To open the fourth session, session moderator Krishnendu Roy, the Robert A. Milton Endowed Chair and the director of the Center for Immunoengineering at Georgia Tech, agreed with earlier discussions about the value of embracing variability during the discovery and early-phase clinical trial stages. However, he said, during the final product manufacturing phase, reproducibility, consistency, standardization, and principles of quality by design, including defined critical quality attributes, are essential. Panelists discussed addressing variability in the manufacturing setting in order to meet quality standards. Carl Burke from biotherapeutics development at Johnson & Johnson discussed the assessment and mitigation of the range of variability associated with input materials. Michelle Myers, a senior director in cell process development at GlaxoSmithKline, described how elements of quality by design can be used to manage variability in the manufacturing setting. Erik Finger, an assistant professor in the Department of Surgery at the University of Minnesota, continued the discussion about cryopreservation with a focus on vitrification as a strategy to reduce variability in regenerative medicine.
A brief overview of transformative drug discovery technologies was offered by Burke, starting with monoclonal antibodies in the 1970s, gene therapy in the 1980s, and cell therapies in the 1990s. Traditionally there has been a 30-year cycle of invention, he said, from building intellectual property to establishing companies, establishing proof of concept, and building manufacturing capacity, ultimately ending with the provision of stable products. Although there have been several recent licensures of cellular therapies, he said, the field in general is still in the manufacturing building stage and remains a long way from stable products.
The field is still in a learning stage, Burke said, and there are many challenges and potential sources of variability in developing cell and gene therapy technologies. These include, for example, analytical challenges (e.g., assay variability, assay controls, sample handling), sample and data management challenges (chain of custody, integrated databases), and process challenges (source cells, raw materials, transduction efficiency, cryogenic handling, process controls). There has been a great deal of development in, for example, the area of autologous CAR T cell production (see Chapter 4), but there is still much to be learned, he said. It had been suggested earlier in the workshop that “the process is the product.” Depending on one’s perspective, this statement can be somewhat controversial, Burke said. For an allogeneic product it is important to have a stable cell line as the input material as well as a reproducible process. For an autologous product the input material is variable, and it may be necessary to have an adaptable process (within a defined process envelope) to be able to achieve the desired product.
Sources of Variability of Input Materials
Assessing the variability of input materials is complicated by a variety of issues, Burke said. First, there is usually a limited quantity of different lots of material to compare, and small sample size and a small number of variables affect statistical certainty. Second, standards and platforms have not yet been developed. Lastly, the early development stage consists of many process changes and different sources of input materials, which also complicate the assessment.
Apheresis material for autologous cell therapy is subject to variability related to its limited quantity (i.e., product yield), the differences between patients (individual attributes, health status, treatment history), and the differences across apheresis facilities (equipment, collection processes, freezing techniques), Burke said.
In addition to the cell source, other input materials are needed for product manufacturing; these are often referred to as raw materials or ancillary materials. These materials are also subject to variability. New materials are often needed for the manufacture of an evolving technology such as regenerative cell therapies. Because these are new materials, there is limited experience using them in manufacturing, he said, and as demand for these new materials increases due to the scale-up of regenerative medicine products industry-wide, it is often necessary to use multiple sources, which can lead to inconsistencies. In addition, he said, inconsistencies can also emerge as vendors scale up to meet demand—for example, when transitioning from the use of research grade materials to GMP materials, which is why the use of research materials is best avoided, if possible.
When selecting raw materials, Burke said, manufacturers seek media, process components, and excipients (inactive ingredients involved in the formulation of pharmaceutical products) that are high purity, multi-compendial grade, and produced under GMP and avoid any reagent-grade or non-compendial grade materials. The grades and attributes of raw materials can be confusing, Burke said, because there are multiple compendia that can apply. The “lure of non-GMP excipients,” he said, especially at the start of a project, is that they are readily available and cost less than the GMP materials. But the ramifications of using these non-GMP materials becomes apparent when product development advances to later stages and to scale-up. There is often inadequate documentation of the source of the supplier’s raw materials, he said, and some do not know the actual source. There is often insufficient change control (i.e., the manufacturer of a raw material might make process changes and not appreciate that these changes affect the users of the raw material). Non-GMP materials can often have batch-to-batch variability. In addition, incorporating non-GMP materials into products can ultimately impede investigations as well as corrective and preventative actions. In summary, GMP materials are the product of controlled equipment and manufacturing processes using defined and consistent procedures that are documented and tested. Using GMP raw materials makes it possible, Burke said, to understand the impact of critical material attributes on the regenerative medicine product’s critical quality attributes.
The field of regenerative medicine is novel with regard to both the supplier of input materials and the developer, Burke said. Many of the input materials can be custom items (e.g., media, custom-made connections and tubing), and it can take several lots of product to begin to understand how the various components are performing. For some input materials the proprietary nature of components and processes can be a barrier to product development. When the specific attributes of raw materials are unknown to the product developers, it can hamper the ability to understand and control the impact of certain input components on process and performance. This
lack of knowledge also limits investigations, Burke said, as it is not known what changes might have been made to the components of certain proprietary input materials.
Controlling the Variability of Raw Materials
One approach to controlling the variability of raw materials is to establish a close relationship with the vendor, Burke said. He suggested working in partnership with the vendor to ensure that the vendor’s incoming source material is appropriately screened, that there is appropriate testing and process control, that the vendor provides notification of process changes, and that the vendor conducts release testing to ensure the quality of batches. It is also important that researchers establish incoming testing requirements and remediate certain input components in the product process if needed (e.g., supplementation if certain components are subject to degradation).
There are many raw materials involved in the development of cell therapies, Burke noted, listing several of the most critical components, including cell culture and cryopreservation media, transduction materials (e.g., plasmids), activation and isolation materials (e.g., coated beads), and consumables (e.g., flasks, single-use kits), and he said these can often be custom materials or acquired from a single or sole supplier.1 It is important to establish strategic collaborations with suppliers of unique components, Burke reiterated, including sourcing and quality agreements. When making the transition from an early clinical (i.e., research) setting to a late development and manufacturing (i.e., commercial) setting, it can be a challenge to acquire the necessary quantities of pharmacy-grade materials, he said.
As an example, Burke described the sourcing of raw materials for a phase III allogeneic cell therapy project. Half of the necessary materials were from a single vendor (either sole-sourced or single-sourced). This presents a risk for long-term manufacturing, he said, as the unavailability of any one of those vendors or a problem with one of the vendor’s input materials can hinder the development program.
For most medicinal products, Myers said, the paradigm has been to focus on whether the product works, and any concerns about whether
1 “Sole source” refers to the only available supplier of a material. “Single source” means that there is more than one source available, but others are not qualified or suitable.
it could be manufactured or supplied were secondary. For ex vivo gene therapies and autologous cell therapies the paradigm has been flipped, she said, and the focus is on making sure the product can be supplied to the patients who need it.
The question of whether a product can be supplied is not trivial. The processes used to supply small, early clinical trials are often academic in nature, Myers said, and they involve steps that are not easily scaled up or industrialized. Process changes are generally necessary if the product is to be supplied to a larger population of patients, and these changes are a source of variability. Even if the issues related to generating a larger supply were not a concern, process changes would still be inevitable, Myers said, as technology is constantly evolving. One result is that changes to the analytical methods are also necessary throughout development, she said. The variability that stems from process change makes these complex products even more complex, Myers said.
The goal, she said, is to “implement process and analytical improvements that will provide greater process robustness and enhanced product quality while minimizing the impact of changes.” To achieve this, Myers proposed using elements of quality by design, which she defined as “a systematic approach to development that is based on sound science and risk management.” It includes predefined objectives and focuses on product and process understanding and process control.
Quality by Design
Quality by design begins with defining the critical quality attributes of a product; that is, those attributes with the potential to affect product safety or efficacy, Myers explained. These critical quality attributes are the foundation for control strategies to manage product quality across the product life cycle, she said. Critical quality attributes are not analytical methods or specifications, but they should be based on robust data where possible, she said. The control strategy for a given attribute might or might not include the need to test for that attribute on product release.
As an example, Myers showed a plot of a critical quality attribute as a function of a clinical outcome for a rare disease product. She described how, in a team that includes both clinicians and statisticians, it is possible to “map” clinical endpoints as a function of product characteristics. Embracing the variability across the clinical data to better understand the process and the product requires handling large volumes of data, she said. This requires an information technology infrastructure that can manage big data as well as collaboration across the clinical team and the chemistry, manufacturing, and controls team, which, she noted, may not typically work closely together.
Once the critical quality attributes have been identified, it is important to ensure the consistency of the analytical methods used to assess them, Myers said. To do this, she suggested using a systematic approach, based on science and risk management, to identify the sources of variability and to implement controls early in the development process. Each analytical method can be thought of as a separate process, she said, and it is important to understand how the equipment, materials, methods, or analysts might contribute to variability of the assay. It is important to consider sources of variability early in development, Myers said, as going back later to identify correlations is extremely difficult.
With the critical quality attributes established, proposed process changes can be systematically evaluated using a risk-based approach that examines their potential impact on the product. Myers described several examples of typical process changes that might be considered when scaling up to treat larger patient populations and improve the supply chain robustness. These included changes to the vector production process, such as switching from an adherent culture system to a suspension culture system; changes in the cell manufacturing process, moving from manual cell manipulation to implementing automation; and changing the final product formulation from fresh cells to a cryopreserved product. Each of these proposed changes should be systematically assessed as a function of the established critical quality attributes, she reiterated. The assessment calculates the potential risk that an attribute would be affected as a result of the process change. For example, changing the vector production process as described above is assessed to present a high risk of affecting potency attributes, including the infectious titer and the infectivity, but there is determined to be less risk of impact to the safety attribute of residual plasmid DNA, and there is minimal concern about impact on some of the other safety attributes of the vector (e.g., mycoplasma, endotoxin, adventitious virus). Changes to the vector process must also be evaluated for any impact to the cell process, she added.
Systematically evaluating each proposed process change for its potential impact on a given critical quality attribute helps drive the development work and generate the data necessary to support a given change, Myers said. These risk assessments also provide the information needed to define the comparability strategy for evaluating the process pre- and post-change, to determine the assays required for product characterization, and to document the rationale for the process change. An ongoing challenge is securing enough donor material to be able to run side-by-side analyses of the old and revised processes, Myers said. She also said that there are other elements of quality by design (e.g., critical material attributes) that can be used to help control variability in manufacturing processes to the extent possible in order to deliver a consistently reliable product.
Advances in tissue cryopreservation can help achieve the goal of putting “the right tissue in the right patient at the right time,” Finger began. As a transplant surgeon, Finger said, he thought it would be ideal to be able to harvest an organ from a person, freeze it, and then transplant it at a time that was convenient to both the recipient and the surgical team. Unfortunately, this is not currently possible, he said, because organs have shelf lives. For example, a kidney for transplantation might last 24 to 36 hours under cold storage, while a heart might last only 4 to 6 hours. The challenge is to extend the time that organs and other tissues can survive prior to transplant.
At its simplest, Finger said, transplantation involves one designated donor, one organ that is collected and briefly stored, and one designated recipient to be treated. A more complex scenario would involve an organ bank, where multiple tissues are harvested from multiple different donors, held in cold storage at the organ bank, and then matched to the most appropriate recipient. On a larger scale, tissue might be collected from hundreds or thousands of donors, cryopreserved, and used as source material for the production of mesenchymal stem cells, or perhaps for high-throughput pharmacology testing.
There are several temperature ranges associated with tissue preservation, Finger said. Clinical organ preservation for transplantation usually involves a cooler whose temperature is kept between 0 and 4°C. Another strategy in development is normothermic machine perfusion, in which an organ is kept viable under near-normal physiological conditions by perfusion with warm oxygenated blood. This strategy can extend preservation by up to 24 hours, Finger said. Researchers are also studying the potential of storage at temperatures just below the freezing point to extend preservation by several days. This approach was inspired by species of Arctic fish and other hibernating animals that can survive in subzero temperatures. Arctic fish, for example, have adapted to survive in temperatures that are below the freezing point of their blood, Finger said. These fish produce an antifreeze protein that binds to ice crystals in their blood and inhibits ice crystal growth. Finger showed a brief video of Rana sylvatica, the wood frog, freezing and thawing in nature. This species retains urea and increases its glucose levels for use as cryoprotectants to prevent it from fully freezing in subzero temperatures. Ultimately, Finger said, the goal is to be able to cryopreserve and store organs for the long term, similar to the routine cryopreservation of human cells.
During conventional cryopreservation, ice crystals form in the solution outside the cells, Finger said. Solutes are excluded from the ice as it freezes, increasing the concentration of solutes in the remaining solution. In response to the osmotic gradient, water leaves the cells, and the cells begin to shrink. This results in increased concentrations of solutes inside the cell, Finger explained, and the increased concentration of salts inside the cell lowers its freezing point and helps to preserve the cell as ice crystals form. For successful conventional cryopreservation, he said, cooling needs to occur at a controlled rate of about 1°C per minute so that the cell has time to undergo osmotic shrinking.
There are limitations to conventional cryopreservation. In addition to the solution effects from the osmotic gradient, ice crystals that form during cryopreservation can damage cell membranes and intracellular structures, leading to cell damage or death and the release of immune mediators. Conventional cryopreservation only works for cells in suspension or for very small embryos, Finger said. In larger tissues, he explained, the ice formation disrupts the complex tissue architecture. Other approaches to cryopreserving tissue are needed that can avoid or manage ice formation, he said.
“It is thermodynamically unfavorable for water to undergo the phase transition from liquid to solid,” Finger said. When cooled slowly below 0°C, water can remain in a supercooled liquid state. As cooling continues past –10°C, the push toward the transition from liquid to solid increases, he said, and eventually ice will spontaneously form. However, Finger continued, if the water is cooled very, very rapidly, the viscosity of the solution will increase in parallel to a point where it is prohibitive for the molecular rearrangements necessary for ice crystal formation. Through this process of “vitrification,” water enters a stable glasslike state (i.e., becomes a noncrystalline amorphous solid).
Around 25 years ago, Finger said, Greg Fahy found that a rabbit kidney could be vitrified—that is, cooled to a glassy state without ice crystals—and theoretically remain stable indefinitely. The greater challenge, he said, is how to thaw and recover the tissue from the vitrified state. Ice crystals can form during re-warming if the tissue is not heated extremely rapidly, he said. Furthermore, the warming process must be uniform, or else the tissue can develop stress fractures and large cracks (potentially splitting the organ).
Tissue damage can be prevented, to an extent, by the addition of a cryoprotective agent. Finger listed four generations of cryoprotective agents (6M glycerol, DP6, VS55, and M22) and said that each solution has a spe-
cific critical warming rate and critical cooling rate (i.e., the rate of change in temperature that must be exceeded to avoid ice crystal formation). The newer-generation cryoprotective agents have much lower critical warming and cooling rates (e.g., a critical warming rate of 32,000°C/min for 6M glycerol versus 0.4°C/min for M22). For comparison, Finger said that water would need to be heated at a rate of 10,000,000°C/min to avoid ice crystal formation, which he described as practically impossible.
The rapid, uniform warming of vitrified tissue is accomplished by magnetic hyperthermia (Manuchehrabadi et al., 2017). In this process, the tissue is loaded with iron oxide nanoparticles and subjected to radiofrequency excitation in a magnetic field. As they oscillate, the nanoparticles generate heat. A complex coating is added to the iron oxide nanoparticles to ensure that they are biologically inert and to prevent clustering, Finger said. The first-generation nanowarming system is a small device that fits in a 15-milliliter vial and generates a uniform magnetic heat, he said. Small organs loaded with iron oxide nanoparticles can be recovered from a vitrified state by being added to the vial and heated in this way.
Summarizing the process, Finger said the critical steps for vitrification and rewarming of tissue include the following:
- Cannulating the organ for vascular perfusion
- Loading cryoprotective solutions in a stepwise fashion
- Loading nanoparticles
- Cooling the sample in a controlled-rate freezer (at a rate greater than the critical cooling rate)
- Transferring the cooled sample to liquid nitrogen storage
- Performing rapid heating via a radiofrequency generator (nanowarming)
- Undergoing a stepwise washout of the cryoprotective agent and nanoparticles
- Assessing the viability and function of tissue
Finger showed photographs, magnetic resonance images, and computed tomographic scans of a range of different organs as a way to illustrate the homogeneous magnetic particle distribution and successful vitrification and rewarming. The goal now, he said, is to scale up the nanowarming process from the current 80-milliliter system to a 1-liter system that could accommodate a human organ.
Applications of Vitrification and Nanowarming
Nanowarming techniques can now be applied to cell clusters, flat substances (e.g., skin), and organs, Finger said. This opens up new possi-
bilities for using vitrification for the cryopreservation of tissue engineering products, including, for example, high-throughput pharmacologic testing using thousands of tissue slices from one donor that have been banked and thawed as needed, and also the testing of donor variability in a large batch after hundreds of samples have been collected and stored.
Implementing Quality by Design Concepts and Controls as Early as Possible
Many technologies that are developed by industry originate in academic research laboratories, Roy observed, and he asked the speakers how early in the product development process the type of controls they had discussed should be implemented. Quality by design is not something that academicians generally think about, he said, and early intervention by industry could help the industry downstream move the product forward more quickly.
Early intervention is helpful, Burke said, suggesting that industry help by performing parallel research projects, while acknowledging that such an approach would compete for resources. Changes were made to the CAR T cell project very early in development, Burke said, and the development timeline was compressed, with clinical studies moving from phase I into phase IIb very quickly. There might be much that is unknown early on, Myers agreed, but she said that it is important to implement controls where possible, as early as possible, in the specific process that will be used for production. There are areas where control is more difficult, such as apheresis or other starting material, she said. Performing the assessments and identifying the critical raw materials and the critical parameters early on may be helpful in driving development, she said.
There is a common understanding that academia is focused on basic research, and development is industry’s job, Roy said, but that mentality needs to change. Development should be taken into account early on, he said, and he suggested that not doing so is delaying the translation of innovations to patients. In the case of CAR T cell products, Burke said, the research, development, supply chain, and manufacturing groups are situated physically close to each other, which creates a sense of integration.
Many startup companies work quickly to reach the phase I clinical trial stage so that they can be acquired by large companies, a workshop participant said. This type of pressure makes it difficult to create a quality by design process. How, the participant asked, can this rush to the clinic be prevented, or how might best practices be encouraged? One issue is that small startups and larger companies have different goals, Myers said. The
goal for GlaxoSmithKline is commercialization, while the goal for some of the smaller biotechnology companies is to be acquired. Education may help small companies and academic researchers understand the benefits of systematic science and a risk-based approach to quality, she said, as well as to understand that implementing process controls improves efficiency and the economics of the development process.
Products will be delayed in reaching patients if startup companies and academic researchers do not start taking the concepts of quality by design into consideration, Roy said. Taking an investigational product into human trials is exciting, he said, but it does not make a product. It is important to consider how to de-risk those early successes for industry so that products can reach the patient. Small companies often seek to acquire the new findings or techniques published by academia, Finger said, which can affect the way that academia conducts business. For example, it can affect how an academic laboratory protects confidential or potentially proprietary data and when the laboratory decides to release or publish it.
One suggestion offered by a workshop participant was to ask students at the Wharton School of Business at the University of Pennsylvania to conduct a study of the process improvements that happen when there is more integration across departments, such as the close integration mentioned by Burke among the departments working on CAR T cell therapy. The idea of engaging those with expertise in business to help study efficiencies in therapeutic product development is a good one, Myers said.
Small companies selling their technology sometimes convey very optimistic timelines which can, in certain cases, lead to delays, Siegel said. This is in part because of choices that result in, for example, the need to regenerate cell lines or redo animal studies. Siegel noted that large pharmaceutical companies have a due diligence process and evaluate the timelines, anticipated market size, and other factors and calculate the potential value of the product. However, companies with less experience might acquire the product and then realize that there are more hurdles and potential delays than they anticipated. The field is moving to a place where academic researchers should consider the types of controls that would help to facilitate more rapid development (e.g., knowing cell source characteristics such as how donors were screened, how cells were treated, antibiotics used), Siegel said. A small group within the Forum on Regenerative Medicine is developing a perspective paper to serve as a guide for academic researchers and biotechnology companies on product attributes that translate well to the clinic and to large-scale production, he continued. There are many actions that, if done correctly from the start, can protect value in the longer term, he said.
Fingerprinting Raw Materials for Quality Monitoring
The speakers were asked about ways to define the critical material attributes for raw materials and how to build a successful relationship with vendors. Raw material chemical fingerprinting is difficult in part because each company knows its own issues and failures with its raw materials, a participant noted, but it is very difficult to see broadly across the industry and to identify vendors that could be encouraged to improve their raw materials in order to secure the supply chain. Some of the hesitancy of companies to be more transparent revolves around failures, Burke said. Usually when an issue with a raw material is identified, it is related to a failure, and sometimes that failure is in a manufactured product. Companies generally want to solve the issue before providing any information. Burke said he was supportive of more openness and added that there had been some side discussion of, for example, registries.
Resources for Clinical and Commercial Product Translation
Taking a product from the laboratory bench to the clinic and commercialization has long been recognized as a challenge, Hubel said, noting that the path between academic laboratory success and a product’s commercialization is often referred to as “the valley of death.” She told workshop participants about the National Institutes of Health’s (NIH’s) Centers for Accelerated Innovations and the Research Evaluation and Commercialization Hubs programs, which are administered by the National Heart, Lung, and Blood Institute. Both programs are designed to help NIH-funded researchers translate their innovations to the clinic and to a commercial product. These centers and hubs serve as a nucleus where diverse teams can come together and share expertise and resources.2 Another workshop participant mentioned the work of the Standards Coordinating Body for Gene, Cell, and Regenerative Medicines and Cell-Based Drug Discovery, a nonprofit organization working to facilitate the development of standards.3 He encouraged other workshop participants to get involved and added that students can become involved as well.
This page intentionally left blank.