生物检定(Bioassay)方法验证学习笔记

Biological assays (also called bioassays) are an integral part of the quality assessment required for the manufacturing and release of biological products. Biological products include biotherapeutics, vaccines, and cell and gene therapies. Bioassays commonly used for potency estimation can be distinguished from chemical tests by their reliance on a biological substrate (e.g., animals, living cells, functional complexes of target receptors, and immunological reagents). Because of multiple operational and biological factors arising from this reliance on biology and biochemical reactions, bioassays typically exhibit a greater variability than chemically based tests do. They produce measures of functional activity or potency relative to a standard, and rely on a fundamental assumption that the test and standard contain the same active constituents.

生物测定（亦称生物检定）是生物制品生产与放行所需质量评估中不可或缺的组成部分。生物制品包括生物治疗药物、疫苗以及细胞与基因治疗产品。常用于效价测定的生物检定与化学检测的区别在于，其依赖生物反应体系（如动物、活细胞、靶受体功能复合物及免疫试剂）。由于这种对生物学与生化反应的依赖会引入多种操作与生物因素，生物检定通常表现出比化学检测更高的变异性。生物检定给出的是相对于标准品的功能活性或效价值，其基本假设是供试品与标准品含有相同的活性成分。

Bioassays are one of several methods with procedures and acceptance criteria that control critical quality attributes of a biological product. As described in the International Council for Harmonisation guideline 

, section 2.1.2, bioassay methods may measure an animal’s biological response to the product, a biochemical or physiological response at the cellular level, enzymatic reaction rates, or biological responses induced by immunological interactions such as ligand- and receptor-binding. Over time, the scope of bioassay approaches is likely to expand. Therefore, this 

 general chapter emphasizes validation approaches that provide flexibility to adapt to new bioassay technologies, new biological products, or both.

生物检定是用于控制生物制品关键质量属性的若干方法之一，包含相应程序与可接受标准。正如人用药品注册技术要求国际协调理事会（ICH）指南 ICH Q6B《生物技术 / 生物制品的质量标准 — 检测方法与可接受标准》 第 2.1.2 节所述，生物检定方法可测定动物对产品的生物学响应、细胞水平的生化或生理反应、酶促反应速率，或由配体–受体结合等免疫相互作用诱导的生物学响应。随着时间推移，生物检定方法的适用范围可能不断扩大。因此，本美国药典（USP）通则强调采用具有灵活性的验证思路，以适配新的生物检定技术、新的生物制品，或二者兼具。

Good manufacturing practice requires that test methods used for assessing compliance of pharmaceutical products with quality requirements should meet appropriate standards for accuracy and reliability. Assay validation is the process of demonstrating and documenting that the performance characteristics of the method underlying a procedure meet the requirements for the intended application and that the assay is thereby suitable for its intended use. 

 describes the assay validation parameters that should be evaluated for separation methods supporting small-molecule pharmaceuticals. Although the evaluation of these validation parameters is straightforward for these methods, their interpretation and applicability for bioassays and biological products is unclear. This chapter addresses bioassay validation from the point of view of the measurement of potency rather than mass or other physicochemical properties, with the purpose of aligning bioassay performance characteristics with uses of bioassays in practice.

良好生产规范（GMP）要求，用于评估药品是否符合质量要求的检测方法应满足适当的准确度与可靠性标准。方法验证是通过证明并记录方法的性能特征满足预期用途要求，从而确认该方法适用于其预定用途的过程。《药典方法验证》〈1225〉描述了用于小分子药物的分离分析方法所需评估的验证参数。尽管这些参数对化学方法的评估较为直观，但其在生物检定与生物制品中的解读与适用性尚不明确。本章从效价测定而非质量或其他理化特性的角度阐述生物检定验证，旨在使生物检定性能特征与其实际应用场景相匹配。

Assessment of bioassay performance is a life cycle process (see 

 and 

). Bioassay validation (called 

 in these 

 chapters) is a stage in the life cycle when bioassay method design and development has been completed and the SOP has been fully documented. Bioassay validation is guided by a validation protocol describing the goals (validation parameters and acceptance criteria), design (sample selection and replication strategy), and analysis of data from the validation study.

生物检定性能评估是一个全生命周期过程（参见《分析方法生命周期》〈1220〉与《生物检定通则 — 概述与术语》〈1030〉）。生物检定验证（在上述 USP 章节中称为方法性能确认）是生命周期中的一个阶段，此时生物检定方法的设计与开发已完成，标准操作规程（SOP）已完整成文。生物检定验证遵循验证方案，方案中说明验证目标（验证参数与可接受标准）、试验设计（样品选择与重复策略）以及验证研究的数据分析方法。

Parameters that will be addressed in this chapter are relative accuracy, IP (includes repeatability), linearity, and range, using what has become known as a 

. Definitions of these parameters are given in 

. Other parameters, which are discussed in 〈1225〉 and 

, such as LOD and LOQ, are not considered in this chapter because they are mostly irrelevant to the uses of bioassay. These may be useful, however, in the validation of ancillary assays such as those used to score responders or measure response in an in vivo bioassay.

本章将论述的参数包括：相对准确度、中间精密度（IP，包含重复性）、线性与范围，所采用的方法通常称为稀释线性研究。这些参数的定义见第 3 节 “生物检定验证参数”。其他在〈1225〉与 ICH Q2 (R2)《分析方法验证指南》 中讨论的参数（如检测限 LOD 与定量限 LOQ）不在本章范围内，因为它们与生物检定的常规用途基本无关。不过，这些参数对于辅助方法的验证可能具有价值，例如用于判定应答者或测定体内生物检定响应的方法。

Repeatability has been omitted from the list of validation parameters because this is a component of IP, and there is no use in the bioassay that informs an acceptance criterion on this parameter. In addition, this is frequently built into the bioassay SOP as an acceptance criterion for replicate variability. Within-run replicates should nevertheless be included in the validation design to assess alternative release procedure formats (combinations of within-run and between-run replicates) and to facilitate design of other procedures.

重复性未列入验证参数清单，因其已包含在中间精密度中，且在生物检定中无适用场景可为此参数制定可接受标准。此外，重复性通常已作为平行样变异性的可接受标准纳入生物检定 SOP。尽管如此，验证设计中仍应纳入批内重复，以评估不同放行程序模式（批内与批间重复的组合），并为其他程序的设计提供支持。

Bioassays are also used during product development to address potency changes over process steps and the impact of formulation. Thus, selectivity (lack of interference from sample matrix components) is best evaluated during development and before the use of the bioassay to support process and formulation development. This is described in 

 and includes assessment of similarity (or linearity) in manufacturing process or spiked samples. Similarly, the ability of the bioassay to detect sample degradation should be addressed before or during bioassay validation.

生物检定还用于产品开发阶段，评估工艺步骤中的效价变化及处方的影响。因此，专属性（即不受样品基质成分干扰） 最好在开发阶段、在使用生物检定支持工艺与处方开发之前完成评估。相关内容在《生物检定的设计与开发》〈1032〉中描述，包括对生产工艺样品或加标样品的相似性（或线性）评估。同样，生物检定检测样品降解的能力应在验证前或验证期间进行评估。

For purposes of these bioassay chapters, specificity will be taken to be aligned with product identity, and this, along with selectivity, will not be covered in further detail in this chapter.

为本系列生物检定章节之目的，特异性将与产品鉴别相关联，本章不再对其与专属性展开详细论述。

Although robustness is not a requirement for validation, 〈1032〉 recommends that it be assessed as part of the method life cycle. Factors that are explored in a robustness study include operating conditions (e.g., incubation temperature and time, pH, etc.) that are suspected of impacting bioassay performance. These robustness factors can be studied simultaneously using a multifactorial approach [i.e., using design of experiments (DOE)]. This is typically done prior to validation in order to establish bioassay parameter limits in the method SOP or to verify previously specified ranges. This is consistent with the life cycle view that validation is a continuous process, where information that informs bioassay design and development is part of the overall assessment of validity of a bioassay method.

尽管耐用性并非验证的强制要求，〈1032〉仍建议将其作为方法生命周期的一部分进行评估。耐用性研究考察的因素包括可能影响生物检定性能的操作条件（如孵育温度与时间、pH 等）。这些耐用性因素可采用多因素试验设计（DOE）同步研究。该工作通常在验证前完成，以便在方法 SOP 中设定参数限度，或验证预先规定的范围。这与生命周期理念一致，即验证是一个持续过程，为生物检定设计与开发提供依据的信息均属于方法有效性整体评估的一部分。

It is important to note that studies that evaluate the potency of diluted (or different doses of) samples are not appropriate for assuring reliable measurement of degraded samples. This is because a bioassay measures activity, not mass. While dilution can be used to measure a change in the response relative to amount (or dose), it does not address changes in meaningful factors that may impact biological activity such as change in protein conformation. In fact, change in dose (or dilution) is the basis of dose response in the bioassay, which renders it useless for assessing meaningful changes in a biological property of the sample. Given the inability to assess relative accuracy of a degraded sample (i.e., the expected potency is unknown), similarity serves as a proxy for the accuracy of relative potency measurement of degraded samples. Changes in potency of degraded samples should be accompanied by characterization of the biophysical structure of the material to verify the stability-indicating nature of the method.

需要重点说明的是：评估稀释样品（或不同剂量样品）效价的研究，不适用于确保对降解样品的可靠测定。因为生物检定测量的是活性，而非质量。虽然稀释可用于测定响应相对于量（或剂量）的变化，但无法反映可能影响生物学活性的实质性改变，如蛋白质构象变化。事实上，剂量（或稀释）变化正是生物检定剂量–响应关系的基础，这使其无法用于评估样品生物学性质的实质性改变。由于无法评估降解样品的相对准确度（即预期效价未知），相似性可作为降解样品相对效价测量准确度的替代指标。降解样品的效价变化应伴随物料生物物理结构的表征，以验证方法的稳定性指示特性。

As described in 〈1030〉, this chapter will conform with terminology introduced in a Stimuli article titled 

. In that article, an analytical method refers to operational components comprised of instrumentation, reagents, standards, sample preparations, calibrations, controls and suitability criteria on the sample and the system. The output of a method is a 

. By contrast, the term 

 refers to the use of the method to make a decision, which might detail a design (including a replication strategy), a sampling plan, and calculations resulting in RV. While bioassay may be used for various investigations, this chapter will concentrate on its use in a release procedure.

如〈1030〉所述，本章将遵循一篇题为《区分分析方法与分析程序以支持 USP 分析方法生命周期模式》的评述文章中提出的术语。在该文中，分析方法指由仪器、试剂、标准品、样品制备、校准、对照以及样品与系统适用性标准组成的操作组件，方法的输出是测量结果。与之相对，分析程序指使用方法做出决策的过程，可详细说明试验设计（包括重复策略）、取样计划及产生可报告值（RV）的计算过程。尽管生物检定可用于多种研究，但本章重点关注其在放行程序中的应用。

Following 〈1220〉 and 

, the basis for validation parameter acceptance criteria can be the ATP for the release procedure. This is informed by a prediction of the probability of failure to meet the potency specification and should be proposed for the RV for the release procedure. Once specified for a procedure the precision requirement can be translated into a requirement on the IP of the method. In this chapter, the acceptance criteria for precision will be primarily described for the method (the IP) in keeping with the need to compare performance characteristics among bioassay methods (e.g., IP of in vivo versus in vitro methods; informal comparison of a bioassay platform method across molecules) and as a baseline for development of other procedures using the bioassay method. When clarity is essential, precision requirements will be specified for both the method and the release procedure.

根据〈1220〉与 ICH Q14《分析方法开发指南》，验证参数可接受标准的依据可以是放行程序的分析目标概况（ATP）。其来源于对效价标准不合格概率的预测，并应针对放行程序的可报告值（RV）提出。一旦为程序规定了精密度要求，即可转化为对方法中间精密度（IP）的要求。本章中，精密度的可接受标准主要针对方法（即 IP）进行描述，以满足不同生物检定方法之间性能特征比较的需求（如体内与体外方法的 IP 比较、同一平台生物检定方法在不同分子间的非正式比较），并作为使用该生物检定方法开发其他程序的基线。在需要明确区分时，将同时规定方法与放行程序的精密度要求。

While validation has historically been viewed as the verification of individual validation parameters, the concept of TAE has been introduced to unite these. One approach combines bias and variability into a statistical prediction interval that is compared to the acceptance criterion of a procedure and has the advantage of recognizing the overall space of validation parameter acceptability versus that of a single pair of pre-specified requirements. This can be used to adjudicate a failure to meet one or another of the criteria on individual parameters or as the primary basis for validation with an ATP that has been expressed in terms of TAE.

尽管验证在历史上被视为对单个验证参数的确认，总分析误差（TAE） 概念的引入将这些参数统一起来。一种思路是将偏倚与变异性整合为统计预测区间，并与程序的可接受标准比较，其优势在于识别验证参数可接受的整体区间，而非单一预设指标。TAE 可用于判定单个参数是否不符合标准，或作为采用 ATP（以 TAE 形式表达）进行验证的主要依据。

The specifics of the validation parameters and the analysis plan should be clearly stated in the validation protocol. Due to restrictions on size of the study, however, the use of statistical approaches such as noninferiority and equivalence testing and the multiplicity of decisions required by some regulatory authorities (e.g., the need to meet requirements at each validation level) can result in elevated risk of failing one of the individual validation parameter acceptance criteria. This risk is partially mitigated using TAE and further managed using a holistic approach such as MEM. With MEM the statistical power of concluding acceptable performance can be enhanced by combining data across levels while assessment of validation parameters can be reduced to only a few statistical tests.

验证参数的具体内容与分析计划应在验证方案中明确说明。然而，受研究规模限制，采用非劣效性、等效性检验等统计方法，以及部分监管机构要求的多重决策（如在每个验证水平均满足要求），可能导致单项验证参数不合格风险升高。该风险可通过 TAE 部分缓解，并通过模型辅助验证法（MEM） 等整体思路进一步管理。采用 MEM 法，可通过合并不同水平的数据提高判定性能可接受的统计功效，同时将验证参数的评估简化为少数统计检验。

Failure to meet the acceptance criterion on IP (and thereby the variability of the RV) may be due to the approach taken to determine study size (e.g., focused on relative accuracy). This feature of employing a statistically rigorous approach but being penalized when the criteria aren’t met (because assumptions were incorrect or when authorities request a different analysis) can be mitigated using an approved protocol that includes appropriate corrective action. As suggested in 〈1220〉, that corrective action might be to modify the procedure format (replication strategy) with the commitment to re-evaluate method performance after sufficient experience with the bioassay. This can be addressed using continuous performance verification (in conjunction with Bayesian analysis) or prior knowledge for a platform bioassay technology.

中间精密度（IP）不满足可接受标准（进而影响可报告值 RV 的变异性），可能源于确定研究规模时的思路（如仅聚焦相对准确度）。采用严格统计方法却因假设不当或监管机构要求不同分析而无法满足标准的问题，可通过包含适当纠正措施的获批方案缓解。如〈1220〉所述，纠正措施可包括修改程序模式（重复策略），并承诺在积累足够生物检定应用经验后重新评估方法性能。这可通过持续性能确认（结合贝叶斯分析）或平台化生物检定技术的先验知识实现。

For the purposes of this chapter, the term 

 will be used to represent the full implementation of the bioassay method (including assessment of system and sample suitability and analyses performed to estimate relative potency). The goal of the bioassay validation will be to assess accuracy and linearity and to estimate both within-run and between-run variabilities of the method. Elaboration of more detailed components of bioassay variability can be found in 〈1032〉 and 

, and are summarized in 

.

为本章之目的，试验运行（run） 一词指完整执行一次生物检定方法（包括系统与样品适用性评估，以及用于估算相对效价的全部分析）。生物检定验证的目标是评估准确度与线性，并估算方法的运行内与运行间变异性。生物检定变异性更详细的组分可参见〈1032〉与《生物检定分析》〈1034〉，并在附录 B《生物检定精密度原理》中总结。

Bioassay validation is in some respects similar to validation of other analytical methods. Key among the similarities is demonstration of fitness for use. This extends from bioassay design and development and is continually managed throughout the bioassay life cycle. For purposes of this chapter, 

 will mean that the bioassay method, and the procedures that are developed using the method, are capable of supporting key biological product decisions. In the case of a release procedure, for instance, performance characteristics should be such as to limit the risks of making a poor release decision [e.g., out of specification (OOS)]. In this regard, the release specification is essential to defining the performance requirements for that use.

在某些方面，生物检定验证与其他分析方法的验证具有相似性，其中最核心的共同点是证明适用于预期用途。这一要求贯穿于生物检定的设计、开发，并在其整个生命周期中持续实施与管理。就本章而言，适用于预期用途是指：生物检定方法以及基于该方法建立的分析程序，能够支持对生物制品的关键决策。例如在放行程序中，方法的性能特征应能降低做出错误放行决策的风险（如结果超标 OOS）。在这一意义上，放行标准是定义该用途下性能要求的关键依据。

It is expected that the design of the bioassay method and associated suitability criteria have been completed prior to implementation of the validation and documented in the SOP. This includes but is not limited to identification and alignment of key structural factors, e.g., plates and plate layout in a cell-based bioassay and cages and cage position in an in vivo method. Robustness (or optimization) studies should have been carried out to establish ranges on key factors such as incubation temperatures and times, and ages and weights of animals. Those ranges should be a part of the SOP and adhered to during the validation.

生物检定方法的设计及相关适用性标准，应在验证实施前完成，并在标准操作规程（SOP）中形成文件。这包括但不限于关键结构要素的识别与统一，例如细胞法生物检定中的培养板及布局、体内方法中的动物笼具及笼位等。应已开展 ** 耐用性（或优化）** 研究，以确定关键条件的范围，如孵育温度与时间、动物年龄与体重等。这些范围应纳入 SOP，并在验证期间严格遵守。

Suitability criteria should be proposed as part of the bioassay SOP, while a provisional release procedure format is used for calculation of validation acceptance criteria. This may be updated according to findings during the validation. Any update should be directed, however, towards improving the performance or efficiency of the procedure. Suitability criteria may be updated, including elimination, when there is sufficient experience to determine the long-term impact on bioassay control. The provisional release procedure format should be changed immediately after validation when it has been determined that this is insufficient to deliver an adequately precise RV, while further changes can occur after sufficient experience with the procedure.

适用性标准应作为生物检定 SOP 的一部分进行拟定，同时采用暂定放行程序模式用于计算验证可接受标准。该模式可根据验证中的发现进行更新，但任何更新均应以提升程序性能或效率为导向。在积累足够经验、能够判断对生物检定控制的长期影响后，适用性标准可进行更新（包括删减）。若经确认暂定放行程序模式无法提供精密度满足要求的可报告值（RV），则应在验证后立即修改；后续进一步调整可在获得足够程序使用经验后实施。

The use of corrective action and information coming from continued performance verification is particularly important when statistical criteria (e.g., equivalence and noninferiority testing) are used to assess conformance of validation parameters to their acceptance criteria. This is because the validation study may not be powered sufficiently to verify satisfactory bioassay performance. This also reflects the need to be vigilant to bioassay performance throughout its life cycle.

在采用等效性、非劣效性检验等统计标准评估验证参数是否符合可接受标准时，纠正措施的应用以及来自持续性能确认的信息尤为重要。这是因为验证研究的统计效力可能不足以充分证明生物检定性能满足要求，也体现了在方法整个生命周期中持续关注其性能的必要性。

The bioassay validation design should include important ruggedness factors with as many levels as is practically feasible. Introduction of factors should take precedence over replication of the procedure (i.e., replication of the provisional release procedure format) to optimize the use of resources and to best represent the impact of changes in these factors on long-term performance of the method. Attention to factors rather than release procedure format provides information to verify the provision release procedure and to effectively design other procedures using the bioassay. Factors that have been resolved during development may be included during validation when these are critical or may interact with another factor (e.g., temperature may be re-explored if the validation includes a heat sensitive reagent). The design of the bioassay validation should likewise acknowledge the range of potencies that will be encountered during routine release, as well as during long-term and accelerated stability evaluations.

生物检定验证设计应纳入重要的稳健性因素，并在实际可行前提下设置尽可能多的水平。引入这些因素应优先于对程序的重复（即对暂定放行程序模式的重复），以优化资源利用，并更真实地反映这些因素变化对方法长期性能的影响。关注因素而非放行程序模式，可用于验证暂定放行程序，并有效设计基于该生物检定方法的其他程序。在开发阶段已确定的因素，若属于关键因素或可能与其他因素存在交互作用，可在验证阶段重新纳入考察（例如，验证中使用热敏试剂时可重新考察温度）。生物检定验证设计同样应覆盖日常放行以及长期、加速稳定性研究中可能遇到的效价范围。

In bioassay validation, 

 refers to the relationship between observed and expected log potencies rather than between signal and dose (i.e., dose response). Their regression (log observed versus log dose) should ideally have a slope equal to 1.0. Any deviation from a unit slope indicates proportional bias (i.e., bias between results across a range). Proportional bias has an impact on comparisons of potency results across a range (e.g., between stability time points or in stability slope, between results before and after a process change, etc.).

在生物检定验证中，线性是指观测效价对数值与预期效价对数值之间的关系，而非信号与剂量（即剂量–响应）之间的关系。其回归方程（观测对数值 vs 剂量对数值）理想斜率应等于 1.0。任何偏离单位斜率的情况均表明存在比例偏倚，即不同效价水平间测定结果的偏倚。比例偏倚会影响不同水平下效价结果的比较（例如稳定性考察各时间点之间、稳定性斜率之间、工艺变更前后结果之间等）。

Relative accuracy and linearity should ideally be prospectively addressed during bioassay design and development. Strategies such as randomization should be incorporated into the bioassay method to mitigate the impacts of factors that can generate bias in relative potency determination or comparisons. Steps in relative potency determination, such as similarity testing, can have a direct impact on RB or nonlinearity of potency determination. Thus, similarity acceptance criteria should be established during development to minimize these impacts prior to validation.

相对准确度与线性宜在生物检定设计与开发阶段进行前瞻性控制。应将随机化等策略纳入生物检定方法，以降低可能导致相对效价测定或比较出现偏倚的因素影响。相对效价测定中的相关步骤（如相似性检验）会直接影响相对偏倚（RB）或效价测定的线性。因此，应在开发阶段建立相似性可接受标准，在验证前将这些影响降至最低。

Determination of IP is perhaps the most important goal of bioassay validation, particularly if relative accuracy and linearity have been successfully addressed during bioassay design and development. The validation design should support the estimation of VCs associated with both within-run and between-run sources of variability or of meaningful ruggedness factors such as analysts and reagent lots. Reliable estimates of VCs can be used either to verify that the provisional release procedure format meets the precision requirement in the ATP or to update the format, as well as to design other procedures using the bioassay method (e.g., standard qualification, stability studies).

中间精密度（IP）的确定是生物检定验证中最重要的目标之一，尤其当相对准确度与线性已在设计与开发阶段得到良好控制时。验证设计应支持对批内、批间变异来源以及分析员、试剂批次等重要稳健性因素对应的 ** 方差成分（VC）** 进行估算。可靠的方差成分估算可用于：验证暂定放行程序模式是否满足分析目标概况（ATP）中的精密度要求、更新程序模式，以及设计基于该生物检定方法的其他程序（如标准品标定、稳定性研究）。

CIs should be calculated for the validation parameters using methods described in 

 and Burdick and Graybill (

). Conformance of validation parameters to their acceptance criteria should be carried out using either equivalence or noninferiority testing when the validation has been designed with sufficient power to perform a statistical assessment. Failure to satisfy the noninferiority criterion for IP may result in a change to the procedure format.

应按照《分析数据 —— 解析与处理》〈1010〉及 Burdick 与 Graybill（1）所述方法计算验证参数的置信区间（CI）。若验证设计具备足够统计效力，应采用等效性或非劣效性检验判定验证参数是否符合可接受标准。若中间精密度不满足非劣效性标准，可能需要对程序模式进行修改。

The remainder of this section will discuss bioassay validation design, the bioassay validation protocol, bioassay validation data analysis, documentation of bioassay validation results, and continued performance verification including bioassay maintenance.

本节后续内容将论述生物检定验证设计、生物检定验证方案、验证数据分析、验证结果文件化，以及包括生物检定维护在内的持续性能确认。

2.1 Bioassay Validation Design 生物检定验证设计

The bioassay validation should include samples supporting the range of the product specification and extended to ensure the reliability of each run used to determine the RV for a test article. Alternatively, the release procedure can include the provision to repeat the series of runs using alternative doses when some runs fail suitability. The bioassay range represents the extreme potency levels that show acceptable relative accuracy, IP, and linearity. Additional consideration should be given to the extended range of real-time and accelerated stability results.

生物检定验证应采用覆盖产品质量标准范围的样品，并可适当扩展，以确保每一次用于计算供试品可报告值（RV）的运行结果均可靠。作为替代方案，放行程序可规定：若部分运行未通过适用性要求，可采用其他剂量重新进行系列运行。生物检定的范围是指能够满足相对准确度、中间精密度与线性要求的极端效价水平。同时还应考虑实时与加速稳定性结果所覆盖的扩展范围。

In order to achieve representative estimates of relative accuracy, validation samples should be produced from the standard material (or the source of the standard material). This is to reduce the impact of the uncertainty of measurement of another sample on the estimate of bias. A minimum of three potency levels should be included in the validation, while five levels are recommended for a reliable assessment of linearity and to mitigate the risk of a restricted bioassay range (i.e., due to failure at one or the other of the validation sample extremes).

为获得具有代表性的相对准确度估算，验证样品应采用标准品（或标准品原料）制备，以降低其他样品的测量不确定度对偏倚估算的影响。验证中应至少包含三个效价水平，推荐采用五个水平，以便可靠评估线性，并降低因验证样品某一极端水平不合格而导致生物检定范围受限的风险。

It is expected that bioassay development (as detailed in 〈1032〉) will yield a method design based on knowledge of the within-run factors that affect the assessment of similarity and measurement of potency. The design of the validation should reflect knowledge of the long-term (between-run) factors that might influence the measurement of potency. It is helpful to include all sources associated with large variations and those where the effects of a factor are poorly understood. During development, a laboratory might alternatively employ “robust design” (

), which is a strategy that optimizes settings of within-run factors while simultaneously minimizing the variability due to between-run factors.

预期生物检定开发（详见〈1032〉）将基于对影响相似性评价与效价测定的运行内因素的认知，形成方法设计。验证设计则应体现对可能影响效价测定的长期（运行间）因素的认知。纳入所有变异较大的来源以及作用尚不明确的因素是有益的。在开发阶段，实验室可选择采用 “稳健设计”（2），即在优化运行内因素设置的同时，最小化运行间因素带来的变异。

A release procedure format that results in RV for a test material may be specified during development and used to test clinical materials. This format may be provisional owing to incomplete knowledge about the impacts of factors that may impact bioassay variability over its life cycle. During validation, IP is studied using combinations of factors that are formulated into validation runs. Validation runs should be conducted the same as bioassay method runs to be representative. Factors may be incorporated into validation runs using strategies that alter these in a planned way.

在开发阶段可规定能够得出供试品可报告值（RV）的放行程序模式，并用于临床样品检测。由于对可能影响生物检定整个生命周期变异性的因素认知尚不充分，该模式可为暂定模式。在验证阶段，通过将多种因素组合纳入验证运行，对中间精密度进行研究。验证运行的执行应与常规生物检定运行保持一致，以保证代表性。可采用计划性变更策略将因素纳入验证运行。

Using DOE, validation runs can be balanced across ruggedness factors (i.e., equal numbers of runs can be performed at all levels of the factors). 

 shows an example of a multifactor DOE that incorporates multiple analysts, multiple cell culture preparations, and multiple reagent lots into the validation plan.

借助试验设计（DOE），验证运行可在各稳健性因素间实现均衡设计，即在所有因素水平下开展数量相等的运行。表 1 给出了一个多因素 DOE 示例，将多名分析人员、多次细胞培养制备以及多批试剂纳入验证方案。

This represents a full factorial design where each analyst performs the bioassay with each of the 2 cell preparations and each of the 2 reagent lots. Balance is noted in having 

 = 4 runs at each level of the three factors in the design. Within-run replicates may be performed within some or all runs to estimate a within-run component of variability and to facilitate the design of bioassay procedures (see 

).

这属于全因子设计，每名分析人员均使用 2 种细胞制备物和 2 批试剂分别进行生物检定。设计中三个因素的每个水平均有 n=4 次运行，体现了均衡性。可在部分或全部运行内开展批内重复试验，以估算批内变异成分，并为生物检定程序的设计提供支持（见第 5 节 生物检定验证研究结果的应用）。

Fractional factorial designs may be employed to manage the validation study size when many factors have been identified. It is important to note that this use of DOE is not to estimate factor effects. Fractional factorial designs are used to maintain balance across factors while accommodating two-way interactions of factors.

当考察因素较多时，可采用部分因子设计控制验证研究规模。需要明确的是，此处采用试验设计（DOE）的目的并非估算各因素的效应，而是在兼顾因素间二阶交互作用的同时，保持因素水平间的均衡性。

Designs using nesting (e.g., replicates nested within plate, plates nested within analyst) may likewise be used during validation. An example of a nested design is shown in 

.

验证过程中同样可采用嵌套设计（例如，重复组嵌套于培养板内，培养板嵌套于分析人员内）。图 1 给出了嵌套设计的一个示例。

Figure 1. Example of a nested design using two analysts.

Components of variability can be estimated from the validation results from one or a combination of these designs. Designs using only two levels of a factor (e.g., two analysts) are insufficient to obtain a reliable estimate of its contribution to IP, while analysis of routine control data (as part of continued performance verification) should be used to support the decision to replicate a factor in the release procedure design. Nevertheless, significant sources of variability may have been identified during bioassay development. In this case, the validation should confirm their impact and their variability can be used to develop a bioassay format that meets the requirement for precision in the procedure ATP.

可通过上述一种或多种设计的验证结果估算变异成分。仅采用两个水平的因素（如两名分析人员）不足以可靠估算其对中间精密度（IP）的贡献，而对日常质控数据的分析（作为持续性能确认的一部分）应用于支持是否在放行程序设计中对该因素设置重复的决策。尽管如此，在生物检定开发期间可能已识别出显著的变异来源。在此情况下，验证应确认其影响，且这些变异可用于开发满足分析目标概况（ATP）中程序精密度要求的生物检定模式。

An additional consideration in bioassay validation design is the number of validation runs. The number of runs (and within-run repeats) is driven by the risks associated with conclusions drawn from the validation data analysis. This topic is discussed further in 

.

生物检定验证设计中的另一项考虑是验证运行次数。运行次数（及运行内重复次数）由验证数据分析所得结论的相关风险决定。该主题将在 2.3 生物检定验证样本量中进一步讨论。

As discussed in 

, it is less desirable to employ the provisional release procedure format within the design of the validation than to dedicate validation runs to inclusion and replication of important long-term sources of variability. Estimates of variability due to these components can be used to verify the precision of the provisional release procedure format or to explore alternative formats. In addition, a validation that delivers reliable estimates of long-term VCs (estimates of the variance due to individual factors) can support different procedures using the bioassay method, with incorporation of significant factors as well as various numbers of within-run and between-run replicates into their designs. This is particularly valuable because different uses of the bioassay (e.g., lot release, qualification of new standards, comparability of a new product manufacturing process, or stability studies) will have different performance requirements, requiring different designs and formats to support these.

如第 1 章引言中所述，在验证设计中采用暂定放行程序模式，不如在验证运行中专设并重复考察重要的长期变异来源。通过估算这些成分导致的变异，可用于验证暂定放行程序模式的精密度或探索替代模式。此外，若验证能提供可靠的长期方差成分（VC）估算（即各独立因素导致的方差估算），则可支持采用该生物检定方法的不同程序，将显著因素以及不同数量的运行内与运行间重复纳入其设计中。这一点尤为重要，因为生物检定的不同用途（如批次放行、新标准品标定、新产品生产工艺可比性或稳定性研究）具有不同的性能要求，需要不同的设计与模式予以支持。

This approach might be thought of as “validating the bioassay method” rather than “validating the release procedure”, and its importance is illustrated for the validation of an in vivo (animal) potency bioassay. While the final validated format for an in vivo bioassay procedure may require multiple runs to have sufficient precision in the RV, it is typically impractical to design an effective validation with the release procedure in its design.

该思路可理解为 **“对生物检定方法进行验证”，而非“对放行程序进行验证”**，其重要性可通过体内（动物）效价生物检定的验证加以说明。尽管体内生物检定程序最终经验证的模式可能需要多次运行才能使可报告值（RV）具备足够精密度，但在验证设计中纳入放行程序通常不切实际。

In fact, the burden of validating a bioassay procedure is generally related to the underlying variability of the method. A procedure for a highly variable method (in vivo, plaque count, TCID50) is likely to include multiple replicates to achieve the ATP requirement for precision of the RV. This is no less important for more precise methods where more factors and replicates of those factors can result in more robust information regarding bioassay performance.

事实上，生物检定程序的验证负担通常与方法本身的变异程度相关。高变异方法（体内法、空斑计数法、TCID₅₀）的程序可能需要设置多个重复以满足 ATP 对可报告值（RV）精密度的要求。这对于精密度更高的方法同样重要，因为纳入更多因素并对这些因素设置重复可获得更耐用的生物检定性能信息。

It is important to note, however, that to obtain a representative estimate of the IP of the method, the validation should be performed in accordance with restrictions that have been imposed during development (or as part of the method SOP). This includes the maximum number of independent runs (e.g., plates) that can be performed by an analyst in a fixed period of time.

但必须注意，为获得具有代表性的方法中间精密度（IP）估算，验证应按照开发期间设定的限制条件（或作为方法 SOP 的一部分）执行。这包括分析人员在固定时间内可完成的独立运行（如培养板）最大数量。

2.2 Bioassay Validation Protocol 生物检定验证方案

A bioassay validation protocol should include the validation design with between-run factors (and within-run if these interact with between-run factors), the number of validation runs, the design layout (crossed or nested), the validation parameters with justified acceptance criteria, and a proposed data-analysis plan.

生物检定验证方案应包含：纳入运行间因素的验证设计（若运行内因素与运行间因素存在交互作用，也应纳入）、验证运行次数、设计布局（交叉或嵌套）、带有合理性论证的可接受标准的验证参数，以及拟定的数据分析计划。

The validation might be formulated based on TAE rather than on individual parameters (see 

). Use of TAE should be clearly specified in the protocol as the primary basis of evaluating fitness for use or as an adjunct to a parameters-based validation.

验证可基于 ** 总分析误差（TAE）** 而非单个参数进行设计（见 3.5 总分析误差）。在方案中应明确规定将 TAE 作为评估适用性的主要依据，或作为基于参数验证的辅助手段。

A brief summary of the method SOP (the bioassay principle, design, and data analysis) is useful to justify possible sources of variability (e.g., potential ruggedness factors). System and sample suitability criteria should be established beforehand as part of design of the bioassay method. Because these may be based on limited data and a limited collection of samples, they may be proposed as tentative and can be updated based on results from the validation (or after suitable experience during routine testing). However, the suitability criteria should not be changed if, during development, these have been established to limit impact on relative potency determination (i.e., have been established to ensure accuracy or precision). They might be changed if they are based instead on variability of the suitability parameter rather than impact, and the change is supported by other measures of acceptable performance of the method.

对方法 SOP 的简要概述（生物检定原理、设计及数据分析）有助于论证潜在的变异来源（如潜在的稳健性因素）。系统与样品适用性标准应作为生物检定方法设计的一部分预先建立。由于这些标准可能基于有限的数据和样品集合，可拟定为暂定标准，并可根据验证结果（或在日常检验中获得充分经验后）进行更新。但是，若在开发期间已确定这些标准用于限制对相对效价测定的影响（即为确保准确度或精密度而设定），则不得修改。若这些标准仅基于适用性参数的变异而非其影响，且变更得到方法可接受性能的其他指标支持，则可进行修改。

Action should be prespecified when a validation run fails to meet a suitability criterion. As these should be rare, it is acceptable to remove the run without replacement. Note that this might result in an imbalanced design, requiring special software or statistical support to analyze the remaining validation data, and will reduce the statistical power if the validation uses a small number of runs. Note also an abundance of failed validation runs (more than is predicted by the method for deriving suitability criteria) should be regarded as a validation failure.

应预先规定当验证运行不满足适用性标准时的处置措施。由于此类情况应较少发生，可删除该运行且不补充。注意这可能导致设计不均衡，需要专用软件或统计支持分析剩余验证数据，且若验证运行次数较少，会降低统计效力。同时需注意，若验证运行失败数量过多（超出建立适用性标准方法的预期），应视为验证失败。

Inability to meet validation acceptance criteria may result in validation failure, a limit on the range of potency that can be measured in the bioassay, or a modification to the release procedure format (i.e., the replication strategy) to achieve the desired precision in RV.

不满足验证可接受标准可能导致验证失败、限定生物检定可测定的效价范围，或修改放行程序模式（即重复策略）以实现可报告值（RV）的预期精密度。

2.3 Bioassay Validation Sample Size 生物检定验证样本量

The application of statistical tests, including the assessment of conformance of validation parameters to their acceptance criteria, involves risks. One risk is that the validation result does not meet its acceptance criterion although the property associated with that result is satisfactory (called a type II error for an equivalence test; also, producer’s risk or false negative); another, the converse, is that the validation result meets its acceptance criterion although the property is truly unsatisfactory (called a type I error for an equivalence test; also, consumer’s risk or false positive).

统计检验的应用（包括评估验证参数是否符合可接受标准）涉及风险。一种风险是尽管结果对应的属性实际合格，但验证结果不满足可接受标准（等效性检验中的 II类错误，亦称生产者风险或假阴性）；另一种相反风险是尽管属性实际不合格，但验证结果满足可接受标准（等效性检验中的 I 类错误，亦称消费者风险或假阳性）。

The two types of risk can be simultaneously controlled via determination of the number of runs (and within-run replicates) to be conducted in the validation. Details for how this is done are reserved for 

.

可通过确定验证中执行的运行次数（及运行内重复次数）同时控制这两类风险。具体实施方法详见附录 A—— 生物检定验证示例。

2.4 Bioassay Validation Data Analysis 生物检定验证数据分析

A thorough analysis of the validation data should include graphical and statistical summaries, including validation parameter estimates (including CIs) and a conclusion regarding success or failure to meet their acceptance criteria. The analysis should follow the specifics of a data-analysis plan outlined in the validation protocol. Data should be analyzed in accord with the scale (e.g., log transformed) determined during development and specified in the method SOP (see 

).

对验证数据的完整分析应包括图表与统计汇总，其中包含验证参数估算值（含置信区间 CI）以及关于是否满足可接受标准的结论。分析应遵循验证方案中概述的数据分析计划细则。数据应按照开发期间确定并在方法 SOP 中规定的尺度（如对数转换）进行分析（见附录 C—— 数据与统计考量，C1 分析尺度）。

Analyses are typically performed on data at each validation level. This includes VC analysis to estimate the contributions of within- and between-run (or individual factor) sources of variability to the bioassay IP and ANOVA to estimate RB and its associated CI. Linearity is assessed across levels using regression analysis.

通常在每个验证水平下对数据进行分析。这包括用于估算运行内与运行间（或各独立因素）变异来源对生物检定中间精密度（IP）贡献的方差成分（VC）分析，以及用于估算相对偏倚（RB）及其置信区间（CI）的方差分析（ANOVA）。采用回归分析评估各水平间的线性。

Statistical success requires evidence of conformance of an estimated parameter to its acceptance criterion. It is not sufficient to fail to find a significant departure from a target (e.g., RB = 0%; sometimes called a difference test). Instead, an equivalence (or noninferiority) test using a CI or CB should be employed to show that a validation parameter is likely to be within appropriately justified acceptance criteria (see 

).

统计合格需要有证据证明估算参数符合其可接受标准。仅未发现与目标值存在显著偏离（如 RB = 0%，有时称为差异性检验）并不充分。相反，应采用基于置信区间（CI）或置信边界（CB）的等效性（或非劣效性）检验，以证明验证参数极有可能在经过合理论证的可接受标准范围内（见〈1010〉附录 3：等效性与非劣效性检验）。

An equivalence approach for assessment of relative accuracy will be illustrated in 

. An upper CB on IP can be used as statistical evidence of noninferiority (relative to the acceptance criterion) while prediction intervals are used with TAE. These may impact validation study sample size and will be illustrated as an "Advanced Consideration" in 

.

附录 A—— 生物检定验证示例中将举例说明用于评估相对准确度的等效性方法。中间精密度（IP）的单侧置信上限（CB）可用作非劣效性的统计证据（相对于可接受标准），而预测区间与总分析误差（TAE）结合使用。这些可能影响验证研究样本量，并将在附录 A—— 生物检定验证示例中作为 “高级考量” 进行说明。

If specified in the data analysis plan, MEM can be used to analyze validation data across validation sample levels while TAE might be proposed instead of the individual parameter approach.

若数据分析计划中有规定，可采用模型辅助验证法（MEM） 跨验证样品水平分析验证数据，也可提出采用 TAE 替代单个参数方法。

2.5 Documentation of Bioassay Validation Results 生物检定验证结果文件记录

Bioassay validation results should be documented in a bioassay validation report. The report should include the raw data and intermediate results (e.g., VC estimates) that facilitate reproduction of the bioassay validation analysis and the design of procedures using the bioassay method. Estimates of validation parameters should be reported at each validation level and overall if planned. Deviations from the validation protocol should be documented with justification. The conclusions from the study should be clearly stated with references to follow-up action(s) as necessary. Follow-up actions can include amendment of system or sample suitability criteria as well as a change to the release procedure format. Since the change in format is a mathematical function of the intermediate validation results there is no need for revalidation after a format change.

生物检定验证结果应记录在生物检定验证报告中。报告应包含原始数据与中间结果（如方差成分估算值），以便于重现生物检定验证分析并设计采用该生物检定方法的程序。应报告每个验证水平下的验证参数估算值，若有计划还应报告总体估算值。偏离验证方案的情况应附带合理性说明文件化。应清晰陈述研究结论，并在必要时注明后续措施。后续措施可包括修订系统或样品适用性标准以及变更放行程序模式。由于模式变更是基于中间验证结果的数学函数，因此模式变更后无需重新验证。

It is good practice to include results of other studies used to build validity into the bioassay method (e.g., robustness studies) or of other validation parameters (selectivity and specificity) in the final validation report.

良好规范是在最终验证报告中纳入用于支持生物检定方法有效性的其他研究结果（如耐用性研究）或其他验证参数（专属性与特异性）。

2.6 Continued Performance Verification 持续性能确认

Once a bioassay has been validated it can be implemented for routine testing of late development and commercial product materials. However, in keeping with a life cycle approach, it is important to monitor its behavior over time (Stage 3, continued performance verification; see 〈1220〉). This is partly accomplished using SPC charts (see 〈1010〉) on suitability parameters including potency of control samples. The purpose of these charts is to detect an aberrant run that might impact potency determination of samples tested in that run or a trend (i.e., a shift or drift) in method performance over time. If a trend is observed in an SPC chart, the cause should be investigated. If an assignable cause has been identified and resolved, the SPC limits may be reevaluated to address future abnormal behavior in bioassay method performance.

生物检定一经验证，即可用于后期开发与商业化产品的日常检验。然而，遵循生命周期理念，持续监测其随时间的表现至关重要（第 3 阶段：持续性能确认；见〈1220〉）。这可部分通过对适用性参数（包括对照样品效价）建立统计过程控制图（SPC） 实现（见〈1010〉）。控制图的目的是识别可能影响该次运行中样品效价测定的异常运行，或方法性能随时间的趋势（即偏移或漂移）。若在 SPC 图中观察到趋势，应调查原因。若已识别并解决可归属原因，可重新评估 SPC 限度，以应对未来生物检定方法性能的异常表现。

It is important to note that SPC monitoring of bioassay performance should be established early in the bioassay life cycle. Early establishment of processes for monitoring bioassay performance helps assure the link of potencies determined during development to testing of commercial product.

必须注意，生物检定性能的 SPC 监测应在生物检定生命周期早期建立。尽早建立监测流程有助于确保开发期间测定的效价与商业化产品检验结果的关联性。

Routine analytical control can be linked to life cycle events by collecting appropriate metadata such as critical reagent sources or lots, analysts, real time levels of critical conditions, and other factors that may influence bioassay performance. Collection of such data facilitates product or assay investigations and can be used to improve knowledge about factors that impact bioassay performance.

日常分析控制可通过收集适当的元数据（如关键试剂来源或批次、分析人员、关键条件的实时水平及其他可能影响生物检定性能的因素）与生命周期事件关联。收集此类数据有助于产品或检定方法的调查，并可用于深化对影响生物检定性能因素的认知。

Continuous performance verification should also include maintenance associated with planned changes during routine use of a bioassay method (i.e., post-approval changes). This includes bioassay method transfer, standard qualification, and method bridging. Planned changes should be supported by comparability protocols that are designed to assess risks to patients as well as manufacturing and supply.

持续性能确认还应包括与生物检定方法日常使用期间计划变更相关的维护（即批准后变更）。这包括生物检定方法转移、标准品标定及方法桥接。计划变更应得到可比性方案的支持，可比性方案旨在评估对患者以及生产与供应的风险。

Definitions and formulae of bioassay validation parameters are presented here.

本节展示生物检定验证参数的定义与计算公式。

3.1 Relative Accuracy 相对准确度

The relative accuracy of a relative potency bioassay is the relationship between measured and expected relative potency. A common approach to demonstrating relative accuracy for relative potency bioassays is by construction of expected potencies by dilution of the standard material. If the standard material is filled to 100% potency then early efforts should be made to preserve the source material for formulation into validation samples.

相对效价生物检定的相对准确度是测定相对效价与预期相对效价之间的吻合程度。证明相对效价生物检定相对准确度的常用方法是通过稀释标准品制备预期效价样品。若标准品效价为 100%，则应尽早采取措施保存原料用于配制验证样品。

Using dilutions of the standard is often referred to as a dilutional linearity study. The results from a dilutional linearity study can be assessed for the estimated RB at individual expected potencies:

使用标准品稀释液的方法通常称为稀释线性研究。可通过稀释线性研究的结果评估单个预期效价水平下的估算相对偏倚（RB）。

RB = 100 × 

It should be noted that altered samples (i.e., samples that have been modified chemically or by temperature) cannot be assessed for relative accuracy due to lack of information about their true value. They may, however, be used to expand the bioassay range based on their measured values in the validation.

See 

 for more information.

3.2 Intermediate Precision and Format Variability 中间精密度与模式变异性

IP is the variability of the bioassay method, as performed over the long term during which there are changes in method conditions and in influential laboratory factors. Since the validation is performed during a short period of time in its life cycle, the bioassay validation is designed to simulate some of these conditions and factors in order to approximate the true IP of the method (see 

).

中间精密度（IP）是指生物检定方法在长期运行过程中，因方法条件与关键实验室因素发生变化而表现出的变异性。由于验证仅在方法生命周期的较短时间段内开展，因此生物检定验证通过模拟部分此类条件与因素，以尽可能接近方法真实的中间精密度（见 2.1 生物检定验证设计）。

During bioassay design and development many method factors (e.g., robustness factors) that contribute important amounts of variation should be identified and appropriately managed (e.g., as ranges in the bioassay SOP) or be adequately replicated in the bioassay method design. It is nevertheless important to exploit this development-based knowledge when planning a bioassay validation, and to include factors in the validation that were not previously assessed or were not assessed in a multivariate design.

在生物检定设计与开发阶段，应识别出对变异贡献显著的诸多方法因素（如耐用性因素），并对其进行适当管控（如在生物检定 SOP 中规定范围），或在生物检定方法设计中进行充分重复。尽管如此，在制定生物检定验证计划时，充分利用这些基于开发阶段的知识，并在验证中纳入此前未评估或未在多因素设计中评估的因素，仍然十分重要。

IP, expressed as percent geometric coefficient of variation (%GCV) (

), is given by the following formula:

中间精密度以几何变异系数百分比（% GCV）（3）表示，计算公式如下：

In many cases there is insufficient replication of validation factors to be able to get reliable estimates of their contributions to IP. These can be used, however, to inform the laboratories of qualitative differences in variability among factors as a basis for improvement or to inform procedure design. As described previously, individual combinations of validation factors will be called runs for purpose of further discussion.

在许多情况下，验证因素的重复次数不足，无法可靠估算各因素对中间精密度（IP）的贡献。但这些结果仍可用于向实验室提示各因素间变异的定性差异，以此作为方法改进的依据，或为程序设计提供参考。如前所述，为便于后续讨论，将验证因素的每一组合称为一次运行。

Along with IP the format variability of the RV [calculated using the average of (RP) measurements] for a sample tested using 

 within-run replicates in each of 

 runs is estimated by:

与中间精密度一并，采用 nF 次运行、每次运行内设置 kF 个批内重复进行检测时，样品可报告值（RV）的程序模式变异性（通过相对效价（RP）测定值的均值计算）按下式估算：

IP (variability of the method) and format variability (variability of the release procedure) will be referenced throughout the chapter.

中间精密度（方法的变异性）与程序模式变异性（放行程序的变异性）将在本章通篇引用。

3.3 Linearity

It is important to show that the bioassay is linear to support uses of a bioassay method to compare potency measurements (e.g., stability investigations and process characterization studies).

为支持生物检定方法用于效价测定结果的比较（如稳定性考察、工艺表征研究），证明生物检定具备线性至关重要。

Similar to relative accuracy, linearity refers to the association between measured relative potency and expected relative potency but extends this to the trend across a range of potencies. This is sometimes called “trend in relative bias”.

与相对准确度类似，线性是指测定相对效价与预期相对效价之间的关联关系，并将其扩展至整个效价范围内的趋势。这有时也被称为 **“相对偏倚的趋势”**。

Linearity can be assessed using linear regression of estimated natural log potency versus expected natural log potency:

线性可通过估算效价自然对数对预期效价自然对数进行线性回归来评估：

ln(Measured Potency)=

+

×ln(Expected Potency)

Here the slope (

) represents the unit change in natural log measured potency per unit change in natural log expected potency. The expected value for 

 when the bioassay is perfectly linear is 1.0. A deviation from 1.0 can be expressed as proportional bias per Schofield (

) and the 

 Stimuli article 

. Note that the linear regression can likewise be performed on log RB versus known log potency with an expected slope equal to 0.0.

式中，斜率 b 表示预期效价自然对数每变化一个单位，测定效价自然对数的相应变化量。当生物检定呈理想线性时，b 的期望值为 1.0。与 1.0 的偏离可按照 Schofield（4）及 USP 评述文章《测量方法的线性》表述为比例偏倚。注意：同样也可对相对偏倚（RB）的对数值与已知效价对数值进行线性回归，其理想斜率等于 0.0。

3.4 Range 范围

The range is derived from the dilutional linearity study and is estimated to be the interval between the upper and lower validation levels for which the bioassay has acceptable levels of relative accuracy, IP (and indirectly, the variability of the RV), and linearity.

范围来源于稀释线性研究，其估算为验证上限水平与下限水平之间的区间。在该区间内，生物检定的相对准确度、中间精密度（IP）（并间接反映可报告值 RV 的变异性）及线性均处于可接受水平。

3.5 Total Analytical Error 总分析误差

TAE combines bias and variabilities of the RV and manufacturing via a statistical prediction interval for the relative accuracy (the ratio of the average potency to its true value, equal to RB/100 + 1). That prediction interval (sometimes called a β-expectation tolerance interval) estimates the range within which a future value for a tested lot is expected to fall. If the target Prob(OOS) is 1%, a 99% prediction interval can be used to represent the predicted distribution, which can be compared to the ATP deriving from this expectation. Similar to RB the bioassay acceptance criterion for TAE would be stated as a pair of bounds on the prediction interval. The use of TAE will be illustrated as an "Advanced Consideration" in 

.

总分析误差（TAE）通过相对准确度的统计预测区间（平均效价与其真实值的比值，等于 RB/100 + 1），将可报告值（RV）的偏倚、变异性与生产变异性进行整合。该预测区间（有时称为β- 期望容忍区间）用于估算未来某一批次样品的测定值预期落入的范围。若目标超标概率（OOS）为 1%，可使用 99% 预测区间代表预测分布，并与基于该预期建立的分析目标概况（ATP）进行比较。与相对偏倚（RB）类似，生物检定针对 TAE 的可接受标准以预测区间的上下限形式规定。TAE 的应用将在附录 A—— 生物检定验证示例中作为 “高级考量” 进行说明。

Validation acceptance criteria represent the requirements on validation parameters that manage decision risks (e.g., lot release). When there is an existing product specification, acceptance criteria can be justified based on the predicted risk of measurements falling outside of the product specification. This approach is introduced in 〈1220〉. A calculation of the probability of an OOS result is illustrated in 〈1010〉.

验证可接受标准是对用于管控决策风险（如批次放行）的验证参数所提出的要求。当已有产品质量标准时，可接受标准可基于测定结果超出产品标准的预测风险进行论证。该方法在〈1220〉中介绍，超标结果（OOS）的概率计算在〈1010〉中举例说明。

This can be calculated using either an estimate or target for product variability. Prior knowledge from products using similar processes (platform processes) can be used to inform this estimate. Lacking sufficient manufacturing history and final specifications, an approach using “state of the art” (e.g., IP no more than 20% for an immunoassay-based method) can be employed to drive bioassay design, development, and validation. Nevertheless, the approaches described here may still be used as a guide.

计算可采用产品变异性的估算值或目标值。使用相似工艺（平台工艺）产品的先验知识可为该估算提供依据。在缺乏充分生产历史与最终质量标准的情况下，可采用 **“行业先进水平”** 思路（如基于免疫分析的方法中间精密度 IP 不超过 20%）指导生物检定的设计、开发与验证。尽管如此，本文所述方法仍可作为指导。

As described previously, the validation acceptance criteria should be based on the release procedure ATP and be specified for both RB and IP. An approach using TAE can be adapted to meet the same standard of OOS risk as the individual parameter approach.

如前所述，验证可接受标准应基于放行程序的 ATP，并对 ** 相对偏倚（RB）与中间精密度（IP）分别规定。采用总分析误差（TAE）** 的方法可达到与单个参数方法相同的超标风险控制标准。

4.1 Determining Validation Acceptance Criteria Using Probability of OOS 使用OOS概率确定验证可接受标准

Estimation of the risk of an OOS release result uses the fact that relative potency measurements are typically nearly log-normally distributed, thus requiring log transformation. For a randomly selected lot from a manufacturing distribution with target relative potency of 100% and lot-to-lot variance (natural log scale) of 

, the RV from the bioassay will be normally distributed on the natural log scale, with a mean (

) that incorporates the bioassay RB and a variance that incorporates the format variability. Note that Greek symbols (

 and 

) are used in this development because they are assumed to be restrictions on true values of each parameter. Roman symbols (

 and 

) will be used later when referring to estimates of their respective parameters. The manufacturing mean (

) in natural log scale is:

对OOS放行结果风险的估计，利用了相对效价测量值通常近似服从对数正态分布这一事实，因此需要进行对数转换。对于从目标相对效价为100%且批间方差（自然对数 scale）为 

 的生产分布中随机选取的一批产品，生物测定的RV值将在自然对数尺度上呈正态分布，其均值 (

) 包含了生物测定的RB，方差则包含了形式变异性。请注意，在此推导中使用希腊字母 (

 和 

) 是因为它们被假定为对每个参数真实值的限制。在后续提及各自参数的估计值时，将使用罗马字母 (

 和 

)。生产均值 (

) 在自然对数尺度上为：

due to the offset from 1.0 (100%) from the RB of the bioassay.

这是由于生物测定的RB导致偏离了1.0（100%）。

The RA variance (

 in natural log scale) is defined as:

RA方差（自然对数尺度上的 

）定义为：

where 

 and 

 are the number of runs and number of replicates within each run of the RA format.

其中 

 和 

 分别是RA形式的运行次数和每次运行内的重复次数。

The observed manufacturing variability is the sum of the true production process variability (

) and RA variability:

观察到的生产变异性是真实生产工艺变异性 (

) 与RA变异性之和：

This can be expressed in terms of IP as follows:

这可以用IP表示如下：

which becomes:

when there are no within-run replicates (

).

当没有运行内重复 (

) 时，上式变为：

Finally, the probability that a random lot from such a process will fall outside of the LSL and USL specification limits for the relative potency is determined by RB, IP, and 

 via:

最后，来自这样一个过程的随机批次其相对效价落在规格下限（LSL）和规格上限（USL）之外的概率，由RB、IP和 

 通过下式确定：

where 

 represents the standard normal cumulative distribution function. The distribution of the measured potency (natural log scale) and associated Prob(OOS) are depicted in Figure 2.

其中 

 代表标准正态累积分布函数。测量效价（自然对数尺度）的分布及相关的OOS概率 Prob(OOS)

 如图2所示。

Figure 2. Illustration of calculation of Prob(OOS).

The validation acceptance criterion on IP should be obtained from knowledge regarding the relative percentages of product and RA variability to the total observed manufacturing variability.

中间产品（IP）的验证可接受标准，应基于产品自身变异与可报告值（RA）变异在总制造变异中所占相对比例的相关认知来制定。

In the absence of knowledge about the true product variance, it is reasonable to assume that it is equal to some proportion of the overall manufacturing variance. This may come from the development targets for process and analytical variability or from prior knowledge of their relative contributions. Implementation of this strategy will be illustrated in 

.

在缺乏产品真实变异数据的情况下，可合理假定其变异占整体制造变异的一定比例。该比例可来源于工艺与分析变异的研发目标，或基于二者相对贡献的先验知识确定。附录 A《生物检定验证示例》将对该策略的实施进行具体说明。

It is noted that these calculations can easily be adapted to the case where the manufacturing distribution true mean relative potency is not 100%. This approach will also be illustrated as an "Advanced Consideration" in 

.

需要说明的是，若制造分布的真实平均相对效价并非 100%，上述计算方法也可轻松适配调整。附录 A《生物检定验证示例》的 “进阶考量” 部分将对此方法进行示例阐述。

This approach for setting acceptance criteria aligns with other concepts in the chapter (e.g., TAE). It replaces a previous formulation that used the Cpm, which is a variant of the traditional process capability index that breaks apart components for process variability, variability of the RV, and RB.

该可接受标准制定思路与本章其他理念（如总允许误差 TAE）保持一致。它替代了此前采用Cpm 指数的计算方式；Cpm 是传统过程能力指数的一种变体，可将工艺变异、可报告值（RV）变异与相对生物效价（RB）变异进行分解。

5. USES OF BIOASSAY VALIDATION STUDY RESULTS 生物检定验证研究结果的应用

Bioassay performance information can be used to develop bioassay procedures, to determine the ranges in results that can be considered outside the noise of the bioassay (critical fold difference), to establish the number of digits to report in bioassay measurements, and as a guidepost against which other sources of information on bioassay variability can be leveraged.

生物检定性能数据可用于：制定生物检定操作规程、判定超出检定噪声范围的结果区间（临界倍数差异）、确定生物检定测定结果的报告有效位数，同时也可作为参照，整合利用其他来源的生物检定变异信息。

5.1 Using Validation Results to Determine Procedure Format 利用验证结果确定操作规程格式

VC estimates can be used to predict the variability for different bioassay procedures and thereby can determine a format that has a desired level of precision (satisfies the procedure ATP). The formula for format variability is presented in 

 and will be illustrated for different selections of number of runs and numbers of replicates within runs in 

.

变异系数（VC）估算值可用于预测不同生物检定方案的变异水平，进而确定满足预期精密度要求（符合规程可接受性能目标 ATP）的操作格式。格式变异计算公式详见 3.2 节《中间精密度与格式变异》，附录 A《生物检定验证示例》将针对不同实验次数与单次实验重复数的组合进行示例计算。

5.2 Using Validation Results to Determine Critical Fold Difference 利用验证结果确定临界倍数差异

Estimates of between-run and within-run variability can also be used to determine the sizes of differences (fold differences) that can be distinguished between samples tested in the bioassay. For 

 runs, with 

 replicates within each run, using an approximate two-sided critical value from the standard normal distribution (

 = 2 corresponding to 95% confidence), the 

 between RVs for two samples that are tested in the same runs of the bioassay is estimated by:

实验间变异与实验内变异的估算值，也可用于判定生物检定中可区分的样品差异幅度（倍数差异）。对于n

 次独立实验、每次实验设置k

 个重复的场景，采用标准正态分布近似双侧临界值（z=2，对应 95% 置信水平），同一批实验中测定的两个样品间可报告值（RV）的临界倍数差异估算公式为：

It is noteworthy that this uses the formula for format variability (under the square root) because the comparison between samples tested in the same run has the same variability as the "comparison" of a test sample to the standard sample.

When samples have been tested in different runs of the bioassay, the critical fold difference is estimated by (assuming the same format is used to test the two series of samples):

The 2 under the square root accounts for a comparison of two measurements made in two independent bioassay procedure runs using the same format.

根号下的系数 2，用于表征采用相同试验方案、在两次独立生物检定试验中对两个测定结果进行比较的情形。

From this the laboratory can assess differences in results from a bioassay with a fixed format or design a procedure (a bioassay format) that has suitable precision to detect a practically meaningful fold difference between samples when the samples are tested together in the same bioassay runs (e.g., when testing samples from a process study) or in different bioassay runs (e.g., when testing samples from a stability study). Since testing samples in different runs results in less sensitivity for detecting differences, this may be a useful consideration for study design. Note that if this is used to design a comparative procedure, it would be useful to consult a bioassay statistician to account for the risks (type 1 and type 2) in the study comparison.

实验室可据此评估采用固定试验方案时生物检定结果的差异，或设计具备适宜精密度的试验方案（生物检定格式），以便在同次生物检定试验（如工艺研究样品检测）或不同次生物检定试验（如稳定性研究样品检测）中，检出样品间具有实际意义的倍数差异。由于在不同试验批次中检测样品会降低差异检出灵敏度，这一点对试验设计具有重要参考价值。需注意，若采用该方法设计对比试验，建议咨询生物检定统计专家，以评估试验对比中的风险（Ⅰ 类错误和 Ⅱ 类错误）。

In addition to variability, nonlinearity (proportional bias) in bioassay results across an appropriate range should be included in the calculation of critical fold difference.

除变异度外，计算临界倍数差异时，还应纳入生物检定结果在适宜范围内的 ** 非线性（比例偏差）** 因素。

5.3 Using Validation Results to Determine Significant Digits 利用验证结果确定有效数字位数

The number of significant digits in a reported result from a bioassay is related to the precision of the RV. In general, RV from a bioassay with %GCV between 2% and 20% will support two significant digits. The number of significant digits should not be confused with the number of decimal places—reported values equal to 1.2 and 0.12 have the same number (two) of significant digits. This standard of rounding is appropriate for log-scaled measurements that have constant variation on the log scale and proportional rather than additive variability on the original scale. Note that rounding occurs at the end of a series of calculations when the final result (RV) is reported and used for decision-making, such as conformance to specifications or a requirement for the slope from a stability analysis. Thus, if the final measurement is an RV from multiple runs, rounding should not occur prior to determination of the RV. Likewise, specifications should be stated with the appropriate number of significant digits. ASTM International Standard E29-08 (

) provides further information about significant digits and rounding.

生物检定报告结果的有效数字位数与可报告值（RV）的精密度相关。通常，当生物检定的几何变异系数（% GCV）在 2%～20% 之间时，可报告值结果可保留两位有效数字。有效数字位数不应与小数点后位数混淆 —— 例如 1.2 和 0.12 均为两位有效数字。该修约标准适用于对数尺度测量值，此类测量值在对数尺度上变异恒定，在原始尺度上呈比例变异而非加和变异。

需注意，修约仅应在一系列计算的最终环节进行，即对用于决策的最终结果（可报告值）进行修约，例如判定是否符合质量标准、满足稳定性分析的斜率要求等。因此，若最终测定值为多次试验的可报告值结果，在确定该可报告值前不得进行修约。同理，质量标准也应采用适宜的有效数字位数进行表述。ASTM 国际标准 E29‑08 (5) 提供了有效数字与数值修约的更多相关信息。

5.4 Confirmation of Long-Term Bioassay Variability 生物检定长期变异度的确认

The estimate of IP from the validation can be highly uncertain due to a small number of validation runs and often underestimated due to omission of unrealized long-term ruggedness factors. The estimate can be confirmed or updated by analysis of control sample measurements such as the variability of a positive control or the residual variability from long-term stability studies. This analysis can be done with the control prepared and tested like a test sample (i.e., same test sample design factors such as number of replicates). This assessment should be made as part of formal continued performance verification and carried out after sufficient experience in implementation of the bioassay method, including planned changes associated with bioassay maintenance (e.g., different analysts, different standards, and different cell preparations). Periodic reassessments should be carried out over the bioassay life cycle. If the assessment reveals a substantial disparity of results, corrective action should be considered such as a change in format to maintain the variability requirement (ATP) for the appropriate bioassay procedure (e.g., the release procedure).

受验证试验次数较少的影响，验证阶段得到的中间产品（IP）变异估算值存在较大不确定性；同时，因未纳入未体现的长期耐用性因素，该估算值往往偏低。可通过分析对照样品测定数据对该估算值进行确认或更新，例如阳性对照的变异度、长期稳定性研究的剩余变异等。该分析可采用与供试品相同的制备及检测方式开展（即重复次数等试验设计因素保持一致）。

此项评估应作为正式持续性能确认的一部分，在生物检定方法经充分实践应用后开展，涵盖检定方法维护相关的计划性变更（如不同检验人员、不同标准品、不同细胞制备等）。在生物检定整个生命周期内应开展定期再评估。若评估发现结果存在显著偏差，应考虑采取纠正措施，如调整试验方案，以维持对应生物检定流程（如放行检测）的变异度要求（可接受性能目标 ATP）。

Based on an appropriate risk assessment, bioassay performance should be re-evaluated whenever a substantial change is made to the method. This includes but is not limited to a change in technology or transfer to another laboratory. A “partial validation” (i.e., a study to assess parameters that a risk assessment determines might be impacted by the change) or verification may be required. The re-evaluation should depend upon the ATP on the parameters that may be impacted by the change, a bridging study that compares the current and the modified methods or strategic use of data obtained from system suitability (including a positive control). The change management approach should be adequately documented in the company’s quality system. Some companies may wish to file their approach to changes as part of their regulatory dossier.

在开展充分风险评估的基础上，若检定方法发生重大变更，应对其性能进行重新评估，包括但不限于技术变更、方法转移至其他实验室等情形。必要时需开展部分验证（即通过试验评估经风险分析判定可能受变更影响的参数）或方法确认。重新评估工作应依据受变更影响参数的可接受性能目标（ATP）开展，可通过比对现有方法与变更后方法的桥接研究，或系统适用性（含阳性对照）数据的策略性应用完成。变更管理流程应在企业质量体系中完整记录，部分企业可将其变更管理方案纳入申报资料。