Most oncology trials we cover test a new drug against placebo or an older comparator. SKYSCRAPER-06 (NCT04619797) did something harder and more honest: it put Roche’s investigational anti-TIGIT antibody tiragolumab, added to atezolizumab and chemotherapy, directly against the current first-line standard — pembrolizumab plus the same chemotherapy — in previously untreated advanced non-squamous non-small-cell lung cancer. The result, per the ClinicalTrials.gov results record, is unambiguous and it is not the one the sponsor wanted.

This was a Phase II/III trial that randomized 542 patients (tiragolumab arm, 269; pembrolizumab arm, 273), with two co-primary endpoints. On investigator-assessed progression-free survival, the pembrolizumab arm reached 9.89 months versus 8.31 months for the tiragolumab arm (hazard ratio 1.27; 95% CI 1.02–1.57). On overall survival, the gap widened: 23.1 months versus 18.89 months (HR 1.33; 95% CI 1.02–1.73). Both confidence intervals sit entirely above 1.0, meaning the tiragolumab regimen did not merely fail to win — it performed measurably worse on both primary endpoints. Objective response rate, a secondary endpoint, followed the same direction (50.2% vs 56.6%).

A hazard ratio of 1.33 for death is not a near-miss; in a head-to-head design, it is the standard of care defending its position.

The safety data sharpen the concern. All-cause deaths were recorded in 146 of 269 patients (54.3%) on tiragolumab versus 117 of 273 (42.9%) on pembrolizumab; serious adverse events occurred in 55.1% versus 52.2%.

What we cannot yet say: the registry record does not establish the cause of the excess deaths, the contribution of any single component, or how these numbers will read once peer-reviewed with full subgroup and exposure detail. Association is not causation, and a registry posting is not a published manuscript. But the core message is already clear. For TIGIT as a target in this setting, this is a hard negative — and a useful reminder that adding to an active regimen can subtract.

AI-drafted under the direction of Armando Cuesta, MD. Sources: ClinicalTrials.gov NCT04619797 (registry and API v2 results record).