PFS and OS results
The addition of pembrolizumab was associated with improved PFS across most protocol-specified subgroups, Dr. Colombo and colleagues noted.
After a median follow-up of 22 months, the 12-month PFS rate in the biomarker-selected population (all patients with a PD-L1 CPS ≥ 1) was 45.5% for patients in the pembrolizumab group versus 34.1% in the placebo group. This translated into a hazard ratio (HR) for progression on pembrolizumab of 0.62 (P < .001).
The respective PFS rates in the ITT population were 44.7% and 33.5%, with an HR for progression of 0.65 (P < .001) with the checkpoint inhibitor.
In patients with PD-L1 CPS ≥ 10, the respective rates of PFS and the HR were 44.6%, 33.5%, and 0.58 (P < .001).
OS rates were also significantly improved, he noted.
The 12-month and 24-month OS rates in all patients with PD-L1 CPS ≥ 1 were 75.3% and 53%, respectively, for patients assigned to pembrolizumab versus 63.1% and 41.7% in patients assigned to placebo, translating to an HR for death with pembrolizumab in this group of 0.64 (P < .001).
In the all-comers (ITT) population, respective 12- and 24-month OS rates were 74.8% and 50.4% with pembrolizumab versus 63.6% and 40.4% with placebo. This difference translated into an HR for death with anti-PD-1 of 0.67 (P < .001).
Among patients with the higher PD-L1 levels (≥ CPS 10), the respective OS rates were 75.7% and 54.4% with pembrolizumab versus 61.5% and 44.6% with placebo (HR 0.61, P < .001).
Dr. Mirza emphasized that “we did not see any efficacy of pembrolizumab in the biomarker-negative population,” with an HR for PFS of 0.94 and HR for OS of 1.0 in this subgroup.
The most common grade ≥ 3 adverse events were anemia, which occurred in 30.3% of patients assigned to pembrolizumab compared with 26.9% in the placebo group, and neutropenias, which occurred in 12.4% and 9.7% of patients, respectively. One patient in the pembrolizumab group died from an immune-related event, encephalitis.
Despite his enthusiasm for the regimen, Dr. Mirza tempered it by pointing out that there was an imbalance in the sample sizes regarding histology, and a potential bias introduced by the failure to stratify by tumor histology.
He noted that in other studies checkpoint inhibitors have had only modest activity against adenocarcinomas, which were more frequent in the placebo group in KEYNOTE-826, resulting in a potential positive bias in favor of pembrolizumab.
KEYNOTE-826 is funded by MSD. Dr. Colombo has disclosed consultant, research, and promotional speaking activities for multiple companies. Dr. Mirza has disclosed personal financial interests with Merck and other companies.
A version of this article was first published on Medscape.com.