As KEYNOTE-181 had three primary endpoints (overall survival in each of three populations), P values required for statistical significance were defined accordingly. “The study was positive if one of the primary endpoints was met,” Dr. Kojima explained at the symposium, which was sponsored by the American Gastroenterological Association, the American Society for Clinical Oncology, the American Society for Radiation Oncology, and the Society of Surgical Oncology.
Some 35% of trial patients had a PD-L1 CPS of 10 or greater. In this population, median overall survival was 9.3 months with pembrolizumab and 6.7 months with chemotherapy. The hazard ratio was 0.69, with the P value (.0074) meeting that predefined for statistical significance in this population (less than or equal to .0085). The 12-month rate of overall survival was 43% and 20%, respectively.
About 64% of trial patients had squamous cell carcinoma histology. In this population, median overall survival was 8.2 months with pembrolizumab and 7.1 months with chemotherapy. The hazard ratio was 0.78, but the P value (.0095) did not meet that predefined for statistical significance in this group (less than or equal to .0077).
Finally, in the entire intention-to-treat population, median overall survival was identical, at 7.1 months, with pembrolizumab and with chemotherapy. The hazard ratio was 0.89 in favor of the antibody, but the P value (.0560) did not meet that predefined for statistical significance in this population (less than or equal to .0077).
A similar pattern was seen for other outcomes, with patients having PD-L1 CPS greater than or equal to 10 deriving greatest benefit from pembrolizumab over chemotherapy in terms of progression-free survival (hazard ratio, 0.73), response rate (21.5% vs. 6.1%), and median duration of response (9.3 vs. 7.7 months).
“Toxicity profiles were in line with previous reports of each treatment. No new safety signals were observed,” Dr. Kojima reported. Pembrolizumab was associated with a higher rate of immune-mediated and infusion reactions (23.2% vs. 7.4%), but lower rates of most gastrointestinal and hematologic adverse events.
Dr. Kojima disclosed ties to Oncolys BioPharma, Astellas, Amgen, MSD, Ono Pharmaceutical, and Shionogi. Merck Sharp & Dohme sponsored the trial.
SOURCE: Kojima T et al. GI Cancers Symposium , https://meetinglibrary.asco.org/record/169377/abstract.