This year marks a major landmark for FabOhio, Inc. It marks the beginning of our 58th year providing containment to the Nuclear and Heavy Processing Industries. It also marks the beginning of our 28th year providing containment solutions and advice to the Pharmaceutical Industry.
To celebrate the 28th Anniversary of involvement with the Pharmaceutical professionals we would like to present the gift of major findings and events that have been created by your industry since the first inception of flexible containment. Many of these were of landmark status since much of the discovery process led to changes in development, work practices and procedures, and strategies developed for confirmation of performance criteria, adopted within the industry for both in-house and regulatory compliance.
Some of the topics have become folklore, and as such, modifications have been introduced over time which have led to doubts, or even outright failures of containment technology solutions that began 25 years ago. We hope that these disclosures which were developed by committed practitioners, including major contributions by operators, within the industry, and often working with FabOhio Inc. experts and products, will add value to your work while clarifying many misconceptions.
We are held under client confidentiality for actual company and site names, and sadly are unable to give credit to individuals involved with the amazing evolution of these findings, our agreements do allow us to share the information for the benefit of the industry as a whole.
While the information is a collection of mostly summaries with a few detailed stories, we may be able to share more information about topics you have interest in. Recognizing that our clients may preclude sharing, we may be able to afford you limited access to some of the practitioners, although many are now unavailable for discussion due to career advancement or retirement.
FabOhio, Inc. thanks all of the participants for the many hours and dedication committed to challenging the status quo in a primarily steel and glass industry. We are proud to be able to share their stories for the benefit of the industry, especially since little of their exploits is available in the public domain.
Consequences of Introducing Flexible and Rigid Containment into the Workspace
One would think that adding a confined containment solution to the Pharmaceuticals workspace would be a minor change. Today it is because of all of the benefits that it brings. But it wasn’t always so!!! Compare that with the Nuclear Industry with over 70 years of experience where both flexible and rigid containment evolved. The Industry would never consider using airflow regulated containment because you cannot control the emission pathway. FabOhio Inc. grew from 1963 to the present, servicing both the Nuclear and Heavy Industry needs for control of materials releases based on our advanced knowledge of the capabilities of flexible film materials, e.g. Polyethylene, Polyurethane, and Polyvinyl Chloride.
In the Pharmaceutical Industry, exhaust hoods and Biological Safety Cabinets were considered to be the standard approach to containment. The significant difference between airflow control and confined spaces is dispersion versus consolidation, i.e. consolidation within an enclosed space is the major factor in achieving migration control for materials of high potency. Changing conventional wisdom from materials releases transported in an airflow, to one of confining materials in a small enclosed space will prevent contamination of adjacent surfaces, air-ducts, and airborne migration throughout a facility, has been a slow laborious process. While the advantages seem obvious, the culture has been slow to recognize them.
In 1993, when the first client proposed testing a flexible glovebag as a solution to overcome the chemophobia expressed by dispensing operators, an entire series of demands were created which have forever changed practices in the Pharmaceutical Industry. Many of the demands evolved into better practices which have since been adopted, often without knowledge or understanding of either their source or correct application. The sad consequence is that without knowledge of the events leading to better solutions, far too many mistakes have led to erroneous applications and assumptions which discredit the technology without cause. Much greater use of both Flexible and Rigid Containment solutions would benefit the entire industry if the misconceptions could be righted.
These synopses address many areas that while seemingly unrelated to containment have come to the fore and been addressed as peripheral needs affecting Quality, Worker Health and Safety, and Environmental needs and demands, while also improving Operations throughout drug search, testing, manufacture, and disposal. In a way this is a summary of the first 25 year of most of the detailed studies undertaken to prove the viability of the technology while highlighting erroneous ‘common knowledge’.
Before even considering allowing the use of Flexible Containment in their traditional steel and glass facilities, Management insisted on proof of performance. Since the initial concept was Worker Protection, it seemed appropriate to perform personnel monitoring - established procedures recognized by OSHA and the Occupational Health Profession in the US, to conform with the Action Level (1/2 the Occupational Exposure Level) compliance guidelines. Sadly this approach only measures an average outcome due to whatever exposure sources are present including, but not exclusive to, the Flexible or Rigid Containment. Furthermore, there were too many unknowns about the potential source materials, i.e. the Actual Pharmaceutical Ingredients (API’s), for any sampling method to be representative of a generic containment performance. Performance testing using an API under development was deemed unacceptable based on risk, availability, and materials cost. At the time, the analytical detection limit achieved by the dedicated Occupational Health Support Laboratory was 20% of the API exposure guideline (later to become the Occupational Exposure Level - OEL) of 0.2µg (equivalent to 0.04µg/m3) with a quantitation limit of 40% (0.08µg/m3), which limited proof of compliance performance - typical of so many drug substances.
The first challenge was selection of a surrogate material having sub-microgram analytical quantification. The material needed to be dusty. Secondly, the actual dustiness value was needed as a precursor to placing reliance on any measured values. Thirdly, a test location free from the surrogate material was required. The surrogate selected had an analytical sensitivity of 0.05µg/m3, which was inadequate for the task when dealing with an OEL of 0.2µg/m3, but the redeeming factor was the extreme dustiness of the material which far exceeded any other acceptable material (>10 times).
Prof Dane Kildsig, Professor of Pharmacy and Pharmaceutical Engineering, Purdue University, conducted research to select the most appropriate of the three available methods for measuring ‘Dustiness Index’ (DI) as it applies to processing in the Pharmaceutical Industry (Containment in the Pharmaceutical Industry, Ed. James P Wood, Publ. Marcel Dekker). The Heubach Method was selected as the most relevant. The surrogate DI was 11% when unmilled, but 14% when milled. Alternate materials of ‘high’ dustiness were corn starch (7% DI) - a complex mixture of molecules leading to analytical limitations, and milled lactose (~3µm median diameter and 1% DI). Lactose is a common excipient in the industry, used in formulation and finishing operations, consequently the high native background levels exclude it from consideration for Research and Development (R&D) and Production finishing, but it is ideal for bulk R&D and Production.
The sampling strategy relied upon the model developed by EPA for assessing clearance after an asbestos abatement (Asbestos Hazard Emergency Response Act (AHERA) protocol). This was the model selected after the EPA review of the myriad of radiation models developed over 50 years by Health Physicists monitoring radiation distribution in nuclear facilities. The model can deliver a statistically sound database as a result of statistical randomization of sample locations within a spatial distribution, e.g. in a room footprint or around a piece of equipment. With three repetitive studies, a 95% (2 standard deviations) Confidence Level can be defined, even with disparate data. When samples placed at potential fugitive release points mimic the random array, the data can be compiled together to achieve a higher order of statistical compliance. The initial database can be used repeatedly to monitor containment performance throughout the lifecycle of installation, use, and disposal of Flexible Containment as well as hard goods (Rigid Containment).
The ~15% overall failure rate of personal sampling pumps is eliminated by using a vacuum source set to deliver -19 inches of mercury, which provides airflow to all fixed sample locations via a manifold fitted with critical orifices rated for 2 l/min airflow. A vacuum pump rated for 3.5ACFM will deliver adequate airflow for up to 14 sample locations.
Vertically oriented open-face filter cassettes with 0.8 µm pore size mixed cellulose fiber filters are used (0.45 µm filters produce a higher flow restriction, requiring a higher vacuum setting resulting in reduced airflow). Vertically oriented samples accept depositing and airborne particles eliminating particle-cut size bias. Isokinetic sampling (matched airflow to inlet velocity) is unfeasible under such test conditions due to the multiple airflow vectors.
The final key to achieving realistic containment performance data was to use operators to perform the activities during test cycles. The first flexible enclosure to be tested was a dispensing glovebag designed for drumto-drum transfers.
First Flexible Enclosure to be Tested
An existing lipped dispensing table was fitted with a glovebag designed to fit over the raised edges and secured with adhesive tape. The table was set up in an unused chill room having no air exchange, i.e. a large confined space, making this a worst case scenario with no airflow dilution. Once the door was closed, sampling was started and two operators performed dispensing operations - attaching the glovebag to the table, attaching the bulk material and empty drums, dispensing, and detaching both drums before repeating the operation for a total of 3 dispensations over a two hour period. Sampling continued for 2 additional hours to allow collection of airborne materials before the door was reopened and and the room vacated. Without further studies it was obvious that the glovebag installation was inadequate due to failure of the tape seal around the corners of the raised table edge where material bunching was unavoidable.
FabOhio, Inc. advised the use of a glovebag with an integral floor with the bagging sleeves attached. Three discrete studies using different operators proved the viability of this approach when no detectable surrogate was found on any of the 33 samples (5 statistically random samples around the enclosure, 3 sleeve, 2 personal, and 1 room for each study). Paired laboratory and field spike and blank samples confirmed sample recovery and analysis to be quantitative.
With this confirmation, the studies were extended to use of a glovebag for materials handling in a pilot plant charging room. In this case a potent intermediate was being handled. Ventilation was fully operational providing a normal operating level of airflow dilution. After 3 studies, the 5 samples around the glovebag, including the 2 sleeves in each study, showed no detectable material (<0.02µg, equivalent to <0.04µg/m3). Several operator and room samples showed material levels up to 0.2µg/m3 OEL. The sample blanks and spikes were recovered quantitatively (~98%). Follow-up involved a detailed review of the impact of room cleaning procedures. Room contamination can lead to multiple forms of contamination including the tools and operators clothing, hair, and other body parts.
A solvent transfer room on a lower level had experienced several release incidents involving the intermediate material, followed by laborious cleaning before further processing was permitted. Sampling was normally performed using up to 4 samples mounted on tripod stands following clean-up after the room was dry. The entire cleaning cycle took 3 days (1 day for cleaning and room drying, 1 day for clearance monitoring, and 1 day to run and deliver laboratory analyses providing priority support.
Repeated cleaning cycles were always required before achieving clearance. Processing time loss was always >2 weeks and frequently over a month with operators required to wear protective air supplied coveralls at all times.
The cleaning process involved use of a ‘water saver’ nozzle connected to a 200 gallon clean water tank to irrigate all the floor, walls, and equipment in the room. The process took ~ 2 hours with the operators in full air supplied Personal Protective Equipment (PPE). The intermediate was known to be water insoluble.
Airborne Dust Testing and Cleaning Remediation Studies
A more robust sampling protocol was needed before initiating further studies. The method was modified to conform with AHERA to generate both a statistically robust data set for decision making, as well as a robust database. The protocol established 5 randomly placed samples on tripods (5 ft high) within the room. Once samples were started, all room surfaces were swept within 3 inches using an air nozzle at 80 psig (simulating the leaf blower cycle under AHERA), i.e. aggressive air monitoring.
Results from the modified sampling method consistently showed contamination on room surfaces even after room cleaning. There was sufficient contamination that operators entering any room, hallway, or stairway were likely to become contaminated during their work shift. This explains the apparent failure of the glovebag based on personal sampling alone, i.e. a data artifact anomaly using the SMEPAC/ISPE guidance. Based on the statistically randomized sampling protocol results the glovebag was not giving rise to fugitive emissions.
An alternative test was reviewed by the laboratory. Reaction Vessel cleanliness is partly based on swab tests of internal vessel surfaces. A study set out to qualify swab testing as a means of establishing room cleanliness. The laboratory prepared a study using 6 different surfaces common to the processing equipment and room surfaces - 2 different grit finish stainless steel surfaces, Polyethylene, Teflon®, a glass vessel man- way cover, and insulation cladding (aluminum). Spikes of a water-soluble API used for instrument calibration (2µg, i.e. 10 times higher than the detection limit) were inoculated onto six different areas of each of the surfaces and repeated for 4 recovery studies by 2 laboratory personnel and 2 field practitioners. The swabbing procedure used for vessel clearance was used. Of the 144 samples, only one was at the detection level, with 143 being ‘non-detect’. Swab testing was abandoned as inappropriate for external surfaces and finishes.
Commercial steel coupons were purchased having different surface finishes. After degreasing and cleaning they were inoculated as before. Recovery was measured by immersing the spiked coupons in a measured volume of distilled water and sonicating for 5 minutes. The wash water was collected. The irrigation wash was repeated for a total of 3 washes. In all cases the first wash level was higher than subsequent washes, and only the electropolish (ep) finish showed quantitative removal of contaminant:
ep 400 grit > 2B mill finish > 25 grit finish > 80 grit finish > 125 grit finish > 400 grit finish
(>95%) (~65%) (~55%) (~15%) (~9%) (<5%)
These finding are contrary to conventional wisdom but easily explained when considering the physical properties of the cleaning agent - distilled water with a surface tension of 70dynes/cm^2.
The findings were substantiated when room cleaning was changed to what is now known as fog/mist cleaning. The method relies on using water having a high surface tension (approaching 70 dynes/cm^2). The higher the surface tension, the smaller the fog-droplet size allowing the droplet to enter the grit grooves. Addition of surfactants destroys the cleaning efficiency by lowering the surface tension.
Adoption of this cleaning procedure allowed for effective cleaning after a single cleaning cycle, vastly improving processing time while reducing the frequency of processing failures.
Residual Contaminant Testing
Riboflavin is commonly used for tracking decontamination efficacy. It has been used fto test cleaning of glovebox surfaces. Unfortunately the observation efficiency under fluorescent light is ~1% in water, depending on the surface and visual acuity.
A better detection level is desired for materials having OEL’s in the sub-µg range. Acid Yellow has been used for observing surface damage, e.g. study of bearing surface damage. When in solution in water it appears yellow with a green overtone, and green under UV light, much like the common antifreeze ethylene glycol. At low concentrations the primary yellow color disappears leaving the green fluorescent component observable to ~40 ppb in water, 25,000 lower level than Riboflavin.
Like most dye materials, Acid Yellow milling is not possible due to melting. When diluted ~10:1 with other materials, which act as heat buffers, it can be milled successfully using a chilled shear mill, e.g. a Wright mill ($4,000) to particle sizes of ~3µm median particle diameter. The milled blend can then be dispersed using a dust generator. This allows for quantitative worst case contamination of room surfaces. Quantitative contamination (the challenge) combined with post-cleaning aggressive air monitoring allows for evaluation of both the cleaning procedure and operator performance.
In a study comparing deluge cleaning (2 - 3 hours) with fog/mist (>4 hours) for a large epoxy coated charging room surface, both deluge and fog/mist cleaning showed equivalency for a water soluble material (lactose + Acid Yellow), while previous pilot plant experience showed that fog/mist was far superior for a water insoluble material. The reason to use fog/mist for soluble materials is the difference in water waste volume. In a large room the difference is >400 gal for deluge cleaning and ~20 gal for fog/mist. Waste storage and destruction is a large budget item in potent compound manufacture, with dedicated piping to, and within, the waste arbor and the waste holding tanks.
Reality Based Room Sampling
How does one evaluate a work environment as it contributes to worker health and cross contamination potential? With containment measures deployed and used on a daily basis, how do you evaluate the impact on the facility environment over time? What is the environment exposure challenge on a 24/7 based workspace?
Although the airflow in a room is usually engineered for ~100 linear ft/min, many factors disrupt the local velocity around and between equipment, making isokinetic sampling unfeasible. Any obstructions, e.g. worker movement, equipment placement, room geometry, have major impact on the local airflow velocity. Using smoke and bubble generators in tandem with air velocity measurements quickly reveals velocities ranging between stagnant air and >300 linear ft/min within an apparently inactive room.
The only constant is the location in which a sample is placed. Using the AHERA principle, repeat samples on the same statistically random array within a room will create a set of data which can be checked for homogeneity and repetitious performance over time. Knowing that each sample location is unique prevents absolute spatial comparisons, however the value of the array is that each location can be compared with itself as new data sets are generated. In practice, a containment controlled room will result in a repeat pattern of performance unless an upset is observed, at which time it becomes necessary to take remedial action. If a disruption is observed and only one or more sample values increase, the release is localized. When all of the values increase, especially near any personnel or materials access doors, adjacent locker rooms, and hallways may well show evidence of a breach which requires immediate preventive measures. The confidence in such progressive data archives can quickly exceed 99% (3 standard deviations) over time.
To generate such a data set, sampling must be continuous on a 24/7 basis. Pumped air samples and laboratory support becomes a herculean task. Resorting to tools used to monitor clean rooms, settling plates become the obvious choice. When set out for sufficient time periods at the same locations they will sample continuously for 24 hours/ day until collected for analysis. Being a passive sampling device they will collect settling material at whatever rate the airflow delivers to them at that location.
The challenge is to devise an appropriate sampling period. This will depend on the analytical sensitivity. It was found that sampling for 30 days for a materials having an OEL of 0.2µg/m^3 was an appropriate period. While not ideal, it is more reliable than random period air sampling for up to 8 hours for the purpose of measuring occupational exposures over the same time frame.
In a facility having 17 discrete containment zones, 18 sets (17 zones plus a shared air supply set in hallways) of 5 sample locations + 1 control spike + 1 blank for each set will generate 126 samples for analysis/month. In addition, the laboratory will retain paired spike and blank samples (36 additional assays) for comparison purposes. At ~$100 cost to prepare and analyze the samples this amounts to a facility cost of ~$16,200/month. This is considerably less that the cost for room decontamination, lost production time, and reduced product yield. With such a program, operator moral improves immeasurable but is definitely observable.
Having a well regulated facility afforded an environment for additional monitoring study development. With four 5-channel particle counting instruments available, the packaging room was selected for simultaneous monitoring at existing sample locations during a packaging operation. Within the first hour of the 3 hour packaging operation, a particle sampler located at the air exhaust intake showed a spontaneous rise in counts in the 0.2 - 5µm particle ranges. The process was terminated and operators withdrew from the room. The particle counts decayed steadily over a 20 minute period to the original baseline levels. The room was patrolled with a particle counter with no evidence of continued release. Further processing occurred without incident. A simultaneous air sample taken at the top of a screw filter feed showed an equivalent 4-hour average exposure of 0.07 µg/m3 during the 4 hour sampling period. The transfer point was enclosed in a flexible enclosure and no further release incidents were experienced.
Particle counting had obvious merit for monitoring particles in a high potency manufacturing environment. To test this premise, a 16-channel particle counter was commissioned along with a multi-point sampling system. The system was installed in a new pilot plant facility in Ireland. The particle size ranges were focused on the particle range below 10 µm with the hope that the instrument would allow discrimination of particle sizes indicative of the physical operation at which a particle was released. After 9 months of operation the system did appear to fulfill that requirement during fire-testing of the pilot plant process equipment. During late September, with the entire pilot plant shut down, an unexpected event occurred. At ~2 pm, the smallest particle range (0.25µm) count started to ramp up at all 16 sample locations, followed by the next higher range (0.35µm), and each successive range (0.45µm, 0.55µm, 0.7µm, 1.0 µm, etc. up to 8µm) increased progressively until by ~5 pm the instrument counts saturated. With no processing activity in the building there was no logical explanation for the occurrence. The answer was evident when exiting the building. The first sign of fog formation was creeping over the site! The particle counter had been responding to the initial phase of fog formation revealing initiation of water micro-droplet formation over the 3 hour period of observation long before it became visible.
While particle counting had limited value in a facility without air-conditioning, further random testing in appropriately air-conditioned facilities verified the discrimination capability of particle counting, i.e. a particle size spectrum. As a tool for performing facility screening for equipment leak points, and clearance after a clean-up operation, particle counters have merit when dealing with solid materials and droplets in the subµg OEL range.
The value of including sample spikes along with blanks far exceeds the cost and inconvenience. Without proof that accompanying sample sets are unaffected by shipping, storage, and handling from the laboratory to the sampling location and back, it is hard to establish the credence of any data.
Multiple studies have shown that the addition of spiked samples allows the laboratory to track sampling failures throughout the life cycle of a sample and to validate the quantification of an accompanying sample set. Without confirmation of a recoverable spike, quantitative or non-detectable laboratory analytical values for airborne samples are indefensible.
During the first three months of deployment of settling plates mentioned earlier (Reality Based Room Sampling), the API spikes for the 18 settling plates (37 cm diameter Petri Dishes) were inoculated at 0.2µg, a value which is 10 times the detection limit for the assay. Recoveries were within the range 95 -103% providing a Confidence Coefficient of 95% (>2 standard deviations (𝛔)). The confidence interval of the recovery was constant, even when the finite recovery decreased over over a 6 year period of monitoring, providing a confidence value of >99% (3𝛔).
The spike was later overwhelmed during a facility upset, to be discussed later.
Over a 6 year period, the spike recovery of the API was observed to steadily decrease on a monthly basis from >95% to ~55%. When the laboratory prepared new standards they confirmed that their original calibration standard had degraded at the rate observed from the field spikes. Having knowledge of the rate of deterioration enabled correction of the few finite sample values with confidence. While the Development Laboratory had been unable to biologically destroy the API, the slow degradation was welcome from the environmental perspective. The ability to calibrate the rate of degradation allowed the laboratory to retroactively audit prior sample reports for studies worldwide, taken intermittently, which would have remained uncorrected without the existence of a reliable and progressive reference base.
The initial three month of deployment of the intermediate material spikes for the 18 settling plates (also 37cm diameter Petri Dishes) were also inoculated at 0.2µg, a value which is 10 times the detection limit for the assay. Repeated spike recoveries showed an anomaly within the range 5 -14%. This significant loss of material could have been due to either instability, interaction with other materials, or volatility resulting in vapor loss. Subsequent spikes were inoculated at 2.0µg and repeat recoveries were within the range 85 - 92% for over 6 years (>99% confidence level) suggesting that the volatility of the intermediate was sufficient to lose ~0.2µg over a 30 day period. Management changed the spiking level back to 0.2 µg at the 6 year point and lost the value of the study validity during the subsequent facility upset.
Vapor pressure of materials is rarely considered when setting workplace OEL’s, and for the most part this does not present a problem. However, as the values decrease to the sub-µm/m3 range it can become a significant factor.
In the case of the intermediate, after the settling plate anomaly, retracing many months of spike recoveries for airborne samples, a pattern developed of reduced spike recoveries to 85 - 88% during a 4 hour sampling period. While this was noted at the time of data evaluation it did not initially raise concern since it was well within the +/-25% OSHA/NIOSH validation criteria for occupational samples. With the realization that vapor loss was occurring, the impact of the air sampling data was reevaluated, but no cause for change was found. The implications for change in the OEL were minimal at this level, but it did raise a red flag for future studies. It also explains why particulate sampling methods for high potency materials cab be difficult to validate.
During a toxicology review, the proposed OEL for another intermediate was due to be set at a sub-µg/m3 level. A review of the vapor pressure showed that this level was a factor of >30,000 lower than the vapor pressure of the material. This meant that the chemists had to re-evaluate their process to eliminate the intermediate from the reaction since no existing PPE has a sufficiently high protection factor to allow for shipping, storage, or handling the material, including dispensing and charging.
Once an OEL is established, performance parameters can be established for all methods used for protecting the workers. Cost consideration often become the driving force when deciding which method of combination of containment and PPE is to be used.
Air-supplied full coverage protective suits are in widespread use for potent compound handling across the Pharmaceutical Industry. However the associated use costs far exceed those of flexile containment which can eliminate the need for PPE in many instances:
For a both 8 and 12 hour work shifts having 3 break periods, an operator will don and discard 4 or 5 suits/day for a 4 day work week, a total of 16 - 20 suits /week/worker. At >$400/suit + FOB charges + ordering + warehousing + distribution + wash down + waste drumming & 90 day storage + shipping + incineration, for an annual cost of >$40,000/operator/year. A significant budget drain for any pilot plant or bulk production plant on a yearly basis when the number of shifts, operators, and visitors (chemist, engineer, quality, safety, etc.) is considered.
Using flexible containment as the alternative, single enclosures cost within a range of ~$600 - ~$1,800 with durability adequate for short duration campaigns (up to 6 months). Tear-down and disposal costs amount to 1 -2 hours of operator labor, drumming, 90 day storage, shipping, and incineration.
Similar arguments can be applied to laboratory, development, and all materials handling situations, including fill/finish. The same arguments apply for rigid containment solutions. Delivery time for flexible and rigid enclosures often becomes a driving factor along with campaign plans once the decision to use a confined enclosure is made. Flexible confinement enclosures are frequently used as prototypes for capital purchases of rigid confinement hardware for long terms manufacturing campaigns.
The single most important factor in successful deployment of flexible or rigid enclosures is the attitude of the operators. Implementation and continued use depends on how well the operators have been educated on the many reasons for their use. A lone operator who has buy-in issues has been known to cause contamination of an entire facility.
Operators are quick to discover an operator who intends to continue working in the old style. Their responses range from avoiding working with them, requesting management to redeploy them to the menial tasks, rather than entering processing areas. Alternatively, redeployment to other facilities Operators who work diligently to minimize the need for PPE strongly resist any actions that will force them into the PPE mode.
An R&D facility containing a laboratory having an adjacent walk-in hood and a bench-hood with a pass-through was selected to synthesize a potent compound. Performance testing had shown both hoods to be below capability standards. Working with the development team, we designed an enclosure to fit inside the hood spaces which was required within 4 weeks. The intended procedure was: 1) synthesis in the walk-in hood, 2) transfer the filtered product through the pass-through to the bench-hood for drying preparation, 3) passing the dryer trays outside the hood to a free standing vacuum dryer, 4) returning the dried product to the bench-hood for dispensing and packaging, 5) overpacking the product packages, and 6) finally passing through a double airlock into the main laboratory space. Used equipment was to be stored in a doghouse built into the enclosure.
The walk-in hood enclosure section was a floor to ceiling design with the enclosure venting through HEPA filters into the rear of the hood. The enclosure was connected at the pass-through to the bench-hood enclosure which further connected to an external enclosure sealed over the door of the vacuum dryer. The walk-in hood enclosure was accessed through a soft-sided double airlock.
After one successful lot, and with the enclosure still in place, 2 additional lots were requested immediately. With no lead time, two modifications were made to the already contaminated enclosure: 1) an additional glove-sleeve, and 2) an access sleeve. A bench-type glovebag was volunteered by other researchers and the needed parts were attached to the enclosure to enable an immediate resumption of processing.
Under normal circumstances, in-use modification of glovebags is not recommended due to a potential for loss of containment. It is also not cost effective. In this case, our representative was available and accomplished the modification safely. The laboratory did not have to wait for a new enclosure to be shipped, plus decontamination of the existing enclosure, equipment, and hood surround was avoided (saving at least a week).
The afore-mentioned 17 containment zone facility had a successful 6 years of controlled performance. The settling plate strategy revealed one zone in which release events were frequent and remedial procedures were introduced for resolving event occurrences.
Management Transition: The Safety and Health support team leader was changed after 6+ years. Operator turnover after the 6+ years was significant at ~85% resulting in progressive loss of earlier knowledge and experiences of how to prevent materials migration. The new management decided that the settling plate strategy was of no Occupational Health value. The monthly settling plate reports were ignored. During the first month of the new strategy, the settling plate in the less stable zone showed a higher than normal contaminant hit which was not responded to. By the second month, not only the hit zone, but also the outer hallway samples showed accumulation of contaminant. Within 3 months, every one of the 17 zones and all hallways showed contamination. When the annual personnel sampling program was instigated 5 months after the first hit, worker exposures were at an all time high. Airborne area samples in all containment zones, the hallways, and the shared spaces including control room, conference room, and offices confirmed the settling plate data trend. New PPE requirements were instituted with all operators entering a processing zone wearing air supplied full protective suits and following stringent decontamination procedures.
While settling plates are not a direct measure of operator exposure, they will indicate the overall health of a facility in which operators are working, including when excursions occur.
The ensuing cost of lost production, decontamination, equipment replacement, e.g. computers, office equipment, etc., quality clearance, and management acceptance was exorbitant. The ongoing costs of the settling plate program were far lower than the loss of facility and confidence of Operators, Quality, and Management.
Widespread Need for Worker Protection and Containment Concerns
Potent compounds and allergenic materials retain their potency even after leaving a processing facility for storage, distribution, and utilities providers, e.g. solvents arbors, or as waste streams.
Credence is rarely given to concerns protecting operators in facilities outside of the actual production zone or building. This has led to cross contamination in shared maintenance shops, warehousing, vehicles, and other sites, as well as on-site shared services such as waste handling. Maintenance is a high risk function for processing facilities because they are often involved in line breakage, contaminated equipment transfer, and open shell cleaning and parts replacement, removal and calibration of sensors when in service, etc.
Any efforts taken in the the production facility to contain releases in such areas yield significant pay back when one considers the cost of decontaminating warehouses, maintenance shops, etc. Consolidating and minimizing the facility waste streams e.g. fog/mist room cleaning instead of deluge wash-down, can lead to optimization of waste flow to the treatment plant by minimizing piping contact and a reduction of dedicated storage capacity.
Such dedicated piping areas can be provided with localized flexible containment enclosures around sample points, mechanical equipment such as mills, flanges where frequent leak patterns are observed, and all open pipe maintenance operations. Quality is not involved in the selection of flexible materials for such utility areas and as such the quoted storage life for production purpose e.g. 3 years based on stability tests, are not relevant. Polyurethane has been shown to be stable for outdoor use through 20 years of seasonal change in Europe in utility applications, retaining its flexibility and remaining leak tight, while showing minor yellow coloration.
Large Scale Applications & Third World Acceptance
FabOhio, Inc. was called in to assist an antibiotic processing plant in Brazil where production was limited to one day/week with the operators in full PPE. Every operator who had previously worked in the building was hypersensitive to the antibiotic.
Processing equipment had been replaced and a new processing flow was needed to overcome challenges with open dispensing, charging the 10 ton capacity blender, and the outflow into the automated packaging line.
In the new flow pattern, dispensing occurred directly into the mixer by incorporated an overhead track along which bulk product bags were moved over the mixer inlet. The bags had an extended neck to allow at least two make-and-break connections onto a multi-ring canister connected to the blender inlet. The carriage support incorporated a load cell with computer documentation and feedback. The blender discharge to the packaging line was enclosed in a flexible Polyurethane sleeve. The oil-coated blended material was not dusty and operator contact was eliminated except for maintenance incidents when cartridge respirators, coveralls, and gloves were used.
Work practices followed the route: 1) the bulk bag is mounted onto the track and located above the blender, 2) the neck is securely connected to the canister, 3) the tare weight logged, 4) bag discharge neck partially opened to control flow out of the bag until the desired charge is delivered, 5) the neck tied and separated ready for a further charge of the remaining bulk material, 6) the carriage is moved along the track and 9) the bag unloaded, 8) The next bulk material is loaded onto the carriage and the blender charge completed, etc. 9) the blender is started and the oil additive metered in using spray nozzles, 10) blend for the validated mixing time, 11) the packaging machine is started and material fed to the hopper intake through the shrouded blender discharge, 12) blender discharge is closed after completion in preparation for the next lot.
The operators elected to use coveralls, gloves, and respirators for the charging operation in the event of a breach.
Prior production of 1 lot per week was resulting in hyper-sensitization of the operators involved. The new process allowed the plant to produce 3 lot/per day, 6 days/week as needed with rare cases of hyper-sensitization (1/month) over a one year period traceable to failure to follow procedure.
The facility upgrade rapidly paid for itself allowing the plant to accept more business and a selection of antibiotics processes.
Both FDA and EMA have audited sites using localized containment. Many questions were raised with few definitive answers. In 2005, EMA took exception to handling of high potency materials in a shared facility, which was one of their major regulatory drives based on lack of performance documentation and poorly managed examples across the industry. The facility under challenge was required to prove that the practice was not conducive to cross-contamination, and was reproducibly contaminant free throughout the warehousing, and dispensing areas for the API.
A study design was created using randomized sampling while using a glovebag for dispensing within a dispensing room under negative pressure, with access through a decontamination and disrobing room, and an outer changing room which accessed the shared corridor to the warehouse. Three separate studies were performed within both the dispensing room and the outer hallway. Aggressive air sampling was performed using a one hour air blast at 80 psig swept over every room surface, followed by 3 hours of airborne dust sampling, i.e. the AHERA-like protocol. The filter cassettes were randomly mounted at breathing zone height in a vertical orientation as discussed earlier. Filter analysis was performed by the Quality Laboratory who further developed the analytical method to a quantitation level of 0.04nanogram (ng) and a detection limit of 0.02ng. The spiking level was 1 ng using a reference calibration standard solution. The studies were conducted on the third, fourth, and final (5th) day of production before the weekly total facility clean-down was begun. Each study generated 10 airborne samples plus 2 spike samples and blanks, along with the laboratory retained spikes and blanks. Of the 30 airborne samples all but one were below the detection limit with the exception showing a trace, i.e. at the detection of ~0.02ng. The spiked field samples all returned ~1ng, confirming sampling and analysis method performance.
During a return visit, the data was requested by, and delivered to, the EMA audit team who took it back for their laboratory staff to evaluate. The facility received a clean bill of health and use of a shared facility justified when appropriate steps are taken to function as a fully contained operation(s).
Several years earlier, a similar study was conducted in a development dispensing room operating with an air cross-flow table and a glovebag. The room was accessed through a holding room used as a storage area for bags of development API’s. The 3 study cycles showed full containment using a glovebag. Unfortunately, samples located in the holding room revealed contamination due to the dispensed material. A further exploratory study in the holding room revealed airborne contamination due to all of the stored materials. Quality were duly notified and appropriate measures taken to isolate all of the development materials (a total of >200 Kg of the 4 materials in various stages of clinical trials and representing a major investment) until remedial measures could be established.
Long Range Benefit - A Precautionary Tale
A FabOhio, Inc. client had purchased a 1960’s era Belgian Pharmaceutical facility having an extensive pilot plant capability. While the laboratories had been upgraded to improve the exhaust hoods and floor layout, all >5 Kg scale equipment was still housed in an aircraft hanger style building with a perimeter mezzanine supporting the reactors, with the filtration, drying, and pack-out below on the main floor.
The siting afforded the company a presence in the European Union (EU).
Based on 1990’s initiatives, several development projects involved synthesizing materials of high potency. Corporate Research deemed that the site had to show capability to handle these potent compounds in the ~30Kg lot size targeted for 2nd and 3rd stage clinical trials plus extensive laboratory testing and process finalization for a New Drug Application.
The pilot plant structure was such that a single release of a potent material would lead to uncontrolled material migration throughout the space with settling on 40+ year old surfaces and equipment throughout the area - all of which would make clean-up inconceivable.
The fate of the site came to a head in May, 2005, when site management were told to deliver ~30 Kg (one lot) of a prospective potent drug by year-end or face site closure.
3 commercial ‘containment’ resources were contacted for proposals on how to contain the process materials after they were delivered from the reactor, i.e. for filtration, drying, and packaging. Of the 3, one simply left the site before finishing a walk-through, a second proposed welding a delivery pipe onto the vessel (not realizing the validation requirements of the industry, with only 7 months before actually delivering product), while FabOhio, Inc. proposed a flexible Polyurethane multi-chamber processing suite enclosing: 1) the reactor and bottom valve protruding below the mezzanine into 2) a dedicated room accessing 3) a second room containing the filter/dryer and pack-out area. Liquid waste disposal piping to be connected from the filter/dryer through piping the wall to a reefer for off-site disposal. The suite required access through a personnel robing and disrobing room with a decontamination shower for both operator and materials passage.
The suite would be suspended externally using Velcro loops stretched around a frame support of extruded aluminum which would be reusable. Airflow would enter the enclosure through the personnel access, being drawn through to the filtration/dryer room exhaust creating 5 successive negative pressure zones. The quote was for €12,000 plus shipping for the enclosure, with the frame (~€6,000 plus shipping for a US fabricator) and ventilation system (blower, ducting, HEPA filter, and louvered vent (~€4,000) to be sent for bid based on the sites preferred EU fabricators). The enclosure could be delivered on site within <2 months of order placement.
With the project being critical to the sites future, site management elected to request complete bids from potential EU vendors based on our proposed design with the intent of establishing a working relationship within the EU. One vendor in the UK responded with a bid for €45,000 with the entire suite delivered on-site and assembled by the beginning of November. This would provide 6 weeks for the site to process the requested product. The bid was awarded in early July.
By mid-November, all the vendor could do was promise delivery. A corporate US containment resource consulted with FabOhio Inc. before travel, and was on-site at the request of the vendor (with a window of availability until Nov. 23rd). After much pressure by the site, the vendors vehicle finally arrived on-site mid-day Nov. 22nd, three days later than their most recent commitment. The two technicians had no idea of what they were expected to do! Their vehicle had tools, pieces of building scaffolding, 2 boxes containing 1) the vessel bottom valve enclosure and 2) the personnel access/ decontamination shower access rooms (both fabricated by FabOhio Inc., US, and delivered to the vendor in mid-August), and a roll of 12 mil polyethylene film. In their haste to prepare for travel several components were left behind in their fabrication shop.
The scaffolding support frame was assembled by noon on Nov. 23rd. The FabOhio Inc. prefabricated decontamination shower with personnel access and bottom valve enclosures were installed within 1 hour. The rest of the enclosures were assembled using Polyethylene roll film with the scaffolding on the inside, i.e. not reusable. The film ends had to be double wrapped around the tubing because of the low tack of Polyethylene and the tape adhesive. Assembly took an additional 10 hours, with the bottom valve enclosure finally sealed into the roofing. The technicians expected to drive back to England immediately because Nov. 24th was a holiday in Belgium, however the site insisted they stay to complete installation of the blowers and vent louvers, and ventilation balancing on Nov. 25th and 26th.
Operations immediately started with only a 3 week window. Typical of such planning, corporate requested 3 lots of product rather than one originally scheduled. The site was able to ship the 3 lots before Dec 15th. and proved capability.
Additional containment projects were performed in the following year in both the pilot plant and the small scale laboratory (<5Kg) utilizing flexible Polyurethane containment. In the first pilot plant, the top of a vessel was enclosed within a large glovebag. The enclosure was designed for one lot use. When 3 lots were requested for immediate delivery. The spare glovesleeve needed was removed from a backup glovebag and safely installed during production, allowing completion of the project.
This was a rare instance when fabrication and delivery of a new enclosure from the US to the site would not have allowed on-time delivery of product and desperate measures needed to be taken.
The small scale laboratory required an enclosure into which a filter/dryer could be rolled for unloading, the product transferred to drying trays and transferred into an attached vacuum dryer. The product was to be removed when dry, and packaged. Fog/ mist cleaning lances were to be installed in the shell for cleaning the filter/dryer before removal. The enclosure was supported on an external aluminum frame.
The project from design to delivery of packaged product was completed within 6 weeks.
These 3 instances were sufficient to establish the viability of the site for potent compound manufacture, guaranteeing recognition of their capability and continued support of company goals. Strategically this ensured their presence as an investment within the EU.
Particulate Free Enclosures - Bio-Technology, Nano-Particles, Engineered-Particles
FabOhio, Inc. learned a lot about micro-particulates, both solid and liquid in 1975 during an emergency situation involving the bacterium later to be known as “Legionnaires Disease”. The nutrient source for the infestation was a heterogenous range of microand nano-particles. Excluding them, and sampling for them, in the environment relied on technology developed in the 1960’s for controlling industrial pollutants.
Emerging technology is frequently confounded by the inability to ensure a particle free environment. The inability to exclude contaminants during studies in both the inorganic and biological disciplines confounds researchers when micro-contaminants are present which create artifacts. Particulate free conditions require preparation using sophisticated cleaning and measurement techniques which are uncommon.
HEPA and UHEPA filters are inadequate for collection of particulates of <0.2µm median particle diameter. They are smaller than the wavelength of visible light, consequently they cannot be observed using particle counters or optical microscopes for evaluation. Currently there are no real-time instruments to measure <0.2µm particles in either static or dynamic environments. Particles of these diameters are typically formed under some form of major physical stressor, e.g. thermal, pressure, etc., producing shock nucleation and resulting in formation micro-particles having the same static charge - which prevents agglomeration. They remain in air suspension for hours rather than seconds, and do not settle on surfaces where other particles have already coated, i.e. only as a monolayer coating. The smallest in the range are of the same dimension as larger gas molecules and are in constant bombardment by such. Even the largest particles (0.19µm) are subject to molecular jostling. These features contribute to the failings of conventional filtration, e.g. HEPA, UHEPA, 0.8µm and 0.45µm sampling filters.
Using impaction filtration allows us to both collect and clean particulates from an air-stream. This is adapted to flexible enclosures which can be inflated with an initial environment 99.7% free from <0.2µm particles. Exhaust and refill cycles (removal of air followed by replacement with cleaned air) allow the enclosure to be further purged, while a recycle loop through a double-filtration cycle can reduce an already low particle environment by orders of magnitude based on exponential air-cleaning. The same principle is used to pre-clean a pass-through before opening to the main enclosure.
Lessons Learned While Justifying Acceptance Of Containment Enclosures And The Long Term Impact on the Pharmaceutical Industry
- The ability to consolidate materials within an enclosure means that stray air currents, e.g. eddy currents following people movement, and vortices due to body parts disturbing an airflow are insignificant when compared with airflow control of a room or exhaust hood where particle dispersion becomes uncontrollable.
- Effective physical containment prevents exposure of operators
- Effective physical containment prevents materials migration - both from the process and into the process
- Conventional performance testing using the SMEPAC/ISPE protocol provides personnel exposure data within an enclosed space which may, or often may not provide information about the performance of a containment enclosure, and is not statistically robust.
- Randomized sampling based on the AHERA model provides the engineering capability of a containment enclosure which can be repeated throughout the lifetime of the enclosure.
- Surrogate sampling only provides meaningful data about containment performance when dustiness characteristics are known, the material is representative of the API, and must not create a quality challenge, e.g. using one API to test for an another API must not create cleaning problems.
- Engineering performance testing must not be biased by variables such as sample flow-rate, cassette opening size, and airflow direction, i.e. eliminate particle size cutpoint bias.
- Field and laboratory sample blanks and spikes are a critical aid in supporting the integrity of a survey data set, especially when the sample results are at, or approaching critical decision points such as non-detect, detectable but not quantifiable, quantifiable, OEL, and corresponding Action Level.
- Isokinetic sampling for both equipment and personnel is unfeasible due to the multitude of variables existing in any workplace.
- To ensure reliable performance, localized flexible enclosures are best designed with every face integrated and openings created using flexible sleeves.
- Cleaning of room and equipment surfaces is not as simple as washing down the walls (deluge or pressure washing) which requires copious amounts of wash-water and repetition producing large volumes of contaminated waste-water.
- Fog/mist room and equipment cleaning for both water-soluble and insoluble materials can be accomplished using a single application and creates less waste-water (<1/50th).
- The AHERA sampling protocol is also effective for large space air sampling, e.g. processing room, hallways, maintenance shop, etc. delivering repeatable and statistically valid results which can pass Agency scrutiny.
- Particle counters can be used for tracking emission releases and airborne residual after clean-up, as well as tracking airborne particle decay rates when used in airconditioned facilities, i.e. no condensable gases or vapors present.
- Acid Yellow when milled with a surrogate such as lactose is an effective dye substance for testing tracking and cleaning performance.
- When establishing OEL’s, developing particulate sampling methods, and PPE decisions, materials vapor pressures must be considered.
- For effective containment of a process or processing facility, a fully trained and committed operator is a must, and all testing should be performed using operators.
- The best containment enclosures are always achieved when operators have input to design decisions.
- Flexible containment enclosures can be of any size within a given space, from a simple pipe flange or torch protection, to a multi-room suite with a cascading pressure drop, and even multi-level.
- Flexible containment performs equally with rigid containment, and as prototypes are an effective, economical, and timely way for creating optimum final stainless steel glovebox/enclosure designs costing upward of $100,000.
- Flexible enclosures have been successfully adopted on many occasions to rehabilitate existing facilities which would otherwise raise serious Risk-MaPP performance outcomes.
- Flexible enclosures are an ideal method for achieving fast processing changes and turn-arounds, e.g. third-party manufacturing and toll operations.
- Optimum cost effectiveness is achieved using well designed flexible glovebags, rooms, suites, etc. by delivering timely solutions and requiring a minimum of clean-up.
- For reducing and solving the 3 major materials release issues - Worker (Occupational Health), Materials Migration (Quality), Waste Consolidation (Environmental), adoption of flexible containment solutions has no equal based on cost, timeliness, and operator acceptance.