Sasanlimab Plus BCG In CREST Improves Event-Free Survival
Key Takeaways
- Adding sasanlimab to BCG induction and maintenance was associated with longer investigator-assessed event-free survival than BCG induction and maintenance alone.
- The induction-only combination arm did not significantly differ from the BCG maintenance comparator, and interim overall survival showed no difference between groups.
- Adverse events were more frequent in the sasanlimab-containing arms, while global health-status quality-of-life change versus the comparator did not reach a clinically meaningful difference.
The CREST trial was a global, open-label, three-arm randomized phase 3 study in 1,055 adults across 140 centers in 14 countries, with 352 patients in Arm A, 352 in Arm B, and 351 in Arm C. Eligible patients had BCG-naive high-risk NMIBC with T1, high-grade Ta, and or CIS, and Ta or T1 disease had to be completely resected within 12 weeks before randomization. Randomization was stratified by CIS status and geographic region, and the primary endpoint was investigator-assessed event-free survival for Arm A versus Arm C. Arm A received subcutaneous sasanlimab 300 mg every 4 weeks for up to 25 cycles plus BCG induction and maintenance, Arm B received sasanlimab plus BCG induction only, and Arm C received BCG induction and maintenance, with a 2 December 2024 cutoff and median EFS follow-up of 36.3 months.
At 36 months, the estimated event-free survival rate was 82.1% in Arm A and 74.8% in Arm C. Arm B did not significantly differ from Arm C for event-free survival, with a hazard ratio of 1.16, a 95% confidence interval of 0.87 to 1.55, and a one-sided P value of 0.8439. Prespecified subgroup analyses favored Arm A in patients with CIS and in those with T1 tumors. Subgroup estimates were 0.53 for CIS and 0.63 for T1, with corresponding 36-month event-free probabilities of 83.0% versus 71.8% and 81.3% versus 72.2%. This pattern favored the combination only when BCG maintenance was retained.
Among patients with CIS at randomization, complete response at any time occurred in 89.8% of Arm A and 85.2% of Arm C. Median duration of complete response was not reached in either arm, and 36-month durability after complete response was 91.7% with Arm A and 67.7% with Arm C. Biopsy-confirmed complete response rates were similar between arms by both investigator assessment and central review. Patient-reported global health status on the EORTC QLQ-C30 changed by minus 5.7 points in Arm A and minus 1.0 point in Arm C, which did not meet the study threshold for a clinically meaningful difference. Baseline PD-L1 high and low distribution in analyzed tumor tissue was similar across treatment groups.
Treatment-related adverse events were more common in the sasanlimab-containing arms, occurring at any grade in 87.1% of Arm A, 79.0% of Arm B, and 70.2% of Arm C. Grade 3 or higher treatment-related events occurred in 29.1%, 21.8%, and 6.3%, respectively, while any-grade immune-related adverse events occurred in 42.6% of Arm A and 46.8% of Arm B. Grade 3 or higher immune-related adverse events occurred in 15.7% and 14.1%, and immune-related hepatitis was the most common severe immune-related event. Discontinuation due to treatment-related adverse events involved sasanlimab and BCG in both combination arms, systemic corticosteroids were given in 69 and 70 patients in Arms A and B, and no treatment-related deaths occurred in Arms A or C, although two occurred in Arm B. Interim overall survival showed no difference between groups. Median overall survival was not reached, and 91 deaths occurred overall, with 32, 30, and 29 deaths in Arms A, B, and C, while the open-label design and low number of deaths limited survival interpretation.