We’ll discuss the critical need to scale manufacturing, even in the early stages of development, to ensure that these therapies reach the patients who need them.
Guests:
- Emma Chan, Director of Technical Development at Orchard Therapeutics
- Matt Hewitt, Chief Technical Officer at the Manufacturing Business Division at Charles River Laboratories
Setting the scene: scaling cell and gene therapies
[00:00:05] Stuart Lowe The cell and gene therapy industry is rightly proud of its achievements to date. New approvals are being announced almost monthly and treatments for indications with much larger patient populations are on the horizon. Though with high price tags and cumbersome manufacturing methods, are we doing enough to ensure that the benefits of these therapies are going to be accessible to all those in need? In order to bring down costs and streamline manufacturing, where should we turn? Traditional methods of processed development rely heavily on legacy equipment and labor intensive handling steps. Decisions that are taken early on in a therapy's development can have a dramatic effect on the ability to scale manufacturing, and this can result in considerable time and expense in deploying the processes in production. So today we ask, can new ways of thinking about process development allow cell and gene therapies to scale to meet patient demand? Join me, Stuart Lowe, as we uncover the future of cell and gene therapy. Hello and welcome back to Invent: Life Sciences, two part series on accelerating the development of cell and gene therapies. In the last episode, we reflected on the progress to date in making cell and gene therapies a viable treatment option and the importance of developing with the patient in mind. Ultimately, this means we need to think about manufacturing scale even at early stages. If we don't, it threatens our ability to deliver the doses needed to meet future demand. In this episode, I'm joined by two experts, Emma Chan, director of technical development at Orchard Therapeutics, and Matt Hewitt, chief technical officer at the Manufacturing Business Division at Charles River Laboratories. Emma and Matt, join me to help decode my next question. How? How can we design with the end in mind and ensure broad patient access? How can process developers benefit from technology innovation? And how do we continue to move forward? I wanted to start this conversation with Emma.
Introducing the guests
[00:02:38] Emma Chan My name's Emma Chan and I'm currently working as a director of technical development at Orchard Therapeutics. So I'm leading a team that's primarily focused on optimizing our HSC hematopoietic stem cell gene therapy platform. I guess not necessarily this subtype. My most of my background has been in cell and gene therapies, but much more T-cell focused. So I did a PhD in post-doc at UCL at the Institute of Child Health, which is a really important center for for cell and gene therapies. And I have a lot of active clinical trials and lots of really great research. And so I was working on anti-HIV therapies there using plenty viruses to modify modified T-cells, but it was all kind of very preclinical proof of concept research. And then after several years there, I moved to Autolus, which at the time was a small spin-out company from UCL, and working on them, working on CAR T-cells. So that was kind of a big step for me and I learned a lot there because I was there from some some of the very early days before we were treating any patients and did a lot of the early development work to reach that kind of phase one trials.
[00:03:50] Stuart Lowe I wondered what sort of decisions Emma had to make during process development to give programs the best chance of success in clinical trials.
Early process development decisions
[00:04:03] Emma Chan I think it's important yet to sort of to map out realistically know how long your phase one, two trial will be, how long your pivotal trial will be, when you're likely to to file how many patients you'll be treating along that that time to make sure you get enough data and that you can, and if you do want to change processes that you plan your comparability study enough to make sure you get the the development data and the clinical data, maybe pre-clinical animal studies as well, if you need to, to make sure that you have enough data to support.
[00:04:31] Stuart Lowe If you are talking to a company who's who's working with a particularly novel cell type, there's not not very good equipment out there at the moment. What should they consider first when they're looking at developing their process?
Fit for purpose manufacturing at early stages
[00:04:45] Emma Chan I mean, I think even if there's not the ideal equipment, I think that's that realistically, there's probably enough options that you can get a decent system. And if it's not a fully automated perfect process, I think that's fine. I think realistically, first, in human phase one, two trials are probably not going to be the final commercial process that people will be using. So I think it's it's having something that you can develop a decent process so that you've got a good enough product to be able to treat some patients and get some get some data. And it's like we were saying, before, even if clinically it doesn't work that way, you you still gain some useful knowledge in terms of the manufacturing process and what whatever equipment you're using is, is capable of. So I think that's kind of having a, I guess, like a fit for purpose manufacturing process.
[00:05:34] Stuart Lowe Yeah. For your kind of your stage as well.
[00:05:37] Emma Chan Yes. Yeah. And I mean if, if it has some not ideal manual processes, I think it's still okay because chances are you're maybe treating a dozen patients so you can get away with, you know, having a few people in the lab sign in isolator doing [00:05:53]pipetting [0.0s] and things like this. If you're if you're treating hundreds of patients, it's it's going to be difficult to be feasible like that. But I think early on, I think it's important to just have a an appropriate process that you can start your clinical trial and then and then you'll understand as well what bits of the process are not performing as well as you like. And then you can look for some new technologies that are more suitable and which kind of increase the robustness and reliability of the process.
[00:06:20] Stuart Lowe There's not the ideal time to to look at different options than around the kind of phase one. Once you kind of start to think, actually, if I've got a bigger trial or bigger patient population, I'm going to have to make changes anyway.
[00:06:33] Emma Chan Yeah. [00:06:34]Energy [0.0s] cell and gene therapy, it's usually a phase one slash two trial. You can probably have I think it's a bit more having a process that that delivers a good product that should be safe and efficacious, and then you can see how it works clinically. I think ideally before you then start your pivotal trial, you don't really want to implement any process changes or any significant process changes during the pivotal trial. I mean, maybe I'm just biased because I work in process development, but I think one of the most exciting things I've seen from when I started in the field, 14 years ago, the manufacturing process, cell therapies were all very manual. So you basically had to have a tissue culture back and you'd have to go into an isolator. It would be very tedious. You have to be very well trained to to do these kind of fiddly procedures in an isolator. When I was at UCL and then when I started Autolus, we were probably some of the first people to get the Miltenyi Prodigy, which is obviously like the kind of probably the most popular automated device for cell therapy manufacturing. And since then, yeah, I've done a lot of work by Autolus under Orchard on the prodigy. But since those early days when that was kind of the the main piece of equipment available, there's now a lot of different options coming out. And I think companies are catching up because obviously it takes a long time to develop these devices. But new things are coming out commercially available. And then also there's a lot of like beta test devices available where manufacturers are trying to work with companies to fully develop these these equipment. So with with all these new devices coming out, it kind of opens up the opportunity to develop new manufacturing processes, new have different cell types to be able to target more disease types. So it increases the flexibility and options for drug producers.
[00:08:19] Stuart Lowe With Emma's experience in therapy development and equipment selection, I was curious to find out where she would like to see more equipment being made available and asked her where she saw the gaps.
Gaps in equipment and infrastructure
[00:08:33] Emma Chan I guess, in terms of gaps in the market, yes, some of it is, some of it, I think is being consistent, consistently felt like companies are working to produce more devices with more capabilities. There's different equipment. So things like the Prodigy [00:08:46]and ice, [0.1s] because you can basically do everything on one piece of equipment, but depending on what your processess. You might prefer a modular option. So we use one piece of equipment to isolate your cells of interest. One, to do a gene modification, one device, to do an expansion and so on. I think if that's the case, then there is some there's some areas where it would it would be interesting and useful to be able to have sort of compatibility of different equipments if you're using them from different suppliers. So I think now maybe an individual device works well. And if you're working in a process development lab, it's nice. But then if you're if you're looking longer term and trying to expand manufacturing so that you're making hundreds or thousands of batches a year, then it's also the kind of infrastructure that supports that. So moving from one device to another, how you're tracking everything. So it's a bit more maybe the logistics to support these devices that now needs to take out to speed to be able to manufacture as much as we hypothetically would like to do.
[00:09:45] Stuart Lowe What do you end up having to do at the moment then if you're using well-tuned pieces of equipment, but then you need to transfer to another one?
Manual interventions and the need for integration
[00:09:51] Emma Chan So, I mean, at the moment it would all be very manual. So, you know, you would have operators just basically taking a bag of cells from one device, sealing off the bag, welding onto something else. But then with that, you you emphasized you need a lot of manual intervention, so it's more time consuming and costly because you need the operators doing this. And then there's also the kind of risk of the traceability every time you're taking cells off one equipment to another, you need to make sure all of the labeling and tracking is is appropriate. So often like now, I think there's more systems coming in for sort of automated track and trace and barcode labeling and scanning everything. So you can kind of get barcode to scan off one piece of equipment, then transfer it to the next and make sure everything, again, for this is for autologous, so obviously when the traceability is really crucial.
[00:10:41] Stuart Lowe Just thinking maybe one step backwards into more into kind of your domain on the process development side, is that about automation needs there as well?
Automation needs in process development
[00:10:50] Emma Chan I guess one of the the challenges we have in terms of automation with the way things are going with automated devices is having good scaled down models, because historically if you're culturing in a 500 mL bag, you could culture in a 10 mL bag and it's kind of a scale down. Whereas if you've got a bioreactor, the minimum volume is a liter. You know, it's not easy to get out of mini bioreactor to replicate this. So when you're doing kind of characterization studies or big development studies, you can't run 10 of these bioreactor side by side because the scale of it would be way too much. So I think actually in terms of specifically for development needs, that's that's something that is kind of a new challenge now based on these automated systems. I mean, there are some companies that are addressing this where they kind of they are developing their bioreactors in mind to have a mini PD suitable version and then that so then the data should be directly scalable so that, you know, if you're running now at 10 mL, it will scale up to your kind of manufacturing scale and the data is representative. I guess actually, then the other kind of similar one in the in the development space is then looking at analytical in terms of sort of automation is a nice one. If you can have maybe sort of robots again when you're doing large screening studies, if just for some of the laborious steps, things like, you know, just basic things like making up PCR mixes or doing flow cytometry, where you've got dozens of samples, and these could maybe either be used in development for screening studies or in manufacturing as you scale up the number of products you're making if you're making tens or hundreds of batches.
[00:12:33] Stuart Lowe Yes, of course. Then you've kind of got this massive backlog of of QC samples to process as well.
[00:12:37] Emma Chan Yes. Yeah. And it kind of gets to the point where if we have these really nice, streamlined automated manufacturing processes, processes is potentially the QC that is then going to become the, the right limiting step. So you end up with an enormous QC lab for a fairly small manufacturing facility. So these are kind of other things that need to to develop and improved to be able to support the, the larger productions.
[00:13:02] Stuart Lowe Have you've been involved in introducing any new assays into into some of your manufacturing, because I know with cell and gene therapies, a lot of the mechanism of action and potency are quite difficult to characterize. Is that something that you've been involved in?
The challenge of potency assays
[00:13:17] Emma Chan Not so much directly, but I know, like, yeah, within our team orchard and then just experience again and I think this is definitely one of the challenges of cell and gene therapy is to be able to get a good potency assay that really is reflective of how the drug will work in patients and be indicative of whether it's going to be efficacious or not. So I think from from my experience and and also discussions I've had with people in the industry, I think having a range of different potency assays as soon as you start treating patients is important. And even if there's not necessarily a regulatory requirement to have these assay set up. For phase one from experience, it is always good to have at least some level of potency assays there available so you can start collecting this data because this is the it's the most difficult data to get but is probably maybe the most crucial in terms of analytics so that you can gather data as you start treating patients. You can gather data from these various different assays to see how it correlates with the clinical efficacy that you see.
[00:14:14] Stuart Lowe What do you need to do to make sure they're kind of fit for purpose? As you go forward to manufacturing because something that works well at the pre-clinical stage might be quite hard to implement later on.
[00:14:25] Emma Chan Yeah, I think it's it's very, I mean, it's definitely always one of the challenges with cell and gene therapy. So some of the potency assays that I've worked on can be quite laborious. There's often a lot of maybe an extended cell culture, so they're not the easiest assyas to run in QC and you get a lot of biological variability, which is them one of the challenges and why you may need to develop a couple of different potency assays to be able to to tolerate that variability, especially when you're using making autologous products. Yeah, it's definitely a difficulty. So having some understanding of what the the kind of ranges that you would are likely to see in your products.
[00:15:08] Stuart Lowe Finally, I wanted to get her thoughts on the role of CDMOs in speeding up development. What other sorts of companies in this ecosystem do you find kind of useful when you're developing or when you're trying to speed things up?
The role of CDMOs in therapy development
[00:15:26] Emma Chan Something that I really noticed actually the last conference I went to, there were a lot of CDMOs. So I don't know if it maybe seems like the industry generalists almost stop trying to do their own manufacturing and think we'll just we'll just, you know, come up with a construct and disease show like mechanism of action develops and potency and then we'll transfer everything to a to a CDMO to do the actual manufacturing, because I think it's so difficult that you need so much infrastructure and support. It's not an easy thing to do. And I think the time and money it takes to set up a manufacturing facility, I think is sort of a small biotech would be reluctant to do that. So you kind of have the safety net if you just transfer things to a CDMO, it's it's a lot cheaper and quicker with less risk because you don't have to build your own facility.
[00:16:15] Stuart Lowe This is a different business model. Isn't or is different, you know, just at the time when you're doing difficult clinical trials which which are expensive in themselves, you're also being asked to invest in infrastructure for for for manufacturing at scale as well. It's quite a big ask.
[00:16:30] Emma Chan Yeah. And I think to have your manufacturing facility ready in time, you need to start investing before you're really sure if the clinical trials are working so well. So it's yeah, it's a really big, big risk in that sense. So I think the CDMO is a good option. And yeah, I mean some of the ones that I've spoken to just generally, you know, at various conferences and things that some will even have platforms. So you literally just give them your construct, they'll make you make you a they've got plasmid backbones, they can make you a lentivirus using their manufacturing process. They've got a CAR T process that you basically will just take your land and and plug into their system. So if you you could have quite a small biotech or even maybe an academic center that just comes up with the construct, it makes it relatively quick and easy to get into clinic so that you you don't need to do all of the process development side, which is, I guess, you know, resources and skills in itself. You could leverage it from another another company.
[00:17:28] Stuart Lowe And what would you say is a good aspiration for maybe the next 10 to 15 years? Do you think we can get to much wider patient populations? And maybe what's the kind of one or two things that we have to do better in order to, you know, together?
Looking ahead to future innovations
[00:17:44] Emma Chan I mean, I think it's definitely things are definitely improving. And I think a lot of the things we talked about so things that platform manufacturing processes, having infrastructure CDMO to make things more accessible to companies is important. Again, the regulators being willing to work and change their perception to suit specific cell and gene is progressive and it takes time, but it is progressing. I think kind of new innovations. The one everyone is always talking about is AI technology and how that can be used. And I mean, I've discussed with various people how potentially it would be really nice, particularly in the autologous field, how you could use AI to predict from your starting material for this patient how the cells will grow. Maybe you can modify your manufacturing process specific for the patient. Maybe in 10 to 15 years, it's possible. I think it's, you know, it's a long way off, but it's something that I think is very exciting.
[00:18:41] Stuart Lowe It was a great conversation with Emma. Here are my key takeaways from our discussion. Planning for scale is key. Understanding your manufacturing needs early on and developing your process accordingly will save you time and effort down the line. Automation on its own is not sufficient. The infrastructure to support product handling, tracking and QC is essential to avoid creating new process bottlenecks. CDMOs have an important role to play. CDMOs bring facilities, personnel and experience to the table, giving therapy developers the option to outsource manufacturing. Someone who spent his career developing technology to support the cell and gene therapy industry is Matt Hewitt, and I was interested to hear his viewpoint on the partnerships needed to advance this space.
Introducing Matt Hewitt
[00:19:40] Matt Hewitt My name is Matt Hewitt, and I'm the chief technical officer for what is now called the Manufacturing Business division at Charles River Laboratories. So that encompasses several business lines for us. One is CDMO, which is plasmid vector and cell therapy manufacturing. Another business line is our biologics testing service, which we have four sites around the world, four at the moment. We have five sites for CDMO. And the last business line is, is microbial solutions, which does a lot of work in the microbial and sterility assessment space, and that has 13 sites worldwide. I have been with Charles River for just three years. I think about a week ago, it was my three year anniversary. I joined just after some acquisitions in the CDMO space that brought those capabilities in the Charles River. Prior to that, I had very similar position within a single business unit at Lonza, and one of my higher profile responsibilities in that role was to bring the Cocoon manufacturing technology to the market. Most are familiar with, and has become the basis for Galapagos decentralized manufacturing network that they're putting together for their therapies.
[00:20:56] Stuart Lowe There's a great kind of like almost potted history of the of the industry as well, and the organizations you've worked with and the therapies that you've worked on. One question is how have you seen the field develop in the past, say, 15 years, and do you think we are doing enough to get these therapies to patients?
[00:21:18] Matt Hewitt Yeah, I mean, I think it's a fantastic question. So I think we have made a lot of strides in the last certainly seven years since the Kymriah approval. Certainly we've had other, you know, so in gene therapies that were approved prior to, you know, Kymriah and Luxturna, like Glybera, like, you know, Provenge and not discounting their contributions to the space because they have been quite significant. Everybody is typically keyed on those, you know, the approval of Kymriah, Luxturna as kind of the the opening of the next phase of cell and gene therapy or advanced therapy.
[00:21:54] Stuart Lowe Yeah. So so there's a couple of things I want to pick up on there. So the first is a really don't want to hold back clinical progress. There's a kind of a speed element to this. And then the second one is, is designing with the with the end in mind. Now, McKinsey, we talked about McKinsey earlier. They project the drug development costs have risen 140 percent over the past 10 years. And they think that there are opportunities to bring medicines to market at 25 percent cheaper cost and 500 days faster. Now, I think this is a kind of a general over the whole of the pharmaceutical industry. But do you think that applies to cell and gene? Do you think we are doing enough? You've given two examples, but is there more we could be doing to make that kind of prediction a reality?
Fast test, fast fail: accelerating timelines
[00:22:40] Matt Hewitt Yeah. So I think in the latest McKinsey report, they're talking, you know, two plus billion for development, including all the ancillary programs which may have failed. I mean, I think they're talking in the realm of like seven to eleven years for gaining approval. If you look at, you know, they issued a similar report so a couple of years ago on cell and gene therapy or advanced therapeutics in general. And they were actually showing that those numbers are a fraction of the traditional therapeutic numbers. And I would also say that what they also showed there is that your actual success rates do rise pretty significantly, you know, from that. And I'll just mention that as well, because they did put out another report more recently on AI drug development, which everybody is quite interested in, and they were showing some, it looks like, advantages of using AI to help drug drug development through. And they have enough data to suggest it's beneficial to phase one clinical not seeing a difference as yet in phase two. But they're they're chalking that up to not about data yet, which is probably certainly the case. But when it comes to advanced therapeutics, we are seeing calculative success rates on various reports up to 20 percent. And then we're also seeing the timeline to approval in some cases, depending on modality, depending on your indication that you're addressing in the three to five year range. So you're talking about very, very fast timelines. And I think what we're really getting to here and when you couple this with certain regulatory landscapes and there are some that are better than others for this. What we're really trying to set up here is a fast test, fast fail scenario. Right. So you want to get to a position where you can get to a go no go point as quickly and cost effectively as possible. And that's really what we're working to to achieve with our partners and clients. We have said there are various other regulatory landscapes I've worked in where this is, you know, where people see. You know, I would say Israel is certainly one area where that has been mentioned. And I have worked in Israel on a couple of cell therapy trials. We also have worked with, you know, TGA down, down in Australia. They have quite a streamlined system as well. I think FDA is certainly getting there with some of the work speed programs that they've now opened up. We're working on, you know, we have the bespoke gene therapy consortium here in the US with FDA. They're now working on the equivalent for cell therapy. And, you know, I do sit on that committee as well working to that. So so we are getting there. But that's the point. And we should be cascading this out as much as possible so we can get to a fast test, fast fail because the no's are just as important as the yeses, because then you know that I mean, then essentially you're not having good money chase after bad.
[00:25:35] Stuart Lowe That's definitely true. You talked about that kind of it's always amortizing the no's if you're not going false positives going through. That's also a really important thing, too.
[00:25:45] Matt Hewitt Yes.
[00:25:46] Stuart Lowe So just thinking about the manufacturing side now, so there's there's an analytics piece, but how how can we support cell and gene therapy developers more in thinking about scale? Because, you know, you and I can agree that it's very important. But how how do those conversations go down? Because process developers just want to develop and they want to kind of get to the first clinical trial. But how can we kind of instill that kind of discipline to think within the mind?
The supply challenge and scaling bottlenecks
[00:26:16] Matt Hewitt So maybe I'll start out with some marketing metrics to give a sense of what we're doing right now. It's very illuminating. Okay. So if we look at the number of commercial batches which were produced in 2023 across all the approved cell therapies, it's around 9,200 batches. Okay. If we look at the second line approval of [00:26:41]CureVac, [0.0s] which is primarily here in the US, that patient population per year is 80,000. Okay. Now you get a sense of what's going on here. Christi Shaw, who is the former CEO of Kite, did an interview with McKinsey and then about 18 months ago and in that interview, she talked about that five years. At that time, it was about five, six years post approval for that. They are still really addressing about two to three out of every ten patients which are eligible for their commercial cell therapies. There was another survey that was conducted that became an ASCO Abstracts in 2022 or 2023. They surveyed 17 U.S. clinical centers, and they found that out of all of the patients that were eligible for commercial parties, only 25 percent actually received one. And that 25 percent that did receive one, the weighted on average or the median wait time was six months. Another 25 percent went on to enroll in a different 30 clinical trial and the remaining 50 went into a different clinical trial or unfortunately, went to the hospice and passed. So we're talking about still massive unmet need. And IQVIA released a report late last year looking at the market and how how it's evolving. And they had a graph that was very telling on commercial cell therapy spend and gene therapy spend. The gene therapy spend is dominated primarily [00:28:19]Zolgensma. [0.0s] It is taking up, but it's a bit slower. If you look at the cell therapy commercial spend, however, it is a very impressive slope in a positive direction. All these things kind of combine to point to is, is that we are it's a 90 to 95 percent probability. We are not limited by the patient adoption in the space of these therapies.
[00:28:41] Stuart Lowe No. It's supply, isn't it.
[00:28:43] Matt Hewitt Its supply, we are limited by the sheer number of batches we can produce per year. So we have to start thinking about how we're going to scale these processes more effectively. Certainly that's going to be through close process. I think everybody is agreeing with that. And I the time I guess one of my taglines is, you know, nobody disagrees that we need more automation. The actual conversation is when do we implement it? Because a lot of these phase one are these pre-commercial biotechs or biopharmas, they are, they have very high pressures and tight timelines based on funding. They get things in the clinic and test all patients. So they might take a somewhat inefficient process that might be open, might be manual, and then they do a two step. First step is just get in the clinic, see if this works on patients. If it does, then let's do a parallel process, analytical development project, optimize everything and close the process. And and if you look at the surveys on this and Serbia's done this and I've done this in the past, when I was at Lonza, there is a plurality of of developers which are waiting to enter phase one or even in phase two to start thinking about automation. And that's the exact reason. They say, look, we want to test this to make sure it works before we spend the time and money to to close and automate the process. I think in general, though, we are seeing acceleration of development of systems that will enable us. So if we look at in the last just three months and ISCT may there are three systems alone that launched, the Ori Biotech IRO, the Excel Foundry and then Cytiva expansion. And this is not including certainly Cellular Origins which is solely [00:30:34]TTP [0.0s] spin-out, Solaris, Multiply Labs. So you have six companies which are focused right there in addition to, Miltenyi, Lonza, Terumo, Cytiva and a couple of others that are that are, you know, we're talking almost, you know, upwards of a dozen plus companies just focused on how do we properly scale manufacturing. So we're making a lot of good progress. I don't think this is a surprising that we're now seeing this seven years post. I don't know if anybody was willing to commit to investing in these types of technologies until I think we have some more.
[00:31:13] Stuart Lowe Yeah. So there's kind of like a market confidence first and then you can kind of address individual therapies like, yeah, okay, these things can work.
[00:31:21] Matt Hewitt And there's a lot of our articles we have some handwringing around commercial success here. But I mean, I would remind everybody that, you know, in the traditional sense, Yescarta is already a blockbuster. So they did around 1.2 billion of revenue last year. We're probably going to see that only go up. You know, JNJ and Legend are saying that, correct is going to be a five billion dollar a year drug. I mean, there's are some significant numbers.
[00:31:45] Stuart Lowe So Matt seems to think we are turning a corner in terms of people's approach to scaling manufacturing. I wanted to ask Matt about something Emma had mentioned earlier about assay development. What did he think the benefits were of bringing partners into the process early on? And if you're trying to introduce a new assay, so I'm thinking maybe something that's going to get used during the development. What do you need to do? What do you need to be thinking about in order to make sure that that new assay is fit for purpose when you go into manufacturing scale?
Assay development and starting with the end in mind
[00:32:20] Matt Hewitt This is fantastic question. So there was a there was actually a manuscript in nature that was written by McKinsey, you know, now probably a year and a half, two years ago. And they actually looked compare late phase trials in advanced therapeutics to monoclonal antibodies. And they looked at where do you typically run into, you know, quote unquote, problems. And there was one area of significant difference between these two modalities, and they face trials. And it was primarily in CMC. And so within traditional small molecule and biologics development, you know, you might have an 8020 principle where 80 percent is describing the program, its aims, its inclusion exclusion criteria, and the 20 percent is, is on CMC. And advanced therapies that's and especially in cell therapy that is many times switched. Now you're 20 percent talking about a lot of your, you know, clinical protocols, trial design, etc., and maybe not 80 percent, but a much larger portion, let's say, is focused on CMC. And when they looked at and they broke down the pain points of the challenges within CMC and what are the regulators kind of, you know, focused on about a majority of the issues that folks run to are on the analytical side. Because we have to remember that a lot of these programs and genes of interest that we're using and to modify cells are using in AAVs or LNPs, you know, these are all custom GOIs for indications which we may not be able to find. So many times these assays are having to be developed to Novo. Many times we're talking about potency assays in this case. And the potency assays become a matrix as well. And they're really a surrogate for functionality and determining that the product does what you say it does. So the way that we look at this is there's kind of two styles of thinking. Let me be clear on that. I think one is the first one I'll go through is is is significantly diminishing in the last two years or so. That one is well, we don't really need a potency assay right now because certainly potency is not required for for phase one clinical. Because efficacy in phase one is not a primary endpoint. It's a secondary endpoint. So there's a very good argument that you don't need it. And actually there is a very good way of thinking that you don't actually need potency until you file your BOI, which I think is technically true. However, it's very risky. And the reason why this kind of line of thinking is significantly diminishing is that FDA has been very clear over the past two years and that a significant clinical benefit alone is not sufficient to get approval. The fifth piece of guidance that I probably didn't didn't mention the first four was actually the potency guidance that FDA released. And we certainly did comment on it. And in that guidance, it's a little bit you know, there's some refinement yet to be had on the guidance, I think. But one key point to that is that do lay out that you are putting your clinical program at risk if you don't have potency assay and early in clinical development because they want to assure that the you know, the therapy is functional as as desired. So the way that we do that in the second line of thinking is, okay, let's get this in as soon as possible. We take a track that we want do as soon as possible, and we work with our partners on that. But we don't necessarily want to hold up clinical activities if we don't need to. So what we'll do there is will slot it in, as we call an FIO assay or for information only. And that way it's not necessarily used for decision making on, say, product release, however, we do collect data. And we actually have a kind of another part of this, which is where if we if we can act and we are increasingly doing this with our partners and clients, you know, if we can connect with them early in the development process, same preclinical, even before they start doing their GLP safety toxicology studies, we can actually connect their CDMO teams with our our discovery or safety teams to ensure that we have good conversation about the assay structure and essentially the assay outputs to have a conversation about is this an assay which can be qualified and later validate that.
[00:36:46] Stuart Lowe Yeah. And you've got people thinking about that early on, so that you've kind of got the answer at the right time.
[00:36:51] Matt Hewitt So that the typical, the typical mantra around that is, you know, starting with the end in mind, you know, that is the, the, the, the tagline. And I think we found that to be quite successful because there are and there there's another thing you can do there that we've had several clients do and partners which is you know prior going into their GLP safety toxicology studies, they actually engage with our development side of our cell therapy CDMO as an example. We can produce product on a platform that will more closely resemble what they're doing in clinics.
[00:37:26] Stuart Lowe Next, I wanted to know what Matt thought of the use of AI in the manufacturing space.
AI as a tool, not a replacement
[00:37:36] Matt Hewitt We're already using that. A lot of folks are moving into those electronic systems. We have to obviously be careful with this as well. There's a lot of talk about AI use in the space. I was on a panel with this with Newsweek at Bio. We're talking a lot about this. And we have to be very careful that it's being seen as a tool rather than a replacement. And I think if you look at the automation and the way that large companies like Amazon, which really kind of lead the way in a lot of these spaces and you read a lot of their position papers and stories, their intent is not to fully replace humans, just to augment humans. And that's what we need to be looking out for this. A lot of service and there are various AI practices and consulting firms that talk about this conversations. I don't want AI in regulated spaces. You're not supposed to have, you know, systems that are learning and spaces that you're not supposed to be learning anything in real time. You're just supposed to be executing.
[00:38:32] Stuart Lowe You don't really want a judgment call. You know, the judgment call should be should be left to the humans, right?
[00:38:38] Matt Hewitt That's correct. Because if the AI, if you have we've had this discussion actually before as well, where if you have a quote unquote, validated AI and it makes the decision, my question to that, okay, is that decision gospel or is it a suggestion? Because we don't typically second guess validated systems. But what if the human disagrees with the AI? So how do you how do you reconcile that? I think as well as the regulators do not hold systems accountable. They hold people accountable. And so the other area that we have looked at and I think there is a smart way to do this, you know, we always talk about doing things and review by exception for anything really, and primarily in batch records, I think is where we focus on this. I'm concerned with that a little bit. And the example I give here is that, you know, whenever you download an app onto your iPhone or Google phone, how often do you read the terms and conditions?
[00:39:34] Stuart Lowe Yeah, I don't think I have ever done that.
[00:39:36] Matt Hewitt Exactly. And everybody ever asked that question. I've only ever had one person say I've actually read them and there is an email buried and the terms and conditions where you can email them and say, I do not want you to use any of my data. You know, that's that's what you have to do, if you can accept to download, but you don't want them to use your data. And that's the temptation is to say, okay, it's on the app store, it's on the Google Play store. It must be okay. So I've got to have except that we know that's not always the case and we need to be very, very careful with that.
[00:40:06] Stuart Lowe We've got several kind of hurdles to jump, but not not necessarily unsurmountable. But but but there's a there's a kind of question about how long is this all going to take? So where do, where do you think we we might be in 10 to 15 years time and how might we get there most efficiently?
A vision for the next 10 to 15 years
[00:40:24] Matt Hewitt No, I think it's great. I see this going kind of two routes and it's something that I was I was moderating a regulatory panel at the ARM on the Med meeting in April. We were actually discussing this one regulatory perspective. I see there hopefully being a bifurcation of the regulatory landscape between larger scale indications or we said normal indications that aren't rare or ultra rare, and then rare and ultra rare. I think for those larger indications, we'll see the traditional frameworks in place. I think we'll see automated manufacturing. I think we're likely to see autonomous robotics, as well as linear robotics in some cases come to bear on on these processes just to scale them and make them more efficient and cost effective. There was, you know, some statements from the in the Investor Day or the Innovation Day over Kite this year or they were saying they have an they believe they have line of sight to where their their cell therapies can have much higher margins for them in the future. And that's going to be based on increasing automation and efficiency of process, as well as analytics. So I think that's kind of the normal thing. We'll see a lot of automation, we'll see a lot of robotics. I just think it's going to be the natural space. You know, we use autonomous robotics and most of most other industries. You know, we're actually a little bit [00:41:43]cloud ball [0.3s] in manufacturing. And if we look at robotics used in controlled spaces, I mean, they're used routinely, as you know, in the semiconductor space. I mean, that that the tolerances for particulates and cleanliness is far and away much higher than what we do in therapeutic manufacturing. I think on the rare and ultra rare space where we might see their general say, I think there's enough work for everybody. But I think what we may see there is and there is some work on this, is, you know, consortia with the academic centers to create kind of centers of excellence or guidelines where academics can actually commercially manufacture ultra rare, rare, indicated for indications where there are small patient numbers.
[00:42:29] Stuart Lowe Okay. So the kind of like exceptional license or or authorization under license or something like that.
Decentralisation and central oversight
[00:42:35] Matt Hewitt And look, there are precedents for this already. If you look at the bone marrow transplant guidelines, there's licensure. There's separate for that. There's been a lot of questions about, you know, can we do that? And I think when you're looking at. You know, treating 30, 40 patients in ultra rare or several hundred a year in rare. This makes sense, right? So it might be better to have the academic centers and certainly we're and others are looking at and in working with academic centers. I mean, we just did a press release recently with Colorado University at Anschutz where we're doing work with them, and we hope to continue to collaborate with them.
[00:43:15] Stuart Lowe Decentralization and closed automation will help as that kind of movement will allow the kind of barrier to entry for doing it in universities and academic centers will be will be easier.
[00:43:25] Matt Hewitt And I think the other thing that they're looking for at that level is help in how to administer the systems as well as the talent for those specialized systems. So we're we can help there, as an example, is that we can use our centralized systems and essentially do a service model. We extend our recording umbrella to some of these centers so we can help them. So they do, they actually do the manufacturing themselves, they do the testing themselves, but we ultimately retain the higher level responsibility and functions of doing what disposition to release with their quality. And then now we again, we were we're stratifying this, we're making it more available, it's making it more efficient and we increase access. And by doing that, we're hoping we're also going to bring cost efficiencies to that space as well. So, I mean, there are multiple ways that we can work together. I don't want to box us into one way or the other. It should be whatever system works best. But I will say that some of the conversations that we've had with regulators, they do seem to be open to multiple flavors, but they do like the idea of creating a central quality oversight mechanism that will create, you know, this parent child complex where the centralized sites and the decentralized sites, because of the systems and the procedures are for all intents and purposes clones of each other.
[00:44:48] Stuart Lowe Yeah, it's almost like a franchise system.
[00:44:51] Matt Hewitt Exactly. And that's what I use. And whenever we start talking about the McDonald's franchise, we start talking about burgers and mcFlurry and french fries and how it's great. You can get essentially the same product no matter where you go in the U.S. or abroad. Right. But actually, if you go to a McDonald's abroad, you won't see differences, right, because they're adapting to a local market. And that's exactly what you'll see in decentralize. You know, you'll see it adapt to the local regulatory structure. But I mean, they do like that central quality oversight because for them, it potentially makes their inspection responsibilities easier because that they inspect the subset of your network because they are clones of each other, because a representative sample.
Final reflections: collaboration and innovation
[00:45:33] Stuart Lowe That's a really interesting vision. And I hope it does become reality, because you're right, it will it will allow more people to to get closer to these therapies. It should bring down costs, like you say. And this is something that is is as possible to do, just hasn't been done before. So that was probably the trepidation. I was really interested to hear Matt's perspective on how innovation is supporting the cell and gene therapy industry. My key takeaways from our conversation were alongside clinical success comes failures, two, companies that can fail fast and move on to more promising drug candidates stand a better chance of backing winners. Invest in the right areas. Some in gene therapy development requires a lot of CMC, and companies need to be looking at data collection and assay development early on in the product lifecycle. Keep humans in the loop. New technology should be seen as an opportunity to augment the role of operators rather than looking to replace them. As we've heard in this episode, there's no one size fits all approach to how we accelerate the development of cell and gene therapies. But the decisions we take early on in development can have a knock on impact even as far as the clinic. Working with partners can help inform that assessment of how innovation can be deployed and what impact it has on your operations. There's a wealth of experience out there. It's just a case of using it at the right time. And if we do this right, there's no doubt that the potential of cell and gene therapies will be felt by patients across the globe.
