Managing Host-Cell Proteins: Robust Risk-Assessment Frameworks for Process-Related Impurities in Biological Products

20-5-de-Zafra-P1-300x300.jpg

Hamster phospholipase B-like 2 (PLBL2) is a host-cell protein (HCP) associated with Chinese hamster ovary cell systems used to express monoclonal antibodies. PSI, THE ONLINE PROTEIN MODEL PORTAL (HTTPS://WWW.PROTEINMODELPORTAL.ORG)

Although biomanufacturing processes are designed to generate highly pure drug substances, some host-cell proteins (HCPs) copurify with target proteins and thus remain in finished drug products. Biopharmaceutical developers are keenly aware that such impurities must be minimized to protect patients. HCPs can activate several kinds of immune responses in treated patients, including production of antidrug antibodies and induction of cross-reactivity with therapeutic proteins (1–5). HCPs also can diminish drug efficacy, potency, and/or stability (6, 7). Thus, regulatory guidances such as ICH Q6B require developers to identify HCPs and related contaminants that could end up in a drug substance, then monitor their removal during purification processes (8).

It’s generally believed that — in many cases — HCPs can constitute 1–100 ng/mg (1–100 ppm) of a biological product, based on results from an HCP enzyme-linked immunosorbent assay (ELISA) (9). However, regulatory agencies have yet to stipulate acceptable HCP ranges, instead working with developers to determine appropriate levels for each new protein product. That situation stems in part from the complexity of HCP detection. Although ELISA remains the “gold standard,” mass spectrometry (MS) and related analytical methods for HCP analysis are gaining in sensitivity and accessibility. Host-protein mixtures contain many species, and HCP identification remains difficult, especially for low-abundance (but potentially problematic) proteins. Developers also have much to learn about clinical impacts of — and acceptable ranges for — particular impurities.

The biopharmaceutical industry continues to devise tools for assessing clinical and product-quality risks posed by HCPs in drug substances. Christina de Zafra and four colleagues at Genentech (Roche) published a useful framework for holistic HCP risk assessment in 2015 (10). Their heuristic helped to catalyze the development of several other tools for HCP risk management, including a multifactorial rubric presented by the BioPhorum Development Group (BPDG) HCP Workstream in 2018 (11, 12). Building upon that work, BPDG released a study of “high-risk” HCPs in April 2021 (13).

I spoke with de Zafra early in 2022 to learn about the impetus behind the framework that her team created and to find out how biologic developers are adapting it to evaluate new products. She emphasized the importance of cross-functional and industry-wide discussions about product-quality concerns. She also highlighted advances among in silico tools for assessment of HCP immunogenicity potential.

Her perspective on HCP assessment is particularly valuable because of her training and industry experience in toxicology. After earning master and doctoral degrees in toxicology from the University of Rochester School of Medicine and Dentistry, de Zafra served as a postdoctoral fellow in the pharmacology program at the University of Colorado Health Sciences Center. She worked in nonclinical safety assessment at Genentech and Amgen, supporting development of monoclonal antibodies (MAbs), fusion proteins, antibody–drug conjugates (ADCs), and oncolytic viruses. In January 2021, de Zafra joined Seagen as a director in the Nonclinical Sciences group. She is a diplomate of the American Board of Toxicology and a member of the international Society of Toxicology. She also has served on the BioPharmaceutical Emerging Best Practices Association (BEBPA) HCP Committee since 2018 and joined the BPDG’s Host Cell Proteins and Other Bio-Residual Process Related Impurities Working Group in 2021.

Frameworks for HCP Management
What drove you to create the risk-assessment framework, and how did you develop its criteria? When we developed the framework, I was a member of Genentech’s Safety Assessment department. I had been recruited onto a task force that was assembled to triage an emerging problem, in which we had detected high levels of an HCP in material for a late-stage program. I’m a toxicologist by training, so the way that I approached the problem was different from how my colleagues in chemistry, manufacturing, and controls (CMC) and analytical sciences were thinking about it. I found myself repeatedly explaining what factors I considered when determining the level of risk presented by the HCP. To facilitate our interactions and to get all the different factors out of my head and down on paper, I built a risk-assessment framework and used it to illustrate my points during our discussions.

The criteria that I considered were fairly simple. I asked my colleagues, what was the HCP in question? What did we know about its function? Where was it located in the host cell? And how different were its hamster and human variants? I wondered whether we had historic experience with that protein from material we’d administered either to animals in a nonclinical setting or to human clinical-trial subjects. And what development stage had the drug program reached? I thought about the product’s proposed indication: Was it designed to treat cancer in adults, or was it in the realm of pediatrics? Obviously, those indications raise different risk factors. And how were we administering the material: by intravenous, subcutaneous, or intravitreal injection? Were we giving the drug as a single dose, or would it be administered every week for a prescribed period to treat a chronic condition?

All those factors needed to be considered to determine what level of risk the HCP posed to the development program. Such deliberation, in turn, would support identification of appropriate mitigation strategies.

At the time, which criteria did you consider to be most important? What criteria do you think deserve more attention today? At least in terms of how the framework was conceived, it isn’t advisable to consider a single risk factor in isolation because such criteria have a high degree of interplay. That said, some criteria might be more important than others depending on your context. Again, consider indication: Oncology and pediatrics raise different risk–benefit discussions, for instance.

Dose regimen and route of administration can be especially important contextually. Some routes are thought to be more or less immunogenic — or more or less tolerant of an immunogenic response. For example, if you’re administering a therapy into a patient’s eye, then generating an immune reaction there can be dangerous. The eye is a relatively closed system, so inflammation can lead to loss of vision. Contrast that with IV administration, which is generally considered to be a nonimmunogenic route.

A program’s development phase can be an important consideration. Again, my experience with this HCP risk-assessment framework stemmed from the identification of a problem with a late-stage program. The pivotal, registration-enabling trial had been underway — and that is about the worst time to find out that you have an issue. But if you can identify impurity concerns before you begin clinical trials, and if you have solid animal data, then you will have time and opportunity to fix your process and eliminate problems before you proceed to human trials. It’s important to acknowledge that there is a balance between speed of development and identification of risks because analytical method development is time intensive and because the quality of therapeutic products generally improves over the course of clinical development as process experience and clinical experience are accumulated.

Since our framework was developed, much work has been done regarding impurity identification, especially for high-risk HCPs. Advances in MS instrumentation are facilitating that work. The BioPhorum group’s 2021 publication is a good example of scientists trying to rank HCP risks by determining which proteins might cause more concern than others (13). HCP identity might be an even more critical factor now than it was originally because we can do so much more analysis on that front than we could even 10 years ago, and MS is used so much more often these days.

Back then, when I asked what the HCP was, that information simply might have been unavailable; my colleagues might have had just a protein sequence and would not have been able to tell me, “It’s phospholipase B-like 2” (PLBL2, opening photo). They might not have known. At that point, we would have needed to disregard the identity criterion and think about all the other factors in the framework. That made it more difficult to perform an accurate risk assessment.

Another example is monocyte chemoattractant protein 1 (MCP-1). In the 2010s, Bristol-Myers Squibb was testing a product that had an issue with biologically active hamster MCP-1 generating adverse injection reactions in patients (4).

From a toxicologist’s perspective, it helps me immensely if my analytical colleagues can tell me the name of a problematic protein. Then I can say whether the protein in question raises concerns because it has inherent biological activity.

The criterion that might need more attention is immunogenicity risk. By immunogenicity, I mean immune responses that produce anti-HCP antibodies or that serve unintentionally as adjuvants to increase the incidence of ADAs. By itself, immunogenicity is not necessarily bad; a treated patient might develop antibodies that have no biological consequence. On the other hand, an immune response can be life threatening. Thus, we greatly need tools to guide our immunogenicity risk assessments.

Much work is going on in that area, although many such tools remain in development stages, and we don’t yet understand their predictive power. Researchers are amassing a database that will increase the predictivity of immunogenicity risk. Such a project will take time to complete. But many resources around the biopharmaceutical industry are being devoted to researching and developing suitable tools and assays. I’m encouraged by that level of commitment, and I’m optimistic that we soon could have several methods in our toolbox for assessing immunogenicity potential.

New Applications
Have you considered adding criteria to the framework, or do you know whether industry colleagues are adapting it to their needs? The framework largely has remained the same since it was published. But last year, I joined the Nonclinical Sciences group at Seagen, and as I got to know people within technical operations and process sciences, I learned that my new colleagues had been discussing how to use the framework here. They had considered adding another consideration: HCPs per therapeutic dose. If you know a product’s therapeutic dose, then you can compute the amount of HCPs administered therein. Based on that and on whatever information you have about your antibody’s backbone, you might be able to determine whether the presence of a given HCP could exacerbate your therapeutic’s inherent immunogenicity. Moreover, the team at Seagen broke down the risk assessment further, separating biological activity and immunogenicity risk and identifying the key pieces of data to request from collaborators in Research and Clinical Development for incorporation into the risk assessment.

Could your risk-assessment framework work for impurities other than HCPs? The framework definitely can be applied to other types of process- and product-related impurities. All the same factors must be considered to determine what risks an impurity presents to patients and what type of action should be taken to decrease the levels present. Thus, the framework is highly transferable.

At Seagen, we’ve been using the framework to consider other product-quality attributes. We sometimes have access to historical data that can help us gauge an impurity’s level of concern. When we do, we can correlate particular attributes with incidences of adverse effects. So we are finding ways to combine the framework with prior knowledge and historical information to make informed decisions about how much risk an impurity presents.

Best Practice for HCP Management
When should HCP risk assessment begin, and how would you approach that activity? HCP evaluation should begin as early as possible during process development. Beginning early provides ample opportunities to make process changes and understand their impacts. At Genentech, I worked on a project during which the drug substance used for preclinical studies had an HCP burden in the thousands of parts per million (ppm). We characterized the material, performed a risk assessment, and determined that the burden was too high, even considering the product’s indication for oncology. Just from a basic product-quality perspective, we believed that we could manufacture a better therapy, and ultimately, a simple process adjustment enabled us to diminish the HCP burden from thousands of ppm to tens of ppm.

Even if you have a platform process that performs reproducibly, a new therapeutic target can change its outputs in terms of HCP content. Knowing that, many companies are initiating “prior knowledge” initiatives to catalog their experiences with platform approaches. Generally, the more information that you have about a process, the better you can understand the factors that influence HCP populations.

When considering where to start the risk-assessment process, and thinking in terms of the framework, it helps to rely on prior knowledge and your company’s manufacturing experience. If you’ve witnessed a product-quality concern, then you can refer to what you did previously to address it, using lessons learned from that experience. Your company also might have clinical data about the safety and immunogenicity profiles of products with specific quality attributes. Such information can provide scientists with something like an “exposure margin,” a range of attribute levels that can provide guidance about what factors might or might not cause trouble. But if a quality concern is novel, then other factors in the risk-assessment framework might increase in importance. You might need to err on the side of caution and implement process modifications to mitigate associated risks. Quality concerns also should necessitate a high degree of cross-functional discussion.

How would you approach HCP specification setting? That activity is product specific. It goes back to thinking about patient risks and benefits. In terms of the framework that we’ve been discussing, the two factors that I would consider most are indication and dose regimen. An HCP’s identity could be an important criterion as well. Certain factors can increase a patient’s tolerance of a known HCP. But that’s a slippery slope because your process might not always generate the same number of impurities after each run.

In answering your question, I’m thinking primarily about the oncology landscape. I want to emphasize that we do our best to serve patients in that indication. We are wholly invested in safety, and we are always trying to make drug products of the highest quality. However, patients with life-threatening conditions and unmet medical needs might be willing to take on a relatively higher degree of risk to realize a significant potential benefit.

You noted increasing implementation of MS for HCP analysis. Why does it remain an orthogonal method to an ELISA? Historically, ELISA has been the industry’s gold standard, partly because it is relatively easy to perform and reproducible. MS requires specialized equipment and technical expertise to perform an assay and interpret the resulting data. MS also generates more data than an ELISA does, raising questions about how to manage all that information. Sure, you can run an MS-based risk assessment, but what do all of those data points mean, how do you talk with regulators about them, and which ones do you include in a regulatory filing? Those and many other reasons help to explain why MS has yet to become the gold standard. Several companies are working to explore use of MS as a release method, but difficulties arise with incorporating it into a quality environment, and such efforts are still at preliminary stages. Currently, MS remains a key method to support HCP characterization, especially for root-cause investigation.

What HCPs do you want to learn more about? I am curious about “hitchhiker” proteins and about what you do when an HCP coelutes with your target molecule. How do you deal with that? One example is amyloid beta (Aβ). When developing an anti-Aβ antibody (e.g., to treat Alzheimer’s disease), your process might generate host proteins that have bound to the recombinant Aβ antibodies.

I am also interested in high-risk HCPs and why we are finding those so frequently during our HCP evaluations. What is it about a host cell line, protein-production process, and/or particular product that generates such impurities over and over again? Kelvin Lee and other researchers with the US National Institute for Innovation in Manufacturing Biopharmaceuticals (NIIMBL) are exploring whether further genetic manipulation of a host cell line can decrease the likelihood of problematic HCPs remaining in final drug product.

Key Collaborations
How important were cross-functional discussions to the framework that your team developed, and why might the industry need more of such collaborations to improve HCP assessment? One of the project’s most important lessons was that product-quality issues necessitate cross-functional, multifactorial conversations. I’m not an expert in MS, high-performance liquid chromatography (HPLC), or other analytical methods. Conversely, when my analytical and CMC colleagues identify proteins of concern, they feel ill equipped to think about the “bigger picture,” especially concerning clinical risks that such proteins could pose. Product-quality problems need to be addressed during discussions that have all the relevant people in the room.

Cross-functional teams foster learning. I’ve learned a tremendous amount about analytical sciences, CMC, and process development. That’s been immensely gratifying for me. I’m a much more well-rounded biotechnology scientist than I would have been if I were siloed. I also hope that having me in the room has helped my analytical and CMC colleagues to learn about the toxicologist’s perspective.

Such discussions will benefit a company without drawbacks. Scientists can make good, well-informed decisions about their products and processes if the lines of communication are open, and having cross-functional conversations as early as possible can streamline process-development efforts while helping to ensure delivery of high-quality drug products to patients.

Cross-industry discussions have been valuable ways to share knowledge that companies are building internally. The Biopharmaceutical Emerging Best Practices Association (BEBPA) and BioPhorum are great examples of organizations that hold working-group meetings in which participants openly ask and answer questions about process concerns (obviously while being careful about protecting intellectual property). Then these working groups publish about their experiences — e.g., putting information about high-risk HCPs into the public domain. Small companies can benefit significantly from such cross-industry efforts because they might not have the resources or know-how to solve such problems.

What else does the industry needs in terms of HCP identification or monitoring? Together, ELISA and MS do a decent job of calling out HCPs and helping analysts to identify them. Additional tools that would help to inform discussions about HCP risks include those for assessing the immunogenic potential of an identified protein. I believe that companies should continue to direct resources toward developing such assays and tools — be they in silico, in vitro, or whatever. That is an area where I wish I had more (or maybe better) arrows in my quiver.

Let’s return to cross-functional collaboration. When product-quality challenges come up, if CMC and analytical teams don’t know who to talk to, or if it hasn’t become routine for them to reach out to, e.g., a toxicologist, then nobody outside of those functions will know that a concern exists — and then it can snowball into a significant problem.

As a toxicologist, I tend to seek out all the possible risks and think about all the things that need to be fixed. We need to mitigate our risks, after all. Discussions with my colleagues have helped me temper that mentality a bit to understand process-related trade-offs. Yes, we can try to produce totally pure material, but what’s the cost of that? How many steps in the process would be required, and how much material would we throw out because our specifications are too tight? Historically, biotherapeutics have an excellent safety record, and we must keep that in mind when we’re talking about the lengths to which we will go to make a completely “clean” product. So many factors and trade-offs need to be considered, and that is why cross-functional collaboration is of so much value. A middle ground exists. Finding that is just a matter of understanding each scientist’s entry point into the conversation and having everyone realize that we all have the goal of delivering high-quality medicine to people with serious medical needs.

References
1 Fischer SK, et al. Specific Immune Response to Phospholipase B-Like 2 Protein, a Host Cell Impurity in Lebrikizumab Clinical Material. AAPS J. 19(1) 2017: 254–263; https://doi.org/10.1208/s12248-016-9998-7.

2 MacDonald ML, Hamaker N, Lee KH. Bioinformatic Analysis of Chinese Hamster Ovary Host Cell Protein Lipases. AIChE J. 64(12) 2018: 4247–4254; https://dx.doi.org/10.1002%2Faic.16378.

3 Huang L-Y, et al. Use of Toll-Like Receptor Assays to Detect and Identify Microbial Contaminants in Biological Products. J. Clin. Microbiol. 47(11) 2009: 3427 3434; https://doi.org/10.1128/jcm.00373-09.

4 Leister KJ. Identification and Control of Process Impurities and Product-Related Variants in the Development and Manufacture of a Glycoprotein. IBC Life Sciences Product and Process Variants & Impurities Conference: Washington, DC, 21–23 October 2013.

5 Vanderlaan M, et al. Hamster Phospholipase B-Like 2 (PLBL2): A Host Cell Impurity in Therapeutic Monoclonal Antibodies Derived from Chinese Hamster Ovary Cells. BioProcess Int. 13(4) 2015: 18-55; https://bioprocessintl.com/analytical/downstream-validation/hamster-phospholipase-b-like-2-plbl2-a-host-cell-protein-impurity-in-therapeutic-monoclonal-antibodies-derived-from-chinese-hamster-ovary-cells.

6 Chiu J, et al. Knockout of a Difficult-to-Remove CHO Host Cell Protein, Lipoprotein Lipase, for Improved Polysorbate Stability in Monoclonal Antibody Formulations. Biotechnol. Bioeng. 114(5) 2017: 1006–1015; https://doi.org/10.1002/bit.26237.

7 Li X, Richardson DD. Analysis of Trace-Level, High-Risk HCPs: Proteomics Advances for Preventing Degradation of Polysorbates in Biotherapeutic Formulations. BioProcess Int. 19(9i) 2021: 8–13; https://bioprocessintl.com/analytical/product-characterization/activity-based-protein-profiling-of-trace-level-high-risk-hcps-proteomics-advances-for-preventing-degradation-of-polysorbates-in-biotherapeutic-formulations.

8 ICH Q6B. Specifications: Test Procedures and Acceptance Criteria for Biotechnological/Biological Products. International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use: Geneva, Switzerland, 1999; https://database.ich.org/sites/default/files/Q6B%20Guideline.pdf.

9 Bracewell DG, et al. The Future of Host Cell Protein (HCP) Identification During Process Development and Manufacturing Linked to a Risk-Based Management for Their Control. Biotechnol. Bioeng. 112(9) 2015: 1727–1737; https://dx.doi.org/10.1002%2Fbit.25628.

10 de Zafra CLZ, et al. Host Cell Proteins in Biotechnology-Derived Products: A Risk Assessment Framework. Biotechnol. Bioeng. 112(11) 2015: 2284–2291; https://doi.org/
10.1002/bit.25647.

11 Wang F, et al. Host-Cell Protein Risk Management and Control During Bioprocess Development: A Consolidated Biotech Industry Review, Part 1. BioProcess Int. 16(5) 2018: 18–25; https://bioprocessintl.com/analytical/downstream-development/host-cell-protein-risk-management-and-control-during-bioprocess-development-a-consolidated-biotech-industry-review-part-1.

12 Wang F, et al. Host-Cell Protein Risk Management and Control During Bioprocess Development: A Consolidated Biotech Industry Review, Part 1. BioProcess Int. 16(6) 2018: 42–47, 64; https://bioprocessintl.com/analytical/downstream-development/host-cell-protein-risk-management-and-control-during-bioprocess-development-a-consolidated-biotech-industry-review-part-2.

13 Jones M, et al. “High-Risk” Host Cell Proteins (HCPs): A Multi-Company Collaborative View. Biotechnol. Bioeng. 118(8) 2021: 2870–2885; https://doi.org/10.1002/bit.27808.

Brian Gazaille is associate editor of BioProcess International, part of Informa Connect; [email protected]; 1-212-600-3594.

You May Also Like