Early ctDNA Testing Provides Another Tool for Predicting Survival Outcomes With Tebentafusp in Metastatic Uveal Melanoma

The incorporation of circulating tumor DNA (ctDNA) testing into standard practice for patients with metastatic uveal melanoma receiving tebentafusp-tebn (Kimmtrak) may provide a more accurate assessment of the agent’s efficacy compared with radiographic response, and better inform the decision to continue or discontinue treatment, according to Ryan J. Sullivan, MD.1

According to prior results from the phase 2 trial IMCgp100-102 trial (NCT02570308), the bispecific T-cell engager tebentafusp improved overall survival (OS) in patients with previously treated metastatic uveal melanoma regardless of their RECIST v1.1 best response. Subsequent exploratory analysis suggested that early reduction in ctDNA levels was a better indicator of patients’ long-term OS benefit than radiographic-based response criteria.2

Analysis of ctDNA in the phase 3 IMCgp100-202 trial (NCT03070392) of patients with previously untreated metastatic uveal melanoma receiving first-line tebentafusp confirmed that a deeper reduction in ctDNA levels was associated with longer OS.Of the 123 evaluable patients with detectable ctDNA (n = 123/202) at baseline, 88% experienced ctDNA reduction by week 9 of treatment. Notably, these patients achieved an overall response rate equating to only 10%. One-year OS rates for patients with undetectable and detectable ctDNA were 90% and 62%, respectively (HR, 0.43; 95% CI, 0.28-0.67). These findings were presented at the 2023 AACR Annual Meeting.1

“The most important takeaway from the data we presented is that ctDNA [reduction] seems to be more reflective of [OS for] a patient with uveal melanoma starting on tebentafusp than their first scan,” said Sullivan, an associate professor of medicine at Harvard Medical School and associate director of the Melanoma Program at Massachusetts General Cancer in Boston, Massachusetts. “[This] justifies the use of ctDNA as a marker to monitor patients who are on tebentafusp.”

In an interview with OncLive®, Sullivan discussed previous research on tebentafusp in uveal melanoma that led to the inception of this phase 3 trial, clinical implications for data confirming the predictive value of ctDNA levels, and future avenues for research that could clarify and improve the adoption of early ctDNA testing as a standard practice in uveal melanoma.

OncLive: Could you provide some background on this phase 3 trial? What previous research prompted this analysis of the relationship between ctDNA and survival outcomes in patients with previously untreated metastatic uveal melanoma?

Sullivan: Tebentafusp is [part of] a new class of molecule. On one end, [it] acts like a T-cell receptor and binds to a specific protein on tumor [cells] called gp100, and on the other end of the molecule is a T-cell engager. I like to describe this agent as an “evil matchmaker,” [because] it binds to cancer and then brings in the cell that’s going to kill it. The interesting thing about this drug, [which is both] cool and challenging, [is that] you measure [its] benefit [through] OS [improvement].

The randomized [phase 3 IMCgp100-202 trial] presented at the [2021] AACR Meeting led to the FDA approval of the drug in January of 2022. The challenge [with these] data, however, is that most patients who benefit [from tebentafusp] aren’t having radiographic responses. It then becomes very difficult to know [whether a patient] whose scans aren’t better, or even look a little bit worse [is] still benefitting from the drug.

This trial [evaluated] tebentafusp vs investigator’s choice of either the PD-1 inhibitor pembrolizumab [Keytruda], the CTLA-4 inhibitor ipilimumab [Yervoy], or the chemotherapy dacarbazine. When you compare the outcomes of patients [who had a] response, they did equally well [with] tebentafusp [vs] the control therapy. That’s not entirely surprising, [as] patients who respond to treatment generally do better than patients who don’t respond to treatment. The real difference in the survival curves was in the patients who had stable disease or progressive disease, suggesting that there’s something more than just having a scan that’s reflective of benefit. The point of the work that we presented [at this year’s meeting] was to determine whether we could use ctDNA as an indicator of benefit that we’re not getting from the scans.

There were [also] some data that had previously been presented from the phase 2 [IMCgp100-102] study of tebentafusp, which enrolled patients who had uveal melanoma and had previously been treated with other therapies. From that dataset, the investigators looked at ctDNA at baseline and on treatment to see if there was any significant reduction in ctDNA, and whether that correlated with benefit. [The data confirmed that] it does. Interestingly, [a majority] of patients had detectable ctDNA in that setting. Clearance of ctDNA was associated with the best outcomes, and 50% clearance was associated with better outcomes. [A ctDNA clearance rate] less than that was associated with [worse] outcomes. Patients [who] didn’t have any reduction of ctDNA did quite poorly.

What findings from the phase 3 study were presented at the 2023 AACR Annual Meeting?

In this randomized study, we looked at the patients who received tebentafusp. Of those [202] patients, 61% had detectable ctDNA at baseline. We measured [ctDNA] again at 9 weeks. The platform we used was a targeted exome platform across a number of hotspot mutations that are common in uveal melanoma. A positive [result] was based on results of this commercial assay that have been published before.

Ultimately, we found that most patients had some reduction in ctDNA. Of those that had a reduction, [48% and 27% of patients] had either greater than 50% reduction or total clearance, [respectively]. About 30% of patients had clearance in their ctDNA. If you cleared your ctDNA, you did way better than if you didn’t clear your ctDNA [in terms of] OS. You are much more likely to survive longer if your ctDNA [decreased to] 0 after being detectable at baseline. In patients who had progressive disease, a 50% reduction in ctDNA level was associated with a significant difference in [OS vs] patients who didn’t have [a 50% reduction in their ctDNA level].

Given these results, how could the use of ctDNA testing in clinical practice better inform the continuation of tebentafusp in patients with uveal melanoma?

[Testing for] ctDNA is becoming increasingly accepted as a standard for patients with certain malignancies. These data support its use in uveal melanoma, particularly in patients who are being treated with tebentafusp. [These data] suggest that if we have a patient in front of us who’s receiving tebentafusp and their scans show new or growing disease, but their ctDNA is significantly lower, it probably makes sense to continue therapy for a little bit longer.

We don’t have data suggesting that continued monitoring of ctDNA should be done, but it makes sense that it should be done in clinical practice. I monitor ctDNA [levels] using commercial assays in my patients on tebentafusp. That additional data helps [me] decide [whether] it is time to move on to something else, or if we should keep pushing [the current treatment] a bit further. Patients may continue to receive benefit from a therapy that we can’t detect by imaging.

Ultimately, this work should help clinicians make decisions with their patients about whether to continue a drug that’s [perhaps] radiographically disappointing. We [can more assuredly tell patients] that [the scan] may seem disappointing, but that there are data suggesting that [they] are probably benefiting [from tebentafusp] because of the significant degree of reduction in ctDNA [levels], and maybe [they] should keep [receiving] it.

Were these findings consistent with expectations according to previous research?

Previous research suggests that ctDNA is reflective of tumor burden. [This is] not necessarily in uveal melanoma, but for cancer in general. Commonly, a reduction in ctDNA will be [associated with] a reduction in tumors in a scan down the road. [However], with immunotherapy, sometimes there’s a mismatch, where the direction of the ctDNA goes down, but the scans [initially] look a bit worse. Often, the scans catch up to the ctDNA [further in the disease course]. That phenomenon is called pseudoprogression, which you sometimes see with immune checkpoint inhibitors. The data are consistent with that phenomenon because the drug is associated with OS benefit but not response rate benefit, and there are clearly some patients who benefit from treatment beyond progression.

I’m not sure that’s what we’re seeing with tebentafusp. We just don’t have enough biopsies and data to understand what we’re seeing. [We have] tumors that are growing, but patients who are surviving longer and having a reduction of ctDNA. It’s something that could be studied further.

There are patients who are experiencing more benefit than what the scan is showing us. We can use a different marker of tumor burden, in this case ctDNA, to help select the patients who would be most likely to continue to benefit from therapy. [This outcome] is not in line with prior data, but it’s consistent with what we thought we’d see based on what we know about the drug and the data from previous trials with tebentafusp in patients with uveal melanoma.

What remaining questions regarding the utility of ctDNA testing in this patient population still need to be addressed through continued investigation?

[There are] a few important things to follow-up [with regarding] this work. First, a longer [time frame for the] collection of ctDNA is important. That [could look like] an industry-sponsored trial, or academic investigators who are treating patients, collecting [data] at specific time points, and then analyzing the ctDNA.

There are a few unanswered questions, including the best timing of ctDNA [evaluation]. We [collected data at] 9 weeks [in this trial], because that’s when we had the most samples, and it was around [the time] when patients were getting scans. When I follow a patient’s ctDNA in clinical practice, I do it a few weeks before their scans, because I’ll get [the results] back around the same time. When I sit down with a patient, I [can] say, ‘This is what your scan shows, this is what your ctDNA [levels] show, and this is what we’re going to do.’ It’d be nice to understand the biological nature of the reduction. Does it go down quickly? Does it take a little while to go down? Over time, can we use [ctDNA levels] to predict [which] patients are going to continue to benefit [from treatment]? We won’t be able to answer that with this dataset, but we can answer that in other studies.

The other thing we need to think about is [whether] there are other tools we can use, [such as] scanning modalities or PET tracers, that may help us to better understand what’s happening [when] a patient is predicted to have additional survival benefit based on their ctDNA, [but their] tumor is not shrinking. There are some cool PET tracers that are being developed against CD8, granzyme B, or other factors that might be reflective of an immunotherapy response. We’re interested in applying those tools to our patients who are receiving tebentafusp. That’s going to be an important part of future endeavors, as well. Those [tests] could all be done together. Any study looking at novel imaging markers needs to include ctDNA [testing], as part of [their study design] based on the data that we’ve presented, and data presented in the prior trials of tebentafusp. That’s the most important next step that can be taken.

What should your colleagues ultimately take away from this presentation?

Perhaps [ctDNA testing] should not [be the] sole justification to treat [a patient] beyond progression, but it’s another useful tool [to inform this decision]. This was a large, randomized phase 3 trial, and it [may be] the best data set we’re going to have for frontline tebentafusp. This biomarker can help prognosticate OS early on in treatment, and potentially predict continued benefit from continued treatment [with tebentafusp].

[These results] support the use of [ctDNA testing] in standard clinical practice. I’m not sure we’re going to have a better data set to justify it [going forward]. Ultimately, it comes down to [whether] we can get it paid for, because it’s not necessarily cheap. If I was an insurance company trying to determine whether to continue to pay for tebentafusp [when] the scan showed [disease] progression, I’d want some other indication to [justify my decision]. More importantly, if I was a patient [who] may not have a lot of time left, I want to know that the treatment I’m getting is potentially helping me. It’s a lot easier to [decide to] stay on the treatment if my scans look good. If my scans don’t look good, I don’t want to stay on a therapy that’s not healthy. I [would] want to go to something different. From a patient and a provider standpoint, we want to know that we’re giving a drug [that will benefit the patient], particularly after there’s some evidence suggesting it’s not working. We want something to [confirm that] we need to continue [this treatment].

The take-home message is that we can use this tool to help mitigate the concern [stemming] from a scan that doesn’t look as good as we’d hoped and justify continuing a therapy that is biologically helping that patient.

What other research were you excited to see presented at this year’s meeting?

[As] an investigator in the study, I was excited to see data [from the phase 2b KEYNOTE-942 trial (NCT03897881) of] the Moderna personalized cancer vaccine [mRNA-4157/V940] and pembrolizumab [Keytruda] vs pembrolizumab [alone] in the stage III adjuvant setting for patients with melanoma.

[This vaccine] is a long time coming. There’s been a lot of phase 1 and single-arm data with personalized cancer vaccination, but it’s hard to understand whether it’s helping or not. You can look at T-cell subsets, and whether you’re getting an immunologic response to that type of therapy. [However, we] now see proof of concept in melanoma, and in the field in general, that neoantigen vaccination can be associated with better outcomes. There’s a lot more to be done before [this option becomes] a standard of care [for] patients, but that was the most exciting [implication]. Having been part of some phase 1 and phase 2 trials for a few of these [cancer vaccines], it’s gratifying to see that [this approach] works. That field will continue to build upon this work.

Disclosure: Dr Sullivan reports working as an independent contractor for Novartis, Pfizer, Bristol Myers Squibb, and Merck; he also reports a grant/contract with Merck.


  1. Sullivan R, Collins L, Rodrigues M, et al. Early ctDNA reduction is associated with better overall survival in the Ph 3 trial of tebentafusp in previously untreated metastatic uveal melanoma. Cancer Res. 2023;83(suppl 7):1035. doi:10.1158/1538-7445.AM2023-1035
  2. Carvajal RD, Butler MO, Shoushtari AN, et al. Clinical and molecular response to tebentafusp in previously treated patients with metastatic uveal melanoma: a phase 2 trial. Nat Med. 2022;28(11):2364-2373. doi:10.1038/s41591-022-02015-7

Leave a Reply

Your email address will not be published. Required fields are marked *

Skip to content