Safe pharmaceutical manufacturing, as we know it today, could not exist without process validation. Introduced in the 1970s, when FDA was still known as the Bureau of Drugs, the concept was originally designed to protect the public against the impacts of variability in sterility assurance, but was extended broadly to drugs and devices.
It aimed to ensure that any process would do what its developers intended it to, and make products of reproducible quality. The concept worked.
It also became a monster. Over time, misunderstandings about validation promoted a rigid, bureaucratic approach to manufacturing. Concerns have led some companies to lock in inefficiencies, generate reams of defensive documentation, and reject new technologies or approaches, even if they would improve both process and product.
On November 16, FDA quietly released a long awaited new draft guidance, which, once approved and finalized, would update its 21-year-old guidance on process validation. Anyone working in the drug industry should read this document.
In addition to being far more engineering-focused and much more clearly written than the previous guidance, the new document stands out from the prior guidance in that it:
- outlines, clearly, a life-cycle approach to process validation
- presents the essence of what Quality by Design is, without going into the terminology
- emphasizes the importance of process control, calling out the importance of both QA professional and the line operator in providing feedback and
continued process verification.
- specifically mentions process stability and process capability
- mentions the potential role of simulation
- touches on the importance of continuous operator training
- highlights the need for a crossfunctional approach, even specifying the different disciplines that should, ideally, be involved.
- discusses the need for data analysis, within and between product batches, mentioning the most relevant places to find this information:
- discusses how PAT and real-time data access would change the approach to Performance Qualification
So what was so wrong with the prior guidance? I decided to check the wording of the 1987 document more carefully. It made me wonder about a few things….for one thing, where on earth did the magic number three (for the infamous “three validation batches”) come up? It wasn’t in that document.
The earlier document had already paved the way for QbD by highlighting the importance of drug development, yet it seemed to focus more on data collection than interpretation. It was also somewhat diffuse, taking many paragraphs to discuss control limits, and including lengthy asides on process examples that sounded like religious commentary on sacred scripts.
Some companies are already taking a very modern approach to process validation, and developing techniques that would allow processes to be adjusted on the fly. Last month, a meeting of the American Institute for Chemical Engineers (AIChE) in Philadelphia offered a number of practical examples showing how these concepts are being applied.
One expert from a U.K. company showed how multivariate models can be used to adjust automatically for differences in raw material sourced from different suppliers so that the resulting product’s quality is always consistent.
That is the future, but for many operations today it might as well be science fiction.
Recent informal surveys in our magazine have found that pharmaceutical development and manufacturing are still using mainly univariate rather than multivariate data, and that many companies aren’t even performing process capability analysis. The sigma level of drug manufacturing, experts say, remains around 3. At the same time, some FDA reviewers and inspectors may not be in synch with the new guidance. Audience members at one AIChE session asked why some FDA reviewers appear less interested in hearing about their carefully developed “design spaces” than in specific process values.
Even if there’s more work to be done, documents like this draft guidance provide clarity and a basis for open discussion and benchmarking. FDA is accepting comments on the document until January 19, via http://www.regulations.gov). I urge you to write with any constructive criticism. In the meantime, write us and let us know what you think.