Two years after drafting a comprehensive revision of the 1987 process validation guidance, the FDA finalized the document this year. The revision elaborates on modern quality by design (QbD) techniques for developing a process, analyzing risks, and monitoring for control. The initial draft update remains largely intact, with some important adjustments focused on clarifying the FDA’s intent for how the industry is expected to validate its processes.
1 — Minor Changes: The guidance includes more references to the Code of Federal Regulations (CFR), an added glossary, and references to subject matter and the risk-based ASTM E250007 verification standard. Expectations for tracking operator errors have been deleted. To clarify different Stage 2 (process qualification) activities, the guidance now refers to technical evaluation as process performance qualification (PPQ).
Stage 1: Process Design
2 — Criticality: The guidance specifically eschews categorizing parameters and attributes according to their criticality. This differs from industry’s practice ofidentifying critical process parameters (CPPs) and quality attributes (CQAs). Instead, the guidance illustrates that criticality is a function of process role and product impact, both of which may change over the lifecycle of a process. The degree of control required should be determined by the risk a given parameter or attribute poses to process and product. Problematic parameters that negatively affect product quality need to be mitigated with additional testing (e.g., process analytical technology, PAT) and monitoring (e.g., statistical process control, SPC).
3 — Viral Safety: The agency removed expectations that viral inactivation and impurity clearance studies be “performed under GMP conditions.” The clarified intent is that these studies have “quality unit oversight,” not that scientists gown up to perform their experiments in Class 10,000 laboratories.
Stage 2: Process Qualification
4 — Less Testing: A noticeable change removes process testing at conditions that pose a high risk of process failure. The FDA is more interested in companies proving acceptable operating ranges than in their testing processes to failure. The guidance is clear: If QbD design space studies demonstrate process performance and control, a company need not test its commercial-scale process at operating range extremes. Such testing is an expensive and often unnecessary burden that some inspectors have insisted upon — and that some companies routinely perform.
5 — Batch Release: The guidance elaborates on concurrent release of PPQ batches. They must meet all lot-release and PPQ protocol criteria.
6 — And More Testing: If Stage 2 does not demonstrate reproducible process performance, the agency expects that additional studies will be done and that process changes may be warranted. So companies should have a review gate before commencing PPQ batches.
Stage 3: Continued Process Verification
7 — Legacy Products: To address industry concerns, the FDA clarified that companies with legacy products should leverage existing process development and qualification data and focus on Stage 3 (ongoing process monitoring) for such products. The guidance states that implementation of its recommendations “for legacy products and processes would likely begin with the activities described in Stage 3.” This should alleviate industry worries that the guidance could create an expectation for substantial new characterization studies of processes with an established history of operating in control. If a process has a history of failing in-process or final specifications and is fraught with execution deviations, however, then additional characterization studies may be needed to help a company better understand (and improve) that process.
8 — Enough Is Enough: The industry was concerned about the burden of continuing Stage 2 testing “through the process verification stage” (Stage 3). Statisticians use 20–30 data points to assess statistical control, and some companies incorrectly interpreted this as implying that 20–30 lots would be needed. (Short-run SPC focuses on data collected, not the number of lots. These techniques rely on far fewer lots to determine process control.) The guidance stops short of instructing specifically when to reduce intensive testing, but it clarifies that companies should consider process complexity, production volume, process knowledge, and experience in making that determination. By applying the principals of QbD (e.g., orthogonal test methods), statistics, and risk analysis, a company can more quickly prove that its process is in control and lower the risk of an upset.
9 — Variability: The guidance appropriately replaces the term process drift with undesired process variability, illustrating the agency’s understanding of variability’s two forms. Commoncause variability is predictable, natural variation that’s always present. Special-cause variability is unanticipated and can cause a loss of process control. For example, a poor batch of raw materials can cause a process specification failure. That should be eliminated through root-cause analysis: e.g., by evaluating and setting appropriate raw-material specifications.
10 — Process Improvement: The FDA clarified that one goal of Stage 3 process monitoring is to identify “potential process improvements” rather than insisting that a “process should be improved.” So the agency understands that although monitoring and study certainly identify opportunities for change, they are not necessary provided that a process is in control. It may not be cost effective or risk-appropriate to change a process that performs well without substantial benefit to cost, quality, and/or safety.
Peter K. Watler, PhD, is principal consultant and chief technology officer of Hyde Engineering + Consulting, 400 Oyster Point Boulevard, Suite 520, South San Francisco, CA 94080; 1-650-588-2660; firstname.lastname@example.org; www.hyde-ec.com.