Pathogenic bacteria enriched in tumour tissues

Metagenomic studies have provided abundant clues to the role of gastric microbiota in GC. Several genera of bacteria, including Streptococcus and Fusobacterium, have been consistently identified as enriched bacteria in GC tumour tissues (figure 1). It could be an important hint to distinguish pathogenic microbes that promote gastric carcinogenesis.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Non-Helicobacter pylori microbial dysbiosis in gastric cancer. Oral, gastric mucosal, gastric juice and intestinal microbiota are associated with gastric carcinogenesis. The taxa, mainly genera, that are consistently altered across different studies are summarised. Clues for potentially pathogenic or beneficial microbes can be derived from the enriched or depleted ones, respectively, in the metagenomic sequencing data, while their causality remains to be assessed. Streptococcus anginosus and Epstein-Barr virus are currently the most well-studied pathogenic microbes for gastric tumourigenesis and are marked with asterisks.

Streptococcus is commonly present in the oral cavity and gastrointestinal tract. In patients with GC, Streptococcus enrichment in gastric premalignant lesions or tumour tissues is consistently reported by multiple studies.19 23 30–32 At species level, the abundance of S. anginosus was found to be enriched in the gastric mucosa of patients with GC compared with superficial gastritis in a Chinese cohort involving 205 gastric biopsy tissues.7 Another study of 276 patients with GC also indicated the enrichment of S. anginosus in gastric tumorous microhabitat.20 In addition to tumour tissues, the abundance of S. anginosus and S. constellatus is also increased in faecal samples of patients with intraepithelial neoplasia and early or advanced GC.33 Consistently, the genus Streptococcus and the species S. anginosus are enriched through three meta-analyses.27 28 34

Fusobacterium is another microbial constituent of the oral cavity, while its enrichment has been associated with GC regardless of H. pylori infection.21 Multiple Asian studies have confirmed the increased abundance of gastric Fusobacterium in patients with GC.24 35–37 Compared with precancerous stages, the abundance of Fusobacterium in GC is also increased.7 Fusobacterium is mainly enriched in distal GC,31 while its abundance is also higher in GC tissues compared with matched non-tumour tissues.30 32 In line with studies with single cohort, the enrichment of Fusobacterium in GC tissues could be confirmed by several meta-analyses.27 28 34 At species level, F. canifelinum and F. nucleatum are highly abundant in patients with GC.25

Numerous studies and meta-analyses have identified several more gastric microbes that are enriched in GC. These include Prevotella, Veillonella, Parvimonas and Peptostreptococcus, as reported by a meta-analysis involving 825 gastric tissue biopsies.27 Another meta-analysis with 1270 gastric biopsies revealed the enrichment of Actinobacillus, Actinomyces, Parvimonas and Rothia in patients with GC.28 Meanwhile, a meta-analysis with by far the largest cohort of 2198 individuals demonstrated that Fusobacterium, Leptotrichia and species S. anginosus are enriched in gastric tissues of patients with GC.34 In general, results from different studies are highly heterogeneous, indicating the large variations of gastric microbiota among and across distinct populations. The observed heterogeneity could be attributed to the diversity in sample size, population characteristics (eg, age, ethnicity, dietary habits), gastric pathology and sequencing techniques among different studies, while the integration of different datasets using standardised methodology can reduce heterogeneity.38

Bacteria enriched in oral, gastric juice and faecal samples are associated with GC

In addition to gastric tissues, the microbial community in saliva and gastric juice is also altered in GC (figure 1). It was reported that individuals with periodontal diseases are correlated with increased risk of GC,39 suggesting the potential role of proinflammatory oral microbes in gastric carcinogenesis. Indeed, Streptococcus and Corynebacterium are enriched in salivary samples of patients with GC.40 P. stomatis, Johnsonella ignava, Neisseria elongate and N. flavescens, which are oral opportunistic pathobionts related to periodontal diseases, are also enriched in patients with intestinal metaplasia.41 Moreover, in oral swab samples, Leptotrichia and Gemella are enriched in patients with GC.42 Of note, similar to the gastric microbiota, inconsistent altered microbial taxa have been reported across different studies with large variations in oral microbial community.

The association of oral microbiota with GC could be attributed by the constant flow of oral microbes into the stomach, or in reverse, the impaired acid secretion in GC leads to higher oral bacterial residence. Hence, microbial communities in saliva, gastric juice and mucosa are tightly linked. Veillonella was found to be enriched in gastric juice in a Korean cohort,35 while Sarcina and Brevundimonas were enriched in gastric fluid of GC in a Chinese cohort.37 Lipopolysaccharide-producing bacteria, including Neisseria, Prevotella and Veillonella, are increased in the gastric juice of patients with bile reflux gastritis or GC.43 These findings therefore imply the importance of maintaining oral hygiene to reduce the potential flow of pathobionts from the oral cavity to the stomach.

Although microbial communities in the lower gastrointestinal tract might have indirect and less impacts on gastric carcinogenesis, their associations cannot be ignored (figure 1). Several taxa that are enriched in gastric mucosa, including Streptococcus, Fusobacterium, Prevotella and Veillonella, are also increased in faecal samples of patients with GC.29 33 42 44 45 At species level, S. anginosus and S. constellatus are prominently increased in faecal abundance.33

Causal associations of pathogenic bacteria and GC

It is frequently questioned whether microbial dysbiosis is the cause or consequence of gastric carcinogenesis. In hypergastrinemic insulin-gastrin (INS-GAS) mice with H. pylori infection, microbiota-depleted germ-free mice develop gastric lesions slower than specific pathogen-free mice.46 Meanwhile, transplantation of the altered microbiota promotes the development of gastric premalignant lesions in the recipient germ-free INS-GAS mice.47 Similarly, transplanting gastric microbiota from patients with intestinal metaplasia or GC into germ-free mice induces premalignant lesions in mouse stomach.48 These findings therefore indicate that microbial dysbiosis could contribute to the development of GC.

Mechanistic links between pathogenic bacteria and GC

To date, several studies have provided mechanistic insights into the links between pathobionts and GC. As aforementioned, the enrichment of S. anginosus is consistently observed in mucosa biopsies of patients with GC from multiple studies. Mechanistically, our previous study demonstrated that S. anginosus produces proinflammatory cytokines, including Ccl20 and Ccl8, to induce acute inflammation in the stomach.49 The surface protein TMPC on S. anginosus mediates its attachment and colonisation of gastric tissues, hence serving as a virulence factor to induce gastric carcinogenesis for long-term infection.49 Upon attachment, TMPC can interact with the annexin A2 receptor on gastric epithelial cells to activate the oncogenic mitogen-activated protein kinase pathway, which further induces the phosphorylation of extracellular signal-regulated kinase and c-Jun N-terminal kinases in mice (figure 2).49 Compared with H. pylori which is mostly depleted in GC, S. anginosus is consistently involved in different stages of gastric carcinogenesis from precancerous lesion, mucosal atrophy, intestinal metaplasia, gastric dysplasia cascade and finally to malignancy.49 In addition, another GC-enriched oral bacterium F. nucleatum has been shown to promote GC development and metastasis in mice through exosome-induced modulation of miR-885-3 p/ephrin type B receptor 2/phosphoinositide 3-kinases (PI3K)/Akt signalling (figure 2).50 Propionibacterium acnes is also another GC-enriched bacterium that can trigger anti-inflammatory M2 polarisation of macrophages through toll-like receptor 4/PI3K/Akt pathway, subsequently promoting the growth of GC cells (figure 2).51 In mice, P. melaninogenica promotes gastric inflammation by activating signal transducer and activator of transcription 3 signalling pathway (figure 2).43 Collectively, while current studies focus heavily on microbial alterations and function in GC, how microbial metabolism and their metabolic products (ie, metabolites) affect gastric carcinogenesis are mostly unclear.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Potential microbiota-correlated mechanisms in the pathogenesis of gastric cancer. Helicobacter pylori mainly serves as the trigger for the initiation of gastric carcinogenesis, while non-H. pylori microbiota can potentially exert their impacts throughout the pathogenesis of gastric cancer through various mechanisms. Virulence factors could be produced by pathogenic microbes to promote the tumourigenesis process. In addition, metabolites are produced by microbes to activate intracellular signalling pathways in host cells. Microorganisms can also attach to the host cell through receptors to induce genetic/epigenetic changes and alter intracellular signalling (for example, the binding of surface protein TMPC from Streptococcus anginosus to the annexin A2 receptor on host cells to activate the pro-tumourigenic mitogen-activated protein kinase (MAPK) signalling). Viruses are incorporated within host cells and exert their function on the cells (for example, Epstein-Barr virus (EBV)-induced hypermethylation of host cells). Furthermore, microbes can modulate the tissue immune microenvironment and affect gastric carcinogenesis. Created with BioRender. EBNA-1, EBV nuclear antigen 1; EphB2: ephrin type B receptor 2; ERK, extracellular signal-regulated kinase; GPR109A, G protein-coupled receptor 109A; HOPX, homologous domain protein homologous box; IL, interleukin; JNK, c-Jun N-terminal kinase; miRNA, microRNA; PD-L1, programmed death ligand 1; PI3K, phosphoinositide 3-kinases; STAT, signal transducer and activator of transcription; TLR4, toll-like receptor 4.

Pathogenic fungus and virus in gastric carcinogenesis

Besides bacteria, fungi and viruses could also be involved in the gastric carcinogenesis processes (figure 1). Similar to bacteria, the diversity and abundance of gastric fungi are decreased in patients with GC.52 53 Fungal dysbiosis in GC is characterised by an increased abundance of Cutaneotrichosporon, Malassezia, Solicoccozyma and Archaeorhizomyces, in two independent Chinese cohorts.53 54 Meanwhile, the enrichment of Candida species C. albicans and C. tropicalis was also detected in GC samples and correlated with poorer survival in an American study.55 Consistently, C. albicans is the most significantly enriched fungi in GC from another Chinese cohort, accounting for around 22% of the overall fungi abundance.52 Co-occurrence of Candida with several bacterial species, including Streptococcus and Lactobacillus, was observed in gastric tumours.55 Fungal dysbiosis in the stomach can also upregulate inflammatory pathways including cytokine and chemokine signalling.53 55 With impaired immune response, patients with GC especially those in advanced stages, are susceptible to infection by opportunistic pathogenic fungi. Yet again, whether these GC-enriched fungi are the cause or consequence of immune dysregulation, and whether they serve as oncogenic pathogens, require further investigation.

Several pathogenic viruses are capable of infecting eukaryotic cells, such as Epstein-Barr virus (EBV). EBV infection accounts for around 7–9% of global GC cases each year, and can induce significant genomic and epigenomic change to promote carcinogenesis.56 For example, EBV-induced gene amplification and expression of programmed death ligand 1 (PD-L1) facilitate GC cells to evade from T cell immunity.57 EBV induces hypermethylation of cell cycle-related gene CDKN2A and p14^ARF promoter by methyltransferase in GC cells.58 EBV-related virus latency gene products, such as EBV nuclear antigen 1, latent membrane protein 2A and microRNAs, also foster carcinogenic processes by inducing epigenetic dysregulation and aberrant mRNA transcription.59 Meanwhile, although several studies reported the potential linkage between other viruses (eg, human papillomavirus, human herpesvirus, hepatitis virus) and GC, none of them have established causal role in gastric carcinogenesis.60 In general, to date, viral community in human stomach remains largely unexplored. Given that other studies have reported the significance of gut virome in other cancers (eg, colorectal cancer), future research is suggested to assess viral alterations of the gastric microbiota in GC development.

Beneficial bacteria in tumour tissues

Probiotics Bifidobacterium and Lactobacillus are lactic acid bacteria that are tolerant of low pH. Several studies revealed that some strains of Bifidobacterium and Lactobacillus are able to survive in the stomach under the strong acidity, digestive enzymes and bile salts.61

Bifidobacterium is generally depleted in GC, compared with chronic gastritis.18 22 A Singaporean cohort reported that Bifidobacterium abundance is reduced by 62-fold in early gastric neoplasia and low-risk intestinal metaplasia.62 Consistently, a meta-analysis indicated a decreased abundance of Bifidobacterium in GC.27 Very low faecal abundance of Bifidobacterium was observed in patients with GC, suggesting indirect interaction between depleted Bifidobacterium and gastric carcinogenesis.63 64 Meanwhile, B. longum could decrease cell viability and angiogenesis of GC cells.65 In a xenograft model of GC, administration of heat-killed B. bifidum inhibits tumour growth.66 Collectively, these preclinical findings have demonstrated anti-tumourigenic effects of Bifidobacterium, implying its potential clinical application against gastric carcinogenesis.

In contrast, current evidence on the alteration of Lactobacillus in GC seems counterintuitive. Although a study reported a decreased abundance of Lactobacillus in GC tissues,62 the majority of studies identified the enrichment of Lactobacillus in patients with GC from different populations.17 19 21 26 30 31 37 Meta-analyses further ascertained the enrichment of Lactobacillus in GC.28 Lactobacillus enrichment was also observed in faecal samples of patients with GC.29 44 Based on this evidence, question on whether virulent rather than probiotic species of Lactobacillus mediate gastric carcinogenesis is raised. However, metagenomic studies demonstrated that L. gasseri, L. reuteri and L. oris, which are all well-recognised probiotics with previously reported tumour-suppressive effects, are enriched in GC.25 51 Meanwhile, a preclinical study demonstrated that L. plantarum could induce apoptosis in GC cells.67 Given the adaptation of Lactobacillus in human stomach, it is possible that its enrichment in GC is attributed to its dominance within the tumour microenvironment, and further mechanistic investigations are required.

Probiotic species are mostly depleted in GC. In a meta-analysis of 825 gastric tissue biopsies, the abundance of Bacillus and Blautia is decreased in GC.27 Decreased abundance of Faecalibacterium and Roseburia was also observed in faecal samples of patients with GC in comparison with healthy individuals.42 44 In general, many GC-depleted species are beneficial ones that can confer a wide range of health advantages, thus suggesting the prophylactic potential of probiotics against GC. Nevertheless, there are currently no consensual probiotics in gastric carcinogenesis, and further explorations in this field are required to translate preclinical findings into clinical settings.

Mechanistic links between beneficial bacteria and GC

Several preclinical studies have revealed the potential link between beneficial bacteria and GC. Administration of heat-killed B. bifidum could induce Akt-p53-dependent apoptosis of GC cells in xenograft mouse models.66 Short-chain fatty acids, particularly butyrate, are produced by a wide range of probiotics, and they are known to exhibit tumour-suppressive effects. For GC, intratumoural CD8⁺ T cells could be activated by butyrate through G protein-coupled receptor 109A (GPR109A)/homologous domain protein homologous box (HOPX) pathway to suppress GC carcinogenesis in mice.68 Butyrate also decreases the levels of immunosuppressive PD-L1 and interleukin (IL)-10 in tumour-associated macrophages, thereby restraining GC tumour growth (figure 2).64

Beneficial fungus and virus in gastric carcinogenesis

Currently available information on beneficial fungi or viruses in GC is extremely sparse. A decreased abundance of Rhizopus is observed in a Chinese cohort, and several species of Rhizopus have been identified to inhibit cancer growth in mice.53 Other Chinese studies reported the depletion of Aspergillus, Phialocephala, Phaeoacremonium, Umbelopsis, Ustilago, Panellus and Cyberlindnera in patients with GC.52 54 55 Although no study has portrayed alterations for viruses in gastric carcinogenesis, modified viruses can be used to deliver therapeutic transgenes with selective replication in tumour cells, thereby inhibiting GC and enhancing treatment efficacy.69–71 For instance, oncolytic herpes simplex virus armed with thrombospondin-1 harbours antitumour efficacy through direct viral oncolysis and anti-angiogenesis in GC.69 The delivery of tumour-suppressor gene TIPE2 into GC cells by adenovirus could inhibit cancer metastasis in mice.70 Chimeric oncolytic poxvirus CF17 directs peritonea to prevent malignant ascites and improve survival of mice with GC peritoneal metastases.71 To date, most GC-depleted fungi and viruses are unexplored for their roles in cancer; hence, extensive observational and mechanistic investigations of gastric mycobiota and virome are needed.

Effects of H. pylori on the gastric microbiota

Although H. pylori initiates gastric carcinogenesis, accumulated studies have revealed its significant depletion from precancerous stage to GC.7 17 18 20 21 30 32 37 Its decrease in abundance could be attributed to the atrophic niche, loss of specialised glandular tissue and impaired gastric acid secretion in GC. Despite its depletion along gastric carcinogenesis, the impact of H. pylori on the gastric microbiota is long-lasting of which its infection decreases gastric microbial richness and diversity.25 27 34 36 72 73 A higher abundance of non-Helicobacter Proteobacteria was observed in H. pylori-infected children,74 while enriched gastroenteritis-inducing bacteria Campylobacter was found in H. pylori-infected adult patients with gastritis or precancerous lesions.75 Another study reported that H. pylori colonisation is associated with enriched Acinetobacter, which commonly affects immunocompromised individuals.76 Helicobacter is also positively correlated with another well-acknowledged pathobionts, Fusobacteria.17 In GC tumour tissues, H. pylori infection is associated with an increased abundance of S. anginosus,33 suggesting that S. anginosus may promote gastric carcinogenesis after H. pylori, forming a so-called ‘H. pylori initiation–non-H. pylori acceleration’ cascade (figure 2).

Effects of H. pylori eradication on the gastric microbiota

H. pylori eradication can in turn restore the diversity and richness of gastric microbiota.34 72 73 76–79 After H. pylori eradication, the abundance of beneficial bacteria including Bifidobacterium and Lactobacillus is significantly increased.77 Commensals such as Firmicutes, Bacteroidetes, Actinobacteria and Cyanobacteria also increase in the gastric mucosa upon H. pylori eradication.72 78 However, it should be noted that H. pylori eradication neither ascertains the restoration of gastric microbiota nor complete abolishment of gastric carcinogenesis. For example, Actinobacteria enrichment after eradication could lead to severe dysbiosis and failure of microbial restoration in H. pylori-infected patients.80 In a Japanese study, H. pylori eradication partially restores microbial diversity in patients with early GC who underwent endoscopic resection, yet the abundance of commensals is still lower than H. pylori-negative individuals.81 Moreover, among patients receiving H. pylori eradication, enriched pathogenic S. anginosus as well as depleted Roseburia and Sphingomonas are associated with persistent gastric inflammation; whereas enriched oral bacteria Peptostreptococcus, Streptococcus, Parvimonas and Prevotella, as well as depleted F. praustznii, are associated with the development of precancerous lesions.79 Fortunately, the supplementation of probiotic Bifidobacterium and Lactobacillus may partially assist the restoration of post-eradication microbial dysbiosis.82 These findings demonstrated that gastric carcinogenesis still occurs under the absence of H. pylori, due to the enrichment of non-H. pylori pathobionts.

Synergistic effects of pathogenic microbes and H. pylori

Although many studies have illustrated how H. pylori induces changes in the gastric microbiota, a few studies have investigated the synergistic effects of gastric pathobionts on H. pylori-induced changes. Coinfection of H. pylori and Staphylococcus salivarius promotes gastric pathogenesis and secretion of proinflammatory cytokines in germ-free INS-GAS mice, compared with H. pylori mono-infection.83 This indicates that gastric pathobionts may work collectively or even synergistically with H. pylori to promote gastric carcinogenesis, yet more studies are necessary to clarify these microbial interactions.

Beneficial microbes inhibit H. pylori

Numerous evidence has revealed the co-exclusive correlations between beneficial microbes and H. pylori.17 20 27 However, due to the dominance of H. pylori in the stomach, it is doubtful whether the co-exclusive correlations among gastric microbes occur due to nutritional and habitat competition of H. pylori, or its antimicrobial effects on beneficial microbes. Hence, the antagonistic effects of different Lactobacillus species on H. pylori have been investigated. In H. pylori-infected mice, L. plantarum supplementation significantly depletes Helicobacter, thereby decreasing H. pylori-induced inflammatory cell infiltration and increasing the abundance of probiotics such as Bifidobacterium.84 Moreover, a study evaluated the effects of 32 strains of lactic acid bacteria on H. pylori and identified that Limosilactobacillus fermentum MN-LF23 and L. gasseri MN-LG80 have the highest efficacy to inhibit H. pylori growth by 80–90%.85 In addition, virulence factors of H. pylori can also be suppressed by Lactobacillus. For instance, L. plantarum ZJ316 downregulates H. pylori virulence genes to reduce proinflammatory cytokines and infiltrated immune cells, thereby alleviating H. pylori-induced gastritis in mice.86

Bifidobacterium also shows antimicrobial effects on H. pylori. Previous studies found that Bifidobacterium inhibits both antibiotic-sensitive and antibiotic-resistant strains of H. pylori through the secretion of antimicrobial peptides.87 At species level, B. bifidum CECT 7366 could inhibit the viability of H. pylori in vitro and attenuate H. pylori-induced damage on gastric tissues in mice.88 B. breve also demonstrated robust inhibitory effect on H. pylori in vitro and alleviated H. pylori-associated gastric inflammation in rats.89 Thus, these findings imply that lactic acid bacteria Lactobacillus and Bifidobacterium have certain commonalities in antagonising H. pylori, which may be due to their production of lactic acid or competition for the binding sites on gastric epithelial cells.

The application of probiotics to suppress H. pylori growth and enhance eradication rate has long been evaluated in clinical studies.90 91 The beneficial effect of probiotics was confirmed in a meta-analysis, which demonstrated that combining bismuth-containing quadruple therapy with Bifidobacterium quadruple viable bacteria tablets markedly improves H. pylori eradication rate and reduces adverse events.90 Intriguingly, another meta-analysis on randomised controlled trials showed that probiotics generally have beneficial impacts on H. pylori eradication, while their effects could become stronger with higher doses and mixed probiotic strains.91 Taken together, these experimental and clinical findings largely support the inhibitory roles of probiotics against H. pylori, thereby reducing the risk of H. pylori-induced gastric carcinogenesis.

Diagnostic microbial biomarkers

Current common clinical biomarkers for GC involve cancer antigen (CA)72-4, CA19-9 and carcinoembryonic antigen, whereas their diagnostic sensitivity is merely around 50%.93 Meanwhile, microbiota represents a new avenue for GC diagnosis, and microbial markers developed for distinguishing GC are summarised in table 1. To detect premalignant lesions, a diagnostic panel of gastric microbial biomarkers showed an area under the curve (AUC) of 0.79 in H. pylori-positive individuals.23 62 For GC, different diagnostic panels of gastric microbial biomarkers have been established with AUC values ranging from 0.82 to 0.96.7 25 27 28 For example, a panel of gastric microbial signatures (Acinetobacter, Peptostreptococcus, Lactobacillus), which were identified by random forest, could accurately distinguish GC from superficial gastritis with an AUC value of 0.96 in a meta-analysis integrating 1270 individuals.28 Moreover, microbial dysbiosis index, which is calculated by integrating enriched and depleted operational taxonomic units in samples, can be used for GC diagnosis, with AUC values of 0.87–0.91 to distinguish GC from superficial gastritis.17 26 Apart from bacteria, gastric fungi may also serve as diagnostic biomarkers for GC. However, the accuracy of fungal biomarkers for GC diagnosis is relatively low with AUC values of 0.71–0.80.52–54

Although gastric mucosal microbial biomarkers have shown promising diagnostic potential, it remains invasive and resource-demanding to collect gastric mucosal samples. To facilitate clinical application, studies have evaluated various non-invasive approaches using oral or gastric fluid samples for GC diagnosis. Using a random forest model, microbial biomarkers in the saliva could distinguish GC from non-malignant lesions with an AUC of 0.91.40 Microbial signatures in gastric fluid can also be used to diagnose GC but with lower accuracy.34 It is noteworthy that oral and gastric fluid microbiota are easily influenced by extrinsic factors, for example, the use of oral rinses. Therefore, standardised sampling workflow should be established before clinical applications.

Faecal microbiota is another non-invasive alternative for GC diagnosis. Microbial biomarkers in faecal samples could be used to detect gastric premalignant lesions23 and GC.29 44 45 For instance, Oceanobacter or Syntrophomonas discriminates GC from chronic gastritis with AUC values of 0.91 or 0.94, respectively.18 A large-scale multicentre study also reported a panel of faecal microbial signatures comprising of S. anginosus and S. constellatus that exhibits great performance in detecting both early and later stages of GC.33 In general, faecal microbial biomarkers exhibit similar diagnostic performance to gastric mucosal microbiota, with less invasiveness and greater convenience and cost-effective. Still, faecal samples are predominantly composed of intestinal bacteria, which may not reflect the gastric microenvironment to the same extent as gastric commensal microbes.

Predictive microbial biomarkers

Microbial biomarkers can also be used to predict outcomes of patients with GC (table 2). For example, patients with a higher abundance of Methylobacterium, Prevotella and Fusobacterium in GC tumour tissues are associated with poorer overall survival.18 94 A higher abundance of Halomonas and Shewanella in gastric mucosa is also associated with poorer survival.95 Collinsella, Blautia, Anaerostipes and Dorea are more abundant in patients with advanced GC than in those with early GC.96 Moreover, increased risk of vessel carcinoma emboli was observed in patients with GC with a higher abundance of Methylobacterium, Oceanobacter and Syntrophomonas.18 Firmicutes is also enriched in faecal samples of patients with GC with lymph node metastasis.97 Currently, translating microbial biomarkers into clinical practice for patients with GC also encounters challenges. Apart from known factors such as age and sex, unknown confounders affecting an individual’s microbiota still exist, and there are no established guidelines for sampling or product efficacy. There is also a lack of consensus on the appropriate timing, location and manner in which microbial diagnostic or prognostic biomarkers should be employed. All these issues need to be thoroughly addressed prior to clinical application.

Targeting H. pylori by probiotics during GC treatment

It is common in clinical practice to use probiotics as adjuvants in H. pylori eradication. However, multiple studies have consistently reported that patients with H. pylori-positive GC have better clinical outcomes compared with H. pylori-negative patients.8 98 For example, H. pylori-positive patients have more favourable outcomes under immunotherapy with higher intratumoural levels of PD-L1⁺ cells and non-exhausted CD8⁺ T cells.8 These findings suggest that targeting H. pylori by probiotics during treatment course might alter immune responses and affect clinical outcomes. In-depth mechanistic investigations and clinical trials are required to examine and ascertain the effects of probiotics and their interactions with H. pylori during GC treatment.

Beneficial bacteria in GC treatment

Gut microbes are closely associated with various GC therapeutics. Gut microbial abundance and diversity in patients with advanced GC are markedly altered by gastric surgery,96 whereas oral administration of Clostridium butyricum among patients receiving gastrectomy could decrease postoperative inflammation and enhance immunity.99 A Lactobacillus probiotic cocktail also reduced gut dysbiosis and inflammation after gastrectomy among patients with GC and rat models.100 In patients with GC after radical gastrectomy, Roux-en-Y reconstruction induces the colonisation of butyrate-producing bacteria, which are associated with the alleviation of postoperative colitis by downregulating inflammatory pathways.101

Chemotherapy and radiotherapy are conventionally used for patients with GC. Transplanting faecal microbiota from healthy obese donor to cachectic patients with metastatic gastro-oesophageal cancer prior to chemotherapy could improve response and survival in a phase II clinical trial.102 In addition, butyrate-producing bacteria increase the efficacy of oxaliplatin by activating IL-12 signalling pathway in both mice and humans.103 Other bacteria that produce short-chain fatty acids also demonstrate protective effects against radiotherapy-induced injury.104 However, the exact effect and mechanism of gut microbiota on the efficacy and outcome of radiotherapy for GC need to be further assessed.

The response rate of patients with advanced GC to immunotherapy is only 10–26%.105 Gut microbiota alterations are associated with the gastric immune microenvironment. Enriched Stenotrophomonas and Selenomona are correlated with increased infiltration of immunosuppressive cells such as regulatory T cells and plasmacytoid dendritic cells into the microenvironment of GC.106 Methylobacterium is negatively correlated with the production of transforming growth factor beta and intratumoural infiltration of CD8⁺ tissue-resident memory T cells.18 Meanwhile, a higher abundance of Lactobacillus is associated with better response to immune checkpoint blockade and survival in patients with GC.107 108 In addition, microbes can exert immunomodulatory effects through their functional metabolites. For example, microbes-derived butyrate suppresses the expression of PD-L1 and IL-10 in tumour-associated macrophages in mouse models of GC.64 Butyrate could also enhance the function of CD8⁺ T cells in the microenvironment of GC through GPR109A/HOPX pathway (figure 2).68 Nevertheless, it is important to pinpoint that the role of microbes in immunotherapy for GC, especially at cellular and molecular levels, remains massively unclear.