Exelixis' Zanzalintinib Suffers Mixed Final Phase 3 Results in Colorectal Cancer, Clouding FDA Label Strategy
On June 22, 2026, Exelixis announced results from the final analysis of its Phase 3 STELLAR-303 trial evaluating zanzalintinib—the company's next-generation multikinase inhibitor and successor to Cabometyx—in combination with Roche's immune checkpoint inhibitor Tecentriq (atezolizumab) in previously treated metastatic colorectal cancer (mCRC).
The trial missed one of its dual primary endpoints, failing to achieve statistical significance in overall survival (OS) in a key subpopulation of patients without active liver metastases (non-liver metastases, or NLM). Despite the clinical setback, analysts believe FDA approval remains highly probable because the trial successfully met its other dual primary endpoint—overall survival in the overall study population—earlier in the year.
The mixed final results, however, introduce regulatory uncertainty regarding the drug's eventual label, particularly whether the FDA will restrict marketing claims or exclude the NLM subpopulation.
Fierce Biotech detailed the final trial outcomes and analyst reactions:
"A phase 3 trial for the next-generation multikinase inhibitor in patients with previously treated metastatic colorectal cancer (mCRC) failed to achieve statistical significance on a dual primary endpoint measuring overall survival in a subpopulation, Exelixis announced Monday... Despite the flop, analysts at Leerink Partners and William Blair argued that the FDA will likely still approve zanzalintinb in third-line mCRC because the Stellar-303 trial previously met the other primary endpoint—OS in all study participants regardless of their liver metastases status." — Angus Liu, Fierce Biotech
Final STELLAR-303 Efficacy Outcomes
The Phase 3 STELLAR-303 trial evaluated the zanzalintinib-Tecentriq combination against Bayer's Stivarga (regorafenib) in previously treated patients with non-microsatellite instability (non-MSI)-high mCRC.
- Subpopulation Miss (NLM OS): At the final analysis of OS in patients without active liver metastases (the NLM subgroup), the combination demonstrated only a numerical 17% overall survival improvement, which did not cross the threshold for statistical significance. Patients in the combination arm lived a median of 15.9 months compared to 12.7 months in the control arm. This represents a clinical deterioration from the earlier topline readout, where the NLM subgroup had shown a 23% OS benefit.
- Overall Population Success: Earlier in the clinical program, the trial met its other dual primary endpoint, showing a statistically significant 20% OS improvement (a 1.5-month median OS extension) in the overall randomized population. Under the trial's statistical design, meeting either of the two dual primary endpoints constitutes a successful trial.
Serial Changes in Trial Design
The STELLAR-303 study has drawn regulatory attention due to multiple protocol amendments:
- 2022 Launch: Originally initiated with OS in the overall randomized population as the sole primary endpoint.
- 2024 Amendment: Exelixis modified the design to make OS in the NLM subgroup the sole primary endpoint. This pivot was prompted by the failure of a rival combo (Merck and Eisai's Keytruda/Lenvima) in all-comer metastatic CRC, where post-hoc analysis suggested that immunotherapies are more effective in NLM patients due to a less immunosuppressive microenvironment.
- 2025 Amendment: Following early positive signals, Exelixis re-upgraded OS in the overall population to a dual primary endpoint alongside NLM OS.
While these serial modifications do not block a potential approval, they may invite close FDA scrutiny during the ongoing review process.
Commercial Stakes and Label Implications
The FDA accepted Exelixis' New Drug Application (NDA) for zanzalintinib plus Tecentriq in late-line mCRC in February 2026, setting a PDUFA target decision date of December 3, 2026.
Leerink Partners models zanzalintinib's unadjusted global peak sales in colorectal cancer at $1.3 billion, with the NLM subgroup accounting for approximately half of that commercial opportunity. Failing to secure a statistically significant claim in this specific subpopulation could weaken the drug's marketing message. However, because the overall population endpoint was met and the NLM subgroup trended positively, analysts remain optimistic that the FDA will grant a broad label without excluding NLM patients.