Multiple Linear Regression Predictive Modeling of Colloidal and Fluorescence Stability of Theranostic Perfluorocarbon Nanoemulsions

Abstract

Perfluorocarbon nanoemulsions (PFC-NEs) are widely used as theranostic nanoformulations with fluorescent dyes commonly incorporated for tracking PFC-NEs in tissues and in cells. Here, we demonstrate that PFC-NE fluorescence can be fully stabilized by controlling their composition and colloidal properties. A quality-by-design (QbD) approach was implemented to evaluate the impact of nanoemulsion composition on colloidal and fluorescence stability. A full factorial, 12-run design of experiments was used to study the impact of hydrocarbon concentration and perfluorocarbon type on nanoemulsion colloidal and fluorescence stability. PFC-NEs were produced with four unique PFCs: perfluorooctyl bromide (PFOB), perfluorodecalin (PFD), perfluoro(polyethylene glycol dimethyl ether) oxide (PFPE), and perfluoro-15-crown-5-ether (PCE). Multiple linear regression modeling (MLR) was used to predict nanoemulsion percent diameter change, polydispersity index (PDI), and percent fluorescence signal loss as a function of PFC type and hydrocarbon content. The optimized PFC-NE was loaded with curcumin, a known natural product with wide therapeutic potential. Through MLR-supported optimization, we identified a fluorescent PFC-NE with stable fluorescence that is unaffected by curcumin, which is known to interfere with fluorescent dyes. The presented work demonstrates the utility of MLR in the development and optimization of fluorescent and theranostic PFC nanoemulsions.

Keywords

• ¹⁹F magnetic resonance imaging
• near infrared fluorescence
• design of experiments
• multiple linear regression
• perfluorocarbon nanoemulsion