Genomic pathway of gut microbiome to predict efficacy of nivolumab in advanced gastric cancer: DELIVER trial (JACCRO GC-08).

Abstract

161 Background: Nivolumab (Nivo) monotherapy demonstrated survival benefit in previously treated gastric cancer (GC) patients (pts) (Kang YK, et al. Lancet 2017). About 60% of the pts did not respond to Nivo in late-lines, and its predictive markers are needed in GC. Several studies revealed that the efficacy of anti-PD-1-based immunotherapy was associated with composition of gut microbiome in various types of cancers, but little is known about GC. We therefore investigated whether genomic information in gut microbiome will serve as predictors for Nivo in advanced GC. Methods: The observational/translational study (DELIVER trial: UMIN000030850) has enrolled 501 pts with advanced GC treated with Nivo alone from March 2018 to August 2019 to discover gut microbiome biomarkers using fecal samples which were collected before treatment with Nivo. Primary endpoint was the relationship between genomic pathway in gut microbiome and efficacy of Nivo, whether progressive disease (PD) or not at the first evaluation according to RECIST criteria. Secondary endpoints included several relationships between markers of microbiome and clinical outcomes. Genomic data were measured by genome shotgun sequence at central lab. Biomarkers were analyzed by Wilcoxon rank sum test in the first 200 pts (training cohort). The top 30 candidates in an ascending order of p-value were validated in the last 300 pts (validation cohort) using Bonferroni method. Results: One hundred eighty of 200 pts and 257 of 301 pts were available for massive metagenomic and clinical data in the training cohort and validation cohort, respectively. The PD was 62.2% (95%CI 54.7-69.3) in the training cohort and 53.2% (95%CI 47.0-59.4) in the validation cohort. More diverse microbiome was observed in pts with non-PD compared to pts with PD. Although there was no statistically significant pathway to be validated for a primary endpoint under Bonferroni method, we found that bacterial invasion of epithelial cells pathway in the KEGG pathway was associated with clinical outcome of Nivo in both training cohort ( p= 0.057) and validation cohort ( p= 0.014). Upregulation of the pathway was related to PD at the first evaluation for Nivo treatment. An exploratory analysis of genus showed that Odoribacter and Veillonella were associated with tumor response to Nivo in both cohorts. Conclusions: Our translational study indicated for the first time that bacterial invasion of epithelial cells pathway in gut microbiome may potentially become a novel biomarker for treatment with Nivo in advanced GC. In addition, we found gastric cancer-specific gut microbiome to predict response to immune checkpoint inhibitors. Clinical trial information: UMIN000030850.