The investigation on the prediction possibility of the oral bioavailability of drugs in human based on structural parameters and the bioavailability of drugs in preclinical animals

Abstract

Background: Experimental determining of the drug´s oral bioavailability in human is costly and time consuming, and usually is performed on preclinical animals. It is believed that animal data are predictive of human data; however, there are some doubts regarding the reliability of this issue. The aim of this study is to clarify whether it is possible to predict correlation between human and animal bioavailability data based on structural parameters. Method: Oral bioavailability data of drugs for human and preclinical animals (rat and dog) were collected from the literature. Structural descriptors (Abraham solvation parameters, topological polar surface area (TPSA), logarithm of partition coefficient (logP) and logarithm of distribution coefficient at pH=6.8 (logD6.8)) were calculated by ACD/Labs software. Data were divided into two classes by percentage deviation (PD) of oral bioavailability between human and preclinical animals (PD<40%: class I and PD>40%: class II). Classification-based models were used to predict the class of each drug using structural parameters. Results: The results of this study revealed that logD6.8 is the main parameter for evaluating correlation between oral bioavailability of human and animals. Moreover, the developed models by logistic regression based on logP, TPSA and Abraham solvation parameters are capable of predicting the class of a drug with 75% accuracy. Conclusion: The structural parameters and developed models in this study could be used to find the compounds which have an acceptable correlation between oral bioavailability in animals i.e., rat and dog and human.