatic metabolic CL int ), log D and f u data were taken from studies in the literature. ANDROMEDA (by Prosilico; version 1.0) prediction software was used for in silico predictions of CL int for carboxylic acids not used in the training set of its CL int -model. Results and Discussion Mean and maximum differences between highest and lowest reported in vivo CL int predicted from hepatocyte CL int were 210- and 1,476-fold (n=8), respectively. Corresponding estimates for in vitro f u were 19- and 50-fold, respectively. The data set with the apparently highest number of carboxylic acids contains 39 carboxylic acids with in vitro CL int and log D (both measured at the same laboratory), in vivo CL int and in vitro f u . 18 carboxylic acids were excluded as their in vitro CL int was below the limit of quantification. The correlation coefficient (R 2 ) for log hepatocyte predicted in vivo CL int vs log in vivo CL int was 0.34. The corresponding R 2 for log D vs log in vivo CL int was 0.40 (0.47 for 64 carboxylic acids). The Q 2 (forward-looking R 2 ) for in silico (ANDROMEDA) predicted and measured log in vivo CL int for 12 carboxylic acids was 0.86. The corresponding R 2 for hepatocytes and log D were 0.67 and 0.66, respectively. ANDROMEDA produced a lower maximum prediction error compared to hepatocytes and also predicted the in vivo CL int for all carboxylic acids out of reach for the hepatocyte assay. Conclusion Very large interlaboratory variability was demonstrated for plasma protein binding and hepatocyte assays. Log D, and especially ANDROMEDA, outperformed the hepatocyte assay for the prediction of CL int of carboxylic acids in vivo in man.