MADRID — Including Pembrolizumab (Keytruda) in the first-line treatment of metastatic HER2-positive gastric or gastroesophageal cancer improved progression-free survival (PFS), but potential harm was seen in PD-L1-negative patients, the randomized KEYNOTE-811 trial showed.

In patients receiving standard chemotherapy plus trastuzumab, median PFS improved from 8.1 months with placebo to 10.0 months with pembrolizumab (HR 0.73, 95% CI 0.61-0.87), Yelena Janjigian, MD, of Memorial Sloan Kettering Cancer Center in New York City, reported at the annual European Society for Medical Oncology (ESMO) congress.

The PFS benefit was limited to the large subgroup of patients with PD-L1 overexpressing tumors, as measured by a combined positive score (CPS) of 1 or greater. Median PFS in this group improved from 7.3 months with placebo to 10.9 with the PD-1 inhibitor (HR 0.71, 95% CI 0.59-0.86), according to the findings, which were published concurrently in The Lancet.

While not mature, overall survival (OS) analyses at different time points favored the pembrolizumab-treated patients in both the intent-to-treat (ITT) population and PD-L1-positive subgroup, while OS appeared worse in the PD-L1-negative subgroup:

• ITT: 20 months with pembrolizumab vs 16.8 months with placebo (HR 0.84, 95% CI 0.70-1.01)
• PD-L1-positive: 20 vs 15.7 months, respectively (HR 0.81, 95% CI 0.67-0.98)
• PD-L1-negative: 16.1 vs 22.3 months (HR 1.61, 95% CI 0.98-2.64)

Based on the findings, pembrolizumab maker Merck announced in June that it would be working with the FDA to update the current indication in metastatic gastric or gastroesophageal cancer to limit treatment to PD-L1-positive patients.

While OS in the PD-L1-positive group was not a primary endpoint of KEYNOTE-811, “the sample size and survival advantages are substantial enough to warrant a change in clinical practice for this subgroup of patients,” wrote Elizabeth C. Smyth, MD, of the Churchill Hospital in Oxford, England, and Raghav Sundar, MBBS, PhD, of the National University of Singapore, in a commentary accompanying the study.

They also suggested that the results “raise provocative questions regarding early regulatory approvals based on preliminary trial datasets,” adding that the potential harm in patients with PD-L1-negative tumors, who had become eligible for treatment with pembrolizumab, emphasized the “importance of long-term outcomes.”

In 2021, the FDA granted accelerated approval to the checkpoint inhibitor in combination with trastuzumab and standard fluoropyrimidine- and platinum-containing chemotherapy for the first-line treatment of locally advanced unresectable or metastatic HER2-positive gastric or gastroesophageal junction adenocarcinoma. That approval was based on an early analysis of KEYNOTE-811 showing that the addition of pembrolizumab resulted in improved overall response rates over placebo (74% vs 52%).

After the ESMO presentation, discussant Florian Lordick, MD, PhD, of the University of Leipzig Medical Center in Germany, said that ESMO had updated its guidelines to recommend pembrolizumab only for patients with a CPS score of 1 or above.

“Don’t give pembrolizumab to those patients who are negative for PD-L1 and are HER2-positive,” commented Lordick.

Data analysis for KEYNOTE-811 included 698 patients enrolled across 168 medical centers in 20 countries worldwide. Patients were randomly assigned to receive chemotherapy (fluorouracil plus cisplatin or capecitabine plus oxaliplatin) and trastuzumab with either pembrolizumab or placebo.

Patients had a median age of 63 years, 81% were male, most were white (62%) or Asian (34%), and 85% had a CPS score of 1 or greater.

Grade ≥3 treatment-related adverse events (TRAEs) occurred in 58% of patients in the pembrolizumab group versus 51% in the placebo group. TRAEs that led to death occurred in four patients in the pembrolizumab group and three in the placebo group (1% each).

The most common TRAEs of any grade were diarrhea (47% in the pembrolizumab group vs 42% in the placebo group), nausea (44% in each group), and anemia (31% vs 33%).