In response to my article, High Potency Regulations (Contract Pharma, November/December 2013, bit.ly/1gzrYth), Mr. LeBlanc and Ms. Ginsbury assert, “We would like to see EMA take a bold step back and return the issue to where it belongs — in the hands of the GMP/GDP working group, addressing ‘certain’ highly potent products. There is no scientific justification for applying the ADE/PDE tool to products that everyone seems to agree have been safely made since the 1/1000 criterion was adopted in the early 1990s. This approach is not one of turning the clock back; it is recognizing what we have learned about safety, efficacy and quality of products during the past 20 years and is a genuinely risk-based approach.”
There are several troublesome issues in the statement above.
1. “…return the issue where it belongs – in the hands of the GMP/GDP working group, addressing “certain” highly potent products.”
The GMP/GDP working group began the journey to update the GMP requirements with respect to the need for dedicated facilities in 2004 and published a concept paper in 2005. This document states:
“An expert agreement should be obtained when and for which “certain” substances separate production
buildings should be mandatory. At the same time a definition of “exceptional cases” should be given as
to when production in campaigns may be acceptable in the same building.
“The decision process for production conditions should be science-based for each product, taking into account quality risk management tools as described in ICH Q9. Production conditions for each product should be scientifically justified and established during product development and be addressed as part of a company´s quality system.”
Note in the above statement from the EMA it is expected that each product is evaluated.
In 2008 the GMP/GDP working group updated the status of dedicated facilities GMP revisions and stated:
“At present the GMP/GDP Inspectors Working Group agrees that the Guide should indeed identify those products
in which the use of dedicated facilities is mandatory. It is acknowledged that the drafting group needs input from toxicological/pharmacological experts in order to develop this part of the guidance so that it is based on
well-grounded scientific principles and this expertise has now been made available. EMEA’s Safety Working
Party will be asked to endorse the outcome of this expert input before the new guidance is released for public consultation.”
With the above statement the EMA GMP/GDP working group is acknowledging the need for subject matter experts to assist in developing a scientific approach. If this group is acknowledging the need for professional assistance how can industry tell them it is not necessary?
During the time from 2005 until the publication of this update in 2008, the EMA working group was provided with proposals on how to use scientific risk-based approaches rather than a list of compounds or classes of compounds to determine the need for dedicated facilities. The main thrust of these proposals was that the compound alone (the hazard) does not represent the risk of cross-contamination; manufacturing processes, volumes, product mix, GMP compliance history, etc. all needed to be factored into the assessment to determine the probability of exposure. The risk is the combination of the probability of occurrence of harm (exposure) and the severity of that harm (adverse health effect of the compound). Hence a case-by-case assessment is needed to determine the need for dedicated facilities. It is quite possible that one manufacturer producing the same product as another manufacturer would require a dedicated facility while the other has controls in place that allow the use of shared facilities. EFPIA was one of the organizations suggesting this approach in its November 2005 resource document and ISPE, through its Risk-MaPP initiative, also supported this approach.
A case-by-case evaluation does not necessarily mean each and every product in a facility is evaluated equally. Similar to using a matrix approach in cleaning validation, the worst-case scenario must be understood, justified and used as the basis for the risk assessment.
2. “There is no scientific justification for applying the ADE/PDE tool to products that everyone seems to agree have been safely made since the 1/1000 criterion was adopted in the early 1990s.”
Who is everyone? Which products does everyone seem to agree have been safely made since the 1/1000 criterion was adopted? Does everyone have the same list of safely made products? Clearly there are products where the 1/1000 criterion is not adequate — otherwise the statement would not have a qualifier, and a change for “certain” products would not be warranted. One only needs to think of a cytotoxic antineoplastic agent that is given at a dose of 500 mg/day. Applying the 1/1000 criteria would yield a dose of 500 mcg/day, which is at least two orders of magnitude too high for this kind of product.
So again we are back to determining “certain” — which again takes us back to a hazard-based response to managing the risk of cross-contamination and requiring dedicated facilities. How do we determine which products fall into the category requiring ADE/PDE calculations versus those that do not? If a true scientific risk-based approach is applied, the outcome should determine whether or not a company (or even a site within a company) can safely manufacture medicinal products in shared facilities.
The ADE/PDE categorizes the hazard of the compound. The same data that are used to determine the appropriate therapeutic dosage(s) is in fact part of the same data used to determine the ADE/PDE. The difference between the two is the application of the data; finding the appropriate therapeutic dosage where there is the desired effect on the body whereas finding the ADE/PDE looks for the dosage that causes no adverse effect (NOAEL) on the body. Once the NOAEL is determined, factors are applied to address any uncertainties/variability in the data. These factors are applied by toxicologists. Toxicologists also assist in determining the appropriate therapeutic dose(s).
The ADE/PDE is just the starting point and is not in and of itself a risk assessment. To determine the risk of cross-contamination, an assessment of the facility’s operations — including historical GMP compliance record, processes, and robustness of Quality Systems — must be undertaken. The assessment of the facility, procedures and operations should differentiate the risk at companies and even facilities. The manufacturing of one of the products “that everyone seems to agree have been safely made since the 1/1000 criterion was adopted” at different companies/sites in a true scientific risk-based approach could have a wide range of risk levels. So if one company does not truly understand or apply the GMPs and has visible residue/dust in its facility, its level of risk should be higher than a company with impeccable GMP compliance. Part of the risk assessment process — whether a Risk-MaPP approach or not — is to match perception with reality. Just because a company claims to have followed one approach or another does not ensure they did so or did so properly.
To further expand upon how the risk is determined not only by the hazard (the compound) but by the processes, review the following table. It represents three different (real) facilities that all manufacture acetaminophen tablets/capsules in shared facilities with their corresponding cleaning limits based on the 1/1000 criterion and the ADE. For Facilities A and C, the 1/1000 criterion is below visual acuity and if manual cleaning is used, some type of chemical analysis is needed after each API changeover cleaning, to ensure the equipment is clean to the limit. Whereas with the ADE-based limit, visual detection is possible, thus eliminating the need, cost, and time for a chemical analysis. Facility B is clearly within visual acuity with either set of limits and in this case the visual limit is lower than the calculated values and would be the overriding factor in assessing cleanliness.
In reviewing the portfolio of a large generic facility that manufactures approximately 400 products with approximately 180 different APIs, 48% of the APIs were brought from below visual acuity to within visual acuity, saving the time and cost of analytical analysis between API changeover by using the ADE criteria. Only 8% of the APIs saw lower limits using the ADE criteria.
The U.S. FDA and EU GMPs do not set acceptance limits; it is up to each company to establish and justify acceptance limits. Then they have to meet those limits for each and every cleaning between different products/APIs. If they do not meet those limits, an investigation should occur to determine the root cause of the failure and whether or not the products are safe.
So again the misconception that, because a limit is lower, it is safer is turned on its head — the application of the limit and the risk assessment will determine the risks, not just the limit itself. In facilities A and C, with the 1/1000 criterion in a manually clean process with only visual inspection, neither would not be in GMP compliance (cannot verify the limits are met) and get a higher risk for cleaning than Facility B or if the ADE criterion were used, since these limits can be visually detected.
3. “This approach is not one of turning the clock back; it is recognizing what we have learned about safety, efficacy and quality of products during the past 20 years and is a genuinely risk-based approach.”
Given the recent FDA warnings on the use of acetaminophen, it is hard to accept that just because a product has been manufactured for 20 years or so, it is presumed safe, efficacious and of high quality. As time goes on, more data becomes available and with this new data new directions may be warranted. Just as in the case of new data on the use of acetaminophen, new data — actually existing data used in new ways — provides a new approach that is better and safer than the existing 1/1000 criterion approach. There is nothing risk-based about blindly applying 1/1000 to a low therapeutic dose to determine safe limits for potential cross-contamination. The 1/1000 criterion was determined by taking three factors of 10 based on:
“The first is that pharmaceuticals are often considered to be non-active at 0.1 of their normal prescribed dosage;
the second is a safety factor; and the third is that the cleaning validation program must be robust, i.e., be vigorous
enough that it would be considered acceptable for quite some time in a world with ever tightening standards.”
Note the third factor is to buy some time to account for “ever-tightening standards”. Well, maybe that time limit has run its course.
Some other misconceptions worth clearing up are discussed below.
- Each product has a separate ADE/PDE: Actually the ADE/PDE is associated with the API(s) in the product, so while a company may have thousands of products, the number of APIs would significantly less.
- It is costly if ADE/PDE values need to be obtained for all products: While there may be more effort required from toxicologists to determine and document these values, the overall cost to the company will be a net savings when factoring in the impact on verifying cleaning for product contact surfaces or as indicated by the EMA the overall savings in safely sharing facilities. Once the value is determined for an API it would only need review or revision if more pertinent information became available such as additional studies or adverse events. In addition, there is a service that has some ADEs, which can be automatically downloaded with a monograph for a reasonable fee. Also many ADE/PDE values have been set for APIs in the generic market, so companies need to find a way to share information or divide the workload.
- Obtaining ADE/PDE values for all products is a burden and may lead to drug shortages: Hiding behind the possibility of drug shortages to negate the use of good science is not a patient-centered response. Many companies are relying on CMOs to help produce their products. As clearly stated by the regulatory agencies, companies and their CMO partners are jointly responsible for the quality of the products produced at the CMO. A CMO typically has several different clients. In order to manage their risks of cross-contamination across all products and clients, the CMO must either dedicate facilities (via segregation or campaign) or safely manufacture in shared facilities. To show that the CMO is safely manufacturing in shared facilities, it could apply risk management principles to its operations, which includes setting appropriate acceptance criteria and properly assessing the risks. If the CMO cannot control its risks it must either dedicate facilities or decline the project to manufacture the product. This is a cost of ensuring quality to clients and can be a good business development tool. Clearly a client will have more comfort with a CMO that has a good understanding of its risks than another that doesn’t.
- Low hazard APIs are less risky than high hazard APIs: It cannot be stated enough that hazard alone does not determine the risk. Adherence to GMPs is a significant factor in deter mining the risk as part of the probability of occurrence, so to insist whether or not shared facilities can be safely used is separate to GMP compliance is incorrect. Many facility audits around the world, as well as Warning Letters and 483s, indicate there are many lapses in GMPs and many of these lapses can lead to cross-contamination. This tends to be more of an issue with facilities manufacturing lower hazard products, since they are made in greater quantity and operator safety is not an overriding concern.
- Cross contamination occurs only from inadequate cleaning of shared equipment: While the inadequate cleaning of shared equipment is a significant risk of cross-contamination it is not the only mode that may lead to cross contamination. Other modes of cross-contamination include mix-up, mechanical transfer and airborne sedimentation. The ADE/PDE value is needed to assess the risks from these other modes as well. In a solid dosage facility that manufactures low hazard or even medium hazard materials, the processes tend to be open, allowing dust to settle on non-product contact surfaces such as uniforms, gloves, tables, and the exterior of product contact equipment. This dust is now available as a potential cross-contaminate through the mechanical transfer mode. The GMPs call for visually clean facilities but this is not feasible during the actual manufacturing processing when open systems are used. The ADE/PDE values help assess if this is a high risk or not.
- Walsh, Andy “Cleaning Validation for the 21st Century: Acceptance Limits for Active Pharmaceutical Ingredients (APIs): Part I” Pharmaceutical Engineering July/August 2012
- Fourman, G., and Mullin, M., “Determining Cleaning Validation Acceptance Limits for Pharmaceutical Manufacturing Operations,” Pharmaceutical Technology, April 1993, www.pharmtech.com
Stephanie Wilkins, PE is president of PharmaConsult Us and participated on the PDA’s commenting team and was the technical lead for ISPE’s commenting team on the EU revisions to Chapters 3 and 5 and the Health Based Exposure Limits Guidance, as well as task team lead for ISPE’s Risk-MaPP Baseline® Guide. She can be reached at firstname.lastname@example.org.