Will PAT and quality by design ever become approaches that will be binding or required?
To the extent that it becomes an expectation that you incorporate some type of predictive scientific design methodology to
be deserving of approval, then yes. The agency has not said that it will be required, in fact these initiatives are being
launched on a voluntary basis. However, if they become a standard way of doing business, they are going to be required in
practice.
Will the technology for a specific operation, such as blending, ever require that it need to be taken in real time?
No. I don't think they will require any one specific way to do it. The essence of quality by design rests on the assumption
that there is a relatively small number of truly independent variables and by understanding the inter-relation between variables,
between process input and process output, you can determine the critical parameters you need to measure. And for some of the
critical variables that you may not be able to measure in real time, you can measure other properties that are correlated
to them closely enough so that you control and measure an important property using the correlation to the measurable variable.
Fundamentally, there is no need to really require any prescribed rigid approach. If you understand enough about the science
and the relationships, you can get the critical information by various means.
Looking back, can you comment on any mistakes that have been made in your field?
I think the fundamentals that we are developing today should have been developed much earlier, but I would not call it a mistake.
The current situation is the outcome of many years of doing business under a certain set of regulatory assumptions. I don't
think we should blame industry for that. Besides, the fact that we are developing that new framework now is very good. This
has been a change. NSF is supporting academic efforts, and academia, industry, and FDA now are working very closely in several
forums toward developing strategic plans for technology development, technology incorporation, and basically creating reservoirs
of objective, predictive knowledge that can be shared, for example, for developing product and process design methods as
well as creating large amounts of new coursework and training material that is essential for the development of the human
resource. I think this collaboration is unprecedented. The government, through the National Science Foundation, made a grant
of close to $40 million to our university consortium to really take a hard look at how we develop products and processes.
And industry has come in large numbers to enhance the government contribution with additional funding, equipment, and knowledge
to create a high impact program. I think this is unprecedented. The ERC was able to attract over 30 members in just six months
for a research and education program. The FDA was one of the first partners to join. I don't know of any other time this has
happened.
Why is this collaboration happening now?
It is happening now because now it has become necessary. Before it was very difficult. Even if we had been trying to put
in place a predictive toolbox that could be used for continuous improvement, we had a regulatory framework that made its application
impossible, at least to a significant degree, because making changes to approved processes was very difficult. Before, there
was no structural reason why it should happen and now there is. Also, there are large political reasons in the back of everyone's
minds. Among many statistics recently published, one is that 80 or 85% of drugs are taken by people who are 65 years and older
and the number of such people in the US population is expected to double in the next 15 years or so, so we need to make changes
that will help lower the cost of healthcare. Even if drugs are only a small fracton of healthcare costs, they are an easily
identified target, so we have seen a lot of interest from colleagues in helping us develop methods that might decrease the
cost of development and manufacturing.
|
