AB154

Domvanalimab is Arcus Biosciences' Fc-silent anti-TIGIT monoclonal antibody, partnered globally with Gilead in a 2020 deal that put roughly $2B on the table [1][6]. It's an immune checkpoint blocker - same broad family as Keytruda - aimed at a target that has gotten brutalized in the clinic. Roche's tiragolumab, the lead anti-TIGIT antibody, missed its primary PFS endpoint in two Phase 3 NSCLC trials in 2022, with subsequent OS analyses also showing no benefit, raising real doubt about whether TIGIT inhibition does anything useful in humans [2][3]. Arcus and Gilead are running two Phase 3 trials that should settle the question: STAR-121 in front-line gastric/GEJ cancer (n=1040, enrollment complete, OS primary) and PACIFIC-8 in Stage III unresectable NSCLC post-chemoradiation (n=860, recruiting, PFS primary, AstraZeneca-sponsored on a durvalumab backbone) [4][5]. Readouts span 2026-2027. Domvanalimab is deliberately Fc-silent, engineered to not engage immune effector cells via its Fc region, which Arcus argues makes it mechanistically distinct from tiragolumab and vibostolimab (both Fc-enabled IgG1s). Whether that distinction translates into clinical wins is the entire investment thesis.

Novel compound, never approved anywhere. Database lists Phase 1 based on a stale link to NCT04999761 (a Taiho platform study), but the drug itself is in Phase 3 across two indications. STAR-121 (Arcus-sponsored, upper GI) and PACIFIC-8 (AstraZeneca-sponsored, Stage III NSCLC) are the late-stage programs that matter [4][5]. No FDA breakthrough therapy or fast-track designation has been publicly disclosed. The drug originated at Arcus Biosciences and was licensed to Gilead in May 2020 under a 10-year collaboration that paid Arcus $175M upfront, $200M equity, with milestones potentially exceeding $1.5B [1][6]. Gilead exercised its formal option on domvanalimab in 2021 for an additional $725M. STAR-121 is active-not-recruiting (enrollment complete at 1,040 patients) with OS primary, readout expected late 2026 into 2027. PACIFIC-8 is still recruiting toward 860 patients with PFS primary. The Phase 2 ARC-7 trial in PD-L1-high (TPS ≥50%) 1L NSCLC showed a numerical PFS improvement for domvanalimab + zimberelimab (Arcus's proprietary anti-PD-1 antibody, the same target class as Keytruda) + etrumadenant (an adenosine A2A/A2B receptor antagonist that targets a separate immunosuppressive pathway) over zimberelimab alone, but with overlapping confidence intervals and a small sample [7]. That trial was the basis for moving into Phase 3.

TIGIT (T-cell immunoreceptor with Ig and ITIM domains) is a brake pedal on T cells and natural killer cells. When engaged, it tells the immune cell to stand down. Tumors learn to express PVR (CD155), the ligand for TIGIT, on their surface. When tumor PVR binds T-cell TIGIT, the T cell stops killing [8]. TIGIT also competes with CD226 (DNAM-1), an activating receptor on T and NK cells that binds the same PVR ligand - so TIGIT inhibition is meant to both release the brake and free up PVR for CD226-mediated activation. Blocking TIGIT with an antibody should release that brake and let T cells attack the tumor. This is the same logic as PD-1 blockade (Keytruda, Opdivo), and the field expected TIGIT to be the natural next checkpoint to hit, usually in combination with anti-PD-1 to get additive immune activation. The mechanism's validation is shakier than it looked five years ago. Preclinical mouse data was strong. Early human signal in the Phase 2 CITYSCAPE trial (tiragolumab + atezolizumab) suggested a benefit in PD-L1-high NSCLC. But Phase 3 confirmation collapsed: SKYSCRAPER-01 missed its primary PFS endpoint in 1L NSCLC (2022), with OS data later confirming no benefit, and SKYSCRAPER-02 in small cell lung cancer also failed on PFS and OS [2][3]. Whether the target is wrong, the antibody design is wrong, or the patient selection criteria are wrong remains unresolved. Domvanalimab's Fc-silent design is Arcus's bet that Fc engagement was the problem - that Fc-active antibodies inadvertently deplete the very effector T cells they should be unleashing.

The node references NCT04999761, a Taiho-sponsored Phase 1 platform study in Japan where AB154 is one of many investigational agents. That's a small data point; the real action is elsewhere. STAR-121 (NCT05568095) is the highest-stakes trial: a Phase 3 randomized comparison of domvanalimab + zimberelimab + chemotherapy vs pembrolizumab + chemotherapy in 1L upper GI cancer (gastric, GEJ, esophageal adenocarcinoma). Chemo backbone is investigator-choice FOLFOX or CAPOX, reflecting current SOC. Enrollment of 1,040 patients is complete, with stratification by PD-L1 CPS (≥5 vs <5) and MSI status; the trial enrolls all-comers rather than PD-L1-selected patients. Primary endpoint is OS. The comparator is current standard of care (Keytruda + chemo), which is the right design but also the hard bar to clear [4]. Sponsor: Arcus Biosciences. PACIFIC-8 (NCT05211895) tests durvalumab + domvanalimab vs durvalumab alone after concurrent chemoradiation in Stage III unresectable NSCLC. AstraZeneca runs it, enrollment target 860, primary endpoint PFS. This is a clean add-on design: does adding TIGIT blockade improve on the established PACIFIC standard of care [5]? EDGE-Gastric and ARC-10 in 1L NSCLC are earlier-phase studies still generating supporting data. Recruitment health is not the bottleneck across the program. Efficacy is.

Our model estimates a 44% chance this drug is eventually approved. It starts from the historical base rate for Phase 3 drugs in this area (about 48%), then adjusts using ten facts about the trial and sponsor. What moves the number most: it is helped by an unusually multi-arm design (17 arms) and a non-randomized design; it is held back by the sponsor's thin or weak approval record and its few secondary endpoints. The other facts land near average for this stage, so they leave the estimate roughly where the base rate put it.

Efficacy risk is the dominant concern. TIGIT inhibition has produced almost no positive Phase 3 readouts in oncology. Roche's tiragolumab missed primary PFS in SKYSCRAPER-01 (1L NSCLC, 2022) and SKYSCRAPER-02 (extensive-stage SCLC, 2022), with subsequent OS analyses also negative [2][3]. Merck's vibostolimab Phase 2 signals have been weak. Ociperlimab (BeiGene, partnered with Novartis until Novartis returned rights in 2023) had its Phase 3 AdvanTIG program discontinued - another class data point in the negative column. Belrestotug (GSK, licensed from iTeos) is in Phase 3 NSCLC and is the most direct comparator on the program calendar, though it uses an Fc-enabled rather than Fc-silent design, so a belrestotug win wouldn't fully validate Arcus's Fc-silent thesis and a belrestotug miss wouldn't fully invalidate it. The Arcus argument is that domvanalimab's Fc-silent design avoids the mechanism failure that hit tiragolumab, but this is a hypothesis, not validated. If Fc engagement matters in the wrong direction (depleting useful effector T cells), Fc-silencing helps. If it doesn't matter, domvanalimab fails for the same reason. Safety risk is moderate. Anti-TIGIT antibodies in combination with PD-1 have shown manageable immune-related adverse events: colitis, hepatitis, endocrinopathies, broadly similar to PD-1 monotherapy. No mechanism-specific black-box safety signal has emerged. Execution risk is low. Enrollment is complete or progressing across the Phase 3 program, and Arcus, Gilead, and AstraZeneca are all credible sponsors. Commercial risk is real even with positive efficacy. The 1L NSCLC market is crowded; Keytruda dominates and pricing is anchored. Upper GI is more open if STAR-121 hits, but Keytruda + chemo and Opdivo + chemo are entrenched. Payers will scrutinize anti-TIGIT add-ons unless the OS benefit is unambiguous (HR <0.8, clean p-value). Marginal wins won't generate the pricing use Arcus and Gilead need.

Worth watching closely. This is the single most important pharma readout for the TIGIT class. STAR-121 is the readout that matters. If it hits OS with a clean hazard ratio (<0.85), Arcus and Gilead validate Fc-silent anti-TIGIT and salvage a class that investors had largely written off. Arcus stock would re-rate meaningfully, and the program could generate $1-2B in peak sales in upper GI. Gilead's $725M option exercise looks vindicated. If STAR-121 misses, the read-through is harsh: three Phase 3 misses across multiple sponsors and molecules would be conclusive evidence that TIGIT as a single-target strategy is dead. Arcus's financial cushion shapes how survivable a miss is: per recent 10-Q disclosures Arcus held roughly $1B in cash and investments with annual operating burn in the $300-400M range, plus Gilead funding a meaningful share of domvanalimab dev costs - that puts independent runway into ~2027 even on a STAR-121 miss [1]. Arcus is not Gilead-dependent for immediate survival, but a miss would force a pipeline reorientation toward etrumadenant (adenosine pathway) and casdatifan (HIF-2α), with the multi-billion-dollar TIGIT thesis evaporating. A belrestotug readout from GSK in the same window is a useful sentinel - concordant misses across Fc-enabled and Fc-silent TIGIT antibodies would close the chapter on the target entirely. Check back: STAR-121 readout window is mid-2026 to mid-2027. PACIFIC-8 will follow. Arcus reports quarterly; watch Q3 2026 and Q4 2026 calls for enrollment updates, interim safety reads, and any DSMB communications. Gilead's discussion of the Arcus collaboration on quarterly calls is typically brief but telegraphs confidence level. Relevant comparison node in the BioCosm map is tiragolumab (Roche) - already at Phase 3, multiple failures recorded.