The STELLAR-303 final analysis is out, and zanzalintinib’s NLM subgroup didn’t make the cut. Exelixis reported Monday that the combo with atezolizumab produced only a numerical 17% OS improvement in patients without active liver metastases, not statistically significant. Median OS: 15.9 months versus 12.7 months for regorafenib.
The trial carried two co-primary endpoints, and it had already met the other one. At the final ITT analysis, the zanzalintinib-atezolizumab combo showed a statistically significant 20% OS improvement versus regorafenib. Meeting one of two co-primaries satisfies FDA’s multiple-endpoints framework under ICH E9, so Exelixis’ NDA stays live with a PDUFA date of December 3, 2026.
The label question is what’s keeping analysts busy. Leerink Partners models zanzalintinib at $1.3 billion in peak CRC sales globally, with the NLM subgroup accounting for roughly half. A label restricted to ITT-only patients cuts roughly $650 million of that opportunity — or FDA could approve on ITT evidence with a label broad enough to cover NLM patients.
I read the endpoint history carefully. Exelixis redesigned Stellar-303’s primary twice: first to NLM as the sole primary in 2024 after Merck and Eisai’s LEAP-17 missed in the same population, then re-adding ITT as a co-primary after that data came in strong. That double shift is what FDA reviewers interrogate under ICH E9(R1)‘s estimand framework: whether each amendment was driven by external evidence or the sponsor’s own emerging results. The company’s justification, citing LEAP-17 and the NLM immunosuppression hypothesis, will carry real weight in the review.
CEO Michael Morrissey called NLM significance important “from a marketing point of view.” Worth auditing any dual-primary statistical analysis plan before FDA signals how it reads this one in December.
Rebecca Lauren