At the latest CMC Strategy Forum on Quality by Design, held July 19–20 in Bethesda, Maryland, discussions focused on an unlikely subject: noncritical process parameters. The core question was, How much regulatory oversight do they require?
The question about noncritical parameters arose in a broader discussion about how companies handle their manufacturing control strategy in the context of Quality by Design. For example, according to the ICH Q8 guidance, companies may make postapproval movements within a design space without regulatory oversight. But does that apply equally on the edges of a design space?
Noncritical parameters, which may not be captured in the design space, also present ambiguity.
At the Forum, industry professionals argued that companies should not be required to include extensive information about noncritical parameters in their regulatory filings, and instead should have the flexibility to manage these parameters through their internal quality systems.
Regulators, however, had a different view. They acknowledged that critical and noncritical parameters should be treated differently from a regulatory point of view, but did not want to completely lose the ability to monitor any of them.
“If we take noncritical process parameters out of the group of things that are visible to the agency, that raises concerns,” said Steven Kozlowski, director of the FDA’s Office of Biotechnology Products. “We need a way of having companies include some detail on change control for noncritical parameters, such that if a big change is made, the company would have to involve the agency.”
Is That Parameter Really Noncritical?
Kozlowski’s concern stems from uncertainty that risk assessments used to classify parameters as noncritical are accurate, particularly if those assessments are carried out early in product development.
“The understanding of criticality evolves during development and the product lifecycle,” Kozlowski said. “If some adverse event arises later that you think may be related to a potential noncritical quality attribute, then you should revisit it.”
On that point, industry members agreed.
“There seems to be a misconception that noncritical attributes disappear,” said Rohin Mhatre, vice president of biopharmaceutical development at Biogen Idec. “If you see random changes in a quality attribute, you would examine all potential triggers, including parameters not deemed critical.”
Review Branch or Inspectorate?
Minimizing the information filed about noncritical process parameters would shift their oversight from regulatory agencies’ review branches, which approve the license applications, to the compliance branches, which carry out inspections of manufacturing sites.
For members of industry, that shift would be preferable, because it would keep long QbD submissions from getting even longer.
“Already when we look at a QbD filing, there is a fair amount of information that we are including, so I would prefer not to add the quality management system in there,” said John Towns, senior director of CMC regulatory affairs at Lilly. “We want to get to the point where regulators are confident with how we’re handling our quality management throughout the product lifecycle, but I'd prefer to have that linkage occur between the review and compliance branches. I would prefer to handle it in inspection.”
“But inspections only last a week or so,” cautioned Kozlowski. “That is not a big slice in time.”
Industry-Wide Approaches, or Case-By-Case Negotiations?
Participants also questioned what regulatory guidance documents should say regarding what information companies must provide regulators about noncritical parameters, and more broadly about their overall manufacturing control strategies, including movement within and at the edges of a design space.
“We need to be able to translate this discussion into something we can share with people outside this room,” said Brian Withers, director of CMC regulatory affairs at Abbott, an industry rapporteur for the ICH Q11 guideline on drug substance development and manufacturing.
“A lot of these issues about critical and noncritical process parameters are fundamentally review issues and will be handled in negotiations with the review team,” noted Patrick Swann, deputy director of the Division of Monoclonal Antibodies in the FDA’s Office of Biotechnology Products. “Yet I share Brian’s concern. Can’t the guidance provide a little bit more than saying ‘the fundamental issues are review issues’?” Failing to provide guidance, he said, would lead to heterogeneity, with important matters being handled differently from reviewer to reviewer.
“Perhaps we can suggest a risk classification system linked to the control strategy,” suggested Kowid Ho, a quality assessor at the department for evaluation of biological products at AFSSAPS, the French health regulatory agency. Depending on the level of risk associated with a given process parameter, he said, different actions would be taken.
“For example, if a range were exceeded for a given parameter, in some cases that might require an extensive internal investigation that may lead to a regulatory submission, whereas for other parameters it might require intermediate levels of testing,” he said. “Others might not require much action at all.”
“That’s a very good idea,” responded Kozlowski. “Companies would manage their noncritical process parameters in their quality systems, but would share information in advance about how they would do that.”
Other participants fleshed out this idea, suggesting that companies could present the information in a relatively high-level tabular format, describing the types of changes they could foresee making, either within the design space or outside it, including the additional testing they would do and how they would report it.
Stefanie Pluschkell, an associate research fellow in global CMC at Pfizer,countered that no matter what was included in guidance documents, case-by-case negotiations would always be necessary.
“We may propose something and the regulators may disagree, in which case we may need to add more parameters to a design space,or shrink the design space, or add more to our control strategy,” she said. “These are not things you can put in a guidance; they will need to be negotiated.”
Also, Pluschkell cautioned, excessive standardization of the rules might hinder companies’ ability to make postapproval manufacturing improvements without regulatory oversight.
“Quality by Design is about continuous improvement,” she said. “If we don’t have that, industry won’t see the gain in doing Quality by Design.”
Do We Trust Risk Assessments?
Jon Coffman, a principal engineer in biotherapeutic pharmaceutical science at Pfizer, commented that it was ironic that the group was focusing so much energy on noncritical parameters. The purpose of risk assessments, of course, is to do just the opposite—to concentrate on the areas of greatest concern.
He presented this as a challenge to everyone in the room. “So what it comes down to is this: Do you believe in quality risk management?”
Kozlowski responded that the process of letting go is tricky, and requires a certain level of trust. “It's a bit like when your child goes off to college,” he said. “At first, you want to have a lot of phone calls. Eventually, over time, you may become comfortable that he or she is acting responsibly and needs less attention.”