Summary
Implementation and long-term maintenance of Quality-by-Design (QbD) with a contract manufacturing organization (CMO) requires mutual in-depth knowledge of each other’s capabilities, quality systems and expectations. QbD requires thorough product understanding, data trending and analysis on an on-going basis as it relates to critical process parameters and quality attributes, analytical method performance and quality metrics; this brings additional challenges for both partners and as the industry moves towards higher volumes of data through continuous manufacturing, PAT and real-time release. This article describes current practices/experiences and discusses strategies for a more efficient, consistent and scalable approach.
Introduction
Quality-by-Design (QbD) is now a well-established and commonly employed approach to the development and commercialization of products in the pharmaceutical industry. According to a 2012 survey conducted by Kourti and Davis1, eleven of the twelve pharmaceutical companies surveyed indicated they use QbD to some capacity in the development process. The benefits were clearly articulated and companies agreed that QbD was the right philosophy for better product and process understanding. Interestingly, when asked if QbD was employed at third-party manufacturers, responses from the same companies were much more variable. The range of responses indicated that QbD is used entirely, selectively, or in some cases, not at all. The common list of reservations included tight timelines, engagement of the Contract Manufacturing/Research Organization (CMO/CRO) too late in the development process, and Quality Systems infrastructure limitations. As pharmaceutical companies look to outsource more of their operations to third-party manufacturers, the inconsistent application of QbD at CMOs must be understood and addressed.
QbD is, by definition, a data-driven process with the goals of achieving maximum process understanding and control focused on science and quality risk management. Several rounds of design of experiments (DoEs), criticality and risk assessments are needed in order to establish the overall control strategy and the ongoing process verification and improvement approach. Therefore, good documentation and knowledge transfer practices are critical to success, particularly for commercializing products throughout a network of external suppliers. How does this information get captured, transferred and implemented in global QC laboratories and manufacturing floors throughout the lifetime of a commercial product?
The early identification and selection of a commercial CMO is essential for the most effective QbD implementation. In a QbD framework, a CMO is not only a manufacturing facility, it is a partner in the continuous improvement cycle for the product. It is expected that the CMO will be equally vested in the technical knowledge of the product, especially in preparing for and hosting regulatory preapproval inspections, as well as in the identification of potential opportunities for improving process capability and efficiency. As with any CMO, the facility must have robust quality systems established, however these systems need to be built over flexible risk management platforms that: (1) enable QbD-specific paradigms (e.g. criticality, design spaces, etc.), (2) promote continuous improvement and (3) allow the adoption of new technology and approaches, such as continuous manufacturing (CM) and real-time release (RTR). Also, in order to support ongoing process verification and improvement (in addition to the traditional annual product quality review process), information management tools need to be built to capture data on an on-going basis (as close to real-time as possible) and the resources allocated to analyze data within a meaningful timeframe to benefit the process and the overall product quality.
There is still a high degree of variability in the documentation needed for QbD implementation (including knowledge transfer) among different clients, which might present challenges for the CMO’s quality systems. At the beginning of a CMO-client relationship, each party arrives with a set of existing documents, knowledge and procedures, and a mutual desire to succeed in a new project and relationship. Table 1 summarizes some documentation examples with sole ownership* by either the client or the CMO. There are also some competing forces/interests. The key to a successful relationship, one built to endure through the lifetime of the product, is to move as many items as possible into the shared ownership realm while minimizing the competing obstacles. But the first step to achieve that is to identify these differences; only then can both parties acknowledge/ embrace them and/or find creative ways to reconcile them.
Table 1. A snapshot of QbD documentation/systems in the early stages of a Client-CMO relationship.
Data management and differences in quality systems could be obstacles to the successful implementation of QbD at a CMO and are discussed in the following sections.
Knowledge and Data Management Challenges
The development history must be transferred completely to the CMO to ensure the product knowledge necessary for inspections and shared ownership of CPV and CI. The earlier the CMO is involved, the sooner the development history becomes shared knowledge rather than a unidirectional flow of information from client to CMO. To illustrate the challenges associated with knowledge and data management when manufacturing a QbD product at a CMO, consider ICH Q8 (R2)2. The client is responsible for development activities: defining the Quality Target Product Profile, establishing the CQAs, linking material attributes and PPs to CQAs through a risk assessment, establishing a design space and implementing a control strategy. Tech transfer to the ultimate commercial site usually begins as early as clinical manufacturing for Phase II/III trials, but can occur after the product’s regulatory approval. Even though both parties recognize the importance of this development knowledge transfer, it often takes a back seat to the practical considerations of incorporating the filed/intended control strategy within the CMO’s systems. In practical terms, this also means that both parties need to be committed to documenting and discussing key development experiments (writing analytical method and process development reports, justification of specification documents, risk assessments preceding the DoE designs, etc.) and ensure that this acquired knowledge withstands the test of time such as personnel and/or corporate changes.
The volume of data necessary to support CPV over the lifecycle of a commercial QbD product can be extensive. Consider as an example the PPs associated with mixing, milling, wet granulation, drying, compression and film coating presented in Nagar et. al.3. For a standard tableting process, there may be more than 80 process parameters used for process monitoring or control. These process parameters could be tied to 5-10 CQAs as measured by validated analytical methods, the capability of which may be measured by a handful of analytical method performance attributes for each method. In total, these ~100 parameters must be captured and shared with the client for real-time or periodic data trending, as the client has joint responsibility for ongoing commercial production and continuous improvement. This data is also required as part of the annual product review process. The process of moving data from CMO systems (e.g. batch or laboratory records, SAP or LIMS systems, etc.) to the client has currently a heightened level of scrutiny, and should be performed through validated closed systems given industry/regulatory focus on data integrity. Recent advances in data historians and data repositories have helped, but CMOs may not be staffed or equipped to support these data archiving and transfer activities and must be retrofitted for these solutions. The systems must also be scalable as the volume of data increases for manufacturing platforms with real-time monitoring.
Reconciliation of Quality Systems
One of the first challenges encountered at the beginning of a CMO-Client relationship (even for traditional processes) is the “reconciliation” of the Quality Systems already in place in each of the companies4. It is extremely important that both companies have a deep understanding of each other’s procedures and systems and find ways to address any differences or needs (e.g. acceptance, change, implementation, etc.). This is particularly important for products developed using a QbD approach, where every company and/ or product might have a different flavor for terminology and regulatory expectations. For example, a manufacturing facility should be prepared to encounter the following scenarios and be able to incorporate them into their quality systems and documentation: process design space limited by an equation resulting from a model, tiered approach to criticality of process parameters and operational steps, instructions for process excursions outside the normal operating range or a critical quality attribute tested by parametric release. The Quality Agreement and/or additional customized documents could provide the means to document and address how to respond to these special situations. Also, given the process understanding offered by the QbD approach, a tiered system for deviations is warranted. Deviations from the Normal Operating Range (NOR) but within the Proven Acceptable Range (PAR) or Design Space Limits (DSL) as appropriate, present no risk to product quality and they can be addressed as minor deviations or observations throughout production. Note that there are different terms that have been used. Manufacturing facilities unaccustomed to QbD are likely to experience an increase in the company’s minor deviations or observations in the batch records, especially with the large amounts of data generated for a product made using a continuous process equipped with on-line IPCs and RTRT. It is important that both parties understand the implications of this, not necessarily from a compliance standpoint (as there may be no batch quality impact), but with a focus shifted toward continuous improvement.
Considering the focus on data trending and continuous improvement for QbD products, it is important to proactively define, identify, and investigate out of trend (OOT) results. Early alignment on the statistical process control (SPC) rules which would trigger OOT results is paramount, as well as the extent of the resulting investigations, given that the values might not impact whether or not the product meets specifications. Without carrying the same regulatory weight as an OOS, a philosophical difference arises between CMOs and clients when asked, ‘does an OOT value justify retesting of the product’, or ‘does an OOT require an investigation’. The identification of OOTs becomes more difficult if the number of batches is not sufficient to use the more defined SPC rules and the use of “scientific judgment” is required (e.g. orphan or pediatrics products). Close collaboration among subject matter experts from both parties is essential to decide the best course of action. As with minor deviations/observations, an increase in the number of OOTs might be encountered given the increased amount of data and real-time trending focus, so resources might become an obstacle. Nonetheless, the identification, investigation and trending of these OOT values are critical to a culture of continuous improvement.
Another key aspect is the understanding and mapping of root causes for deviations. Trending of quality metrics will not be effective for continuous improvement unless the definitions of root causes and how these are determined are clear. For example, in some systems “human error” is not seen as the ultimate root cause (in a 5-why analysis) but a much more specific reason such as “procedure not clear”, “training expired”, “failure to follow method” or “sample preparation issue” is defined.
Realization of Benefits
When operations reach a high level of efficiency and CMO and client systems are reconciled and working together, the full benefits of QbD can be realized. Shared ownership means both companies are driving towards the same end-goal: the consistent production of highquality drug products. Shared ownership translates directly to a reduction in process and analytical deviations over time, continuous improvement of operations and analytical testing, increased process capability indices for process parameters and CQAs (Figure 1), and a decrease in Quality Events (Figure 2). When the CMO and client truly reach shared ownership, continuous improvement will be driven equally by both partners, trends will be identified and addressed in a timely way, inefficiencies between mismatching Quality systems will be minimized and the number of Quality events reduced.
Figure 1. Example of the benefits of continuous improvement and data trending
Figure 2. Example of the benefits of Quality Metrics trending and continuous improvementEven though the implementation of QbD products might require a higher level of personalized effort, dedication and engagement, it will result in improved control of product quality and efficiency gains given the reduction in the number of quality events over time. These benefits will in turn create a fruitful partnership and overall team sense of accomplishment and motivation.
Acknowledgements
The authors would like to thank Tom Gandek and Michele Kercher-DiVerdi for their insightful comments and suggestions on this manuscript.
References
- Kourti T., Davis B. The Business Benefits of Quality by Design (Qbd). Pharmaceutical Engineering. 2012, 32(4).
- ICH Q8 (R2). http://www.fda.gov/downloads/Drugs/.../Guidances/ucm073507.pdf. Retrieved 19-Jan-2016.
- Nagar M., Singhai S., Chopra V., Bala I., Trivedi P. A study over effects of process parameters on quality attributes of a tablet by applying “quality by design”. Der Pharm Lettre, 2010, 2(2): 370-392.).
- Alghabban A. What to Look for in Selecting a CRO/ CMO and How to Ensure the Right Choice: A Quality Assurance Perspective. Pharmaceutical Outsourcing. https://www.pharmoutsourcing.com/Featured-Articles/172751-What-to-Look-for-in-Selecting-a-CROCMO- and-How-to-Ensure-the-Right-Choice-A-Quality-Assurance-Perspective/ Retrieved 27-Jan-2016.
* The term “ownership” used throughout this manuscript does not mean possession of legal rights, title or interest.