ESMO 2025 – no starring role for Exelixis’s Cabometyx follow-on

Exelixis’s Cabometyx successor zanzalintinib has succeeded in the pivotal Stellar-303 relapsed colorectal cancer trial, ESMO heard on Monday, but the discussant said more information on the project was needed to understand the asset’s role in this setting.

Toxicity looked problematic, with five treatment-related deaths in the active cohort of the trial, which combined zanzalintinib with Roche’s Tecentriq. This is particularly concerning as the main aim with zanzalintinib is to improve on the safety profile of Cabometyx – and the latest data raise more doubts about whether it can do so.

The ESMO discussant, Vall d'Hebron University Hospital’s Dr Elena Elez, said she wanted to understand the toxicity profile of zanzalintinib better, and also to see quality-of-life data, as well as mature results from a key subgroup, patients without liver metastases.

This group is where zanzalintinib has produced the biggest benefit so far. Stellar-303, which tested zanzalintinib plus Tecentriq versus Stivarga in patients with third-line non-MSI-high colorectal cancer, once had a primary endpoint of overall survival in patients without liver mets. However, this was changed to OS in an intent-to-treat population, which also included those with liver mets – something that looked like an attempt by Exelixis to gain a broad label.

The company toplined a win in Stellar-303 in June, and the data presented at ESMO, and published simultaneously in The Lancet, concerned the final OS analysis in the ITT population. Here, median OS was 10.9 months with zanzalintinib plus Tecentriq, versus 9.4 months with Stivarga, representing a 20% reduction in the risk of death that was statistically significant with a p value of 0.0045.

OS in patients without liver mets is a co-primary endpoint. Data here are still immature, but it’s apparent that the ITT benefit is being driven by this subgroup. Median OS was 15.9 months with the zanzalintinib/Tecentriq combo, and 12.8 months with Stivarga.

Meanwhile, the corresponding numbers in patients with liver mets were 8.9 and 7.7 months respectively.

OS data in Stellar-303 in third-line non-MSI-high colorectal cancer

Source: Dr Anwaar Saeed & ESMO 2025.

On the question of whether zanzalintinib plus Tecentriq will become a new option for patients with previously treated colorectal cancer, Elez concluded that it “could be”. She added, though, that the final data in patients without liver mets would be needed.

She also flagged an increase in grade 3 treatment-related adverse events with zanzalintinib plus Tecentriq, at 56% versus 33% with Stivarga. Grade 3 or higher hypertension was seen in 15% of patients receiving the combo.

As for the aforementioned deaths, there were two cases of intestinal perforation that were deemed related to zanzalintinib, and one due to “altered state of consciousness” deemed related to the combo. There were also two deaths deemed related to Tecentriq, from renal failure and pneumonitis, and one death related to Stivarga, of jejunal perforation.

Despite all this, Exelixis plans to file zanzalintinib with the FDA this year for colorectal cancer.

Cabometyx, which is set to come off patent in 2030, isn’t approved in colorectal cancer, but it is marketed for renal cancer, where zanzalintinib’s next big readout will come: data from the first-line Stellar-304 trial testing an Opdivo combo are due in the first half of 2026.

However, another late-stage trial, Stellar-305 in first-line head and neck cancer, was recently discontinued.