Pembrolizumab as second-line therapy improved overall survival (OS) in patients with advanced or metastatic esophageal cancer and high PD-L1 expression compared with chemotherapy, according to findings from the phase 3 KEYNOTE-181 study presented at the 2019 Gastrointestinal Cancers Symposium.
Pembrolizumab was not significantly better than chemotherapy in patients without PD-L1 expression or in those with squamous cell histology, although a prespecified statistical compensation might have prevented the achievement of a significant difference between the treatment arms in patients with squamous cell carcinoma.
“These data suggest that pembrolizumab should be considered a new standard of care for patients with PD-L1 CPS [combined positive score] of 10 or more with metastatic esophageal cancer in the second-line setting,” said lead investigator Takashi Kojima, MD, Department of Gastroenterology and Gastrointestinal Oncology, National Cancer Center Hospital East, Kashiwa, Japan, who presented the KEYNOTE-181 findings.
Patients whose disease progresses after first-line chemotherapy have a poor prognosis and limited treatment options, Dr Kojima said.
KEYNOTE-181 Details
KEYNOTE-181 was a global open-label study of 628 patients with locally advanced or metastatic esophageal cancer whose disease had progressed after first-line chemotherapy. Patients were randomized to monotherapy with pembrolizumab 200 mg every 3 weeks for up to 2 years or to standard chemotherapy (ie, paclitaxel, docetaxel, or irinotecan) at the investigator’s discretion.
Of the 628 patients enrolled, 64% had squamous cell carcinoma of the esophagus and 35% had adenocarcinoma of the esophagus or Siewert type I adenocarcinoma of the gastroesophageal junction. The study had 3 prespecified primary end points of OS in the subgroup with a PD-L1 CPS ≥10 (N = 222), the subgroup with squamous cell carcinoma (N = 401), and the full intention-to-treat (ITT) population.
At the data cutoff point, pembrolizumab treatment was ongoing in 9 patients, and 5 patients completed 2 years of pembrolizumab therapy.
In the subgroup of patients with PD-L1 CPS ≥10, the median OS was 9.3 months in the pembrolizumab arm versus 6.7 months in the chemotherapy arm (hazard ratio, 0.69; P = .0074). In this subgroup, the 1-year OS was 43% with pembrolizumab and 20% with chemotherapy; the 18-month survival was 26% and 11%, respectively.
In the patients with squamous cell carcinoma, the median OS was numerically better with pembrolizumab versus chemotherapy (8.2 months vs 7.1 months, respectively), but the difference did not achieve statistical significance (P = .0095), which was set at .0077 because the P value had to be shared across the 3 populations to reduce the possibility of type I error.
In the ITT population, median OS was 7.1 months in each arm.
The objective response rate (ORR) was higher with pembrolizumab compared with chemotherapy in all 3 patient populations. In those with a PD-L1 CPS ≥10, the ORR was 21.5% in the pembrolizumab arm and 6.1% in the chemotherapy arm; in the squamous cell carcinoma subgroup, the ORR was 16.7% versus 7.4%, respectively; and in the ITT population, it was 13.1% versus 6.7%, respectively.
The median duration of response in the subgroup with a high PD-L1 CPS score was 9.3 months in the pembrolizumab arm compared with 7.7 months in the chemotherapy arm. In the squamous cell carcinoma population, median duration of response was 8.5 months with pembrolizumab compared with 10.7 months with chemotherapy.
As expected, pembrolizumab had far fewer grade 3 to 5 treatment-related adverse events than chemotherapy, 18.2% versus 40.9%, respectively.
In the ITT population, “the study failed to meet its primary end point of OS. So pembrolizumab is not indicated in unselected esophageal cancer patients,” said invited discussant Harry H. Yoon, MD, MHS, Co-Chair, Gastroesophageal Cancer Disease Group, Mayo Clinic, Rochester, MN. He speculated that in the squamous cell carcinoma subgroup, the OS difference in favor of pembrolizumab might have been significant if there had been only 2 primary end points instead of 3.
Dr Yoon called the results in the high PD-L1–expressing subgroup “clinically meaningful and statistically significant,” pointing out that patients in this subgroup who were randomized to pembrolizumab had better Eastern Cooperative Oncology Group performance scores than those randomized to chemotherapy and had a higher proportion of patients with squamous cell carcinoma, who are known to respond better to pembrolizumab.
Nevertheless, there is biologic rationale for pembrolizumab to perform better in patients with high PD-L1 expression, he said, which explains the higher response rates in tumors that had at least some PD-L1 expression (CPS ≥1) than in tumors that had no PD-L1 expression, 16% versus 6%, respectively.
