The 5-year data from the phase III IMCgp100-202 study demonstrate durable, long-term overall survival (OS) benefit with the novel agent tebentafusp—a first-in-class, bispecific, soluble T-cell receptor (TCR) therapeutic—in individuals with previously untreated metastatic uveal melanoma (UM).
At 5 years, median OS was longer with tebentafusp than with investigator’s choice (IC; 21.6 vs 16.9 months) in the intention-to-treat population. A comparison between groups yielded a stratified hazard ratio (HR) of 0.67.
“There was an early separation between the Kaplan-Meier curves, which was sustained throughout the duration of the study. At the 5-year mark, there was a doubling of OS from 8 percent with IC to 16 percent with tebentafusp,” said Dr Paul Nathan from Mount Vernon Cancer Centre, Northwood, UK, at AACR 2026.
Of note, 44 percent of tebentafusp survivors only received tebentafusp, while 86 percent of IC survivors crossed over to receive tebentafusp at a later date.
The OS benefit with tebentafusp was sustained even after removing the effect of subsequent therapy (HR, 0.52). [AACR 2026, abstract CT029]
Key subgroups
“The benefit or trend-to-benefit favouring tebentafusp over IC was evident across most subgroups,” Nathan said. The OS benefit was observed irrespective of disease location, tumour burden, metastatic lesion size, alkaline phosphatase (ALP) level, and lactate dehydrogenase level at baseline.
According to Nathan, UM is highly hepatotropic. In the study, 95 percent of participants had liver involvement at baseline, and tebentafusp showed benefit regardless of whether the disease was confined to the liver (unstratified HR, 0.57) or had spread to extrahepatic sites (unstratified HR, 0.70).
The most pronounced treatment effect was observed in participants with a tumour burden <3 cm (unstratified HR, 0.54) and those with baseline ALP greater than the upper limit of normal (unstratified HR, 0.56).
Even patients with known poor prognostic factors (eg, large tumours [≥10 cm]) had an OS benefit (unstratified HR, 0.70).
Other endpoints
Tebentafusp also outperformed IC in disease control rate (46 percent vs 27 percent). Eleven percent of participants in the tebentafusp group had a complete/partial response, whereas in the IC group, only 5 percent had a partial response.
Among patients whose best objective response was progressive disease, tebentafusp-treated patients did substantially better than those who received IC (median OS 15.1 vs 10.1 months; stratified HR, 0.61).
ctDNA: Robust indicator of benefit
Sixty-one percent of participants had detectable ctDNA at baseline. Of these, 81 percent had a molecular response at week 9 (37 percent had ctDNA clearance and 44 percent had ≥50-percent reduction).
The molecular response rates in the partial response (PR) and stable disease (SD) groups were 93 percent and 86 percent, respectively.
“In the subgroup of patients within the SD group who experienced any tumour reduction that was not [sufficient] enough to reach PR, they did just as well as the partial responders (molecular response rate 92 percent). This means that any ctDNA reduction results in benefit,” Nathan explained.
Nearly three-quarters (71 percent) of tebentafusp-treated patients who had survived ≥5 years had undetectable ctDNA at baseline; the remaining 29 percent achieved clearance by week 9 on treatment.
Median OS was longer among participants with undetectable vs detectable ctDNA at baseline (27.3 vs 18.4 months; HR, 0.50), among those who did vs did not clear their ctDNA by week 9 (29.6 vs 10.2 months; HR, 0.32), and among those with ≥50 vs <50 percent reduction in ctDNA levels by week 9 (19.8 vs 9.1 months; HR, 0.41).
Of note, the median OS was similar between patients who achieved ctDNA clearance at week 9 and those who had undetectable ctDNA at baseline. “[This implies that] if we can get rid of ctDNA at 9 weeks, the patient will do just as well as if he did not have detectable ctDNA at baseline,” Nathan explained.
“[Taken together, these findings suggest that] ctDNA is a more sensitive and robust marker of tebentafusp activity than radiographic response,” Nathan said.
When asked about the clinical implications of ctDNA in this setting, Nathan noted that it can be useful when discussing with patients at baseline. “If a patient has undetectable ctDNA at baseline, you know that their prognosis is superior to that of untreated patients with detectable ctDNA at baseline.”
“The second is, if there is ctDNA reduction or normalization at week 9, you can look your patient in the eye and say, ‘Indeed, we have evidence that tebentafusp is having a desirable effect on your [condition]’,” he continued.
Post-progression survival
In tebentafusp recipients who had treatment beyond radiographic progression (TBP; n=139), 27 percent (n=38) had post-progressive disease reduction; only 1 of 28 patients in the IC group did so.
However, selection bias was possible because patients in this subgroup tended to have favourable baseline characteristics.
After adjusting for differences in covariates using a stepwise selection Cox model, tebentafusp remained an important factor in the survival benefit among patients who had TBP, even after accounting for baseline covariates (adjusted HR, 0.61).
Two important ‘firsts’
In a previous report, UM was referred to as a “relatively rare but deadly cancer.” Despite the satisfactory local disease control with first-line treatment alternatives (ie, resection, radiation, enucleation), long-term survival remain uncertain, with a risk for liver metastasis. Hence, there remains an unmet need for early treatment of small melanocytic lesions or melanomas in the eye to achieve disease control, preserve vision, prevent metastases, and improve survival. [Eye (Lond) 2017;31:241-257]
“IMCgp100-202 is a practice-changing international trial that represents two important ‘firsts’: It is the first positive phase III study in metastatic UM, a refractory disease that is distinct from cutaneous melanoma. Second, tebentafusp is the first TCR therapeutic to demonstrate a survival benefit,” Nathan said.
This open-label trial included 378 patients with advanced UM who were HLA-A*02:01-positive and had no prior liver-directed therapy (except surgery) or systemic therapy in the advanced setting. The participants were randomized 2:1 to receive tebentafusp or IC of pembrolizumab, ipilimumab, or dacarbazine. Pembrolizumab was the most common IC (82 percent).
These results reinforce those reported in the primary analysis (HR for death, 0.51; p<0.001; HR for disease progression or death, 0.73; p=0.01) and the updated OS (HR for death, 0.68) and safety analyses at 3 years. [N Engl J Med 2021;385:1196-1206; N Engl J Med 2023;389:2256-2266]