FDA Regulatory Panel: 2023 ISPE Aseptic Conference
On 7 March 2023, ISPE continued its tradition of concluding the 2023 ISPE Aseptic Conference with a regulatory panel question and answer session. Attendees were invited to submit questions to the FDA representatives. This article offers highlights from the discussion.
At this year’s session, most questions focused on technical pharmaceutical queries related to sterile products, whereas last year’s panel had a lot of questions around audit practices during the pandemic and learnings from virtual and hybrid inspections. The insightful questions and open discussion made for a very successful session, and participants went home with information that will help them in their jobs.
A highlight of the session was the mutual praise from industry and regulators on how well the two parties worked together during the pandemic to bring vaccines and drugs to patients in record time. Please note that views expressed by the panelists are not necessarily representative of the position of the FDA, and that questions and responses are lightly edited for clarity.
The FDA representatives in this panel were:
We hear people discuss the concept of second air. First air refers to the filtered air that has not been interrupted prior to contacting exposed product and product contact surfaces with the potential to add contamination to the air prior to reaching the critical zone. Do you align the phrase with the potential to add contamination? Because if you look at most all filling systems, there are all sorts of surfaces, tubing, stainless steel parts, and the filling needles themselves dancing over the tops of the open units being filled. How is this considered not interrupting first air or is first air only interrupted by the gloved hand of an operator?
I don‘t think many of us would want to subscribe to the concept of second air as appropriate. The standard is first air, and that means that there is no major blockage of air that would cause a perturbance, or disruption, of air in ISO5 zone, and special attention should be afforded at the area of exposed sterile vials, syringe barrels, or bags. This needs to be demonstrated in the smoke study qualifications. By design, machines are done in the right way these days, avoiding any potential air disruptions as much as possible. If not, then you need to go back to the drawing board with your vendor or go to another vendor.
How will the FDA use Annex 1 in its final version?
Annex 1 is aligned and in harmony with the FDA Guidance for Aseptic Processing from 2004. The FDA has been involved in the development of the final version of Annex 1 via PIC/S (Pharmaceutical Inspection Convention/Pharmaceutical Inspection Co-Operation Scheme), which meant a lot of long and early calls in the committees. The FDA’s guidance has different topics and details in it that are additive to Annex 1, but they are compatible. The FDA’s guidance was written to last for a long time to provide explanations and technical content that facilitate cGMP compliance, without being overly prescriptive. We describe the principles that facilitate voluntary compliance by the industry. While local policies are in general alignment with PIC/S guidelines, there are some aspects that add further depth and explanations in regional guidelines from EU as well as other PIC/S member states on various topics.
With the mutual recognition agreement between the EMA [European Medicines Agency] and the FDA, do you think the FDA will also expect PUPSIT [pre-use post-sterilization integrity testing] for products released to the US?
Our guidance does say that integrity testing can be performed prior to processing and that it should be routinely performed post use. It is similar to the language in Annex 1, which indicates that pre- and post-use testing should be done noting that an alternative pre-use approach including a risk assessment might be scientifically justified as sufficient on its own. Our guidance may not be identically worded to Annex 1, but it is equivalent and aligned.
Does the FDA have expectations for education with virtual reality (VR) technology? It can visualize smoke tests, which otherwise is difficult to understand. Aseptic operation depends a lot on skill of operators, so using VR could be valuable.
FDA does not promote the use of one tool over another. With that said, anything that helps employees understand aseptic technique and the importance of smoke studies is valuable.
Have you considered using your smoke studies for training? Properly performed smoke studies are really a wonderful tool to show your operators how to mitigate potential risks while performing interventions on the line. Operators can easily see the air flow patterns and disruptions caused by their movement. And then they might be able to perform those interventions in a better manner because they know exactly what the consequences of their actions are.
If a fully automated process with a gloveless isolator does not require operator intervention in the critical zones, could that filling machine potentially be placed in a grade D or even less stringent room?
You must consider your sterile core where your filling is happening. Even though it‘s operating closed, you‘re going to need to open it for any maintenance procedures, for any preparation procedures, for cleaning procedures, and every time you‘re setting it up for another filling process. So, you want that sterile core protected. You want it in a higher grade than a D because you don‘t want to impact the cleanliness of that sterile core in any operations that you need to do when it‘s open. You want the surroundings as clean as possible. We would not recommend a D or less stringent room for this kind of operation.
Could you comment on the background requirement for sterility test isolators? Is it okay to install them in CNC [controlled, not classified] or even noncontrolled areas?
We assume this is a closed isolator. There’s a paragraph that was included in the original aseptic processing guidance that was published back in 1987, and I consider it to be policy poetry. It basically says that the sterility testing ISO 5 environment should be protected so that it is “comparable” to the aseptic processing operation. It includes this advisement because of the concern that a positive sterility test might be erroneously invalidated due to the assumption that an unreliable laboratory caused a false positive, even when the drug tested was actually truly nonsterile. So, if the sterility testing conditions are far more vulnerable to contamination than the aseptic production operation, there may be a subjective tendency, although ill-advised, to deem any sterility positive result to be a false positive.
We preserved the same principle in the 2004 rewrite of the guidance. A laboratory problem should not be masking a real non-sterility problem in production. I’m assuming for this question that the set-up for production is probably an open isolator, a closed RABS [restricted-access barrier system], or something similar. Closed isolators provide a highly robust barrier between the surrounding environment and the internal ISO 5 environment. For that reason, a closed sterility test isolator in CNC (controlled non-classified) would generally be fine. Obviously, the isolator needs to be properly designed and controlled, and the lab environment needs to be clean and orderly to enable CNC to be appropriate for the closed sterility testing isolator.
What are the most common observations in the past year, and how should a company respond to them?
For me, the two most common observations are not having the data to support your room classifications and media fills that are not representative of routine manufacturing. During inspections, I find beautifully written reports that are delightful to read. However, I‘m always going to look at the data to support the classifications justified by those reports and that‘s where I‘ll find problems. There is no raw data to support the claims made in those reports. The second most common observation I find is media fills that are not representative of routine manufacturing. Whether it‘s not documenting all the interventions during routine production so it is unknown how many interventions should actually be done during a media fill or not trending interventions, or the interventions performed during a media fill are not representative of the production process.
First, I‘ve seen a lot of procedures that are either inaccurate or they‘re lacking the detail for production to perform their tasks correctly and consistently. That is a big one. Another one is that qualifications and validation reports and the data from the qualification don‘t get transferred into the process parameters or into the procedures to represent the way that the system was validated. Whether it‘s cleaning validation, autoclave validation, or disinfectant efficacy qualifications, they don‘t get transferred over into the procedures to get executed the way that they were validated.
We are seeing a lot of issues with smoke studies. There are deficient air flow patterns that are obvious to us when we‘re watching the videos that are not being caught by the firms. During regulatory meetings, we’ll discuss these concerning airflow patterns with the company representatives and provide specific time frames in their videos where we see issues with their smoke studies that they did not catch.
In general, when you respond to our observations, we want to see a robust, comprehensive, systemic response—not a narrow response that only focuses on specific examples cited on the FDA 483. For example, there is an issue with one investigation. The company opened up that one investigation again, but they may need to expand their review to their full investigation system and consider all of the equipment that might be similar to the one piece that had an issue with or all media fills—not just the one that had an issue—and then provide supporting evidence for commitments. Finally, there is the topic of time frames: When are you going to implement some of the corrective actions, and what are you going to do in the meantime?
Concerning the last point of timely close-out of investigations, we also train our investigators to ensure that companies do not prematurely close out. Lower complexity investigations can certainly be closed out in 30 days, but we recognize some investigations necessitate extensive CAPAs that will take much longer than 30 days to fully implement. Merely citing companies on the 483 for not being in the 30-day time frame, on its own, would not suffice. The inspection should be evaluating whether certain high complexity investigations that will require extensive remediations, such as redesign of the process, may require longer timeframes. With that said, some action plan may need to be in place by Day 30. Companies should have a way forward and it should be documented in their system that they have created at least a plan of action, and interim controls. So, if a firm will need to embark on an extensive root cause analysis and remediation process, closing out some of the more difficult deviations or failures in 30 days can be very difficult. What we are looking for, in such cases, is valid explanation in your system and a sustainable remediation of a serious issue that really will prevent reoccurrence. Upgrading to new technologies can be a 6-12-month endeavor to initiate, and we would rather see those plans mapped out, with appropriate interim controls in place.
Please also don’t forget to evaluate the product that is on the market and that could also have been impacted by the issue discussed in the observation.
How often should gloves for RABS [restricted-access barrier systems] be autoclaved?
You should look at the frequency based on your processing durations and how you are using the gloves. Check what the manufacturer recommends, because some gloves have a certain number of qualified autoclave cycles, and may need to replaced after multiple resterilizations. There are different requirements based on the gloves that you are using. What I would typically do is look at the firm‘s program of how they are autoclaving or treating their gloves and see what their use is, how their campaign manufacturing is, and see what their rationale is behind supporting their program for autoclaving their gloves. There is not a simple answer.
Risk equals exposure. This is a core theme behind contamination control strategy in Annex 1. We very much subscribe at FDA to its purpose to ensure a lifecycle approach that adapts the strategy for ensuring ongoing state of control of the sterile operation based on iterative knowledge gained over time. Based on the factors that Laura mentioned, you’ve got to sterilize RABS gloves with appropriate frequency, with an understanding that cumulative contamination risk may increase over the course of a multi-day campaign, and these decisions should be based in good science and sound risk management.
What is the reason for the extra year given to section 8.123 in Annex 1 for sterilization and loading of lyophilizers for implementation?
We refer to the EU for Annex 1, but firms should be following an appropriately stringent frequency for lyophilizer sterilization already and justify a risk rationale when not sterilizing each batch. Fully automated transfers can be the basis for that rationale, and the EU authorities wanted to provide a chance for firms to conduct risk assessments for existing equipment and to automate lyophilizer loading and unloading if possible. This does not mean that an investigator cannot currently cite a firm for their sterilization practices and frequencies if they are not scientifically justified.
Will it be possible to retrofit traditional filling lines to comply with Annex 1? If yes, what will be the minimum requirement? In yesterday’s panel discussion, it seemed that the panel members did not fully agree on whether going from 30 to five open door interventions per batch would be compliant, whereas another mentioned that zero open door interventions during production should be the minimum requirement. What is the minimum requirement for those lines? This is the dilemma in question many pharma companies are struggling to answer.
Some of the warning letters I’ve been involved with recently have addressed deficiencies related to older “traditional” filling lines. Firms respond many times that they are going to retrofit the lines. And it has not been an easy road for these firms. There is a lot to consider when you retrofit a line by putting a RABS around it. One of the things that we want to think about is whether it’s even a good idea to start with such a project.
Let’s compare it to similar situation: What if you have an old car? At what point do you say to yourself, “I’ve been putting in all this money. I’ve been fixing it, but maybe I should just should have gotten a new car instead, something that is more reliable.” It’s not going to break down as often. Maybe you want to move to a newer line that is more stable, a line that’s not going to require as many interventions. A line that’s not going to require as much maintenance as these older lines.
More modest improvements are possible for existing lines, but attempting to retrofit a traditional line to a RABS can be difficult and they often lead to poor outcomes, as Brooke noted. I understand why there was disagreement on the panel yesterday, as there is a lot of debate these days about whether it is wise for aseptic personnel in the year 2023 to continue to directly interact with sterile drugs during an aseptic processing operation without the barrier wall protection afforded by isolators and closed RABS.
With the US being a member of PIC/S, are products manufactured and marketed in the US required to meet Annex 1?
PIC/S membership is extremely valuable, as it actively supports convergence on inspector training. Besides convergence on inspection practices through shared training, there are additional benefits from policy engagement that lead to aligned thinking. The regulators weren‘t talking to each other en masse through any formal mechanism before we started working with PIC/S. I mentioned already that there is the question of sovereign policies and other local aspects. But PIC/S provides guidances that are authoritative and they do generally provide strong baseline standards that we refer to in our regulatory work.
Does the FDA support the requirement for species-level identification for all microbial recoveries in equivalent grade A and B areas?
Yes. We expect you to identify all isolates to species level wherever possible in critical areas and immediately surrounding clean areas, as well as personnel in aseptic processing rooms.
The Emergency Use Authorization approval for the COVID-19 vaccines was a major success. Can the FDA adopt principles of how the EUA was implemented to expedite approval of drugs?
I will take that question, as I was member of the team that evaluated the vaccines, including the Pfizer vaccine, for the last three years. The agency was working on all parts of the approval process in parallel, the same as industry. And so instead of all the different steps being branched out and one after another, they were stacked and in parallel, which helped industry accelerate the development of the COVID-19 vaccines. We also stacked our review, looking at the use authorization for the vaccines, and we stacked it with multiple resources we borrowed from other parts of the agency.
This meant that there were other drugs and other applications that we put on the back burner because we put all our resources into the COVID-19 vaccine approval, or authorization and then approval. We couldn’t do this the normal way. This was a process that was justified for this health emergency, and we would not be able to do it on all drug applications. It was quite an experience and an amazing effort at the agency on all levels of management. But no, not doable under normal circumstances.
For indirect contact parts, is it necessary to employ sterilization? There are lines with no autoclave. Is there a way to accept decontamination with bioburden assessment and robust pre-cleaning, that is, if you can remove the parts?
You should autoclave the contact parts like guard rails. If they‘re removable I am curious as to why there are sites with no autoclave. I would expect if you can take format parts apart, you will autoclave it.
When introducing rapid micro methods, it’s quite difficult to prove the equivalency to current environmental monitoring, as they are essentially monitoring different things. RMM [rapid microbial monitoring] is for real-time process monitoring, whereas petri dishes are used for microbial analysis. What would be a recommended approach to introduce RMM?
Both technologies are intended to detect microbial contamination. The difference is in the frequency. The traditional use of settle plates is monitoring for a finite amount of time. That’s much less than continuous monitoring can achieve. There are some real advantages with RMM if it is continuous monitoring, if it is implemented properly and is a well-designed, valid method.
There are some systems out there that look extremely promising. Some are used for microbial monitoring in the isolator and are also monitoring nonviable particles. The bottom line is that our recommended approach is to show comparability. Environmental monitoring programs should detect a departure in state of control by providing an early warning system that can enable proactive action related to any in-process product that may be compromised. It is also critical for microbes to be identified. Microbial identification remains a basic, core element of monitoring the suitability of the ISO 5 aseptic processing environment.
For gloveless isolators, is it okay to omit contact plate monitoring?
No, we recommend that end-of-campaign monitoring should be incorporated.
How often should a disinfectant efficacy program be reevaluated and what principles are most commonly missed when creating a program?
You would want to reevaluate your disinfectant efficacy program if your facility isolates have changed. It should be part of your program to be monitoring the isolates in your facility. You also would want to reevaluate your program if you are adding or changing any cleaning agents, as well as when you are seeing any drastic changes in your monitoring data.
The part that is most commonly missed when we are looking at a disinfectant efficacy program and when we‘re looking at the qualification of the cleaning agents is that the different contact surfaces in the facility are not always included. Sometimes we see very common surfaces missing from the studies, like stainless steel.
When are anaerobic media fields needed when filling under nitrogen headspace? Initial or regular?
Initially if it’s under nitrogen cover and then periodically after.
How often should anaerobic monitoring be performed for anaerobic processes?
This should be based on risk assessment: maybe monthly or quarterly in typical situations.
Does the FDA approve the use of laser oxygen head space analysis in place of manual inspection for aseptic process simulation?
I’m not sure how it would be an effective surrogate. There are methods that showed some promise in the past that used laser technology to help detect microbes in media fill vials, but I have not seen data at this time that would support oxygen head space as an adequate surrogate.
Can Brooke Higgins please repeat the statistics you shared regarding human contamination impacts on product from yesterday morning? How many pounds of microbes do we carry? I want to highlight this data when I talk hygiene, behavior procedures.
I’m so excited about this question: Microbes outnumber our own human cells 10 to 1 and make up about 1% to 3% of our body mass. In a 200-pound average adult, it’s 2 to 6 pounds of bacteria.
Section 8.67 [Annex 1]: How and how frequently should the side rails in cooling zones of dehydrogenation tunnels be sterilized or disinfected?
I’m interpreting that this would be sterilizing the cooling zone in your tunnel, and it is something that that we do cite for, if a company doesn’t know how to sterilize their cooling zone. When you open your tunnel to grade C and do maintenance—for example, replace HEPA filters—you really need a way to disinfect/sterilize it and usually it’s dry heat sterilization. Often, there is also a sterilization of the cooling zone at intervals in routine production.
We have an open prefilled syringe isolator with a grade A air supply over two bag openers. Will outfitting an active mousehole with a grade A air supply hood, with surrounding room, with one-step material airlock, with palletizing of tubs classified and qualified as ISO 8 with class C microbial limits be acceptable?
In the surrounding area, class C microbial limits would usually be acceptable because we expect ISO 8 in operational state, or better, as the background for isolators.
What are the FDA’s thoughts on using phenotypic ID systems for microbial identification?
They have always been accepted if they sufficiently identify the isolate. Our environmental monitoring section of the associated guidance says that whenever possible, identify to species level. But if not that, at least to genus level. So as long as the phenotypic or biochemical ID system can do that reliably, it would usually suffice. ID systems that are less precise in terms of differentiation and can’t reliably get to genus or species level would be insufficient. With that said, many firms are moving to genotypic systems these days. At minimum those methods use 16s sequencing to look for 500 base pair homology, or even higher capability genomic methods, that make comparisons to an established microbiological genome library that helps identify the species.
What is the impact of Annex 1 on loading of the lyophilizer? Would manual loading using as isolator with a half suit be acceptable?
We would want to see your risk assessment and qualification, and the data to justify that set-up. These are typical documents I would want to review.
Is quality by design a regulatory requirement or industry best practice? Are QbD rules enforced?
Quality by design was a government initiative. We were trying to promote quality by design plus quality systems to create the 21st century approach to pharmaceutical manufacturing. We intended to create more momentum toward modernization because industry was considered to have too much obsolescence in its facilities, and quality was suffering. There was good regulator and industry consensus that a new paradigm was needed. What is a requirement under the actual regulations is designing your process, equipment, facility, operations, and procedures adequately to assure ongoing robust control. The word ”design” is used in the GMPs many times. So, in effect assuring quality through sound design is a requirement.
Adding to that, ICH Q8 for process development includes quality by design and design space.
Yes, ICH Q8 does establish a standard for the development of products and processes, but here at this conference we are talking more about the facility design. As Brooke was saying, your design may need to be revisited during the life cycle, and that applies to your product, process, and facility.
When I teach on ICH Q 8, it’s always about why are we using the parameters that we’re using? Why is 500 RPM in compounding good? Why is it not 450? Why is it not 600? That kind of data comes from a quality by design approach of having a scientific data to justify your process.
Absolutely, acceptable ranges for parameters as well as critical raw material attributes must be justified by good science. And it is hard to do science-based design without a strong quality system and quality assurance department. So, the substance you are discussing is exactly right and it’s beyond how QbD may be understood by some. ICH Q8 normalizes it well and provides standardized approaches that are effective in designing your process to support successful commercialization.
How is the use of the real-time viable monitoring viewed as a replacement for settle plates and active sampling?
For air monitoring, it has the potential to be a replacement. We are not generally bullish on the use of settle plates; we prefer active air monitoring. What is always necessary, irrespective of which type of monitoring method you use, is identification of the microbes that you detect. The microbe isolated from an ISO 5 environment provides pivotal information regarding state of control of an operation and when investigating the source of contamination.
We would like to add one thing about phenotyping methods. You may use API strips or Vitek or similar methods, which is probably a routine approach to environmental monitoring and identification. However, if you have a sterility failure or a media fill failure or other kind of investigations where you’re trying to make links, it may be appropriate to use sequencing for those investigations. It would be appropriate for an investigator to ask that question.
Whether you genotyped or did whole genome sequencing—anywhere from ribotyping all the way up to whole genome sequencing—should depend on how serious the issue is. Sometimes phenotypic methods are inconclusive about even what the genus is. And sometimes it’s very important to figure out what the species is to make matches between slants you might have in your refrigerator from that week of environmental monitoring and the sterility test isolate; you could make an actual exact or close match if you make the right methodology choices in the micro lab.
TGA did a beautiful job around 2000 stating in their guidelines that when you are doing a sterility investigation or media fill investigation, you should consider the use of genotyping. The FDA said the same thing in the 2004 aseptic guidance, so it is in international regulatory guidance.
Just based on recent news, drug shortages continue to be a problem. Last week, a company just folded up. There’s been other companies that left, and we’ve had such excellent work and presentations in this conference about issues that need to be resolved. What can we do as an industry and with health authorities to improve the success rate of that outcome so that companies stay in business and continue to supply what’s needed?
In a market economy, there are some companies that fail and there are myriad reasons. Sometimes companies have data integrity issues, sometimes they have manufacturing issues, sometimes it’s nothing to do with that. They may have made some poor business choices related to drug application filings and marketing. The generic industry is particularly competitive.
We know that 60% of shortages, as our data has shown consistently, are due to manufacturing quality issues. And, at ISPE, you have a Facilities of the Future Conference that promotes the demystification of modern beneficial technologies. There are a lot of great contemporary technologies that companies should embrace now and without delay to maintain the “c” in cGMP to ensure business continuity through dependable operations and to optimize their conformance with regulatory requirements. We think that the more industry moves toward reliability engineering and keeping your eye squarely focused on risk management and the patient, it will be to the advantage of all stakeholders. Manufacturing determines quality. Choosing robust manufacturing technologies will create the needed conditions for dependable quality and supply. Modernization of the sterile industry as discussed in this conference includes technologies such as isolators and robotics, and full adoption of such technologies will go a long way towards putting an end to shortages been associated with non-sterility risks.
Outreach is important. We’re here talking to all of you and I’m excited that there are 20 students attending this year, which is phenomenal because then they get involved earlier in their education and career.
It’s just more a follow-up question on drug shortage and supply. I believe there is a backlog in applications due to the prioritization because of the pandemic. Can you speak a little bit on what the FDA is doing differently to be able to accelerate or get back to normal? Has the agency had some staff augmentation to help and get ahead of this issue?
During the pandemic, CBER and CDER experienced similar challenges. CDER had many EUAs and essential medicines priorities that required special attention. We also did a lot of mitigating regulatory work and exercise of discretion during the pandemic to keep supply in the marketplace during persistent bottlenecks in the supply chains. Filters and other single-use systems were just a few such examples. FDA is a wonderful, collegial agency. We worked well together in an emergency and even lent resources across Centers and Offices to help each other. The pandemic has had some residual effects, but we’re catching up now. We are now getting over it and I hear we are aiming to get back to full speed with inspections.
I also want to compliment the industry because during the pandemic 100 years ago, and in fact even just 20 years ago, we wouldn’t have had an mRNA vaccine and could not have deployed life-saving drugs at such a remarkable speed.
My question is around gloveless isolators specifically and the inspection history there. Have you seen any observations with these closed systems that you would not have seen with a more traditional isolator with gloves on it? Or alternatively, is there anything that you specifically look for when you‘re inspecting a facility or reviewing a product that uses a closed gloveless isolator versus a more traditional system?
It‘s more of the same in terms of the basic cGMP requirements. There might be some shift in our critical control point mindset in terms of what to cover; for example, in terms of robotic arms, we may look at maintenance, precision, mechanical system integrity, automation algorithms, IT software, and so on.
But it‘s still aseptic processing. It just gets more into the realms of automation and digitalization. Having sensors in the right places and understanding how to mitigate system failures. For example, having redundant sensors usually makes sense. I’ve seen sensors fail and that has caused big failures at the biggest companies in the world in isolator technology in their CIP/SIP systems. The drain valve didn’t close totally, and the sensors were not detecting it, so a back siphonage problem ultimately led to contamination of the isolator with spore forming bacteria. And this happened more than once.
This is more of a compliment to the nonbureaucratic approach that the regulators took during the pandemic. I think you did a good job. Thank you. And if the vaccine industry does not win the Nobel Peace Prize, there‘s something wrong with the prize.
On behalf of everyone at the FDA, we appreciate that acknowledgement. Thank you.
It is amazing how fast the industry came up with the mRNA vaccines and other therapeutics. Within a year or so, industry and regulators were able to support the deployment of vaccines and therapeutics that literally saved tens of millions of lives around the world. Kudos to the industry for your tremendous accomplishments during extremely challenging times.
ISPE thanks the panelists for their open discussion of audience questions. We are looking forward to the next regulatory panel at the 2024 ISPE Aseptic Conference in Vienna, Austria, 12-13 March 2024. For more, please visit information ispe.org/conferences