Optimizing Mycophenolate Therapy in Renal Transplant Patients Using Machine Learning and Population Pharmacokinetic...

Area of Science:

• Pharmacology and Transplantation
• Computational Biology and Bioinformatics

Background:

• Mycophenolic acid (MPA) is a vital immunosuppressant for renal transplant recipients.
• Variability in MPA pharmacokinetics necessitates personalized dosing to optimize efficacy and minimize toxicity.

Purpose of the Study:

• To couple population pharmacokinetic (PopPK) modeling with machine learning (ML) for improved MPA therapy optimization.
• To identify key covariates influencing MPA pharmacokinetics and explore predictive factors for individualized dosing.

Main Methods:

• Developed two PopPK models for different MPA formulations using data from 76 renal transplant patients.
• Applied ML techniques (PCA, ensemble trees) to identify predictive factors and assess MPA concentrations.
• Utilized Monte Carlo simulations to evaluate drug exposure under varying clearance conditions.

Main Results:

• Total daily dose and post-transplant time were key covariates affecting MPA clearance.
• MPA dose, urea, and post-transplant time significantly predicted plasma MPA levels with high accuracy (R² > 0.91).
• Saliva MPA levels showed potential as a complementary monitoring tool, though plasma monitoring remains superior.

Conclusions:

• Integrating ML with PopPK modeling enhances understanding of MPA variability in renal transplant recipients.
• The developed PopPK/ML models offer a foundation for personalized immunosuppressive therapy strategies.
• This approach supports individualized MPA dosing to balance efficacy and reduce toxicity.