Thirty-five patients with CRC diagnosed by pathology and requiring surgical treatment at the General Hospital of Ningxia Medical University from September 2023 to September 2024 were collected. Among these, five patients were excluded due to the presence of other primary malignant tumors and patient/family refusal to participate. Thirty patients were screened according to the inclusion and exclusion criteria. Thirty fecal samples were collected, and 60 samples were collected from CRC tissues and paracancerous tissues (5 cm from the cancerous tissues). Thirty stool samples were collected from healthy volunteers.

The inclusion criteria for CRC patients were the following: (1) patients with a clear histopathological diagnosis; (2) patients who had not received surgery, radiotherapy, chemotherapy, or immunotherapy; and (3) patients who had signed the informed consent.

The exclusion criteria for CRC patients were the following: (1) having inflammatory bowel disease; (2) Received chemotherapy within 2 weeks before sampling; (3) Used antibiotics/probiotics within 2 weeks before sampling. The inclusion criteria for healthy volunteers were as follows: (1) no infectious diseases; and (2) no history of special diets. The exclusion criteria for the healthy volunteers were as follows: (1) with family history of colorectal adenoma or CRC. (3) Used antibiotics/probiotics within 2 weeks before sampling. The study was approved by the ethics review committee (Approval No.KYLL-2025-1827), and all study subjects signed an informed consent form.

Stool samples: The subject should first empty their bladder and flush the toilet to prevent urine contamination. Place a small basin lined with tissue paper into the toilet. Defecate directly into the center of the basin. Using the collection spoon, stir the stool thoroughly from the middle section and scoop out approximately 5 grams (filling the collection box to 2/3 capacity). Immediately tighten the box lid. Place the collection box into a resealable bag, transfer it promptly into a dry ice container, and store it in a -80°C freezer within 24 hours.

Tumor samples: Tumor tissue specimens must be collected immediately upon surgical resection; the interval from tissue devitalization to the initiation of processing should not exceed 30 min to minimize ischemia- and metabolism-induced alterations of the microbial community. To prevent RNA degradation and changes in the microbial community, the samples were immersed in an RNA stabilizer (RNAlater) and stored frozen at -80°C.

Initial blood test data were collected from the CRC patients after admission to the hospital. This included a white blood cell (WBC) count, hemoglobin (Hb), relative ratio of neutrophils to lymphocytes, blood urea nitrogen (BUN), serum creatinine (Scr), aspartate aminotransferase (AST), alanine aminotransferase (ALT), total bilirubin (TBIL), albumin (ALB), alkaline phosphatase (AKP), gamma-glutamyltransferase (GGT), carcinoembryonic antigen (CEA), and glycoantigen 199 (CA199), as well as the pathologic Tumor, Node, Metastasis(TNM) stage of the resected tumors, number of lymph node metastases, and size of the tumors.

Total DNA of the microbial community was extracted from the fecal samples of the CRC patients and healthy volunteers according to the instructions of the FastPure Stool DNA Isolation Kit (MJYH, Shanghai, China). The integrity of the DNA was detected using 1% agarose gel electrophoresis, and DNA concentration and purity were determined using the NanoDrop2000 (Thermo Fisher Scientific, USA).

The above extracted DNA was used as a template, and the upstream primer 338F (5’-ACTCCTACGGGGAGGCAGCAG-3’) and downstream primer 806R (5’-GGACTACHVGGGTWTCTAAT-3’), which carried the barcode sequence, were used. PCR amplification of the V3–V4 variable region of the 16S rRNA gene was carried out as follows: pre-denaturation at 95°C for 3 min, 29 cycles of denaturation at 95°C for 30 s, annealing at 53°C for 30 s, and extension at 72°C for 45 s, followed by a stable extension at 72°C for 10 min with storage at 10°C. The PCR instrument was an ABI GeneAmp^® Model 9700. The PCR reaction consisted of 10 μL 2×Pro Taq, 0.8 μL (5 μM) forward primer, 0.8 μL (5 μM) reverse primer, and 10 ng/μL template DNA. The reaction mixture was supplemented with ddH2O to 20 μL. Three replicates were performed for each sample. PCR products from the same samples were mixed and analyzed by 2% agarose gel electrophoresis to detect fragment size. The target bands were excised, gel-purified on 2% agarose gels, and quantified using a Synergy HTX (Biotek, USA).

The purified PCR products were used for library construction using the NEXTFLEX Rapid DNA-Seq Kit (Bioo Scientific, Austin, Texas, USA): (1) junction linkage; (2) removal of junction self-linkage by magnetic bead screening; (3) enrichment of the library template by PCR amplification; and (4) recovery of the PCR products by magnetic bead recycling to obtain the final library. The second-generation sequencing was performed using the NextSeq 2000 PE300 platform from Illumina (Shanghai Meiji Biomedical Technology Co., Ltd.), with an average Q30 value of 96.6% for the raw data. The third-generation sequencing uses Pacbio. Although Q30 is not used as a base quality assessment, the raw data has achieved high accuracy in obtaining Hifi reads after quality control using SMRT Link 11.0 software.

The second-generation sequencing was performed using the Illumina NextSeq 2000 platform (PE300) at Shanghai Majorbio Bio-pharm Technology Co., Ltd. The detailed procedure is as follows:

The third-generation sequencing uses Pacbio. Purified products of PCR were pooled in equimolar and DNA library was constructed using the SMRTbell prep kit 3.0 (Pacifc Biosciences, CA, USA) according to PacBio’s instructions. Purified SMRTbell libraries were sequenced on the Pacbio Sequel IIe System (Pacifc Biosciences, CA, USA) by Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China). PacBio raw reads were processed using the SMRTLink analysis software (version 11.0) to obtain high-quality Hifi reads with a minimum of three full passes and 99% sequence accuracy. Hifi reads were barcode-identified and length-filtered. For bacterial 16S rRNA gene, sequences with a length < 1,000 or >1,800 bp were removed. The Hifi reads were de-noised using DADA2 (Callahan et al., 2016) plugin in the Qiime2 (Bolyen et al., 2019) (version 2020.2) pipeline with recommended parameters, which obtains single nucleotide resolution based on error profiles within samples. DADA2 denoised sequences are usually called amplicon sequence variants (ASVs). To minimize the effects of sequencing depth on alpha and beta diversity measure, the number of sequence from each sample was rarefied to 6004 which still yielded an average Good’s coverage of 99.90%. Taxonomic assignment of ASVs was performed using the classify-consensus-blast(Blast) consensus taxonomy classifier implemented in Qiime2 and the NT taxon -core -16s database (v2024).

After demultiplexing, the resulting sequences were quality filtered with fastp (v0.19.6) (Chen et al., 2018) and merged with FLASH (v1.2.11) (Magoč and Salzberg, 2011). Then the high-quality sequences were denoised using DADA2 (Callahan et al., 2016) plugin in the Qiime2 (Bolyen et al., 2019) (version 2020.2) pipeline with recommended parameters, which obtains single nucleotide resolution based on error profiles within samples. DADA2denoised sequences are usually called amplicon sequence variants (ASVs). To minimize the effects of sequencing depth on alpha and beta diversity measure, the number of sequence from each sample was rarefied to 20937, which still yielded an average Good’s coverage of 99.90%. Taxonomic assignment of ASVs was performed using the Naive bayes consensus taxonomy classifier implemented in Qiime2 and the SILVA 16S rRNA database (v138) using confidence threshold of 0.7.

All data analyses were performed on the Meggie BioCloud platform (https://cloud.majorbio.com). Alpha diversity indices, including species richness (Sobs), Shannon diversity, and Simpson indices, were calculated using mothur software (http://www.mothur.org/wiki/Calculators, version 1.30.2). Differences in alpha diversity between groups were assessed using the Wilcoxon rank-sum test. Principal coordinate analysis (PCoA) based on the Bray-Curtis distance algorithm was used to test the similarity of the microbial community structure among the samples. Common/endemic core species were obtained using Venn diagram analysis. A non-parametric Wilcoxon signed-rank test was used to assess differences in paired samples. Species were selected for correlation network diagram analysis based on Spearman correlation |r| > 0.6, p < 0.05. The Gut Microbiota Health Index (GMHI) is calculated using Python (version 2.7.10) software and the following formula.

Calculation formula:

(Gupta et al., 2020).

The Microbial Dysbiosis Index (MDI) was calculated using the R vegan package (version 2.4.3) in R (version 3.3.1) and Python (version 2.7.10), and is defined as: MDI=log10[(total abundance in genera increased in disease group)/(total abundance in genera decreased in disease group)] (Gunathilake et al., 2020).

In this study, the dilution curves (Figures 1A–D) illustrate a gradual increase in species richness with an increase in the number of sequences, which eventually leveled off. This indicates a gradual attainment of stability in species diversity, suggesting that the amount of experimental sequencing data was reasonable. Additionally, it implies that a greater amount of data yields diminishing returns in the discovery of new species. Alpha diversity serves as a measure of the abundance and diversity of microbiota within samples. In this study, the observed richness (Sobs),the Shannon diversity index and the Simpson diversity index(For the Simpson diversity index (D), smaller values indicate higher diversity.) were utilized for evaluation. For the tissue samples, the Shannon index was higher in the paracancerous tissue compared to the cancerous tissue (Figure 1E). Conversely, the Simpson index showed the opposite trend (Figure 1H). Although the differences were not statistically significant, there was a tendency for the paracancerous tissue to exhibit higher alpha diversity than the cancerous tissue.

The Shannon index was lower in the fecal samples from patients with CRC compared to healthy individuals (Figure 1F). Conversely, the Simpson index showed the opposite trend (Figure 1I). While these differences were not statistically significant, the alpha diversity tended to be greater in healthy individuals than in individuals with CRC.

Comparing tissue and fecal samples from patients with CRC, the Shannon index was lower in the cancerous tissue samples than in the fecal samples (Figure 1G). This indicates that the diversity of the community was higher in the fecal samples than in the cancerous tissue samples from the same CRC patients. Similarly, the Simpson index was higher in the cancerous tissue samples than in the fecal samples, suggesting that the diversity of the fecal microbiota was greater than that of colorectal cancer (CRC) tissues (Figure 1J). This difference was statistically significant (p < 0.01), suggesting that the presence of dominant species in the feces from CRC patients reduced the Simpson index.

For paracancerous tissue samples from CRC patients and fecal samples from healthy individuals, the Sobs index was higher in the paracancerous tissue samples than in the fecal samples. This indicates that paracancerous tissues in CRC exhibited a higher richness compared to feces (Figures 1K–M). The difference was statistically significant in the Sobs index (p < 0.01).

Comparing differences in bacterial genera between cancerous and paracancerous tissues from CRC patients (Table 1), it was found that Fusobacterium had a significantly higher proportion in the cancerous tissue samples (p < 0.05). Ruminococcus, Faecalibacterium, Bifidobacterium, and Blautia had a significantly higher proportion in the paracancerous tissue samples than cancerous tissue samples (p < 0.05) (Figure 9A). Comparing feces from CRC patients and healthy individuals at the genus level (Table 2) revealed that Bifidobacterium, Blautia, and Ruminococcus accounted for a higher percentage in feces from healthy individuals, with a statistically significant difference between the cancerous and paracancerous tissue samples (p < 0.05). Akkermansia, Bacteroides, Mediterraneibacter, and Enterococcus accounted for a statistically higher proportion in the fecal samples from CRC patients than healthy individuals (p < 0.05) (Figure 9B). At the species level, fecal samples from CRC patients had a statistically higher percentage of Akkermansia muciniphila and Fusobacterium varium than the fecal samples of healthy individuals (p < 0.01) (Figure 9C). Comparing fecal and cancerous tissue samples from CRC patients (Table 3) at the level of analyzed genera revealed a significantly higher percentage of Faecalibacterium, Bifidobacterium, and Blautia in the fecal samples than cancerous tissue samples (p < 0.05). Fusobacterium, Methylobacterium/Methylorubrum, and Acinetobacter accounted for a higher percentage in the cancerous tissue samples in CRC patients, and the difference was statistically significant (p < 0.01) (Figure 9D). Comparing paracancerous tissues from CRC patients with feces from healthy patients (Table 4) at a genus level revealed that Ruminococcus, Bifidobacterium, Blautia, Anaerobutyricum, and Fusicatenibacter were in high abundance in feces from healthy individuals (Figure 9E). Paracancerous tissue samples from CRC patients had a high percentage of Enterococcus, Akkermansia, Stenotrophomonas, and Pantoea (p < 0.05).

By analyzing the correlation of different bacterial genera in cancerous tissues and clinical factors in CRC patients (Table 5), Escherichia-Shigella, for example, showed a statistically significant positive correlation with tumor size (p < 0.05). Methylobacterium/Methylorubrum showed a statistically significant positive correlation with tumor stage (p < 0.05). Parvimonas, Gemella, Peptostreptococcus, and Streptococcus were negatively correlated with the number of tumor lymph node metastases, and the difference was statistically significant (p < 0.01). Parvimonas, Gemella, Peptostreptococcus, and Streptococcus were negatively correlated with CEA values. The difference was statistically significant (p < 0.05). Parvimonas, Gemella, Peptostreptococcus, and Streptococcus were positively correlated with TBIL, and the difference was statistically significant (p < 0.05) (Figure 10A).

At the species level, Escherichia coli was positively correlated with the number of lymph node metastases with a statistically significant difference (p < 0.05). Parvimonas micra, Peptostreptococcus stomatis, Fusobacterium nucleatum, and Gemella morbillorum were negatively correlated with CEA and the number of tumor lymph node metastases, with a statistically significant difference (p < 0.05). Pantoea agglomerans was negatively correlated with tumor size (Figure 10B).

The present study found that CRC patients had lower alpha diversity than healthy individuals, a result that is consistent with several studies (Zhao et al., 2024; Darnindro et al., 2025). A decline in alpha diversity may be attributed to the proliferation of pathogenic bacteria and a reduction in beneficial bacteria. In the gut of CRC patients, the enrichment of opportunistic pathogens (such as Fusobacterium nucleatum and Escherichia coli) and the depletion of butyrate-producing bacteria (such as Roseburia spp. and Faecalibacterium spp.) may impair intestinal barrier function and promote inflammation. This may lead to structural imbalance of the microbiota and impaired metabolic function, ultimately resulting in microbial dysbiosis (Li et al., 2023). The alpha diversity of tumor tissues was lower than that of adjacent tissues (Figure 1E), suggesting that metabolic characteristics of the TME (such as hypoxia and lactate accumulation) may directly inhibit microbial survival. For instance, elevated lactate levels within tumor tissues can alter the local pH, suppressing the growth of strict anaerobes such as Bifidobacterium spp., while simultaneously promoting the proliferation of acid-tolerant bacteria such as Enterococcus spp., thereby fostering a tumor-promoting microenvironment (Certo et al., 2021). Furthermore, fecal samples from CRC patients exhibited a reduced Shannon index (Figure 1F), indicating a trend toward simplified species distribution within their gut microbiota. This was characterized by the overgrowth of dominant taxa and a concomitant reduction in the abundance of commensal or beneficial bacteria, such as butyrate producers. Consequently, the ecological balance of the gut microbial community was disrupted.

Notably, the lower alpha diversity observed in cancerous tissues compared to adjacent paracancerous tissues (Figure 1E) reflects heterogeneity between the local microbial community within tumors and the overall gut microbiota. For instance, while the gut microbiota of CRC patients has shown enrichment of Enterobacteriaceae, the intratumoral communities are characterized by a predominance of Fusobacterium nucleatum (Obuya et al., 2022). This heterogeneity highlights the need for future studies to incorporate multi-compartment sampling (e.g., mucosal, tissue, and fecal specimens) to comprehensively assess microbiota alterations.

The beta diversity analysis revealed a significant separation in the gut microbiota structure between CRC patients and healthy individuals (Figures 3B, C, p< 0.01). This finding has also been confirmed by other studies (Messaritakis et al., 2024). PCoA further demonstrated that although microbial composition at the genus level did not reach statistical significance between cancerous and adjacent paracancerous tissues (p = 0.326), a distinct separation trend was observed (Figure 3A). This trend may be potentially attributable to limited sample size or substantial inter-individual heterogeneity. Hierarchical clustering analysis (Figure 2A) revealed that cancerous tissue samples formed independent clusters, suggesting the presence of a tumor-specific microbial signature. This signature was characterized by an elevated abundance of Escherichia-Shigella and Fusobacterium (Figure 6A). Notably, Escherichia-Shigella contributes to CRC development through the production of the genotoxin colibactin, which induces DNA damage (Mo et al., 2020). Fusobacterium spp. promotes tumor proliferation by activating the Wnt/β-catenin signaling pathway (Valciukiene et al., 2023).

At the species level, CRC patients exhibited increased abundance of Akkermansia muciniphila (Figure 9C). Under healthy conditions, controlled mucin degradation by Akkermansia muciniphila promotes mucus layer renewal and provides nutrients (e.g., SCFAs) to other commensals, thereby supporting barrier function. However, recent studies have indicated that in advanced CRC stages, Akkermansia muciniphila may exacerbate tumor cell proliferation and invasion through excessive mucosal degradation. Furthermore, disruption of the mucus layer may activate the TLR4–NF-kB pathway, driving macrophage polarization toward the M1 phenotype and amplifying inflammatory responses within the TME (Zhang et al., 2024). This context-dependent functional duality underscores the stage-specific nature of microbial contributions to disease pathogenesis, mandating future analyses that stratify patients according to clinical disease stage.

This study revealed that the GMHI was significantly higher in paracancerous tissues compared to cancerous tissues (p < 0.001), whereas the MDI showed an inverse pattern (Figures 4A, E). The reduced GMHI likely reflects depletion of protective taxa, such as Blautia spp. and Faecalibacterium spp., within the TME. This depletion may remodel the TME through multiple pathways, including metabolic dysregulation, immunosuppression, expansion of pro-tumorigenic bacteria, therapy resistance, and dysregulated cross-organ signaling. Notably, Blautia and Faecalibacterium are major producers of SCFAs, such as butyrate and propionate. SCFAs enhance antitumor immunity by inhibiting histone deacetylase (HDAC) activity and suppressing pro-inflammatory cytokine release (e.g., IL-6, TNF-α) via activation of G protein-coupled receptors (e.g., GPR43/109A). Specifically, butyrate inhibits tumor cell proliferation through HDAC suppression (Jiang et al., 2024). Conversely, elevated MDI correlated with enrichment of pathogenic bacteria, such as the pathobiont Bacteroides fragilis, which secrete enterotoxins that activate the STAT3 signaling pathway, thereby promoting CRC stem cell expansion (Zhou et al., 2023).

Healthy individuals exhibited significantly higher GMHI values than the CRC patients (Figure 4B), suggesting that GMHI may serve as a complementary biomarker for CRC screening. Furthermore, fecal samples demonstrated a higher GMHI than the cancerous tissue samples (Figure 4C), indicating that the stool microbiota retained residual healthy signatures while the local tumor-associated microbiota exhibited more severe dysbiosis. Our analysis revealed significant compositional differences between cancerous and adjacent paracancerous tissues in CRC patients, with adjacent tissues already displaying dysbiotic features resembling those in the cancerous tissues. This suggests that microbial imbalance may precede tumorigenesis and become progressively compartmentalized during cancer progression (Mo et al., 2020). As a distal sampling compartment, the fecal microbiota may retain stronger systemic signatures of host–microbiota interactions (e.g., immunomodulatory metabolites). The compositional divergence between fecal and tumor-resident microbiota reveals the multi-layered nature of host–microbial crosstalk. Fecal communities reflect systemic ecological equilibrium with tumor-localized microbiota actively shaping the microenvironment and driving malignant transformation (Valciukiene et al., 2023). Collectively, these findings necessitate future investigations integrating multi-compartment sampling (feces, tumor tissue, adjacent mucosa) with spatiotemporal dynamic analyses to delineate the pathological contributions of microbiota throughout carcinogenesis.

Escherichia coli: Strains producing colibactin induce double-strand DNA breaks through DNA alkylation. This damage activates host DNA damage response pathways (e.g., p53 signaling), but persistent genomic insult leads to genomic instability, thereby driving tumorigenesis (de Souza et al., 2024). In this study, an elevated abundance of Escherichia coli was observed in the cancerous tissues of CRC patients (Figure 6A), suggesting its potential synergistic role in tumorigenesis through the driver-passenger model. Colibactin initiates tumorigenesis by inducing critical oncogenic mutations (e.g., Adenomatous Polyposis Coli (APC), Kirsten rats arcomaviral oncogene homolog(KRAS)). Other gut microbes or inflammatory microenvironments may further promote tumor progression. For instance, colibactin-induced cellular senescence is accompanied by a senescence-associated secretory phenotype (SASP), which recruits immune cells and fosters a tumor-promoting microenvironment (Dalmasso et al., 2024).

Bacteroides fragilis: Enterotoxigenic Bacteroides fragilis (ETBF) was significantly enriched in both cancerous tissue samples and fecal samples of CRC patients. The Bacteroides fragilis toxin (BFT) gene carried by ETBF strains is recognized as a pivotal oncogenic driver. ETBF promotes carcinogenesis through a toxin–inflammation–genomic instability–metabolic dysregulation cascade, establishing a tumor-promoting microenvironment and serving as a key driver in CRC pathogenesis (Scott et al., 2022).

Akkermansia muciniphila: This bacterium exhibits a context-dependent role in CRC, functioning as a double-edged sword. It may prevent tumorigenesis by maintaining barrier integrity, yet contribute to tumor progression via mucin degradation in the TME. Through mucin degradation, Akkermansia muciniphila generates metabolites, including SCFAs, which promote mucus layer renewal and enhance tight junctions between intestinal epithelial cells, thereby reducing pathogen and toxin translocation. Additionally, its outer membrane proteins (e.g., Amuc_1100) activate Toll-like receptor (TLR) signaling pathways, inducing regulatory T cell (Treg) differentiation and mitigating chronic inflammation—a key contributor to CRC pathogenesis (Fan et al., 2023). A study (Zhu et al., 2023) have reported enrichment of Akkermansia muciniphila or its genetic signatures in tumor tissues of advanced-stage CRC patients. This enrichment may promote cancer cell proliferation by activating oncogenic pathways such as Wnt/β-catenin. Additionally, mucin degradation products (e.g., sialic acid) may be utilized by tumor cells to support immune evasion or metastatic progression.

Bifidobacterium: A higher abundance of Bifidobacterium was observed in healthy individuals (Figure 9B). These bacteria enhance intestinal barrier function by secreting exopolysaccharides (EPS) and suppressing excessive inflammatory responses by inducing regulatory T cell (Treg) differentiation (Kainulainen et al., 2022).

Faecalibacterium prausnitzii: A study (Nittayaboon et al., 2022) have documented significantly reduced abundance of Faecalibacterium prausnitzii in the gut of CRC patients. As a primary butyrate producer, its depletion (Figure 6C) likely contributes to diminished colonic butyrate concentrations in CRC. Butyrate exerts tumor-suppressive effects through HDAC inhibition, inducing cancer cell apoptosis, suppressing proliferation, and reversing epithelial-mesenchymal transition (EMT). The loss of butyrate-mediated HDAC inhibition may consequently promote tumor progression (Luo et al., 2023).