Prediction of pharmacokinetic/pharmacodynamic properties of aldosterone synthase inhibitors at drug discovery stage using an artificial intelligence-physiologically based pharmacokinetic model
Objective:
This study aimed to develop an artificial intelligence–physiologically based pharmacokinetic (AI-PBPK) model to predict the pharmacokinetic (PK) and pharmacodynamic (PD) profiles of aldosterone synthase inhibitors (ASIs). The goal was to facilitate early-stage drug discovery by enabling the selection of candidates with high potency and selectivity.
Methods:
An AI-PBPK model was constructed on a web-based platform, combining machine learning techniques with a traditional PBPK framework to simulate the PK behavior of ASIs. Baxdrostat, selected as the reference compound due to its extensive clinical data, was used for model calibration and validation based on published sources.
Once validated, the model was applied to predict PK parameters for Baxdrostat, Dexfadrostat, Lorundrostat, BI689648, and the 11β-hydroxylase inhibitor LCI699. PD effects were estimated using predicted plasma free drug concentrations.
Results:
The model successfully predicted the PK/PD characteristics of the five compounds, demonstrating that these properties can be inferred from molecular structure within an acceptable margin of error. This approach offers a valuable tool for screening and optimizing ASI lead compounds during early drug development.
Conclusion:
The AI-PBPK model provides a promising framework for the prediction of ASI pharmacokinetics and pharmacodynamics based on chemical structure. Continued refinement and validation are expected to improve its predictive performance and broaden its applicability in drug discovery workflows.