Preliminary data suggest that a certain pattern of gut microbes may be useful in predicting which patients with advanced gastric cancer are likely to benefit from treatment with the immunotherapy nivolumab (Opdivo).
Researchers have demonstrated bacterial invasion of the epithelial cell pathway in the gut microbiome and suggest that this could potentially become a novel biomarker.
“In addition, we found gastric cancer–specific gut microbiome predictive of responses to immune checkpoint inhibitors,” said study author Yu Sunakawa, MD, PhD, an associate professor in the Department of Clinical Oncology at the St. Marianna University School of Medicine, Kawasaki, Japan.
Sunakawa presented the study’s results during the Gastrointestinal Cancers Symposium (GICS) 2021, which was held online this year.
The gut microbiome holds great interest as a potential biomarker for response. Previous studies suggested that it may hold the key to immunotherapy responses. The concept has been demonstrated in several studies involving patients with melanoma, but this is the first study in patients with gastric cancer.
Nivolumab monotherapy has been shown to provide a survival benefit with a manageable safety profile in previously treated patients with gastric cancer or gastroesophageal junction (GEJ) cancer, Sunakawa noted. However, fewer than half of patients responded to therapy.
“The disease control rate was about 40%, and many patients did not experience any tumor degradation,” he said. “About 60% of the patients did not respond to nivolumab as a late-line therapy.”
In the observational/translational DELIVER trial, investigators enrolled 501 patients with recurrent or metastatic adenocarcinoma of the stomach or GEJ. The patients were recruited from 50 sites in Japan.
The primary endpoint was the relationship between the genomic pathway in the gut microbiome and efficacy of nivolumab and whether there was progressive disease or not at the first evaluation, as determined in accordance with RECIST criteria.
Genomic data were measured by genome shotgun sequence at a central laboratory. Biomarkers were analyzed by Wilcoxon rank sum test in the first 200 patients, who constituted the training cohort. The top 30 candidates were validated in the last 300 patients (the validation cohort) using the Bonferroni method.
Clinical and genomic data were available for 437 patients (87%). Of this group, 180 constituted the training cohort, and 257, the validation cohort.
The phylogenetic composition of common bacterial taxa was similar for both cohorts.
In the training cohort, 62.2% of patients had progressive disease, as did 53.2% in the validation cohort. The microbiome was more diverse among the patients who did not have progressive disease than among those who did have progressive disease.
The authors note that although there was no statistically significant pathway to be validated for a primary endpoint using the Bonferroni method, bacterial invasion of epithelial cells in the KEGG pathway was associated with clinical outcomes in both the training cohort (P = .057) and the validation cohort (P = .014). However, these pathways were not significantly associated with progressive disease after Bonferroni correction, a conservative test that adjusts for multiple comparisons.
An exploratory analysis of genus showed that Odoribacter and Veillonella species were associated with tumor response to nivolumab in both cohorts.
Sunakawa noted that biomarker analyses are ongoing. The researchers are investigating the relationships between microbiome and survival times, as well as other endpoints.
Still Some Gaps
In a discussion of the study, Jonathan Yeung, MD, PhD, of the Princess Margaret Cancer Center, Toronto, Canada, congratulated the investigators on their study, noting that “the logistical hurdles must have been tremendous to attain these data.”
However, Yeung pointed out some limitations and gaps in the data that were presented. For example, he found that the ratio of the training set to the validation set was unusual. “The training set is usually larger and usually an 80/20 ratio,” he said. “In their design, the validation set is larger, and I’m quite curious about their rationale.
“The conclusion of the study is that a more diverse microbiome was observed in patients with a tumor response than in those without a response,” he continued, “but they don’t actually show the statistical test used to make this conclusion. There is considerable overlap between the groups, and more compelling data are needed to make that conclusion.”
Another limitation was the marked imbalance in the number of patients whose condition responded to nivolumab in comparison with those whose condition did not (20 vs 417 patients). This could have affected the statistical power of the study.
But overall, Yeung congratulated the authors for presenting a very impressive dataset. “The preliminary data are very interesting, and I look forward to the final results,” he said.
The study was funded by Ono Pharmaceutical and Bristol-Myers Squibb, which markets nivolumab. Sunakawa has received honoraria from Bayer Yakuhin, Bristol-Myers Squibb Japan, Chugai, Kyowa Hakko Kirin, Lilly Japan, Nippon Kayaku, Sanofi, Taiho, Takeda, Yakult Honsha. He has held a consulting or advisory role for Bristol-Myers Squibb Japan, Daiichi Sankyo, and Takeda and has received research funding from Chugai Pharma, Daiichi Sankyo (inst), Lilly Japan (inst), Sanofi, Taiho Pharmaceutical, and Takeda (inst).
Gastrointestinal Cancers Symposium (GICS) 2021: Abstract 161, presented January 15, 2021