Skip to main content

Microbiome signatures associated with clinical stages of gastric Cancer: whole metagenome shotgun sequencing study

Abstract

Background

Gastric cancer is one of the global health concerns. A series of studies on the stomach have confirmed the role of the microbiome in shaping gastrointestinal diseases. Delineation of microbiome signatures to distinguish chronic gastritis from gastric cancer will provide a non-invasive preventative and treatment strategy. In this study, we performed whole metagenome shotgun sequencing of fecal samples to enhance the detection of rare bacterial species and increase genome sequence coverage. Additionally, we employed multiple bioinformatics approaches to investigate the potential targets of the microbiome as an indicator of differentiating gastric cancer from chronic gastritis.

Results

A total of 65 patients were enrolled, comprising 33 individuals with chronic gastritis and 32 with gastric cancer. Within each group, the chronic gastritis group was sub-grouped into intestinal metaplasia (n = 15) and non-intestinal metaplasia (n = 18); the gastric cancer group, early stage (stages 1 and 2, n = 13) and late stage (stages 3 and 4, n = 19) cancer. No significant differences in alpha and beta diversities were detected among the patient groups. However, in a two-group univariate comparison, higher Fusobacteria abundance was identified in phylum; Fusobacteria presented higher abundance in gastric cancer (LDA scored 4.27, q = 0.041 in LEfSe). Age and sex-adjusted MaAsLin and Random Forest variable of importance (VIMP) analysis in species provided meaningful features; Bacteria_caccae was the most contributing species toward gastric cancer and late-stage cancer (beta:2.43, se:0.891, p:0.008, VIMP score:2.543). In contrast, Bifidobacterium_longum significantly contributed to chronic gastritis (beta:-1.8, se:0.699, p:0.009, VIMP score:1.988). Age, sex, and BMI-adjusted MasAsLin on metabolic pathway analysis showed that GLCMANNANAUT-PWY degradation was higher in gastric cancer and one of the contributing species was Fusobacterium_varium.

Conclusion

Microbiomes belonging to the pathogenic phylum Fusobacteria and species Bacteroides_caccae and Streptococcus_anginosus can be significant targets for monitoring the progression of gastric cancer. Whereas Bifidobacterium_longum and Lachnospiraceae_bacterium_5_1_63FAA might be protection biomarkers against gastric cancer.

Peer Review reports

Background

Gastric cancer is one of the major health problems worldwide, ranking fifth in incidence and third in cancer-related mortality, as reported in the latest published global cancer statistics [1]. Long-term studies have confirmed that the development of gastritis with precancer lesions such as atrophic gastritis or intestinal metaplasia increases the risk of gastric cancer [2,3,4].

Gastritis diagnosis in clinical practice relies primarily on invasive endoscopy and histological examination [5], which cannot be performed frequently and easily. Hence, monitoring disease progression with non-invasive methods and detection of biomarkers are in high demand for prevention and treatment strategies for gastric diseases.

A series of studies have affirmed the role of microbiomes other than Helicobacter pylori, a well-known carcinogen [6], in gastric lesions [7,8,9,10]. In the gastrointestinal tract, trillions of microorganisms colonize the mucosal surface and lumen, constantly releasing immunomodulatory molecules that interact with and shape the immune system [11]. Analyses of alterations in gastric mucosa microbial changes at different stages of gastritis including superficial gastritis, atrophic gastritis, intestinal metaplasia, and gastric cancer found that shifts in gastric microbial composition are associated with progression toward a more advanced form of gastric disease [10, 12, 13].

Next-generation sequencing of fecal samples produced tens of millions of reads per sample, allowing for comprehensive analysis of both rare and abundant microbes with high genome sequencing coverage [14]. Furthermore, at this depth of sequencing, de novo prediction of genes is also possible [15].

Notably, the majority of microbiome studies on gastric diseases are limited to gastric mucosal samples utilizing 16S rRNA sequencing. In contrast, we performed whole metagenome sequencing to enhance the detection of bacterial species, examining both rare and abundant species, and employed multiple bioinformatics approaches to investigate the potential targets of the microbiome that could serve as indicators for distinguishing between chronic gastritis and gastric cancer.

Materials and methods

Study setting and sample

The study participants were recruited from the Kaohsiung Medical University Chung-Ho Memorial Hospital, as well as from multisite Taipei Medical University hospitals, which include Taipei Medical University Hospital, Wanfang Hospital, and Shuang-Ho Hospital. The participants were grouped into chronic gastritis (CG, n = 33) and gastric cancer (GCA, n = 32). Within these groups, further sub-groups were categorized: chronic gastritis without precancer lesions (non-intestinal metaplasia, NIM, n = 18), chronic gastritis with pre-cancer lesions (intestinal metaplasia, IM, n = 15), early-stage gastric cancer (Phase I and II, n = 13) and late-stage gastric cancer (Phase III and IV, n = 19). The cancer stage was clinically determined according to American Joint Committee on Cancer staging manual 8th Edition. The diagnosis of chronic gastritis with NIM, chronic gastritis with IM, and gastric cancer were confirmed through histological examination of endoscopic mucosal biopsies conducted by pathologists.

Exclusion criteria were participants with; any significant infectious disease requiring intensive antibiotic treatments within 6 months before fecal sample collection, a history of alcohol/substance dependence, any disease that needed immunosuppressant therapy, inflammatory bowel disease, indeterminate colitis, irritable bowel syndrome, colitis, persistent or chronic diarrhea of unknown etiology, and recurrent Clostridium difficile infection.

This study was approved by the Institutional Review Boards of Kaohsiung Medical University (IRB No. KMUHIRB-G(I)-20,200,024), Taipei Medical University (IRB No. N202108054), and Hebrew SeniorLife (IRB No. 2019–50) in Boston, MA, USA. All participants provided informed consent to participate in this study.

The study collected de-identified clinical and survey information from participants, only including data relevant to the research objectives. The procedures conducted in this study adhered to ethical standards established by the institutional and/or national research committees, following the principles of the 1964 Helsinki Declaration and its later amendments or equivalent ethical standards.

Fecal sample collection and DNA extraction

The stool samples for the study were collected using OMNIgene-GUT tubes (OM-200, DNA Genotek). Each participant collected approximately 1 g of stool at home following the user instructions provided by the manufacturer. The collected samples were then returned to clinicians at Kaohsiung Medical University and Taipei Medical University hospitals. Since OMNIgene-GUT tubes do not need a cold chain, the collected samples were stored at room temperature for a period of up to 2 months (https://www.dnagenotek.com/ROW/products/collection-microbiome/omnigene-gut/OM-200.htm).

For DNA extraction, subsampled stool specimens of approximately 100 mg were processed using the QIAamp PowerFecal Pro DNA Kit from Qiagen (catalog number 51804) [16]. All lysis, separation of impurities and purification procedures adhered to the manufacturer’s protocol provided by the QIAamp PowerFecal DNA Kit. The QC criteria were applied to ensure the reliability of the extracted DNA. These criteria included a minimal DNA concentration of 30 ng/μl with no serious degradation observed by gel electrophoresis with a DNA fragment length over 1 kb and the total amount was higher than 300 ng.

Whole metagenome shotgun sequencing (WMGS)

Next-generation sequencing library construction

Next-generation sequencing library preparations followed the protocol of the VAHTS Universal DNA Library Prep Kit for Illumina (ND607–01, Vazyme Biotech). For each sample, 200 ng genomic DNA was randomly fragmented to sizes less than 500 base pairs (bp) using a sonication method with an S220 Focused-ultrasonicator from Covaris. The fragments underwent End Prep Enzyme Mix for end repair, 5′ phosphorylation and dA-tailing in a single reaction, followed by T-A ligation to attach adaptors to both ends. The adaptor-ligated DNA was subjected to size selection using beads. and fragments of approximately 470 bp (with the approximate insert size of 350 bp) were recovered. Each recovered DNA was amplified by PCR using P5 and P7 primers, with both primers carrying sequences that can anneal with the flowcell to perform bridge PCR and P5/P7 primer carrying index allowing for multiplexing. The PCR products were purified using beads, validated and quantified by Qubit 3.0 Fluorometer (Cat No Q33216, Invitrogen).

The resulting library preparations were subjected to validation and quantification using a Qubit 3.0 Fluorometer (Cat No Q33216, Invitrogen). With the raw sequencing data, both the amount and the quality of the sequencing data were checked by the software Seqtk (v1.2-r94).

Covariate information

Clinical information used as covariates included age, sex, BMI, comorbidities (hypertension, dyslipidemia, and diabetes), and Helicobacter pylori infection/eradication history. Two different Food Frequent Questionnaires (FFQs) for recent diet information were collected along with fecal samples from all the subjects who participated in the study. The FFQ survey included checklists assessing recent and regular dietary habits about food types (fish, meat, vegetables, dairy products, etc.), food intake frequency, prebiotics, and probiotic use.

Statistical analyses

Baseline characteristics of subjects

Baseline characteristics of subjects were expressed as mean ± standard deviation (SD), and frequency or proportion (percentage) and were compared using the unpaired Student t-test, and Fisher’s exact test or chi-square test, respectively. FFQs were tested by the Mann-Whitney U (Wilcoxon rank-sum) test between CG and GCA. To examine whether food consumption was associated with any differential microbiome features as well as any covariates, we performed Hierarchical All-Against-All Significance Testing (HAllA) [17].

Microbiome taxonomic and functional profiles

We used Whole Metagenome Shotgun (wmgx) workflow (https://github.com/biobakery/biobakery_workflows#whole-metagenome-shotgun-wmgx) in biobakery pipelines [18] to process the paired-end raw metagenome shotgun sequencing FASTQ files. The first step involved filtering low-quality or irrelevant reads from the metagenome shotgun sequencing data. This was done using the KneadData tool (version 0.70). Taxonomic profiles of shotgun metagenomes were generated using MetaPhIAn2 (version 2.7.8). MetaPhIAn2 utilizes a library of clade-specific markers to provide profiling of various taxonomic groups, including bacteria, archaea, eukaryotes, and viruses. Functional profiling was performed by HUMAnN2. HUMAnN2 constructed a sample-specific reference database based on the pangenome of a subset of the species detected by MetaPhlAn2 in the sample. This allowed for the determination of the abundance profiles of gene families (UniRef90s). The information on which species contributed to these genes was stratified by StrainPhlAn and could then be summarized into higher-level gene groupings. Protein-coding sequences in the constructed pangenomes were pre-annotated to their respective UniRef90 families [19]. UniRef90 represents a comprehensive and nonredundant protein sequence database.

Normalization and filtering process

Normalization plays a crucial role in differential abundance analysis, especially when dealing with metagenome sequencing data, as differences in sequencing depth can make read counts incomparable across samples. We used the total sum scaling (TSS) [20] that aimed to address the heteroscedasticity of the samples observed in the samples, thereby stabilizing the variance of the data [21] after removing archaeal and viral taxonomies in the samples. After normalizing the raw measures into relative abundances, we limited our analysis to only microbial features at each taxon level that were prevalent and abundant with mean relative abundance > 0.01% in at least 10% of the samples.

Microbiome community diversity

The Shannon index [22] was used to measure α-diversity. Alpha diversity assesses the diversity of species within a single sample or group of samples. To identify differences in alpha diversity between groups, an independent two-sample t-test was applied. β-diversity was computed using Bray-Curtis dissimilarity and summarized using weighted and unweighted principal coordinates analysis (PCoA) [23]. Statistical differences in beta diversity metrics between groups were tested by permutation multivariate analysis of variance (PERMANOVA).

Differentially abundant microbiome features

Multiple bioinformatics approaches

We employed a comprehensive approach to minimize the likelihood of false positive or false negative findings. For univariable association analysis, Wilcoxon-Sum Rank test, RNASeq (EdgeR) implementing empirical Bayes estimation, exact tests, generalized linear models and quasi-likelihood tests based on the negative binomial distributions [24], Linear discriminant analysis effect size (LEfSe) that can highlight features that are particularly relevant in distinguishing different classes or groups [25] were used. For multivariable analysis, backward stepwise multivariable generalized linear regression analyses were performed adjusting for age and sex utilizing MaAsLin (Multivariate Association with Linear Models) [26]. Additionally, Random Forest variable of importance (VIMP) was computed for feature selection using the randomforestSRC R package [27]. HAllA method was implemented to test correlation among all pairs of FFQ and species abundance. HALLA model tests for correlation among all pairs of variables in a high-dimensional dataset, and prioritizes statistically promising candidate variables. HAllA utilizes hierarchical false discovery correction to limit false discoveries and loss of statistical power attributed to multiple hypothesis testing.  All the comparisons were two-tailed and the False Discovery Rate (FDR) method was used for multiple testing corrections with adjusted p values (q values) in all approaches. All analyses were performed in the biobakery pipeline [18], Microbiomeanalyst platform [28] and R version 4.1.2.

Results

Baseline characteristics and FFQ

A total of 33 CG and 32 GCA fecal samples were collected. Within CG, there were NIM (n = 18) and IM (n = 15). Within GCA, there were early-stage GCA (n = 13) and late-stage GCA (n = 19). The mean age was younger in CG but was not significantly different. Sex, BMI, hypertension, dyslipidemia, diabetes, and Helicobacter pylori history (eradication history) were comparable between the CG and GCA groups (Supplementary Table 1).

Among the 44-food frequency questionnaire (FFQ) items, 5 items were significantly different between the two groups after multiple testing corrections (q < .05) (Supplementary Table 2). Notably, mushroom consumption displayed significant association with Dorea_formicigenerans and Phascolarctobacterium_succinatutens in HALLA clustering (q = 0.007 and 0.010, respectively). Since both showed higher abundance in GCA in biomarker differential analysis in species, we removed these 2 species from the final results to prevent false positives from possible confounding effects of mushroom consumption.

Microbiome community diversity

Microbiome community analysis for alpha (Shannon index) and beta diversity (PCoA) demonstrated no statistically significant difference in phylum, genus, or species between the 2-group comparison of CG and GCA. These results remained consistent when extended to the four-group comparison involving NIM, IM, early-stage GCA, and late-stage GCA (Supplementary Fig. 1 & 2). A whole taxa profile for 65 samples is available in Supplementary Table 3.

Differential abundances

Phylum

In the Wilcoxon-Sum Rank test, Actinobacteria and Fusobacteria exhibited significantly different abundances between the two groups among 7 phyla. Actinobacteria demonstrated a higher abundance in individuals with CG, while Fusobacteria were more abundant in those with GCA (Table 1).

Table 1 Differential abundance by Wilcoxon Rank-Sum Test between chronic gastritis and gastric cancer in Phylum

In LEfSe analysis, for the 2-group comparison, only Fusobacteria was significantly higher in GCA after multiple testing corrections (q = 0.041, LDA score 4.27). In LEfSe analysis, among the 4-group comparison, Fusobacteria presented the highest abundance in the late-stage GCA but was not statistically significant (q = 0.27, LDA score 4.47) (Fig. 1 & Supplementary Table 4). Age and sex adjusted MaAslin model indicated that Fusobacteria was nominally significant (p = 0.029) with beta 0.685 (se 0.33) but multiple testing corrections turned to null (q = 0.309) (Supplementary Table 5).

Fig. 1
figure 1

Differential abundance in phylum between chronic gastritis and gastric cancer. Footnote: (A) Y axis: relative abundance by percent value

Genus

In the genus level analysis, 64 genera were used for final statistical analysis. Fifteen genera were significant in Wilcoxon-Sum Rank test. Veillonella, Sutterellaceae_unclassified, Fusobacterium, Parabacteroides, Phascolarctobacterium, Sutterella,Oscillibacter, Haemophilus and Coprococcuscontributed to GCA whereas Acinetobacter, Enterococcus, Adlercreutzia, CollinsellaPseudomonas and Bifidobacterium to CG (Supplementary Table 6). The same results were found in LEfSe analysis before multiple testing (Fig. 2 (A) & Supplementary Table 7). LEfSe analys in 4-group comparison showed Acinetobacter contributed to NIM, Anaerotruncus and Adlercreutzia to IM and Fusobacterium, Oscillibacter and Parabacteroides to late-stage GCA (Fig. 2 (B) & Supplementary Table 8).

Fig. 2
figure 2

LEfSe analysis between chronic gastritis and gastric cancer in the genus

In EdgeR analysis, Eubacterium, Collinsella, Pseudomonas and Morganella were significantly less but Lactobacillus is more abundant in GCA in the 2-group comparison. In 4-group comparison, Megamonas and Pseudomonas were more abundant in NIM whereas Eubacterium showed higher abundance in IM than the other groups (Table 2).

Table 2 Significant genera by EdgeR analysis

In MaAsLin analysis (age and sex-adjusted), none of them surpassed multiple testing corrections but most of them remained nominally significant (p < .05) (Supplementary Table 9).

Species

After filtering low-abundance species (refer to normalization and filtering process in method section), 156 species remained for further analysis. In the Wilcoxon Rank Sum test, 29 species had significantly different abundances between CG and GCA. Fusobacterium_mortiferum is among one of them (p = 0.02) (Fig. 3). Dorea_formicigenerans and Phascolarctobacterium_succinatutens that showed significant association with mushroom consumption in HALLA analysis also presented significance, we removed these 2 species in a later analysis (Supplementary Table 10). Using the Random Forest algorithm, classification of CG and GCA showed modest out of bag (OOB) error rates of 21.9, and 27.3%, respectively. In 4 group classifications, Random Forest presented high OOB error, NIM (66.7%), IM (60%), early-stage GCA (100%), and late-stage GCA (42.1%) (Supplementary Fig. 3).

Fig. 3
figure 3

Heat Tree by Wilcox-Rank Sum test. Footnote: Red color means higher abundance in chronic gastritis (CG). Green color means higher abundance in gastric cancer (GCA)

In the age and sex adjusted MaAslin analysis in species, Bacteria_caccae was the most significant species among a total of 16 nominally significant species (p < .05) but none of them surpassed the multiple testing (Supplementary Table 11). Additional validation through the Random Forest VIMP (Fig. 4 & Supplement Table 12) analysis, 11 species were finally selected as important features having significant VIMP scores above 10 as well as significant findings from MaAsLin results (Fig. 5 & Table 3). Bifidobacterium_longum, Enterococcus_faecium and Lachnospiraceae_bacterium_5_1_63FAA showed higher abundance in CG compared to GCA whereas Bacteroides_caccae, Bifidobacterium_dentium, Streptococcus_anginosus, Coprococcus_catus, Lactobacillus_fermentum, Parabacteroides_distasonis, Oscillibacter_unclassified and Lactobacillus_mucosae presented higher abundance in GCA (Fig. 5 & Table 3).

Fig. 4
figure 4

Top 30 features by Variable importance by Random Forest. Footnote: Red-colored bars are significant species within 95% CI of VIMP scores

Fig. 5
figure 5

Important species between chronic gastritis and gastric cancer by MaAsLin and VIMP (VIMP score > 2) A Bacteroides_caccae Bifidobacterium_longum C Streptococcus_aginosus D Lactobacillus_fermentum E Parabacteroides_distasonis F Oscillibacter_unclassified

Table 3 Important species differentiating CG and CGA

EdgeR analysis among the 4-group comparison, Eubactrium_rectale was significantly higher in IM (logFC 2.120, SE: 14.127, q = 0.002). We also explored 4 group comparisons in important features found in Table 3. Although they were not statistically significant in EdgeR results, Bacteroides_caccae presented a higher abundance in late-stage GCA, whereas Bifidobacterium_longum showed a higher abundance in IM, and Lactobacillus_mucosae showed a higher abundance in early-stage GCA (Fig. 6).

Fig. 6
figure 6

Distribution of microbiome abundance across 4 groups in important 4 species Bacteroides_caccae Bifidobacterium_longum C Lactobacillus_mucosae Eubacterium_rectale

Metabolic pathway analysis

Age, sex and BMI adjusted MasAsLin analysis indicated that the superpathway of L-lysine, L-threonine and L-methionine biosynthesis I and II, superpathway of pyrimidine ribonucleosides salvage and superpathway of N-acetylglucosamine, N-acetylmannosamine and N-acetylneuraminate degradation were more enriched in GCA. In particular, GLCMANNANAUT-PWY; superpathway of N-acetylglucosamine, N-acetylmannosamine and N-acetylneuraminate degradation was associated with Fusobacterium_varium (Supplementary Table 13).

Discussion

The overall composition and community diversities of the microbiome were similar between the CG and GCA groups irrespective of the specific subgroups within these categories. These results align with previous findings reported in other studies [29, 30].

In particular, we found that an enrichment of microbiota belonging to the phylum Fusobacteria was significantly associated with GCA, which has been confirmed in multiple studies [29, 31,32,33]. The genus Fusobacterium was frequently abundant in patients with gastric cancer, and a receiver operating characteristic curve analysis revealed that species Fusobacterium_nucleatum exhibited a diagnostic ability for gastric cancer [29]. The distribution of genus Fusobacterium in tumor tissues was demonstrated [31]. Fusobacterium_nucleatum, which originates from the oral cavity, can potentiate the carcinogenesis of colorectal cancer involving the activation of Wnt target genes which increase the secretion of proinflammatory cytokines and evade anticancer immune response [32, 33]. Hsieh et al have shown that Fusobacterium_nucleatum colonization leads to a worse prognosis in GCA patients with H. pylori positivity [29]. However, we didn’t detect Fusobacterium_nucleatum from fecal samples in our study, instead, we detected Fusobacterium_mortiferum and Fusobacterium_varium by WMGS. This might be due to the sample difference, most of the previous study findings for Fusobacterium_nucleatum were from samples collected from stomach tissues. Fusobacterium_mortiferum was significantly enriched in the GCA group in univariate analysis and Fusobacterium_varium was significantly associated with the microbial metabolic pathway of GLCMANNANAUT-PWY which was highly enriched in the GCA group in our study. Despite limited reports about Fusobacterium_varium, in a study among the patients with Fusobacterium infections in Korea, patients with Fusobacterium_varium infections were older and had a higher proportion of nosocomial infections than the other groups. The Fusobacterium nucleatum and Fusobacterium_varium groups showed higher in-hospital mortality than the other patients with Fusobacterium species [34]. Fusobacterium_varium as well as species belonging to Fusobacteria might be potential targets to study in the future in microbiome research.

As well-known pathogenic species, Clostridium perfringens, Clostridium perfringens 13, Clostridium perfringens A99 and Escherichia coli K-12 substr. also share GLCMANNANAUT-PWY. These species can generate carbon and nitrogen sources through these N-acetylglucosamine, −mannosamine and -neuraminic acid degradation pathways [35] which might provide sources of N-nitroso compound (NOC) in affected patients.

It is known that patients with GCA have higher NOC levels than healthy subjects [36]. Genera Veillonella and Lactobacillus which were found significantly high in the GCA group in our genus analysis, contributed to gastric carcinogenesis by stimulating the production of NOCs [37, 38]. Veillonella was significantly lower in gastritis subjects than in gastric adenoma or advanced gastric cancer subjects in the previous study [39].

At the species level, we also found a few protective species that were more abundant in CG than GCA. Bifidobacterium longum were higher in CG and according to the previous study, they were more abundant in the cancer patients who responded well to chemotherapy than non-responders [40]. Bifidobacterium_longum strains regulate oxidative stress by regulating the production and accumulation of ROS (reactive oxygen species), thereby reducing the symptoms of Inflammatory Bowel Disease [41]. Lachnospiraceae_bacterium_5_1_63FAA belongs to Lachnospiraceae which has been linked to protection from colon cancer in humans, mainly due to the association with the production of butyric acid, a substance that is important for both microbial and host epithelial cell growth [42].

Whereas a few pathogenic species were also detected. Among them, Bacteroides_caccae, a pathogenic species previously found in cultures from infections in the appendix and the peritoneal abdomen [43] was higher in GCA.

Bacteroides_caccae degraded the mucus [44], which would lead to a condition of a “leaky gut” and therefore increased the permeability of the intestinal barrier. Clinical and experimental data suggested the importance of intestinal hyperpermeability in the inflammatory changes of various diseases including GI cancers [45]. Streptococcus_anginosus showed higher habitation in gastric tumors in the previous study [46]. Oscillibacter has been positively correlated with gut permeability [47].

However, some of the results are controversial from the previous findings. Bifidobacterium_ dentium, appeared to protect mucin glycans which is vital in the gut barrier [48]. Lactobacillus_fermentum UCO-979C is a good probiotic for the protection against H. pylori infections [49]. Enterococcus_faecium is the main causative agent of infection in humans and frequently demonstrates resistance to vancomycin, ampicillin, and other antimicrobials [50]. Coprococcus was less abundant in colon cancer compared to healthy individuals, although there was no evidence for its protective role against colon cancer [51]. Parabacteroides_distasonis attenuates toll-like receptor 4 signaling and Akt activation and blocks colon tumor formation in high-fat diet-fed azoxymethane-treated mice [52]. Lactobacillus mucosae has been reported as cardio-protective [53].

Based on these findings, we assume CG can also present a degree of gastric disease related microbial perturbations which can result in higher abundance of pathogenic microbiota found in CG as well.

In this study, we detected species-level microbiome markers and associated metabolic pathways by using WMGS, which enabled us to find novel species belonging to Fusobacterium known to be associated with poor prognosis of gastric diseases or gastric cancer. We also applied multiple bioinformatics approaches encompassing various patient characteristics and food consumption history which would affect the microbiome composition and delineate potential microbiome signatures that could be utilized as diagnostic or treatment biomarkers.

We also have limitations to address. First, due to the small sample size, the generalization of these results would be limited. However, multiple previous studies generated similar results, especially in the Asian population, which might validate this study’s results. Second, most study findings did not surpass multiple testing correction but we incorporated Random Forest VIMP scores to add further evidence to species found in MaAsLin multivariable models. We assume this study added clinically meaningful new findings to the previous microbiome studies in the gastric disease area. Third, medication use might have affected the gut microbial composition, however, due to the lack of information on medication use, we could not adjust this factor in the multivariable model. Fourth, Helicobacter_pylori is a well-established risk factor for gastric cancer, and it is worthwhile to investigate the association between Helicobacter_pylori and other microbiome features. However, we used fecal samples in this study, and Helicobacter_pylori which usually inhabits the upper gastric region was not detected, which limited additional exploration on Helicobacter pylori. Finally, although we sub-grouped the patients in order to dissect the different microbiome features corresponding to different stages of gastric diseases, the cross-sectional approach itself has limitations portraying only a snapshot of the time we collected the fecal samples. Therefore, designing longitudinal time-varying fecal sampling approaches within individuals is warranted.

Conclusion

CG and GCA share similar microbial community characteristics. However, several distinctive microbiome pathogenic features including Fusobacteria, Bacteroides_caccae, and Streptococcus_anginosus might be represented as signature indicators for the progression of CGA. In addition, Bifidobacterium_longum, and Lachnospiraceae_bacterium_5_1_63FAA might be protective biomarkers against advanced gastric diseases.

Availability of data and materials

Metagenomic raw sequencing data have been deposited in SRA database with BioProject ID PRJNA1065874. Raw data can be accessed here: https://www.ncbi.nlm.nih.gov/bioproject/PRJNA1065874

Abbreviations

CG:

Chronic gastritis

FFQ:

Food Frequency Questionnaire

GCA:

Gastric Cancer

IM:

Internal Metaplasia

NIM:

No Internal Metaplasia

WMGS:

Whole metagenome shotgun sequencing

References

  1. Bray F, Ferlay J, Soerjomataram I, Siegel RL, Torre LA, Jemal A. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2018;68(6):394–424.

    Article  PubMed  Google Scholar 

  2. Pimentel-Nunes P, Libânio D, Marcos-Pinto R, Areia M, Leja M, Esposito G, et al. Management of epithelial precancerous conditions and lesions in the stomach (MAPS II): European Society of Gastrointestinal Endoscopy (ESGE), European Helicobacter and microbiota study group (EHMSG), European Society of Pathology (ESP), and Sociedade Portuguesa de Endoscopia Digestiva (SPED) guideline update 2019. Endoscopy. 2019;51(4):365–88.

    Article  PubMed  Google Scholar 

  3. Song H, Ekheden I, Zheng Z, Ericsson J, Nyrén O, Ye W. Incidence of gastric cancer among patients with gastric precancerous lesions: observational cohort study in a low risk Western population. Bmj. 2015:27(351).

  4. Meining A, Riedl B, Stolte M. Features of gastritis predisposing to gastric adenoma and early gastric cancer. J Clin Pathol. 2002;55(10):770–3.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  5. Rugge M, Meggio A, Pennelli G, Piscioli F, Giacomelli L, De Pretis G, et al. Gastritis staging in clinical practice: the OLGA staging system. Gut. 2007;56(5):631–6.

    Article  PubMed  Google Scholar 

  6. Ishaq S, Nunn L. Helicobacter pylori and gastric cancer: a state of the art review. Gastroenterol hepatol from bed to bench. 2015;8(Suppl 1):S6–S14.

    Google Scholar 

  7. Sohn S-H, Kim N, Jo HJ, Kim J, Park JH, Nam RH, et al. Analysis of gastric body microbiota by pyrosequencing: possible role of Bacteria other than Helicobacter pylori in the gastric carcinogenesis. J cancer prevent. 2017;22(2):115–25.

    Article  Google Scholar 

  8. Ferreira RM, Pereira-Marques J, Pinto-Ribeiro I, Costa JL, Carneiro F, Machado JC, et al. Gastric microbial community profiling reveals a dysbiotic cancer-associated microbiota. Gut. 2018;67(2):226–36.

    Article  CAS  PubMed  Google Scholar 

  9. Schulz C, Schütte K, Koch N, Vilchez-Vargas R, Wos-Oxley M, Oxley A, et al. The active bacterial assemblages of the upper GI tract in individuals with and without Helicobacter infection. Gut. 2018;67(2):216–25.

    Article  CAS  PubMed  Google Scholar 

  10. Aviles-Jimenez F, Vazquez-Jimenez F, Medrano-Guzman R, Mantilla A, Torres J. Stomach microbiota composition varies between patients with non-atrophic gastritis and patients with intestinal type of gastric cancer. Sci Rep. 2014;4(4202)

  11. Belkaid Y, Hand T. Role of the microbiota in immunity and inflammation. Cell. 2014;157(1):121–41.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  12. Eun C, Kim B, Han D, Kim S, Kim K, Choi B, et al. Differences in gastric mucosal microbiota profiling in patients with chronic gastritis, intestinal metaplasia, and gastric cancer using pyrosequencing methods. Helicobacter. 2014;19(6):407–16.

    Article  CAS  PubMed  Google Scholar 

  13. Coker O, Dai Z, Nie Y, Zhao G, Cao L, Nakatsu G, et al. Mucosal microbiome dysbiosis in gastric carcinogenesis. Gut. 2018;67(6):1024–32.

    Article  CAS  PubMed  Google Scholar 

  14. Sims D, Sudbery I, Ilott N, Heger A, Ponting C. Sequencing depth and coverage: key considerations in genomic analyses. Nat Rev Genet. 2014;15(2):121–32.

    Article  CAS  PubMed  Google Scholar 

  15. Human Microbiome Project Consortium. A framework for human microbiome research. Nature. 2012;486(7402):215–21.

    Article  Google Scholar 

  16. QIAGEN. QIAamp® PowerFecal® Pro DNA Kit Handbook: For the isolation of microbial genomic DNA from stool and gut samples. 2019. https://www.qiagen.com/us/resources/resourcedetail?id=8896817a-253f-4952-b845-0aab796813ce&lang=en.

  17. Ghazi A, Sucipto K, Rahnavard G, Franzosa E, McIver LJ, Lloyd-Price J, et al. High-sensitivity pattern discovery in large multi-omic datasets. ISMB Proceedings. 2022. https://huttenhower.sph.harvard.edu/halla.

  18. McIver L, Abu-Ali G, Franzosa E, Schwager R, Morgan X, Waldron L, et al. bioBakery: a meta'omic analysis environment. Bioinformatics. 2018;34(7):1235–7.

    Article  CAS  PubMed  Google Scholar 

  19. Suzek B, Wang Y, Huang H, McGarvey P, Wu C. UniRef clusters: a comprehensive and scalable alternative for improving sequence similarity searches. Bioinformatics. 2015;31(6):926–32.

    Article  CAS  PubMed  Google Scholar 

  20. Paulson JN, Stine OC, Bravo HC, Pop M. Differential abundance analysis for microbial marker-gene surveys. Nat Methods. 2013;10(12):1200–2.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  21. Song SJ, Amir A, Metcalf JL, Amato KR, Xu ZZ, Humphrey G, et al. Preservation methods differ in fecal microbiome stability. Affect Suitab Field Stud mSystems. 2016;1(3):e00021–16.

    Google Scholar 

  22. Simpson E. Measurement of diversity. Nature. 1949;163:688.

    Article  Google Scholar 

  23. Goodrich JK, Di Rienzi SC, Poole AC, Koren O, Walters WA, Caporaso JG, et al. Conducting a microbiome study. Cell. 2014;158(2):250–62.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  24. Robinson M, McCarthy D, Smyth G. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics. 2010;26(1):139–40.

    Article  CAS  PubMed  Google Scholar 

  25. Segata N, Izard J, Waldron L, Gevers D, Miropolsky L, Garrett W, et al. Metagenomic biomarker discovery and explanation. Genome Biol. 2011;12(6):2011–2.

    Article  Google Scholar 

  26. Morgan X, Tickle T, Sokol H, Gevers D, Devaney K, Ward D, et al. Dysfunction of the intestinal microbiome in inflammatory bowel disease and treatment. Genome Biol. 2012;13(9):2012–3.

    Article  Google Scholar 

  27. randomForestSRC. Fast Unified Random Forests with randomForestSRC. version 3.2.2. https://www.randomforestsrc.org/index.html.

  28. Dhariwal A, Chong J, Habib S, King I, Agellon L, Xia J. MicrobiomeAnalyst: a web-based tool for comprehensive statistical, visual and meta-analysis of microbiome data. Nucleic Acids Res. 2017;45(W1):W180–W8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  29. Hsieh Y-Y, Tung S-Y, Pan H-Y, Yen C-W, Xu H-W, Lin Y-J, et al. Increased abundance of Clostridium and Fusobacterium in gastric microbiota of patients with gastric Cancer in Taiwan. Sci Rep. 2018;8(1):158.

    Article  PubMed  PubMed Central  Google Scholar 

  30. Deng Y, Ding X, Song Q, Zhao G, Han L, Ding B, et al. Alterations in mucosa-associated microbiota in the stomach of patients with gastric cancer. Cell Oncol. 2021;44(3):701–14.

    Article  CAS  Google Scholar 

  31. Liu D, Zhang R, Chen S, Sun B, Zhang K. Analysis of gastric microbiome reveals three distinctive microbial communities associated with the occurrence of gastric cancer. BMC Microbiol. 2022;22(1):184.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  32. Brennan CA, Garrett WS. Fusobacterium nucleatum - symbiont, opportunist and oncobacterium. Nat Rev Microbiol. 2019;17(3):156–66.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  33. Proença MA, Biselli JM, Succi M, Severino FE, Berardinelli GN, Caetano A, et al. Relationship betweenFusobacterium nucleatum, inflammatory mediators and microRNAs in colorectal carcinogenesis. World J Gastroenterol. 2018;24(47):5351–65.

    Article  PubMed  PubMed Central  Google Scholar 

  34. Lee S, Baek Y, Kim J, Lee K, Lee E, Yeom J, et al. Increasing Fusobacterium infections with Fusobacterium varium, an emerging pathogen. PLoS One. 2022;17(4)

  35. Caspi R, Billington R, Fulcher CA, Keseler IM, Kothari A, Krummenacker M, et al. The MetaCyc database of metabolic pathways and enzymes. Nucleic Acids Res. 2018;46(D1):D633–D9.

    Article  CAS  PubMed  Google Scholar 

  36. Xu L, Qu Y, Chu X, Wang R, Nelson H, Gao Y, et al. Urinary levels of N-nitroso compounds in relation to risk of gastric cancer: findings from the shanghai cohort study. PLoS One. 2015;10(2)

  37. Zhang S, Shi D, Li M, Li Y, Wang X, Li W. The relationship between gastric microbiota and gastric disease. Scand J Gastroenterol. 2019;54(4):391–6.

    Article  PubMed  Google Scholar 

  38. Wang L, Zhou J, Xin Y, Geng C, Tian Z, Yu X, et al. Bacterial overgrowth and diversification of microbiota in gastric cancer. Eur J Gastroenterol Hepatol. 2016;28(3):261–6.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  39. Park JY, Seo H, Kang C-S, Shin T-S, Kim JW, Park J-M, et al. Dysbiotic change in gastric microbiome and its functional implication in gastric carcinogenesis. Sci Rep. 2022;12(1):4285.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  40. Balar A, Weber J. PD-1 and PD-L1 antibodies in cancer: current status and future directions. Cancer Immunol Immunother. 2017;66(5):551–64.

    Article  CAS  PubMed  Google Scholar 

  41. Yao S, Zhao Z, Wang W, Liu X. Bifidobacterium Longum: protection against inflammatory bowel disease. J Immunol Res. 2021;23(8030297)

  42. Meehan C, Beiko R. A phylogenomic view of ecological specialization in the Lachnospiraceae, a family of digestive tract-associated bacteria. Genome Biol Evol. 2014;6(3):703–13.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  43. Wexler H. Bacteroides: the good, the bad, and the nitty-gritty. Clin Microbiol Rev. 2007;20(4):593–621.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  44. DeGruttola A, Low D, Mizoguchi A, Mizoguchi E. Current understanding of Dysbiosis in disease in human and animal models. Inflamm Bowel Dis. 2016;22(5):1137–50.

    Article  PubMed  Google Scholar 

  45. Fukui H. Increased intestinal permeability and decreased barrier function: does it really influence the risk of inflammation? Inflamm Intest Dis. 2016;1(3):135–45.

    Article  PubMed  PubMed Central  Google Scholar 

  46. Liu X, Shao L, Liu X, Ji F, Mei Y, Cheng Y, et al. Alterations of gastric mucosal microbiota across different stomach microhabitats in a cohort of 276 patients with gastric cancer. EBioMed. 2019;40:336–48.

    Article  Google Scholar 

  47. Lam Y, Mitchell A, Holmes A, Denyer G, Gummesson A, Caterson I, et al. Role of the gut in visceral fat inflammation and metabolic disorders. Obesity. 2011;19(11):2113–20.

    Article  CAS  PubMed  Google Scholar 

  48. Engevik M, Luk B, Chang-Graham A, Hall A, Herrmann B, Ruan W, et al. Bifidobacterium dentium Fortifies the Intestinal Mucus Layer via Autophagy and Calcium Signaling Pathways. LID - https://doi.org/10.1128/mBio.01087-19 [doi] LID - e01087–19. mBio. 2019;10(3):01087–19.

  49. García A, Navarro K, Sanhueza E, Pineda S, Pastene E, Quezada M, et al. Characterization of Lactobacillus fermentum UCO-979C, a probiotic strain with a potent anti-Helicobacter pylori activity. Electron J Biotechnol. 2017;25:75–83.

    Article  Google Scholar 

  50. Murray BE. Vancomycin-resistant Enterococcal infections. N Engl J Med. 2000;342(10):710–21.

    Article  CAS  PubMed  Google Scholar 

  51. Ai D, Pan H, Li X, Gao Y, Liu G, Xia L. Identifying gut microbiota associated with colorectal Cancer using a zero-inflated lognormal model. Front Microbiol. 2019;10(826)

  52. Koh GY, Kane AV, Wu X, Crott JW. Parabacteroides distasonis attenuates tumorigenesis, modulates inflammatory markers and promotes intestinal barrier integrity in azoxymethane-treated a/J mice. Carcinogenesis. 2020;41(7):909–17.

    Article  CAS  PubMed  Google Scholar 

  53. Ryan PM, Stolte EH, London LEE, Wells JM, Long SL, Joyce SA, et al. Lactobacillus mucosae DPC 6426 as a bile-modifying and immunomodulatory microbe. BMC Microbiol. 2019;19(1):33.

    Article  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgments

We would like to thank Dr. Wei-Kuang Chi and Dr. Chi-feng Chang from Development Center for Biotechnology (DCB) for facilitating the initial meetings with the Department of Industrial Technology (DoIT) for approval to conduct this research. In addition, we would also like to thank Yin-Hsien Ho from Yourgene Health (Taiwan) Co., Ltd. for overseeing and coordinating Yourgene’s laboratory experiment progress.

Funding

The project was supported by grants from the Development Center for Biotechnology (grant number: 110VE013); grants from Kaohsiung Medical University Hospital (KMUH109-0R03), Kaohsiung Medical University Research Center Grant (KMU-TC112A02); and self-financing resources from Taipei Medical University Hospital and Yourgene Health (Taiwan) Co., Ltd.

Author information

Authors and Affiliations

Authors

Contributions

SJ: Study design, Data analysis, Visualization, Interpretation, Manuscript draft and revision. YL: Funding acquisition, Study design, Interpretation, Project coordination, Manuscript draft and revision. YHH**: Funding acquisition, Study conceptualization, Project Director, Manuscript revision. DCW**: Funding acquisition, Supervision of clinical study at Kaohsiung Medical University Hospital site for this research project, Clinical consultation, Manuscript revision. CCC**: Funding acquisition, Supervision of clinical study at Taipei Medical University Hospital site for this research project, Clinical consultation, Manuscript revision. MHT: Management of clinical data and collection of clinical samples. YKW: Management of clinical data and collection of clinical samples. ICW: Administration of clinical project. CJL: Management of clinical data and samples. MSW: management of clinical data and collection of clinical samples. TSC: management of clinical data and collection of clinical samples. MYC: management of clinical data and collection of clinical samples. PJH: management of clinical data and collection of clinical samples. WYK: management of clinical data and collection of clinical samples. MJT: Data analysis. HCL: Execution of wet lab experiments. CYL: Data analysis.

Corresponding authors

Correspondence to Chun-Chao Chang, Deng-Chyang Wu or Yi-Hsiang Hsu.

Ethics declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Boards of Kaohsiung Medical University (IRB No. KMUHIRB-G(I)-20200024), Taipei Medical University (IRB No. N202108054), and Hebrew SeniorLife (IRB No. 2019–50) in Boston, MA, USA. All participants provided informed consent to participate in this study. The study collected deidentified clinical and survey information from participants, only including data relevant to the research objectives. The procedures conducted in this study adhered to ethical standards established by the institutional and/or national research committees, following the principles of the 1964 Helsinki Declaration and its later amendments or equivalent ethical standards.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Jeong, S., Liao, YT., Tsai, MH. et al. Microbiome signatures associated with clinical stages of gastric Cancer: whole metagenome shotgun sequencing study. BMC Microbiol 24, 139 (2024). https://doi.org/10.1186/s12866-024-03219-2

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12866-024-03219-2

Keywords