OR WAIT 15 SECS
Duncan Low is a scientific executive director in process development at Amgen Inc.
Anurag S. Rathore is a professor in the Department of Chemical Engineering at the Indian Institute of Technology Delhi and a member of BioPharm International's Editorial Advisory Board, Tel. +91.9650770650, firstname.lastname@example.org.
Evaluate and communicate risk to stakeholders.
Enhanced process and product understanding are the basic tenets of Quality by Design (QbD). Although significant advances have been made, appropriate characterization and management of raw materials remain a concern for the regulatory authorities. Because of the large number of raw materials that typically are used in biotech processes, a QbD-based approach for raw material management must be based on scientific knowledge and risk analysis. This strategy will ensure that adequate characterization is performed on those raw materials that have the most effect on process consistency and product quality. Part 1 of this article presented an approach for raw material management following QbD principles. In Part 2, we discuss how to conduct risk assessments for biotech raw materials, along with examples of risk assessment tools that can be used.
The contribution of raw materials to product quality, safety and process performance is considerable, in part because of the potential to introduce contaminants. It is therefore important in designing processes and selecting raw materials that careful thought be given to understanding their purpose and suitability for the intended use. Raw materials also are a major source of variability, and steps must be taken to minimize any negative effects arising from materials and their sources. There essentially are two parts to this: prevention through assessment, inspection and control of incoming materials; and intervention, in which processing conditions are modified to account for variability, sometimes through continued monitoring of materials and their effect on process performance.
Anurag S. Rathore
Part 2 of this 20th article in the Elements of Biopharmaceutical Production series discusses how to conduct risk assessments, handling risk communication to stakeholders, and give examples of applying risk assessment tools for analyzing raw materials in biotech processes.
There are a number of different risk assessment tools available in a range of detail and complexity, and it is important to use a methodology suited to the purpose of the assessment. The approach begins with identification of risk factors through brainstorming, in which ideas are recorded with minimal critical appraisal; or Fishbone or Ishikawa diagrams, which offer a more structured approach. These can be used either to map the process generally, working backwards from the endpoint and identifying all possible risks at a given point, or to structure the evaluation around a given sent of questions at each point (e.g., the 6M's: material, method, measurement, machine, man, and mother nature).1,2 After they are identified, risks can be analyzed by a variety of tools, such as preliminary hazard analysis (PHA), failure mode effects analysis (FMEA), failure mode effects criticality analysis (FMECA), What-if?, fault tree analysis (FTA), event tree analysis, cause consequence analysis, hazard and operability analysis (HAZOP), and hazard analysis and critical control points (HACCP).
A comparison of some risk assessment tools is given in Table 1. PHA scores risk based on severity and occurrence on a five-point scale, with the highest scores representing the greatest risk. FMEA adds detection to the scoring system. By adding the extra scoring category, it is relatively simple to advance a PHA assessment to FMEA, which may become appropriate as more information is known. Additional criteria may be added to distinguish between risks to more or fewer critical quality attributes (CQAs), bringing it to a FMECA assessment. HAZOP assessments commonly are used to evaluate risk to operators or environmental risks.
Table 1. Comparison of different risk assessment tools
The next stage of the risk assessment is risk evaluation, in which the calculated risk index rating is compared to established criteria to determine high-risk items and whether additional information is required, and what mitigation activities are required to bring risk to an acceptable level.
Finally, communication is an important part of any risk management plan. It is essential that the outcome of the risk assessment is communicated to stakeholders and management and the recommendations for further mitigation or acceptance of residual risk are taken at the appropriate level.
Before performing a risk assessment, it is necessary to establish a list of CQAs from the target product profile (TPP) of the molecule in question to establish what is being affected by the material in question.3,4 These can be product related, such as aggregates, deamidated isoforms, incorrectly folded product, or different glycoforms; or process related, such as DNA, host cell protein, cell culture medium components, or purification components.5 Drug product materials also may include materials of construction, dimensionality, labels, and instructions. A more detailed list of product quality attributes, and a discussion of quality attributes is given elsewhere.6
A risk assessment is also a valuable exercise in knowledge management, because it brings together cross-functional teams from quality, manufacturing, and early- and late-stage development. Teams should assemble internal and external knowledge, and information from in vitro, animal, and human clinical studies. The quality of the risk assessment can be highly dependent on the team composition and dynamics, and it is critical to have experienced team members and facilitators available to ensure that all voices are heard. It is strongly recommended to run the meeting face-to-face. It is also recommended that the team begin with the most critical operations near the end of the process and work backwards, because these are the most effective. Lastly, teams should benchmark their scoring against the outcome of previous assessments for similar products and materials to ensure a consistent view of risk.
Risk assessment examples of materials commonly used in antibody purification are provided in this section. The risk assessments that follow are not specific to any product or material, but represent risks that may be present in a class of materials. No specific supplier is identified or scored in the examples, but they are mentioned where there may be general concerns over the source.
The purpose of upstream processes is to express the product of interest. Materials are chosen to maximize productivity and consistency without impairing quality. The process also must be protected from adventitious agents, which could result in the loss of a batch, but are unlikely to affect quality because no product would reach patients. A hypothetical scoring of selected materials used for upstream unit operations is given in Table 2. Complex nutrients such as hydrolysates and yeast extracts are added to cell culture media at various points from thaw to production. They affect growth, titer, cell viability, and product quality, and are a known cause of variability. Functional performance testing is required at great cost to distinguish between lots. Complex nutrients may be a source of viral contamination, resulting in a high severity score: S = 9. This usually is controlled by suppliers, resulting in a lower occurrence score: O = 3. Viral contamination would be observable in the bioreactor, with a detectability of D = 3. Complex nutrients also contain immunogenic materials which typically are removed by downstream processing, so occurrence in the final product is low, and removal of host cell protein can be considered a surrogate marker for the presence of high molecular weight (immunogenic) material, so detectability is good for that parameter as well.
Table 2. Selected upstream materials
There are known process problems with growth media, which can result in differences in product variants (glycosylation, protease activation) caused by inherent variability and the presence of trace elements. These variants may have similar clinical properties; there may also be variability in titer, cell viability, and filterability, resulting in a lower severity score (S = 7). Such lot-to-lot variability has been observed (O = 5). The variability in product performance is clear if cell growth is affected, but variability in product quality may not be observed until further downstream. Detectability is good (D = 3) and in many cases, downstream purification is likely to remove the variants. Guard filters are used as peripheral filters to protect the main process streams from microbial contamination. They may be 0.45 or 0.22 μm filters that can be used elsewhere as sterilizing filters. Guard filters have contact with process fluids and may be a source of leached materials. Severity scores are low (S = 3), and occurrence and detection levels also are low because contact times are limited and filters typically are characterized for leachable materials. The low risk attached to such filters mean that it is relatively straightforward to qualify alternate sources. Complex nutrients and growth media, on the other hand, may require more extensive qualification, but because a disruption in supply could have severe consequences on process performance and product quality, it is important to consider and qualify alternate sources.
The main purpose of downstream processes is to achieve the desired levels of product purity; the secondary objective is minimizing product loss. Product variants, process by-products such as host cell protein and DNA, and process materials such as media components have to be removed and the product transferred to the final formulation. Purification typically is through a combination of chromatography and filtration steps, but may include other process steps (precipitation, partitioning, and crystallization) as appropriate. Hypothetical scoring for a downstream step is provided in Table 3.
Table 3. Selected downstream materials
The materials used in chromatography resins generally are innocuous, although there may be some risk for leachables, notably after long-term storage. These usually are characterized by the supplier, and the information is available as a regulatory support file or drug master file. Solvents (ethanol, benzyl alcohol) frequently are used as preservatives, and this also represents the major risk for residual solvents later in the process. Ethanol is a class 3 solvent and represents a low severity (S = 3). Solvent is present in the material as shipped and after storage, but not in the material as used, because resins are washed by a validated washing process before use. Additional consideration should be given for affinity resins where the ligand leaches into the process and specific removal steps may be required for its removal. Protein A is not particularly toxic and is used in approved FDA processes, but does have biological activity as a superantigen,7 which would increase the severity score (S = 5) and require specific removal steps and controls, but is detectable (D = 3). Resins are the primary agent for product purification and poor performance would have a high impact on product quality and yield (S = 7). However, they are not the final step and typically are used in combination with other purification steps. Resin quality is good, and poor performance is seldom observed (O = 3), and normally is detected (D = 3) by changes in purity or yield. Column performance can be tracked by in-line procedures such as transition analysis to monitor continued fitness for purpose. Resins are not interchangeable because the base matrix and resulting profile of impurities from nonspecific binding will vary, depending on the source. Because this represents a single point of failure, supplier selection is critical, and risk may be managed by using different suppliers at different points in the process or in different processes. This is highlighted by the recently published case study from Toro, et al.8 The authors' company received a supply chain notification from a supplier of an ion-exchange resin. Laboratory experiments performed thereafter showed minor differences in the chromatographic profile and pressure drop across the column at different flow rates. The new resin was therefore deemed comparable to the old resin. However, when packed in 20-cm radial columns, significant differences were noticed in the packing properties of the old and the new resins.
Particle size analysis through a laser-diffraction particle size analyzer showed that the average size for the new resin was 126.9–141.7 mm versus 127.2–130.1 mm for the old resin. Thus, the difference in particle size could not be the reason for the difference in packing behavior. Per the authors, "Since we received the SCN sent by our chromatography media supplier, a significant amount of time and resources have been allocated to identify differences caused by that simple change in the supplier's resin liner source." They further state, "That investigation is ongoing, as is additional experimentation on our part." This case study exemplifies the problems faced when changes are made in critical raw materials.
In-line filters are used for maintaining sterility of process streams, concentration, buffer exchange, and virus removal. Filters usually are made of non-interactive materials, and the chief concern, as with guard filters, is the presence of leachables. The effect on the process depends on the specific purpose of the filters; where product passes through the filter there is minimal contribution to the process, but where product is retained, there is the potential for product losses (S = 5). Occurrence and detectability scores typically are low. In the case of viral filters, there are no specific detection methods for viruses. However, filters are validated for viral clearance and if the filter remains integral after use, this can be taken as evidence of successful performance.
Compendial chemicals are used as buffers and cleaning agents. These are well defined and typically well understood. However, problems may arise from handling issues such as clumping or deliquescence, which could affect dispensing, extraneous matter, insoluble materials, and filterability (S = 3). Compendial materials generally are intended to have minimal interaction with product, but the presence of trace components may result in undesirable adducts or modifications. These usually are not of concern until the final stages of the process where there are no remaining steps between the product and the patient. Final formulation and packaging processes transform drug substance to drug product. All remaining product components (active pharmaceutical ingredient, packaging, delivery systems, instructions, labels, and secondary packaging) should be viewed as high risk because of their proximity to the final product.
This article presents an approach for managing raw materials in the QbD paradigm. It should be understood that risk assessments are limited by what is known at a given point in time. In a recent example, it was noted that the distribution of peaks on a column in the purification train showed differences after a process transfer from one site to another. Although it caused no difference in product functionality, the cause of the variability was sought. After investigation, the two sites were found to be using different sources for a media component. Further investigation showed that the two sources differed in their content of a trace metal ion, which was a cofactor for an enzyme whose activity could explain the differences in peak distribution. Continuous monitoring of processes and the use of techniques such as multivariate data analysis are prerequisites for continuous improvement and superior process performance.9–12
The authors would like to thank Jennifer Mercer, Amgen Inc., Thousand Oaks, CA, for her review of the manuscript and her helpful comments.
Anurag S. Rathore, PhD, is a biotech CMC consultant and a faculty member at the Indian Institute of Delhi, India, +91-9650770650, email@example.com. Rathore also is a member of BioPharm International's editorial advisory board. Duncan Low is a scientific executive director in process development at Amgen Inc., Thousand Oaks, CA.
1. International Conference on Harmonization. Q9, Quality risk management. Geneva, Switzerland; 2005.
2. ASTM Standard E-2476. Standard guide for Risk Assessment and Risk Control as it Impacts the Design, Development, and Control of PAT Processes for Pharmaceutical Manufacture. ASTM International. West Conshohocken, PA; 2009.
3. Rathore AS, Winkle H. Quality by Design for pharmaceuticals: regulatory perspective and approach. Nature Biotechnol. 2009;27:26–34.
4. Rathore AS. A roadmap for implementation of Quality by Design (QbD) for biotechnology products. Trends Biotechnol. 2009 Sep;27(9):546–53.
5. Rathore AS. Follow-on protein products: scientific issues, developments and challenges. Trends Biotechnol. 2009 Dec;27(12):698–705.
6. CMC Biotech Working Group A-Mab: a Case Study in Bioprocess Development. 2009. Available from: http://www.ispe.org/cs/explore_by_topic/pqli_resources and http://www.casss.org/displaycommon.cfm?an=1&subarticlenbr=286.
7. Low D, Phillips J. 2009. Evolution and integration of Quality by Design and process analytical technology. In: Rathore AS, Mhatre R, editors. Quality by Design for biopharmaceuticals. Hoboken, NJ: Wiley Interscience; 2009. p. 255–86.
8. Toro A, Colon J, Melendez-Colon V, Rivera JH. Changes in raw material sources from suppliers. Bioprocess Int. 2010 Apr;8(4):50–5.
9. Kirdar AO, Chen G, Rathore AS. Combining near-infrared (NIR) spectroscopy and multivariate data analysis (MVDA) for screening of raw materials used in the cell culture medium for the production of a recombinant therapeutic protein. Biotechnol Progr. 2010;26:527–31.
10. Lanan M. QbD for raw materials. In: Rathore AS, Mhatre R, editors. Quality by Design for biopharmaceuticals. Hoboken, NJ: Wiley Interscience; 2009. p. 193-210.
11. Kirdar AO, Green KD, Rathore AS. Application of multivariate analysis for identification and resolution of a root cause for a bioprocessing application. Biotechnol Prog. 2008;24:720–6.
12. Rathore A, Johnson R, Yu O, Kirdar, A, Annamalai A, Ahuja S, et al. Applications of multivariate data analysis in biotech processing. BioPharm Int. 2007;20(10):130–4.