基于红外和拉曼光谱的晶型含量预测多种模型的构建及对比分析研究

With the general improvement of the polymorphism of drugs in Chinese pharmaceutical industry, the concern and demand for the quality control of drug polymorphs are increasing day by day. Infrared spectroscopy and Raman spectroscopy have been applied for quantitative analysis of crystal forms. They have the advantages of simple sample preparation, simple operation, lossless analysis and high penetration rate. Infrared spectroscopy and Raman spectroscopy combined with stoichiometry to establish the prediction model of crystal form content can effectively solve the problem of quantitative analysis of crystal form in binary or ternary complex phase system. However, this method is not included in the Chinese Pharmacopoeia 2015 edition, the main reason is the lack of systematic and in-depth research about the modeling parameters and chemometrics methods which are important to the accuracy of the prediction models..In order to remedy the above lack of the research, this project will establish the quantitative prediction models of crystal form concentrations in binary or ternary polymorphic mixtures. During the modeling process, the influences of several factors on the prediction models, such as the number of samples, the modeling range, spectral pre-processing methods, chemometrics methods and so on, will be systematically studied and compared. Finally, a process of the modeling method for quantitative analysis of the polymorphic mixture will be established, and a principle for the choices of modeling parameters will be suggested. The study will provide the scientific research data and management basis for the selection of quality control methods of polymorphic pharmaceuticals in China.

随着对晶型药物认识的普遍提高，我国对药物晶型质量控制的关注与需求日益增长。红外与拉曼光谱用于晶型定量分析，具有样品前处理简单、操作简便、无损分析、普及率高等优势。红外与拉曼光谱结合化学计量学建立晶型含量预测模型，可有效解决二元或三元复杂物相体系中的晶型定量分析问题。但这两种方法未被收录到现行中国药典的晶型指导原则中，主要原因是影响建模成效的参数与化学计量方法多元化，但相关的有针对性的系统考察与深入探讨研究却十分缺乏。.为弥补上述的研究欠缺，本项目拟依托6个药物混晶体系，采用红外与拉曼两种互补的光谱技术，在建立晶型含量定量检测模型的基础上，系统考察并综合分析样本数量、建模区间、光谱预处理方法、化学计量算法等等多种因素对定量模型的影响程度，建立适用于药物晶型含量预测的流程化模型构建方法，提出建模参数选择与优化的原则及建议，为我国晶型药物质量控制方法的选择，提供科学的研究数据和科学的管理依据。

随着对晶型药物认识的普遍提高，我国对药物晶型质量控制的关注与需求日益增长；同时随着共晶药物（属晶型药物范畴）研究成为新的热点方向，晶型药物的定量质量控制面临着更复杂、更多元的混晶体系难题。红外与拉曼光谱结合化学计量学建立晶型含量预测模型，可有效解决多元复杂物相体系中的晶型定量分析问题。.本项目采用红外光谱、拉曼光谱两种技术方法，分别建立了7种晶型/共晶混晶体系、共计14个含量预测模型，涉及的体系包括：①米非司酮2种晶型混晶体系（单一两相体系）；②吡罗昔康2种晶型混晶体系（单一两相体系）；③吡嗪酰胺3种晶型混晶体系（单一三相体系）；④柚皮素卡马西平共晶混合体系（混合三相体系）；⑤盐酸小檗碱反丁烯二酸共晶与辅料混合体系（复合两相体系）；⑥蓓萨罗丁川芎嗪共晶混合体系（混合三相体系）；⑦蓓萨罗丁-蓓萨罗丁川芎嗪共晶-辅料混合体系（复合三相体系）。.在晶型定量模型的构建过程中，项目主要考察了经典线性回归与结合化学计量学的两种方法，在光谱技术结合化学计量学的研究中，重点考察了光谱区间（低波数特征波段、高波数特征波段、全波段）、化学计量学模型（PCR、PLS、ANN）、光谱预处理方法（一阶导数、二阶导数、中值、降噪、MCS）等建模参数，为建立适用于药物晶型含量预测模型构建的流程化方法，提出建模参数选择与优化的原则及建议提供了基础科学数据支撑。