|
PDA Journal of Pharmaceutical Science and Technology PDA制药科学与技术杂志 How Many Batches Are Needed for Process Validation under theNew FDA Guidance? 在FDA新指南下工艺验证需要多少批？ Harry Yang MedImmune LLC, One MedImmune Way, Gaithersburg, MD 20878©  DA, Inc. 2013 ABSTRACT: The newly updated FDA Guidance for Industry onProcess Validation: General Principles and Practices ushers in a life cycleapproach to process validation. While the guidance no longer considers the useof traditional three-batch validation appropriate, it does not prescribe thenumber of validation batches for a prospective validation protocol, nor does itprovide specific methods to determine it. This potentially could leave manufacturersin a quandary. In this paper, I develop a Bayesian method to address the issue.By combining process knowledge gained from Stage 1 Process Design (PD) withexpected outcomes of Stage 2 Process Performance Qualification (PPQ), the numberof validation batches for PPQ is determined to provide a high level ofassurance that the process will consistently produce future batches meetingquality standards. Several examples based on simulated data are presented toillustrate the use of the Bayesian method in helping manufacturers makerisk-based decisions for Stage 2 PPQ, and they highlight the advantages of themethod over traditional Frequentist approaches. The discussions in the paperlend support for a life cycle and risk-based approach to process validationrecommended in the new FDA guidance. 摘要：最新的FDA工艺验证行业指南:一般原则和实践引入了工艺验证的生命周期方法。而指导不再考虑使用传统的三批适当的验证,它并没有在一个事先的验证方案中规定也不提供具体的方法来决定验证批次的数量。这可能会让企业感到困惑。在本文中,我将开发一个贝叶斯方法（Bayesianmethod）来解决这个问题。通过结合从阶段1(PD)工艺设计中获得的工艺知识以及阶段2工艺性能确认（PPQ）所期望的结果,PPQ中验证批次的数量应提供一个高水平的保证,这种保证使得生产工艺能够持续的生产出符合质量标准标准的产品。在说明贝叶斯方法（Bayesianmethod）使用中会给出几个基于模拟数据的例子以帮助企业在阶段2PPQ中做出基于风险的结论。并且这个方法相比于传统的频率学论（Frequentist）方法有着明显的优势。本文中的讨论旨在为FDA新指南中建议的基于生命周期和风险的工艺验证方法提供支持。 Introduction 概述 In January 2011, theU.S. Food and Drug Administration (FDA) published the long-awaited FDA Guidance for Industry on Process Validation: General Principles and Practices (1). The guidance represents a significant shift of regulatoryrequirements from the traditional “test to compliance” at the end of process developmentto the current “quality by design” throughout the life cycle of the product andprocess. Process validation is no longer a set of documented evidencesdemonstrating consistency of the process in producing several consecutivebatches of commercial-scale product that meet pre-specified specifications. Instead,it is a science and risk-based development paradigm that yields product qualitythrough designing the process so that it is capable of consistently producingacceptable quality products within commercial manufacturing conditions. Theapproach is most apparent in the new regulatory definition for process validation:“the collection and evaluation of data, from the process design throughcommercial production, which establishes scientific evidence that a process is capableof consistently delivering quality” (1). The new guidance approaches processvalidation in three stages: 2011年1月，美国食品药品管理局（FDA）发布了期待已久的工艺验证指南：一般原则和规范。这份指南代表了法规要求的重大转变，即从传统在工艺后期开发的“测试至合规”转变为现在贯穿于产品和工艺生命周期的“质量源于设计”。工艺验证不再是通过收集若干连续产出预定标准的商业化生产来证明工艺的一致性的证据。相反，它是一门科学及基于风险的发展范例，通过良好的工艺设计，在商业化生产条件下有能力持续生产出合格的产品。最明显的变化是监管部门对工艺验证的新定义：“收集并评估从工艺设计阶段一直到商业化生产的数据，用这些数据来确立科学证据，证明该工艺能够始终如一地生产出优质产品”（1）。新的指南将工艺验证划分为3个阶段： Stage 1—Process Design(PD): The commercial manufacturing process is defined during this stage basedon knowledge gained through development and scale-up activities. 阶段1-工艺设计（PD）：在这个阶段确定了商业化生产工艺，这基于工艺开发以及放大活动中获得的知识。 Stage 2—ProcessPerformance Qualification (PPQ): During this stage, the PD is evaluated to determinewhether the process is capable of reproducible commercial manufacturing. 阶段2-工艺性能确认（PPQ）：在这个阶段，评估工艺设计（PD）以决定工艺是否有能力重现商业化生产。 Stage 3—ContinuedProcess Verification: Ongoing assurance is gained during routine productionthat the process remains in a state of control (1). 阶段3-持续的工艺核实：持续保证在日常的生产中工艺始终处于受控状态。 The new process validationguideline aligns activities at each of the three stages with the other existingguidelines, including ICH Q8(R2) Pharmaceutical Development, 2006; ICH Q9Quality Risk Management, 2007; ICH Q10 Pharmaceutical Quality Systems, 2007;ICH Q11 Concept Paper, 2011 (2–5). These encourage the use of modernpharmaceutical development concepts, quality risk management, and qualitysystems at all stages of the manufacturing process life cycle. 新的工艺验证指南在3个阶段中的每一阶段的活动都与其他指南相适应，包括ICH Q8(R2)药品开发，2006；ICH Q9质量风险管理，2007；ICH Q10 制药质量系统，2007；ICH Q11 概念文件，2001（2-5）。这鼓励运用现代化的制药发展概念，如质量风险管理，贯穿于生产工艺生命周期每个阶段的质量系统等。 By and large, the new processvalidation guidance has been received favorably by the industry, as it allows manufacturersto demonstrate the performance of the manufacturing process, using not onlydata from commercial-scale studies including process qualification, but alsothose from laboratory- and pilot-scale experiments conducted during the processdesign stage. However, this guidance also brings about some new challenges. Atthe top of the list is determination of the number of validation batches forStage 2 PPQ, using process knowledge obtained from Stage 1 PD. Ever since theFDA published its original guidelines on GeneralPrinciples of Validation in 1987 (6), three-batch validation has been widelyviewed as industry best practice in the past 25 years. However thethree-batch-validation rule lacks in scientific basis. It is conceivable thatwhen data from Stage 1 suggest that the process is robust and that the processvariations are well understood and under control, Stage 2 PPQ is more likely tosucceed. Under those circumstances, an extended Stage 2 PPQ with numerousbatches might add little benefit than what is already gleaned from Stage 1 PD.On the other hand, if the process demonstrates inconsistent performance orlarge variations that are difficult to control during Stage 1, its passingthree-batch PPQ acceptance criteria does not necessarily imply that it will providea high level of assurance that batches produced in the future will consistentlymeet pre-specified specifications. As a result, determination of the number ofbatches required for Stage 2 PPQ needs to take into account knowledge andunderstanding of the process gained from Stage 1, including process capability,sources of variations, in-process control, quality attributes, or validation parametersused to characterize product quality, as well as practical limitations inmanufacturing the number of batches sufficient for PPQ within a period of time. 总的来说，新的工艺验证指南已顺利被行业接受，因为它允许企业可以用商业化生产过程及工艺确认中的研究数据，也可以用工艺设计中实验室规模或小试规模的数据来证明工艺的性能。然后，该指南同样带来了一些新的挑战。首要的问题就是如何运用阶段1中获得的工艺知识来决定阶段2PPQ中的验证批次的数量。自从FDA在1987年发布最初的工艺验证指南：验证的一般原则，在过去的25年里，3批验证批次被行业普遍采用。然而，3批验证批次缺少科学依据。显而易见的是，从阶段1中获得的数据越有力，工艺中的变异就越容易理解和控制。在这种情况下，在阶段2PPQ中大量的验证批次可能只会带来少量的收益，因为阶段1PD中收集的信息已经足够多了。另一方面，如果在阶段1中已经证明工艺不稳定或存在着较大的变异并难以控制，即使其通过3批PPQ的可接受标准也不能说明工艺在将来的连续生产时能否符合预定的质量标准。因此，决定第2阶段PPQ所需的运行次数必须考虑在阶段1中获得的工艺知识与理解，包括工艺的稳定性，变异的来源、中间控制、质量属性或用于表征产品质量的验证参数，以及一定时间内，对生产充足PPQ批次数量的实际限制因素。 While the new guidanceno longer endorses the one-size-fits-all approach, it does not prescribe thenumber of validation batches for PPQ; the burden is on the manufacturers tomake a rational proposal, in light of data collected during Stage 1 PD andexpected outcomes of Stage 2 PPQ. However, how to combine data from both stagescan be challenging. In statistics, there are two types of inferences; one iscalled Bayesian and the other Frequentist. Through application of Bayes’ rule(7), the former provides a framework for combining new data with prior beliefsto calculate a posterior probability regarding parameters or random variables aboutwhich statistical inference is to be made. In the context of processvalidation, data concerning process performance during Stage 1 constitute priorbeliefs while data collected from Stage 2 PPQ are new evidences. One example ofthe prior belief is that the probability for batches produced during Stage 1 topass specifications varies according to certain statistical distribution. Thisis often true because raw materials, reagents, experimental conditions, andproduction scale change during early process development. When combined throughthe Bayes’ rule, the prior beliefs and new evidence result in an updatedestimate to the probability for future batches produced after PPQ to meetquality standards, producing a more sensible estimate of the number ofvalidation batches required for PPQ. By contrast, Frequentist methods rely onan assumption that the true probability for batches to meet quality standardsis a parameter that does not change. As a result, they rely on data such as thosefrom Stage 2 PPQ collected under the same manufacturing conditions to makeinference about the parameters of interest, having limited means to incorporateprior beliefs or data from Stage 1 with those from Stage 2 PPQ. 虽然新的指南不再认同一刀切的验证方法，但是它并没有为如何决定PPQ的验证批次数量提供解决方案。而作为企业则有责任根据阶段1PD中收集的数据以及阶段2PPQ中期望的输出结果做出合理的决策。然后，如何结合2个阶段的数据可能有点困难。在统计学上，这里有2种类型的推论。一种是贝叶斯法（Bayesian）另外一种是频率论（Frequentist）。通过运用贝叶斯规则（7），前者提供了一个框架，用于结合新的数据和先验概率以计算从统计学上推断出的关于工艺参数或随机变量的后验概率。在工艺验证的背景下，阶段1中建立先验信念所涉及工艺性能有关的数据以及阶段2PPQ中收集的数据都是新的证据。先验信念一个例子是阶段1中生产并通过标准按照某一统计学分布的概率。通常结果为真（Ture），因为在早期的工艺开发中原料、试剂、试验条件、生产规模都会改变。当通过贝叶斯规则合并，先验信念和新证据则产生了关于PPQ以后生产出符合质量标准批次概念修正后的预测以及PPQ所需验证批次更为合理的预测。相比之下，频率论方法依赖的依据是假设符合质量标准的批次的实概率是不会变的参数。因此，它们依赖于从阶段2PPQ中在同样的生产条件下收集的数据，以推断兴趣参数，而对于合并先验信念或阶段1的数据和阶段2的数据的手段则有限。 | 
|手机版|蒲公英|ouryao|蒲公英
( 京ICP备14042168号-1 ) 增值电信业务经营许可证编号：京B2-20243455 互联网药品信息服务资格证书编号：(京)-非经营性-2024-0033
发表于 2015-9-2 16:40:03
置顶卡
变色卡
楼主
