Quantitative Analysis of Amorphous Form in Indomethacin by Near Infrared Spectroscopy Combined with Partial Least Squares Regression Analysis
Mingdi Liu,Rui Fu,Jichao Liu,Ping Song,Haichao Li,Weibing Dong,Zan Sun
DOI: https://doi.org/10.3390/molecules29225290
IF: 4.6
2024-11-09
Molecules
Abstract:Indomethacin (INDO) is a synthetic non-steroidal antipyretic, analgesic, and anti-inflammatory drug that commonly exists in both amorphous and crystalline states. Its amorphous state (A-INDO) is utilized by pharmaceutical companies as an active pharmaceutical ingredient (API) in the production of INDO drugs due to its higher apparent solubility and bioavailability. The crystal state also encompasses various crystal forms such as the α-crystal form (α-INDO) and γ-crystal form (γ-INDO), with the highly crystalline and insoluble γ-INDO being commercially available. A-INDO, existing in a thermodynamically high-energy state, is susceptible to several factors during the preparation, storage, and transportation of API leading to its conversion into γ-INDO, thus impacting the bioavailability and efficacy of INDO drugs. Therefore, quantitative analysis of the A-INDO/γ-INDO content in INDO API becomes essential for controlling the production quality of INDO. The primary objective of this study is to investigate the feasibility of NIR for the quantitative analysis of A-INDO in INDO API, and to further elucidate its quantitative analysis mechanism. The NIR spectral data were collected for A-INDO and γ-INDO binary mixture samples with different resolutions, and these spectra were then selected and reconstructed using the interval partial least square (iPLS) method. Different pretreatment methods were employed to enhance the reconstructed spectra by highlighting relevant eigen information while eliminating invalid information caused by environmental factors or physical characteristics of samples. The most suitable PLSR model for quantitative analysis of A-INDO within the range of 0.0000–100.0000% w/w% was established, screened, and validated. From various perspectives, including distribution of spectral effective information, impact of resolution on PLSR model performance, variance contribution/cumulative variance contribution of PLSR model principal components (PCs), PCI loadings, relationship between spectral scores, and A-INDO content, feasibility assessment was conducted for the quantitative analysis of A-INDO in INDO using NIR spectroscopy. Additionally, a detailed investigation on the quantitative analysis mechanism of the optimal PLSR model was undertaken including the correlation between the characteristic peaks of spectra and information regarding hydrogen groups or hydrogen bonds in A-INDO or γ-INDO molecules. This study aims to provide theoretical support for the quantitative analysis of A-INDO in INDO API as well as serve as a reliable reference method for API quantification and quality control in similar drugs.
chemistry, multidisciplinary,biochemistry & molecular biology