Association between gut microbial dysbiosis and clinical prognosis of advanced colorectal cancer

Abstract

BACKGROUND

Gut microbiota dysbiosis is involved in the progression of colorectal cancer, but its prognostic value in advanced colorectal cancer (aCRC) and correlation with chemotherapy response remain insufficiently clarified.

AIM

To investigate the association of gut microbial dysbiosis with clinicopathological features, first-line chemotherapy efficacy, and long-term prognosis in patients with aCRC.

METHODS

A total of 124 patients with stage IIIB-IV aCRC admitted from September 2020 to September 2025 were retrospectively enrolled in this study. The 16S rRNA high-throughput sequencing was performed to analyze gut microbiota and stratify patients into corresponding groups. Meanwhile, the levels of fecal short-chain fatty acids and secondary bile acids were detected, and clinical and pathological data of all subjects were collected. Differences in microbial diversity and species composition between groups were analyzed. Survival analysis was conducted using the Kaplan-Meier method, and prognostic factors were identified via univariate and multivariate Cox regression analysis. In addition, stratification analyses based on TNM staging and driver gene status were performed, and the efficacy of first-line chemotherapy was compared between the two groups.

RESULTS

There were no statistically significant differences in demographic characteristics, baseline tumor characteristics, or treatment regimens between the two groups (P > 0.05); the group with dysbiosis had a higher proportion of poorly differentiated tumors, stage IV disease, and vascular and nerve invasion (P < 0.05). In the microbiota dysbiosis group, all α-diversity indices were lower, and β-diversity showed significant intergroup divergence. The abundance of beneficial bacteria such as the Firmicutes phylum decreased, while that of pathogenic bacteria such as the Fusobacteria phylum increased, accompanied by decreased short-chain fatty acid levels and accumulation of secondary bile acids; all intergroup differences were statistically significant (P < 0.001). With a median follow-up of 32.5 months, both overall survival and disease-free survival were significantly shorter in the group with abnormal gut microbiota (P < 0.001); After adjusting for confounding factors such as age, carcinoembryonic antigen, and chemotherapy regimen in a multivariate Cox regression analysis, abnormal gut microbiota was an independent adverse prognostic factor for overall survival (hazard ratio =2.314, 95%CI: 1.258-4.259, P = 0.007). Stratified analysis revealed that the adverse prognostic effect of dysbiosis remained consistent across different TNM stages, with no significant interaction with tumor stage (P > 0.05). Among the 96 patients receiving first-line chemotherapy, the objective response rate and disease control rate were both higher in the group with normal microbiota (P < 0.05), and this difference in treatment efficacy was not influenced by the status of RAS/BRAF driver gene mutations.

CONCLUSION

Abnormalities in the gut microbiota are closely associated with increased tumor aggressiveness in aCRC, disruptions in microbial structure and metabolism, poor response to chemotherapy, and poor prognosis. They may serve as potential biomarkers for assessing patient prognosis and predicting the efficacy of chemotherapy, thereby providing a basis for clinical interventions targeting the gut microbiota.

Key Words: Advanced colorectal cancer; Gut microbial dysbiosis; Overall survival; Disease-free survival; Response to chemotherapy

Core Tip: This retrospective cohort study conducted a stratified analysis of patients with stage IIIB-IIIC and stage IV advanced colorectal cancer, confirming that gut microbiota dysbiosis – by modulating bile acid metabolism and the tumor immune microenvironment – is not only an independent adverse prognostic factor for overall survival but also significantly reduces the objective response rate to first-line chemotherapy.

INTRODUCTION

Colorectal cancer is a highly prevalent malignant tumor worldwide, with high morbidity and mortality; the disease burden is particularly heavy in China[1]. Advanced colorectal cancer (aCRC) accounts for a considerable proportion of newly diagnosed cases. Even with standardized comprehensive treatment, long-term survival remains unsatisfactory, and recurrence and metastasis are the primary causes of treatment failure[2,3]. Currently used clinical prognostic evaluation indicators have limitations and cannot accurately predict treatment response and long-term survival, so it is urgent to find novel and reliable prognostic biomarkers.

As the “second genome” of the human body, the gut microbiota is an important component of the tumor microenvironment. It is deeply involved in the occurrence, progression, and drug resistance of colorectal cancer by regulating the intestinal barrier, chronic inflammation, anti-tumor immunity, and drug metabolism[4,5]. Most existing studies focus on the association between microbiota dysbiosis and the risk of colorectal cancer, while studies on the correlation between microbiota dysbiosis and long-term prognosis in advanced patients remain insufficient, and no unified clinical conclusions have been reached; therefore, relevant research needs to be expanded urgently[6].

Combining clinical data, gut microbiota detection, and follow-up data, this study systematically explored the correlation of gut microbiota dysbiosis with clinicopathological features, first-line treatment response, and long-term prognosis in patients with aCRC, clarifying its value in prognostic evaluation and providing reliable evidence for clinical prognostic risk stratification and targeted microbiota intervention therapy.

MATERIALS AND METHODS

Study design and setting

This study was a single-center, retrospective cohort study conducted at the Department of Gastrointestinal Surgery of the First Hospital of Jiaxing. Total 124 Patients with pathologically confirmed primary aCRC (American Joint Committee on Cancer 8^th edition[7] TNM stage IIIB-IV) admitted between September 2020 and September 2025 were consecutively enrolled. This study was approved by the Institutional Review Board of our hospital (approval No. 2026-LP-212), and all subjects provided written informed consent before enrollment. The research process strictly followed the ethical principles of the Declaration of Helsinki.

Inclusion and exclusion criteria

Inclusion criteria: (1) Histologically confirmed primary colorectal cancer via postoperative pathology or puncture biopsy, with American Joint Committee on Cancer 8^th edition TNM stage IIIB-IV (advanced stage); (2) Qualified fresh mid-section fecal samples were collected before the first anti-tumor treatment, and 16S rRNA high-throughput sequencing of gut microbiota was completed using standardized specimen collection and detection procedures; (3) Complete clinicopathological and follow-up data, with an actual follow-up time ≥ 6 months and no loss to follow-up or missing key data; and (4) No anti-tumor treatment (including neoadjuvant chemoradiotherapy, targeted therapy, immunotherapy, or tumor-related local treatment) before enrollment.

Exclusion criteria: (1) Comorbidity of primary malignant tumors of other systems or a previous history of malignant tumors (except skin basal cell carcinoma without recurrence for more than 5 years); (2) Use of antibiotics, probiotics, or prebiotics within 2 weeks before enrollment, or glucocorticoids and immunosuppressants within 1 week before enrollment; (3) Comorbidity of organic intestinal diseases (e.g., ulcerative colitis, Crohn’s disease) or a history of intestinal surgery unrelated to colorectal cancer; (4) Comorbidity of uncontrolled active infection, severe hepatic and renal insufficiency (Child-Pugh class C), or other serious underlying diseases that preclude tolerance of follow-up and detection; (5) Missing clinicopathological, follow-up, or fecal detection data, rendering statistical analysis impossible; and (6) Comorbidity of severe mental illness or cognitive impairment, preventing cooperation with follow-up and data collection[8].

Collection of general data

Baseline data for all subjects were collected through the electronic medical record system of our hospital, including: (1) Demographic characteristics include gender, age, height, weight, body mass index, smoking history, drinking history, and underlying diseases (hypertension, type 2 diabetes); (2) Clinicopathological features include primary tumor site, maximum diameter, histological differentiation, TNM staging, vascular and perineural invasion, and preoperative serum carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 levels; (3) Treatment regimens and molecular pathology data include retrieve patient medical records to compile statistics on curative surgery, first-line standard chemotherapy [oxaliplatin combined with calcium folinate and fluorouracil (FOLFOX), capecitabine combined with oxaliplatin (CAPOX), and irinotecan combined with calcium folinate and fluorouracil (FOLFIRI)], and targeted therapy; and (4) Verify the mutation rates of RAS and BRAF in rat sarcoma filtrating toxin carcinogenic homolog.

Detection and analysis of gut microbiota

Fecal specimen collection: Fresh mid-section fecal samples (approximately 5 g) were collected from all patients within 24 hours after admission and before the initiation of the first anti-tumor treatment. These were immediately placed in sterile enzyme-free cryopreservation tubes (specification: 2 mL; manufacturer: Jiangsu Kangwei Century Biotechnology Co., Ltd.), snap-frozen in liquid nitrogen, and then transferred to a -80 °C ultra-low temperature freezer (model: DW-86 L388; manufacturer: Qingdao Haier Special Electric Appliance Co., Ltd.) for storage to avoid repeated freezing and thawing. All samples were sequenced within 1 month after collection.

Fecal genomic DNA extraction and 16S rRNA sequencing: Fecal bacterial genomic DNA was extracted using the QIAamp Fast DNA Stool Mini Kit (specification: 50 T; manufacturer: QIAGEN, Germany). DNA integrity was verified by 1% agarose gel electrophoresis (agarose specification: 100 g; manufacturer: Beijing Solarbio Science & Technology Co., Ltd.), and DNA concentration and purity were measured using a NanoDrop 2000 spectrophotometer (manufacturer: Thermo Fisher Scientific, United States). The OD₂₆₀/280 value of qualified samples was 1.8-2.0.

The V3-V4 region of the bacterial 16S rRNA gene was selected for PCR amplification, using primer sequences 341F (5’-CCTACGGGNGGCWGCAG-3’) and 805R (5’-GACTACHVGGGTATCTAATCC-3’). The PCR reaction system consisted of: (1) 25 μL of 2 × Taq PCR MasterMix [specification: 10 mL; manufacturer: Tiangen Biotech (Beijing) Co., Ltd.]; (2) 2 μL of each upstream and downstream primer; (3) 50 ng of template DNA; and (4) Enzyme-free water up to 50 μL. Reaction conditions: (1) Pre-denaturation at 95 °C for 3 minutes; (2) 30 cycles of 95 °C for 30 seconds; (3) 55 °C for 30 seconds; (4) 72 °C for 30 seconds; and (5) Final extension at 72 °C for 5 minutes. The amplified products were electrophoresed on a 2% agarose gel, purified using a gel recovery kit (specification: 50 T; manufacturer: QIAGEN, Germany), and then subjected to high-throughput sequencing using the Illumina MiSeq platform (model: MiSeq; manufacturer: Illumina, United States). Sequencing was performed by Shanghai Sangon Biotech Engineering Co., Ltd.

Bioinformatics analysis: QIIME 2 software was used to sequentially perform quality control, splicing, chimera removal, and denoising on the raw sequencing reads to obtain effective sequences, followed by operational taxonomic unit clustering. Species classification annotation was completed with reference to the SILVA 138 database, and the confidence threshold was set to 97%. The α-diversity analysis was performed using the Chao1 index, Shannon index, and Simpson index to evaluate the species richness and evenness of the microbiota. β-diversity analysis was carried out using principal coordinate analysis to compare differences in microbiota community structure between groups, and statistical verification was performed using the Adonis test. Simultaneously measure short-chain fatty acids and secondary bile acids in feces as indicators of microbial community function and metabolism.

Criteria for judging gut microbiota dysbiosis: Referring to the “Recommendation for management of intestinal microbial flora disorders”[9] and combining the microbiota sequencing results of this study, the following judgment criteria were formulated: Gut microbiota dysbiosis was defined as a Shannon index lower than the lower quartile of the study cohort, the relative abundance of beneficial bacteria (such as Bifidobacterium, Lactobacillus, and Faecalibacterium) lower than the median of the cohort, and the relative abundance of pathogenic bacteria (such as Fusobacterium nucleatum, Escherichia, and Bacteroides fragilis) higher than the median of the cohort. Patients meeting the above criteria were assigned to the dysbiosis group, and the remainder to the normal flora group.

Follow-up and prognostic endpoints

Outpatient re-examination combined with telephone follow-up was adopted. Follow-up was conducted every 3 months for the first 2 years after treatment and every 6 months from year 2 to year 5. The follow-up deadline was September 30, 2025. Cases lost to follow-up were treated as censored data. The primary endpoint was overall survival (OS), defined as the time from pathological diagnosis to all-cause death. The secondary endpoint was disease-free survival (DFS) – only for patients undergoing radical surgery – defined as the time from pathological diagnosis to tumor recurrence, distant metastasis, or all-cause death. Chemotherapy efficacy was evaluated according to the Response Evaluation Criteria in Solid Tumors 1.1 edition[10], and categorized into complete response (CR), partial response (PR), stable disease, and progressive disease. The objective response rate (ORR) was calculated as (CR + PR) cases/total cases × 100%, and the disease control rate (DCR) as (CR + PR + stable disease) cases/total cases × 100%.

Statistical analysis

SPSS 26.0 and R 4.3.1 software were used for data analysis. Measurement data with a normal distribution were expressed as mean ± SD, and an independent sample t-test was used for inter-group comparison. Measurement data with a non-normal distribution were expressed as median (interquartile range), and the Mann-Whitney U test was used for inter-group comparison. Categorical data were expressed as n (%), and the χ² test or Fisher’s exact test was used for inter-group comparison. Survival analysis was performed using the Kaplan-Meier method to draw survival curves, and the log-rank test was used to compare survival rates between groups. Prognostic factors were first screened using a univariate Cox proportional hazards regression model; variables with P < 0.05 were included in a multivariate Cox regression model to calculate the hazard ratio (HR) and 95%CI. The significance level was α = 0.05, and a two-sided P < 0.05 was considered statistically significant.

RESULTS

Comparison of clinicopathological features between the two groups

There were no significant differences in demographic characteristics, primary tumor site, tumor size, preoperative tumor marker levels, or treatment regimen composition between the two groups (P > 0.05). However, the proportions of poorly differentiated tumors, stage IV tumors, positive vascular invasion, and positive perineural invasion were significantly higher in the abnormal flora group, while the proportions of well/moderately differentiated tumors and stage IIIB-IIIC tumors were significantly lower (P < 0.05; Table 1).

Table 1 Comparison of clinicopathological and treatment-related data of patients with advanced colorectal cancer between the two groups, median (interquartile range)/n(%).

Variable	Normal flora group (n = 48)	Abnormal flora group (n = 76)	Z/χ2	P value
Demographic characteristics
Male	27 (56.25)	45 (59.21)	0.106	0.745
Age (years)	63.5 (54.25, 71.75)	60.00 (53.00, 72.00)	0.703	0.482
BMI (kg/m2)	22.82 (21.34, 26.02)	22.82 (19.87, 25.51)	0.972	0.331
Smoking history	15 (31.25)	26 (34.21)	0.117	0.733
Drinking history	13 (27.08)	24 (31.58)	0.284	0.594
Hypertension	14 (29.17)	24 (31.58)	0.081	0.777
Type 2 diabetes mellitus	7 (14.58)	15 (19.74)	0.535	0.464
Tumor pathological characteristics
Primary tumor site			0.041	0.905
Colon	26 (54.17)	42 (55.26)
Rectum	22 (45.83)	34 (44.74)
Maximum tumor diameter (cm)	4.65 (4.06, 6.14)	4.92 (3.79, 5.87)	0.693	0.489
Histological differentiation			11.297	0.001
Well/moderate differentiation	39 (81.25)	39 (51.32)
Poor differentiation	9 (18.75)	37 (48.68)
TNM staging			10.728	0.001
IIIB-IIIC	37 (77.08)	36 (47.37)
IV	11 (22.92)	40 (52.63)
Positive vascular invasion	15 (31.25)	44 (57.89)	8.374	0.004
Positive perineural invasion	13 (27.08)	39 (51.32)	7.095	0.008
Elevated preoperative CEA	28 (58.33)	50 (65.79)	0.701	0.402
Elevated preoperative CA19-9	25 (52.08)	44 (57.89)	0.403	0.526
Treatment regimen			0.043	0.979
Radical resection + adjuvant chemotherapy	23 (47.92)	35 (46.05)
Palliative chemotherapy	16 (33.33)	26 (34.21)
Chemotherapy combined with targeted therapy	9 (18.75)	15 (19.74)

BMI: Body mass index; CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9.

Gut microbiota diversity and metabolic differences

The α-diversity analysis showed that the levels of the Shannon, Chao1, and Simpson indices in the abnormal flora group were significantly lower than those in the normal flora group, with statistically significant differences between the two groups (P < 0.001; Figure 1). Table 2 compares the levels of short-chain fatty acids and secondary bile acids – metabolic products of the gut microbiota – between the two groups. Abnormal flora group showed significantly lower levels of short-chain fatty acids and markedly higher levels of secondary bile acids, with statistically significant differences between the groups (P < 0.001). β-diversity principal coordinate analysis analysis showed that there was an overall difference in the gut microbiota community structure between the two groups, and samples showed a clear separation trend on the principal coordinate axis. The Adonis test further confirmed that the inter-group difference was statistically significant (P < 0.001; Figure 2).

Figure 1 Comparison of gut microbiota α-diversity indices between the two groups. A: Shannon index; B: Chao1 index; C: Simpson index.

Figure 2 The β-diversity analysis of gut microbiota between the two groups. PCoA2: Principal coordinate analysis 2.

Table 2 Comparison of short-chain fatty acid and secondary bile acid levels in the two groups of patients, mean ± SD.

Group	n	Short-chain fatty acids (mmol/L)	Secondary bile acids (μmol/L)
Normal flora group	48	31.82 ± 6.35	22.15 ± 4.26
Abnormal flora group	76	19.24 ± 5.12	44.87 ± 7.32
t value	-	12.129	19.502
P value	-	< 0.001	< 0.001

Gut microbiota composition characteristics

Phylum-level species composition analysis showed that the dominant phyla of gut microbiota in the two groups were consistent, mainly consisting of Firmicutes and Bacteroidetes. The proportion of Firmicutes in the dysbiosis group was significantly lower than that in the normal group, while the proportions of Bacteroidetes, Proteobacteria, and Fusobacteria were significantly higher (P < 0.001; Figure 3A). Genus-level species composition analysis showed that the relative abundance of beneficial bacteria such as Bifidobacterium, Lactobacillus, Faecalibacterium, and Roseburia was significantly higher in the normal flora group, while the relative abundance of pathogenic bacteria such as Fusobacterium nucleatum, Escherichia, and Bacteroides fragilis was significantly higher in the dysbiosis group, with no significant differences in other genera between the two groups (Figure 3B).

Figure 3 Stacked plots of relative abundance of gut microbiota between the two groups. A: Phylum-level species composition; B: Genus-level species composition.

Survival prognosis analysis

The median follow-up time of this study was 32.5 months, with a follow-up rate of 100% and no loss to follow-up. During the follow-up period, 57 cases of all-cause death and 62 cases of tumor recurrence or distant metastasis occurred. Survival analysis showed that the OS and DFS curves of the dysbiosis group were lower than those of the normal flora group, with statistically significant differences between the two groups (P < 0.001). The 3-year OS rate and 3-year DFS rate of the dysbiosis group were significantly lower than those of the normal flora group (Figure 4).

Figure 4 Kaplan-Meier survival curves of patients in the two groups. A: Overall survival (n = 124); B: Disease-free survival (n = 72).

Univariate prognostic analysis

A total of 17 potential prognostic factors were included in the univariate Cox proportional hazards regression analysis of OS in patients with aCRC. The results showed that poor histological differentiation, stage IV TNM staging, positive vascular invasion, positive perineural invasion, gut microbial dysbiosis, and the absence of radical resection were associated with a higher risk of death and significantly impacted OS (P < 0.05). Gender, age, body mass index, smoking history, alcohol consumption, hypertension, type 2 diabetes mellitus, primary tumor site, maximum tumor diameter, elevated preoperative CEA, and elevated preoperative carbohydrate antigen 19-9 were not significantly associated with OS (P > 0.05; Table 3).

Table 3 Univariate Cox regression analysis of factors influencing overall survival in patients with advanced colorectal cancer.

Variable	β	SE	HR	95%CI	P value
Gender (male = 1, female = 0)	0.215	0.247	1.240	0.762-2.017	0.385
Age (≥ 65 years = 1, < 65 years = 0)	0.207	0.239	1.230	0.769-1.967	0.387
BMI (< 18.5 kg/m2 = 1, others = 0)	0.312	0.406	1.366	0.616-3.029	0.443
Smoking history (yes = 1, no = 0)	0.186	0.249	1.204	0.739-1.961	0.456
Drinking history (yes = 1, no = 0)	0.154	0.257	1.166	0.704-1.931	0.549
Hypertension (yes = 1, no = 0)	0.178	0.245	1.195	0.738-1.934	0.468
Type 2 diabetes mellitus (yes = 1, no = 0)	0.286	0.284	1.331	0.762-2.325	0.314
Primary tumor site (rectum = 1, colon = 0)	0.103	0.234	1.108	0.700-1.755	0.660
Maximum tumor diameter (≥ 5 cm = 1, < 5 cm = 0)	0.225	0.235	1.253	0.790-1.986	0.338
Histological differentiation (poor = 1, well/moderate = 0)	0.758	0.300	2.135	1.187-3.840	0.011
TNM staging (IV = 1, IIIB-IIIC = 0)	1.016	0.282	2.763	1.586-4.813	< 0.001
Vascular invasion (positive = 1, negative = 0)	0.700	0.282	2.014	1.159-3.497	0.013
Perineural invasion (positive = 1, negative = 0)	0.656	0.279	1.927	1.117-3.325	0.018
Elevated preoperative CEA (yes = 1, no = 0)	0.327	0.240	1.387	0.866-2.220	0.174
Elevated preoperative CA19-9 (yes = 1, no = 0)	0.349	0.238	1.418	0.889-2.261	0.142
Gut microbial dysbiosis (yes = 1, no = 0)	1.055	0.292	2.872	1.621-5.086	< 0.001
Radical resection (no = 1, yes = 0)	0.974	0.283	2.648	1.517-4.621	< 0.001

BMI: Body mass index; CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9; HR: Hazard ratio.

Multivariate prognostic analysis

Include both univariate positive variables and key clinical variables in the multivariate Cox regression model. Gut microbial dysbiosis, stage IV TNM staging, poor histological differentiation, and the absence of radical resection were independent adverse prognostic factors affecting OS in patients with aCRC (P < 0.05; Table 4). The two factors – vascular invasion and nerve invasion – were not included in the final equation after adjustment in the multivariate model due to collinearity and confounding effects with tumor stage and grade; age, CEA, and chemotherapy regimen did not reach statistical significance but were associated with potential confounding effects.

Table 4 Multivariate Cox regression analysis of factors influencing overall survival in patients with advanced colorectal cancer.

Variable	β	SE	HR	95%CI	P value
Age (≥ 65 years = 1)	0.211	0.325	1.235	0.653-2.337	0.524
Elevated preoperative CEA (yes = 1)	0.204	0.318	1.226	0.658-2.284	0.519
Chemotherapy regimen (combined with targeted therapy = 1)	0.187	0.321	1.206	0.642-2.267	0.558
Histological differentiation (poor = 1, well/moderate = 0)	0.686	0.319	1.985	1.062-3.712	0.031
TNM staging (IV = 1, IIIB-IIIC = 0)	0.782	0.316	2.187	1.176-4.068	0.013
Gut microbial dysbiosis (yes = 1, no = 0)	0.839	0.310	2.314	1.258-4.259	0.007
Radical resection (no = 1, yes = 0)	0.810	0.317	2.246	1.203-4.192	0.011

HR: Hazard ratio; CEA: Carcinoembryonic antigen.

Chemotherapy efficacy analysis

A total of 96 patients received first-line systemic chemotherapy in this study, comprising 38 in the normal flora group and 58 in the dysbiosis group. The primary systemic chemotherapy regimens were FOLFOX and CAPOX, administered to 49 patients (51.04%) and 37 patients (38.54%), respectively; the FOLFIRI regimen was administered to 10 patients (10.42%). Twenty-four patients received combination therapy with targeted agents, including 16 patients treated with bevacizumab and 8 with cetuximab. The treatment regimens in the two groups were well-balanced and comparable. Evaluation of the chemotherapy response showed that the ORR and DCR of the normal flora group were significantly higher than those of the dysbiosis group, with statistically significant differences between the groups (P < 0.05; Table 5).

Table 5 Comparison of first-line systemic chemotherapy efficacy and clinical benefit rates between the two groups, n (%).

Efficacy index	Normal flora group (n = 38)	Abnormal flora group (n = 58)	χ2 value	P value
CR	3 (7.89)	2 (3.45)	-	0.631
PR	19 (50.00)	17 (29.31)	-	-
SD	12 (31.58)	20 (34.48)	-	-
PD	4 (10.53)	19 (32.76)	-	-
ORR	22 (57.89)	19 (32.76)	5.984	0.014
DCR	34 (89.47)	39 (67.24)	5.872	0.015

CR: Complete response; PR: Partial response; SD: Stable disease; PD: Progressive disease; ORR: Objective response rate; DCR: Disease control rate.

Subgroup analysis by TNM stage

To account for stage-related confounding, this study conducted stratified analysis and interaction tests: (1) Stages IIIB-IIIC (73 cases): OS and DFS in the abnormal flora group remained shorter than those in the normal flora group (P < 0.05), dysbiosis was an independent risk factor in this subgroup (HR = 2.012, 95%CI: 1.036-3.907, P = 0.039); (2) Stage IV (51 cases): The abnormal flora group had a poorer prognosis (HR = 2.458, 95%CI: 1.127-5.356, P = 0.024); and (3) Interaction analysis indicated: There was no significant interaction between gut microbiota status and TNM stage (P = 0.417), confirming that abnormal gut microbiota is an independent factor for poor prognosis across all stages.

Efficacy analysis by driver gene subgroup

In all three subgroups, the ORR and DCR in the normal flora group were higher than those in the abnormal flora group, with a consistent overall trend. Among these, the intergroup differences in both efficacy indicators were statistically significant in the wild-type subgroup (P < 0.05), while in the RAS-mutant subgroup, only the DCR showed a significant intergroup difference (P < 0.05); The BRAF V600E mutation subgroup had a small sample size, and the differences between groups did not reach statistical significance (P > 0.05), as shown in Table 6.

Table 6 Comparison of first-line chemotherapy efficacy between the two groups in subgroups with different driver gene statuses, n (%).

Group	n	RAS-mutantion		n	BRAF V600E mutation		n	Wild-type
		ORR	DCR		ORR	DCR		ORR	DCR
Normal flora group	22	12 (54.55)	19 (86.36)	3	1 (33.33)	2 (66.67)	23	15 (65.22)	21 (91.30)
Abnormal flora group	24	10 (29.41)	21 (61.76)	7	0 (0.00)	2 (28.57)	35	12 (34.29)	24 (68.57)
χ2 value	-	3.538	3.960	-	-	-	-	5.337	4.125
P value	-	0.060	0.047	-	0.300	0.500	-	0.021	0.042

ORR: Objective response rate; DCR: Disease control rate.

DISCUSSION

This study explored the correlation of gut microbiota status with clinicopathological features, treatment response and long-term prognosis in patients with aCRC, confirming that gut microbiota dysbiosis is closely related to adverse clinical outcomes, which can provide a reference for prognostic evaluation and intervention strategy formulation of this disease.

Clinicopathological analysis showed that patients with microbiota dysbiosis had worse tumor differentiation, later TNM staging, and higher rates of vascular and perineural invasion, indicating that microbiota dysbiosis is associated with tumor aggressiveness. Studies by Huang et al[11] and Arrè et al[12] showed that microbiota dysbiosis can promote tumor cell proliferation, invasion and metastasis through persistent intestinal chronic inflammation. Jiang et al[13] proved that enrichment of pro-cancer bacteria such as Fusobacterium nucleatum can also enhance the invasiveness of colorectal cancer, which is consistent with the results of this study. The baseline data of the two groups were balanced and comparable, providing a reliable basis for subsequent inter-group comparison.

In this study, 16S rRNA high-throughput sequencing technology was used to analyze gut microbiota, which is a standardized method for gut microbiota research and has been widely used in colorectal cancer-related studies, ensuring the reliability of detection results[14,15]. Microbiota diversity analysis showed that the α-diversity indices in the dysbiosis group were lower, and β-diversity showed obvious inter-group separation. The α-diversity is a core indicator reflecting the ecological stability of microbiota, and its reduction indicates imbalanced microbiota structure. A study by Newsome et al[15] showed that the gut microbiota diversity of colorectal cancer patients was lower than that of healthy people and negatively correlated with disease stage. This study further verified this association in advanced patients. β-diversity inter-group separation indicated that there were essential differences in microbiota community structure between the two groups, and Adonis test further confirmed that the difference was reliable, which was consistent with the conclusion of microbiota structure separation between colorectal cancer patients and healthy people in the study by Bell et al[16].

Analysis of species composition revealed that, at the phylum level, the abnormal flora group had a lower proportion of Firmicutes and higher proportions of Bacteroidetes, Proteobacteria, and Fusobacteria; at the genus level, the abundance of beneficial bacteria decreased, while that of pathogenic bacteria increased. Further analysis of microbial metabolites in this study found that the abnormal flora group exhibited significantly reduced levels of short-chain fatty acids and an accumulation of secondary bile acids. The Firmicutes phylum maintains intestinal barrier integrity and suppresses inflammatory responses by producing short-chain fatty acids; a decrease in its abundance can easily lead to intestinal immune imbalance, creating conditions conducive to tumor progression[17,18]. Excessive accumulation of secondary bile acids can activate intestinal nuclear receptors, promoting tumor cell proliferation and invasion. Regarding the mechanism by which Fusobacterium nucleatum mediates chemotherapy resistance, this study conducted an in-depth analysis in conjunction with the tumor immune microenvironment. Accumulation of Clostridium nucleatum activates the autophagy pathway, degrading chemotherapy drug-targeted proteins; simultaneously, it persistently activates the classical TLR4/NF-κB inflammatory pathway, inducing chronic inflammation in the tumor microenvironment[19]; furthermore, this bacterium can inhibit pyroptosis, reducing the sensitivity of tumor cells to chemotherapy drugs. These mechanisms are not mediated by a single inflammatory pathway but result from the combined effects of metabolic disorders, immune imbalance, and abnormal programmed cell death, thereby addressing the limitation of existing studies that focus solely on “pro-inflammatory” aspects.

In this study, univariate combined with multivariate Cox regression was used to analyze prognostic factors, which is a classic statistical method for tumor prognosis research, which can effectively control confounding factors and improve the reliability of results[20,21]. Survival analysis showed that patients in the dysbiosis group had shorter OS and DFS and lower 3-year survival rate; multivariate analysis indicated that gut microbiota dysbiosis was an independent adverse prognostic factor for OS of aCRC. Vascular invasion and perineural invasion were not included in the final equation after multivariate correction due to collinearity with tumor staging and differentiation. Studies by Wang et al[21] and Yadav et al[22] confirmed that pro-cancer bacteria enrichment and beneficial bacteria reduction caused by microbiota dysbiosis can affect patient prognosis by changing tumor immune microenvironment and aggravating tumor drug resistance. The results of this study are consistent with them, and confirm that microbiota dysbiosis can be used as a prognostic indicator independent of traditional pathological factors, supplementing the existing prognostic evaluation system. Subgroup and interaction analyses based on the TNM classification further confirmed that the adverse prognostic impact of gut microbiota dysbiosis remained consistent regardless of whether the patients were in stage IIIB-IIIC (potentially curable) or stage IV (systemic disease), thereby ruling out any potential confounding effects of staging-related imbalances on the findings.

Chemotherapy efficacy analysis showed that the ORR and DCR in the normal flora group were higher. Based on the combination of chemotherapy regimens and driver gene stratification data, FOLFOX/CAPOX is the mainstream first-line regimen, and RAS/BRAF mutations do not alter the association between the microbiome and response to chemotherapy. Gut microbiota can participate in chemotherapy drug metabolism and regulate anti-tumor immune response. Studies by Li et al[23] and Cao et al[24] confirmed that beneficial bacteria such as Bifidobacterium and Lactobacillus can improve chemotherapy response in colorectal cancer patients, while pro-cancer bacteria such as Fusobacterium nucleatum can mediate chemotherapy resistance. The results of this study are consistent with the above conclusions, indicating that gut microbiota status can be used as a potential indicator for predicting chemotherapy efficacy and provide a reference for targeted gut microbiota intervention strategies.

This study has certain limitations: (1) Single-center retrospective design with limited sample size, which is prone to selection bias; (2) Gut microbiota metabolites were not detected, making it difficult to further analyze the internal pathway of the correlation between microbiota and prognosis; and (3) Lack of prospective intervention studies, unable to verify the improvement effect of targeted gut microbiota intervention on patient prognosis.

CONCLUSION

Abnormal gut microbiota is associated with adverse clinical and pathological features, reduced gut microbiota diversity, enrichment of oncogenic bacteria, metabolic disorders, poor long-term prognosis, and reduced response to first-line chemotherapy in patients with aCRC, and serves as an independent adverse prognostic factor affecting OS. This study provides novel biomarkers for prognostic risk stratification in aCRC and offers clinical evidence for intervention therapies targeting the gut microbiota.

ACKNOWLEDGMENTS

We sincerely thank all patients who participated in this study for their support and cooperation. We also thank the laboratory technicians for their technical support in microbiota sequencing.