Mucosal microbiome dysbiosis in gastric carcinogenesis

Abstract

Objectives: We aimed to characterise the microbial changes associated with histological stages of gastric tumourigenesis.

Design: We performed 16S rRNA gene analysis of gastric mucosal samples from 81 cases including superficial gastritis (SG), atrophic gastritis (AG), intestinal metaplasia (IM) and gastric cancer (GC) from Xi'an, China, to determine mucosal microbiome dysbiosis across stages of GC. We validated the results in mucosal samples of 126 cases from Inner Mongolia, China.

Results: We observed significant mucosa microbial dysbiosis in IM and GC subjects, with significant enrichment of 21 and depletion of 10 bacterial taxa in GC compared with SG (q<0.05). Microbial network analysis showed increasing correlation strengths among them with disease progression (p<0.001). Five GC-enriched bacterial taxa whose species identifications correspond to Peptostreptococcus stomatis, Streptococcus anginosus, Parvimonas micra, Slackia exigua and Dialister pneumosintes had significant centralities in the GC ecological network (p<0.05) and classified GC from SG with an area under the receiver-operating curve (AUC) of 0.82. Moreover, stronger interactions among gastric microbes were observed in Helicobacter pylori-negative samples compared with H. pylori-positive samples in SG and IM. The fold changes of selected bacteria, and strengths of their interactions were successfully validated in the Inner Mongolian cohort, in which the five bacterial markers distinguished GC from SG with an AUC of 0.81.

Conclusions: In addition to microbial compositional changes, we identified differences in bacterial interactions across stages of gastric carcinogenesis. The significant enrichments and network centralities suggest potentially important roles of P. stomatis, D. pneumosintes, S. exigua, P. micra and S. anginosus in GC progression.

Keywords: Gastric cancer; mucosal microbiome dysbiosis; oral bacteria.

Figure 1

Mucosal microbiome dysbiosis across stages of gastric carcinogenesis. Decreased microbial richness, estimated by Chao1, in IM and GC compared with SG (A). Model-free feature screening and logistic regression were used to select differentially abundant bacteria adjusted for age, gender and Helicobacter pylori status. Log2 fold change relative abundances of GC-enriched and GC-depleted bacteria compared with SG, q values <0.05 (B). Significantly increased percentage of oral bacteria were observed in GC compared with SG, AG and IM. AG, atrophic gastritis; GC, gastric cancer; IM, intestinal metaplasia; SG, superficial gastritis (C). (Statistical significance was determined by Mann-Whitney U test, *p<0.05).

Figure 2

Correlation strengths of gastric cancer (GC)-enriched and GC-depleted bacteria increased with disease progression. Correlation networks of GC-enriched and GC-depleted operational taxonomic units (OTUs) in: superficial gastritis, atrophic gastritis, intestinal metaplasia and GC. SparCC algorithm was used to estimate correlation coefficients and adjust for compositional effects. Cytoscape V.3.4.0 was used for network construction. A subset of significant correlations with strengths of at least 0.2 were selected for visualisation. The size and colour of the nodes correspond to weighted node connectivity (WNC) scores). Peptostreptococcus_OTU16 (3), Parvinomonas_OTU35 (6) and Streptococcus_OTU68 (8), Dialister_OTU151 (13), Slackia_OTU174 (14) have significant WNC scores (p<0.05).

Figure 3

Gastric cancer enriched markers with significant centralities. Fold change analyses of gastric cancer-enriched markers with significant weighted node connectivity scores in microbial association network, figure 1. p values were adjusted by FDR method, *q<0.05 (A). The diagnostic performance of Peptostreptococcus_OTU16 Parvinomonas_OTU35, Streptococcus_OTU68, Dialister_OTU151 and Slackia_OTU174 indicated by receiver operating characteristic curve analysis in discovery cohort, area under the receiver-operating curve of 0.82 (B). AG, atrophic gastritis; GC, gastric cancer; IM, intestinal metaplasia; SG, superficial gastritis.

Figure 4

Helicobacter pylori infection reduces the number of gastric microbiome interactions. Correlation strengths of gastric cancer-enriched and gastric cancer-depleted bacteria in Helicobacter pylori negative (HPN) and Helicobacter pylori positive (HPP) samples: superficial gastritis (HPN: n=12; HPP: n=12), atrophic gastritis (HPN: n=22; HPP: n=21), intestinal metaplasia (HPN: n=18; HPP: n=18), gastric cancer (HPN: n=7; HPP: n=7). SparCC algorithm was used to estimate correlation coefficients and adjust for compositional effects. (Statistical significance was determined by Fisher’s exact test).

Figure 5

Validations of GC associated bacteria in independent cohort. Correlations of the fold changes of gastric cancer (GC)-enriched and GC-depleted bacteria compared with superficial gastritis (SG) between Xi’an (discovery) and Inner Mongolia (validation) cohorts. Spearman correlation coefficient r=0.506, p=0.00437 (A). The diagnostic performance of Peptostreptococcus_OTU16, Parvinomonas_OTU35, Streptococcus_OTU68, Dialister_OTU151 and Slackia_OTU174 indicated by receiver operating characteristic curve analysis in validation cohort, area under the receiver-operating curve of 0.81 (B).