For patients with unresectable esophageal squamous cell carcinoma (ESCC), definitive concurrent chemoradiotherapy (dCRT) remains a standard treatment approach. However, disease progression or recurrence remains common, and unlike in some other tumor types, no immunotherapy-based consolidation strategy has yet been established after dCRT.
The phase III SKYSCRAPER-07 trial explored whether immune checkpoint inhibition after dCRT could improve outcomes in this setting, evaluating atezolizumab either alone or combined with the anti-TIGIT antibody tiragolumab.
The study, titled “Atezolizumab with or without tiragolumab in unresectable esophageal squamous cell carcinoma following definitive concurrent chemoradiotherapy (SKYSCRAPER-07): a randomised, phase III study,” was published on August 21, 2026, in Annals of Oncology.
Authors: R-H. Xu, B.C. Cho, M. Chen, K. Muro, L. Wyrwicz, B. Li, S. Gao, C. Hsu, A. Turpin, F. Lordick, E. Cha, Q. Wu, S. Kim, J. Cai, L. Zhang, J. Fan, L. Wang, K. Sully, L. Wang, K. Goodman, M.A. Shah, and I. Chau.
The SKYSCRAPER-07 Trial
SKYSCRAPER-07 (NCT04543617) was a global, randomized, double-blind, placebo-controlled phase III trial conducted across 166 centers in 28 countries or regions.
The study enrolled adults with unresectable stage II–IVA ESCC, as well as selected patients with stage IVB disease limited to supraclavicular lymph node metastases, who were considered suitable for definitive chemoradiotherapy. All patients had an ECOG performance status of 0–1 and had completed platinum-based dCRT without radiographic disease progression.
A total of 760 patients were randomized 1:1:1 to receive atezolizumab 1200 mg plus tiragolumab 600 mg, atezolizumab 1200 mg plus placebo, or placebo plus placebo every 3 weeks for up to 17 cycles. The median age was 66 years, 74.2% of patients were male, and 62.8% were Asian. Approximately half of the population had stage III disease before dCRT. PD-L1 expression, assessed using the tumor area positivity (TAP) score, was ≥1% in 86% of patients and ≥10% in 36%.
The trial used a hierarchical statistical testing strategy. Investigator-assessed progression-free survival (PFS) for atezolizumab plus tiragolumab versus placebo was tested first, followed by overall survival (OS) for the same comparison and then OS for atezolizumab alone versus placebo.
Earlier Presentation at ESMO 2025
SKYSCRAPER-07 was previously presented by Ian Chau, MD, at the ESMO Congress 2025. The trial showed that atezolizumab plus tiragolumab did not significantly improve investigator-assessed PFS versus placebo, with median PFS of 20.8 versus 16.6 months (HR 0.82; 95% CI 0.65–1.03; p=0.0947). Median OS was 38.6 versus 36.4 months (HR 0.91; 95% CI 0.70–1.18).
In contrast, atezolizumab alone showed a clinically meaningful improvement in outcomes compared with placebo. Median PFS was 29.1 versus 16.6 months (HR 0.74; 95% CI 0.58–0.93), while median OS was not reached versus 36.4 months (HR 0.69; 95% CI 0.52–0.91). Because the tiragolumab-containing arm failed the first endpoint in the hierarchical testing sequence, the atezolizumab-versus-placebo OS comparison could not be formally tested for statistical significance.
Tiragolumab Addition Did Not Improve Outcomes
At a median survival follow-up of 25.0 months, the trial did not meet its first primary endpoint. Median investigator-assessed PFS was 20.8 months with atezolizumab plus tiragolumab compared with 16.6 months with placebo. The difference was not statistically significant, with an HR of 0.82 (95% CI 0.65–1.03; p=0.0947).
Median OS was 38.6 months with atezolizumab plus tiragolumab and 36.4 months with placebo, corresponding to an HR of 0.91 (95% CI 0.70–1.18; descriptive p=0.4772).
Because the first endpoint in the hierarchical testing sequence was not met, subsequent endpoints could not undergo formal confirmatory statistical testing. The findings therefore showed no added clinical benefit from combining tiragolumab with atezolizumab in this post-dCRT setting.
Atezolizumab Alone Showed a Different Signal
A different pattern was observed in the atezolizumab-alone arm. Median investigator-assessed PFS reached 29.1 months with atezolizumab compared with 16.6 months with placebo, corresponding to an HR of 0.74 (95% CI 0.58–0.93). Independent review showed a similar result, with an HR of 0.64 (95% CI 0.50–0.82).
Median OS had not been reached in the atezolizumab group and was 36.4 months in the placebo group. The HR for death was 0.69 (95% CI 0.52–0.91; descriptive p=0.0085).
Importantly, although the magnitude and consistency of these results suggested a clinically meaningful treatment effect, the OS comparison between atezolizumab and placebo could not be formally tested because the preceding endpoint in the trial’s statistical hierarchy had failed. The findings should therefore not be interpreted in the same way as a formally statistically significant primary endpoint.
OS benefit with atezolizumab was generally observed across clinically relevant subgroups, including age, ECOG performance status, and geographic region. The investigators also reported a stronger trend toward benefit among patients with PD-L1 TAP scores ≥1%, with the greatest apparent benefit among those with TAP scores ≥10%. Confidence intervals for these subgroup analyses were not adjusted for multiplicity.
What Happened With Tiragolumab?
The comparison between the two immunotherapy-containing groups provided no evidence that adding TIGIT inhibition improved outcomes. Median investigator-assessed PFS was 20.8 months with atezolizumab plus tiragolumab versus 29.1 months with atezolizumab alone, with an HR of 1.13 (95% CI 0.89–1.44). Median OS was 38.6 months with the combination and had not been reached with atezolizumab alone, with an HR of 1.36 (95% CI 1.03–1.80).
These comparisons were not formally tested within the hierarchical statistical framework, but they provide little support for an additional benefit from tiragolumab. The authors noted that the combination arm also had a shorter treatment duration and higher rates of several toxicity measures, raising the possibility that additional toxicity may have affected treatment delivery.
Safety
Among 755 patients included in the safety population, treatment-related adverse events occurred in 74.8% of patients receiving atezolizumab plus tiragolumab, 65.2% receiving atezolizumab alone, and 55.4% receiving placebo.
Grade 3–4 treatment-related adverse events occurred in 16.1%, 9.6%, and 9.6% of patients, respectively. Treatment-related deaths were reported in 1.2% of patients receiving atezolizumab plus tiragolumab, 0.8% receiving atezolizumab, and 1.6% receiving placebo.
Adverse events led to treatment discontinuation in 9.1% of patients in the combination group, 6.8% in the atezolizumab group, and 4.0% in the placebo group. Immune-mediated hypothyroidism was the most frequently reported adverse event of special interest, while immune-mediated pneumonitis accounted for the grade 5 adverse events of special interest reported in the immunotherapy-containing groups.
Patient-reported outcomes showed no clinically meaningful changes in functioning, overall health, quality of life, or disease-related symptoms with either immunotherapy approach compared with placebo.
Takeaway
SKYSCRAPER-07 did not meet its primary endpoint of investigator-assessed PFS for atezolizumab plus tiragolumab versus placebo, and no OS difference was observed with the combination. In contrast, clinically meaningful improvements in both PFS and OS were observed with atezolizumab monotherapy compared with placebo.
Because the first endpoint in the hierarchical testing sequence was not met, formal statistical testing of subsequent efficacy endpoints, including OS with atezolizumab versus placebo, was precluded. Nevertheless, the investigators reported consistent findings between investigator- and independently assessed PFS and concluded that the data indicate a treatment effect for atezolizumab monotherapy.
The study also had limitations. Only eight Black or African American patients were enrolled, and approximately two-thirds of participants were Asian. In addition, PD-L1 status was assessed using tumor samples collected before dCRT. Because chemoradiotherapy can alter PD-L1 expression, pre-dCRT PD-L1 status may not reflect PD-L1 expression at the start of atezolizumab consolidation therapy.
Overall, the addition of tiragolumab to atezolizumab did not demonstrate additional benefit. The study demonstrated clinically meaningful improvements in PFS and OS with atezolizumab monotherapy compared with placebo following dCRT in patients with unresectable ESCC, although formal statistical testing was precluded by the hierarchical testing plan.
The full article is available in Annals of Oncology.