Increasing correlation between oral and gastric microbiota during gastric carcinogenesis

Abstract

Background/aims: Recent research has increasingly focused on the role of the gastric microbiome in the development of gastric cancer. We aimed to investigate the changes in the microbiome during gastric carcinogenesis in structural and functional aspects, with a specific focus on the association between oral and gastric microbiomes.

Methods: We collected saliva, gastric juice, and gastric tissue samples from 141 patients at different stages of gastric carcinogenesis and processed them for microbiome analysis using 16S rRNA gene profiling. The alpha and beta diversities were analyzed, and the differences in microbiome composition and function profiles were analyzed among the groups, as well as the correlation between changes in the oral and gastric microbiomes during carcinogenesis.

Results: We observed significant differences in microbial diversity and composition between the disease and control groups, primarily in the gastric juice. Specific bacterial strains, including Schaalia odontolytica, Streptococcus cristatus, and Peptostreptococcus stomatis, showed a significant increase in abundance in the gastric juice in the low-grade dysplasia and gastric cancer groups. Notably, the correlation between the oral and gastric microbiota compositions, increased as the disease progressed. Predictive analysis of the metagenomic functional profiles revealed changes in functional pathways that may be associated with carcinogenesis (ABC transport and two-component systems).

Conclusion: During gastric carcinogenesis, the abundance of oral commensals associated with cancer increased in the stomach. The similarity in microbial composition between the stomach and oral cavity also increased, implying a potential role of oral-gastric bacterial interactions in gastric cancer development.

Keywords: 16S rRNA; Carcinogenesis; Gastric cancer; Microbiome.

Figure 1

The alpha diversity (A) and beta diversity (B, C) at each stage of gastric carcinogenesis in the gastric juice, gastric tissue (antrum, body), and saliva samples. The alpha diversity was analyzed using the observed Chao1, Shannon, and Simpson indices, while the beta diversity was assessed using the Bray–Curtis method. The disease stages were categorized as control, low-grade dysplasia (LGD), high-grade dysplasia (HGD), and gastric cancer (GC). *p < 0.05, **p < 0.01, ***p < 0.001.

Figure 2

Microbial composition and linear discriminant analysis Effect Size (LEfSe) analysis across gastric cancer stages for the gastric juice, gastric tissue (antrum, body), and saliva samples. (A) The microbial composition changes at the phylum level in response to gastric carcinogenesis in the tissues (antrum and body) and saliva. (B) LEfSe analysis employing a cutoff linear discriminant analysis (LDA) score of 3.5 or higher for further analysis. After using the non-parametric Kruskal–Wallis test with a significance threshold (FDR-adjusted p value; q-value) of 0.05, we selected significant discriminative groups using the LDA within the LEfSe algorithm. Subsequently, data normalization for LEfSe analysis was conducted through relative log expression. Microbial taxa with high proportions in the control, low-grade dysplasia (LGD), high-grade dysplasia (HGD), and gastric cancer (GC) groups were color-coded as follows: red > orange > light blue > blue. The indicated strains have a value of p < 0.05, i.e., Pseudomonadota (formerly Proteobacteria), Bacillota (formerly Firmicutes), and Actinomycetota (formerly Actinobacteria).

Figure 3

Correlation of the microbial composition among different sample types and disease groups. The composition and ratio of strains of each microbial composition in the oral cavity, gastric juice, gastric antrum tissue, and gastric body tissue were confirmed by analyzing the similarity in Spearman’s rank correlation according to the disease stage between each sample. The graph illustrates changes in the degree of similarity based on gastric carcinogenesis. This enabled us to determine how the similarity levels among gastric juice, gastric tissue, and saliva changed in relation to gastric carcinogenesis. The similarity between the samples increased as the disease progressed. Using the Spearman rank test, the correlation coefficient for the relative microbial composition between the two types of specimens during gastric carcinogenesis was calculated. Spearman’s rho and p value were calculated using the cor.test in R. All results displayed in the graph indicate significant positive correlations (correlation coefficient > 0.2; p < 0.05). GJP, gastric juice; ORWP, saliva; TIA, tissue (antrum); TIB, tissue (body); LGD, low-grade dysplasia; HGD, high-grade dysplasia; GC, gastric cancer.

Figure 4

Metagenomic analysis reveals functional differences among the different sample types and disease groups. Linear discriminant analysis Effect Size (LEfSe) analysis identified functional profiles that exhibited changes with a linear discriminant analysis (LDA) score cut-off of 0.25 in response to gastric carcinogenesis. After employing the non-parametric Kruskal–Wallis test with a significance threshold (FDR-adjusted p value; q-value) of 0.05, we selected significant discriminative groups using the LDA within the LEfSe algorithm. Subsequently, data normalization for LEfSe analysis was conducted using relative log expression. Functional profiles with high proportions in the control, low-grade dysplasia (LGD), high-grade dysplasia (HGD), and gastric cancer (GC) groups were color-coded as follows: red > orange > light blue > blue.