A team of investigators from Japan detected several genomic pathways and bacterial genera in the pre-treatment gut microbiome of patients with advanced gastric cancer that were significantly associated with survival times following treatment with nivolumab. These findings were presented at the ESMO World Congress on Gastrointestinal Cancer 2021, which was held 30 June to 3 July.

Dr. Yu Sunakawa of the St. Marianna University School of Medicine in Kawasaki, Japan noted that, although previous studies have found that the efficacy of anti-PD1-based immunotherapy was associated with the composition of the gut microbiome in various types of cancers, little is known about this relationship in gastric cancer. Therefore, Dr. Sunakawa and co-investigators were prompted to conduct the observational/translational DELIVER study (JACCRO GC-08: UMIN000030850), with the aim of identifying novel host-related biomarkers for nivolumab treatment in advanced gastric cancer using foecal samples.

This team has already reported that the genomic pathway may predict poor response to nivolumab in patients with advanced gastric cancer and also identified a gastric cancer-specific gut microbiome as a putative clinical biomarker. (1)

At ESMO Congress on Gastrointestinal Cancer 2021, results of an analysis of data was presented from the DELIVER study that investigated the associations between the genomic pathway and the genus of the gut microbiome with survival time following nivolumab treatment in advanced gastric cancer.

From March 2018 to August 2019, 501 patients with advanced gastric cancer from 67 institutes were treated with single agent nivolumab. Clinical data and foecal samples prior to nivolumab treatment were collected prospectively and genomic data were measured by genome shotgun sequence at a central lab. The biomarker analysis planned to use a Cox regression model in the first 200 patients to form the training cohort. The top 30 candidates in the training cohort were chosen according to an ascending order of p value and were validated by using the Bonferroni method in the last 301 patients, who became the validation cohort.

An association was found between the nucleotide metabolism pathway and both overall and progression-free survival, suggesting this pathway may be a biomarker of nivolumab efficacy in advanced gastric cancer

The training cohort ultimately comprised 190 patients and the validation cohort contained 277 patients who were available for massive metagenomic and survival analyses.

The training cohort demonstrated median overall survival (OS) of 5.78 months and the median progression-free survival (PFS) was 1.79 months. In the validation cohort, median OS and PFS were 5.95 and 1.84 months, respectively.

Regarding the OS analysis, the investigators detected 24 pathways and 6 genera in the top 30 candidates with p value of < 0.05 in the training cohort. Of these, 2 pathways and 1 genus were confirmed as statistically significant in the validation cohort, which included the purine metabolism (hazard ratio [HR] 0.31; p = 0.00064) and peptidoglycan biosynthesis (HR 0.27; p = 0.00088) pathways, as well as Corynebacterium (HR 1.1; p < 0.0001).

The PFS analysis identified 20 pathways, 9 genera, and 1 phylum with p < 0.05 in the training cohort. The nucleotide metabolism pathway and Porphyromonas were associated with PFS in the validation cohort by Bonferroni method (HR 0.39; p = 0.0013 and HR 3.37; p < 0.0001), respectively.

The nucleotide metabolism pathway was found to associate with both OS and PFS, but this observation did not reach statistical significance regarding OS per Bonferroni method. 


Based on these findings from the translational study, the authors concluded that several genomic pathways and genera in the gut microbiome were significantly associated with survival time following nivolumab treatment in patients with advanced gastric cancer.

They also suggested that the nucleotide metabolism pathway could potentially become a biomarker for prediction of the prognosis with nivolumab treatment in advanced gastric cancer.

Funding from Ono Pharmaceutical and Bristol Myers Squibb was disclosed.

Prof. Yelena Y. Janjigian of the Memorial Sloan Kettering Cancer Center in New York, NY, US who discussed the study data said that majority of patients had ECOG performance status 0 and 1 (86%), with 60% patients > third-line setting, 20% had HER2-positive tumours, but PD-L1 and MSI status were not reported. The overall response rate in 282 patients with RECIST measurable disease was 6.7%. Median PFS was < 2 months, OS < 6 months. The data highlights importance of biomarker selection and that we should use anti-PD-L1 therapy in first-line setting.

Prof. Janjigian commented excellent effort and coordination in terms of stool and blood collection (not tumour). Training and validation cohort is a reasonable approach. However, hypothesis was not well defined. The OS outcomes were associated with purine metabolism, peptidoglycan biosynthesis, and Corynebacterium; PFS outcomes with nucleotide metabolism pathway and Porphyromonas; while both OS and PFS associated with nucleotide metabolism pathway. It is surprising to mix pathways and genera in the training and validation cohorts. It would be helpful to comment on the phylum level. The study should have included an assessment of bacterial diversity. It is unclear how nucleotide metabolism pathway was selected for quartile analysis, according to Prof. Janjigian.


  1. Sunakawa Y, Matoba R, Inoue E, et al. Genomic pathway of gut microbiome to predict efficacy of nivolumab in advanced gastric cancer: DELIVER trial (JACCRO GC-08). Journal of Clinical Oncology 2021;39(3_suppl):161-161.


O13 – Sunakawa Y, Matoba R, Inoue E, et al. Gut Microbiome to Predict Survival Time in Advanced Gastric Cancer Treated With Nivolumab: the DELIVER Trial (JACCRO GC-08).  ESMO World Congress on Gastrointestinal Cancer 2021 (30 June – 3 July).