Perioperative Pembrolizumab Plus Chemo Does Not Significantly Improve EFS in MIBC

Adding perioperative pembrolizumab (Keytruda) to neoadjuvant cisplatin and gemcitabine in patients with cisplatin-eligible muscle-invasive bladder cancer (MIBC) did not significantly improve event-free survival (EFS) compared with placebo plus chemotherapy, according to results presented by Jens Bedke, MD, from the phase 3 KEYNOTE-866 trial (NCT03924856) given at the 2026 European Society for Medical Oncology (ESMO) Congress.

The randomized, double-blind, placebo-controlled study was presented during a proffered paper session on genitourinary tumors. Across 155 sites, 907 patients were randomly assigned 1:1 between July 19, 2019, and September 5, 2022. In the experimental arm, 453 patients received pembrolizumab with chemotherapy, and in the control arm, 454 received placebo with chemotherapy.

In the neoadjuvant phase, patients received 4 cycles of pembrolizumab 200 mg or placebo every 3 weeks with cisplatin 70 mg/m² on day 1 and gemcitabine 1000 mg/m² on days 1 and 8. All patients then underwent radical cystectomy with pelvic lymph node dissection, followed by up to 13 adjuvant cycles of pembrolizumab or placebo. Eligible patients were adults with clinical stage T2-T4aN0M0 or T1-T4aN1M0 disease, at least 50% urothelial histology, a glomerular filtration rate of at least 60 mL/min, and an ECOG performance status of 0 or 1. Randomization was stratified by PD-L1 combined positive score (CPS of 10 or higher vs less than 10), clinical stage, and geographic region.

EFS by investigator assessment was the primary end point, with overall survival (OS) and pathologic complete response (pCR) being the key secondary end points. Under the statistical design, OS and pCR could be formally tested only if the EFS result was positive.

At the second prespecified interim analysis, which was the final EFS analysis (data cutoff, January 26, 2026), the median time from randomization to data cutoff was 60.9 months (range, 40.7-78.7). EFS events occurred in 157 patients (34.7%) in the pembrolizumab arm and 168 patients (37.0%) in the placebo arm. The hazard ratio (HR) was 0.88 (95% CI, 0.71-1.10; 1-sided P = .1366), which did not cross the prespecified efficacy boundary of .0225. The median EFS was 74.1 months (95% CI, 72.4-not reached [NR]) with pembrolizumab and was NR (95% CI, 67.7-NR) with placebo. The 24-month EFS rates were 76.7% and 70.0%, respectively, and the 60-month rates were 63.5% and 61.9%.

Results were generally consistent across key subgroups. For patients with a PD-L1 CPS of 10 or higher, the EFS HR was 0.87 (95% CI, 0.65-1.19). For those with a CPS below 10, it was 0.94 (95% CI, 0.68-1.28).

Moreover, the OS was similar between the arms. There were 135 deaths in each arm (HR, 1.00; 95% CI, 0.79-1.27), and the 60-month OS rates were 70.8% with pembrolizumab and 70.9% with placebo. The pCR rate, assessed by central pathology review, was numerically higher with pembrolizumab: 44.4% (95% CI, 39.8%-49.2%) vs 35.8% (95% CI, 31.3%-40.4%), an estimated difference of 8.9% (95% CI, 2.5%-15.1%). Neither secondary end point was formally tested for statistical significance.

Cystectomy was performed in 415 patients in the pembrolizumab arm and 399 in the placebo arm. Of these, 318 and 315 patients, respectively, started adjuvant treatment.

Treatment-emergent adverse events (TEAEs) of any grade occurred in 99.8% of patients who received pembrolizumab and 98.7% of those who received placebo. Grade 3 to 5 TEAEs occurred in 80.2% and 75.1%, and serious TEAEs in 61.8% and 55.2%, respectively. TEAEs led to discontinuation of any drug in 36.7% of the pembrolizumab arm and 26.7% of the placebo arm, and to death in 3.8% and 4.5%. Each arm had 1 treatment-related death: multiple organ dysfunction syndrome with pembrolizumab and hemophagocytic lymphohistiocytosis with placebo.

The most common TEAEs in the pembrolizumab arm were nausea (54.9%), anemia (54.4%), and neutropenia (53.1%). Immune-mediated adverse events and infusion reactions occurred in 28.9% of patients receiving pembrolizumab and 9.4% of those receiving placebo, with grade 3 to 5 events in 8.9% and 1.8%, respectively. Hypothyroidism was the most frequent of these events with pembrolizumab, occurring in 10.4% of patients.

Bedke, a senior consultant in the Department of Urology & Transplantation Surgery at the Eva Mayr-Stihl Cancer Center Stuttgart, Klinikum Stuttgart in Germany, noted that outcomes in the placebo arm exceeded conventional benchmarks for neoadjuvant cisplatin plus gemcitabine followed by cystectomy. He also reported that the safety profile of perioperative pembrolizumab was manageable, with no new safety signals. He concluded that perioperative pembrolizumab with neoadjuvant cisplatin-based chemotherapy is not supported in this population, and that perioperative enfortumab vedotin (Padcev) plus pembrolizumab has become a standard of care in MIBC, regardless of cisplatin eligibility.

Disclosures

Bedke reported receiving research funding to his institution from Astellas Pharma, Daiichi Sankyo Europe GmbH, Johnson & Johnson, Merck Sharp & Dohme LLC (MSD), Pfizer, Roche, and Seagen; reported consulting or advisory roles with Apogepha, Astellas Pharma, AstraZeneca, Bristol Myers Squibb, Daiichi Sankyo Europe GmbH, Eisai, Ipsen, Janssen, MSD, Pfizer, and Roche; reported speakers' bureau participation for Apogepha, Astellas Pharma, Bayer, Bristol Myers Squibb, Ipsen, MSD, and Pfizer; reported travel, accommodations, or expenses paid to his institution by Ipsen and Johnson & Johnson; and received study funding, medical writing, and editorial support from MSD.

Reference

Bedke J, Arkosy P, Nishiyama H, et al. Perioperative pembrolizumab (pembro) or placebo (pbo) + neoadjuvant cisplatin (cis) and gemcitabine (gem) in participants (pts) with muscle-invasive bladder cancer (MIBC): the double-blind, phase 3 KEYNOTE-866 study. Presented at: 2026 European Society for Medical Oncology Congress; October 23-27, 2026; Madrid, Spain. Abstract 4735O