Introduction
FDA has defined Process Analytical Technology (PAT) as “Systems for analysis and control of manufacturing processes based on timely measurements during processing, of critical quality parameters and performance attributes of raw and in-process materials and processes to assure acceptable end product quality at the completion of the process” [1]. The concept of PAT aims at better understanding manufacturing processes, monitoring quality parameters continuously and in a timely manner (preferably in-line or on-line) and thus becoming more efficient in achieving quality assurance while reducing process variability, minimizing deleterious events, and relying less on end-product testing as an absolute arbiter of product quality. The agency drafted “Guidance for PAT, A Framework for Innovative Pharmaceutical Manufacturing and Quality Assurance” (Sept. 2004) in an effort to coordinate and oversee from a compliance vantage such efforts industry wide. This paper is an attempt to view PAT and Rapid Microbiological Methods (RMM) against the backdrop of the Bacterial Endotoxin Test (BET); viewing the BET as a prototype for PAT efforts due to its long, diverse and successful history of implementation. Rapid methods in general seek to employ new technology to provide the real-time results that PAT seeks. After a (very) brief history of microbial surveillance and a short introduction to PAT/RMM, some broad categories of PAT requirements or desired attributes will be contrasted with historic BET efforts including: choice of technology (with subcategories), regulatory history, and quality mindset. This is an effort to examine some criteria that may be useful in determining the goals of any specific PAT effort before embarking upon it.
Broad Context for Microbial Surveillance
It has only been some 330 years since Antonie van Leeuwenhoek first peered through his little home-made lens to discover the world of “animalcules” or microbes. Since then, they have been surveilled and a greater understanding painstakingly gained by differentiating them into categories. Increasing control has been gained over them to lessen their activities in despoiling and poisoning foods, and in causing disease in domesticated animals and humans. Most recently they have been genetically characterized and used as production tools to battle, in some cases, the very diseases they cause.
The micro-organisms proliferating in the oceans, lakes and rivers of the world are mainly Gram negative and thus contain endotoxin. Gram negative bacteria are among the most prevalent organisms in the biosphere and, therefore, endotoxin is among the most prevalent of bio-molecules given its occurrence in them and its great stability. Gram negative microbes are so rife with life they have been found in frozen polar ice and in boiling hot springs. As they grow and die, they leave endotoxin as an artifact or residual skeleton. Many of the infamous plagues or scourges of the world have been caused by Gram negative organisms [typhus (Rickettsia), plague (Yersinia pestis), fecal - water contaminants (E. coli, Salmonella), etc.]. This is the tension in nature that exists between the lower and higher life forms. The higher is built on the lower but the higher remains, preyed upon, infested and assailed by the lower- even as parasitically and symbiotically infested [2]. This ancient war of prokaryotes and metazoans of which Limulus [3] is one of the- if not the- most ancient creatures still around (estimates are 400 – 450 my) has been going on for half to three quarters of a billion years. The presence of germ-line encoded receptors for the artifacts of prokaryotes (endotoxin being one of many) on the internal tissue cells of all metazoans testifies to the depth of the ongoing struggle.
Parenteral manufacturing can be seen as an extension of this ancient war between higher and lower life forms and water for injection (WFI) can be seen as the life blood of the parenteral manufacturing process. Thus the monitoring of this process has come to mimic nature’s own complex efforts (in the bloodstream) employing redundancy via both mobile, upstream detection and downstream quality functions. The advent of the production of parenteral drugs brought with it adverse reactions they called at the time, “injection fever”. Seibert is credited with solving the riddle of this pyrogenic contaminant which came to be known as endotoxin and subsequently endotoxin had to be precluded from parenteral manufacturing processes lest the injection one receives bring with it fever and potentially shock and death.
PAT / RMM Background
Industry and FDA PAT efforts began in concert with the publication of FDAs “cGMPS for the 21st Century” [4] (Aug. 2002) which encouraged the early adoption of new technologies, modernization of quality management techniques, and adoption of risk-based approaches to the monitoring of process control points. In simplistic terms, it encouraged an approach analogous to a method of waging war attributed to a Union general: “git thar furstest with the mostest.” This is to say that the acquisition of knowledge is to be pushed up stream in the process (where), is obtained rapidly (when), and reveals an increased amount of data about the given process parameters (what). Industry developmental efforts currently seem to focus more on the “when” of achieving PAT, sometimes to the exclusion of other relevant criteria.
An excerpt from the FDAs initial guidance (2002 paper):
The PAT guidance… discusses a range of flexible options for qualifying and justifying new technologies, such as modern on-line process analyzers intended to measure and control physical and/or chemical attributes of materials to achieve real time release. By emphasizing the importance of understanding manufacturing processes, the Agency hopes to create win-win approaches for enhancing the use of product and process development knowledge throughout the life-cycle of a product. It is our goal to encourage industry to explore new, efficient pharmaceutical manufacturing technologies with the assurance that we are working to develop new regulatory paths to address the technical and regulatory issues and questions anticipated during the introduction of such new technologies. We believe that a focus on process understanding will facilitate risk-based regulatory decisions and innovation as well as the use of appropriate risk identification, management, and control methodologies.
No PAT/RMM implementation today will proceed without also putting forth first a business case [5] that will clearly set out the advantages versus costs of employing new technology. Follow the link below to access an example case study on “Justifying RMM for Environmental Monitoring.”
Choice of Technology
Complexity - The word I want to use for the best technology is elegance. Many microbiological devices are being developed to enumerate organisms for example; however, a room full of computers and accompanying lasers or flow columns in some cases brings with it a “cure” for a measured pace of results that can be worse than the affliction. By worse, I mean in terms of complex calibration and constant analyst attention and efforts at method and computer compliance that may cast doubt as to the state of inspection readiness that could be maintained. A ten-fold increase in the level of system complexity is not in and of itself an elegant solution. Additionally, by gaining some short time in obtaining the desired information other information can be lost, such as the destruction of organisms enumerated such that they cannot be subsequently identified.
Rapidity /Placement - The reaction of the horseshoe crab’s blood with gram negative bacteria was readily observed by Fredrik Bang in the 1950’s upon dissection of a sick Limuli- its blood had turned to a blue jelly. This lead fairly quickly to Levin and Bang’s being able to develop a gel clot reagent that eventually would supplant the use of the rabbit pyrogen test almost completely. In terms of rapidity, it was not so much that an hour’s gel clot test was so much faster than the half to three hour interval required for the rabbit pyrogen test but rather that it allowed the removal of so much infrastructure needed to care for the rabbits. In some cases today, it is the opposite in that sophisticated infrastructure is being piled up to do the testing. The subsequent move to photometric or kinetic tests from the gel clot assay brought with it a simple spectrophotometer and computer but not an avalanche of sensitive and hard to keep equipment previously alien to the lab.
Some technologies are more amenable to moving “upstream” than others, upstream meaning earlier in the manufacturing process as opposed to the stationary quality control function (batch or endproduct testing). Instrumentation for both on-line (in-line) and at-line (point of use) systems has been developed for BET testing. The former system is plumbed into WFI lines and utilizes a recently developed reagent that is the cloned biosensor occurring in natural Limulus blood amebocytes that is used to make lysate (LAL). A recent, handheld point of use system is available to take upstream, real time measurements utilizing a pre-configured spectrophotometer and prefilled (with reagents and standards) disposable cartridges. These results are ready in approximately 15 minutes per single sampling event. The placement of testing upstream means that samples do not have to be transported to the lab, removing transportation as well as the lab itself including sample receipt, refrigeration, and other testing activities.
Platform - The best technologies develop into broad based, widely applicable platforms. This has been discussed for LAL applications above, however, the most recent application to PAT efforts continues to surprise in terms of the range of capabilities being provided for users by the specific companies involved. In some cases, these technologies are using a platform that has the capability of testing microbiological quality attributes other than endotoxin. In the future, perhaps one will be able to pull a sample (manually or via an in-line sample), test for endotoxin, total organic carbon (TOC), bioburden, set aside a parallel sample for later microbial identification (if necessary), and send the results wirelessly to a centralized software application – all from the same portable or in-line instrument.
Increased Surveillance /Enumeration/ Speciation - James Cooper’s initial application of the LAL test to radionuclide cisternography [6] was followed by a tidal wave over the years of expanded applications for broadening the testing of in process drug manufacturing materials. The amount of testing and thus process control from an endotoxin control vantage over more and more pieces of the manufacturing puzzle (components, raw materials, excipients, API, bulk and finished drug) simply would not have been practical using the rabbit pyrogen test. Not only has it allowed the expansion of testing but it has done so at an ever increasing sensitivity and speed. Early lysates were rated in EU’s at 0.12 EU/mL and now are commonly used down to 0.005 EU/mL. Assay times are now as fast as 15 minutes and some speculate that a 5 minute test may be around the corner.
The detection and enumeration of microorganisms and their artifacts is a broad spectrum of activities depending upon the stage of cleanliness of the articles being tested which is a function of the product or raw material type; it also generally relates to the phase of the manufacturing process over time. One wants to be able to at different times and places in the process (a) count all the viable organisms present (b) count only certain types (i.e. coliforms), and/ or (c) know exactly what species are present. Microbiologically the questions are: “What can we know?”, “When can we know it?”, and “What is the cost benefit dynamic of the old versus the new technology in terms of both the cost of instrumentation/ consumables/expertise/labor and the anticipated manufacturing benefits (increased microbial surveillance, inventory control, etc.).” The choice to know everything immediately is rarely a practical expectation. However, there is now technology that aspires to do this. The use of PCR [7] and DNA microarray [8] can detect thousands of different types of microbes in a water sample and based on relative strength of the individual measurements can provide an estimation of each type detected. However, the additional cost and complexity will not make this a practical reality any time soon. It is also of limited value in sorting out viable versus non-viable organisms (i.e. residual DNA from broken cells remains active).
Technological and Regulatory History
No technology is developed in a vacuum. New methods typically bring with them either the baggage or the benefits of the technology from which they were derived. The early gel clot test had no coattails to ride upon but did have the benefit of the burdensome nature of the rabbit pyrogen test. This made the cost-benefit analysis fairly intuitive and brought about the intensive interest and effort necessary to bring the BET (gel clot test) to fruition. Many current efforts seem to lack even the benefit of generating such excitement when compared to a simple agar plate method. Having established the gel clot test as the USP test, subsequently the efforts to include the photometric tests seemed certain if not slow at the time. The use of kinetic testing as a USP alternate test for in process testing took place for many years before it was finally (the way it seemed at the time anyway) adopted into the USP as an equivalent test (photometric test).
Quality Mindset
One cannot help but wonder if some of the hesitation in implementing upstream monitoring is due to the difference in mindset between manufacturing and quality (oversight) functions. The former strictly performs and the latter gauges. The quality control labs are accustomed to determining appropriate specifications and routinely develop tests to gauge quality attributes but this is a new paradigm for manufacturing areas. Increased monitoring in real time will bring more data and potentially data that one does not know how to integrate into the process (i.e. increased variability may accompany an increase in surveillance or limits may be challenged by the detection of organisms that are stressed and were not detectable before). The process cannot grind to a halt while one determines what to do with the data. A clear strategy has to be set out. Typically issues will be resolved during a parallel testing development phase. FDA has said that they will consider such data as “research” data and it will not be used during this phase to determine compliance unless there are special circumstances. “Data collected using an experimental tool should be considered research data. If research is conducted in a production facility, it should be under the facility’s own quality system” [9]. And, “FDA does not intend to inspect research data collected on an existing product for the purpose of evaluation the suitability of an experimental process analyzer or other PAT tool” [10]. Effective interaction between the quality component- where the expertise for such testing resides and the manufacturing areas seeking to employ the technology will be key to implementing such systems.
Revisiting the Questions
It seems pertinent to expand upon the three questions put forth above for any specific microbiological PAT analytical application. The first question, “What can we know?” is perhaps more relevantly rephrased as “What do we need to know?” to take into account both the quality attribute to be determined and the product type to be tested.
The second question, “When can we know it?” seems to be a current focus of PAT technology development and implementation efforts and in the older quality control vernacular means “When can we get a result?” In upstream testing of processes this question takes on a different meaning in proportion that one is removed from the downstream delay of testing that involves sending samples away to a distant quality infrastructure to await test results. Thus, even older technology employed upstream can serve a greater utility. In some cases the rush to know will be tempered by the fact that immediate knowledge of one quality attribute will have to sit and wait until another, slower attribute is determined.
The third question is perhaps the most important in actually employing new, often costly, technology. “What is the cost-benefit dynamic of the old versus the new technology?” This question perhaps should be married to another question: “Will this result in increased microbial surveillance?” New technology for the sake of new technology may not provide more microbiological knowledge of a process and may not increase the rapidity of bringing such knowledge to light without leaving behind or obscuring other data. If it brings a much greater degree of complexity with it then it may also bring risks to compliance and even to the basic scientific monitoring function that was not inherent in the older, simpler technology. No one wants to be “stuck in the past” using old technology, but progress is made slowly for a reason, the tried and true balances the new and unproven. In the world of BET testing, the old has become new time and time again (taking its cue perhaps from its mascot, the horseshoe crab) and one wonders if some opportunities in advancing current technologies are being overlooked by the constant focus on the new and improved. This ancient-modern synthesis has sometimes been a paradigmatic path that has shown interesting results in terms of the re-invention of various technologies over the years [11].
The old and the new. Photo: public domain (NOAA)
References
1. Process Analytical Technologies (PATs), Applications and Benefits Working Group, 2/28/02
2. The human body harbors a 10-fold greater number of microbial cells than human cells. The commensal flora includes microorganisms that occasionally cause disease, especially when host defenses are impaired (due to immunosuppressive drugs, disruption of anatomic barriers, suppression of bacterial flora with antibiotics, or insertion of artificial surfaces). – Detection and Identification of Previously Unrecognized Microbial Pathogens, by David A. Relman, Emerging Infectious Disease, Vol. 4. No. 3, July-Sept. 1998.
3. The LAL reagent used to detect endotoxin is derived from Limulus (the horseshoe crab)
4. See FDAs: “Pharmaceutical cGMPS for the 21st Century—A Risk-Based Approach: Second Progress Report and Implementation Plan”, 8/2009.
5. See: http://www.pharmamanufacturing.com/ articles/2009/081.html and “Rapid Microbiological Methods and Demonstrating a Return on Investment: It’s Easier Than You Think!”, by Dr. Michael J. Miller, American Pharmaceutical Review, July/Aug. 2009.
6. “Endotoxin as a Cause of Aseptic Meningitis after Radionuclide Cisternography”, by J. F. Cooper and J. C. Harbert, Jour. of Nuclear Medicine, Sept. 1975, Vol. 16, No. 9, Pages 809-813.
7. “Nucleic Acid Amplification –based Rapid Microbiological Methods: Are these technologies ready for deployment in the pharmaceutical industry?”, Claudio D. Denoya, American Pharmaceutical Review, April-March, 2009.
8. “Molecular Biology and DNA Microarray Technology for Microbial Quality Monitoring of Water”, Lemarchand, Masson, and Brousseau, Critical Reviews in Microbiology, 30:145-172, 2004.
9. Guidance for Industry PAT – A Framework for Innovative Pharmaceutical Development, Manufacturing, and Quality Assurance, U.S. Dept. of Health and Human Services, FDA, Sept. 2004.
10. Ibid.
11. See: Sequence-Based Identification of Microbial Pathogens: a Reconsideration of Koch’s Postulates, D. N. Fredricks and D. A. Relman, Clinical Micro. Reviews, Jan. 1996, Vol. 9, No. 1, p. 18-33.
Kevin is a microbiologist at Eli Lilly & Company with 27 years in the pharmaceutical manufacturing environment. He has authored the book “Endotoxins” and “Microbial Contamination Control in Pharmaceutical Manufacturing” as well as journal articles and industry presentations. He has a Bachelors of Science from Texas A&M University and is a long standing member of the LAL Users Group. Readers may contact the author directly at: [email protected]