Tiragolumab

Tiragolumab is Roche's anti-TIGIT antibody, paired with the company's PD-L1 inhibitor atezolizumab (Tecentriq) plus chemotherapy across a multi-tumor Phase 3 program called SKYSCRAPER. The pitch is straightforward: stack a second checkpoint blocker on top of PD-L1 inhibition and get deeper, more durable responses. Execution has been bumpy. SKYSCRAPER-01 in PD-L1-high non-small cell lung cancer missed its progression endpoint in 2022 and sent the TIGIT field into a confidence crisis; the final OS analysis presented at AACR 2025 confirmed no OS benefit either (HR 0.87, 95% CI 0.71-1.08, p=0.22) [5]. SKYSCRAPER-02 in extensive-stage small cell lung cancer also failed [6]. SKYSCRAPER-08 in first-line esophageal squamous cell carcinoma reported a positive Phase 3 readout in 2026 (OS HR 0.70, 95% CI 0.55-0.88, p=0.0024; median OS 15.7 vs 11.1 months), the first clean Phase 3 OS win for any TIGIT antibody [1]. SKYSCRAPER-06 in nonsquamous NSCLC, a head-to-head against pembrolizumab plus chemotherapy, was stopped at interim in 2024 for lack of benefit and published in 2026 [2]. The drug matters because TIGIT is the most-tested second-wave checkpoint target after PD-1/PD-L1, and Roche needs an I-O asset with fresh patent runway to defend its Tecentriq franchise as PD-L1 biosimilars approach. Roche's Pharmaceuticals Division generated CHF 46.2B (~$51B USD) in 2024 [13], and tiragolumab is one of the few internal candidates positioned to backfill the post-Tecentriq oncology gap.

Novel compound. Tiragolumab has never been approved anywhere, for any indication. Roche owns the program outright with no licensing partner [13]. Public disclosures do not include a tiragolumab-specific FDA breakthrough or fast-track designation, though that could change if Roche files in first-line esophageal squamous cell carcinoma on the back of SKYSCRAPER-08. The active development envelope spans Phase 3 in esophageal SCC (SKYSCRAPER-08, reported positive [1, 8]) and nonsquamous NSCLC (SKYSCRAPER-06, stopped early [2]), plus earlier-stage work in hepatocellular carcinoma (Morpheus-Liver [9]), colorectal liver metastases (PURPLE [4]), advanced renal cell carcinoma (NCT05805501 [12]), and pediatric SMARCB1/SMARCA4-deficient tumors run by NCI (NCT05286801 [10]). A rollover study (NCT05862285 [11]) keeps prior trial participants on tiragolumab, which signals Roche sees clinically meaningful responders worth preserving access for. The likely next regulatory move is a filing in first-line esophageal SCC, where standard of care globally is pembrolizumab plus chemotherapy from KEYNOTE-590 [7]. Roche has not publicly committed to a submission date on its 2024 or early-2025 earnings calls. If approved, the launch indication would be narrow; expansion into a Keytruda-style label depends on Phase 3 reads in other tumors that have not started or have already missed.

Think of a killer T cell attacking a tumor as a car that needs both gas (an antigen signal) and no brakes (no inhibitory signal). PD-1 and PD-L1 are one brake. TIGIT (T-cell immunoreceptor with Ig and ITIM domains) is another. TIGIT is a protein sitting on the surface of T cells and natural killer cells. When it binds CD155 on tumor cells or on antigen-presenting cells, it sends a stand-down signal that suppresses immune attack. Tumors exploit this by overexpressing CD155. Tiragolumab is an antibody that binds TIGIT and physically blocks the interaction, so the brake never engages. The full mechanistic story is more interesting. The same CD155 ligand is also bound by an activating receptor, CD226 (DNAM-1). TIGIT outcompetes CD226 for CD155, so high TIGIT not only brakes T cells directly but also shuts out the accelerator. Blocking TIGIT releases both effects simultaneously - the inhibitory signal is removed and CD226-mediated co-stimulation is restored. This is the mechanistic rationale for why TIGIT blockade was expected to amplify anti-tumor immunity on top of PD-1/PD-L1 inhibition and why preclinical synergy with PD-L1 blockade was so striking in mice. The biology is plausible and the receptor is well-characterized. Where the case weakens is translation. Genetic evidence linking TIGIT to cancer outcomes is moderate. Open Targets gives TIGIT-NSCLC an evidence score of 0.41, TIGIT-SCLC 0.35, TIGIT-esophageal SCC 0.30. No biomarker reliably predicts which patients respond to TIGIT blockade - CD155 expression, CD226 co-expression, and tumor-infiltrating lymphocyte density have all been explored as candidate predictors but none has cleared validation. In humans the PD-L1/TIGIT synergy has been inconsistent. TIGIT as a clinically additive checkpoint on top of PD-1/PD-L1 is not yet established by any approved drug. Tiragolumab is the most advanced candidate in the class, and SKYSCRAPER-08 is the first Phase 3 readout that supports the additive mechanism in any setting [1].

NCT04543617 is SKYSCRAPER-08: a randomized, double-blind, placebo-controlled Phase 3 trial in first-line locally advanced unresectable or metastatic esophageal squamous cell carcinoma [1, 8]. Treatment arm: tiragolumab plus atezolizumab plus chemotherapy (cisplatin plus paclitaxel). Control arm: placebo plus atezolizumab plus chemotherapy. Critically, the comparator is atezolizumab plus chemo, not pembrolizumab plus chemo, which is the global standard of care from KEYNOTE-590 [7]. SKYSCRAPER-08 measures the contribution of tiragolumab on top of atezolizumab plus chemo, not whether the tiragolumab triplet beats a Keytruda-based regimen. Co-primary endpoints were overall survival and progression-free survival. Enrollment was 461 patients, predominantly across Asia-Pacific sites. The trial reported in 2026 with both co-primary endpoints met: OS HR 0.70 (95% CI 0.55-0.88, p=0.0024), median 15.7 vs 11.1 months; PFS HR 0.56 (95% CI 0.45-0.70), median 6.2 vs 5.4 months [1]. The OS effect is meaningful (4.6-month median improvement). The PFS effect is statistically robust but numerically modest in absolute terms - only 0.8 months of median PFS gain, with curves separating later in follow-up. The design answers a clean efficacy question, but it leaves a commercial question open: regulators and payers will need to extrapolate from contribution-of-component data to a head-to-head positioning that Roche has not run. The broader SKYSCRAPER program (SKYSCRAPER-06 head-to-head versus pembrolizumab plus chemo in nonsquamous NSCLC [2], MORPHEUS-EC Phase 1b/2 in esophageal cancer [3], Morpheus-Liver in HCC [9]) tests the same combination logic across tumor types, with mixed results so far.

Our model gives this drug a 31% chance of eventual approval. That starts from a historical base rate of about 57% for Phase 3 drugs in this area, then shifts based on ten specific facts about the trial and sponsor. On the positive side, the drug has more secondary endpoints than usual and the sponsor has a strong track record of approvals; on the negative side, the trial uses heavier-than-usual blinding and the earlier-phase results were weak or limited. The remaining facts were close to average and did not move the estimate much.

Efficacy risk is class risk. TIGIT antibodies have failed more Phase 3 trials than they have won. SKYSCRAPER-01 missed PFS in PD-L1-high NSCLC in 2022, and the AACR 2025 final OS analysis showed no OS benefit either (HR 0.87, 95% CI 0.71-1.08, p=0.22) [5]. SKYSCRAPER-02 missed in extensive-stage SCLC [6]. SKYSCRAPER-06, the head-to-head against pembrolizumab plus chemotherapy in first-line nonsquamous NSCLC, was stopped at interim in 2024 for inferior efficacy versus the Keytruda-based comparator [2]. Merck's vibostolimab plus pembrolizumab program in melanoma and lung cancer has produced inconsistent signals. The unanswered scientific question (does TIGIT blockade meaningfully add to PD-L1 blockade) has not been definitively answered. SKYSCRAPER-08 is the first clean Phase 3 win [1], and one positive trial does not validate a target. The SKYSCRAPER-08 PFS delta itself is a soft spot. A 0.8-month median PFS improvement (6.2 vs 5.4 months) is statistically significant by HR but numerically modest. Payers asked to fund a two-checkpoint-inhibitor regimen will press on whether that effect size justifies the incremental cost. Safety risk is moderate. Tiragolumab's safety profile in SKYSCRAPER trials tracks closely with standard checkpoint inhibitor toxicities - immune-related colitis, pneumonitis, endocrinopathies - without a distinct on-target signal that would derail development [1, 2]. Commercial risk is the larger concern. Pembrolizumab plus chemotherapy owns first-line esophageal SCC globally from KEYNOTE-590 [7]. Tiragolumab plus atezolizumab plus chemo enters as a challenger triplet, and Roche has no head-to-head data showing it beats the Keytruda regimen. Competitive risk in TIGIT is concentrated and overlaps directly with Roche's only foothold. Domvanalimab (Gilead/Arcus) has Phase 3 readouts pending in upper GI cancers (STAR-221 in 1L gastric/GEJ/esophageal adenocarcinoma, expected 2026-2027) and NSCLC (STAR-121 in 1L NSCLC and ARC-10 in PD-L1-high NSCLC, expected 2026-2027). STAR-221 in particular overlaps the upper GI space where SKYSCRAPER-08 just established a foothold; a positive STAR-221 with comparable or better efficacy would directly contest Roche's only commercial opening. Ociperlimab (BeiGene) is the third active program in the class with ongoing late-stage work in lung and other solid tumors; an ociperlimab win in esophageal SCC would fragment the market. Absent any head-to-head data among TIGIT agents or against the Keytruda standard, a simultaneous-approval scenario in upper GI would force payer formulary choices on price rather than efficacy. Execution risk is low. Roche runs global Phase 3 trials competently.

Watch, with specific triggers. The SKYSCRAPER-08 OS update at extended follow-up is the data point that decides whether tiragolumab launches as a real product or remains a science exercise. The initial median OS gain (15.7 vs 11.1 months, HR 0.70) is meaningful; if the OS curves stay separated through long-term follow-up, Roche files in first-line esophageal SCC and tiragolumab gets a foothold in a narrow but commercially relevant indication. If the curves converge or the PFS modesty bleeds into OS attrition, the drug stalls; positive Phase 3 in a contribution-of-component design is not enough to displace Keytruda-based standard of care [7]. The second trigger is whether Roche commits to a direct head-to-head trial against pembrolizumab plus chemo, which would be the most informative readout for the class and the riskiest commercial bet - and which Roche has so far avoided in esophageal SCC. The third trigger is competitive: domvanalimab (Gilead/Arcus) STAR-221 in upper GI and STAR-121/ARC-10 in NSCLC read out across 2026-2027. If domvanalimab wins in upper GI, the esophageal SCC foothold becomes contested before launch. If ociperlimab (BeiGene) reads positive in any overlapping indication, the class fragments further. For Roche specifically, this drug matters more than the trial-by-trial data alone suggests. Tecentriq faces biosimilar pressure as PD-L1 patents expire across the late 2020s. A new I-O asset with patent runway into the late 2030s is one of the few credible engines to defend Roche's CHF 46B (~$51B USD) Pharmaceuticals Division revenue base [13]. Check back after the next ASCO/ESMO cycle and at any Roche pharma division update.