zimberelimab

Zimberelimab is Arcus Biosciences and Gilead's PD-1 antibody, licensed in 2020 from China's Gloria Biosciences as GLS-010 and already approved in China under the brand Yu Qi Shu [1]. In the US, the drug is not approved as monotherapy and is unlikely to be - its job is to serve as Arcus's in-house PD-1 backbone for combinations with domvanalimab (anti-TIGIT), quemliclustat (CD73 inhibitor), and other pipeline assets [2][3]. Multiple Phase 3 trials are running in upper GI cancers and non-small cell lung cancer (NSCLC), with the commercial question being whether Arcus's combinations can carve share from Keytruda - not whether the PD-1 antibody itself is differentiated. Investor framing note: Gilead holds US commercialization rights and a 50/50 profit share on co-developed Arcus assets, so the upside on a STAR-221 win flows substantially to Gilead, not purely to RCUS shareholders [13].

Zimberelimab is approved in China for relapsed/refractory classical Hodgkin lymphoma and was greenlit there in 2021 [1]. No FDA approval. The lead US program is the Phase 3 STAR-221 trial in first-line upper GI cancers, combining zimberelimab plus domvanalimab on top of FOLFOX chemotherapy versus standard nivolumab plus chemotherapy (NCT05568095) [4]; this is the Phase 3 successor to the EDGE-Gastric Phase 2. In NSCLC, the Phase 3 STAR-121 study is also active. Supporting Phase 2 data dropped in 2026: the ARC-10 Part 1 study in PD-L1-high NSCLC was published in Lung Cancer (PMID 41830668, unverified - flagged for human confirmation) [2], and a randomized Phase 1 of quemliclustat plus chemo with or without zimberelimab in metastatic pancreatic adenocarcinoma appeared in Nature Medicine (PMID 41912809, unverified - flagged for human confirmation) [3]. No FDA breakthrough or fast track designations for zimberelimab specifically. STAR-221 enrollment is reported complete per Arcus's 2025 10-K, supporting interim readout guidance in 2026 [5]; STAR-121 likely a 2026-2027 event depending on event rates. The drug already cleared regulatory and manufacturing risk through the China approval, which de-risks the CMC side but tells you little about US efficacy bars.

PD-1 is a brake on T cells. When a T cell finds something it should attack - like a tumor - it gets activated, but PD-1 acts as a safety switch that tells the cell to stand down. Many tumors hijack this by expressing PD-L1, the ligand that engages PD-1, which lets the cancer hide in plain sight. Zimberelimab binds PD-1 and blocks that off-signal, so T cells can keep attacking [6]. This is the same trick pembrolizumab (Keytruda) and nivolumab (Opdivo) play, and PD-1 blockade is among the most validated mechanisms in oncology - over a decade of approvals across 20+ tumor types, combined ~$50B in annual sales. So the target itself is bulletproof. The harder question is whether zimberelimab is meaningfully different from pembrolizumab. Available comparative data suggests it isn't. The strategic logic for Arcus isn't pharmacology - it's economics: owning the PD-1 means the company doesn't pay Merck for every combination trial it wants to run, and any combo win flows back to Arcus and Gilead rather than getting split with a pembrolizumab supply deal. The PD-1 is the chassis; the combination partner is where the bet sits. The co-targeted molecule is domvanalimab, an anti-TIGIT antibody. TIGIT is a second inhibitory receptor on T cells and NK cells that suppresses anti-tumor immunity in parallel with PD-1, so the theoretical synergy is releasing two brakes simultaneously. Notably, domvanalimab is engineered as Fc-silent (no FcγR engagement, no Treg depletion), which differentiates it mechanistically from Roche's tiragolumab (Fc-active) - relevant because Fc engagement is one hypothesis for tiragolumab's mixed Phase 3 record.

The key Phase 3 is NCT05568095 - STAR-221 - testing zimberelimab plus domvanalimab plus FOLFOX chemotherapy versus nivolumab plus FOLFOX in first-line locally advanced or metastatic gastric, gastroesophageal junction, or esophageal adenocarcinoma [4]. FOLFOX is the standard backbone first-line chemo regimen for upper GI cancers - oxaliplatin plus leucovorin plus 5-fluorouracil. The active comparator design is unusual and aggressive: Arcus chose to beat nivolumab head-to-head rather than placebo, which means the bar is real efficacy improvement, not just regulatory tolerance. Primary endpoint is overall survival. Phase 2 EDGE-Gastric data showed a 59% objective response rate in the triplet versus historical ~45-50% for nivolumab plus chemo, with a hazard ratio signal that Arcus has called encouraging but is not yet mature [5]. Enrollment target is approximately 1,000 patients and enrollment is reported complete per Arcus's most recent annual filing [5]. Other active studies include NCT04736173 (ARC-10, zimberelimab + domvanalimab in NSCLC, Phase 2, n=169) [7], NCT05633667 in lung cancer (n=192) [8], and NCT05329766 in upper GI Phase 2 (n=332) [9]. The risk in the design: choosing nivolumab as comparator instead of pembrolizumab in gastric is defensible (nivolumab is the approved frontline immunotherapy there) but limits the commercial story if the win is modest.

The model gives this drug a 33% chance of eventually being approved. It starts from the historical approval rate for Phase 3 drugs in this area - about 48% - then adjusts using ten facts about the trial and sponsor. The estimate is pulled up by a larger-than-typical number of participants, the trial's blinding approach, and the sponsor's strong track record of getting drugs approved; it is pulled down by weak results from earlier phases. The remaining facts are close to average for this stage, so they don't move the number much either way.

Three concrete failure modes. First, TIGIT class risk: anti-TIGIT combinations have produced mixed Phase 3 readouts. Roche's SKYSCRAPER-01 in first-line NSCLC missed its primary endpoints [10], while SKYSCRAPER-08 in esophageal squamous cell carcinoma (ESCC) reported a positive PFS and OS signal versus chemotherapy plus atezolizumab [14] - partially supportive for the class but in a different histology than STAR-221's gastric/GEJ/esophageal adenocarcinoma population. ESCC and adenocarcinoma have distinct biology and immune microenvironments, so extrapolation is uncertain. Domvanalimab's Fc-silent design is a differentiator versus tiragolumab and may not inherit the same liabilities, but it has not been validated at Phase 3 yet. If TIGIT biology doesn't add meaningfully to PD-1 in upper GI adenocarcinoma, zimberelimab plus domvanalimab is just a more expensive Opdivo plus FOLFOX. Second, commercial risk even on a win: gastric cancer is a smaller market than NSCLC, payers will demand pharmacoeconomic justification, and nivolumab plus chemo is the entrenched standard with cost dynamics already baked in. A statistically significant but clinically modest OS benefit (say 1.5-2 months) gets approved but doesn't displace nivolumab in real-world use. Third, the PD-1 itself is undifferentiated. If STAR-221 fails, zimberelimab has no independent commercial path in the US - there is no scenario where the FDA approves a fifth PD-1 monotherapy in a saturated class. The asset's entire US value is tethered to combination wins. Safety is the least worrying risk: PD-1 toxicity is well-characterized, and domvanalimab has shown a clean profile in early trials. Biosimilar/patent horizon is a longer-term concern - zimberelimab composition of matter dates to the original Gloria Biosciences IP from the late 2010s, putting standard 20-year horizon expiry in the late 2030s, with potential biosimilar entry shaping terminal value.

Worth watching, but watch the right thing. The signal is not zimberelimab - it's whether anti-TIGIT plus anti-PD-1 can beat anti-PD-1 alone in a real Phase 3 with overall survival as the primary endpoint, in an adenocarcinoma population where the prior partial-positive (SKYSCRAPER-08) doesn't directly extrapolate. STAR-221 interim analysis in upper GI is the inflection point: a clean OS hit with HR < 0.80 would be the first true TIGIT win at Phase 3 in adenocarcinoma and would re-rate Arcus (RCUS, ~$1.5B market cap) substantially. A miss closes the TIGIT chapter and leaves zimberelimab as a China-only asset plus a CD73/HIF-2α option value play. Critical economics caveat: under the Arcus-Gilead 2020 collaboration, Gilead paid $175M upfront, made a $200M equity investment, and is entitled to up to ~$1.225B in opt-in and milestone payments across the collaboration; for US co-developed assets including zimberelimab and domvanalimab, Gilead and Arcus split profits/losses 50/50 with Gilead leading commercialization [13]. So while Gilead's ~$27.1B in FY2024 revenue [15] insulates the parent, an ARCUS investor's upside on a STAR-221 win is real but diluted by the profit split and milestone structure. The Arcus stock is binary on this readout; the Gilead stock is largely insulated either way. Check back at the next Arcus 8-K filing window or ASCO/ESMO 2026 - both have plausible windows for an interim update [11][12]. Secondary watch: the quemliclustat plus zimberelimab pancreatic Phase 1 in Nature Medicine [3] is an unrelated mechanism worth tracking on its own merits, but it doesn't change the gastric thesis. Skip the press releases about zimberelimab in isolation; the drug is a means, not the bet.