Marlana, this has been a great conversation. Let’s close with future directions. Where are we going with tebentafusp? What are the next trials we must do and the next set of questions we have to answer?

Teasing out why there is still such a great response in patients whose best overall response is progressive disease is an important question, because again to your comment, can we stop some treatment in these patients? Are these patients who need to continue to be treated beyond progression, or is this something that is durable? We just don’t know because all of our patients who we have treated thus far, if they’re responding, they’ve stayed on therapy. Combinations are always the next natural move whenever you get a drug under your belt. What do we combine this with? Would we combine it with something like immune checkpoint inhibition because of some of the anecdotes and some of the changes we may see at the tumor level? Do we make it multimodality? Are we potentially combining this with things like liver-directed therapy? Those are questions; there may be questions about schedule. Again, weekly treatment for how long? Do you have any other thoughts about where we’re headed with this?

The last critical one that we are already in discussions about is the adjuvant trial. For the first time ever we’ve gotten effective systemic therapy in the metastatic setting. So it has to be tested in the adjuvant setting.

There is one other piece; this wasn’t compared to Ipi/Nivo [ipilimumab and nivolumab]. Like you said from the outset, in the absence of a clinical trial, lots of patients with metastatic uveal melanoma have been getting combination immunotherapy. If you look even across trials, which are never supposed to do, tebentafusp still looks pretty good. It gives us even more options, this being the first true drug in a uveal-specific randomized trial to show overall survival benefit. It’s amazing for the community and for the patients. It’s an exciting time.

Dr Orloff, I could not have said that better myself. That’s a fantastic way to close this session. Thank you very much for this. This has been a great conversation.

Videos

The next steps in research investigating best practices using novel therapies such as tebentafusp as treatment for uveal melanoma. 

Future Management of Uveal Melanoma

We have the 202 [IMCgp100-202] study, which was basically untreated patients, a clear survival advantage. We have the 102 [IMCgp100-102 study] trial, where it’s a single-arm trial, but overall survival looks good compared to historical controls. Who are the patients you’re going to treat, and are there any patients who are HLA [human leukocyte antigen] eligible who you might not treat with tebentafusp?

No. 1, we screen all our especially high-risk patients, the patients who are not yet metastatic but at least high-risk primary. They have their primary eye tumor treated, and then either by size of primary eye tumor or genetic risk profile, they are deemed to be at high risk of recurrence. We check HLA on everyone just so we have that information because once the patient has that first scan that shows a new lesion, you want to have plan out of the gate. This is likely going to be a good first-line option for all HLA-positive patients. The potential problem is the logistics. There are some patients for whom weekly treatment for whatever reason is not going to be feasible. For those patients, we may need to figure out something else. Because this is a rare cancer, I know you too, we see patients from all over the country and patients who come from all different types of backgrounds and means, and getting weekly treatment can be a lot. The interesting thing though, unlike immune checkpoint inhibition where you worry about baseline autoimmune disease and exacerbating that, obviously there haven’t been any studies yet with tebentafusp. But you could see though that even if someone has an autoimmune disease, because this is so specific, outside of that initial cytokine release general immune stimulation, we may be better set up to treat those kinds of patients just because of the specificity of this drug. I don’t know if you have thoughts about that.

The only thing I would think about is, we’re very careful with surveillance, and we tend to catch patients when they have got low volume recurrence. But in the patient population with significant disease within the liver, I wonder if those patients may benefit from maybe a regional therapy first. I say that because if you look at the data from the 202 trial, the patients who had normal LDH [lactate dehydrogenase], low tumor burden, they are the ones who clearly had the benefit, whereas the hazard ratio crossed 0 for the M1c patients, the larger tumor burden patients. The benefit may be the greatest in the lower tumor burden patients.

Fortunately, like I said it’s not many patients who come to us with greater than 50% of their liver involved out of the gate, but they do exist. I agree, coming from an institution where we do a lot of liver-directed therapy, we always do like to control people’s livers before we get them on some sort of systemic therapy, especially if this drug takes a little while to work. I don’t remember if it was patients you treated, but there were some patients who had significant hepatic disease early on in the 102 study, and they also had some more of the toxicity. In 102 ultimately, there was a bit of a change where we weren’t treating patients with greater than 50% of their liver involved just because of concern about some of the [adverse] effects. You are right, that may be a subset of patients for whom we want to control their liver first, and if we can control their liver, then potentially looking at tebentafusp.

Drs Richard D. Carvajal and Marlana M. Orloff highlight variables that should be considered when assessing the appropriateness for treating patients with uveal melanoma with tebentafusp. 

Uveal Melanoma: Patient Candidacy for Tebentafusp 

The other big category of toxicity is the cytokine release syndrome [CRS]-type symptoms. In addition to counseling patients about the skin toxicity, I always tell them that about 6 hours after the infusion, particularly the first one, you’re going to get fevers, you’re going to get some chills, and your blood pressure is going to start to drop. But it’s all going to go away by 24 hours after induction of infusion. We have looked at the peripheral cytokine changes, and we’ll see it peaks 6 to 8 hours after that infusion, so that correlates with the onset of the fevers and these symptoms. But how have you managed the hypotension in these CRS-type symptoms?

We put everyone on fluids when they come in, about 150 cc an hour, just understanding they may not be drinking a whole bunch while they are there, and then using things like Tylenol or other drugs for fever. Then if we start to see the blood pressure drop a little bit, we fluid challenge them, and sometimes you’re given a couple of boluses and keeping the blood pressure up. I will say that we fortunately, in all the patients we have treated in our institution have not had to use steroids or tocilizumab, which is also used at times for the significant CRS. I have not had any of the really significant ones that make you nervous. We have been able to treat most of our patients with fluids. An important thing is understanding someone’s blood pressure. It’s a bit different if someone comes in and they run a blood pressure of 100 and they drop to 90 [mm Hg] vs they normally run a blood pressure of 170 and they drop to 90 [mm Hg]. It’s knowing when to be aggressive with fluids, but it is important that if you feel like you’re not getting response from fluids, using things like steroids early.

That’s right. We looked at the 102 [IMCgp100-102] experience, and none of the patients required tocilizumab on that study. Steroids were used in fluid refractory patients, but certainly most patients were easily well controlled just with fluid boluses. I have adopted your strategy of starting all patients on standing fluid, and since I have done that, I haven’t had any problems at all.

I can’t imagine that that’s it, but maybe you just have to flood the system a little and give them a little buffer so they don’t get into that danger zone, but you can certainly see really high fever, and we’re talking like 104 °F. For some of these patients who hadn’t had a fever, they can feel really yucky, so you always worry, and you worry about both the blood pressure and also the drop in oxygen. It’s interesting the way cytokine release syndrome is defined; grade 1 CRS is just a fever. Grade 2 is when you have a little bit of blood pressure and a little bit of oxygen need, but it’s responsive to fluids. Then obviously with higher grades you need interventions with things like pressors. It’s being vigilant, exactly as you do, explaining to the patients why they are getting admitted to the hospital, what you are looking for, having a nursing staff that is on top of it, calling you if they notice any change in the vital signs, and being proactive about it.

Advice to help educate patients about symptoms and management of cytokine release syndrome following treatment with tebentafusp for uveal melanoma. 

Tebentafusp for Uveal Melanoma: Cytokine Release Syndrome

When I first started introducing the mechanism of tebe [tebentafusp], I mentioned that gp100 [glycoprotein 100] is in 2 places—on the tumor and in the skin on the melanoma site. There was another interesting abstract that talked about actually coprimary end point of rash with basically overall survival. What did you think about it?

I was intrigued early on before this formal presentation was put out because as you know the majority of the patients will have some sort of skin toxicity, many of them with the first treatment which tends to taper off over time. And many of our patients also develop changes in hair color—white eyelashes, white eyebrows—and anecdotally I thought those patients seem to do well. To have that coprimary end point looking at the 1 week time point if you got rash or not with survival better, and the bottom-line was it was better. If you got a rash, the survival was better. But then if you look at the other, if you look at rash with other variants for outcomes, rash seem to fall out. That is, the patients who develop the rash seem to be a better prognosis group patient population anyway. The data is what it is. I would have liked to have that early marker though and truly significant independent biomarker for earlier response. I was hopeful that would pan out but unfortunately it seems like it did not.

Though again, and I know anecdotes are just anecdotes, but I feel like though those patients that have that significant first week rash, it makes me feel like the drug is doing what it is designed to do. It’s getting where it needs to go. I guess even though, like you said, it may not be an independent biomarker, it can be a little bit helpful even just talking patients through the potential adverse effects basically acknowledging that we want to see at least a little bit of rash. That at least tells us that the drug is doing what it needs to do because again this is weekly. The first 3 doses are given in the hospital for both management of rash and cytokine release, which we can talk about here in a second. But the patients could be pretty miserable with the rash. I tell them, “Hey it’s not for nothing. It is a sign to us that the drug is hitting its mark.” But yes, it would be nice especially because like we just talked about, we are struggling with imaging to help guide us in predicting response and survival, so something really early and obvious like rash would be nice.

: I guess while we’re on the rash topic maybe it’s worth talking about how we manage those adverse effects or what the patterns of the rash are.

When we started, I wasn’t sure what we were going to see. We have all come off the standard immune checkpoint inhibition, dermatitis rash that could be anywhere from just itch with no rash to a full blown almost bullous blistering almost like a Stevens-Johnson [syndrome]. Acknowledging that it was going to be the immune system getting drug over to all the gp100 in the skin. We were prepared for something significant, but at least what I have seen is anything from erythema, just some redness, sometimes it’s more maculopapular and blotchy, but almost like a full body sunburn and skin itself becomes edematous. There are some eyelid and periorbital edema that can happen. I don’t know about you, but I can almost set my watch from the time the drug is infused to when their chin is going to itch. It’s pretty predictable in those patients who are going to get a rash. When it’s going to start, they are going to be miserable for about 24-48 hours. But by the time they show up the next week for the next infusion, it’s gone. Some patients who have a lot of swelling you can see some blistering and dry skin when it goes away. We have used everything from Benadryl to other types of anti-itch topical steroids. I tell patients honestly the best thing is just time.

That’s exactly it. Clearly the severity of the discomfort is worse with the first 2 or maybe 3 infusions, and then maybe there are some lingering dry skin or mild pruritus. But that sunburned exuberant skin tox [toxicity], it’s just for the first few treatments. It’s interesting as I have been treating patients longer and longer and then you see the hair color changes develop later, I started to see the hair color come back, which again is anecdotal.

In those patients where you feel like the color was coming back, are they patients that were still deriving benefit?

: It’s a mixed bag. In some of the cases I was concerned it’s going to precede progression and it has done that in some cases, but in others it hasn’t.

We presented an abstract, specifically MRAE we called it, [or] melanocyte-related adverse event, [on] exactly what you’re saying vitiligo—lightening of the hair, eyebrows, and eyelashes. Vitiligo too, even in standard immune checkpoint inhibitor therapy literature, has been thought of as a positive thing basically acknowledging it’s going after something that is similar to the melanoma. We have shown with tebentafusp that some of those late changes have predicted longer survival. The problem with those is that while rash happens very early those tend to happen late. It may just be that they’re happening because patients are already on drug longer. But we have seen patients who have gone completely white hair and then the color sometimes does come back a little bit. And similar, I am always little worried is this a sign of things to come. But it is another interesting observation.

What to know about treatment response and treatment-related adverse events with tebentafusp for uveal melanoma.

Tebentafusp for Uveal Melanoma: Adverse Events 

You presented nice work trying to tease out the potential confounding effects of post-progression therapies.

This takes us back to an abstract at ASCO 2019 [American Society of Clinical Oncology meeting] where we had observed, and I guess it was how many patients. It was ultimately about 29 patients from a couple of our institutions. A lot of patients that had immune checkpoint inhibitor first got tebentafusp and then got immune checkpoint inhibition afterwards. It did appear that there may have been this re-sensitization of 2 immune checkpoints after exposure to tebentafusp. Again, this was a small cohort of patients but there were patients who say didn’t respond to new checkpoint inhibition, got tebentafusp, and then did respond to immune checkpoint inhibition. Just thinking to ourselves, again going back to the mechanism of action of tebentafusp, if you’re ultimately allowing the immune system to see the cancer maybe then you would ignite some of those natural defense mechanisms like PD-L1 [programmed death-ligand 1] regulation and then you can come in with immune checkpoint inhibition. Interestingly though, in the abstract that you are talking about, it looked like regardless of post tebentafusp treatment the tebentafusp arm still does better. Patients who got tebentafusp then checkpoint inhibition versus investigator’s choice then checkpoint inhibition, the tebentafusp group still does better. But then there is this fancy statistical maneuver that was done that censored patients who got any post therapy and then OS [overall survival] still exist for the tebentafusp arm. Basically, he was trying to model that if you didn’t get patients anything with the OS still stands up versus the immune checkpoint inhibition is that what’s really carrying the prolonged survival in tebentafusp. We really can’t say that. Yes, we ask ourselves is tebentafusp changing the tumor microenvironment? Is it making it more prime for something like checkpoint inhibition? But it may just be doing something else altogether. We really don’t know yet. It’s exciting. 

: It’s really interesting. I am struck. I do have anecdotal cases of some patients who were treated with checkpoint blockade, clear progression, then put them on tebentafusp. They may or may not benefit for a period of time and then I have retreated them and gotten significant radiographic responses that were the same checkpoint inhibitors. They are anecdotes but they are interesting ones.

We all have those anecdotes, which made me a little bit surprised that there wasn’t potentially more to the post CPI [check point inhibitors] after tebentafusp because I thought maybe that’s what was keeping the curve up for prolonged survival, especially for responsive progressive disease.

Potential sequencing strategies with novel therapies such as tebentafusp for uveal melanoma. 

Uveal Melanoma: Sequencing Therapy 

You're right, this is reminiscent to me of the ipilimumab experience where you get the survival benefit despite the low response rate. But even more so here, we've both had patients that we've treated where the radiographic disease course has been remarkable, right? Where it's significant progression and then a flattening of the curve. I'd love to hear your experience of treating patients beyond progression and the benefit that we've seen as presented by Anthony Joshua, MD [head, Department of Medical Oncology, Kinghorn Cancer Centre, Sydney, Australia].

Yes, when you see progression early on and then stabilization, or you see progression early on and then shrinkage in immunotherapy, there's that term pseudoprogression that's been coined, but I don't even think we really know. Without biopsying, which we actually did early on in 102 [IMCgp100-10 phase 2 study], we had a few patients that had really significant progression of some pretty easily accessible tumors. I'm thinking of one as a larger kind of neck mass, and we did see necrosis early on. It's possible that there are patients that either truly have pseudoprogression and it's an immune phenomenon, or it just takes a little while longer to work, because I know that even patients with initially kind of stable disease, sometimes you may see responses first starting at 6 or 9 months—maybe just a delay in response. The interesting thing is there is one abstract at ASCO [American Society of Clinical Oncology meeting] that talked about overall survival in patients whose best overall response was progressive disease. That's it; they had progressive disease from the get-go, and a lot of these patients were treated beyond progressions actually in both arms. There was some treatment beyond progression, but more so in the tebe [tebentafusp.] arm. And their overall survival holds up. They showed that either progressive disease, stable disease, or a PrCR [programmed cell removal] all had better survival than any arm in investigator choice. I think there is something to it. We're still trying to tease that out. Do you have any thoughts about what we could be seeing here?

No, this is a remarkable example of how RECIST [Response Evaluation Criteria in Solid Tumors] fails us. This is a case where we cannot make decisions based upon RECIST. It just does not predict overall benefits. As you brought out, one of the striking curves from Dr Joshua's presentation was that if you look at the tebe survival curve for those just who've progressed, it's higher than the investigator's choice arm, whether they do well or not. And it's just this remarkable decoupling of what the scans are telling us and the true benefit. There's interesting work going on and other changes beyond size that may be able to help us predict which patients are going to benefit or not, but RECIST by itself in this case is not helpful.

Yes, and even what has been coined as kind of IR [interventional radiology] recess, which is basically repeating a scan at 4 weeks, that either confirms what you previously saw or not, also sometimes still doesn't totally help in these scenarios. It is kind of food for thought, moving forward, that we do need to do a better job at using radiology, like you said, radio density of lesions, if there are some other clues we can get. Certainly, we use a lot of MRI [magnetic resonance imaging] of the liver in this. Looking at different aspects of these tumors to help predict which ones maybe more just go into a dormant phase or if there is actually some response that’s happening within the tumors that would then predict that survival.

Some of the challenges then are, I mean over 50% of the patients on 202 were treated beyond progression—and they get benefit—but it’s still unclear to me how to determine how long patients need to be treated beyond progression. Would they have derived the same benefit if we stopped? Because right now, I still have patients on weekly treatment 2 years out. How long do we have to do that for?

That’s a whole other question. Those are patients who you’re saying even had some treatment beyond progression. But even patients who have best overall response of stable disease or PR, weekly treatment is a commitment and there are patients that have gone very long getting this therapy. That’s probably another question we’re going to need to ask ourselves, certainly soon enough.

The last bit on the radiographic findings I found interesting is that the ctDNA [circulating tumor DNA] data that was collected from the 102 trial might be predictive of benefit. I think Dr Joshua presented this as well, where they looked at the patients who progressed and looked at the ctDNA findings. What was interesting is in, a small proportion, in 4 of the patients who had radiographically progressed, they completely cleared their ctDNA. And if you look at the patients who had some odd-fold decrease in the ctDNA, that seem to predict benefit. So maybe there will be ways for us to predict beyond scans.

Yes. And just a little anecdote, again, how scans are kind of what we use to guide ourselves. We see these patients weekly, and you do physical exams, and we have had a handful of patients that something that was palpable was a certain size on the scan, and it was a different size on physical exam about a week later. There is something also dynamically happening in these tumors, so things like ctDNA, where maybe you can map even more consistently, may be helpful in these scenarios because you can’t get scans every week.

Considerations for accurately predicting and identifying response to tebentafusp in uveal melanoma, and insight regarding the appropriate length of treatment for patients with disease progression.

Tebentafusp for Uveal Melanoma: Assessing Treatment Response

: We've had 2 dedicated clinical trials now for the uveal melanoma patient population, the 102 trial and then the 202 trial. The 102 trial was a dose escalation followed by an expansion trial for patients who were previously treated, and the 202 trial was the pivotal randomized trial where patients who were A0201-positive and untreated were randomized on a 2:1 ratio to either tebentafusp or best alternative care. And the best alternative care in this trial was either single-agent PD-1 [programmed cell death protein 1], single-agent CTLA-4 [cytotoxic T-lymphocyte-associated protein 4], or chemotherapy. And in this trial, it turned out that the majority of the patients who were randomized to best alternative care did in fact get single-agent PD-1. But Marlana, I don't know if you want to share some of the high-level big news—the survival advantage for this study.

Yes, even just going back real quick to 102, the patients that were enrolled on that trial were patients who were heavily pretreated, a lot of whom had either prior liver-directed therapy or even prior immune checkpoint inhibitor therapy or chemotherapy. Early on in that setting, there was a 1-year overall survival of potentially 70%-plus in patients who were heavily pretreated. As far as 202, certainly our expectations were high, but we obviously needed to see it in the randomized study. The study survival data had been presented at AACR [American Association for Cancer Research meeting], and the big take-home there was understanding that the IC or investigator choice was a PD-1 dominant group. We're looking at a 1-year overall survival for tebe [tebentafusp.] of 73.2% vs investigator choice, 58.5% with a hazard ratio of 0.51. With that kind of hazard ratio, it was obvious that there was a clear overall survival benefit for tebentafusp. Since that presentation at AACR, there have been a number of presentations teasing out some of the unique aspects of this drug and some of the other trends that we have seen. Rich, do you have any other thoughts about the general 202?

We can’t underestimate how phenomenal these results were. The overall survival of nearly 22 months is just far beyond what I think I would have expected, because based off historical data sets, we would have said the median overall survival's typically about a year, give or take. That 21, 22 months is absolutely phenomenal. And that's in the setting of the hazard ratio of .51, which is incredible, but that's also in the case where the control arm did better than I would have expected.

Yes, especially because PD-1 in uveal, if you look at all the studies, both retrospective and there have been a few prospective series, of a response rate of only about 3% to 5%, mostly in the PD-1 studies. A 1-year OS [overall survival] of 58.5 for basically a PD-1 dominant control group is decent. And there wasn't crossover in these patients, at least at the time of this data, so you ask yourself, OK, what did that patient population ultimately get after the fact? But it wasn't tebentafusp. The bar was high, kind of in both groups, but tebe still outperformed, which is impressive.

And I think the other important thing to highlight is the trial did hit some of its secondary endpoints—for instance, progression free survival, right? The hazard ratio in terms of PFS [progression-free survival] was .73 in favor of tebentafusp, but the medians were not that good. We're talking median PFS was still 3.3 months.

Yes. I remember when ipilimumab came out, and I think that might have been our first inkling that especially with immunotherapies, response rates may not tell us everything. And even PFS may not tell us everything, and I think we're still trying to tease this out. Is the way that we traditionally look at efficacy appropriate for especially unique types of immunotherapies? Is it a problem with imaging? Is response how we describe it on recessed or immune recessed, the appropriate way to be looking at this? Obviously, overall survival is overall survival, and that's kind of the holy grail, but we really use all these other proto-surrogate kind of earlier markers to try to help us predict, but maybe we're not doing such a great job at that, just by looking at things like response rate and PFS.

Implications for treating uveal melanoma with tebentafusp based on responses demonstrated by the IMCgp100-102 and IMCgp100-202 clinical studies.

Tebentafusp Efficacy Data in Uveal Melanoma

Let’s shift gears to tebentafusp, an immunotherapy where there is clear success. I wonder if you can describe for the audience how this drug works? What's its mechanism?

Yes, it goes by a couple of different names now: an ImmTAC or an immune mobilizing monoclonal T-cell receptor against cancer, or simply kind of a TCR [T-cell receptor], but the way that I like to describe it is that it's like a magnet. On one end, it binds a T cell with an anti-CD3 [cluster of differentiation 3] end that would otherwise maybe not necessarily be attracted to the cancer, and then on the other end, it binds the antigen gp100 on the surface of the tumor and basically forces the T cell to engage with the cancer. Again, acknowledging a possible mechanism of innate resistance and just having the immune system kind of not acknowledge uveal melanoma, this is a really high-affinity magnet that just drags the immune system over to the cancer and forces them to engage. That's the mechanism in the simplest terms. As a target, gp100 is pretty uniformly expressed on uveal melanoma. The other place that we see it is melanocytes, most commonly in the skin.

Yes, and with that binding this really optimized T-cell receptor, it's important to highlight that it only recognizes that gp100 protein in the context of HLA-A*0201. It's important for the audience to understand that what is the HLA [human leukocyte antigen] status is how the immune system is able to recognize certain fragments and certain antigens. We think about 40%-50% of the Caucasian population is HLA-A*0201-positive, and this drug was designed only to identify the fragment in that context, so the efficacy of this drug is there just for that portion of the patient population.

It’s important that both providers and patients understand that HLA is basically your immune type. When I describe that to patients, I say, OK, you have a blood type, and your blood type doesn't change. With other immunotherapies, like PD1 inhibitors, looking at biomarkers like PDL1 as something that may predict response, PDL1 may be dynamic, you could rebiopsy and you may see it and you may not, and it may change over time. HLA is one of those things that does not change. It's a simple blood test, but it's like your blood type. I know that patients may get tested, and they may not be A0201 and you may get this question, can we check it again? It's just not something like that. Like you said, it's important because it's that mechanism antigen presentation that interacts with the immune system, and it needs to be there. And the reason they chose this one, as you said, is that it's just the most common in the patient population for which we see uveal melanoma.

Drs Richard D. Carvajal and Marlana M. Orloff comment on tebentafusp’s mechanism of action (MOA) and explain what makes the therapy attractive for uveal melanoma.

Tebentafusp for Uveal Melanoma

Hello, my name is Richard Carvajal. I'm a medical oncologist at Columbia University Irving Medical Center in New York, New York. I'm here today with Dr Orloff, who's an associate professor of medicine at Thomas Jefferson University Hospital. Today, we want to spend some time talking about some really exciting data and some major advances in the field of uveal melanoma, namely the data surrounding a new drug called tebentafusp. Marlana, to start us off, let's set the stage with an overview of uveal melanoma. How would you describe the current state of the field?

For some of our viewers who may not be as familiar with uveal melanoma, it's important to start off by explaining that although uveal melanoma has melanoma in its name, it is very different from cutaneous melanoma. And as many people know, there have been tons of advancements in cutaneous melanoma in the last decades that have totally changed the way in which we treat the disease. Unfortunately, though, that has taught us a lot about the differences with uveal melanoma and the mutations that drive it. Whereas with cutaneous we have

 mutations, in uveal we have different mutations like GNAQ and GNA11. The metastatic patterns in uveal are different; we don't see the lymph node metastasis that you see traditionally early on in cutaneous—rather, we see liver first. What has been most notable is that the immune checkpoint inhibitors have just changed the way we treat cutaneous melanoma; they have not been as effective as they are in uveal. I would say that the current treatment landscape in uveal has to do with 2 major approaches: one being local-regional approaches, liver-directed approaches, because like I said, we do see this disease go into the liver and the liver becoming in many patients, the life-threatening thing, as well as systemic therapies. And in the systemic therapy bucket, immune checkpoint inhibitors are there, things like Opdivo and Yervoy, but we've relied a lot on clinical trials for trying to find effective systemic therapies, whether it's different approaches in immunotherapies or even other types of targeted therapies, again, given the fact that we see those unique G-protein-coupled receptors (GPCR) mutations in uveal melanoma.

: In the field outside of clinical trials, I tend to fall back on combination checkpoint blockade with ipilimumab and nivolumab as the frontline of protocol option if we're not going to do regional therapy. And if we look at the 2 prospective phase 2 clinical trials that have been conducted as well—retrospective data—the best outcomes we see are response rates on the order of 15% to 20% overall survival in the order of 18 to 19 months. And that is just dramatically not as good as what we see with cutaneous disease. I guess the question is: Why do you think that is?

In the simplest way, I like to think of uveal melanoma as just being not easily seen by the immune system. For some reasons when you send these mutational reports on cutaneous melanoma, you'll often get back a number of mutations, a high tumor mutational burden (TMB). We see PDL1 expression, and we just don't see that in uveal. It's typical to just get back a handful of mutations, TMB, tumor mutational burden is often 0 or low, really a lack of PDL1 expression, especially when our metastatic site is the liver. And this has been shown even across other tumor types. I think immune checkpoint inhibitor therapy attacking liver dominant tumors—the liver is a harder organ to crack, if you will. Even in cutaneous, there are patients who have bad liver-dominant disease, and sometimes they may not be the ones that respond best to immune checkpoint inhibition. Unfortunately, there are a lot of cards stacked against us in uveal, but there are probably also mechanisms that we aren't even aware of yet. But certainly, like you said, when you look, compare response rates, durability of response, it's not as good as it is in cutaneous.

Yes, and when we started doing single-agent checkpoint blockade—now that we have the data with combination checkpoint blockade—even though we do it, for a while we were saying this is a nonimmune-responsive disease. For a while many of us thought maybe this is just one of those tumors that is an immune desert. Then, over time, with the collection of biopsies that we've done, work from The University of Texas MD Anderson Cancer Center and elsewhere, just looking at the microenvironment, it just turns out that there are immune cells there, and it's not a desert, but the immunological milieu is just not sufficient or not adequate to engender a response, and maybe that's particularly true in the liver, which is just an immunosuppressive organ.

Yes, and you're right; even when you look at primary uveal melanoma in enucleated specimens, you can see T cells there. When I said before that the immune system is just not as attracted to uveal melanoma in the most kind of endgame sense, it may be that there are just other things, like you said, in the tumor microenvironment that just kind of set the immune system up for failure, that even if it gets there, there are these other resistance mechanisms that we've yet to overcome with traditional immune checkpoint inhibition.

: Yes, and then looking at the data, for instance with TILs (tumor-infiltrating lymphocytes), those are early data that suggested to me that maybe it's not immune resistant, but maybe we're just targeting uveal the wrong way in terms of trying to manipulate the immune system.

Yes, I was trying to simplify it in my head that if standard immune checkpoint inhibition isn't going to get the job done, it's about either making the tumor more of a target to the immune system or somehow forcing the immune system over to the target that it otherwise wouldn't be able to get to. And that's certainly—which I know we'll talk about in a minute—one of the really unique things about the drug tebentafusp and acknowledging that mechanism of resistance and trying to do essentially what we haven't been super-successful with yet with immune checkpoint inhibition.

A brief explanation regarding the difference between uveal melanoma and skin melanoma and an overview of standard treatment approaches used to treat patients with uveal melanoma.

Richard D. Carvajal, MD, Columbia University Irving Medical Center

Articles