Previous research has revealed that the microbial diversity and abundance in breast cancer tissues surpass those in corresponding normal breast tissues, with adjacent normal tissues exhibiting an intermediate composition (7). Various factors such as ethnicity, tumor stage and molecular subtype contribute to the heterogeneity of the breast microbiota (31). Methylobacterium radiotolerans and Staphylococcus consistently emerge as the predominant bacteria in breast tumors based on multi-omics analyses (7,32–34). Additionally, Tzeng et al (26) further demonstrated that Lacibacter and Ezakiella are notably enriched in breast cancer tissues, and their abundance increases with the elevation of tumor stage. By contrast, the relative abundances of Sphingomonas yanoikuyae and Acetobacter aceti are higher in normal breast tissues, while Anaerococcus, Caulobacter and Streptococcus are notably absent in breast cancer (33,34). Despite these insights into microbial community characteristics, further research is required to elucidate the precise mechanisms by which these microorganisms influence processes such as breast cancer cell dissemination, intravasation and extravasation.

Meta-genomic research has revealed substantial alterations in the microbial community composition within tumor tissues of patients with CRC compared with those of healthy individuals (35). Fusobacterium nucleatum has consistently emerged as a prominent member of the microbiota associated with CRC (36). Additionally, Bacteroides fragilis, Parvimonas and Bacteroides have been notably observed to be more abundant in CRC tissues (37,38). Nevertheless, Repass (39) performed a comparative analysis of tumor and adjacent normal tissues from the same cohort of patients, demonstrating that Fusobacterium nucleatum was not prevalent in the majority of CRC samples. Moreover, the authors found no marked distinction in the overall microbiota composition between the two tissue types. This suggests that discrepancies in findings may be attributed to variations in sample origins, population diversity and methodological disparities.

The lungs undergo continuous gas exchange with the external environment, making their microbial communities vulnerable to influences from the oral, nasal and gut microbiota through ‘ascending’ or ‘hematogenous’ routes (40). Meta-genomic studies have verified marked dysbiosis in the lung tissues of individuals with lung cancer, characterized by heightened local inflammatory responses, elevated microbial quantities and the proliferation of specific pro-inflammatory bacteria (41). Notably, potential pathogens such as Brevundimonas and Escherichia are markedly more abundant in cancerous tissues compared with adjacent normal tissues, whereas Corynebacterium, Lachnoanaerobaculum and Halomonas, prevalent in healthy individuals, are markedly diminished (28). Subsequent investigations indicated that Streptococcus and Peptoniphilus are associated with an increased lung cancer risk, whereas the presence of Aggregatibacter offers a protective effect (42).

The local pancreatic microbiota is closely associated with the susceptibility, progression and treatment response of pancreatic ductal adenocarcinoma (PDAC) (43,44). Propionibacterium acnes, one of the earliest bacteria isolated from pancreatic cancer tissues, can induce chronic inflammation and provide an inflammatory microenvironment for tumorigenesis (45). Meta-genomic analysis has revealed that Gammaproteobacteria is the dominant bacterial class in human pancreatic tumors (46). In a PDAC mouse model, Bacteroides and Parabacteroides were notably enriched in tumor tissues, suggesting their potential involvement in cancer progression (47).

It was further found that intratumoral microbial diversity is associated with prognosis in patients with PDAC. Riquelme et al (48) reported that long-term survivors exhibited markedly higher tumor microbial diversity compared with short-term survivors. Moreover, the co-existence and high abundance of Pseudoxanthomonas, Streptomyces, Saccharopolyspora and Bacillus clausii could serve as an independent marker for predicting long-term survival.

Gastric microbiota dysbiosis serves a crucial role in the progression of gastric cancer. Helicobacter pylori (H. pylori) has been designated as a Group I carcinogen by the World Health Organization. However, only ~3% of individuals infected with H. pylori ultimately develop gastric cancer (49,50). Apart from H. pylori, various meta-genomic studies have identified specific microbiota associated with gastric cancer, including Peptostreptococcus stomatis, Streptococcus anginosus, Parvimonas micra, Slackia exigua and Dialister pneumosintes (51). Throughout the continuum from gastritis-gastric adenoma-early-stage gastric cancer-advanced stage gastric cancer, the microbiota composition undergoes dynamic changes. Akkermansia and Lachnospiraceae NK4A136 dominate the gastritis stage, whereas Lactobacillus and Veillonella markedly increase in abundance in gastric cancer tissues compared with normal tissues (30,52). The aforementioned distinctive microbial dysbiosis may expedite the carcinogenic process through mechanisms such as inflammation promotion, metabolic reprogramming and immune suppression. Therefore, monitoring the composition and function of the gastric microbiota holds notable clinical value for the early diagnosis and risk assessment of gastric cancer.

Ovarian cancer ranks among the most aggressive gynecological malignancies. High-grade serous ovarian cancer is considered to originate from either the ovarian surface epithelium or fallopian tubes (53). Previous studies have revealed that dysregulation of the local microbiota may serve a regulatory role in the development and progression of ovarian cancer (54,55). Meta-genomic analyses have identified Roseomonas mucosa and Sphingomonas US_602 as being notably enriched in ovarian tumor tissues, indicating their potential as key microbial players in this disease (24). Additionally, human papillomavirus (HPV) infection has been proposed as a risk factor for ovarian cancer, with higher expression levels of high-risk HPV observed in malignant tissues compared with adjacent normal tissues (56). However, a study by Ingerslev et al (57) found no notable association between high-risk HPV and epithelial ovarian cancer in Caucasian patients, suggesting it may not be related to ethnic differences.

Prostate cancer is the most prevalent malignant tumor in men (58). Previous research has demonstrated that prostate tissue is not sterile, with microbial DNA identified in tumor samples from 87% of patients with prostate cancer, implicating the local microbiota in the disease's onset and progression (59,60). Multi-omics analyses reveal elevated levels of Escherichia and Propionibacterium in prostate cancer, suggesting their potential role in tumor development (61). Additionally, liquid biopsy indicates markedly increased levels of Streptococcus, Peptostreptococcus and Haemophilus in the urine of patients with prostate cancer (62). Furthermore, markers of parasites such as Toxoplasma and Plasmodium are detected in prostate cancer tissues, potentially contributing to tumorigenesis through chromosomal damage and reactive oxygen species generation (63).

Melanoma arises from the malignant transformation of melanocytes and is the deadliest form of skin cancer (64). The skin microbiota varies notably by location: Staphylococcus dominates sebaceous regions, Corynebacterium is prevalent in moist areas and β-Proteobacteria are found in dry regions (65,66). In melanoma tissues, Enterobacter and Streptococcus are notably enriched, facilitating tumor immune evasion by inhibiting CD8⁺ T cell infiltration and reducing chemokine expression (29,67). Notably, Kozmin et al (68) discovered that a commensal Staphylococcus epidermidis strain producing 6-N-hydroxyaminopurine could markedly inhibit B16F10 melanoma growth and decrease ultraviolet-induced tumor incidence, suggesting a novel approach for microbiota-driven prevention and treatment of skin cancer.

Intratumoral microorganisms can drive genomic instability through a dual-pathway mechanism involving direct genotoxicity and indirect activation, with the specific mechanisms being strain-dependent (23). In CRC, Escherichia coli carrying the pks gene island secretes the genotoxin colibactin, inducing somatic base substitutions and insertion or deletion mutations, while the cyto-lethal distending toxin produced by Campylobacter can cause DNA double-strand breaks (69). Aflatoxin B₁ (AFB₁) is metabolized by cytochrome p450 to the reactive AFBO metabolite, which blocks nucleotide excision repair and induces TP53 mutations, promoting carcinogenesis in multiple organs (70). In CRC, Fusobacterium nucleatum activates the E-cadherin/β-catenin signaling pathway via the FadA adhesin, upregulates checkpoint kinase 2 and induces DNA damage (71). Elevated β-glucuronidase in the breast cancer microenvironment can catalyze the release of reactive intermediates, indirectly causing DNA damage (8). In the intestinal mucosa of patients with familial adenomatous polyposis, enterotoxigenic Bacteroides fragilis and Escherichia coli were found to colonize synergistically, accelerating early-stage CRC development through DNA damage and inflammatory pathways (72).

Intratumoral microorganisms colonize TME, influencing immune remodeling and accelerating tumor progression (73). Their viable bacteria, residues and metabolites collectively regulate immune cell functions and inflammatory responses, enhancing immune suppression and promoting pro-tumor phenotypes (74). Enterobacteriaceae and Pseudomonadaceae in breast and pancreatic cancers directly promote tumor cell proliferation by activating the NF-κB/TLR4 inflammatory axis and upregulating the PI3K/Akt signaling pathway (34,75). In CRC, Fusobacterium nucleatum increases intracellular Ca²⁺, facilitates E-cadherin and Krüppel-like factor 4 (KLF4) interaction, promotes KLF4 nuclear translocation and upregulates integrin a5 transcription, driving proliferation, invasion and metastasis (76). In pancreatic cancer, Campylobacter, Selenomonas and Clostridium difficile inhibit immune cell activity via the MET proto-oncogene, receptor tyrosine kinase-protein tyrosine kinase 2 and programmed cell death protein 1 (PD-1) pathways, enhancing tumor invasiveness; these genera, rare in normal tissues, are potential diagnostic markers (77). Fusobacterium upregulates pathways such as cytotoxic T-lymphocyte associated protein 4, JAK-STAT, TNF and PI3K-AKT-mTOR, creating an immunosuppressive microenvironment and remodeling the transcriptome to promote oral squamous cell carcinoma progression (11). Conversely, microorganisms can inhibit tumor progression; for instance, Propionibacterium in normal breast tissue directly inhibits tumor growth by secreting antimicrobial peptides and produces short-chain fatty acids to activate free fatty acid receptors 2 and 3, reducing inflammation and tumorigenesis risk (34).

Intratumoral microorganisms modify the metabolic networks of glucose, lipids, and amino acids within tumors via the ‘metabolite-receptor-signal axis’, influencing the immune microenvironment and thereby affecting tumor progression. In CRC, the anaerobic bacteria Fusobacterium nucleatum, Clostridium spp. and Bacteroides spp. produce short-chain fatty acids that activate pro-inflammatory pathways such as NF-κB, elevating tumorigenesis risk (38). In gastric cancer, Peptostreptococcus spp. enhance phosphatidylcholine hydrolysis and triglyceride synthesis by upregulating phospholipase C and 1-acyl-sn-glycerol-3-phosphate acyltransferase, driving lipid metabolic reprogramming and tumor progression (78). Additionally, the study by Flores-García et al (79) demonstrated that long non-coding RNAs boost glycolysis by upregulating 6-phosphofructo-2-kinase/fructose-2,6-biphosphatase 4, phosphoglycerate kinase 1 and lactate dehydrogenase A, while promoting glutamine catabolism through increased glutamate dehydrogenase 1 and Golgi transport 1A expression, supplying extra energy to tumor cells. These enzymes further activate the hypoxia inducible factor 1α/PI3K/Akt/mTOR pathway, enhancing tumor cell proliferation, survival and invasion. Table I summarizes the compositional differences and action mechanisms of the tumor microbiome (80,81).

Liquid biopsy, a non-invasive technique, utilizes peripheral blood samples to monitor tumor characteristics. Its core detection targets include circulating tumor cells, circulating tumor DNA (ctDNA), cell-free RNA and extracellular vesicles, which have demonstrated applicability in the early diagnosis and dynamic monitoring of various cancers (82). Specifically, ctDNA enables real-time assessment of tumor burden by identifying tumor-specific somatic mutations, copy number variations and epigenetic modifications, such as methylation (83). Furthermore, ctDNA analysis can predict postoperative recurrence in patients with CRC (84).

The bacteremia microbial profile in patients with CRC was closely associated with tumor stage, as reported by Kwong et al (85). Specifically, the positive inflection point of Clostridium perfringens appeared 100 days after the diagnosis of bacteremia, while the peaks of Peptostreptococcus spp. and Fusobacterium spp. occurred at 121 days and 127 days, respectively. Notably, Fusobacterium spp. and Peptostreptococcus spp. were associated with the diagnosis of early-stage CRC, whereas Streptococcus spp. was associated with late-stage CRC. Furthermore, the abundances of bacterial and fungal DNA in peripheral blood were highly consistent with the characteristics of the intratumoral microbiome, suggesting their potential as circulating microbial biomarkers (20). The integration of high-throughput sequencing and machine learning algorithms has enabled the multi-cancer early detection technology based on cell-free circulating DNA to identify molecular signals shared by multiple tumors in a single blood sample (86). Wang et al (87) demonstrated that Lactobacillus and Streptococcus could serve as microbial markers for gastric cancer, aiding in its early non-invasive diagnosis. Collectively, these studies indicate that circulating tumor microbiome markers have the potential to become non-invasive and highly sensitive diagnostic tools, which could expand the application of liquid biopsy in early tumor screening (88).

In situ tissue detection technology, including immunohistochemistry, immunofluorescence and fluorescence in situ hybridization (FISH), serves a crucial role in directly identifying microbial biomarkers within tumor tissues. These techniques are valuable for elucidating the spatial distribution of microorganisms in relation to tumor cells and their involvement in tumorigenesis and progression (89). Notably, FISH stands out as a well-established molecular cytogenetic approach that employs highly complementary fluorescent DNA probes to sensitively and specifically detect and localize bacterial 16S rRNA genes (90).

Studies have shown that analyzing the tumor-associated microbiome using various techniques such as 16S rRNA gene amplicon sequencing and metagenomic sequencing has the potential to target and eliminate tumor-promoting microorganisms or enrich microorganisms that enhance antitumor immunity, thereby regulating the composition of the microbiome and providing new strategies for the diagnosis, prevention and treatment of tumors (91). Chai et al (92) identified specific bacteria, such as Klebsiella pneumoniae and the fungus Paraburkholderia, in intrahepatic cholangiocarcinoma tissues using FISH. In pancreatic cancer, Fusobacterium was independently associated with poor prognosis, suggesting it has a role as a prognostic biomarker (93). Zhang et al (94) showed that the tumor microbiome influences treatment outcomes and survival rates. By examining microbial changes in patients with stage III–IV non-small cell lung cancer, the authors identified Haemophilus parainfluenzae, Serratia marcescens, Acinetobacter junii and Streptococcus constellatus as markers for predicting 2-year survival rates. Yamamura et al (95) reported that high Fusobacterium nucleatum levels in esophageal squamous cell carcinoma indicated poor prognosis, serving as a potential biomarker. Intratumoral microbiota thus holds promise for tumor diagnosis and prognosis, although large-scale, multi-center studies are necessary to confirm their clinical utility.

Microorganism-based adjuvant therapies have demonstrated notable potential to enhance the efficacy of tumor immunotherapy. A study has shown that in patients with CRC receiving anti-PD-1 monoclonal antibody (mAb) immune checkpoint blockade, the abundance of Fusobacterium nucleatum is markedly increased in non-responders, and its metabolite succinate can induce CRC cells to develop resistance to immunotherapy (96). A clinical trial on locally advanced rectal cancer found that the anti-PD-1 mAb dostarlimab exhibited promising therapeutic effects, with some patients avoiding chemotherapy, radiotherapy or surgery (97). Furthermore, a phase I trial reported complete remission in a metastatic melanoma patient resistant to anti-PD-1 therapy after FMT (98).

Emerging strategies, such as oncolytic viruses (OVs) and engineered bacteria, have shown promise in overcoming the limitations of CAR-T cell therapy for solid tumors, owing to their tumor selectivity and programmable immunogenicity (99). For instance, the OV CD19t can induce CD19 expression on tumor cells, thereby enhancing the antitumor response of CD19-CAR-T cells in mouse models (100). Additionally, the combined use of attenuated Brucella strains and adoptive transfer of antigen-specific CD8⁺ CAR-T cells has demonstrated near-complete elimination of tumor growth and proliferation, achieving a 100% host survival rate and effectively overcoming CAR-T cell resistance (101). However, intratumoral microorganisms may also weaken the efficacy of traditional chemotherapy. 5-fluorouracil (5-FU) is a first-line chemotherapy drug for CRC, but Fusobacterium nucleatum and Escherichia coli can metabolically deplete 5-FU and activate the autophagy pathway, reducing the local drug concentration and efficacy (102). Therefore, precise regulation of the intratumoral microbiota is of great importance for optimizing comprehensive tumor treatment.

The commensal microbiota and essential nutrients have demonstrated notable potential in cancer treatment. Multiple studies have revealed the promoting effects of commensal bacteria and their metabolites on the efficacy of immune checkpoint blockade (ICB). For example, Jia et al (103) found that the abundance of commensal Lactobacillus spp. was positively associated with ICB responsiveness, and its metabolite, indolepropionic acid, could improve the immunotherapy outcomes of melanoma, breast cancer and CRC by regulating the stemness program of intratumoral CD8⁺ T cells. Bifidobacterium spp. has been shown to enhance the immunotherapy response by activating the STING signaling pathway (104). Oral administration of live Lactobacillus rhamnosus GG increased the numbers of tumor-infiltrating dendritic cells and T cells, thereby enhancing the antitumor activity of anti-PD-1 therapy (105). Kalaora et al (29) identified human leukocyte antigen-presenting peptide fragments derived from intratumoral bacteria in melanoma, which could be co-presented by antigen-presenting cells and tumor cells. This can increase the diversity of immunogenic antigens and promote T cell activation, thus enhancing the benefits of immune checkpoint inhibitors.

Probiotics can mitigate the side effects of conventional cancer treatments. Linn et al (106) demonstrated that a probiotic mix of Lactobacillus acidophilus LA-5 and Bifidobacterium animalis subsp. lactis BB-12 reduces acute diarrhea in patients with cervical cancer undergoing radiotherapy. Certain commensal microbiota also have direct therapeutic potential. Montalban-Arques et al (107) showed that oral administration of four Clostridium spp. strains prevented and effectively treated CRC in a mouse model, outperforming anti-PD-1 monotherapy. Engineered microbial therapy is an emerging cancer treatment strategy. Canale et al (108) developed an engineered Escherichia coli Nissle1917 that converts ammonia-nitrogen to L-arginine within tumors, enhancing CD8⁺ T cell infiltration and achieving antitumor effects when combined with programmed death-ligand 1 antibodies. Commensal bacteria, their metabolites and engineered strains can reshape the TME through various pathways, offering new strategies and targets for combined immunotherapy.

The differences in the local microenvironments of microbial colonization and the heterogeneity of host cells form a complex interaction network, because both microbial communities and host cells exhibit notable spatiotemporal dynamic changes (109,110). However, current technological constraints hinder the concurrent acquisition of microbial spatial distribution and host single-cell transcriptome information (111,112).

The causal relationship between microbial colonization and tumor progression is difficult to clarify, mainly due to unclear temporal relationships and interference from confounding factors (111). Most existing studies are based on cross-sectional analyses, making it difficult to distinguish whether microbial colonization is a driving factor, a concomitant phenomenon or a secondary result of tumors (8). The dynamic changes in the TME (such as hypoxia or immune suppression) may reversely shape the microbial community, forming a complex feedback loop of bidirectional interaction (70). In addition, individual differences (including genetic background, diet and antibiotic use) and the spatiotemporal heterogeneity of sample collection (such as differences in intratumoral/peritumoral microbiota) need further investigation (70,113).

Research on microbe-host interactions faces a notable lack of standardization, mainly manifested in the absence of unified specifications for sampling methods and bioinformatics analysis, resulting in poor data comparability (69). The complexity of clinical sample collection poses challenges in guaranteeing the quality and quantity of tumor samples (114). Ensuring sample purity and devising strategies to effectively mitigate environmental microbial contamination are crucial objectives during the detection process (4).

Firstly, future research should prioritize the integration of tumor-microbiome-immune multi-omics through systematic analysis of interaction mechanisms using dynamic network modeling (4). To achieve this, advanced technologies such as single-cell transcriptomics, spatial metabolomics and high-throughput sequencing should be combined for synchronous data collection across different scales (115). Machine learning algorithms should be employed to integrate metagenomic, metabolomic and immune cell interaction data to identify key hub molecules (116). Integrating organoid models with dynamic perturbation experiments could validate the predicted networks. This approach aims to develop a quantifiable and controllable ‘microbiome-metabolism-immunity’ computational model, serving as a predictive platform for targeted interventions and advancing research from association to causal mechanisms (117).

Secondly, future research should focus on the development of precision intervention strategies based on the characteristics of the tumor microbiome. Through the integration of multi-omics data such as intratumoral microbiota sequencing, metabolomics and immune microenvironment analysis, a system of predictive biomarkers should be established to guide individualized treatment (118). Utilizing machine learning to analyze tumor-specific microorganisms and validating the interaction between microorganisms and drugs through organoid co-cultures and high-throughput drug screening are essential (119). Moreover, the advancement of microorganism-targeted delivery technologies such as nanocarriers and the design of microbiome combination therapies are crucial (97,120). Ultimately, the goal is to establish a closed-loop transition from microbial diagnostic profiling to treatment decision-making, thereby shifting the paradigm of tumor treatment from a generalized approach to precise microbiota regulation.

Finally, future research should also focus on establishing a highly biomimetic organoid-microbe co-culture system to construct a standardized and high-throughput research and translation platform for the interaction between microorganisms and the host (121,122). Long-term co-culture systems that pair specific organ-derived organoids with complex microbial communities should be developed, and microfluidic-chip technologies should be integrated to simulate key in vivo microenvironmental parameters such as oxygen gradients (123,124). Building on these systems, real-time multimodal monitoring should be employed to analyze the regulatory mechanisms by which microbial colonization influences barrier function, immune responses and drug sensitivity in organoids (125). Additionally, an automated culture-and-detection platform should be established and coupled with artificial-intelligence algorithms to predict the therapeutic effects of microbiota-targeted interventions, thereby accelerating the translation of mechanistic discoveries into clinical treatment regimens (126). This platform will bridge the technological gap between in vitro models and clinical trials and provide accurate prediction tools for microbial targeted therapy.