Hyper-spectra imaging analysis of PLGA microspheres via machine learning enhanced Raman spectroscopy

Abstract

Long-acting injectables (LAI) offer a cost-effective and patient-centric approach by reducing pill burden and improving compliance, leading to better treatment outcomes. Among various types of long-acting injectables, poly (lactic-co-glycolic acid) (PLGA) microspheres have been extensively investigated and reported in the literature. However, microsphere formulation development is still challenging due to the complexity of PLGA polymer, formulation screening, and processing, as well as time-consuming and cumbersome physicochemical characterization. A further challenge is the limited availability of drug substances in early formulation development. Therefore, there is a need to develop novel and advanced tools that can accelerate the early formulation development. In this manuscript, a novel comprehensive physicochemical characterization approach was developed by integrating Raman microscopy and the machine learning process. The physicochemical properties such as drug loading, particle size and size distribution, content uniformity/heterogeneity, and drug polymorphism of the microspheres can be obtained in a single run, without requiring separate methods for each attribute (e.g., liquid chromatography, particle size analyzer, thermal analysis, X-ray powder diffraction). This approach is non-destructive and can significantly reduce material consumption, sample preparation, labor work, and analysis time/cost, which will greatly facilitate the formulation development of PLGA microsphere products. In addition, the approach will potentially be beneficial in enabling automated high throughput screening of microsphere formulations.