In silico predictions of volume of distribution of drugs in man using conformal prediction performs on par with animal data-based models
Fagerholm U, Hellberg S, Alvarsson A, Arvidsson McShane S, and Spjuth O..
In silico predictions of volume of distribution of drugs in man using conformal prediction performs on par with animal data-based models.
Xenobiotica. 51, 12, 1366-1371. (2021). DOI: 10.1080/00498254.2021.2011471
Volume of distribution at steady state, Vss, is an important pharmacokinetic endpoint. In this study we apply machine learning and conformal prediction for Vss prediction and make a head-to-head comparison with rat-to-man scaling, allometric scaling and the Rodgers-Lukova method on combined in silico and in vitro data, using a test set of 105 compounds with experimentally observed Vss. The mean prediction error and percentage with sub 2-fold prediction error for our method were 2.4-fold and 64 percent, respectively. 69 percent of test compounds had an observed Vss within the prediction interval at a 70 percent confidence level, . In comparison, 2.2-, 2.9- and 3.1-fold mean errors and 69, 64 and 61 percent of predictions with less than 2-fold error was reached with rat-to-man and allometric scaling and the Rodgers-Lukova method, respectively. We conclude that our method has theoretically proven validity that was empirically confirmed, and showing predictive accuracy on par with animal models and superior to an alternative widely used in silico-based method. The option for the user to select the level of confidence in predictions offers better guidance on how to optimize Vss in drug discovery applications.