Eric Van Cutsem, MD, PhD

Invited discussant Eric Van Cutsem, MD, PhD, of University Hospitals Leuven, Belgium, commented that both ATTRACTION-02 and KEYNOTE-059 suggest that anti-PD [programmed cell death protein] antibodies have activity in advanced gastric cancer, but their findings differed with regard to the impact of programmed cell death ligand 1 (PD-L1) expression. 

“Nivolumab [Opdivo] was active in Asian patients regardless of PD-L1 expression, and I believe these data are robust,” he said. “The question is whether we can extrapolate that to Western patients. My guess, based on circumstantial data, is that it’s likely we can, but we want to see data in Western patients.”

He continued, “Pembrolizumab [Keytruda] showed higher activity in PD-L1–positive patients but also had some activity in PD-L1–negative patients. In the large cohort 1, the response in a pretreated population was 12%, similar to that in Asian patients in the nivolumab study. In the small group of patients in cohort 2, the results are hypothesis-generating, suggesting there may be some activity when we combine pembrolizumab with chemotherapy in the first line,” he said. 

“Cohort 3 is a small group but more interesting, evaluating a checkpoint inhibitor in the first line as monotherapy. In PD-L1–positive patients they got a nice result—77% had tumor reduction and there were long-lasting responses,” he noted. “Pembrolizumab’s activity in the first-line setting as a single agent in PD-L1–positive patients is very appealing, and this will be looked at in a larger study that is currently accruing.”

The difference between the studies in terms of the impact of PD-L1 expression is hard to explain, but this might be accounted for by the use of different antibodies and platforms, he suggested. In lung cancer, PD-L1 expression data have also come from different antibodies and platforms, and these studies have shown interobserver lack of agreement. 

“It would be nice if we could compare outcomes based on the same antibody and platform,” he commented. While acknowledging caveats regarding cross-study comparisons, Dr. Van Cutsem concluded that survival appears a bit better in PD-L1–positive patients. 

“We need to validate PD-L1 testing and other biomarkers,” he added. “It’s clear that checkpoint inhibitors have activity in some patients with advanced gastric cancer. The challenge is to find the predictive factors.” ■

DISCLOSURE: Dr. Van Cutsem has been a consultant for and/or ison the advisory board of Bayer, Lilly, Roche, SERVIER; and his institution has recevied research funding from Amgen, Bayer, Boehringer Ingelheim, Lilly, Novartis, Roche, Sanofi, Celgene, Ipsen, Merck, Merck KGaA, and SERVIER.