Design of experiment and risk analysis.
Design of experiment (DOE) and risk analysis are recommended in the draft guidance to provide data that support process design.
The emphasis should be to reveal relationships between variable inputs (e.g., component characteristics, processing parameters)
and resulting outputs (e.g., in-process material, intermediates, or the final product).
Early-stage product assurance.
The draft guidance states "early process design experiments do not need to be performed under CGMP conditions." Emphasis during
early stage development are verification rather than validation. "Decisions and justifications of the controls should be sufficiently
documented and internally reviewed," according to the draft guidance. Some examples of early-stage product assurance are outlined
below (4, 5).
- Viral and impurity clearance studies are required because they have a direct impact on product quality
- Cleaning verification would replace validation during the early stages of development
- Sterility must be ensured but may not be completely validated (e.g., if performing sterile filtration on a water-based product,
a filter integrity test may be performed using water-for-injection to demonstrate that the filter still meets its manufacturer's
integrity test value)
- Preliminary specifications, tests, acceptance criteria, and limits should be operational.
Overall, the further along a company is in its product development, the more verification and validation are expected. Process
controls that address variability can help to ensure the product's quality.
"It is important to understand the degree to which models represent the commercial process including differences that may
exist," states the guidance (4). The significance of understanding the process increases with a model that properly reflects
all the variants of the process and product. A good model can be used as a tool for process and equipment design, process
control development, estimating variablity, and training personnel.
Process analytical technology and qualification.
Process analytical technology (PAT) "uses timely analysis and control loops to adjust processing conditions so that the output
remains constant," states the guidance (4). As a result, there is a higher degree of process control. It is unclear, however,
whether maintaining the process within the design space using PAT reduces process-validation testing. The draft guidance places
added emphasis on design (qualification = design plus verification). There are requirements, for example, to challenge the
process under load, to test interventions, and to test stoppage and start-up routines as expected for production.
The common industry practice of testing three batches for PQ may no longer apply. Specific requirements regarding when batches
can be released and the information needed to begin commercial distribution are included in the draft guidance. PQ must have
a higher level of sampling, testing, and scrutiny of process performance to confirm the product's quality during batch processing.
According to the draft guidance, "In the case of PAT strategy, the approach to process qualification will be different from
that for other process designs," but the agency did not explain how the process will be different (4). This difference is
important because additional use of PAT leads to additional validation to verify the technology's capability. Industy will
assume this means an easier validation approach because conformation of control would be readily available, but there are
no examples provided in the draft guidance.
With regard to continued process verification, the goal is to ensure the process remains in a state of control during commercial
operations (3, 4). The draft guidance recommends manufacturers include intra-batch as well as inter-batch variation as part
of their continued verification program. PQ sampling levels should continue until variability is assessed.
The draft guidance suggests that a statistician be involved in the development of the company's data collection plan and in
the selection of statistical methods to evaluate process stability and capability. Included in this review should be quantitative
statistical methods where feasible. Process flow diagrams for commercial manufacturing, for example, should be completed as
- Describe each unit operation and placement in the process
- Describe process monitoring and control points
- Describe components and other processing material inputs
- Describe expected outputs (e.g., in-process and finished products).
In addition, the flow diagrams should preserve a life-cycle approach to facilitate comparison and decision-making regarding