Biomedical Engineering Reference
In-Depth Information
product manufactured without that process step. For example, the absence of an
ion-exchange chromatography polishing step may result in increased host cell proteins
and aggregates and a shift in product charge variants.
Understanding the product attributes affected by a process step would also be
important for a PATapproach. Therefore, multiple material attributes would usually need
to be monitored. It may be more likely to have a design space that combines operating
parameters with material attributes since not all important step functions may be
evaluated by sensors. For very complex process steps such as cell culture, sensors for
relevant material attributes or their surrogates could be very useful. The complexity and
uncertainty regarding attributes of complex products suggest that even with monitoring
of material attributes, movement within the design space will require risk-based
verification by the sponsor's quality system. As described for the more limited
chromatography design space, this verification should be specific to the uncertainties
associated with a particular process step. Thus, there would need to be a risk- and
knowledge-based approach to significant movement within a PAT-based design space.
Also as described above, even in the absence of major adjustments to the process,
verification of process performance is a life cycle issue. For PAT processes, the data from
monitoring through a variety of sensors and parameters can be integrated using
multivariate statistical analysis and/or control. In addition to verifying the design space,
these data would be very useful in process improvement.
2.8 QbD SUBMISSION THOUGHTS
There has been discussion as to the organization of QbD information in submissions to
regulatory agencies. There is no definitive approach for this; however, some possibilities
can be considered. For small molecules, much of the QbD discussion has focused on the
drug product. For biotechnology products, many complex process steps that can impact a
large number of product attributes occur inmanufacture of drug substance. Much of QbD
information can fit into specific sections of the Common Technical Document (CTD) for
registration of pharmaceuticals [48].
QbD information is best conveyed as a story from target profile, to CQAs, to process
development and controls for each unit operation. A sequential (backward or forward)
description of each unit operation would be helpful. Such a narrative should be located in
one place. Possibilities include the manufacturing description or development sections,
or an overarching narrative inModule 2: Quality Overall Summary. Ideally, the narrative
would be sufficient to give review staff a clear picture of the product and process
knowledge [12] and cite or hyperlink to the relevant detailed sections in other parts of the
submission.
The content to support a QbD submission is also being discussed. The goal is to
present knowledge, not raw data. However, there must be sufficient data to allow for
understanding. Graphical representations should be accompanied by appropriate statis-
tics and summary worksheets. Information based on models should describe the models
used and evidence for their validity. Sources for prior knowledge and approaches to risk
assessment and management should be well described. Initially, large amounts of data
