The debate on the number of batches used to determine if a process is validated or not has come under some scrutiny since the update of the US FDA’s Guide to Process Validation in 2011. As a result of the update, industry has tried to come up with a number of ways to help to justify the number (no longer acceptable to just assume that “3” batches is sufficient), and align with the US FDA definition that “The approach to PPQ should be based on sound science and the manufacturers overall level of product and process understanding and demonstrable control”. Assuming that the process design has already been evaluated and shows the process to be capable of reproducible commercial manufacturing, the number of batches that should be used to demonstrate a sufficient Control Strategy should indeed be based on sound science and a risk-based approach.
When should Stage 1 data be assessed?
The risk assessment of Stage 1 for a new product should be done periodically during the design phase, assessing the documented process understanding-that is the Critical Quality Attributes (CQA’s), Critical Process Parameters (CPP’s) and ultimately the Control Strategy for the process. The greater knowledge gained during Stage 1, the lower the risk in Stage 2, hence the number of batches required to demonstrate process reproducibility, may be lower. In the case of a legacy process, the risk assessment would concentrate on historical data and experiences.
What data should be assessed from Stage 1?
The assessment of Stage 1 (be it a new or legacy process) will look at assessing 3 specific sections (as shown in the diagram below):
- Product knowledge. The assessment should concentrate on the process variation that might impact the product safety, efficacy or quality.
- Process Understanding. This attempts to assess the relationship between material attributes, CQA’s, CPP’s and to estimate their variability on the Control Strategy.
- The Control Strategy itself. The factors that should be assessed include raw material specifications, equipment capability versus process requirements and experiences to date with process performance.
With these assessments completed, a residual risk level should be determined. This risk level is directly proportional to the confidence in the process performance, and will help to justify the number of PPQ batches. The higher the residual risk, the more PPQ batches will be required to confirm the process capability for reproducible commercial manufacture, but again, the process should be fully understood and documented. Using a Quality by Design (QbD) approach during Stage 1 should ultimately result in a lower residual risk, and minimise the number of PPQ batches.
Translating the Residual Risk Level into the number of PPQ Batches
A number of approaches to determining the number of PPQ batches were presented during the ISPE Process Validation Conference this year. These include the three examples below, one of which is based on experience and two of which are statistical-based approaches:
- Based on rationales & experience. This approach makes the assumption that three PPQ batches is sufficient to demonstrate reproducibility as the process has a low risk and three batches has been shown historically to be satisfactory for other similar processes. At the same time, where increased residual risk has been determined, an increased number of batches should be the result. It is important that this rationale be clearly justifiable and documented in a site procedure.
- Based on Target Process Confidence and Target Process Capability. Process Capability (CpK) is a statistical tool that can measure the capability of a process to meet quality requirements. Where the risk of process failure has been determined to be low for example, a CpK = 1.0 with 90% confidence seems reasonable, but as the residual risk increases, a higher level of confidence may be needed to show the process is performing at an acceptable level of capability. The level of confidence needed in the CpK needs to be fully defined i.e. at what point will the high level of confidence be reached-at completion of Stage 2 or during Stage 3. Again, it is important to fully document and define this approach, with clear linkage of the Target Process Confidence and the CpK level with the residual risk ranking, with justification provided.
- Expected Coverage. This approach is based on the concept of order statistics-with increasing number of PPQ batches, the probability of a future batch meeting the PPQ acceptance criteria also increases. This increasing probability provides increasing “coverage” and increasing assurance of the overall process robustness. For example, a process with a low residual risk would not demand a high degree of assurance from the PPQ batches, and vice versa. This approach requires the appropriate level of expected coverage that is meaningful for the process against the number of PPQ batches to be run to be determined and documented.
Other approaches may also be suitable for justifying the number of PPQ batches based on process and product knowledge, the control strategy, risk management and quantitative measures. The three approaches detailed above are further detailed in an ISPE Topic paper on Stage 2 Process Validation that is available to members on the ISPE website.