Data-Driven Prediction of Nanoparticle Biodistribution from Physicochemical Descriptors

Area of Science:

• Nanotechnology and Materials Science
• Pharmacokinetics and Drug Delivery
• Computational Toxicology

Background:

• Nanoparticles possess unique properties influencing biological behavior, but predicting their fate remains challenging.
• Current physiologically based pharmacokinetic (PBPK) models often rely on animal data, limiting non-animal approaches.
• Accurate prediction of nanoparticle biodistribution is crucial for safety and efficacy in various applications.

Purpose of the Study:

• To develop a predictive framework for nanoparticle biodistribution using physicochemical properties and PBPK modeling.
• To integrate quantitative structure-activity relationships (QSAR) and multivariate linear regression (MLR) with PBPK models.
• To establish a non-animal alternative for early-stage nanoparticle evaluation and risk assessment.

Main Methods:

• Integrated PBPK modeling with QSAR principles and MLR, using biodistribution data from mice.
• Focused on nondissolvable nanoparticles, employing Bayesian analysis and Markov chain Monte Carlo simulations.
• Utilized physicochemical properties (zeta potential, size, coating) as key predictors.

Main Results:

• The MLR-PBPK framework achieved high predictive accuracy for kinetic indicators (adjusted R² up to 0.9).
• Successfully simulated nanoparticle biodistribution across 18 experiments, identifying key influential properties.
• Zeta potential, size, and coating were the most significant physicochemical predictors of nanoparticle behavior.

Conclusions:

• The developed framework offers a robust, non-animal method for predicting nanoparticle biodistribution and pharmacokinetics.
• This approach supports safe and sustainable by design (SSbD) principles in nanoparticle development.
• The study provides a valuable tool for early-stage nanoparticle assessment, though further refinement with larger datasets is suggested.