Regulation of Aseptic Processing in the 21st Century

Numerous points of special emphasis in the regulation of sterile products have come and gone over the past three decades, but by and large the trajectory of regulatory change has remained the same. The dominant regulatory philosophy is well known to all stakeholders, and in general involves an ongoing reduction of allowable environmental-control levels with an emphasis on the quantitative aspects of microbiological monitoring (1–4). Along the way, new parallel points of emphasis have been added, including the evaluation of airflow patterns (chiefly by smoke study testing), emphasis on high-efficiency particulate air filter integrity, and increasing attention to nonviable particle excursions. Not surprisingly, given this direction, regulators recently have spoken about expanding the correlation between particulate-air quality, room or zone classification, and microbiological air quality.

The emphasis on reducing microbial contamination has worked, although it is possible to debate whether this progress was the result of regulatory emphasis or of ongoing technological advancement. Much of the improvement likely was inevitable, unrelated to regulatory pressure, and purely technological. The increase of automation dramatically reduces the need for human operators to intervene in the process. Examples include automatic weight checking, sterilizing or cleaning in place after equipment setup, depyrogenation tunnels, automatic parts feeding, and automatic lyophilizer loading. The cumulative effect of these advances has been to reduce the number of human operators required in aseptic-processing areas. Because humans are the major source of contamination, automation directly reduces microbial levels.

It is also clear that improved cleanroom garments and cleanroom undergarments have played significant roles in reducing human-borne contamination of aseptic processing environments. An often overlooked reason for improved environmental control is that modern aseptic fill rooms have far higher air-exchange rates than they did two or three decades ago, a fact that deserves emphasis. Clean-air dilution rates have not received the same amount of regulatory attention as have air direction evaluated by smoke tests, and transient and low-amplitude particulate excursions.

Aligning philosophy with metrology

Regulators' increased emphasis on microbiological monitoring has fostered the implementation of advanced processing technology and better cleanroom designs. The effort to "push down counts" was inarguably effective at focusing industry's attention on improving the science and engineering of aseptic processing. Interestingly though, other industries have unquestionably evolved faster in environmental contamination control without similar levels of regulatory emphasis on monitoring. The aseptic beverage industry has already moved beyond using human operators and gone to highly automated and fully enclosed aseptic processing systems. And, of course, the microelectronics industry operates at levels of environmental control well beyond those employed for the aseptic processing of healthcare products. These industries were motivated largely by commercial opportunity and advantage.

Modern biopharmaceutical manned aseptic processing ISO 5 critical zones generally produce contamination recovery rates of substantially less than 1%. A review of data from these cleanrooms confirms that this low-level contamination does not follow any discernible pattern. Small deviations in rate are apparent, but statistical trends are not. In addition, the long relied-upon alert-and-action-level trending approach now has little scientific merit in ISO 5 or cleaner environments. The limit of detection of the monitoring methods is not zero, and, in fact, that limit is necessarily unclear. Analytically, it is not possible to know that zero means no contamination, although it is reasonable in a manned cleanroom to assume that it doesn't. No meaningful difference exists between a plate with one colony forming unit (CFU) and one with three or four CFU, but current regulatory philosophy forces us to pretend that it does. Those required to investigate a normal and completely routine environmental recovery from a manned environment generally understand the utter futility of their exertions, but those who require them to do this busywork apparently fail to grasp that it is a waste of human resources. The resulting lack of mutual understanding is just the sort of thing that breeds cynicism, and ultimately frustration.

The authors believe that unless human operators are made to wear hermetically sealed suits with filtered inlet and exhaust air, the industry has approached the contamination-control baseline in state-of-the-art manned cleanrooms. The substantial improvements observed before the 1990s are no longer possible. As a result, the industry is operating at the ragged edge of method capability.* Put another way, we are operating at the edge of our limit of detection at all times. People often express a need for better monitoring technology to continue on the current regulatory trajectory, but this is quite simply fool's gold. Ultimately, industry must accept the fact that where human-scale aseptic processing is concerned, it can no longer measure the incremental improvements using monitoring approaches that worked well when process capability was substantially lower.

Random, low-level counts, essentially at background levels, are common. Manned rooms will always and inevitably have a background level of microbial count. Conducting investigations, then, into background-level recoveries is a meaningless activity that is unrelated to the assessment or reduction of risk.

Monitoring in the assessment of risk has another drawback in that, more than ever, it requires a significant increase in interventions into the ISO 5 aseptic-processing critical zone. Greater implementation of automation has reduced risk largely by eliminating interventions and reducing the human population within aseptic fill suites. Increasing the intensity of monitoring will reverse this positive trend by requiring more interventions to perform air and surface sampling. In the research that led to the publication of the Akers-Agalloco risk-analysis method, the authors found that monitoring activities were the largest single type of intervention in many modern cleanrooms (5). Operators thus were creating risk in an effort to measure risk.

Advanced aseptic technology

If the discussion shifts from manned cleanrooms to unmanned environments such as isolators and closed restricted access barriers (RABS), monitoring as a means of evaluating performance becomes even more marginalized. In other words, the industry's monitoring metrology is wholly inadequate for the dramatically improved contamination-control capability.** Studies using even advanced spectrophotometric methods targeting and detecting macromolecules associated with microbial life confirm that isolators are indeed exceptionally clean environments (6). Increasing monitoring intensity is therefore both unwarranted and unproductive. Increasingly, monitoring does nothing but confirm expectations; the methods are no longer sensitive enough to detect status changes. Status changes below the limit of detection in isolators and closed RABS (if they even exist) are, based upon the high level of safety attainable with products made in conventional cleanrooms, medically irrelevant.

Continuous and intensified monitoring

Recent years have seen an emphasis on continuous monitoring of environments, specifically as regards total particulate counting. Advocates of settle plates argue that because they can be exposed for 4–5 h, they are effectively a form of continuous monitoring. This notion, however, is exceedingly simplistic.

As previously noted, high air-exchange rates are the most effective means to dilute contaminants that are inevitably released into aseptic environments. Make no mistake, contamination release is unquestionably inevitable in manned environments. Modern cleanrooms routinely provide 500 or more air changes per hour. Assume that a large cleanroom has a total volume of 500 m³ and achieves a mean air-exchange rate of 500/h. The total volume of airflow passing through the room in an hour would be 500/h × 500/m³, which equals 250,000/m³ of air per hour. If personnel placed 1000 monitoring devices in this room, each with the ability to sample 6 m³/h, they should be able to sample just 6000 m³ of the 250,000 m³ of air passing through this environment. This amounts to 0.24% of the air passing through the room in an hour.

The sampling probes could indeed be placed at critical locations, but they would still sample only a small fraction of the air moving through the critical zone. The cost of electronic particle samplers, which sample and report results in real time, is significant, and the modern microbial-detection systems are quite expensive. If such units cost $5000–10,000 and a company used such a vast number of them, the cost obviously would be prohibitive. In this admittedly extreme example of sampling intensity, the initial purchase cost would be $5–10 million. One could argue that two such units, one near the fill head and one near the closure application zone, would suffice. From a regulatory point of view, they might, but the amount of air actually sampled would be trivial. Such an exercise would be of little practical benefit. Passive sampling is even less likely to be of any real value in risk analysis, although it might help ensure a good inspectional experience.

Particulate excursions

If the above is true, why does industry observe nonviable particulate excursions? The confirmed reason is that processes inevitably generate particulates; most processing machinery is belt- or gear-driven and has moving parts. Process equipment generates nonviable particulates, and humans generate viable and nonviable particulates. Friction occurs on accumulation tables and bottom-drive conveyors between containers and equipment and between adjacent containers. Also, routine activities, such as opening a sterilized bag of closures and pouring them into a feed hopper, will generate transient levels of particulate that exceed the ISO 5 class limit (7).

Classifying cleanrooms under operational conditions is logical, or at least appears so. However, processes do generate particulates, and this effect cannot be eliminated. Powders and ointments are known particle generators, and even ordinary liquids can aspirate particles into the environment. Even in advanced aseptic processing in isolators, process-related excursions occur. In fact, if a probe is located too close to a point along the process line where particulate is generated, counts higher than the class level may be observed more or less continuously. This reality of aseptic processing is rarely discussed, which is unfortunate, and precisely why the authors chose to discuss it in this article. Realistically, the only way to ensure that any process with humans and machines is always within the ISO 5 class limit whenever sampled is to turn off the equipment and have the personnel limit their movement to levels significantly less than those required to perform their work. In other words, the conditions would no longer be dynamic.

Particulates are a functional reality in sterile drug manufacturing, whether the product is aseptically produced or terminally sterilized. This reality is why standards call for "essential" freedom from visible particles and why the compendial limits for total particulate are considerably higher than microbial expectations.

It follows, then, that requiring a firm to investigate any and all excursions is essentially pointless, because it would be investigating completely routine and normal conditions. A more appropriate expectation is that firms maintain their operations within a validated state of control and that in the interest of continuous process improvement they evaluate means of reducing particulate levels.

In the authors' experience in aseptic processing, periodic excursions do not routinely manifest themselves in high product-rejection rates or customer complaints. Periodic excursions above ISO 5 in the critical zone as a result of normal process operations do not increase risk to the patient. In fact, in the authors' experience, airborne particulate is less likely to correlate to visible particulate than is container cleaning or closure cleaning and feeding.

Performance expectations have to be realistic. Often excursions can be correlated to a routine activity such as restocking a parts hopper. It is not reasonable to expect a complete absence of airborne contamination in any manned ISO 5 environment, nor is it reasonable to consider that a minor, brief excursion violates classification requirements. It is reasonable to expect firms to design their manned facilities and closed RABS to handle personnel loads and production capacities, which means in general that these facilities must provide sufficient air-exchange rates. Isolators, given their relatively low enclosed volume and absence of personnel, operate perfectly satisfactorily at lower air-exchange rates. Pragmatic engineering and realistic appraisals of performance should carry the day.

Smoke studies and airborne contamination

The last half dozen or so years have been marked by a dramatic increase in the emphasis on studies intended to visualize airflow. The origin of this movement is unclear, but perhaps people thought that smoke tests could somehow resolve things that monitoring no longer could. Over the past few years, regulatory comments have asserted that smoke studies indicated a "lack of sterility assurance." It is puzzling how someone could observe or review a recorded smoke-study test and draw from it far-reaching conclusions regarding sterility assurance. The lack of objective criteria in smoke studies suggests that definitive conclusions of that nature are highly suspect.

The air-movement story in most cleanrooms is the same. The vast majority of aseptic processing areas are designed with air entry above the work zone and with air returns mounted along the walls close to the floor. Thus, the air moves downward, generally at an initial rate > 0.45m/s.^† Generally at 1100–1200 mm above the floor, air in the critical zone comes into contact with a solid horizontal surface, encounters resistance, and, as a result, eddy currents develop. The air travels vertically downward in a unidirectional, but certainly not laminar flow, until it encounters the irregularly shaped, solid structure of the processing equipment oriented perpendicularly to the direction of airflow. Further disturbances to airflow arise from the movement of materials along the conveyor and the operation of such things as stopper-positioning arms, vibratory bowls, various feed mechanisms, and, of course, the filling pumps and related plumbing (8). Airflow in cleanrooms, RABS or isolator systems is still occasionally referred to as laminar flow, although in reality the airflow has significant turbulence and can best be described as "generally" or "mostly" unidirectional. References to "sweeping" movement of air convey the comforting notion of a cleansing effect, but realistically the sweep includes the movement of air over and across open containers and closures.

Airflow visualization or smoke studies can be informative, but only in a general sense. We do not believe that they enable an observer to draw far-reaching conclusions regarding sterility assurance. The studies may have value in assisting a firm in evaluating and tuning air movement in a general sense, but they don't have a major effect on environmental-monitoring results or on process-simulation studies. This fact is not surprising because smoke tests, environmental monitoring, and media-fill tests are not sensitive enough to discern minor differences in airflow, and as previously discussed, air-exchanges rate are more directly related to dilution and removal of contaminants.

The authors are not suggesting that firms curtail smoke studies, but they do caution against overinterpreting the results of such tests. Turbulence is unavoidable, and it is not reasonable to think that airflow can be evaluated precisely using visualization methods. Also, the visualization technique ideally should involve smoke generated in an isokinetic and isothermal manner. Obviously, when the smoke ordinarily used is colder or heavier than ambient air, it will move downward on its own, thus distorting the results. Regardless of how these tests are conducted, the authors see no way that these tests can provide an objective means to evaluate sterility assurance. Rather, they believe that where smoke studies are concerned, beauty is in the eye of the beholder.

Conclusion

The authors believe that the methods that industry has used to evaluate aseptic-processing environments, while of some value, cannot in and of themselves assess sterility. All manned environments are nonsterile, and the ability to prove objectively the sterility of aseptically filled products will remain elusive. To prove sterility in an environment would require a sample of infinite size to be analyzed with a method that has a limit of detection of zero. Such analytical methodology does not exist, and this article has shown that even with extreme sampling intensity, only a small fraction of air passing through an aseptic production environment would be sampled. The authors are concerned that the benefits of continuous total-particulate and microbiological monitoring are not technically significant. Environmental monitoring is a limited means to assess aseptic environments, and the industry has long since passed the point of diminishing returns.

Caution is also in order in the interpretation of nonviable particulate (i.e., total particulate) monitoring. The strict enforcement of class limits in every location within an ISO 5 area under dynamic conditions is unreasonable. Process-generated particulate is, at this point, inevitable, as are brief excursions at locations within the environment. The value of nonviable particulate monitoring is in ensuring that excursions are similar from day to day, in terms of frequency and amplitude, with reasonable allowances for inherent variability.

Smoke studies are another well-meaning initiative, but their value has been exaggerated. No objective parameters exist for smoke-test success, which means that acceptability is a subjective judgment call. No evidence demonstrates that smoke-test visualization can determine sterility assurance.

This article raises questions regarding elements that could be considered cornerstones of current regulatory-compliance strategy in aseptic processing, and thus challenges the status quo. The authors' justification is their belief that the current direction of aseptic processing process control and evaluation needs to be changed. Simply doing more of what the industry has been doing for years is not a good way forward.

Advanced aseptic technologies, including the interventionless systems already in use in some industrial aseptic applications, will require new approaches. Many physical parameters that can be measured may provide better assurance of quality in real time than conventional monitoring does. These parameters include air overpressure, humidity, temperature, air velocity across product entry or exit points, and nonviable particulate monitoring.

Given the cost of energy, the pharmaceutical industry will need to revisit air-supply systems, as other aseptic and clean industries have done already. Green aseptic processing is now possible, and high levels of automation and gloveless isolators promise to reduce costs and enable companies to manufacture products at competitive costs of goods produced in North America and Europe. Such systems are here today, and their use should be encouraged and incentivized by regulatory authorities.

Pharmaceutical aseptic processing sadly lags behind the technological level attained in other industries. This situation should not persist. Companies must embrace new technologies and appropriate process-control systems. To that end, meaningful dialogue between regulatory authorities and industry technology experts is required. The focus of these discussions must be on future technologies and control, rather than preservation of the status quo. The result should be safe, and low-cost aseptically produced medicines.

*The regulatory pressures and advent of validation in the 1970s fostered substantial improvement in many manned cleanrooms that plateaued some 20 years later. **Open restricted-access barrier systems where the enclosure must be opened during either setup or operation perform similarly to manned cleanrooms. †Other velocities would perhaps be even more effective, but regulators have fixed on this value. Its removal from Federal Standard 209C in 1987 suggests that it has little scientific validity, but regulators continue to enforce it.

James E. Akers* is the president of Akers Kennedy and Associates, PO Box 22562, Kansas City, MO 64113, akainckc@aol.com, and James P. Agalloco is president of Agalloco and Associates and a member of Pharmaceutical Technology's editorial advisory board.

*To whom all correspondence should be addressed.

References

1. FDA, Sterile Drug Products Produced by Aseptic Processing (Rockville, MD, Jun. 1987).

2. FDA, Sterile Drug Products Produced by Aseptic Processing—Current Good Manufacturing Practices (Rockville, MD, Sep. 2004).

3. EMA, Annex 1, Sterile Medicinal Products (London, 2009).

4. Task Force on Sterile Drug Products Produced by Aseptic Processing, (with support from the Japanese Ministry of Health, Labor, and Welfare), Guidance for Industry—Sterile Drug Products Produced by Aseptic Processing, (Tokyo, 2005).

5. J. Agalloco and J. Akers, Pharm. Technol. 30 (7), 60–76 (2006).

6. M. Miller and M. Walsh, PDA Annual Meeting (Las Vegas, 2009).

7. International Standards Organization, Standards 14644-1 to 8: Clean Rooms and Associated Controlled Environments (Geneva, 1999).

8. B. Reinmuller, Building Services Engineering, Bulletin 56, (Royal Institute of Technology, Stockholm, 2001).