A Mechanistic Model for Predicting Lung Inflammogenicity of Oxide Nanoparticles

Abstract

This study presents a mechanistic model for identifying oxide nanoparticles that induce a high level of neutrophils in the bronchoalveolar lavage fluid, an important marker for lung inflammogenicity. The model is based on 4 nanoparticles' physicochemical properties, ie, the reactivity, surface charge, wettability, and dissolution. First, I calculate these properties and show that theoretical values reproduce acceptably the experimental measurements. Then, I combine these properties with mechanistic knowledge to build a classification model for the prediction of acute invivo lung inflammogenicity, measured as the total number of polymorphonuclear neutrophils. The model uses reactivity and dissolution properties of nanoparticles as toxicological initiating events, whereas surface charge and wettability are characteristics involved in the interactions between the nanoparticles and the lung surfactant, eventually leading to increased cellular uptake and bioaccumulation. The model is validated on a set of 43 oxide nanoparticles tested invivo to confirm that acute lung inflammation can be described using this mechanistic framework. In addition, I also develop a linear regression model for insoluble nanoparticles to quantitatively predict the polymorphonuclear neutrophil count as a function of reactivity and surface charge. The proposed models are based on mechanistic knowledge and can support the development of adverse outcome pathways, risk assessment frameworks and safe design strategies at early stages of material's R&D.

Keywords: computational toxicology; invivo inhalation; QSAR; mechanistic toxicology; nanoparticles.