Driving Cell Therapy Innovation: Applying Key Lessons from the Evolution and Commercialization of Protein-Based Therapies

DSC_0400-300x157.jpg

(left to right) Panel host Anne Montgomery, Lee Buckler, Edwin Stone, Jon Rowley, Lynne Frick, and Phil Vanek (PHOTO BY LEAH ROSIN)

After many trials and errors — and milestones — regenerative medicine has become a mainstream part of the biopharmaceutical industry, supported by at least 670 companies and clinics of all sizes. But many experiences in the protein-based industry segment can be leveraged to further improve successful commercialization of advanced therapies. At the 2017 Biotech Week Boston conference, BioProcess International editor in chief Anne Montgomery hosted a panel of industry cell therapy experts to discuss key lessons that can be gleaned from the evolution of traditional bioprocessing to support further development of regenerative medicines, to review tools and techniques affecting the current development pathway, and to propose expectations of what may happen over the next decade. Panelists were Jon A. Rowley, founder and chief technology officer of RoosterBio; Lynne Frick, founder of Bioguides; Edwin Stone, cell therapy automation and technology lead at the Technology Partnership; Phil Vanek, general manager of cell therapy strategy at GE Healthcare; and Lee Buckler, president and CEO of RepliCel Life Sciences and a member of the BioProcess International editorial advisory board. All statements made here are those of the individuals and do not necessarily reflect those of the companies they represent.

Innovative Tools and Techniques 
Montgomery: Which tools are making a difference toward proving efficacy and reducing production costs, and which are truly necessary?

Stone: One example I like to use involves my company’s development of the Amgen erythropoietin (EPO) manufacturing line. This produced EPO at commercial volumes using 40,000 roller bottles. With autologous therapies, there can be a sense that we are always treading on new ground. But examples like this show that as an industry, we have conducted scale-out previously.

Scale-out often comes with a set of new challenges that typically you don’t have to think about if you are just increasing volume. Whereas previously we might have been dealing with systems of part counts in the thousands, the Amgen system had a part count in excess of a million. When you are processing at a capacity of a thousand therapies a year, which equates to about a hundred parallel systems, you’re talking about a great number of parts. And with those, you suddenly need to manage reliability problems that have not been seen before.

There was an interesting comment in a previous presentation [today], in which the presenter said that the only failure that his company has seen in the release of 100 therapies was a mechanical failure. That’s a reliability issue. We are going to have to start thinking about different approaches and different ways to design and think about manufacturing equipment for cell therapies. We are not going to be able to just over-engineer this equipment in the way that we have done before.

Montgomery: Can you point to tools that you feel will be especially necessary?

Stone: The lesson learned from the Amgen example is that you stop thinking that you are making a piece of bioprocessing equipment and instead start thinking more like you are making a medical device. You start to think about failure rates. If you have one-in-a-hundred or one-in-a-thousand failure rates, you could work with those and make a diagnosis; those failures might even be something that you could predict. But for one in 10,000, 100,000, or maybe even a million failure rates, you’ll need a different engineering tool box. It is possible to achieve those failure rates. The medical device community does so. It needs to be able to make systems reliably with failure rates down to one in a million or at least one in 100,000. But you can’t entirely rely on quality control (QC) for those rates. You need to consider design tools and statistical design approaches to be able to predict those numbers long before you see them.

If there is one piece of advice that I would give, it is that the data are there already. The subject of big data has been mentioned a lot during this conference, and it will continue to be a major point of discussion in our industry. The data are available. The data points that are “one off,” which we often discard during process development, or those individual failures that we attribute to use errors might not truly be “one-off” points. We should be capturing those data. We should be recording them. Because when problems come up in production, those “one off” points will be pointing to those problems. That’s instead of seeming surprised, thinking that we didn’t know about those problems when we might have known about them years before.

A previous speaker at this conference talked about allogeneic cell therapy development and production. He suggested that manufacturers are going to have to move much more quickly into commercialization. The fact that some regulatory barriers are dropping means that we can commercialize faster. We will need to rely a lot more on data we collected in early development. We need to be a lot more aware of all individual issues that have been happening, bring those data together, and be cognizant of potential problems much earlier.

Buckler: I like the analogy to medical devices. If autologous cell therapies are going to be successful commercially at treating relatively large numbers of indications, then their delivery will be much more akin to what occurs at the point of care with medical devices. That’s true at least for many autologous cell therapies today. It will need to be much more decentralized, much more regional to point of care than the centralized manufacturing model that we’ve seen in the past.

Stone: I agree. Even with the centralized approach, the same problems might exist. I think that even if we stay centralized, the number of bits of equipment that bioprocessing needs makes those processes look a lot more like those of medical devices.

DSC_0404-300x227.jpg

(PHOTO BY LEAH ROSIN)

Defining Key Metrics
Montgomery: As you are assessing some of these similarities, how do you define key metrics for cell therapy bioprocessing, and how do they accelerate innovation?

Buckler: Two needs in the industry right now are for advancements in surrogate biomarkers and for clinical designs that feed into bioprocess improvements as the industry progresses. Feedback loops from bioprocess systems themselves are integral and important, but we are only beginning to tap into them. There also is a need to feed the clinical endpoints and the clinical data that we see — big data, small data, and even observational data.

Rowley: I always focus on the factors that affect cost of goods (CoG), assuming of course that safety and efficacy are established. Once those are figured out, a major issue is the ability to manufacture consistently and within a particular cost structure. A product manufacturer is limited by reimbursements for its drug products, and profitable businesses will need to be built in this industry. So metrics are important.

The best thing that happened to me in my cell therapy career occurred when I got a job at Lonza back in 2008 heading up its new cell therapy R&D division. While at Lonza, I visited its different protein-production sites trying to understand the challenges associated with the manufacture of protein-based therapies. I realized that cell therapy development and manufacturing would go through very similar stages. I attended the BioProcess International conference for the first time in 2008 and listened to the a number of presentations and discussions on challenges. There I realized that cell therapy manufacturing would have these same challenges, and the goal was to try to address them proactively.

I noticed that all biomanufacturers had a consistent metric, which was protein production in grams per liter. Cell therapy has not had a metric similar to this. If you look at this productivity metric over the life of the industry, it follows a very similar trajectory to that of the semiconductor industry, which is essentially Moore’s law — log improvements over time. The semiconductor industry did a number of computations per cm2 of chip per second, which has consistently decreased — not linearly but on a logarithmic scale. Once biomanufacturers started charting grams per liter for monoclonal antibodies, they reached first tens of milligrams per liter, then hundreds of milligrams per liter by the early 1990s. By the 2000s, manufacturers had reached a gram per liter, and at 2010 it was >10 g/L. This rate of biotechnology improvement is the same as the rate the semiconductor industry had. The same log improvements are now happening in stem cell production, at least for mesenchymal stem cells (MSCs), which is the part of the field that I am in, but I am sure for other cell types as well.

One lesson that we need to learn from protein-based biotherapeutics is the importance of having a consistent metric (grams per liter in this case). At RoosterBio, we like to track the productivity metric of millions of cells produced per liter of culture media consumed. However, no one is using that metric yet. It showed up for the first time this year in the regenerative medicine glossary. MSCs are going to have a different number of million cells produced per liter of media than are T cells, but they will be within similar ranges. These metrics are not just technology metrics. Like Moore’s law, they are solid economic metrics as well.

Media are common tools of all cell therapies and are always the biggest cost drivers. We need to understand media productivity. Many conference presentations and discussions focus on x-fold expansion, or maybe cells per liter, but they don’t talk about the amount of media consumed during their processes to get there. Getting productivity metrics that have both a technical and economic basis is going to be key. Such metrics will help you compare apples to oranges — different processes with one another. There always will be differences, but if we are not talking about both economic and technology metrics, then the field isn’t moving forward, and we aren’t learning from the protein-therapy world.

Vanek: I tend to agree with Gary Hamel (London Business School), who believes that if you can’t measure something, you can’t manage it. I would add that such metrics must be normalized around potency in some way. Bulking up cells and making more cells is a great metric as a starting point, it is a production mindset. Those cells eventually might become more potent, and dosages could decrease. So cost would decrease as a function of production but also as a function of potency. We have to weave that into the metric.

Rowley: I completely agree. Earlier I mentioned safety and efficacy and potency. Those are table stakes. Once you get a highly potent cell, then you still need to be able to manufacture it. That is where you get the log improvements over decades, which is going to lead to — maybe in the year 2050 — regenerative medicines being something that you can buy at CVS for a hundred dollars.

Frick: As a non–cell-therapy person who is squarely from the bioprocess space, I am perhaps more practical in my comparison regarding media development discussed previously. I think back to the early days of protein therapies — how the industry didn’t have animal-free, serum-free media and how long it took to characterize media or to have fully chemically defined media. The cell therapy field needs that type of work. It is going to happen much faster than it did for protein therapies, but there is still a lot of variability.

Another issue is stratification of patient response and the tools available to think about it in a different way. One company I previously worked with, IsoPlexis, has a single-cell secreted protein assay, which captures highly multiplexed secreted proteins across thousands of single cells. That technology shows promise in predicting actual patient response, and on the company website there are references to that effect. That perhaps could help us obtain additional information on immunotoxicity prediction and perhaps on characterizing these new therapies during manufacturing. It took years in bioprocessing to come up with a way to characterize the manufacturing process. Of course, the process was the product, and everyone just ran the same process. It was an evolution to get to excellent product characterization. How do we do that more quickly this time?

Buckler: There is an underlying assumption that “more and cheaper” is always better. It might be that we ultimately are looking at a subset of cells that have a certain functional profile tied to clinical efficacy. In that case, the end goal is not having more or producing it less expensively, but rather defining a subset, fraction, or population of cells that are tied to efficacy. With that subset of cells, we are handling a much smaller, but more effective, population of cells.

Rowley: I’ll stick to my guns here: Let the scientists figure out the exact cell, then figure out how to make it consistently at a larger scale and at a lower cost. Once the science is figured out, then the engineers can come in and make it. Once you figure out the exact cell (why we don’t know that is beyond me, but I am an engineer. I’m not trying to fix the biology, but rather the bioproduction side), then you have to make it efficiently. We can’t compare anything at cell therapy conferences yet because we are not focused on those productivity metrics.

Reducing Risk
Montgomery: That feeds into current industry perspectives on risk management. What is the priority at this stage: cost or risk?

Vanek: There is a lot of talk about cost of goods and the reimbursement of the new Novartis cancer drug Kymriah at $475,000. I heard a comment last week at a conference in Berlin about the first reaction to that cost. Although approval of the drug was hailed as great news for the industry, the first reaction from a hospital was, “A drug for $475,000 is crazy!” So we have to temper our enthusiasm a little bit, but obviously there is durability and value of these therapies over a long period.

If you think about what is happening in the industry, there is increasing complexity and a convergence of clinical and production pathways. Autologous cell therapies are more precise and personalized. The whole ecosystem is changing on us. If we start today at the infancy of this industry focusing on cost, I think that it might drive some bad behaviors in how we select and make trade-offs in production methods, in geographies where we produce, and perhaps even in the types of technologies that we adopt. By evaluating manufacturing, distribution chains, and supply chains and building all of those from a risk perspective, consequently our costs will decrease. I agree with what Jon (Rowley) said: You build these capabilities by focusing on derisking, closing systems and connecting systems, and adding a digital layer. These things will come, but I still think that it is too early to focus just on cost. I prefer to focus on risk.

Buckler: I think that we should add another “versus” to this question. Our consideration should be cost versus risk versus benefit. We are now seeing some phenomenal benefits, which we have to assume that engineers will help us solve, and the cost related to that. The risk seems manageable, the benefits seem phenomenal, and the cost seems engineerable. We were talking before this panel about how manufacturers have to plan for success. What is only barely imaginable now we must assume will be possible in a decade from now and a commodity a decade later.

Frick: One other comment on risk: At a meeting with a cell therapy company, a discussion came up about predicting patient response: “What if we can predict whether a therapy will work well, cause CRS, or maybe work only marginally well? We have already engineered it. The product is made. Do we administer it to patients?” This is a real dilemma. One person commented, “Well, if we know that it won’t work, then we are not going to do it.” However, the clinical oncologist in the group said, “Are you kidding? Suppose the patient is my father. Am I not going to try the drug product? Am I going to tell a patient that I am not going to treat him or her after I have made the drug?”

When we think about risk for a cell therapy, we need to do so differently than with a traditional bioprocess. If you are engineering a product autologously, and you have it ready to be made, you need to think about all the possibilities and what you are going to do in the clinic with the patient. It’s a very interesting topic.

Stone: Risk has these two dimensions: patient safety and commercial risk. And the two are not necessarily aligned. That is where traditional bioprocesses and cell therapy processes have some differences. For example, losing a bioprocess batch might result in a stock out or in a big commercial issue, but would it result in patient death? Maybe not so. It looks like we are getting better at having reserve batches and the ability potentially to freeze two batches. But losing an individual patient batch is a very serious concern.

Rowley: My last comment is that cost isn’t the first issue. I always say that quality is number one. But right behind quality is cost. As technologists, it is our job to ensure sure that we can manufacture cost effectively. For the companies that we work for, we are focused also on the safety and potency of our drug products so that once they get to market, we have a business model.

I am influenced by a very early biotechnology writer, Cynthia Robbins-Roth, I read a book of hers back in graduate school called From Alchemy to IPO; The Business of Biotechnology. She wrote that developing drugs that are too expensive to manufacture may never see a commercial end point. This was in the late 1990s. Then Gary Pisano from the Harvard business school wrote The Development Factory: Unlocking the Potential of Process Innovation and said the same thing. So yes, quality is always number one. A drug product has to be safe, it has to be potent, and it has to be effective. So many people are focusing on making the most potent cell, but if you can’t make enough to treat a patient, or enough to treat tens of thousands of patients, do you really have something that is going to have a lasting impact? It just might die on the vine.

When I talk, probably you hear only mention of cost as I tend to gloss over other important things. Yes product needs to be highly potent and extremely safe with closed systems for sure. But then people ignore cost. That has been happening for way too long, and that is why I get up on this soapbox. I firmly believe that when the science has been done and presented at scientific meetings, that is where you talk about potency and efficacy. But we are at the BioProcess International conference, so lets talk about bioproduction.

Anne Montgomery is editor in chief of BioProcess International; Lee Buckler is president and CEO of RepliCel Life Sciences; Jon Rowley is founder and chief technology officer of RoosterBio; Lynne Frick is founder of Bioguides; Edwin Stone is cell therapy automation and technology lead at The Technology Partnership; and Phil Vanek is general manager of cell therapy strategy at GE Healthcare.

You May Also Like