Product and process specifications.
Specification limits for the product and its process must be defined to protect the CQAs of the drug substance or drug product.
These limits may be set based on a transfer function (e.g., how does X influence the Y response) from a characterization study
or may be set statistically (based on some multiplier of sigma and/or risk) for those parameters that show no harm (i.e.,
clinical) and where variation is known. Specification limits will form a key basis for CPP determination. Specification limits
are primarily defined for product control rather than for process control.
Validation of analytical methods.
Limit of detection, limit of quantification, precision, and accuracy must be characterized for all analytical methods, and
method validation must be completed. Once these steps are done, one can trust the numbers and know the error associated with
any statistic of interest. Method validation should be done prior to product and process characterization studies and the
design and implementation of process controls.
Quality risk management for materials and operations.
A formal QRM process should be in place to systematically examine all materials and unit operations for their potential influence
to drug CQAs. Risk-ranking and other QRM tools are used to identify factors and unit operations that hold the greatest risk.
Scientific understanding and historical data are typically the basis upon which potential risks are identified and prioritized.
Candidate CPPs may be identified in this process that will later need to be ruled in or out based on data and identified risk.
Figure 1: Factor response matrix and risk assessment.
Many of the previous steps provide inputs to effective design-space characterization and optimization. DOE and multifactor
studies are used to understand the sensitivity of key product and process parameters relative to drug-product and drug-substance
specification limits. Factor selection prior to DOE generation is the most important step in design-space characterization.
The matrix shown in Figure 1 is used in the identification of the factors and responses that should be characterized and completed as part of risk assessment
prior to DOE design. One should take care to open up the range of the X factors sufficiently to understand their influence
on Y response and to be representative of the normal operational range of the process. Figure 2 shows a clear picture of the design space generated from a characterized process.
Figure 2: Design space characterization. The white area is within the specification limits while the shaded area is not.
Factor effect size and CPP selection.
DOE and multifactor experiments can help to isolate the influence of every factor and interaction on the critical responses
associated with the substance or product. Analysis of the DOE will generate the scaled estimates (one half the change in Y
relative to the change in X) also known as half effects, as shown in Table I.
Table I: Scaled estimates for a purification step.
One can convert the scaled estimate into the full effect (total change in Y relative to change in X) and compare the full
effect to the product specification tolerance. The formulas for conversion are as follows:
- Full Effect = Scaled Estimates * 2
- % of Tolerance = Abs (Scaled Estimates * 2) / (USL-* LSL) for two-sided limits
- % of Design Margin= Abs (Scaled Estimates * 2) / (Average-LSL) for one-sided LSL only
- % of Design Margin= Abs (Scaled Estimates * 2) / (USL-Average) for one-sided USL only
where, USL is the Upper Specified Limit, LSL is the Lower Specified Limit, and the average is the baseline process or product
average from the DOE or other lots.