Page 35 - 《中国药房》2024年9期
P. 35

基于NIRS技术的款冬花药材质控指标定量分析模型的建立
                                                                                                     Δ


                                                                     #
          耿 涛 ,蒋文慧,刘佳伦,兰松平,王柳璎,陈佩林,严寒静,姬生国(广东药科大学中药学院,广州 510006)
                *
          中图分类号  R284;R917      文献标志码  A      文章编号  1001-0408(2024)09-1057-07
          DOI  10.6039/j.issn.1001-0408.2024.09.06

          摘  要  目的  建立基于近红外光谱(NIRS)技术的款冬花药材中款冬酮、水分、醇溶性浸出物和总灰分含量的定量分析模型,
          为款冬花药材及其制剂的快速质量评价提供新思路。方法  参照2020年版《中国药典》,分别采用高效液相色谱法、烘干法、热浸
          法及灰分测定法测定19个产地130批款冬花药材中主要质控指标款冬酮、水分、醇溶性浸出物、总灰分的含量,采集款冬花药材的
          NIRS数据信息,然后采用NIRS结合偏最小二乘法建立样品中上述质控指标的各个定量分析模型,经验证集样品验证后得到NIRS
          含量预测模型。结果  130 批款冬花药材样品中款冬酮、水分、醇溶性浸出物和总灰分的含量范围分别为 0.051 4%~0.103 5%、
          7.75%~10.93%、20.17%~31.12%、7.68%~12.10%。所建立的款冬花药材中款冬酮、水分、醇溶性浸出物和总灰分定量分析模型
          的内部交叉验证决定系数(R)分别为 0.985 8、0.968 4、0.973 4、0.988 0;校正集均方差(RMSEC)分别为 0.001 54、0.187、0.478、
                                2
          0.127;预测均方差(RMSEP)分别为0.001 81、0.212、0.543、0.149;RMSEP/RMSEC分别为1.175 3、1.133 7、1.136 0、1.173 2,均在合
          理范围内(1<RMSEP/RMSEC≤1.2)。验证集样品中上述4个质控指标真实值与模型预测值的平均绝对误差分别为-0.000 36、
          0.061 43、0.144 00和0.010 43,平均预测回收率分别为99.65%、100.72%、100.66%和100.15%。结论  所建NIRS定量分析模型稳定
          性好、测定结果可靠,可用于款冬花药材中相关质控指标含量的快速批量预测。
          关键词  款冬花;近红外光谱技术;款冬酮;快速分析;定量分析模型;质量评价

          Establishment of quantitative analysis model for quality control indexes of Farfarae Flos based on NIRS

          GENG Tao,JIANG Wenhui,LIU Jialun,LAN Songping,WANG Liuying,CHEN Peilin,YAN Hanjing,
          JI Shengguo(School  of  Chinese  Materia  Medica,  Guangdong  Pharmaceutical  University,  Guangzhou  510006,
          China)

          ABSTRACT   OBJECTIVE  To  establish  a  quantitative  analysis  model  for  the  contents  of  tussilagone,  moisture,  ethanol-soluble
          extract  and  total  ash  in  Farfarae  Flos  based  on  near-infrared  spectroscopy (NIRS),  providing  a  new  idea  for  the  rapid  quality
          evaluation  of  Farfarae  Flos  and  its  preparations.  METHODS  Referring  to  the  2020  edition  of  the  Chinese  Pharmacopoeia,  the
          contents of the main quality control indexes tussilagone, moisture, ethanol-soluble extract and total ash in 130 batches of Farfarae
          Flos  from  19  producing  areas  were  determined  by  HPLC,  drying  method,  hot  dip  method  and  ash  assay,  respectively. The  NIRS
          data  information  of  the  medicinal  herbs  of  Farfarae  Flos  was  collected,  and  then  NIRS  combined  with  the  partial  least  squares
          method  was  used  to  establish  the  individual  quantitative  analysis  models  of  the  above  quality  control  indexes  in  the  samples,  and
          the  predictive  model  of  the  NIRS  content  was  obtained  after  sample  validation  with  validation  set.  RESULTS  The  range  for  the
          contents  of  tussilagone,  moisture,  ethanol-soluble  extract  and  total  ash  in  130  batches  of  Farfarae  Flos  were  0.051  4%-0.103  5%,
          7.75%-10.93%,  20.17%-31.12%,  and  7.68%-12.10%,  respectively.  The  internal  cross-validation  coefficients  of  determination (R)
                                                                                                           2
          of the established models for the quantitative analysis of tussilagone, moisture, ethanol-soluble extract and total ash in Farfarae Flos
          were  0.985  8,  0.968  4,  0.973  4,  0.988  0,  respectively;  the  root  mean  square  errors  of  calibration (RMSEC)  were  0.001  54,
          0.187,  0.478,  0.127,  respectively;  the  root  mean  square  errors  of  prediction (RMSEP)  were  0.001  81,  0.212,  0.543,  0.149,
          respectively; RMSEP/RMSEC were 1.175 3, 1.133 7, 1.136 0 and 1.173 2, respectively, which were all within a reasonable range
                                                            (1<RMSEP/RMSEC≤1.2).     The   mean   absolute   errors
             Δ 基金项目 国家自然科学基金项目(No.81773829);广东省科技
          计 划 项 目(No. 2019A141405024);广 州 市 科 技 计 划 项 目(No.  between the true  and model-predicted values of the above four
          202002020071)                                      quality  control  indexes  in  the  validation  set  of  samples  were
             *第一作者 硕士研究生。研究方向:中药资源开发与品质评价。                   -0.000  36,  0.061  43,  0.144  00,  and  0.010  43,  respectively,
          E-mail:495392871@qq.com
                                                             and  the  mean  predicted  recoveries  were  99.65%,  100.72%,
             # 通信作者 教授,硕士生导师,博士。研究方向:中药资源、中药
          质量标准、中药新药研究。电话:020-39353119。E-mail:shengguo_ji@    100.66%,  and  100.15%,  respectively.  CONCLUSIONS  The
          gdpu.edu.cn                                        established  NIRS  quantitative  analysis  model  has  high  stability


          中国药房  2024年第35卷第9期                                                China Pharmacy  2024 Vol. 35  No. 9    · 1057 ·
   30   31   32   33   34   35   36   37   38   39   40