Implementation of Established Conditions: Learnings from a “Sharing Science Solutions” Workshop

The International Council for Harmonisation of Technical Requirements for Pharmaceuticals for Human Use (ICH) guideline Q12 (1) (step 4 sign-off in November 2019) is in the process of being implemented in a number of regulatory regions. The document provides additional frameworks for pharmaceutical life-cycle management. It is intended to support globally harmonized regulatory tools such as established conditions (ECs) and product life cycle management (PLCM) documents to facilitate postapproval changes to chemistry, manufacturing, and controls (CMC). Although a harmonized framework and an agreed-upon plan for managing changes have the potential to be transformative, the practical implementation of the ICH Q12 concepts has yet to be fully realized. On 27 January 2020, a CASSS Sharing Science Solutions Working Group workshop titled “Established Conditions for Biopharmaceutical Products: Ion-Exchange Chromatography” was held in Washington, DC. The meeting brought together a small group of regulators and industry representatives to deliberate issues related to identifying appropriate ECs and reporting categories and presenting them with their underlying justifications in regulatory dossiers.

The workshop hosted a discussion of current experiences with identifying ECs and determining appropriate reporting categories. Topics included initial experiences with a US Food and Drug Administration (FDA) EC pilot program, concepts presented in the final version of ICH Q12, and challenges and opportunities from the perspectives of regulators and industry participants.

For the morning’s presentations, the discussion focused on a set of industry-generated chromatography unit operation case studies for purifying standard IgG monoclonal antibodies (MAbs). Following a general discussion, a working group’s cation-exchange chromatography (CEX) case study — based on an A-Mab case study (2) — was used as the framework for further discussion of EC designs. The EC designs were based on a traditional approach, enhanced approach, and performance-based approach as described in ICH Q12.

Morning Session
Dr. Steven Kozlowski (director of the FDA’s Office of Biotechnology Products, Center for Drug Evaluation and Research) made introductory remarks in the first session. Representatives from three companies that have been working toward potential implementation of ICH Q12 approaches then presented on development work within their companies.

The first presenter was Sally Anliker from Eli Lilly, who spoke on “Considerations for Shifting to an EC Paradigm.” She noted that in addition to potential realization of some ICH Q12 goals, such as simplified postapproval changes, implementation of the guideline also can introduce some potential difficulties. There might not be acceptance of including non-ECs in “established conditions sections” (formerly “commitment sections”). Companies also might not maintain some information in the EC sections, which would lead to having out-of-date information in the dossier’s EC sections and potentially subsequent confusion during inspection.

Anliker used an anion-exchange (AEX) chromatography unit operation as an example of the process for identifying ECs and reporting categories and potential outcomes of that process. A risk-based approach would focus on potential risks to product quality if a change were made. The approach would use current guidance to inform reporting categories, including prior approval, moderate and low notifications (for established conditions), and no reporting (for non-ECs). Risk assessments would be based on testing using validated models and design of experiments (DoE), confirmational results from clinical and full-scale batches, prior knowledge, and consideration of practical impact to critical quality attributes (CQAs). Out of nearly 30 process parameters included in the evaluation, about 10 would be proposed for registration as ECs. Only a few of those were considered to be critical and thus fall under the prior-approval reporting category. All parameters would remain part of the control strategy and be maintained through the internal Pharmaceutical Quality System (PQS). To provide clarity regarding which elements of a section are ECs, all ECs would be listed in tabular form, and the narrative would be considered non-EC. Information included in narrative form alone would consist of supporting information that helps explain, for example, the execution of the process and what the numerical parameters in the EC tables mean and how these parameters are tested. Anliker concluded by recommending that EC proposals be clear and complete, all proposed reporting categories be included, and the Quality Unit should review and be aligned with the proposals.

Vandana Chauhan from F. Hoffmann-La Roche presented “Defining Established Conditions for a Marketed Product: Case Study.” She highlighted the approach that Roche is developing for defining ECs and reporting categories, and she provided an example set of parameters designated as ECs and non-ECs as well as the corresponding EC reporting categories for a project that is part of the FDA ECs pilot program. Chauhan presented specific examples of the justifications behind those designations. Her case study was based on a CEX unit operation for an approved product that had been developed using an enhanced approach. The approach to defining ECs also was based on assessing the risk of an impact to CQAs and on using product-specific studies and prior knowledge. Similar to the previous example, narrative descriptions were considered to be non-ECs, and specific process parameters (including some noncritical parameters) and select in-process controls were identified as ECs.

The Roche example included the concept of differential reporting within the same parameter. Start and stop parameters for elution collection were separated into a prior-approval reporting category (for changes to the upper end of the range) and a notification-reporting category (for changes to the lower end of the range). Chauhan’s conclusion both supported the potential benefit from the establishment of ECs and reinforced that open questions still. Those questions include how the control strategy might play into the assessment of risk and whether it would be desirable or acceptable to have the reporting category for an individual parameter vary depending on the magnitude of the change relative to risk level.

Although directly comparing the example ECs and reporting categories in both presentations is not truly appropriate (given that they involve different products and processes), some similarities and differences are worth noting. For example, a number of parameters that are foundational to a chromatography operation run in bind-and-elute mode were categorized similarly. Column load and key elution buffer parameters were identified in both examples as ECs with prior approval reporting categories in both examples. However, although resin bed height and flow rate also were identified as ECs in both examples and were given notification reporting category designations in both examples, their notification reporting levels were different: changes being effected in 30 days (CBE-30) versus an annual report (AR). Similarly, although microbial in-process controls were included as ECs and given notification reporting category designations in both examples, their notification-reporting levels were different. One other notable difference was the identification of resin reuse cycles as an EC, with low-level reporting in one example and as a non-EC in the other. Those differences might be due to product-specific differences. However, differences in thinking about how to support EC and reporting-category designations also were evident during all discussions.

The final talk of the morning took the concept of ECs to the next level. Amy Morrison from Biogen spoke on “Looking to the Future: Performance-Based Established Conditions.” She noted that risk assessments and sufficient product and process understanding are foundational to enhanced control strategies. Her case study involved a hydrophobic-interaction chromatography (HIC) unit operation fed from an AEX unit operation.

The proposal included a feed-forward control strategy that varied the number of column cycles based on the level of high–molecular-weight species (HMWS) in the AEX output. Based on significant process understanding, the feed-forward control strategy was one of the few HIC process parameters designated as an EC. Inclusion of that type of adaptive control presents additional challenges. For example, potential infrequent triggering of different cycle numbers could exclude some scenarios during process performance qualification (PPQ), and there would be minimal additional at-scale data to support those scenarios.

One approach to verifying such a strategy (including ECs) is using a concurrent process validation protocol in addition to submitting high-quality data from small-scale and pilot-scale models used to develop the strategy. Implementing such enhanced and performance-based ECs should lead to manufacturing flexibility, maintenance of high-yield processes with few batch rejections, and a decrease in the number of ECs required for a unit operation.

Afternoon Session
Group Activity: In the afternoon session, teams of industry and regulatory attendees worked together to identify ECs for a CEX unit operation and to assign reporting categories to those parameters. Different groups considered that case study from the perspective of a traditional approach (two groups), enhanced approach (three groups), or enhanced development with a performance-based approach to establishing ECs (one group). Given the timeframe of the workshop, participants were provided with only basic product and process information. Enhanced-approach teams also received a brief summary of DoE-based study results.

Similarities and differences arose among the groups’ conclusions, even those arrived at using the same type of approach. The groups using information derived from traditional development identified most process parameters as ECs because limited product-specific data were available to help them exclude potential impact on product quality. Attendees noted that prior knowledge could be used to support the risk assessments involved. However, the groups had insufficient information in the case study to determine the relevance of much of the prior knowledge available. Based on gaining clarity regarding postapproval changes requiring a submission, they concluded that identifying ECs themselves can support the ICH Q12 goal of more predictable and efficient management of postapproval changes. The full potential of ICH Q12 concepts and harmonization might not be realized if a traditional approach is followed because a smaller amount of product-specific data is available to develop an understanding of process impact and risk-based conclusions to support many potential non-EC and lower reporting category proposals.

The groups basing ECs and reporting categories on an enhanced approach identified about half the number of process parameters as ECs relative to the number of parameters identified as such with the traditional approach.

There was a small difference in the number of parameters determined to be ECs across enhanced-approach groups. Although each group identified all critical process parameters (CPPs) as ECs, participants divided non-CPPs into ECs and non-ECs differently, and the reporting categories proposed were not identical. The groups noted some difficulty in justifying extrapolation beyond the data available and that the width of the range studied might influence data interpretation. The diversity of assessments was not linked to the attendees’ backgrounds. For example, there was diversity of conclusions among industry participants, among regulators, and among those with similar technical backgrounds.

The group studying the performance-based approach assumed the use of process analytical technology (PAT) and assumed that a model had been developed to provide control based on outputs. The number of ECs was limited to three process outputs and the model and PAT sensor, but a prior-approval reporting category was assigned to all five ECs. Participants noted that routine model maintenance also would be needed.

Workshop discussions that followed the morning presentations were an integral part of the afternoon group case study work. These discussions indicated that although there is keen interest in implementing ICH Q12 concepts to achieve greater harmonization in postapproval changes, many challenges, differences in understanding, and questions remain. Even traditional approach discussions were more problematic than expected.

Although some parameters seemed noncritical at first, little data existed to support an actual understanding of the relationship between process parameters and quality attributes. Making assumptions was easy, but justifying reporting categories was difficult because of the limited understanding of the level of potential risk to product quality from process parameter changes. Participants noted that being able to use prior knowledge — derived from the fundamental principles of the unit operation, literature, data from other products, and so on — would improve significantly the ability to identify and potentially downgrade ECs and reporting categories. However, the general understanding was that applicability of prior knowledge must be justified and that verification studies might be required.

Discussions related to data interpretation, the role of magnitude and direction of change, and control over parameters highlighted increased uncertainty and differences in understanding how ECs and reporting categories should be established. Although most participants agreed that all parameters can become CPPs when varied over broad ranges, there were divergent thoughts on how to address that, and some participants had considered that in more detail than others. The size of the range studied can influence the interpretation of results. For example, if a narrow process parameter range has been assessed, no impact to product quality might be observed, but that might not be an appropriate indicator of potential product quality impact related to parameter changes beyond the studied range.

Extrapolation beyond the data was difficult to justify as part of the case study. Although extrapolation should be possible, there was no conclusion regarding what would be needed to provide an acceptable level of support. How to define prospectively the magnitude of change to which an applicant would want to extrapolate also was unclear. The ranges for many process parameters had practical limits that could be driven by physical (e.g., equipment), chemical, and biological principles. Although considering those limits as part of the EC and reporting category assessment should be possible, how that should be approached or communicated is unclear.

The directionality and magnitude of potential future changes can affect the level of risk to product quality (and thus the proposed reporting category). For example, widening the upper limit of protein load onto the column can be considered to be a higher risk to product quality than widening the lower limit. Increasing a load by only a small amount is lower risk than is increasing a load significantly. Although implementing that concept could improve flexibility, it also would make the EC and reporting category proposal more complex, and the broad regulatory acceptance of the use of that concept is unclear. Participants also disagreed on whether and how controllability of a parameter should be included as a factor in EC or reporting category assessment.

Participants noted the need for considering an overall process and its controls. Not having the full picture for the unit-operation case study led to circular arguments and an inability to determine whether a parameter needed to be an EC for the unit operation under consideration. One example of that related to processing temperature: Some participants understood that temperature does not need to be an EC because it is controlled within the facility. Other participants argued that temperature is controlled within the facility because it is a CPP (per ICH definition), so temperature should be an EC.

Another example of the need for assessing the overall picture is a parameter (e.g., elution conductivity) that is critical but can be controlled during buffer preparation rather than as a parameter specified as an EC for a unit operation (and potentially included in a different section or subsection of a dossier). In general, including control over a change as a buffer parameter was thought to be acceptable. However, that parameter then would need to be an EC for buffer preparation.

Other examples, including the assumption that a process parameter already is controlled by a previous step (e.g., by the previous step’s elution buffer), confirmed the conclusion that the entire set of ECs for a process needs to be assessed to ensure comprehensiveness.

Participants also considered the relevance of the overall PQS. An effective PQS is required for using the ICH Q12 framework. However, the intent of ICH Q12 is not to create a requirement for additional descriptions of a company’s PQS requirements in a dossier. They also noted that separating ECs from some good manufacturing practice (GMP) requirements is not always easy. For example, some aspects of bioburden monitoring (e.g., in-process controls with registered action limits or acceptance/rejection criteria) typically are considered to be ECs. Aspects of bioburden monitoring also are critical GMP elements and are part of inspectional assessments.

Questions for Future Discussions
Although a number of questions were answered and common understanding was reached in some areas, questions remained unanswered. For example, are some parameters understood to “always” be critical? Can the reporting category for all non-CPP ECs be assumed to be low-level notification (e.g., annual reporting)? Will it be acceptable to propose different reporting categories, or even different EC/non-EC categories for a single process parameter, for different magnitudes or directions of change? Which in-process control tests need to be included as ECs? Are process parameters that affect non-CQAs or performance parameters and key performance indicators (KPIs) (always) ECs? With respect to the latter questions, workshop participants disagreed on non-CPPs/KPIs such as step yield that often are considered by industry to be business risk mitigations but that might be considered by regulators as important indictors of process control and/or market supply risk controls.

Participants concluded that benefits from the ICH Q12 framework could be envisioned for some products, and work toward that goal will continue to move forward at a number of companies. In addition to the technical questions above, some theoretical and practical questions regarding benefit and upfront resource requirements also remain. One question focused on actual benefit of ECs when changes are infrequent. Still in question is the magnitude of benefit from the implementation of ECs for products that have undergone traditional development, which may include many products with breakthrough designations that undergo rapid development.

Participants also noted that because designation of reporting category is not a requirement for declaring ECs, some benefit may be realized for products that have undergone traditional or rapid development by undertaking a less-intensive assessment to identify ECs to be proposed in a dossier at the time of initial submission. Determining a reporting category or updating those ECs could be undertaken when additional time and data are available.

Participants also agreed that additional separate workshops related to process and analytical ECs would be useful, especially if a more holistic process could be considered. Such workshops could provide for additional discussions on considerations for flexibility in the approach to proposing ECs and reporting categories and other concerns that had not been considered fully by all participants before this workshop.

References
1 Q12: Technical and Regulatory Considerations for Pharmaceutical Product Lifecycle Management. International Conference on Harmonisation of Technical Requirements for Pharmaceuticals for Human Use, 2019; https://database.ich.org/sites/default/files/Q12_Guideline_Step4_2019_1119.pdf.

2 A-Mab: A Case Study in Bioprocess Development. CASSS Sharing Science Solutions CMC Biotech Working Group, 2009; https://cdn.ymaws.com/www.casss.org/resource/resmgr/cmc_no_amer/cmc_amab_case_study/A-Mab_Case_Study_Version_2-1.pdf.

Corresponding author Sarah Kennett is technical regulatory director for biologics at Genentech, a member of the Roche Group. Kristopher Barnthouse is senior director of process development, API at Janssen Pharmaceutical R&D, LLC. Christopher Downey is CMC and product quality review chief at CDER, FDA. Minh Luu is director of technical regulatory affairs at Genentech, a member of the Roche Group. David Robbins is director in biopharmaceutical development at AstraZeneca. Kimberly Wolfram is director of global regulatory affairs CMC at Biogen; 600 Massachusetts Ave NW #300, Washington, DC 20001; [email protected].