Tebentafusp Offers Hope, But Non–HLA-Restricted Options Remain Key in Metastatic Uveal Melanoma

Five-year survival data with tebentafusp-tebn (Kimmtrak) show that patients with HLA-A*02:01–positive metastatic uveal melanoma can live for years with their disease, but most patients still need additional options, making targeted approaches within the HLA-A*02:01–negative population, such as the investigational combination of darovasertib plus crizotinib (Xalkori), essential to closing the gap, according to Alexander N. Shoushtari, MD.

In the final analysis of the phase 3 IMCgp100-202 trial (NCT03070392), conducted after a minimum follow-up of 5 years, previously untreated patients with HLA-A*02:01–positive metastatic uveal melanoma who received tebentafusp (n = 252) achieved a median overall survival (OS) of 21.6 months (95% CI, 19.0-24.3) vs 16.9 months (95% CI, 12.9-19.5) with investigator’s choice of pembrolizumab (Keytruda), ipilimumab (Yervoy), or dacarbazine (n = 126; HR, 0.67; 95% CI, 0.54-0.85).¹ The 5-year OS rates were 16% (95% CI, 11%-21%) and 8% (95% CI, 4%-14%), respectively. These 5-year data were first presented at the 2026 AACR Annual Meeting. In an exploratory analysis of the tebentafusp arm, patients who cleared circulating tumor DNA (ctDNA) by week 9 (n = 45) had a median OS of 29.6 months (95% CI, 23.0-37.0) vs 10.2 months (95% CI, 8.3-13.2) among those who did not (n = 78; HR, 0.32; 95% CI, 0.21-0.49).¹

For patients with HLA-A*02:01–negative disease, first-line darovasertib plus crizotinib (n = 210) improved median progression-free survival (PFS) by blinded independent central review to 6.9 months (95% CI, 5.6-8.3) vs 3.1 months (95% CI, 1.8-4.2) with investigator’s choice of pembrolizumab, nivolumab (Opdivo) plus ipilimumab, or dacarbazine (n = 103; HR, 0.42; 95% CI, 0.30-0.59; P < .0001) in the phase 2/3 OptimUM-02 trial (NCT05987332), according to data presented at the 2026 ASCO Annual Meeting.² The objective response rate was 37.1% (95% CI, 30.6%-44.1%) vs 5.8% (95% CI, 2.2%-12.3%), respectively. In the third quarter of 2026, IDEAYA Biosciences completed a new drug application submission to the FDA to support potential US approval of the combination in metastatic uveal melanoma.³

“It’s very nice to have very specific treatments for [the HLA-A*02:01 population], but even [for those patients], the vast majority will need another treatment down the road. The rising tide will lift all boats when [a treatment] is not HLA specific,” Shoushtari said in an exclusive interview with OncLive®.

In the interview, Shoushtari discussed how he frames the 5-year tebentafusp data for patients, the role of ctDNA in interpreting imaging, how HLA-A*02:01 status shapes treatment selection, the biology that sets uveal melanoma apart from cutaneous melanoma, and the potential of darovasertib in both metastatic and primary uveal melanoma.

Shoushtari is an associate attending physician on the Melanoma Service and the Cellular Therapy Service at Memorial Sloan Kettering Cancer Center in New York, New York.

OncLive: How do you balance the realistic hope offered by the 5-year IMCgp100-202 data with tebentafusp against the reality that many patients still die of metastatic uveal melanoma?

Shoushtari: The duality is a good way to put it. As a medical oncologist, [I am] in front of someone who’s generally in a great deal of stress, sadness, shock, anger, [or] any mix of emotions. It is a boon to tell someone, “Listen, this is not the death sentence that you imagine it might be,” which some people still imagine when they get a stage IV diagnosis.

But we are not yet at the point in uveal melanoma that we are starting to get to [with] some [patients with] cutaneous melanoma, where you can actively reframe their mindset and say, “Don’t worry so much; the majority of people will live 5 years with stage IV disease if you have sun-exposed melanoma.”

How can ctDNA help interpret imaging in patients receiving tebentafusp, and what will it take to incorporate it into real-time treatment decisions?

Broadly across solid tumor oncology, more physicians and, increasingly, patients [are] motivated to check cell-free DNA [cfDNA] as a way to look for residual disease in some cases. Metastatic uveal melanoma is one of the better-studied scenarios where cfDNA could be helpful. We are still limited; not everybody has detectable ctDNA or cfDNA at baseline. But it can often be helpful [with] agents like tebentafusp, where you know that [patients] can benefit even when the scan shows not a lot of shrinkage or even [a] slight increase in [the] size of some lesions.

In terms of where we need it to go, which is incorporating this into trials and proving that these decisions are as good as, if not better than, using our scans, we’re still a little bit lagging. It’s hard to do that. You need a very large study to say we are going to use ctDNA as one of our primary or co-primary end points. So far, other investigators are incorporating it, and we’re doing our best to use it as an embedded, usually secondary, aim. But it’s hard to get ctDNA in a large phase 3 trial [and say], “We’re going to use this contemporaneously, in real time, [and] give it back to the physicians to make the decision along with their patients. We will give that data as fast as a CT scan or an MRI gives us data, and then we’re going to make a better decision.” That’s what’s been a little bit missing in our disease, and it’s not for lack of trying.

Those days will come, but right now we’re still embedding it largely on the back end. We’re getting it months later, pooled and analyzed after the study is over, and saying, “Look how promising this treatment is.” That’s coming, but we have yet to get high-quality data to say this is better than RECIST.

As a result, in my own practice, I can’t say that I universally use it every single time. It can sometimes be useful, but sometimes the scans really do speak for themselves, or if [patients] had symptoms at the start and those symptoms improve, that speaks for itself. We’re still, unfortunately, in this heterogeneous situation where we don’t have clear guidance that everybody has to use it, but it’s nice to have it available for select cases.

How does HLA-A*02:01 status shape your treatment approach in metastatic uveal melanoma, and what does it mean for patients who are not eligible for HLA-restricted therapies?

HLA-A*02:01 has a big role in the clinic, with the FDA approval of tebentafusp and ongoing clinical trials targeting different antigens, like PRAME, that are HLA specific. PRAME [itself] is not HLA specific, but targeting it right now is being done only in [patients with] HLA-A*02:01.

This is a weakness for us as physicians. We have to acknowledge to some patients, “Listen, we’re hoping [you have] 02:01, because it adds the option of tebentafusp, or it may add a clinical trial to our option list.” But even in the most enriched population for 02:01, which is typically in Northern and Western Europe, it’s about 60% of people that don’t have that allele. Where I practice in New York City, it’s a little bit lower than that; probably only about 30% to 35% have it. I was just talking to my colleagues in Brazil, and for them it’s 20%. East Asia is also quite low.

It’s a weakness to not be able to offer every treatment that you want to every single patient. At the same time, as we’ve said, we’re still not doing well enough with 02:01, so we need to improve that anyway. We continue to study 02:01 and hope that down the road, some of those lessons can spill over into other HLA types. But realistically, treatments that are HLA specific are years down the road for non-02:01 [types], at least in melanoma.

As physicians, we have a moral imperative to continue to push treatments from companies that have a good track record and a scientific basis that we know are not HLA specific, and say, “This is really important because this can reach everybody.” That’s more short term than saying, “02:01 works; now let’s use the same approach in 03:01 or 01:01 or 29:02.” That’s years and years away, unfortunately, because it’s a different molecule, and it’s going to take years to prove that.

How should clinicians understand the biological divide between uveal melanoma, which is driven almost entirely by GNAQ and GNA11 mutations, and cutaneous melanoma, and why has it put uveal melanoma drug development on a separate track?

The perspective I use scientifically, and in explaining it to patients, is that every cancer has something wrong in its growth signal pathway that allowed it to become a tumor and allowed it to spread and grow. It’s just a question of knowing what that trigger is; we may have good treatments for it, or we may not. That happens in skin melanoma, too.

But it has historically been a huge issue [in uveal melanoma, where], as you say, almost everyone has GNAQ, GNA11, or a little less common [alteration] like CYSLTR2. Up until very recently, we have not really had a targeted treatment that can shrink or stabilize disease in a way that’s tolerable and durable. Of course, with recent data using darovasertib with crizotinib, we’re starting to target, in that case, the PKC and MET signaling pathways that are activated by GNAQ and GNA11.

In contrast, 99% of cutaneous melanomas do not use [the same signaling pathways] to send their growth signals, and PKC inhibition and MET inhibition are not going to be effective for 99% of those cutaneous melanomas. In that sense, it’s nice to see dedicated trials targeting the mutations that are enriched in uveal melanomas, or in the 1% of skin melanomas that have those mutations, which are called malignant blue nevus–type tumors. They have a certain histologic appearance and genomic subsets. We are carving them out, finally, with some treatments that can start to be helpful.

How do you interpret the magnitude of PFS benefit seen with darovasertib plus crizotinib in OptimUM-02, and where could the combination fit alongside tebentafusp in closing the gap for patients with HLA-A*02:01–negative disease?

The magnitude of PFS benefit with darovasertib and crizotinib is extremely significant. It is, by any sense in modern clinical oncology, an advance. It’s not a subtle difference. If you’re a patient, you’re not just looking at averages and saying, “If I’m on immunotherapy, the average is going to be 3 months on treatment or something, and then it’s only going up to 7 months on treatment.” I don’t think that’s how we should explain the data to our patients.

Average time to growth is useful. It’s a useful research tool; it helps us understand if treatments are working. [But] in front of a person who’s dealing with this problem, they are one person, and they are going to want to know, “How long can this work for me, realistically?” To that extent, with more time, we look at the right end of that curve and say, “What are the odds that you’re on this for 1 year or 2 years?” For those types of benefiters, it’s life-changing to be on a treatment and be told, “You have a chance to be on this treatment for 1 year or 2 years,” which is what tebentafusp really did, too. Having one more tool in our toolkit to offer patients that is not HLA restricted is helpful.

Darovasertib plus crizotinib, like any cancer medication, is not for everybody. There are specific weaknesses of that regimen that you need to talk to your doctor about to make sure that you’re fit and ready, but that’s the case with every cancer treatment. Hopefully [there will be] an FDA approval in 2027, which is what we’re all openly rooting for, and I would be shocked if that didn’t happen, quite frankly. It’s going to be a great option if it becomes available.

How significant would neoadjuvant darovasertib be for globe salvage and vision preservation in patients with primary uveal melanoma?

There is a significant minority of people who, for example, can’t get radiation; they need their eye removed. The idea of finally having a medication that can quickly shrink the tumor and allow for saving the eye and radiating the eye, or, theoretically, if you were going to radiate this much of your visual field, maybe radiating less of your visual field and improving vision long term, would be really clinically useful.

As a field, we’re excited that that’s being investigated. This is [the] typical path for lots of drugs in all types of solid tumors: If it works in the advanced setting, let’s try to get it to people sooner in their disease course and help more people earlier on. Now that we have this tool, it’s one of the first tools that the field believes, a little bit more broadly, could be effective in the neoadjuvant setting. Hopefully this is the first of many more in my career to improve and iterate and say, “It worked in stage IV; let’s bring it to more people in earlier stages.”

What gives you the most hope about where the uveal melanoma field is heading?

Compared with the beginning of my career, a little over a decade ago, the pace of improvement is accelerating for uveal melanoma. That really gives us on the medical side hope, and I hope that it gives patients and their loved ones some hope as well.

References

1. Piperno-Neumann S, Rutkowski P, Hassel JC, et al. Five-year survival with tebentafusp in metastatic uveal melanoma. Ann Oncol. 2026;37(9):1266-1277. doi:10.1016/j.annonc.2026.05.695
2. Orloff MM, Ramelyte E, Butler MO, et al. Darovasertib plus crizotinib vs investigator’s choice as first-line treatment for patients with HLA-A2 negative metastatic uveal melanoma: primary results from the OptimUM-02 trial. J Clin Oncol. 2026;44(suppl 17):LBA9503. doi:10.1200/JCO.2026.44.17_suppl.LBA9503
3. IDEAYA Biosciences reports financial results for the third quarter 2026 and provides a business update. News release. IDEAYA Biosciences. November 3, 2026. Accessed November 12, 2026.