Developing and setting specifications in the QbD paradigm should be a data-based exercise that allows for incorporating data
as it becomes available during the product lifecycle. Each specification has three key elements:7,8
a quality attribute or parameter that the specification is targeting
an analytical method that is used to perform the test
quantitative or qualitative acceptance criteria that determine acceptance or rejection of a manufacturing lot.
The first step is to perform a risk assessment to determine which quality attributes are important to the clinical performance
of the product (safety and efficacy).7 All product quality attributes are considered and assessed for risk; this approach is illustrated in Figure 1. Data available
on the product or from other platform products are analyzed.
Figure 1. Illustration of an approach for setting specifications for product quality attributes
After attributes have been identified for specifications, the next step is to come up with justifiable acceptable criteria.7,8,13,16,17 As discussed above, setting product specifications requires sifting through a variety of analyzing data from sources (clinical
studies, animal models, pharmacokinetic studies, analysis of manufacturing lots, etc.) This necessitates the use of appropriate
statistical tools. Multivariate techniques are more effective than univariate methods for the two-step process of first treating
the raw data and then combining data sets to produce meaningful results.17
Here are some key considerations that should be kept in mind when setting specifications:2,5,6,16
- Some raw materials such as cell banks may require specifications to ensure that no contaminants are introduced into the product. Excipients
added to the final drug product are expected to meet pharmacopoeial specifications.
- A subset of in-process controls may have specifications associated with them to ensure the absence of contamination or removal of an impurity.
- Validation studies may demonstrate that some process-related impurities such as DNA and HCPs do not require specifications.
Drug substance and drug product specifications are designed to measure the identity, purity, and potency of the product.
Control limits may be set based on process and analytical method variability (such as ±3 standard deviation). These limits can be useful
for identifying any trends, thus triggering an investigation to find the root cause before lot failure is observed.