Yang Yuangui, Jin Hang, Zhang Jinyu, Wang Yuanzhong
Yunnan Academy of Agricultural Sciences, Institute of Medicinal Plants, Kunming 650200, China.
J AOAC Int. 2018 May 1;101(3):732-738. doi: 10.5740/jaoacint.17-0304. Epub 2017 Oct 24.
The saponins of Paris spp. have antimicrobial, immune-stimulating, and antitumor biological properties. In this investigation, FTIR and ultra-HPLC (UHPLC) were used for the determination of total steroid saponins in different species of Paris from Yunnan Province, China. A 52-sample calibration set and a 26-sample validation set for partial least-squares regression (PLSR) and support vector machine regression (SVMR) combined with FTIR and UHPLC were investigated. The optimal parameters C and γ were screened by a grid search with a sevenfold cross-validation. The results indicate that pretreatment with the combination of standard normal variate, second derivative, and orthogonal signal correction had the best performance. When comparing the SVMR and PLSR models, linear PLSR had better performance than nonlinear SVMR for the determination of total steroid saponins in different species of Paris. The highest total saponin content was found in P. axialis from Baoshan City (40.92 ± 9.06 mg/g). These results demonstrate that this approach would provide a fast and robust strategy for the QC of Paris in further analyses.
重楼属植物的皂苷具有抗菌、免疫刺激和抗肿瘤生物活性。在本研究中，采用傅里叶变换红外光谱（FTIR）和超高效液相色谱（UHPLC）法测定中国云南省不同重楼属植物中总甾体皂苷的含量。研究了用于偏最小二乘回归（PLSR）和支持向量机回归（SVMR）并结合FTIR和UHPLC的52个样本的校正集和26个样本的验证集。通过七重交叉验证的网格搜索筛选出最优参数C和γ。结果表明，采用标准正态变量、二阶导数和正交信号校正相结合的预处理方法性能最佳。在比较SVMR和PLSR模型时，线性PLSR在测定不同重楼属植物中总甾体皂苷方面的性能优于非线性SVMR。保山市滇重楼的总皂苷含量最高（40.92±9.06 mg/g）。这些结果表明，该方法将为进一步分析重楼的质量控制提供一种快速且可靠的策略。
