In this study, the gastric microbiota composition was characterized between GC cases and controls. Calculation of the MDI by applying compositional analysis of microbiome data is a novel approach to identify the gastric dysbiosis and further to observe the associations with GC risk. Moreover, the metagenomics functions were predicted to identify the pathways associated with GC risk. The MDI was higher for GC cases than healthy controls for the total population. A significant higher MDI was observed for GC cases than controls in the female population (p = 0.002). In females, those who are in the third tertile of the MDI showed a significantly increased risk of GC (OR: 2.66, 95% CI: 1.19–5.99, p-trend = 0.017; model II). NMDS results indicated that the microbial composition of HP positive and negative groups was significantly different (PERMANOVA; p = 0.001). Regarding KEGG pathways, biosynthesis of ansamycins pathway was identified as a critical pathway which is differentially abundant in healthy controls.
In our study, Proteobacteria, Firmicutes, Bacteroidetes, and Fusobacterium were the most dominant phyla in case and control groups. Comparison with the previous research findings, Proteobacteria, Firmicutes, Bacteroidetes, Actinobacteria, and Fusobacteria was identified as the most dominant phyla in the gastric environment which shows consistent results with the current study [19, 22, 37–39]. A study conducted using Chinese and Mexican populations revealed that HP was the predominant member of the microbiota in the gastric environment . A study aimed to evaluate the microbial composition of the gastric mucosa found that the gastric microbial composition of the patients who had non-atrophic gastritis, intestinal metaplasia, and GC of the intestinal type was with Firmicutes and Proteobacteria phyla which account 70% of in each sample . A similar result was observed in our study indicating that Bacteroidetes, Firmicutes, Proteobacteria are dominant phyla in the gastric environment in GC cases and controls. Another study evaluated the gastric microbiota in individuals based on different histopathological stages of GC reported that there were 19 genera with average relative abundance > 0.5% across 60 samples at the genus level where the high abundance genera were Helicobacter, Flavobacterium, Haemophilus, and Serratia . The finding of this study related to the Helicobacter, Haemophilus, and Neisseria genera are consistent with our findings indicating those three genera are most dominant in both GC cases and control groups. A Korean study observed that the Epsilonproteobacteria class corresponding to HP species was the predominant, but the abundance of Bacilli class was relatively increased in the GC group which is consistent with our findings .
A study observed that there were a higher evenness and diversity of the gastric microbiota in the GC group in comparison with chronic gastritis and intestinal metaplasia groups . We observed that the evenness was more or less similar in both GC cases and controls. Interestingly, richness was significantly higher in GC cases in the current study population. A similar finding has been reported in the study compared cancer tissues with non-cancer tissues where there is a high microbial richness in GC tissues . Moreover, we observed that there was a higher Shannon index in the controls than the GC group. Another study that compared the chronic gastritis group with the GC group observed that there is a higher Shannon index in the chronic gastritis group compared to the GC group . A study focused on the relationship between gastric dysbiosis and GC development found that there is an increased richness although the Shannon index is lower in the GC group compared to controls which is similar to our findings . On the contrary, a study conducted on microbiota in gastric mucosa in the GC tissues compared with the non-cancer tissues revealed that there is a significantly higher Shannon index in the cancer group compared with the healthy controls . A study reported that 75.86% was captured by the first two principal coordinates in the PCoA beta diversity plot according to the weighted UniFrac phylogenetic distance measure. They have observed that there was a significant divergence between non-cancer and cancer samples since those samples were clustered separately . A study conducted in Mexica performed an ordination analysis of the 44 taxa between non-atrophic gastritis and GC based on weighted-UniFrac distance measure and reported that there is a significant separation of the microbiota composition between two groups .
In species level, Streptococcus_NCVM and Campylobacter jejuni were differentially abundant in the GC cases while Prevotella melaninogenica was differentially abundant in the controls based on the LEfSe analysis. Although the evidence related to the effects of those bacterial species in GC occurrence is limited, a study has reported that there is a remarkable effect of the metabolic products produced by those Prevotella species such as lactic acid, acetic acid, butane diacid, isovaleric acid, isobutyric acid to the human gastric cell physiology . Furthermore, Prevotella melaninogenica enrichment can create the gastric environment more likely to be acidic by lowering pH than non-atrophic gastritis where there is a restriction to colonize by other harmful bacterial species . A study evaluated the microbiota composition in advanced gastric adenocarcinoma through the shotgun metagenomics approach and they have reported that the cladogram of the gastric microbiome phylogenetically associated with GC and superficial gastritis. Family Porphyromonadaceae, genus Porphyromonas, genus Alloprevotella were enriched in the GC group whereas genus Actinomyces, and genus Atopobium were enriched in the superficial gastritis group. However, when comparing to our study, this study does not have similar taxa which were highly enriched specifically in the GC case group . Based on the cladogram, the Bacilli class was phylogenetically related in GC cases in the current study that is similar to the findings observed by Liu et al. .
Based on the compositional analysis of microbiome data in the genus level, there was a higher MDI in GC cases than controls and it was marginally significant for the total population (p = 0.097) and it was significant for female (p = 0.002). Furthermore, higher MDI showed a significantly increased risk of GC in females (OR: 2.66, 95% CI: 1.19–5.99, p-trend = 0.017; model II). A study conducted to observe the association between gastric dysbiosis of the gastric microbiome and GC risk concluded that there is a higher MDI in GC patients than those who had chronic gastritis (p < 0.0001) . A study carried out to characterize the changes of the microbiome associated with histopathological stages of gastric tumorigenesis observed that there is a significant microbial dysbiosis of gastric mucosa in GC patients with a significant overrepresentation of 21 and reduction of 10 bacterial taxa in GC in comparison to superficial gastritis (q < 0.05) .
In one hand for deriving MDI based on the fold changes of selected genera, Lactobacillus, Diaphorobacter, Acinetobacter, Atopobium, Actinobacillus, and Rhizobium genera are top six genera out of 13 genera enriched in GC cases in the current study population. Particularly, Lactobacillus has a critical role in carcinogenesis because of N-nitroso compounds derived from the metabolism of nitrate/nitrite . In fact, several previous microbiome studies have reported that there is an increment in the abundance of Lactobacillus in GC patients [41–43]. MDI was significantly positively associated with the risk of GC specifically in females in the current study. As a plausible biological mechanism, it has been reported that the gut microbiome is one of the principal regulators of circulating estrogen in females . The gut microbiota secretes β-glucuronidase which is an enzyme that deconjugates estrogens into their active forms where there is a direct regulation of estrogens by gut microbiota. Once the dysbiosis of the microbiota is taken place that is characterized by lower microbial diversity, it can impair the above-mentioned deconjugation process where there is a reduction of the circulating estrogens. The alterations of circulating estrogens may affect to develop several pathological conditions particularly GC in females . Three possible mechanisms have been proposed for the carcinogenesis due to the microbial dysbiosis .
The first mechanism is related to the bacterial-induced chronic inflammation. The inflammatory mediators produced due to chronic inflammation have harmful effects on epithelial, endothelial cells, and extracellular matrix compounds. During this inflammatory process, epithelial and immune cells trigger ROS and reactive nitrogen species (RNS) due to the direct influence of TNF-α, IL-6, and TGF-β . Production of ROS and RNS occurs via induction of NADPH oxidase and nitric oxide synthase. NADPH oxidase catalyzes the superoxide anion leading to superoxide dismutase mediated hydrogen peroxide H2O2 production. Simultaneously, nitric oxide synthase generates nitric oxide (NO), which can be converted into nitrogen dioxide (NO2), peroxinitrite, and dinitrogen trioxide (N2O3) to produce their ROS and RNS. Interestingly, increased expression of NADPH oxidase, nitric oxide synthase, and their ROS and RNS species have been identified in tumor microenvironment . Additionally, cell proliferation, mutagenesis, oncogene activation, and angiogenesis can be facilitated by the inflammatory mediators produced by the above mechanism.
In the second mechanism, NF-κB can be activated and cellular apoptosis can be inhibited. Activation of NF-κB pathway that is related to oncogenic cell signaling in epithelial cells has been identified as a critical pathway for the TNF- α induced tumor growth. NF-κB signaling can be categorized into a “classical” pathway and “alternative” pathway. In classical pathway IκB kinase β (IKKβ) phosphorylates IκBα whereas in alternative pathway IKKα phosphorylates the p100 precursor of the NF-κB p52 subunit. There is an accumulation of the heterodimeric NF-κB transcription factors in the nucleus as a result of the above signaling events. Classical pathway regulates mainly p50/p65/ and p50/c-Rel dimers and the alternative pathway regulates the p52/relB dimers. Other kinases including the unconventional IKK family members, IKKε, and TBK1 can also activate the NF-κB pathway. Several signaling pathways converge on the NF-κB regulators provides significant evidence where cancers can aberrantly stimulate NF-κB . Further, it has an effect to activate pro-inflammatory cytokines, oncogenes, and to induce cancer cell proliferation. In the third mechanism, bacterial substances can act as carcinogenic substances that induce the carcinogenesis process . The integration of those mechanisms can potentially stimulate the carcinogenesis process with the involvement of microbial dysbiosis. Particularly, local concentration of various cytokines including interleukin-1β (IL-1β), IL-6, tumor necrosis factor-α (TNF-α) can be increased due to the microbial dysbiosis. Endothelial cells can be activated by IL-1β to produce vascular endothelial growth factor (VEGF). VEGF eventually generates an inflammatory microenvironment which is helpful for angiogenesis and tumor progression . Also, TNF-α has an ability to produce ROS that can induce DNA damages. A study focused on the relationship between tumor-immune environment associated with GC microbiota in patients who have GC identified that there is a correlation between regulatory T cells and plasmacytoid dendritic cells in the tumor microenvironment that is further associated with the dysbiosis of the gastric microbiota .
Regarding the metagenomics functional pathway results, biosynthesis of ansamycins pathway was highly enriched in controls. It has been reported that ansamycins is a groups of antibiotic produced by strains of several Actinomycetes. Ansamycins have proved to be very potent molecules displaying anticancer, antibacterial, and antiviral activities . The one carbon pool by folate pathway was highly enriched in controls than GC cases. One carbon metabolism mediated by folate cofactor supports the multiple physiological processes including biosynthesis of purines and thymidine, amino acid homeostasis (glycine, serine, and methionine), epigenetic maintenance, and redox defense. While most gut bacteria can synthesize the folate, humans require the dietary folate intake to maintain the physiological processes . It has an essential role in the nucleic acid synthesis. It has been noted that the adequate folate level over the long term may support genome integrity. Based on the evidence of experimental and epidemiological studies, there is a protective effect of the folate towards the colorectal cancer, breast cancer, and pancreatic cancer while there are inconsistent results for the gastric cancer [54–56]. Secondary bile acid biosynthesis pathway was also highly enriched in GC cases than controls. Secondary bile acids can induce reactive oxygen species production, genomic destabilization, apoptosis resistance, and cancer stem cells-like formation. There are diverse signals involved in the carcinogenesis mechanism of bile acids, with a major role of epidermal growth factor receptor, and its down-stream signaling, involving mitogen-activated protein kinase, phosphoinositide 3-kinase/Akt, and nuclear factor kappa-light-chain-enhancer of activated B cells. Bile acids regulate numerous genes including the human leukocyte antigen class I gene, p53, matrix metalloprotease, urokinase plasminogen activator receptor, Cyclin D1, cyclooxygenase-2, interleukin-8, and miRNAs of cancer cells .
There are several strengths of our study. To the best of our knowledge, this is the first study that is associated with the methodological approach that employed the compositional analysis of microbiome data using a novel statistical approach termed “CCREPE” to derive a MDI for a Korean population. The main strength of this approach is it abrogates the spurious correlations when determining the significance of a similarity measure. Second, the sample size of current microbiome-related study is comparatively large with 268 GC cases and 288 healthy subjects relative to previous microbiome studies and it improves the power of statistical analysis to observe the relevant associations between the microbiome and the risk of GC. Third, several potential covariates were considered in a multivariate analysis that are established risk factors for GC development. Those confounding variables are age, smoking, family history of GC, regular exercise, education, occupation, income, and total energy intake throughout the analysis.
However, there are potential limitations associated with the current study. Generally, selection bias and recall bias need to be raised since this study is a hospital-based case-control study. Selection bias might have occurred because healthy subjects were selected from the participants who attended the health screening. They may have a healthier lifestyle due to health concerns compared to those who do not participate in screening. Therefore, healthy subjects might be less representative of the general population. Second, current study is not a follow-up study. Thus, the associations between the gastric microbiome and GC risk can occur without having a causal relation because patients with early GC have changed their microbial profile because of premalignant lesions that already have been progressed and due to their changes in the dietary habits. However, cases included only patients diagnosed with early GC in this study. Thus, the influence of the dietary changes on GC symptoms will be slight. Third, MDI has not been validated although it has been applied in an epidemiological case-control study. Furthermore, since a single sample was measured for the microbial measurements in the current study, the results related to microbial exposure may have less accuracy compared with the microbiome measurements in multiple time points . However, it is important to note that repeating biopsies with those who have normal gastric histology has ethical issues.