For biopharmaceutical products, potency testing is a requirement through all phases of clinical investigation and following market approval. Due to the complexity and inherent variability of this type of analysis, an assay control strategy should be established for each specific product, adapted and improved throughout the analytical life cycle
By Sonja Klingelhöfer at Richter BioLogics
In the last decades, one of the key changes in the pharmaceutical industry has been the shift towards biotechnological products, broadening in diversity and complexity. This has led to increasing demands for the analytical characterisation of these products, and the use of state-of-the-art analytical technologies and methods.
Potency assays are one of the most crucial tests in the analysis of biopharmaceutical products, as they reflect the mode of action of the respective product and should be related to clinical efficacy. Alterations in the product might also result in reduced stability, so potency assays must also be stability indicative, and the demands for these types of tests concerning accuracy and precision are changing during the analytical life cycle. The use of reliable suitability criteria is key to verifying that a method is fit for its intended purpose. This is true for any method and is reflected in its use as a critical quality attribute. For some methods, concrete expectations for the choice of suitability criteria are given (eg, USP chromatography chapter <621>).¹ However, due to the diversity of potency assays, the setting of suitability criteria is much more complex and concrete guidance is lacking.
This article discusses aspects of an assay control strategy for risk mitigation from a contract development and manufacturing organisation (CDMO) perspective. Based on Richter BioLogic’s (RB’s) experience over the last two decades, some of the challenges in defining and correctly setting potency test suitability criteria are outlined.
Regulatory considerations
Suitability criteria for potency testing
Figure 1: Summary of exemplary measures for suitability testing
Aspects of RB’s assay control strategy
Developing, conducting and controlling (cell-based) potency assays for biotechnological products is always a complex task. But from the CDMO lens, the task is even more complex when given the diversity of products produced at a CDMO (eg, recombinant proteins, bacterial vaccines and plasmid DNA in case of RB). One can imagine that the spectrum of potency assays operated by RB’s bioassay group is quite broad, when considering not only the mode of action, but also the assay readouts, designs and statistical evaluation models. In addition, services are provided at all stages of development up to and including market supply. To address this challenge, the group has developed its own phase-dependent best practices regarding assay control strategy – including a variety of statistical tests – with the aim to be efficient but flexible to meet specific customer needs. Whether an assay is newly developed by RB or transferred from a customer, the common goal is to quickly gain control of assay performance, and ton successfully and smoothly progress through assay development and validation.
The review of assays is a hierarchical process, starting with identification of potential errors in assay performance, followed by identification and elimination of outliers. The latter is an aspect of evaluation that is very special for potency assays and is recommended by several guidance documents.5,8,9,11 After this precheck of the assay data, the SSTs are checked and only if they are met is the evaluation of the sample suitability criteria performed separately for each run (see Figure 2).
Figure 2: RB’s hierarchical decision tree of assay control (based on an assay example consisting of two sample runs). After exclusion of any error or outlying value, the suitability criteria are checked, starting with SSTs and followed by sample suitability criteria. (If an assay fails on a system suitability criterion, then a repeat of the whole assay is required; if a sample fails its suitability criterion, then only that sample run needs to be repeated)
By way of example, this article discusses the suitability of different measures of model adequacy. RB generally prefers the use of easily interpretable relative measures as suitability tests, but not only for goodness-of-fit. Especially in the early stages of a project, the use of such measures is advantageous when historical data is limited.
However, not all relative measures can be recommended. For example, the R² metric (coefficient of determination), which is widely used in regression models and can be easily calculated by a variety of different software systems, is not an optimal measure for potency assays (either for linear or sigmoid regression models). The main disadvantage of R² is that it is not an uncorrelated measure, but a combined measure of precision and goodness-of-fit. Another inappropriate relative measure is the lack-of-fit p-value, as the interpretation of the p-value depends on the variability of the respective run.
Where historical data is available, the lack-of-fit sum of squares (LoF SSQ) is an absolute measure, that may be used and is also recommended in USP (however, absolute measures such as LoF SSQ are problematic in the case of transfers or other situations where different equipment may be used). Although RB has established a variety of different measures and tests, the general aim is to limit the final suitability criteria to a set of relevant sensitive and uncorrelated parameters.
Another aspect of model adequacy is the confirmation of the response model. During assay development, the need for transformation or weighting of the response data is checked by additional graphical evaluation. In the example shown in Figure 3, the correct transformation is shown to have a relevant impact on the accuracy of the final result (relative potency changes from 0.91 to 0.98 after transformation). The appropriateness of the transformation is confirmed by almost normal distributed residuals and improved values for the SSTs monitoring model adequacy. Again, the R² metric is relatively insensitive and shows only marginal change after transformation, unlike other measures (data not shown).
Figure 3: Exemplary dose-response curves of a potency run (comparison of standard and sample in duplicates, upper graphs) and the corresponding residual plots (lower graphs). Left: Response (raw) data; Right: transformed response data
Continuous monitoring of assay parameters is also a key factor in assay control as it facilitates improvements. Monitoring of failure rates after validation can trigger method optimisation as well as improvement and refinement of suitability criteria during the life cycle. A high failure rate not only calls into question the robustness and suitability of an assay (regulatory risk), but also represents a commercial risk, as higher quality events can translate directly into higher costs.
Conclusion
The field of biopharmaceuticals is rapidly evolving, and with it the need for reliable bioassay methods to control product potency. Risk mitigation strategies are in the focus throughout the analytical method life cycle. Along the way, the phase-dependent selection of appropriate acceptance criteria (suitability tests) to control the assay performance is key, and continuous monitoring can facilitate improvements. With a keen awareness of customer needs and regulatory expectations, over the decades RB has developed its own phase-dependent best practices for assay control strategy, which remain flexible to meet specific customer needs.
References:
- Visit: usp.org/sites/default/files/usp/document/harmonization/genchapter/
harmonization-november-2021-m99380.pdf - Visit: database.ich.org/sites/default/files/ICH_Q14_
Guideline_2023_1116_1.pdf - Visit: uspnf.com/sites/default/files/usp_pdf/EN/USPNF/usp-nfnotices/
gc-1220-pre-post-20210924.pdf - Visit: doi.usp.org/USPNF/USPNF_M5877_10101_01.html
- Visit: doi.usp.org/USPNF/USPNF_M1354_01_01.html
- Visit: doi.usp.org/USPNF/USPNF_M912_01_01.html
- Visit: doi.usp.org/USPNF/USPNF_M5677_01_01.html
- Visit: doi.usp.org/USPNF/USPNF_M98860_02_01.html
- Visit: edqm.eu/en/d/418688
- Visit: ema.europa.eu/en/ich-q2r2-validation-analytical-proceduresscientific-guideline
- Visit: doi.usp.org/USPNF/USPNF_M99740_05_01.html
Sonja Klingelhöfer is a principal scientist at Richter BioLogics and
holds a PhD in biology. Before joining the pharmaceutical industry,
she worked at DESY in Hamburg, Germany. Sonja gained experience
in drug screening (herbal medicines), before setting up a good
manufacturing practices laboratory specialising in biological assays,
which she managed for over 20 years.
Quelle: European Biopharmaceutical Review | Summer 2025
