SKYSCRAPER-01 confirms tiragolumab added nothing in PD-L1-high lung cancer

Anti-TIGIT antibodies were, a few years ago, the most crowded bet in cancer immunotherapy. TIGIT is a checkpoint on immune cells, and blocking it alongside a PD-L1 inhibitor looked like a way to make immunotherapy work in more patients. Tiragolumab, Roche's candidate, led the field after an early trial hinted at benefit in lung cancer. SKYSCRAPER-01 was the phase 3 test of that hint, and its final analysis is now published. The drug added nothing that reached statistical significance [s1].

What the trial tested

SKYSCRAPER-01 enrolled patients with untreated, locally advanced unresectable or metastatic non-small-cell lung cancer whose tumours were PD-L1-high on central laboratory testing [s1]. They were randomly assigned 1:1 to tiragolumab (600 mg) plus atezolizumab (1,200 mg) or to placebo plus atezolizumab (1,200 mg), given intravenously in 21-day cycles until disease progression, loss of clinical benefit or unacceptable toxicity [s1].

The two primary endpoints were investigator-assessed progression-free survival and overall survival in the primary analysis set, defined as PD-L1-high tumours by the 22C3 assay [s1]. In all, 521 patients were randomised — 262 to tiragolumab plus atezolizumab and 259 to placebo plus atezolizumab [s1]. The design matters here: the comparator arm still received a working immunotherapy, so the trial measured what tiragolumab adds on top, not immunotherapy against chemotherapy.

Restricting the primary analysis to PD-L1-high tumours was itself a bet. This is the subgroup in which the early-phase signal had been strongest, so the trial gave tiragolumab its most favourable ground. A drug that cannot separate from the comparator in the population most likely to respond has little room to argue that a broader population was the problem.

What happened

At the primary progression-free-survival analysis, with a median follow-up of 9.9 months, median progression-free survival was 7.0 months (95% confidence interval, 5.6 to 9.8) with tiragolumab plus atezolizumab and 5.6 months (95% CI, 4.4 to 7.0) with placebo plus atezolizumab [s1]. The hazard ratio was 0.78 (95% CI, 0.63 to 0.97), with P = .02 — which the paper reports as nonsignificant [s1].

That combination — a nominal P of .02 labelled nonsignificant — is not a contradiction. In a trial with two primary endpoints and interim looks, the statistical significance boundary is split and moved so that the overall chance of a false positive stays controlled. A P of .02 that would be convincing in a simple two-arm test did not cross the stricter boundary this design demanded. The apparent 1.4-month gain in progression-free survival was, formally, not a win.

The overall-survival result removes any ambiguity. At the final analysis, with a median follow-up of 17.9 months, median overall survival was 23.1 months (95% CI, 17.7 to 28.8) with tiragolumab plus atezolizumab and 16.9 months (95% CI, 14.6 to 21.3) with placebo plus atezolizumab, a hazard ratio of 0.87 (95% CI, 0.70 to 1.10), with P = .22 — nonsignificant [s1]. The confidence interval crosses one, and the median difference sits on top of overlapping intervals. The authors conclude that tiragolumab plus atezolizumab did not demonstrate a statistically significant progression-free or overall survival benefit over atezolizumab in this population [s1].

Toxicity moved in the expected direction. Grade 3 or 4 adverse events occurred in 41.2% of patients (110 of 267) receiving tiragolumab plus atezolizumab and 33.8% (89 of 263) receiving placebo plus atezolizumab [s1]. There were four treatment-related deaths in the tiragolumab group and two in the placebo group [s1]. The added drug added harm without adding benefit.

The funding and the pattern

SKYSCRAPER-01 was sponsored by Hoffmann-La Roche, the maker of both tiragolumab and atezolizumab, and was registered as NCT04294810 [s1][s2]. A negative trial published in full by its industry sponsor is the system working as intended: the encouraging early signal was put to a hard test, the test came back null, and the null is on the record rather than buried. That record is what allows a whole drug class's promise to be recalibrated honestly.

Limits

The trial was confined to PD-L1-high, previously untreated non-small-cell lung cancer, so it does not speak to other tumour types, lower PD-L1 levels or later treatment lines [s1]. It tested one anti-TIGIT antibody at one dose against one immunotherapy backbone; a different agent or combination could behave differently, though the burden of proof now sits higher [s1].

What to watch

Whether remaining anti-TIGIT programmes report, and whether any of them clears the bar SKYSCRAPER-01 missed. For tiragolumab in this setting, the final analysis is unambiguous: the early signal did not hold, and the addition cost patients in toxicity without buying them time [s1].

This article describes trial results. It is not medical advice, and nothing here should be used to guide treatment decisions.

Sources

• SKYSCRAPER-01: Tiragolumab in Combination With Atezolizumab in Previously Untreated PD-L1-High NSCLC, Journal of Clinical Oncology, 27 July 2026
• A Study of Tiragolumab in Combination With Atezolizumab (NCT04294810), ClinicalTrials.gov