Manufacturing: CAR T-cell Therapy

Expert Profile

Section 1: Current Trends in CAR T-cell Manufacturing

1.1. What are the current CAR-T cell manufacturing approaches?

The first step is taking a sample from the patient and then essentially you need to select the cells you want to use for your CAR T-cells. As a platform, of course, the majority of the people use T-cells. Normally, the selection is done by the antigens CD4 and CD8, but there are people looking into using other subsets of T-cells. Certain manufacturers are looking at using gamma delta, NK cells, and NKT. That’s the first real choice you have, which kind of subset of T-cells, NK, or immune cells we want to use.

Each of those has advantages and disadvantages. Currently, we are focusing on NK which is important because it’s part of the natural immunity. Essentially, the NK cells could be a good allogenic treatment so you could just expand NK and give it to different patients. That will be one way to produce off-shelf therapy.

The second thing is, when you decide what to select, of course, you need to expand the cells and the expansion is really based on which type of cells you select because for T-cells we need a certain type of condition to grow, gamma delta we’ll want something slightly different case. So, you really need to tailor that process to your T-cells, to the cells you’re using for CAR. Then, of course, there is the gene modification type and delivery of the construct which can be changed.

You need to deliver the construct; whatever cells are you using, and the construct can be delivered in different ways. Of course, a lot is done by viral delivery because of course, that’s has been historically the way, but viral delivery has lots of disadvantages. So, first of all, it’s costly, it’s longer, they require specific expertise. So essentially, lots of companies are looking into non-viral delivery. There are different ways now, there is lots going on, the sleeping beauty transposon system, lipidic delivery, electroporation, and all sorts of ways because that would make the process very fast and cheap.

At the moment, we use this viral delivery, and then, of course, you need to make sure that your construct of interest is resulting in a decent percentage. You will need to go back and check, you have all the release assays to assess how much CAR there is and it has to give a certain percentage and all of that and you need to select the cells and you need to make sure that if you are responding to CD8/CD4. Your progression is pure enough, when you guys find the NK you don’t have T-cells otherwise, you don’t give to the patient what you say.

The most important assay is the potency assay, if we talk about CAR T-cells, you need to show that your CAR actually kills what it has to kill. Then, of course, you need to have all the control in place to say, “Yes, the cells have to have this antigen and don’t give them on everyone.” There are all the toxicological studies that you need to ensure that there are no side effects and you need to look at cytokine storm. The last important step is the formulation where you give the cells back to the patient.

1.2. What are the main bottlenecks in current manufacturing processes?

Every single step has a challenge and every single step is very expensive. What is important in all of this is that you need to have a close process where a university comes in with an approval principle to say, “This concept is better than the other, this domain is better than the other, these cells are going to stay longer, integral than the others”.

Of course, machines like CliniMACS Prodigy will help, as it is able to select T-cells and can deliver some of those gene modifications you need. Of course, it cannot be applied to everything. The idea is, whatever system you do has to be closed meaning that, as few people as possible putting their hands in. There are other companies coming up with different systems such as Ori BioTech and Miltenyi so there is growing competition in this space. Everybody’s coming with a slightly different incubator with greater volumes but each of those steps requires resource, compliance with regulatory standards etc. A level of specific training is required to work in these positions, and this is really about to be as responsible as possible with the closed systems.

Then of course cost is a big issue as well as regulation. For example, all the companies building these facilities need to have all the licenses in place. You need to make sure that the building has a certain exit and entrance and there are delicate rules around everything.

What they are doing now is they are building licensed clean rooms and then renting them out. So, the company has to care about creating their own process but know about building a facility that actually has the right rules and is actually approved for this kind of experimentation. So essentially, what Catapult has done is build a series of rooms that the company can just rent for the time they need. As I said, there’s a lot to think about and everything requires lots of investment. If you want to build something for yourself, you really need to be a successful company with lots of investment.

For example, the company Autolus is building up a center just for themselves but because now they managed to build up so many CARs in the pipeline and they are looking to just have a building where they have their own units, their own clean rooms and they will do all of that. To overcome the fact that the process is long, sometimes you require 15 or 20 days just to produce one CART T-cell from one patient in each state. This is costly and they are looking at employing people on a shift in a way that they can run the process 24/7.

In this case, you need to have people that are willing to go and work at six o’clock in the morning or on weekends, but they are looking into that model. Then of course, after you have your cells, the other problem is the distribution where you need to freeze them. If you want to send the cells to your patients somewhere else in the world, then you need to make sure that there are all the cold supply chain and the works and make sure that the delivery is done in a certain way and the cells normally are kept in liquid nitrogen.

This brings the topic of centralized versus decentralized because one thing is freezing the cells in a big storage room somewhere and then post it to where the patient is. The other thing, decentralizing means is to make sure that this hospital can have the possibility to create this process now. So essentially, in that case, the patient is in the hospital or some type of center where you actually take the blood and can almost bring the blood into an adjacent lab to start the process. When the cells are ready to go and are modified in the right way, they can go back to the patient immediately. Of course, centralized makes it easier from one point of view because you do everything in the same place, and you have lots of people working and that’s what they do continuously at companies like Autolus. However, the challenge is the blood has to come from the hospital, from the patient into that center, and then, it has to go back out as a finished product.

1.3. What are the greatest unmet needs in autologous CAR-T cell manufacturing?

On the biological side, there are different issues with the CAR, of course, every cancer is specific, so you need a different construct for every cancer. You need to make sure that the majority of the time, the cancer cells are exactly the same cells that are hidden in the body. So essentially, there is an antigen that you target which is normally more expressed in cancer cells, but it doesn’t mean that it’s not a toll on the normal cells. So, you need to make sure that the CAR is specific.

People came around that way, for example, by double targeting, they are trying to target two things on the cancer cells. So, if the normal cells don’t have both, that is normally the case, the CAR does not get activated, or if there is any issue with the CAR T-cells keep proliferating and maybe themselves become a cancer, then there are, what we call killer genes. So, there is a switch of genes, essentially a gene that we insert into the CAR construct and then with the drug and we can kill the cancer cells. The other problem is that these CAR T-cells can produce a cytokine storm. So, they get too activated, they produce lots of cytokines and the patient has side effects with that so that’s another reason why if you have a way to kill the CAR T-cells if it’s becoming too much, you just switch off the CAR, and essentially and get rid of that in the patient.

Then the other issue is the CAR T-cells are quite good and that’s why it’s been approved in the clinic for blood cancer when they are not really good with solid tumors. Essentially the problem is normally naturally the T-cells and therefore the CAR T-cells live in the blood. Of course, if you want to go in specifically into the liver or somewhere else, then you need to make sure that something is standing to the CAR T-cells to actually go to the liver. People have been coming up with ideas of putting some homing receptors on top of the CAR so there are lots more modifications and emulsification are going on and in a way that you tell, you direct your CAR T-cells into the right place.

The last issue is about the microenvironment which is really about switching off the immunological response because the tumor doesn’t want T-cells around because of course, it is naturally their job to kill them. Therefore, it has to develop all those ways to go around and switch off the immune response. So, that’s where it comes into play the whole story of the checkpoint inhibitors because you can have a combination therapy with a checkpoint in reverse where you essentially switch off the microenvironment, the hematology, not the environment, and the CAR can do the dump or you just have some modification again into the CAR T-cells.

For example, you knock out all the receptors for those checkpoint inhibitors, so the CAR T-cells don’t feel them anymore because they don’t have the receptor for that. Now of course, with the CRISPR/Cas9 we are becoming better at changing one to five genes at a time, and you can rule all those modifications together.

So, you can just go into other receptors to switch off things. Then of course, that makes the process much more complicated because something is like you respond, you put your thinking and you go and something else, in your process, you need to add steps of, “I need to switch off these genes, switch on this gene” because of course, everything has to be checked, everything has to have an assay to test it to actually ensure it was better than what was before. You need to prove that all those modifications have a reason, are not toxic for the patient, don’t provoke other problems in the patient. The process becomes more complicated.

1.4. How is the manufacturing process currently qualified and regulated?

There are agreed standards, called the GMP so, Good Manufacturing Practice for T-cell manufacturing. Of course, we need to follow all these regulations. We need to make sure that our product is GMP compliant, with good clinical practice. It’s a common standard for everybody and you really need to follow it for every single step. So, from the moment it’s self-contained from how the people need to dress in this clean room, how you actually report the data, we need to have two people that check each other, essentially.

Section 2: Future of CAR T-cell Therapy

2.1. What are the next-generation manufacturing technologies for autologous CAR-T?

It will be a mix of different approaches that combine different receptors and having those CAR T-cells much more modified. I think people will start adding homing receptors to knock out checkpoint inhibitors. Essentially, the answer will come from genetic modification and making sure that because now we are able to do it and we can insert the steps in our process. That’s where the answer will come from, will come from heavily genetically modified T-cells. The reality is that cells as they are, will not be able to do the job. We need to go step by step and check every single one of those receptors that we want, or we don’t want. So, the answer is in a much more complex genetic modification.

2.2. What is the role that allogenic CAR-T cell production will play in the future? What are the key opportunities?

Allogenic is going to be the topic that everybody’s going to look into. There are different ways to get allogeneic cells. You can modify the T-cells with the ability of CRISPR so there are some genes that essentially make the T-cells be rejected. Of course, if I take my T-cells, they put it in my body then everything is fine but if I take my T-cells, they put it in your body or someone else, then it will be rejected. So, because of CRISPR and the ability to knock out too many genes at the same time, there is one approach that is not knock out everything that makes the T-cells rejected. Then of course, in this way, you can give it to different people.

The other option is getting NK cells, for example, that don’t get rejected because they don’t kill recognized HSC-1/2. So, you can take my NK cells, and give them to someone else. There are people looking at gamma delta for the same reason because there’s a subset of T-cells that actually don’t get rejected. There are people just looking at what else is there that can be used as a platform for CAR or looking at how we can modify the cells as much in order to be an off-shelf product. Of course, the good thing is, even if the process becomes a little bit complex, you can produce tons of it and then freeze and give it to different patients; so, it’s cheaper. If you are not dependent on a patient base anymore and you find a way to produce tons, you create many off-shelf vials.

So, in order to get the product, you need to solve all the technical problems, costs and regulations. There is also a lack of general training and there is no university that teaches you how to do manufacturing. They teach you about the biology behind it but that’s it. The CAR T-cells are the answer for solid tumors and I’m quite sure that the CAR T-cells are the answer to a lot of liquid tumors but there is a lot to do to get those working for solids.

2.3. How will CAR-T cell manufacturing become more decentralized?

The problem is that I think the two models will live together because it’s a matter of how you want to spend your money. Of course, having a decentralized model is good and is easier for the transporter and you save money on the supply chain. On the other side, you need to make sure that every single place you will not decentralize into has the right facility, the right equipment, and the people are really expensive.

My point is, you need to have a Prodigy system per place in each of these places to have a cleanroom that has to be checked and all of this requires money, etc. It’s very hard to predict that everything will be decentralized, or everything will be patient-based. Of course, if the industry becomes good in allogenic, then you don’t need the patient base anymore, but we are not there yet.

2.4. What will be the impact of these new gene-editing technologies on the clinical outcome of patients?

I don’t think that will make the process faster because the process will be probably the same or maybe longer because you have an extra step. The process is much more dependent on the fact that you need to expand the cells, and this also will grow at a certain speed. They will require the time they require to grow. So, you need to expand the cells fast, modify them and then expand the modified ones until the number is right for the patient. The number is calculated on the number of CAR T-cells per kilo of the patient.

Then what are the means for a patient is you get a more potent CAR, you get a CAR that has fewer side effects, and that is more specific for the cancer you have. You have a CAR that goes exactly where you want the CAR T-cells to be. So, this is about making a better, more efficient fit for CAR T-cells. That’s where the CAR, the genetic modification is playing into, they are not playing to make the process quicker, or probably they make the process a bit more complicated. It’s still doable but you have an extra step.