中药大品种制造关键质量属性表征:沸腾时间状态属性的提取过程在线NIR质量控制研究

Critical quality attribute assessment of big brand traditional Chinese medicine: online NIR quality control research on boiling time during extraction process

沸腾时间状态属性表征是制药企业提取过程质量控制亟待解决的难点问题。该研究以中药大品种生产物料黄柏的中试提取过程为载体,开发提取过程中沸腾时间状态属性的在线NIR质量控制方法。首先,采集黄柏2次提取过程的在线近红外(near infrared, NIR)光谱。通过提取罐透明视窗观察气泡状态,采用人工判断作为沸腾时间状态属性表征的参考,建立了提取过程中沸腾时间状态属性的在线NIR光谱移动窗口标准偏差(moving block standard deviation, MBSD)模型,优化了模型中光谱预处理方法为标准正则变换(standard normal variate, SNV),建模波段为800~1 200 nm,窗口值为4。以0.002 0为MBSD模型阈值,实现了提取过程中沸腾时间状态属性的在线NIR质量控制。进一步,为降低在线NIR光谱噪音和背景信号对模型的影响,采用课题组编写的主成分分析-移动窗标准偏差(principal component analysis moving block standard deviation, PCA-MBSD)模型,优化了PCA-MBSD模型中主成分数为2。以0.000 075为PCA-MBSD模型阈值,建立了可靠性更高的提取过程中沸腾时间状态属性的在线NIR质量控制方法。该研究开发的提取过程中沸腾时间状态属性的在线NIR质量控制方法稳定、可靠,可代替人工判断,实现中药大品种制造中提取过程的数字化。

Assessment of the status property(boiling time) is a challenge for the quality control of extraction process in pharmaceutical enterprises. In this study, the pilot extraction process of Phellodendron chinense was used as the research carrier to develop an online near-infrared(NIR) quality control method based on the status property(boiling time). First, the NIR spectra of P. chinense were collected during the two pilot-scale extraction processes, and the status property(boiling time) was assessed by observing the state of bubbles in the extraction tank using a transparent window during the extraction process, which was then used as a reference standard. Based on the moving block standard deviation(MBSD) algorithm, the assessment model using online NIR spectra for boiling time during extraction process was established. In addition, the model was optimized as follows: standard normal variable(SNV) for spectral pretreatment, modeling band of 800-2 200 nm, and window size of 4. The results showed that, with 0.002 0 as the MBSD model threshold, the boiling time can be accurately assessed using online NIR spectra during extraction process. Furthermore, the principal component analysis-moving block standard deviation(PCA-MBSD) model was developed by our group to reduce the influence of online NIR spectral noise and background signal on the model, and the number of principal components was optimized into 2 in the PCA-MBSD model. The results showed that, with 0.000 075 as the PCA-MBSD model threshold, the boiling time can be accurately assessed using online NIR spectra during extraction process, with improved reliability. This study can provide a assessment method for boiling time during extraction process using online NIR spectra, which can replace the empirical judgment in manual observation, and realize the digitalization of the extraction process for big brand traditional Chinese medicine.