Adjunct Professor at Johns Hopkins University
- Overview of the bispecific antibodies industry and mechanisms of action, including evaluating BiTEs (bispecific T-cell engager) and DARTs (dual-affinity retargeting antibodies)
- Deep dive into Johnson & Johnson’s (NYSE: J&J) Rybrevant (amivantamab-vmjw), Amgen’s (NASDAQ: AMGN) Blincyto (blinatumomab) and Roche’s (VTX: ROG) Hemlibra (emicizumab)
- Evaluation of existing drugs’ trial data and market viability
- Bispecific antibodies market outlook
Specialist (SP): Actually, I can start with the mutation. That’s what is leading to this actual antibody being developed, and the mutation is the exon 20 insertion mutation. That usually occurs from resistance to your traditional TKI inhibitors that are normally prescribed for this type of non-small-cell lung cancer. Normally, non-smallcell lung cancer, they always look for EGFR mutations because PD-1 therapies don’t work very well with this type of mutation. These patients, which are about 1-2% of patients, 3% of patients, will go on this therapy, a traditional TKI, but eventually they develop mutations in the exon 20. With that, people then developed this antibody, Rybrevant, J&J’s antibody. This antibody is a bispecific, not surprisingly, and it’s actually an asymmetric, meaning that one side is different from the other. It’s a traditional antibody shape, it’s a Y shape, using the knob-and-hole procedure to pair them. It has one binding site that binds to EGFR and the other binding site binds to the MET. These are two surface molecules that are involved in signalling to cancer growth.
As I already mentioned, the EGFR 20 mutation is the third most prevalent EGFR mutation and it accounts sometimes up to 9% of all the EGFR mutations. When you look at that, actually these patients do the worst, they have the poorest prognosis when you consider all non-small-cell lung cancer patients, so this is a very high unmet need area, because they eventually develop resistance to traditional TKIs, so this antibody was like a life-changing event for many people. What it does, the mechanism of action is it binds to EGFR and also binds to the MET molecule on the surface, and what that does is that brings those molecules together and they’re unable to signal. This is an interesting molecule, the antibody, because usually TKIs, what they do is they work internally inside the cell, and so a lot of the mutations develop in the internal part, the signalling part. This molecule, the bispecific by J&J, then bypasses that, brings those cells together, and that causes a couple of things to happen, one, again, blocking the signalling, but it also causes endocytosis, internalisation of the signalling molecules, and that of course stops the growth of the cancer, that’s one of the mechanisms.
Another one is trogocytosis, which brings in macrophages and NK cells to cause cyto-cellular toxicity and kills the cancers. The MET is interesting. There have been some studies on it, and the role of MET, even though it is involved in cancer and signalling, a study was done that said it’s not well-defined whether the MET activity is actually contributing to the role of this antibody. It may, what’s going on here is that the Rybrevant, what it does is it prevents resistance. That is, one of the resistance mechanisms to TKIs or EGFR antibodies, for instance, is actually activation of MET. This bispecific is helping to reduce recurrence or resistance to EGFR. It’s a very intriguing mechanism because it brings those two molecules together, preventing them from signalling. Again, a good example of an asymmetric bispecific antibody, and it has a traditional antibody shape. With that, I can stop there.
SP: We definitely talked about that in the first talk. This is where you use the knobs-and-hole procedure to allow two heterodimeric chains to come together, two heavy chains, that allows the pairing, and then they do the Fab exchange or CrossMab technology. This was considered a dual body, actually. It’s an IgG1 dual body, and so, again, a lot of companies use these. These are fairly standard now, where you can get two different chains to pair. This is actually one of the simplest forms. When you really look at the structure, it’s a traditional antibody, it just has two different binding sites. Instead of being a homodimer, it’s a heterodimer, so again, using knobs-and-hole plus CrossMAb or dual antibody exchange. That allows you to then pair different light chains to the two different heavy chains, and that allows that. In this case, this is a fairly straightforward procedure.
Third Bridge (TB): I think what he is describing there, is that for an asymmetric, if you were to split the Y shape of an antibody down the middle, one side would consist of a heavy and a light chain of one type of antigen, and the other side, which was why it would be a bispecific and asymmetric, would be another heavy and light chain with a different antigen. When you have asymmetric, the split is not heterogeneous. Am I correct that those changes occur in the Fc [fragment crystallised] portion on the bispecific antibody?
SP: Yes, there are a couple of places. One is there, the other is when you… the process you just mentioned, one side can actually have two binding sites, which means they match the other side, but it still has two binding sites, so that is considered a bispecific, but it happens to be symmetric, so there are antibodies like that as well. You can develop multiple specificities on both sides, and that way it’s considered a bispecific, and it’s symmetric because each side is the same. For the J&J, it just happens to be a Y shape with two different specificities, sort of a classic bispecific. If you envisioned it, that’s what it would look like, a Y shape molecule with two different… You really need to look into the detail sometimes, they don’t actually show it. There was a Nature Reviews about this. Very clearly shows the differences, and it’s colour-coded, so it’s a very nice way, but yes, there is a lot of confusion on some of the papers. Unfortunately, they don’t show it exactly the way it’s described, but definitely it’s an asymmetric. It doesn’t have to be completely different on one side and the other, it just has different binding sites, that’s it. There are other asymmetrics that look really different. They don’t even have a binding site on one side, it’s some other molecule, so, again, it’s very heterogeneous how they look. I think that’s where some of the confusion can lie.
TB: To your point, what is so fascinating, while this Rybrevant structure is just a simple Y that is asymmetric, bispecifics such as IgM-2323 can have all these different Ys linked, and then you get all these unusual shapes, such as 2323 being pentameric. Then there is this whole concept of the hook effect, where if you overdose, it can create some issues with binding, which is fascinating. When folks are considering higher dosing and wondering if it would be even more efficacious, one thing to realise with bispecific antibodies is you can have this crowding out.
SP: The CHRYSALIS trial was a phase 1 trial, actually. Again, about 1.5-2.5% of patients with non-small-cell lung cancer have this EGFR mutation, and it can be very heterogeneous, so it’s a very difficult population to treat, actually. This is one of their starting points, you’re starting with a very difficult population. They don’t respond to checkpoint therapy, so you’re really relying on some other new therapy, TKIs, new-generation, third-generation TKIs, but also bispecifics. For the CHRYSALIS trial, the bispecific trial of Rybrevant, the median age was 62, so a much older population, 95% adenocarcinoma, about 46% had prior immunotherapy, and then prior therapies were around two. This bispecific was administered at four weeks and then every two weeks thereafter, and then it was given until disease progression or unacceptable toxicity. The ORR, we were talking about that just now, the ORR was 40% but the complete response was 3.7%. The way you have to think about that is, remember this is already a very small population and then they’re refractory to traditional TKIs, and so this area was a high-unmet-need patient population that really did not respond to anything else once they failed TKI.
Of course, there are other TKIs in the pipeline, that’s another talk for another day, but right now there was a high unmet need in this space, and this bispecific, although numbers are small, it is something that allows these patients to have another line of treatment. Otherwise, there is no other treatment available to these patients. There is chemotherapy, of course, but usually that has a very low response rate and these patients are already very sick as it is. Normally, we look at overall response rate. 40%, for this type of population, this is quite a good number, and also this therapy was quite safe. There were some reports of pneumonitis. Again, these are expected. I guess there was some eye blurring, vision impairment, but, again, these all resolved eventually. Again, all these are somewhat expected. A lot of immunotherapies will have side effects like this. They usually resolve on standard therapies or some steroids, for instance.
The data also showed, so this is a fully human EGFR MET bispecific, it was safe and efficacious for this particular population, as I mentioned. I believe the maximum tolerated dose was not identified and that it was recommended to be, I think, 1,400mg eventually. I think there were 81 patients, and that was about it, that was the trial. I know you might say, “You would expect better,” but, again, this is, first of all, a very small population of patients and they’re refractory to TKIs, so they have a response post-TKI relapse. This is encouraging. Of course, you want to see better numbers, and there are newer trials, I believe, where they’re doing combination therapies with Rybrevant. Again, that’s probably another talk. For now, people obviously realise that it’s not the best we could achieve. Probably combination therapies are warranted here.
TB: I think that’s a great summary of the safety and tolerability profile. You’ve hit on some of the ocular issues, but I think you’ve fairly pointed out that they were very minimal. Rybrevant can cause ocular toxicity, keratitis, dry eye and conjunctival redness, but keratitis occurred in just 0.7% and uveitis occurred in just 0.3%. Then another issue was ILD [interstitial lung disease] and pneumonia, and that was just 3.3%, so a very clean profile, even though we’re talking about just 80 or so patients. The biggest issue was infusion-related reactions, at 66%, but 65% of those occurred week one, day one, and most of that was just within the first hour after infusion, so it developed rapidly and moved forward.
SP: : Remember, this is a small niche, 2-3% at most, that have these mutations, so it’s not a large space. You may know Takeda won approval for an EGFR mutation drug as well, and so there is some competition there, I think it’s called Exkivity, and then also AstraZeneca has the market lead right now, Tagrisso, so there are those, but those are TKIs. Remember, these are binding inside the cell. The Rybrevant’s edge is it binds outside the cell, so it’s almost resistant to additional mutations, because there are other mutations that can develop. I think Rybrevant is also being looked at to be combined with lazertinib in a phase 3 trial, so they are looking at combinations. I think that’s something to look forward to, to see how those turn out.
SP: TKIs are very infamous for eventually failing because of the mutations, and I feel that’s going to happen here too with these TKIs, Tagrisso, for instance. Eventually, they will develop mutations. There are other companies looking at other TKIs that can be resistant to those, but if you look at the overall, it is a small chunk of the market, 2-3%, so there is competition there, so it’s hard to say. Again, the antibody, I think, will not have as much resistance as the TKIs, and it may even be better in combination, with the lazertinib, for instance. They may win out in the end because they won’t develop resistance in those particular combinations, but we won’t know that for sure, of course, until that data plays out.
SP: The BiTE is a really interesting molecule. This is not your traditional Y-shaped molecule. What the BiTEs do is, first of all, they’re very small compared to a regular antibody. It’s literally two, they’re called single-chain variable fragments, that are bound together. Actually, it’s not considered asymmetric, but you can think of it as one side binds a T cell, CD3, the other side binds a tumour antigen like CD19. Blincyto, or blinatumomab, that’s your traditional FDA-approved molecule, it’s a bispecific T-cell engager, there is a huge space just for those. Some people think of bispecifics but they only think of that space, CD3, and then CD19 or some molecule. The BiTEs are, like I just mentioned, they’re very small, and they bring cells together, so they’re T-cell engagers, they get the cells very close to each other, the T cells to the cancer, the cytotoxic T cells, and that helps develop a synapse and allows that to overcome cancer’s evasion of the immune system. The ceruminous T cells then get inside the cancer because these BiTEs bring them in there, and that causes killing of cancerous cells. BiTEs are designed to target tumour-associated antigens like CD19, but also CD33, DLL, FLT3 ligand, MUC17, prostate cancer antigen like PSMA, and other ones, like claudin 18.2 in gastric cancer.
There are a variety of antigens that can be targeted, a variety of shapes you can develop from these BiTEs. They can be very tiny to very large, depending on how long you want them to last. For instance, blinatumomab, that is Blincyto, has a very short half-life because it’s a small molecule, it doesn’t have the Fc fragment, and therefore it’s continuously infused, so that’s a down, a caveat, to that. There are other BiTEs out there that have an Fc portion added on to allow it to have a longer half-life, so, again, different types of structures, similar ideas. This is probably one of the biggest markets right now, is the BiTEs. They tend to be pretty easy to make, to develop. You can make a BiTE pretty much to anything. If you have an antibody, you can make a BiTE. Quickly to make, relatively small, so they can penetrate solid tumours better. That’s one of the advantages of these. They can activate any T cell, so it’s off the shelf, you can give it to anybody, there is no MHC restriction here. That’s a quick summary of that, about BiTEs.
SP: The BiTE is interesting because, again, it binds CD3, which is on the T cell, and then the other part, either CD19 or any antigen you just mentioned, so it’s literally a molecule that bridges the T cell to whatever that antigen is that happens to be on the cancer. The CD3, you might ask, “Why CD3? What is that going to do?” CD3 is part of the T-cell receptor, like the on switch, and so when you stimulate it with an antibody, this is not a blocking antibody, this is actually a stimulating, agonistic antibody, it stimulates the CD3 and that onswitch turns the T cell on, and when that happens, the T cell will then produce these granzymes and perforin that we just talked about. Those are the killing mechanisms of the T cell. That’s what literally punches holes in the cancerous cell. The reason you want to have the BiTE is because it’s very tiny and you need to get that T cell right next to the cancer. That’s one reason why a BiTE is so small, that’s why it’s designed like that, because it’s bringing that T cell right up against the cancer cell. It activates it, the T cell then releases the granzymes and perforin. Think of perforation, it’s literally perforating the cancerous cell. The granzymes go in and then kill the cancerous cell. Again, that’s the idealistic mechanism. These BiTEs are so versatile you can choose many antigens.
The antigens you mentioned here, BCMA, which is a multiple myeloma, you may have heard of BCMA CARs. BCMA is an area that’s highly active. BCMA is on multiple myeloma and it’s pretty specific to that, so that’s considered a tumour antigen. CD19, as everyone knows about, is on ALL, DLL, diffuse large B-cell lymphoma, it’s not really specific to cancer, it’s also on regular B cells, but CD19 is one of those markers that has already shown good efficacy, since blinatumomab obviously has shown that, as well as CAR-Ts, so this is CD19. There are other molecules, CD20, that are also on B-cells. There is also CD123, CD33 on AML. Again, all these are specific for these types of cancers. It doesn’t mean that it’s tumour-specific. These are just normal molecules that just happen to be on these cancers. That’s why the solid tumour space is having some issues with bispecifics, because there are not very many targets on solid tumours, but haematologic tumours do have these targets. PSMA, for instance, on prostate, is a target on prostate, but that’s not really considered a tumour antigen, it just happens to be upregulated on prostate. These targets, and there are many other ones, like CD123, CD38, there’s FLT-3 ligand you mentioned, CEA, FKN, and so again, those are just some of the ones that are being targeted. HER2, HER3, but also PD-L1, checkpoints, you can target them as well using these molecules.
SP: That’s the main function of the Fc, one of the main functions of the Fc receptor, that’s the bottom portion of the Y, which a BiTE does not have. That allows longer half-life because what it does is it binds to something called the Fc neonatal receptor. Don’t get me wrong when I say neonatal, it doesn’t mean only neonates have it. We all have that receptor. What it does it is extends the half-life of the antibody for at least 30 days, at least for IgG1s. That Fc is allowing that. You can put mutations on that Fc to either block that, and that will reduce half-life or just eliminate it completely, as a BiTE like blinatumomab doesn’t have that. The Fc also does other things, though. It has effector function that’s involved in antibody-dependent cytotoxicity, but the thing is you wouldn’t want that on the blinatumomab because if you had that, that would kill the T cell. When you design these, you have to think, “Do I want to kill the cell or do I want to bring that cell next to another cell?” You have to be thinking, “What’s the mechanism here?” before you just start doing all these changes. There is a blinatumomab, I believe, being developed where it has an Fc on it, to extend the half-life. I saw that somewhere.
The reason they would do that is that would allow them, because right now it is a 1.25-hour half-life. That’s really short, but that allows you to dose it continuously and then be able to stop it immediately if there are any side effects, because there can be side effects, such as cytokine release syndrome. Not as powerful as, say, CAR-T, but there are side effects to these molecules, so you have to be careful. Modifying that Fc can give you many advantages and also disadvantages. You always have to think, what is the mechanism? What is the molecule targeting? What’s your purpose? Are you bringing another cell to another? Are you targeting just two surface molecules? Do you want to cause antibody-dependent cytotoxicity to help kill the tumour? Again, you have to really be thinking of all these things. Also the half-life, do you want to extend the half-life so you don’t have to keep dosing it, or do you want to have that short half-life to prevent side-effects, if you stop infusion immediately? All these are considered, every patient population is going to be different, every cancer is going to be different, all the targets, and so all of these come into play. You can imagine, this really opens a can of worms, and when you really start digging into this space you do see how many different constructs there are. There are at least 35 different constructs of bispecifics and each one of them has their advantages and disadvantages.
TB: Do bispecifics not have the Fc and then it’s added? I thought they had it and then it was removed.
SP: No, a lot of bispecifics, they’ll be just fragments, they’re called, like blinatumomab. That doesn’t have any Fc at all, it’s just a fragment.
TB: That completely makes sense because that’s what was causing my confusion. Just as a recap, you have the Y of the antibody, and then in the pronged, two-part portion, known as the Fab, we have the antigen-binding portion on each prong. Then you have the single line, and that’s called the Fc. If you don’t have that Fc, you don’t have much of a half-life. If it’s not there to begin with, and you want to have constant infusion so that you can stop, to your point, if you have real potential for CRS, then you’d want a short half-life, but if you don’t, and you don’t want to have constant infusion, and you don’t have an Fc already, you’d want to add that, and that’s what extends the half-life.
SP: Exactly, that’s right.
SP: Initially, when it came out it was approved for a Philadelphia chromosome negative B-cell ALL and they were getting response rates, I believe, in the 40% range. The problem is what happened is there is a lot of resistance to this molecule, and at the same time this was happening, you started to see development of CAR-T cell therapy. CAR-T cells, I think, overwhelmed this particular space. Although this sometimes is preferred, because this is easier to give, it can be given to anybody, it’s not a cellular therapy, CAR-Ts tend to have a longer durability compared to blinatumomab. I think that’s the main reason, that’s how I would visualise it. It’s not that it’s not a successful drug. I think it has been successful. I think it’s the prodigy for many other new drugs that will come out eventually, as we understand them better. I think just because of the actual resistance, usually within 12 months people are resistant to blinatumomab, so with CARs there’s a longer durability and less resistance, supposedly. At least that’s what the data shows so far, but that could always change. That’s probably the main reason.
I think it’s doing what it’s supposed to do, but eventually, remember, when it grabs on to a T-cell the CD3 portion of blinatumomab can bind to any T cell. The T cell might not even be a cytotoxic T cell. You have to understand that it may bind T regulatory cells or other cells that don’t do anything, or exhausted cells with PD1. You can’t control that part. With a CAR-T, you’re given a cell that you know is very specific and it’s a T cell and it’s cytotoxic. I think CAR-T has overshadowed this particular market here, but this is still preferable because it’s not a cellular therapy and it’s not as involved. Cellular therapy is very involved. People still try this, but I think the continuous infusion is one of the downsides, the caveats, to this. They’re looking, I think, at subcu dosing now, as well as the Fc blinatumomab, where it has a longer half-life, so that may change things.
TB: I believe Hemlibra is the only prophylactic treatment for people with haemophilia A with or without factor VIII inhibitors. Interestingly, it can be self-administered subcutaneously and it has multiple dosing options, including once-weekly, every two weeks and every four weeks. I think, given how this has performed, this has been a huge success.
SP: This drug, I’ve got to say, this is the type of drug where it makes you want to do medicine and science, because here you’re correcting a genetic defect, haemophilia, which a lot of people are suffering from, and there hasn’t been a therapy in 20 years or so. This is correcting that through an antibody, a bispecific, it’s another asymmetric antibody. It’s just like J&J’s. It binds to two different sites, two different molecules I mean. What it’s doing is it’s bringing two molecules together, so it’s your Y-shaped molecule again, one binds to factor X and the other binds to factor IXa, and the reason it does that is because normally the factor VIIIa which is missing in these patients, that’s bridging those two molecules together and you get coagulation. When you cut yourself, you want the blood to stop bleeding, you need that factor VIII, VIIIa. What this antibody does, it just mimics that interaction. It brings X to IXa and they come together and you get coagulation. This is not a cancer drug, this is completely different. This is for haemophilia, it’s for correcting a genetic defect that, again, many people have.
It’s had tremendous success. I think 96% of patients, or 96-97%, had reduced bleeds. None of them had to get the other factor VIII transfusions. This is an amazing… This is what bispecifics can do, they can correct something. If you understand the mechanism, you can bring two things together. This is similar to the J&J where it brings EGFR to MET, it brings them together, but that’s a different mechanism, that’s shutting them down, that’s stopping the signalling. Here, this is bringing them together so that they can signal and then cause coagulation. Again, different mechanism, similar idea. This one has an Fc as well, a longer half-life, typical shape of a Y. Again, the design is important here, what you want to do, how you want to bring things together. Yes, this works, it reduced bleeds, and so this is a very promising start to the bispecifics space.
TB: Just as a summary, when we discussed J&J’s Rybrevant, which was approved in May 2021, and whose future is ahead of it, no sales have been broken out by J&J to my knowledge, but certainly interesting in the NSCLC with EGFR exon 20 space. Then we moved forward to Amgen’s Blincyto, which perhaps was overwhelmed by CAR-T, and was approved all the way back in 2014.
Then we finished off with this fantastic drug, Roche’s Hemlibra, which, to your point, makes you love medicine. This was to my knowledge the first new class of medicine in nearly 20 years for the haemophilia A community. That community, which you said was quite large, had been living with the same standard of care for two decades, and now we’ve moved into the subcu, where you can go once every four weeks. I believe that the old standard of care required weekly infusion, so you’ve moved from a potential once-weekly IV to a subcu maybe once every four weeks, and that’s not for everybody, but it just shows you the tremendous advance just in terms of quality of life. That doesn’t even cover that this is prophylactic and has done well to reduce bleeds compared to the prior factor VIII prophylactic. I think it was 68% better than the previous standard of care, so just a wonderful advance for humanity there, not to try to be hyperbolic, but it is performing quite well in the marketplace.
SP: DARTs, really it’s very similar to a BiTE. When you really look at it, it looks similar, it’s just the way it’s designed, and that design is to help in manufacturing, the manufacturing process. The idea is, it’s just like a BiTE, one side binds D3, for instance, the other site binds… and these DARTs can be made into different shapes, they could be Y-shaped, if you like, they could be like a BiTE-type size, depending on your application, what you want to do with it. You could have multiple DARTs on one Y-shaped molecule, and it can be symmetric or it can be asymmetric, so, again, a tetravalent. A DART, the format is actually based on something called the diabody format, and all that is it’s something that separates the variable domains of the heavy and light chains of the two antigen-binding sites, and it puts it onto two separate polypeptide chains. Whereas the two polypeptide chains associate non-covalently in the diabody format, the DART format provides additional stabilisation through some, it’s called a C-terminal disulphide bridge. I know this is complicated, but the point is it’s just a different way they’re designing it, and so far it looks like DARTs are easy to produce in high quality, quantity, that is, and quality.
They actually performed head-to-head with BiTEs, and it looked like it performed or consistently outperformed the BiTE format, so that could be a competitor to the BiTEs in the future, so you have to think about that. These are very small molecules, just like a BiTE, very similar. Again, just designed slightly differently for manufacturing, but can be designed in many multiple formats, tetravalent, bivalent, trivalent, you can do whatever you like with these, you can add many antigen specificities. One other fact is, compared to the BiTE format, the DART format was moderately highly associated with a higher association rate with CD3, so it tends to bind better to CD3. Again, something about the design makes it somewhat more flexible than a BiTE. This could be a competitor. There are a lot of DART antibodies being developed out there, in many different formats, different designs, so that, again, is another one to consider, considering the whole space, as a whole.
TB: I was a little surprised by MacroGenics’ market cap, just under USD 990m. I think the company was the originators of the DART format, and so maybe other players stood on its shoulders and might have had more success. I think that’s a fantastic topic for another day maybe, down the road with you, when we dig into either a MacroGenics-specific call or a DART landscape call.