生物活性检测方法验证专栏

生物活性检测方法量效关系模型的确立问题探讨*

展开
  • 中国食品药品检定研究院 国家药品监督管理局化学药品质量研究与评价重点实验室,北京 102629
第一作者 Tel:(010) 53851589,E-mail:tandj@nifdc.org.cn

收稿日期: 2022-01-10

  网络出版日期: 2024-06-24

基金资助

* 国家药品标准制修订课题(2022Y09);中检院质量安全与能力建设项目

Discussion on how to set up the dose-response model of bioassay*

Expand
  • NMPA Key Lab for Quality Research and Evaluation of Chemical Drug, National Institutes for Food and Drug Control, Beijing 102629, China

Received date: 2022-01-10

  Online published: 2024-06-24

摘要

目的: 探讨生物活性检测方法的量效关系模型确立方式和评估标准。方法: 结合对量效关系模型的最新研究进展,采用回顾性研究方式,对生物活性检测方法中的量效关系模型,从其重要性、常见生物反应的量效关系模型、模型确立前需要解决的问题,到模型确立需要的基本步骤进行梳理和分析。结果和结论: (1)量效关系建模是目前进行方法验证前的重要评价内容,是方法风险评估的出发点。(2)生物反应的量效关系模型较多,不仅包括直线模型,还包括多项式模型、logistic模型、Brain cousins模型、单位点和双位点结合的特异性和非特异性模型,以及质反应数据的probit模型等。(3)在模型确立前,需要结合专业和统计知识进行跨学科交流,找出反应窗,通过预实验确定合理的浓度数和各浓度下重复数,进行合理的实验设计,并事先确定模型的评估方法和接受标准。(4)模型确立的基本步骤包括评估各响应值的独立性、残差的正态性和同质性,以及最后对同类模型的择优选择方式等。

本文引用格式

谭德讲, 韩璐, 段丽, 李娜, 杜颖 . 生物活性检测方法量效关系模型的确立问题探讨*[J]. 药物分析杂志, 2022 , 42(6) : 942 -950 . DOI: 10.16155/j.0254-1793.2022.06.04

Abstract

Objective: To explore the establishment method and evaluation criteria of dose-effect model of bioassay. Methods: Combined with the latest research progress in dose-effect model, the dose-effect model in bioassay was reviewed and analyzed from its importance, the common dose-effect models, the problems to be solved before model establishment, and the basic steps needed for model establishment. Results and conclusions: (1) The modeling of dose-effect relationship is an important evaluation content before bioassay validation at present, and it is the starting point of method risk assessment. (2) There are many model types in bioassay, including not only linear model, but also polynomial model, logistic model, brain cousins model, specificity and non-specificity model for single-site and double-site specific combination for quantitative data, and probit model for qualitative data, etc. (3) Before the establishment of the model, it is necessary to combine professional and statistical knowledge for interdisciplinary communication, find out the bio-response window, determine the number of dose and the number of repetitions at each dose through pre-experiments, conduct reasonable experimental design, and predetermine the evaluation approach and acceptance criteria of the model. (4) The basic steps for model establishment include evaluating whether each response value is independent, whether the residual value is in normal distribution and homogeneity, and finally, the selection of optimal model from similar models, etc.

参考文献

[1] USP 43-NF 38. Vol.5 [S].2020:7300, 8166, 7337, 7320, 8117
[2] EP 10.0[S].2020: 685
[3] FDA Guidance for Industry. Exposure-Response Relationships-Study Design, Data Analysis, and Regulatory Applications [DB/OL].2003. [2021-09-26].https://www.fda.gov/media/71277/download.
[4] ICH E4. Dose-Response Information to Support Drug Registration[S].1994:1
[5] ICH Q2(R1)Harmonised Tripartite Guideline:Validation of Analytical Procedures Text and Methodology[S].2005:1
[6] 中华人民共和国药典2020年版四部[S].2020: 480, 530, 206
ChP 2020. Vol Ⅳ [S].2020: 480, 530, 206
[7] LEBLOND D, TAN CY, YANG H. Confirmation of analytical method calibration linearity[J/OL]. Pharmacopeal Forum, 2013; 39(3):450. [2021-09-18]. https://www.uspnf.com/sites/default/files/usp_pdf/EN/USPNF/pf-legacy-pdf/pf-2013_vol-39. pdf
[8] LEBLOND D, TAN CY, YANG H. Confirmation of Analytical Method Calibration Linearity:Practical Application[EJ/OL].Pharmacop Forum, 2013, 39(5): 793. [2021-09-18]. https://www.uspnf.com/sites/default/files/usp_pdf/EN/USPNF/pf-legacy-pdf/pf-2013_vol-39.pdf
[9] USP<1010>Analytical Data-Interpretation and Treatment[S/OL].2022.DOI: https://doi.org/10.31003/USPNF_M99740_05_01
[10] ICH Press Release: ICH Prepares for Future New Topics [EB/OL]. (2018-06-22) [2021-09-23]. https://www.ich.org/pressrelease/ich-assembly-kobe-japan-june-2018
[11] 李娜, 杜颖, 谭德讲, 等. 生物活性检测方法的定义及分类辨析[J].药物分析杂志,2022, 42(6):931
LI N, DU Y, TAN DJ, et al. Definition and taxonomy of bioassay[J].Chin J Pharm Anal, 2022, 42(6):931
[12] Tecdat. R语言进行数据结构化转换:BOX-COX变换、“凸规则”变换方法[EB/OL]. (2021-04-19) [2022-01-04]. Date Strucureal Transformation by R Language: BOX-COX and Bulging Rule Transformation Methods. http://tecdat.cn/r%e8%af%ad%e8%a8%80%e8%bf%9b%e8%a1%8c%e6%95%b0%e6%8d%ae%e7%bb%93%e6%9e%84%e5%8c%96%e8%bd%ac%e6%8d%a2%ef%bc%9abox-cox%e5%8f%98%e6%8d%a2%e3%80%81%e5%87%b8%e8%a7%84%e5%88%99%e5%8f%98/
Tecdat.Date strucureal transformation by R language: BOX-COX and Bulging rule transformation methods. [EB/OL]. (2021-04-19) [2022-01-04]. http://tecdat.cn/r%e8%af%ad%e8%a8%80%e8%bf%9b%e8%a1%8c%e6%95%b0%e6%8d%ae%e7%bb%93%e6%9e%84%e5%8c%96%e8%bd%ac%e6%8d%a2%ef%bc%9abox-cox%e5%8f%98%e6%8d%a2%e3%80%81%e5%87%b8%e8%a7%84%e5%88%99%e5%8f%98/
[13] 闪念基因. 用实例告诉你什么是“过拟合”与“欠拟合”[EB/OL].(2019-11-14) [2022-01-03].
Flash Gene. Tell you what “overfitting“ and “underfitting” are with examples[EB/OL]. (2019-11-14) [2022-01-03].https://flashgene.com/archives/76532.html
文章导航

/