FDA Lowers Barriers to Process Improvement

In trying to bring pharmaceutical manufacturing regulation into the twenty-first century, the US Food and Drug Administration is doing more than talk. The agency is changing fundamental concepts of regulation and backing those changes up with initiatives to encourage process developers to keep improving their production systems—even after approval.

Eisai Machinery

As the first article in this series showed, some companies have refused to let outmoded thinking or FDA regulatory filings get in the way of streamlining manufacturing processes (1). Many of these companies are taking a "don't ask–don't tell" approach: through design-of-experiments and advanced real-time analytical methods such as near infrared spectroscopy, they are significantly improving their scientific understanding of their processes. Then, they use that knowledge to monitor operations and make processing decisions, but continue to run the validated test methods at the same time. This way, the company focuses on improving processes instead of on filing manufacturing supplements.

Adding process monitoring and control elements in parallel with traditional fixed-target analyses, however, can yield only limited gains. Many industry experts believe that significantly improving pharmaceutical manufacturing processes requires a much more flexible regulatory system, in which communication between industry and the regulators is open and fluid, and in which continuous process improvement is an everyday reality.

FDA already envisions such a world, and they have christened it the desired state (2). The agency has identified the structural problems to reaching that state; now they are trying to pave the road to reach it. Here is what they are doing.

The end to validation's "rule of three"

Pharmaceutical Technology recently asked Brian Hasselbalch, a compliance officer in the Division of Manufacturing and Product Quality in FDA's Office of Compliance, "How would you characterize the current industry approach to process validation?"

His answer was simple: "With the number three."

Hasselbalch was referring to the three batches that companies typically manufacture at commercial scale before applying for marketing approval of a new drug. If the three batches produced passing results, the thinking went, the process worked, and could be considered validated. The origin of that number lies in FDA's 1993 compliance policy guide on process validation, a document formally directed at agency staff but much read by industry.

Table I: Types of regulatory filings for scale-up and postapproval changes to approved new drug applications (NDAs) and abbreviated new drug applications (ANDAs).*

In March 2004, the agency revised the guide, to signal its evolving thinking about process validation (3). In one of the most significant revisions, FDA deleted the reference to three batches at commercial scale as adequate minimum proof of process validity—a number is no longer suggested.

The other significant change, says Joe Famulare, director of the Division of Manufacturing and Product Quality in FDA's Office of Compliance, was in the way the agency refers to those batches. "We no longer call them validation batches because validation is a life cycle approach and three batches do not represent validation," he explains.

Hasselbalch says that both industry and FDA have focused too much on the absolute number of replicant runs. "And in doing so I think we've gotten away from the original intent of process validation, which had a strong design and development element to it," he says. "In fact, a careful read of the 1987 guide shows that we intended always that the replication at scale would be preceded by good design and development work."

Updating the process validation guideline

The 1987 guide Hasselbalch mentions is the 1987 Guideline on General Principles of Process Validation (4), often confused with the compliance policy guide on the same topic. When FDA released the revised compliance policy guide, it also announced its intention to update the much larger 1987 guideline. Although FDA will not comment in detail on the revision until the draft is made public (possibly before 2006), Hasselbalch says the agency plans to reflect more of a "life cycle approach" to validation.

"The design, development, and verification of the process capability has to factor in the information gathered during routine production," says Hasselbalch, because even well-designed and well-developed processes sometimes need adjustment. "Processes may need to be optimized to improve yield or to reduce margins of quality," he says.

So there's hope that the new guideline will make it easier for manufacturers to improve their processes after approval. In the meantime, can changes in the drug filing system also help?

PQAS: New CMC review systems ease postapproval changes

FDA also is considering ways to give manufacturers flexibility by changing the way the chemistry, manufacturing, and controls (CMC) information is submitted in new drug applications—both NDAs and ANDAs.

In Sept. 2004, the Office of New Drug Chemistry announced that it is developing a new system for quality review, to replace the existing CMC review process (5). Through the new "pharmaceutical quality assessment system" (PQAS), the office wants to receive more information in new drug applications about pharmaceutical development, manufacturing science, quality-by-design and its implementation in unit operations, and the scientific rationale behind a sponsor's proposals.

The promise of this new system is that by having companies submit appropriate information about the scientific underpinning of their processes, FDA will be able to grant companies more latitude to make postapproval changes.

The new system also will include a new quality overall summary (QOS). Moheb Nasr, PhD, director of FDA's Office of New Drug Chemistry, says the QOS will prompt companies to carefully review the critical aspects of their applications before submitting them, thus improving the quality of the submissions and saving reviewers time. "Now, we get tons of information without a coherent summary or description of what the application is about," he says, which makes it difficult to identify the critical CMC issues. The old system, he says, was like doing graduate research and handing in your laboratory notebooks; the new system is more like writing a manuscript to summarize and analyze the findings. In this way, he says, "the assessment of the CMC submission starts at the sponsor before it comes to us."

CMC pilot program. To pave the way for the PQAS approach, the agency is conducting a pilot CMC program (6). Through the pilot, FDA hopes to obtain additional information from companies that will help the agency develop and implement the new review system.

The agency is seeking 12 pharmaceutical companies to volunteer to participate in the pilot program by submitting original NDAs. The announcement was made in mid-July; by the end of August, the agency had received five applications and several inquiries, Nasr says.

"The CMC pilot program provides an opportunity for industry ... to share with us, in an open manner, the kind of scientific information they currently have, and provide the justification to allow for regulatory flexibility," says Nasr.

For example, if a company uses design-of-experiments to study the effects of particle size distribution on a drug product's performance, it may be able to prove to FDA that particle-size distribution within a wider range will not significantly affect the performance of the drug product. "So when we set specifications, that specification would be a wider particle-size distribution, since we have the scientific evidence of its impact—or lack of impact—on the performance of the product," Nasr explains.

This approach contrasts with the existing system, in which manufacturers typically submit limited data, based on a single particle size or a very narrow range. "So if you do a single experiment, and you carry it through manufacturing, that specification will be set around the information we have from that single experiment," he says. "When they come to us with a single point, what other choice do we have?"

The international scene: ICH Q8, Q9, and Q10.

Nasr notes that some companies already conduct the experiments needed for such justification, and he believes the pilot, and the new system, will encourage firms to share that data with FDA. "The data ... will result in fewer supplements and provide more freedom to manufacturers to continue to improve their processes," he says.

The new concept of "design space". The concept underlying this approach—of leaving companies in control of a limited set of changes within a certain set of parameters—is the idea of design space, which has been defined in the International Conference on Harmonization's (ICH) Q8 document on pharmaceutical development (7) (see sidebar, "The international scene: ICH Q8, Q9, and Q10").

Design space is a multidimensional space that defines the parameters and acceptable ranges of a process. Its boundaries are the limits of interrelated specifications, including time, within which product quality will still be maintained. Companies that develop and define their design space would have the opportunity for continuous improvement or optimization within the bounds of their design space, without the need to file CMC supplements or reports. In this scenario, postapproval continuous improvement would be managed within a company's quality system and subject to FDA's inspectional oversight for compliance with current good manufacturing practices (CGMPs). "The design space concept was intended to change the definition of change itself, saying movements within the design space are not changes," says Ajaz Hussain, PhD, deputy director of FDA's Office of Pharmaceutical Quality.

Of course, granting companies such flexibility requires that the company be able to demonstrate a high level of process understanding. "[T]he information we will receive from the sponsor, [will] help establish the design space," says Nasr. "We will provide industry the opportunity, using their own existing robust quality systems, to manage manufacturing within that design space." And that design space can be large, he says, if the company has data to show that product quality will be maintained in a large space.

New FDA postmarketing division to support postapproval changes. To make the revamping of the quality assessment system work, Nasr is restructuring his office. Key changes include establishing a new branch for assessing manufacturing science and hiring staff with expertise in areas such as pharmaceutical development, pharmaceutical engineering, and physical pharmacy to complement the chemistry experience the office already had. And in the new system, each submission will be handled by a team of reviewers with varied expertise, rather than by the current single reviewer.

After the restructuring, separate branches will handle premarket submissions (INDs and NDAs) and postmarket submission (CMC supplements). This will accelerate new-drug reviews by allowing one set of reviewers to focus their time and attention on INDs and NDAs without interruptions to review supplements. Second, it will streamline the review of postmarket changes. Currently, the review of postmarketing changes is scattered among 19 chemistry teams in 15 clinical divisions. "Now it would be consolidated to only one division, within our office," Nasr says.

Third, the new postmarketing division will have a greater ability to evaluate the types and frequency of postmarketing supplements. This change, Nasr believes, could allow the agency to develop strategies and guidances related to what kinds of supplements could be eliminated and what kinds of changes could be down-regulated to less-onerous filing categories.

Question-based review for generic-drug filings. At the same time that this new review system is being developed in the Office of New Drug Chemistry, the Office of Generic Drugs is also developing a Question-based Review (QbR) system, that promises manufacturers regulatory relief if they submit more-complete product and process development data. In the current system, ANDA sponsors do not submit product development information or a quality overall summary.

"We will have more information about the way the product was developed and the way the process was developed," says Gary Buehler, director of FDA's Office of Generic Drugs (OGD).

"With that additional information, we will be able to allow certain regulatory relief with respect to supplements." For example, a change that would normally require a company to file a prior approval supplement may only require that the company file a change-being-effected supplement. Thus, the Office of Generic Drugs is applying the concept of design space.

Lawrence Yu, director for science at OGD, says that the office can even go beyond the design space submitted by the sponsor. FDA reviewers have a lot of accumulated knowledge, which they can apply when reviewing drugs. So in some cases, even if a manufacturer has not submitted data for a broader specification range, the agency may be able to grant some flexibility, based on its own knowledge. "We can make decisions based on the sponsor's data and our own internal knowledge base," says Yu. Buehler emphasizes that the agency would not use any proprietary information to allow other applications to get approved. "It's basically the information in the literature and the information that's accrued over the years that is available to us," he says.

OGD hasn't worked out the details yet. Buehler says he expects the increased flexibility to be stated in the approval letter, although he doesn't know what kind of wording will be used.

One way or another, the agency will find a way to make it easier for manufacturers to improve their processes. "We don't want to stand in the way of pharmaceutical innovation," says Buehler. "We don't want to stand in the way of companies being able to modify their process and make the process better."

Comparability protocols—because we learn about processes as we go

But what happens if a company doesn't have that much data available at the time of submission? That happens often. Because delayed approval of new drugs can cost millions of dollars, pharmaceutical companies typically don't want to delay filing by conducting extensive manufacturing studies. Later, through commercial operations, manufacturers often gain experience that would allow them to make process improvements.

FDA has proposed comparability protocols for postapproval CMC changes. In its Feb. 2003 draft guidance, the agency described a way for companies to reduce the paperwork for postapproval changes by showing that what a company intends to change is comparable to what was in the original filing (7). Through such protocols, a change that normally would require a prior approval supplement could be down-regulated, so that it only requires a changes-being effected filing. Similarly, other changes would move down a level from their standard filing requirements.

The problem, industry experts say, is that the comparability protocols guidance focuses on discrete changes and recommends that companies don't combine too many changes in one comparability protocol application. This limits the usefulness of comparability protocols. "This is a step in the right direction, but I think for a true process understanding process change system to work, it has to be holistic," says Richard G. Whitfield, PhD, senior director and team leader of Global Quality Technical Services at Pfizer.

Nasr says the agency understands those concerns and is revising the comparability protocols guidance to try to make it a more useful tool. "We are simplifying and streamlining the process," he says. But he quickly adds that the amount of regulatory relief provided by the agency must be proportional to the level of scientific evidence companies provide about their processes. "The better you understand your process, the more we are comfortable as regulators that you can operate within the design space without a problem. We have the responsibility to assure the safety and efficacy of the product, and we would be reluctant to let go [of control] without such scientific assurance."

More flexibility up front; more accountability later

What this flexibility means is that more responsibility for compliance and maintaining quality shifts from FDA to the company.

"It is not losing control; it's basically making industry more responsible for the product," says Nasr. "We at the agency do not discover drugs, do not develop drugs, we do not manufacture drugs. So what's our responsibility? Our responsibility it to make sure that the drugs, before they come to the market, [are controlled for their] critical quality attributes, to assure their safety and efficacy. And that means we have to evaluate information coming to us to provide that assurance without creating an environment that prevents or delays approvals or continuous improvement."

Hussain agrees that regulatory flexibility won't mean a lack of regulatory control. Freedom to make low-risk changes in well understood processes without filing CMC supplements, he says, would be balanced by control over compliance with current good manufacturing practices. In other words, regulatory scrutiny would be applied through inspections for CGMP compliance. "It's not that continuous improvement change control is not regulated," says Hussain, "In fact, our regulatory hammer for penalizing somebody who doesn't do things right is higher on the GMP side than on the CMC review side."

Old-style validation will remain valid

Hussain also emphasizes that these incentives are only for companies that understand their processes and can demonstrate a state of control. "If other companies are not able to do that, then the current system remains for them," he says.

Famulare agrees. "There can be process validation approaches taken with conventional manufacturing processes that make the best use of data obtained batch to batch, from the use of statistics, to obtain better control over the process," he says.

The hope, however, is that soon no one will want to do it the old way.

References

1. L. Bush, "The End of Process Validation as We Know It?" Pharm Technol. 29 (8), 36-42 (2005).

2. US Food and Drug Administration, Innovation and Continuous Improvement in Pharmaceutical Manufacturing. Pharmaceutical CGMPs for the 21st Century (FDA, Rockville, MD, Sept. 2004) p. 8, www.fda.gov/cder/gmp/gmp2004/manufSciWP.pdf (accessed Sept. 19, 2005).

3. FDA, "Sec. 490.100. Process Validation Requirements for Drug Products and Active Pharmaceutical Ingredients Subject to Pre-Market Approval (CPG 7132c.08)," (FDA, Rockville, MD, March 2004), www.fda.gov/cder/gmp/processvalidation.htm (accessed Sept. 19, 2005).

4. FDA, Guideline on General Principles of Process Validation (FDA, Rockville, MD, May 1987).

5. FDA, "ONDC's New Risk-Based Pharmaceutical Quality Assessment System," (FDA, Rockville, MD, Sept. 29, 2004), www.fda.gov/cder/gmp/gmp2004/ondc_reorg.htm (accessed Sept. 19, 2005).

6. FDA, "Submission of Chemistry, Manufacturing, and Controls Information in a New Drug Application Under the New Pharmaceutical Quality Assessment System; Notice of Pilot Program," Fed. Regist. 70 (134), 40719–40720 (July 14, 2005).

7. FDA, Draft Guidance for Industry: Comparability Protocols — Chemistry, Manufacturing, and Controls Information (FDA, Rockville, MD, Feb. 2003).

8. FDA, "International Conference on HarmoniZation; Draft Guidance on Q8, Pharmaceutical Development; Availability," Fed. Regist. 70 (26), 6888–6889 (Feb. 9, 2005).

9. FDA, "International Conference on HarmoniZation; Draft Guidance on Q9, Quality Risk Management; Availability," Fed. Regist. 70 (151), 45722–45723 (Aug. 8, 2005).

10. Richard G. Whitfield, "Reaping the Rewards of PAT from the Regulators: PAT and Postapproval Changes," paper presented at the Institute of Validation Technology's Process Analytical Technology conference, Philadelphia, PA, June 13–14, 2005.