BARCELONA—Pembrolizumab did not significantly improve overall survival (OS) or progression-free survival (PFS) compared with paclitaxel as second-line therapy in patients with advanced gastric or gastroesophageal junction cancer that had progressed after one line of chemotherapy (containing a platinum and fluoropyrimidine). Results were reported from the KEYNOTE-061 randomized, open-label, controlled phase III trial reported at the ESMO World Congress on Gastrointestinal Cancer (Lancet 2018;392(10142):123-133).
The lack of statistical benefit was despite the fact that the analysis was with patients who had been selected because they had PD-L1 combined positive scores (CPS) of 1 or higher—indicating at least a 1 percent positive staining rate in their tumors—marking them as theoretically ideal candidates for therapy with this checkpoint inhibitor antibody that targets PD-1 molecular features.
The findings call into question the simplistic concept that targeting a molecular feature on a particular tumor will necessarily be sufficient to work as a method of treating that cancer.
Lead author Kohei Shitara, MD, Chief of the Department of Gastrointestinal Oncology at the National Cancer Center Hospital East in Kashiwa, Japan, did note that sub-group analysis yielded positive findings for some patients and that pembrolizumab had a better safety profile than paclitaxel.
“The trial did not meet the primary endpoints to show survival improvement [or] PFS improvement,” he said. But some of the patients showed “very durable benefit” with pembrolizumab. The sub-group study confirmed that patients with good performance status and having at least 10 percent staining for PD-L1 expression (defined as a CPS score of 10) had improved prospects—as did those with tumors expressing high microsatellite instability (MSI).
A total of 592 patients were enrolled in KEYNOTE-061 at a total of 148 medical centers in 30 countries. Of those, 395 had PD-L1 CPS scores of at least 1 and were randomized to be treated with pembrolizumab (196 patients) or to join the control group (199 patients) allocated to paclitaxel therapy.
After a median follow-up of 8 months, 7.8 percent of patients had completed their pembrolizumab therapy or were still being treated with it.
Median OS was 9.1 months for patients treated with pembrolizumab and 8.3 months among those on paclitaxel treatment—a non-significant difference. There was also no statistically significant difference between the arms of the study in the numbers of patients alive at 12 months or at 18 months follow-up.
Median PFS was 1.5 months with pembrolizumab as compared with 4.1 months in patients receiving paclitaxel. But responses to the antibody were more durable (a median of 18.0 months compared with 5.2 months) than with the taxane.
A marked difference emerged in toxicity. Grade 3-5 drug-related adverse event rates were 14.3 percent with pembrolizumab and 34.8 percent with paclitaxel.
When Shitara was asked why the fully human monoclonal antibody (IgG1) ramucirumab had not been combined with paclitaxel for the control group, he explained this was because the need for a multinational cohort of patients constrained the researchers to use paclitaxel alone because it had been available and approved globally as a second-line agent.
“In the U.S. and Japan, ramucirumab with paclitaxel was already an approved combination in 2014 when the trial began but was not available in most countries. [But] paclitaxel had been approved by regulatory authorities in all countries participating in this trial,” he said.
The lack of superiority for pembrolizumab (even to an older, possibly less-efficacious regimen) underlines the fundamental clinical implications of the KEYNOTE findings.
In long-term follow-up, a benefit from pembrolizumab emerged in some patients. There had been “clinically meaningful” 12-month and 18-month survival estimates (about 40% and 26%), and subgroup analyses had suggested the effectiveness of pembrolizumab “might be more pronounced in patients with better performance status, greater levels of PD-L1 expression, and in those with tumors found to have high levels of microsatellite instability,” Shitara said.
He noted that their findings about the association between high MSI and pembrolizumab sensitivity added more data to confirm study findings that had already led to the recent licensing of the antibody for some solid tumors with high MSI.
Summing up his take on the KEYNOTE-061 findings, Shitara said the non-superiority of pembrolizuab therapy in patents with CPS 1 scores of PD-L1 expression had important clinical implications. “This means that CPS 1 is not a sufficient biomarker in the second line to use pembrolizumab. But this trial showed that specific sub-groups may achieve benefit with pembrolizumab even in second line.”
Sarah Maxwell and Peter M. Goodwin are contributing writers.