Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2025 Apr 28:16:1578117.
doi: 10.3389/fphar.2025.1578117. eCollection 2025.

Prediction of pharmacokinetic/pharmacodynamic properties of aldosterone synthase inhibitors at drug discovery stage using an artificial intelligence-physiologically based pharmacokinetic model

Mengjun Zhang  1 , Keheng Wu  2 , Sihui Long  1 , Xiong Jin  1 , Bo Liu  3
Affiliations

Prediction of pharmacokinetic/pharmacodynamic properties of aldosterone synthase inhibitors at drug discovery stage using an artificial intelligence-physiologically based pharmacokinetic model

Mengjun Zhang et al. Front Pharmacol. .

Abstract

The objective of this study is to develop an artificial intelligence-physiologically based pharmacokinetic (AI-PBPK) model to predict the pharmacokinetic (PK) and pharmacodynamic (PD) properties of aldosterone synthase inhibitors (ASIs), enabling selection of the right candidate with high potency and good selectivity at the drug discovery stage. On a web-based platform, an AI-PBPK model, integrating machine learning and a classical PBPK model for the PK simulation of ASIs, was developed. Baxdrostat, with the most clinical data available, was selected as the model compound. Following calibration and validation using published data, the model was applied to estimate the PK parameters of Baxdrostat, Dexfadrostat, Lorundrostat, BI689648, and the 11β-hydroxylase inhibitor LCI699. The PD of all five compounds was predicted based on plasma free drug concentrations. The results demonstrated that the PK/PD properties of an ASI could be inferred from its structural formula within a certain error range, providing a reference for early ASI lead compounds screening and optimization. Further validation and refinement of this model will enhance its predictive accuracy and expand its application in drug discovery.

Keywords: AI-PBPK; CYP11B2; PK/PD simulation; aldosterone synthase inhibitors; machine learning.

PubMed Disclaimer

Conflict of interest statement

Author KW was employed by Yinghan Pharmaceutical Technology (Shanghai) Co., Ltd. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Figures

None
Steps for predicting a compound’s PK using the AI-PBPK model and predicting enzyme inhibition by combining it with the Emax model are shown. In Step 1, the compound’s structural formula is input into the AI model to generate key ADME parameters and physicochemical properties of the compound. In Step 2, these parameters are used in the PBPK model to predict PK profiles of the compound. In Step 3, a PD model is developed to predict the inhibition rate of aldosterone synthase and 11β-hydroxylase based on the plasma free concentration of the drug.
FIGURE 1
FIGURE 1
Biosynthetic pathways of aldosterone and cortisol.
FIGURE 2
FIGURE 2
Workflow for predicting PK profiles of compounds using the AI-PBPK platform.
FIGURE 3
FIGURE 3
Observed and predicted (before and after calibration) plasma drug concentrations of 2.5 mg Baxdrostat over time.
FIGURE 4
FIGURE 4
Observed versus predicted PK profiles of Lorundrostat and Dexfadrostat at different dosages. (A) Simulated and observed profiles for a single dose of Lorundrostat (5, 10, 20, 50, 100, 200, 400, 800 mg). (B) Simulated and observed profiles of Lorundrostat at multiple doses (40 mg, 120 mg, 360 mg). (C) Simulated and observed profiles for a single dose of Dexfadrostat (1 mg, 2 mg, 4 mg, 8 mg, 12 mg, 16 mg). (D) Simulated and observed profiles of Dexfadrostat at multiple doses (4 mg, 8 mg, 16 mg).
FIGURE 5
FIGURE 5
Logarithmic coordinate plots of predicted versus observed values of AUC0-24 and Cmax for Baxdrostat, Dexfadrostat and Lorundrostat. The solid line represents the linear regression fit of the data. (A) Predicted versus observed AUC0-24 with 2-fold and 0.5-fold deviation lines. (B) Predicted versus observed Cmax with 2-fold and 0.5-fold deviation lines.
FIGURE 6
FIGURE 6
Predicted plasma drug concentration versus time curves for five ASIs at different dosages. (A) Blood drug concentration-time curves of Baxdrostat at SAD. (B) Blood drug concentration-time curves of Baxdrostat at MAD. (C) Blood drug concentration-time curves of BI689648 at SAD. (D) Blood drug concentration-time curves of BI689648 at MAD. (E) Blood drug concentration-time curves of Dexfadrostat at SAD. (F) Blood drug concentration-time curves of Dexfadrostat at MAD. (G) Blood drug concentration-time curves of LCI699 at SAD. (H) Blood drug concentration-time curves of LCI699 at MAD. (I) Blood drug concentration-time curves of Lorundrostat at SAD. (J) Blood drug concentration-time curves of Lorundrostat at MAD.
FIGURE 7
FIGURE 7
Predicted inhibition rate versus time curves for five ASIs at different dosages. (A) Enzyme inhibition over time for Baxdrostat at SAD. (B) Enzyme inhibition over time for Baxdrostat at MAD. (C) Enzyme inhibition over time for BI689648 at SAD. (D) Enzyme inhibition over time for BI689648 at MAD. (E) Enzyme inhibition over time for Dexfadrostat at SAD. (F) Enzyme inhibition over time for Dexfadrostat at MAD. (G) Enzyme inhibition over time for LCI699 at SAD. (H) Enzyme inhibition over time for LCI699 at MAD. (I) Enzyme inhibition over time for Lorundrostat at SAD. (J) Enzyme inhibition over time for Lorundrostat at MAD.
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Andersen K., Hartman D., Peppard T., Hermann D., Van Ess P., Lefkowitz M., et al. (2012). The effects of aldosterone synthase inhibition on aldosterone and cortisol in patients with hypertension: a phase II, randomized, double-blind, placebo-controlled, multicenter study. J. Clin. Hypertens. (Greenwich) 14 (9), 580–587. 10.1111/j.1751-7176.2012.00667.x - DOI - PMC - PubMed
    1. Arafat M., Sarfraz M., AbuRuz S. J. P. (2021). Development and in vitro evaluation of controlled release Viagra® containing Poloxamer-188 using Gastroplus™ PBPK modeling software for in vivo predictions and pharmacokinetic assessments. Pharm. Basel, Switz. 14 (5), 479. 10.3390/ph14050479 - DOI - PMC - PubMed
    1. Awosika A., Cho Y., Bose U., Omole A. E., Adabanya U. (2023). Evaluating phase II results of Baxdrostat, an aldosterone synthase inhibitor for hypertension. Expert Opin. Investig. Drugs 32 (11), 985–995. 10.1080/13543784.2023.2276755 - DOI - PubMed
    1. Azzam K. A. (2023). SwissADME and pkCSM webservers predictors: an integrated online platform for accurate and comprehensive predictions for in silico ADME/T properties of artemisinin and its derivatives. Kompleksnoe Ispolzovanie Mineralnogo Syra= Complex use mineral Resour. 325 (2), 14–21. 10.31643/2023/6445.13 - DOI
    1. Balhara A., Kale S., Singh S. (2022). “Physiologically based parmacokinetic (PBPK) modelling,” in Computer aided pharmaceutics and drug delivery: an application guide for students and researchers of pharmaceutical sciences. Editor Saharan V. (Singapore: Springer Nature Singapore; ), 255–284.

LinkOut - more resources