Impact of gut microbiota on atypical endometrial hyperplasia and endometrial cancer: a comprehensive analysis of microbial composition and metabolomic profiling

Abstract

Background

Endometrial cancer (EC) is one of the most common malignant tumors in women, and in recent years, the role of gut microbiota in tumorigenesis has gradually gained attention. Previous studies have shown that the gut microbiome is closely related to the occurrence of various cancers, but the specific mechanisms through which gut microbiota contribute to the development of endometrial cancer (EC) remain unclear. This study aims to analyze the gut microbiome characteristics of atypical endometrial hyperplasia (AEH) and EC, and to explore key gut microbial species and metabolites, providing evidence for the etiological research and early screening of EC and AEH.

Methods

This study selected 24 AEH or EC patients from the Gynecology Department of Gansu Provincial Maternity and Child Care Hospital between February 2023 and October 2023. The patients were divided into the AEH group (n=7) and the EC group (n=17), with 24 healthy women selected as a control group. Fecal and serum samples were collected, and 16S rRNA gene sequencing was performed using the Illumina MiSeq platform. Serum metabolomics analysis was conducted using LC-MS technology. Spearman correlation analysis was used to explore the associations between gut microbiota and metabolites, and potential gut microbial biomarkers were evaluated using ROC analysis.

Results

The study found that as AEH progressed to EC, significant changes occurred in the composition of the gut microbiota, particularly in Klebsiella, whose abundance increased from 0.264% in the control group to 0.809% in the AEH group and 6.092% in the EC group, with significant differences (P<0.001, FDR=0.026). LEfSe analysis identified Megamonas, Klebsiella, Escherichia, and Akkermansia as potential biomarkers. ROC analysis showed that the AUCs of Megamonas/Klebsiella for EC were 0.864/0.838, and the AUCs of Escherichia/Akkermansia for AEH were 0.744/0.920. Metabolomics analysis revealed significant enrichment of glycerophospholipid metabolism in both the EC and AEH groups, with significant differences in lipid metabolism in the EC group. Correlation analysis indicated significant positive correlations between Enterococcus and hypoxanthine, inosine in the EC group (r=0.686, 0.637, P<0.05).

Conclusion

This study reveals the dynamic changes in the gut microbiome during the development of endometrial lesions, especially the increasing abundance of Klebsiella as AEH progresses to EC, suggesting that it may play a key role in the occurrence and progression of EC. Furthermore, significant changes in lipid metabolism further support the role of gut microbiota in regulating lipid metabolism in EC pathogenesis. This study provides new insights into the role of gut microbiota in endometrial cancer and offers a theoretical basis for early diagnosis and personalized treatment strategies based on gut microbiota.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12866-025-04568-2.

Introduction

Endometrial cancer (EC) is one of the three most common malignancies of the female reproductive system. According to the American Cancer Society, approximately 66,200 new cases and 13,030 deaths from EC are expected in the United States in 2023, with a 1% annual increase in mortality rate [1–3]. In China, new EC cases account for 23.7% of the global total, and the mortality rate represents 30.2% [4], making it a significant public health issue. Atypical endometrial hyperplasia (AEH) is a precursor lesion to endometrioid adenocarcinoma (Bokhman type I), with similar risk factors shared between the two [5, 6]. Therefore, identifying risk factors and conducting early screening and diagnosis is crucial.

The gut microbiome is one of the most complex ecosystems in the human body, closely linked to infections, autoimmune diseases, and metabolic disorders [7]. With the advent of high-throughput sequencing and metagenomics technologies, increasing evidence has pointed to the potential role of the gut microbiome in tumor initiation and progression. Pathogenic bacteria in the gut can activate specific pathways, leading to cell cycle disruption and DNA damage, thereby promoting tumorigenesis [8–10]. Moreover, gut microbes can modulate immune responses and the production of inflammatory factors through metabolites such as short-chain fatty acids, thus promoting tumor progression [11, 12].

Identifying tumor-specific gut microbiota helps to understand the biological mechanisms of tumors and provides potential targets for early screening. In the field of gynecological oncology, research has primarily focused on differences in microbial distribution. For example, cervical cancer patients show increased α-diversity of gut microbiota, with a higher abundance of specific microbial species closely related to cervical cancer risk [13, 14]. However, there has been less research on precancerous lesions (such as cervical intraepithelial neoplasia, AEH, etc.), which limits the application of gut microbiota in early screening of gynecological cancers. In contrast, studies on gastrointestinal cancers have shown that changes in the gut microbiota can occur as early as the precancerous stage, suggesting that the dynamic changes in gut microbiota may span the entire cancer development process [15]. Therefore, investigating the role of gut microbiota in gynecological tumor precursors may offer new insights into early diagnosis and precise screening.

Although research has revealed the regulatory role of gut microbiota in various metabolic disorders, the specific mechanisms by which gut microbiota contribute to EC initiation and development remain underexplored. In recent years, more studies have focused on the relationship between gut microbiota and EC, with increasing emphasis on mechanisms and clinical applications. Initial studies found that β-glucuronidase in gut microbiota could regulate the estrogen-gut microbiome axis, potentially leading to estrogen imbalance and increasing the risk of EC [16]. Subsequently, researchers discovered that the gut metabolite urolithin A could influence the depolymerization and migration of actin filaments in endometrial cancer cells [17]. A 2022 study first indicated that the genus Succinivibrio promotes EC development directly by affecting fatty acid metabolism [18]. These findings reveal the regulatory pathways of gut microbiota-host metabolites in EC metabolic reprogramming. However, there is still a lack of in-depth research on the changes in gut microbiota during the progression from AEH to EC. Therefore, to further investigate the composition of the gut microbiome in patients with endometrial lesions and its relationship with disease onset and progression, this study examines EC, AEH, and healthy populations using 16 S rRNA sequencing and untargeted metabolomics analysis. It compares the differences in gut microbiota abundance and diversity between groups, explores serum metabolites related to changes in microbiota, and investigates the characteristic changes of the gut microbiome in the process of endometrial lesions, providing new perspectives and directions for EC etiological research and the optimization of precise screening strategies.

Materials and methods

Inclusion and exclusion criteria of participants and ethical considerations

This study was conducted at Gansu Provincial Maternal and Child Health Care Hospital from February to October 2023. A total of 82 participants were recruited, and after strict inclusion and exclusion criteria were applied, a computer-assisted random sampling method was used to select the final sample. Among the candidates who met the inclusion criteria, stratification was performed based on pathological diagnoses (endometrial cancer, atypical endometrial hyperplasia, and healthy controls). The SPSS random number generator was used to independently select samples within each stratum. Ultimately, 48 female participants were included, comprising 17 endometrial cancer (EC) patients, 7 atypical endometrial hyperplasia (AEH) patients, and 24 healthy controls. All participants were Chinese women aged between 18 and 60 years.

The inclusion criteria were as follows: (1) participants aged between 18 and 60 years; (2) members of the case group diagnosed through histopathological examination and meeting at least one of the following conditions: (a) Endometrial pathology consistent with the World Health Organization (WHO) classification of atypical endometrial hyperplasia; (b) Pathologically confirmed endometrioid adenocarcinoma after standard hysterectomy, with diagnosis based on the FIGO 2009 surgical staging criteria; participants must have signed informed consent and be willing to comply with the study protocol.

The exclusion criteria included: (1) a history of estrogen-dependent diseases; (2) a history of other malignancies; (3) pregnant or breastfeeding women; (4) participants who had received chemotherapy, radiotherapy, or cellular immunotherapy prior to enrollment; (5) participants who had used antibiotics, probiotics, or other medications affecting gastrointestinal function within 3 months prior to enrollment; (6) participants who had received immunosuppressants or hormonal treatments within 3 months prior to enrollment; (7) a history of diarrhea or other gastrointestinal diseases within 3 months prior to enrollment; (8) vegetarians or individuals with special dietary habits.

This study was approved by the Ethics Committee of Gansu Provincial Maternity and Child-Care Hospital (Approval No. 2022-040-KYSL). All research procedures strictly adhered to ethical guidelines, and all participants voluntarily enrolled after being fully informed and signing a written informed consent form.

Sample collection

Fecal samples, each about the size of a soybean (approximately 1.5 to 3 g), were collected either prior to surgery for the patients or on the day of health check-ups for the control group. The samples were placed in 7-mL sterile collection tubes and immediately stored at -80 °C in an ultra-low temperature freezer. To ensure the stability of the microbiota, all participants were instructed to fast for at least 12 h prior to sampling to minimize the potential impact of food intake on the gut microbiota. Additionally, 2.5 mL of venous blood was collected from each participant and immediately transported to the laboratory. The blood was centrifuged at 4000×g for 5 min at room temperature to separate the serum. Subsequently, 1 mL of the separated serum supernatant was aliquoted into two cryovials, labeled with identification numbers, and rapidly stored at -80 °C in an ultra-low temperature freezer.

PCR amplification and sequencing methods

PCR amplification was carried out using specific primers 338 F (ACTCCTACGGGAGGCAGCAG) and 806R (GGACTACHVGGGTATCTAAT). The reaction mixture had a total volume of 20 µL, containing 30 ng of quality-verified genomic DNA and fusion primers. The PCR amplification conditions were as follows: initial denaturation at 98 °C for 1 min, followed by 30 cycles consisting of denaturation at 98 °C for 10 s, annealing at 50 °C for 30 s, and extension at 72 °C for 60 s; a final extension step was performed at 72 °C for 10 min. Purified amplification products were then subjected to Agencourt AMPure XP magnetic bead purification, dissolved in Elution Buffer, and prepared for library construction. The library fragment size and concentration were assessed using the Agilent 2100 Bioanalyzer to ensure quality. After confirming the library quality, sequencing was performed on the Illumina MiSeq platform. Filtered sequencing data were then used for subsequent analysis.

Bioinformatics analysis

Data filtering and tag linking

The raw sequencing data were initially filtered. A sliding window strategy with a 25-bp window was applied. If the average quality score within the window was below 20, both the window and its subsequent sequences were discarded. If the trimmed read length was < 75% of the original, the read was discarded. Adapter contamination was removed by setting the overlap region between adapters and reads to 15 bp, allowing for 3 mismatches. Reads containing N bases and low-complexity reads (defined as sequences with ≥ 10 consecutive identical bases) were also removed. Finally, samples were identified based on barcode and primer sequences, with no mismatches allowed between the barcode and sequencing reads.

Sequence assembly was conducted with FLASH (Fast Length Adjustment of Short Reads, v1.2.11), which utilized overlapping relationships to assemble paired-end reads into a single sequence, yielding high-variable region tags. The assembly conditions included: (1) a minimum matching length of 15 bp; (2) an overlap region mismatch rate of 0.1 (for detailed data, see the appendix).

OTU clustering analysis

USEARCH software (v7.0.1090) was employed to cluster the assembled tags into operational taxonomic units (OTUs). Initially, UPARSE was used to perform clustering at 97% similarity, resulting in the representative sequences for each OTU. Subsequently, UCHIME (v4.2.40) was employed to detect and remove chimeras that may have been generated during the PCR amplification process. For 16 S and ITS sequences, chimera removal was achieved by aligning the sequences with established chimera databases. The 16 S chimera database used was the gold database (v20110519), while the ITS chimera database used was UNITE (v20140703). Depending on the sequencing region, comparisons were performed against the full-length ITS, ITS1, or ITS2 regions. For 18 S sequences, a De novo method was used to remove chimeras. Lastly, the usearch_global method was applied to align all tags with the OTU representative sequences, generating an OTU abundance table for each sample (see appendix).

Microbial community diversity analysis

Microbial community diversity in the gut was analyzed using QIIME2 (v2023.2) to compute the Shannon index, Chao index, and observed OTUs, which were used to assess alpha diversity. Beta diversity analysis was conducted using the Bray-Curtis distance matrix and weighted UniFrac distance. Principal coordinates analysis (PCoA) was utilized to visualize the differences in community structure among groups. All analyses were performed in R (v4.4.1) with the vegan package (v2.6-4), and a significance threshold was set at P < 0.05.

Microbial community differences and linear discriminant analysis

DESeq2 (v1.12.4) was employed to analyze the microbiota’s relative abundance data at the phylum, family, and genus levels. A negative binomial distribution model was used to detect differentially abundant taxa between groups, using the criteria of FDR-corrected P < 0.05 and |log2FoldChange| > 2. Additionally, LEfSe analysis (LDA score > 3.0) was conducted to identify microbial biomarkers that significantly contribute to the differences between groups.

Metabolomics analysis

Serum sample extraction

Metabolites in serum samples were analyzed using a high-resolution mass spectrometer in both positive and negative ion modes. Peaks were detected using XCMS software, and preliminary identification was performed based on the mass-to-charge ratio (m/z) and retention time. Subsequently, metaX software was employed to match the detected metabolites against the HMDB and KEGG databases, providing primary identification results. To improve accuracy, a secondary mass spectrometry library was used to compare the sample data, yielding high-confidence metabolite identification results. For data processing, quality control was applied to remove low-quality peaks (with QC sample missing more than 50% or actual samples missing more than 80%). Missing values were imputed using the K-Nearest Neighbors (KNN) method, and normalization was performed using the Probabilistic Quotient Normalization (PQN) method. Finally, metabolite identification and relative quantification results were obtained.

Serum metabolite detection

The samples were sequentially arranged in the ultra-high pressure liquid chromatography system, and pre-separation was carried out using an ACQUITY UPLC BEH T3 column. The chromatographic conditions were set as follows: column temperature at 50 °C, flow rate at 0.3 mL/min, with mobile phase A being a 0.1% formic acid aqueous solution, and mobile phase B being a 0.1% formic acid-acetonitrile solution. The gradient elution program is detailed in the appendix.

Mass spectrometry data were collected using a Q-Exactive high-resolution mass spectrometer, operating alternately in positive and negative ion modes. The ion source parameters were as follows: sheath gas pressure at 10 psi, auxiliary gas pressure at 40 psi, ion transfer tube temperature at 350 °C, and spray voltage at + 3800 V (positive ion mode) / -3100 V (negative ion mode). Data-dependent acquisition (DDA) mode was applied, with a full scan range of 70-1050 Da (resolution 70,000 @ m/z 200), automatically selecting the top 3 precursor ions with signal intensity > 1 × 10⁵ for fragmentation (resolution 17,500 @ m/z 200). QC samples were introduced after every 10 samples to monitor system stability, and systematic errors were corrected by normalizing the inter-sample variation using the QC samples.

Metabolomics data processing and analysis

The data processing workflow is as follows: First, metabolites with missing values exceeding 50% in QC samples or 80% in actual samples are excluded during quality control. Subsequently, missing values are imputed using the K-Nearest Neighbors (KNN) method. Data normalization is performed using Probabilistic Quotient Normalization (PQN), referencing the median peak area ratio from the QC samples. The identified metabolites were annotated using the KEGG database, and significant differential metabolites were selected using a PLS-DA model. The selection criteria were: fold change > 1.5 or < 1/1.5, P < 0.05, and VIP > 1. Finally, KEGG pathway enrichment analysis was conducted on the differential metabolites, with P < 0.05 considered significant.

Statistical analysis

Statistical analyses were conducted using SPSS 26.0 software. The normality of continuous variables was assessed using the Shapiro-Wilk test. Normally distributed data were presented as mean ± standard deviation (Mean ± SD), and group comparisons were performed using one-way analysis of variance (ANOVA), with multiple comparisons corrected using the Tukey-Kramer method. Non-normally distributed data were presented as median (interquartile range, IQR), and group differences were analyzed using the Kruskal-Wallis H test. Categorical variables were presented as frequency (percentage), and group comparisons were made using Pearson’s chi-square test. Spearman rank correlation analysis was conducted to assess the association between specific gut microbiota and serum metabolites. Additionally, five-fold cross-validation was used to assess the consistency of Spearman’s correlation coefficient by randomly splitting the dataset and calculating the coefficient for each validation, resulting in the average and standard deviation. Receiver Operating Characteristic (ROC) analysis was applied to evaluate the predictive power of potential gut microbial biomarkers for endometrial lesions, and the area under the curve (AUC) was calculated to assess discriminative performance. All tests were two-tailed, with a significance threshold set at P < 0.05.

Results

Baseline characteristics analysis

A total of 48 participants were enrolled, including 17 patients with endometrioid adenocarcinoma (EC group), 7 patients with atypical endometrial hyperplasia (AEH group), and 24 healthy controls (N group). The baseline characteristics are detailed in Table 1.

Table 1.

Baseline characteristics of study groups. Continuous variables are presented as mean ± SD or median (IQR) as appropriate. Categorical variables are shown as number (percentage). *P < 0.05 was considered statistically significant

Category	Variables	EC Group (n = 17)	AEH Group (n = 7)	N Group (n = 24)	P value
Demographic Characteristics
	Age (years) 1	50.52 ± 7.26	47.28 ± 4.42	33.87 ± 7.28	<0.001*
	Menopausal status2	10(58.8%)	0(0.0%)	0(0.0%)	0.009*
	Gravidity1	3(2–4)	2(1–3)	2(1–2)	0.005*
	Parity1	2(1–3)	1(1–2)	1(0–1)	0.017*
Metabolic Characteristics
	BMI(kg/m2)1	27.88 ± 5.55	25.30 ± 2.97	21.31 ± 2.81	<0.001*
	Obesity2	7(41.2%)	1(14.3%)	0(0.0%)	<0.001*
	Waist circumference (cm)1	95.32 ± 24.38	79.14 ± 5.61	74.27 ± 6.51	<0.001*
	Hip circumference (cm)1	96.05 ± 18.67	93.57 ± 3.99	92.72 ± 5.87	0.680
	Diabetes mellitus2	4(23.5%)	2(28.5%)	0(0.0%)	0.800
	Fasting blood glucose (mmol/L)1	5.48 ± 1.28	5.15 ± 0.91	5.01 ± 0.43	0.652
	Hypertension2	8(47.0%)	1(14.2%)	0(0.0%)	0.140
	Systolic blood pressure (mmHg) 1	134.58 ± 22.45	119.85 ± 9.09	108.62 ± 11.95	<0.001*
	Diastolic blood pressure (mmHg) 1	86.17 ± 12.17	81.85 ± 10.96	66.87 ± 8.08	<0.001*
Tumor Characteristics
	2009 FIGO stage2
	I	15(88.2%)	-	-
	II	0(0.0%)	-	-
	III	1(11.7%)	-	-
	IV	0(0.0%)	-	-
	Histological grade2
	G1	13(76.4%)	-	-
	G2	4(23.5%)	-	-
	G3	0(0.0%)	-	-
	Tumor markers1
	CA125(u/ml)	21.45(11.29–32.01)	19.15(16.81–30.51)	-	0.968
	HE4(pmol/L)	64.81(52.17−131.32)	52.45(44.06–59.81)	-	0.105

There were significant differences in menopausal status between the three groups (P = 0.009). The proportion of postmenopausal women in the EC group was 58.8% (10/17), significantly higher than that in the AEH (0/7) and N groups (0/24). There were also statistically significant differences in gravidity (P = 0.008) and parity (P = 0.017) among the groups.

In terms of metabolic characteristics, the EC group exhibited significant metabolic disturbances, with both BMI and waist circumference significantly higher than those of the AEH and control groups (P < 0.001). In the EC group, 41.2% (7/17) of patients were classified as obese (BMI > 28 kg/m²), a significantly higher proportion compared to the AEH and control groups (P < 0.001). Blood pressure measurements indicated that both systolic (134.58 ± 22.45 mmHg) and diastolic (86.17 ± 12.17 mmHg) pressures were significantly higher in the EC group compared to the AEH and control groups (P < 0.001).

Regarding tumor characteristics, according to the 2009 FIGO staging criteria, 15 cases (88.2%) in the EC group were classified as Stage I, while 2 cases (11.7%) were classified as Stage III. The pathological grade was predominantly well-differentiated (G1), accounting for 76.4% (13/17), while moderately differentiated (G2) accounted for 23.5% (4/17). Serum tumor marker analysis showed no significant differences in CA125 (P = 0.968) and HE4 (P = 0.105) levels between the EC and AEH groups.

OTU clustering and microbial diversity analysis

Fecal samples from 48 participants were subjected to paired-end sequencing to construct a short fragment library. After quality control, an average of 76,419 reads per sample was obtained, with 47,975 valid data points on average, yielding a quality control success rate of 62.77%. The sequencing data were clustered at a 97% similarity threshold, resulting in 470 distinct OTUs. The Venn diagram (Fig. 1) illustrates that the N, AEH, and EC groups contained 1,182, 853, and 1,142 OTUs, respectively, with 766 OTUs shared by all three groups. Compared to the AEH and EC groups, the N group had a higher number of unique OTUs (8, 81, and 104, respectively).

Fig. 1.

Venn Diagram of OTUs Across Groups. The Venn diagram shows the distribution of OTUs among the N, AEH, and EC groups. There are 766 shared OTUs across all groups, with 104 unique to N, 8 unique to AEH, and 81 unique to EC

Based on α-diversity indices (Table 2; Fig. 2), no significant differences were observed in species richness and diversity levels among the N, AEH, and EC groups (P > 0.05). The Sobs index for the EC group was 420.12, and the Chao index was 635.16, both slightly higher than those of the N and AEH groups. The higher Sobs index in the EC group indicates a wider distribution of gut microbiota species, reflecting the increased complexity of the gut microbiota in the cancer state [19]. The higher Chao index suggests the presence of more undetected species in the gut microbiota of the EC group, indicating enhanced heterogeneity.

Table 2.

Analysis of α-diversity indices among the three groups

	Sobs	Chao	Ace	Shannon	Simpson	Coverage
N Group	413.0416667	604.6197133	733.2931503	2.737661417	0.189339833	0.996856125
AEH Group	392.2039935	598.6099481	788.90468	2.536753857	0.197486714	0.996947286
EC Group	420.1176471	635.1563734	766.8038203	2.864007412	0.169637059	0.996693412
P value	0.45372	0.19475	0.27607	0.58278	0.49943	-

Fig. 2.

Box plot of intergroup differences in α-diversity. (A) Simpson index, (B) Simpson index, (C) Sobs, (D) Chao, (E) Ace, and (F) Coverage. The five lines from bottom to top represent the minimum, first quartile (Q1), median, third quartile (Q3), and maximum, respectively.

The AEH group exhibited a lower Shannon index and a relatively higher Simpson index, indicating the possible dominance of specific microbial taxa, which could lead to reduced community evenness. Similar features have also been observed in the gut microbiota of patients with colorectal polyps and colorectal cancer [20, 21].

Further visualization of the differences in gut microbiota among all study participants (Fig. 3 and 4) using unweighted PCoA analysis showed significant separation between the N, AEH, and EC groups (P < 0.001), whereas the weighted PCoA analysis did not reveal significant separation among the three groups (P = 0.05), suggesting that the abundance distribution of high-abundance taxa remained relatively stable across the groups. A comprehensive analysis of β-diversity data indicated that the microbial communities in the N, AEH, and EC groups had distinguishable structural features.

Fig. 3.

Principal coordinates analysis (PCoA) of gut microbiota based on unweighted Bray-Curtis distance. The x- and y-axis scales represent the projection coordinates of the sample points on the two-dimensional plane

Fig. 4.

Principal coordinates analysis (PCoA) of gut microbiota based on weighted Bray-Curtis distance

Gut microbiota composition analysis

At the phylum level (Fig. 5), Bacillota and Bacteroidota were the core dominant phyla in the gut microbiota of the N, AEH, and EC groups. The relative abundances of Bacillota in the N, AEH, and EC groups were 48.14%, 31.87%, and 45.24%, respectively, while the relative abundances of Bacteroidota were 40.55%, 43.12%, and 33.48%, respectively. However, the Bacillota/Bacteroidota ratio was skewed in the AEH and EC groups: the Bacillota/Bacteroidota ratio in the EC group (B/B = 1.35) was significantly higher than in the AEH group (B/B = 0.74). A higher B/B ratio is associated with inflammation or abnormalities in metabolic pathways such as tryptophan and lipid metabolism [22, 23], and may play a role in the progression of EC. Additionally, Pseudomonadota abundance was higher in the AEH (18.5%) and EC (17.1%) groups than in the N group (4.2%), but the difference between the groups was not statistically significant (P > 0.05).

Fig. 5.

Bar plot of gut microbiota composition at the phylum level. Unannotated taxa at this classification level and species with relative abundances below 0.5% in the samples were grouped into "Others"

At the family level (Fig. 6), Bacteroidaceae was the dominant family in all three groups (N group: 21.32%, AEH group: 40.94%, EC group: 23.99%). The abundance of Enterobacteriaceae was elevated 6.8-fold in the AEH group (17.26%) and 5.4-fold in the EC group (13.59%) compared to the N group (2.52%). Selenomonadaceae was significantly enriched in the EC group (H = 8.12, P = 0.003). Selenomonadaceae is associated with inflammatory cytokines, such as increased levels of IL-1β and TNF-α [24]. In addition, Akkermansiaceae was specifically elevated in the AEH group (H = 3.80, P < 0.001). Akkermansiaceae can enhance intestinal mucosal barrier function, alleviate inflammation, and help maintain intestinal immune homeostasis [25, 26], suggesting that these bacteria play corresponding regulatory roles at different stages of endometrial lesions.

Fig. 6.

Bar plot of gut microbiota composition at the family level

At the genus level (Fig. 7), 65 genera with significantly different abundances were identified between the N, AEH, and EC groups (P < 0.05, FDR < 0.05, detailed information on the genera can be found in the appendix). Although Phocaeicola was the most abundant genus in all three groups (N group: 13.72% ± 5.21%, AEH group: 31.84% ± 12.37%, EC group: 17.27% ± 8.43%), no significant differences in its abundance were observed between the groups (P = 0.343). The AEH group exhibited specific enrichment of Escherichia (H = 15.36, P = 0.048), Klebsiella (H = 0.80, P < 0.001), Akkermansia (H = 3.60, P = 0.003), and Hallella (H = 0.43, P < 0.001), whereas the EC group showed significant enrichment of Megamonas (H = 7.48, P < 0.001) and Succinivibrio (H = 0.65, P = 0.003). The relative abundance of Escherichia in the AEH group (15.36% ± 4.82%) was significantly higher than in the N group (1.94% ± 0.67%) and EC group (5.91% ± 2.15%). The abundance of Megamonas in the EC group (7.48% ± 3.21%) was 7.0-fold and 93.5-fold higher than in the N group (0.94% ± 0.35%) and AEH group (0.08% ± 0.02%), respectively. The AEH group was characterized by the enrichment of genera such as Escherichia and Klebsiella, whereas the EC group exhibited a distinctive enrichment of Megamonas. Further LDA analysis is needed to identify gut microbial biomarkers associated with the progression of endometrial lesions.

Fig. 7.

Genus-level differential species analysis based on the Kruskal-Wallis test. In each subplot, the left column shows the mean relative abundance of the genus in each group; the middle column displays the confidence intervals obtained from the statistical testing process; and the right column presents the P value and false discovery rate (FDR) value

Identification of gut bacterial genera positively correlated with the progression of endometrial lesions

Based on the analysis of gut microbiota abundance differences among the N, AEH, and EC groups, four bacterial genera were identified as positively correlated with disease progression: Rothia, Klebsiella, Butyrivibrio, and Luoshenia. The abundance of these genera exhibited a significant increasing trend with disease progression, as shown in Table 3.

Table 3.

Relative abundance of gut bacterial genera positively correlated with the progression of endometrial lesions across groups

Genus	N(mean ± SD)	AEH(mean ± SD)	EC(mean ± SD)	P value	FDR
Klebsiella	0.264 ± 0.775	0.809 ± 1.887	6.092 ± 15.443	< 0.001	0.0260
Rothia	0.024 ± 0.040	0.071 ± 0.015	0.086 ± 0.033	< 0.001	0.0005
Butyrivibrio	0.0	0.0005 ± 0.001	0.003 ± 0.004	< 0.001	0.0170
Luoshenia	0.0001 ± 0.0008	0.001 ± 0.004	0.027 ± 0.081	0.001	0.0488

The average relative abundance of Rothia increased from 0.024% in the N group to 0.071% in the AEH group and 0.086% in the EC group (P < 0.001, FDR < 0.001). Butyrivibrio was undetectable in the N group, with an abundance of 0.0005% in the AEH group and 0.003% in the EC group (P < 0.001, FDR = 0.017). The average relative abundance of Luoshenia increased from 0.0001% in the N group to 0.001% in the AEH group and reached 0.027% in the EC group (P = 0.001, FDR = 0.048). Klebsiella showed a sharp increase in abundance, from 0.264% in the N group to 0.809% in the AEH group and 6.092% in the EC group, representing a 23.1-fold increase compared to the N group (P < 0.001, FDR = 0.026). Notably, the relative abundance of Klebsiella in the EC group exceeded 6% of the total microbiota, significantly higher than that of other genera in the same group. The gradient changes in the abundance of these genera suggest a close association between the dynamic shifts in the gut microbiota and the progression of endometrial lesions, providing a foundation for further exploration of the underlying mechanisms.

LEfSe analysis of gut microbiota

LEfSe analysis, combining statistical significance with biological relevance, revealed hierarchical differences in gut microbiota at the genus level between the AEH and EC groups, as shown in Fig. 8 and 9. The EC group exhibited significant enrichment of Megamonas (LDA = 3.90, P = 0.0093), Klebsiella (LDA = 3.70, P = 0.0030), and Succinivibrio (LDA = 3.08, P = 0.0003). The characteristic genera of the AEH group were Escherichia (LDA = 3.67, P = 0.0057) and Akkermansia (LDA = 3.27, P = 0.0025).

Fig. 8.

LEfSe LDA bar plot. The plot displays significantly different taxa (biomarkers) with LDA scores greater than the predefined threshold, which is set to 3.0 by default. The length of each bar represents the LDA score

Fig. 9.

LEfSe cladogram. Nodes in different colors represent microbial taxa that play an important role in the corresponding group. Each colored node indicates a biomarker, and the legend in the upper right corner shows the names of these biomarkers. Yellow nodes indicate taxa that are not significantly associated with any group. From the center to the outermost ring, the circles represent taxa at the phylum, class, order, family, and genus levels, respectively

In the N group, the abundance of Bacteroides (LDA = 2.54, P = 0.0207) and Lactobacillus (LDA = 2.67, P = 0.0347) decreased by 2.5–3.1 times in the AEH and EC groups. Bacteroides, a key fiber-degrading bacterium in the gut, plays a vital role in the production of short-chain fatty acids (SCFAs), which are essential for maintaining intestinal barrier integrity and immune function. A reduction in its abundance may indicate a decline in intestinal metabolic function, further affecting immune tolerance in the gut [27]. Lactobacillus, an important probiotic, regulates intestinal pH by producing lactic acid and modulates the gut immune system. Its reduced abundance may impair intestinal barrier function, increasing the risk of dysbiosis [28]. The decreased abundance of certain gut microbiota may also lead to a reduction in immune tolerance and damage to the intestinal barrier, thereby promoting inflammatory responses within the gut microenvironment and creating favorable conditions for tumor initiation and progression. These results, based on effect size and feature extraction from LEfSe analysis, identify Megamonas/ Klebsiella and Escherichia / Akkermansia as potential gut microbiota biomarkers for EC and AEH, respectively.

Differential metabolites and pathway enrichment analysis in the EC and AEH groups

To investigate the serum metabolic characteristics in patients with endometrial lesions, this study applied Liquid Chromatograph-Mass Spectrometry (LC-MS)-based untargeted metabolomics to systematically analyze the serum metabolomes of the AEH, EC, and N groups. Based on selection thresholds of VIP > 1, P < 0.05, and |log2FC| ≥1, a total of 3,917 differential metabolites were identified in the AEH group (1,304 upregulated, 2,613 downregulated), and 4,390 differential metabolites in the EC group (1,493 upregulated, 2,897 downregulated). All metabolites were subjected to quantitative quality control (CV < 50%). Based on the cross-validation parameters (as shown in Fig. 10), the PLS-DA model demonstrated that the principal components t1 (12.50%) and t2 (6.50%) significantly separated the metabolic profiles of the groups. The model parameters R²Y = 0.896 and Q²=0.654 indicate a high level of model fitting and predictive power, respectively, with significant inter-group differences (P = 0.01). The PLS-DA model further confirmed the significant separation of the metabolic profiles among the three groups, suggesting that the differential metabolites identified in this study are highly reliable.

Fig. 10.

PLS-DA Score Plot (left) and Model Validation (right). The PLS-DA score plot shows the first principal component on the x-axis and the second principal component on the y-axis. The PLS-DA model validation uses a permutation test to assess whether the model suffers from overfitting, with 100 permutations. The x-axis represents the correlation between the permuted Q²and R²values and the original Q²and R²values, while the y-axis shows the R²and Q²values obtained from each permutation

As shown in Fig. 11, the secondary metabolite classification revealed that the differential metabolites in the serum of the AEH and EC groups were primarily categorized under lipids and lipid-like molecules. Further screening of metabolites with high expression levels, sorted by descending VIP value, showed that the EC group was characterized by significantly upregulated metabolites, including hypoxanthine and inosine (Fig. 12), while the AEH group exhibited significantly upregulated metabolites such as oleamide, stearamide, and LysoPE 20:5 (Fig. 13). KEGG pathway enrichment analysis revealed that both the glycerophospholipid metabolism and purine metabolism pathways were significantly enriched in the EC and AEH groups, with both groups’ core metabolic pathways primarily focused on lipid metabolism.

Fig. 11.

Number of differential metabolites among the three groups. The x-axis represents the number of differential metabolites, and the metabolite categories are based on secondary identification

Fig. 12.

Volcano plot of differential metabolites between the EC group and the N group. The x-axis represents the log₂ fold change of average intensities between the two phenotypes, while the y-axis shows the statistical significance (–log₁₀ P value). A higher y-axis value indicates greater statistical significance

Fig. 13.

Volcano plot of differential metabolites between the AEH group and the N group

As shown in Fig. 14, further analysis of the glycerophospholipid metabolism pathway reveals that glycerol-3-phosphate is catalyzed by LPAAT to form 1-acylglycerol-3-phosphate, which is then converted into phosphatidylcholine and phosphatidylethanolamine. These are further synthesized under the regulation of CDP-choline and LPCAT. Phosphatidylserine is synthesized by LPSAT and interconverts with other lipids through the CDP-lipid pathway. This metabolic pathway regulates cell membrane biosynthesis, signal transduction, and energy metabolism, thus promoting cancer cell proliferation, migration, and immune evasion, which in turn facilitate the initiation and progression of EC [29, 30].

Fig. 14.

Glycerophospholipid metabolism pathway.This pathway illustrates the key steps in glycerophospholipid metabolism, including the synthesis and interconversion of phosphatidylcholine, phosphatidylethanolamine, and phosphatidylserine. Key enzymes such as LPAAT, LPCAT, LPSAT, and others are involved in these processes. The pathway is sourced from the KEGG database (https://www.kegg.jp/)

Bacterial-Metabolite correlation analysis in the EC and AEH groups

Based on the 18 and 11 significantly different genera identified by LEfSe analysis in the EC and AEH groups, respectively, and 20 significantly different serum metabolites, Spearman’s rank correlation analysis was performed to systematically assess the associations between gut microbiota and metabolites. The results showed significant positive correlations between Enterococcus and hypoxanthine (r = 0.686, P < 0.05) and inosine (r = 0.637, P < 0.05) in the EC group, observed under both positive and negative ion modes (Fig. 15). In the AEH group, Flavonifractor showed positive correlations with oleamide (r = 0.857, P < 0.05) and linoleamide (r = 0.857, P < 0.05), and a negative correlation with dexrazoxane (r = -0.964, P < 0.05) (Fig. 16).

Fig. 15.

Spearman correlation heatmap between differential bacteria and metabolites in the EC group. The x-axis represents significantly different serum metabolites, and the y-axis represents differential bacterial genera identified by LEfSe. The color gradient indicates the Spearman correlation coefficient (r)

Fig. 16.

Spearman correlation heatmap between differential bacteria and metabolites in the AEH group

Additionally, the five-fold cross-validation of Spearman’s correlation analysis revealed an average Spearman correlation coefficient of 0.696 (SD = 0.080) between Enterococcus and hypoxanthine, and 0.683 (SD = 0.073) between Enterococcus and inosine, indicating a high level of consistency in the correlation between Enterococcus, hypoxanthine, and inosine across samples.Hypoxanthine, a naturally occurring purine derivative, facilitates nucleic acid synthesis through the nucleotide salvage pathway. Accumulating evidence indicates that hypoxanthine drives cancer cell metastasis via PI3K/AKT pathway activation, concomitant upregulation of glycerolipid/fatty acid metabolism, and impairment of glycolytic ATP generation [31].Hypoxanthine is a direct catabolite of adenosine. Mounting evidence reveals that adenosine-to-inosine (A-to-I) RNA editing—a widespread epigenetic modification affecting thousands of human transcripts—is globally downregulated, particularly at Alu repetitive elements in tumor tissues. Critically, this dysregulation of A-to-I editing constitutes an emergent layer of epigenetic control implicated in oncogenesis and malignant progression [32, 33].

Further analysis revealed that among the metabolites significantly associated with gut microbiota, 8 were lipids and lipid-like molecules, and 4 were organic acids and their derivatives. This study preliminarily explored the overall association between gut microbiota and metabolites using untargeted metabolomics, covering a broad spectrum of relationships across different metabolite categories. Future studies will apply targeted metabolomics to focus on the correlation between lipids and lipid-like molecules, aiming to further elucidate their specific connections with gut microbiota.

Diagnostic performance evaluation of gut microbial biomarkers

In this study, two potential biomarkers from each group, identified through LEfSe analysis, were evaluated for their diagnostic performance in predicting EC and AEH, with ROC curves plotted to calculate the associated diagnostic indicators.

For predicting EC, Megamonas had an area under the curve (AUC) of 0.864, with sensitivity of 0.764, specificity of 0.875, and a Youden index of 0.640. Klebsiella had an AUC of 0.838, with sensitivity of 0.764, specificity of 0.813, and a Youden index of 0.577.

For predicting AEH, Escherichia had an AUC of 0.774, with sensitivity of 1.000, specificity of 0.458, and a Youden index of 0.458. Akkermansia had an AUC of 0.920, with sensitivity of 1.000, specificity of 0.833, and a Youden index of 0.833.

In summary, Megamonas/ Klebsiella and Escherichia/ Akkermansia can serve as gut microbial biomarkers for diagnosing EC and AEH, offering new perspectives for the early identification and precise classification of endometrial lesions (Figs. 17 and 18).

Fig. 17.

ROC curves of Megamonas and Klebsiella for predicting EC. The x-axis represents the false positive rate, and the y-axis represents the true positive rate. The solid line and dashed line represent the ROC curves of the two genera, and the gray diagonal represents the reference line

Fig. 18.

ROC curves of Escherichia and Akkermansia for predicting AEH

The x-axis represents the false positive rate, and the y-axis represents the true positive rate. The solid line and dashed line represent the ROC curves of the two genera, and the gray diagonal represents the reference line.

Discussion

Numerous studies have reported notable differences in the gut microbiota between patients with endometrial carcinoma (EC) and healthy individuals [18, 34–37]. When the intestinal microbiota becomes imbalanced, changes in its composition and structure affect the host’s metabolic activities and influence the tumor microenvironment (TME), promoting the onset and progression of malignancies. The potential mechanisms are as follows: (1) Gut microbiota triggers an inflammatory response, releasing inflammatory mediators, leading to immune imbalance, DNA damage, and increased mutation rates within the TME [38]; (2) Toxins and metabolites produced by gut microbiota regulate signaling pathways associated with malignant tumor progression, thereby affecting the host’s metabolic function and participating in the entire process of tumor metabolism [39]; (3) Gut microbiota affects systemic immunity and tumor development through mechanisms such as bacterial translocation and antigen cross-reactivity [40]. Energy metabolites and hormones, including lipids, estrogen, and insulin, which influence the development of endometrial cancer (EC), are extensively regulated and affected by the gut microbiota. When the intestinal microbiota is dysregulated, it leads to abnormal regulation of related metabolic axes and pathways, secretion of multiple inflammatory factors, alteration of proto-oncogene and tumor suppressor gene expression, and inducing irregular proliferation of endometrial glands and microvessels, thus increasing the risk of EC development.

However, the exact role of the gut microbiota in the pathogenesis of EC remains poorly understood. This study represents the first comprehensive investigation of alterations in the gut microbiota across various stages of endometrial adenocarcinoma, including AEH and EC. Using 16 S rRNA gene sequencing, we identified four specific genera (Rothia, Butyrivibrio, Klebsiella, and Luoshenia) that showed a progressive increase in abundance throughout the development of endometrial lesions. Notably, Klebsiella exhibited the most notable increase in abundance between the N, AEH, and EC groups. Furthermore, during the progression from AEH to EC, the Bacillota/Bacteroidota (B/B) ratio increased, accompanied by a notable rise in the relative abundance of Bacillota, particularly the genus Megamonas, which was significantly enriched in the gut microbiota of EC patients.

Microbial community analysis across the N, AEH, and EC groups revealed substantial enrichment of Megamonas and Succinivibrio in the EC group, whereas Akkermansia, Klebsiella, and Hallella were more abundant in the AEH group. LEfSe analysis further identified Megamonas, Klebsiella, Escherichia, and Akkermansia as potential biomarkers for endometrial lesions. ROC curve analysis validated the diagnostic potential of the Megamonas/ Klebsiella ratio for EC and the Escherichia/ Akkermansia ratio for AEH, providing new insights into early detection and precise classification of endometrial lesions.

Analysis of serum profiles in patients with endometrial lesions revealed that metabolites associated with lipid metabolism exhibited the most significant alterations. KEGG enrichment analysis of differentially expressed metabolites revealed marked changes in the glycerophospholipid metabolism pathway in both the EC and AEH groups. Correlation analysis further confirmed that lipid and lipid-related molecules were significantly influenced by, and regulated by, the disease-specific gut microbiota in endometrial lesions.

In conclusion, this study underscores the close relationship between gut microbiota and the pathogenesis of EC. It provides key insights into the alterations in specific microbiota and their metabolites during the progression of endometrial lesions, which are essential for a deeper understanding of the underlying pathological mechanisms.

Changes in gut microbiota during the development of endometrial lesions

In this study, we identified specific gut microbiota genera (Megamonas, Klebsiella, and Akkermansia) associated with endometrial lesions, along with a progressive increase in the abundance of four genera—Rothia, Butyrivibrio, Klebsiella, and Luoshenia—during the development of endometrial lesions. Among these, Klebsiella appears to play a pivotal role in the progression of endometrial cancer (EC). A detailed comparison of these endometrial lesion-specific genera with those identified in existing literature further elucidates the complex relationship between the gut microbiota and EC.

First, Megamonas exhibited a higher abundance in the intestines of EC patients, which may be closely linked to its role in regulating host metabolism and immune responses. Megamonas, a genus belonging to the phylum Firmicutes, is an anaerobic, non-motile, non-spore-forming Gram-negative bacterium commonly found in the gastrointestinal tract of both humans and animals [41–43]. Previous studies have shown that increased abundance of Megamonas is associated with several diseases, including obesity, cognitive impairments, non-alcoholic fatty liver disease, and systemic lupus erythematosus [44–47]. Recent research indicates that Megamonas promotes lipid absorption by degrading myo-inositol in the intestine, which may contribute to obesity through this mechanism.

Wu et al. [44] performed high-throughput metagenomic sequencing on fecal samples from 1,005 Chinese participants, profiling their microbiota and identifying a microbial cluster enriched in obesity, with Megamonas as a dominant genus. This cluster exacerbated high-fat diet (HFD)-induced obesity and metabolic disorders and demonstrated independent and cumulative effects on obesity risk, as reflected in the polygenic risk score (PRS) for BMI. Moreover, the study detected the myo-inositol degradation pathway (PWY-7237) in Megamonas. Myo-inositol plays a role in inhibiting the expression of fatty acid transport protein genes and fatty acid uptake, and Megamonas influences the expression of the myo-inositol degrading enzyme iolG, thus promoting lipid absorption in the intestine [44]. Therefore, while this study did not find a significant correlation between Megamonas and lipid metabolism in EC patients, it suggests that Megamonas and other EC-specific genera may indirectly affect host lipid metabolism by mediating intermediary pathways.

TONG et al. [48] analyzed the correlation between gut microbiota and clinical pathological features in ovarian cancer patients. The results showed that patients with shorter survival times had significantly higher levels of Megamonas (P < 0.001) in their gut, whereas patients with longer survival times exhibited a marked decrease in Klebsiella abundance. Klebsiella, a Gram-negative pathogen from the Proteobacteria phylum, is associated with various infections, including pneumonia and bacteremia [49]. Similarly, our study found Klebsiella to be highly enriched in the gut microbiota of EC patients. DONG et al. observed that in a lipopolysaccharide (LPS)-treated endometritis mouse model, the abundance of harmful bacteria such as Klebsiella increased in the gut, which disrupted energy metabolism and primary bile acid biosynthesis [50]. This disruption weakened the intestinal barrier function, allowing harmful bacteria and their toxins to enter the bloodstream, triggering signaling pathways and the release of inflammatory cytokines. These systemic and localized inflammatory responses impaired the normal physiological structure of the endometrium, promoting tumorigenesis, proliferation, and invasion. Thus, Klebsiella plays a key regulatory role in the “inflammation-cancer transformation model” during the development of EC. Notably, recent studies using metagenomics have confirmed the gut enrichment of Klebsiella in EC patients [35].

Based on LEfSe analysis, Akkermansia was identified as a potential biomarker for AEH. Although this genus was not highly enriched in the gut of EC patients, prior studies have reported similar findings. ZHAO et al. [18] found that Akkermansia muciniphila was increased in the gut of EC patients and was positively correlated with serum prealbumin (PAB), while negatively correlated with circulating threonine levels. Additionally, KEGG pathway enrichment analysis in this study highlighted the significant role of amino acids in the serum metabolism of endometrial lesions, which may be closely related to changes in Akkermansia abundance. Branched-chain amino acids (BCAAs), essential for human metabolism, are synthesized by both dietary intake and gut microbes such as Akkermansia [51].

When dysbiosis occurs, the regulatory function of BCAAs on insulin sensitivity decreases. Excess insulin, in combination with estrogen, exerts a synergistic effect that promotes the expression of EC hormone receptors. This synergy arises from the interactive signaling between estrogen and insulin pathways, leading to receptor activation [52]. Branched-chain amino acid transaminase 1 (BCAT1), an enzyme involved in BCAA metabolism, is expressed at higher levels in EC tissues compared to normal endometrium (P < 0.001) and AEH (P = 0.027). BCAT1 activates the mTORC1 pathway through BCAA production, stimulating EC cell proliferation [53]. In conclusion, these findings provide new insights into how gut microbiota influences the onset and progression of EC, and they offer theoretical support for the potential application of gut microbiota as biomarkers for endometrial lesions.

The age range of participants included in this study was 18 to 60 years. Although there were age differences among the study participants, existing literature suggests that the impact of age on the gut microbiota remains relatively stable between the ages of 3 and 70 years [54]. Typically, the composition and diversity of the gut microbiota in children at 3 years of age are similar to those in adults [55], while in individuals over 70 years of age, the gut microbiota composition tends to be affected by changes in digestive and nutrient absorption functions as well as reduced immune activity [56]. However, within the age range of this study, the variations in gut microbiota composition and diversity were relatively small, and the impact of age differences on the study results was minimal. Nevertheless, future studies should aim to control for age differences or adjust for potential confounders through stratification or statistical methods to minimize bias.

Impact of gut microbiota-related metabolites on endometrial lesions

Gut microbiota is closely linked to lipid metabolism. From a biomass perspective, major bacterial phyla, such as Bacteroidetes, Firmicutes, Verrucomicrobia, and Actinobacteria, are capable of producing lipid metabolites [57, 58]. The composition and abundance of the gut microbiota can influence the synthesis and breakdown of dietary and host lipids, potentially affecting cholesterol synthesis, lipid accumulation, adipocyte proliferation, and inflammatory responses, thereby influencing immune function and possibly contributing to tumorigenesis [59]. The relationship between gut microbiota and host lipid metabolism is intricate and multidimensional. On one hand, gut microbiota can finely regulate the expression levels of various metabolites, including short-chain fatty acids, bile acids, amino acids, and lipopolysaccharides (LPS) in plasma and target organs, significantly affecting the host’s metabolic homeostasis and signaling pathways [60, 61]. For instance, gut microbiota converts choline and L-carnitine into trimethylamine (TMA), which is further metabolized to trimethylamine N-oxide (TMAO), inducing lipid dysregulation and inflammatory processes that contribute to tumorigenesis [62]. On the other hand, membrane lipid molecules from common gut bacteria, such as Escherichia coli, Bacteroides, Alistipes, and Prevotella, can be recognized by the host’s innate immune system, activating immune sensors and triggering immune deficiencies and inflammatory responses [63, 64].

Gut microbes also contribute to lipid metabolism disturbances that play a role in the onset and progression of EC. Zhao et al. [18] performed serum-targeted metabolomics and found that fatty acids such as C16:1, C18:1, C20:1, C20:2, C22:6, C24, and C24:1 were significantly enriched in the serum of EC patients. Mass spectrometry imaging further confirmed that C16:1 and C18:1 accumulated more extensively in EC tumor regions compared to adjacent tissues. Both C16:1 and C20:2 showed a positive correlation with the EC-specific bacterium Ruminococcus sp. N15, and the abundance of both Ruminococcus sp. N15 and these metabolites increased synchronously from the control group to EC patients. Ruminococcus plays a significant role in lipid metabolism, primarily by fermenting dietary fibers to produce short-chain fatty acids like butyrate, which affect the host’s lipid metabolism and storage functions. Specifically, Ruminococcus flavefaciens regulates lipid metabolic pathways by promoting fatty acid absorption and utilization [65, 66]. Furthermore, Zhao et al. [18] found that Ruminococcus sp. N15.MGS-5 in EC patients’ gut was positively correlated with serum triglycerides (TG) and negatively correlated with high-density lipoprotein (HDL). Both TG and HDL are associated with EC risk, with EC patients showing significantly higher TG levels compared to healthy controls and endometriosis patients, while LDL levels were lower [67, 68]. The TG/HDL-c ratio in EC patients was significantly correlated with EC’s FIGO stage. After adjusting for confounding factors, the TG/HDL-c ratio was identified as a potential prognostic marker for postmenopausal EC patients [69, 70]. While Ruminococcus is also enriched in the microbiota of the vagina, cervix, fallopian tubes, and ovaries in EC patients [71], our analysis of the gut microbiota in endometrial lesion patients did not detect a specific increase in Ruminococcus abundance.

Our study results show that the gut microbiota in patients with endometrial lesions is significantly correlated with various lipid metabolites, including Enterococcus and Flavonifractor. The role of Flavonifractor in lipid metabolism has been studied extensively. This bacterium metabolizes dietary polyphenols (such as flavonoids) and fiber to generate physiologically active short-chain fatty acids (like butyrate), thereby influencing host metabolism [72, 73]. In EC patients with gut dysbiosis, the concentration of butyrate in host metabolism increases, which, by promoting fatty acid oxidation and enhancing tumor cell utilization of lipids, indirectly supports the metabolic needs of cancer cells [74]. Butyrate, as a histone deacetylase (HDAC) inhibitor, alters chromatin structure, activating or inhibiting the expression of genes related to cancer cell proliferation and migration, thus promoting tumor growth and metastasis [75]. In contrast, Enterococcus strains exhibit strong probiotic potential and play an important regulatory role in various metabolic pathways involved in liver lipid accumulation and lipogenesis. Specifically, in high-fat diet-induced obesity mouse models, the group treated with Enterococcus faecalis EF-2001 showed a significant reduction in lipid accumulation, involving AMPK phosphorylation activation, which inhibited transcription factors (such as SREBP-1c) and key enzymes (like fatty acid synthase, FAS) related to lipogenesis. This effectively lowered triglyceride (TAG) and cholesterol levels in liver and adipose tissue [76, 77]. The adaptability of E. faecalis to lipid membrane components is also noteworthy. When exposed to high concentrations of exogenous fatty acids (such as oleic acid and linoleic acid), the membrane lipid composition of E. faecalis underwent significant changes, including a decrease in the ratio of phosphatidylglycerol (PG) and diacylglycerol (DAG), while the proportion of cardiolipin (CL) increased [76, 78]. These alterations not only demonstrate E. faecalis’s adaptive regulation to lipid environmental stress but may also affect the host’s lipid metabolic pathways by altering the fluidity and structural stability of membrane lipids.

While the intestinal microbiota influences endometrial lesions through the regulation of lipid metabolism, it also interacts with the host’s hormone levels via other mechanisms, such as β-glucuronidase (GUS). Specifically, β-glucuronidase regulates estrogen activity, further promoting the development of endometrial tumors. Therefore, the intestinal microbiota not only plays a role in lipid metabolism but also influences the progression of lesions by modulating hormone metabolism.

GUS is a glycosidase enzyme derived from the intestinal microbiota, primarily exerting its effects through bacteria from the Firmicutes phylum. A meta-analysis has shown that bacteria from the Firmicutes phylum increase estrogen levels in the host, while bacteria from the Bacteroidetes phylum decrease them. In essence, estrogen, a common endogenous glycoside, is metabolized to estrone sulfate in the liver and is extensively regulated by gut microbiota β-glucuronidase (gmGUS) [79]. Research by Sui et al. [80] indicates that the crystal structure loops, mini-loop 1, and flavin mononucleotide (FMN) sequences of GUS proteins derived from Escherichia coli, Bacteroides fragilis, and Streptococcus agalactiae determine the catalytic activity of gmGUS on glycoside substrates, which profoundly affects the proliferation of estrogen-dependent tumor cells such as endometrial cancer (EC). Furthermore, Wang et al. [81] discovered that Pichia terricola M2, a gut fungus with GUS activity, effectively converts icariin into its metabolite icariin II, confirming the role of gut fungi in mediating intestinal glycoside metabolism. Moreover, icariin II significantly inhibits the proliferation of human EC cells. When the intestinal environment achieves homeostasis, gmGUS performs a functional role that helps balance estrogen levels, exerting an inhibitory effect on the development of gynecological tumors. However, when the gut microbiota is dysregulated, overexpression of gmGUS may increase the risk of EC development.

Diagnostic value of gut Microbiota-Based models for predicting endometrial lesions

Gut microbiota biomarkers offer considerable advantages in non-invasive detection. Researchers have attempted to construct gut microbiome screening models and identify core biomarkers, confirming the effectiveness of tumor-specific gut biomarkers for screening across multiple populations. For instance, a large-scale case-control study [82] demonstrated that combining the abundance of Fusobacterium nucleatum (Fn) in fecal samples with immunohistochemical testing improved sensitivity and specificity for colorectal cancer (CRC) detection to 92.8% and 79.8%, respectively. Furthermore, the ratio of Fn to Bifidobacterium (Bb) and Faecalibacterium prausnitzii (Fp) showed promising potential in diagnosing CRC. The results showed that the Fn/Bb ratio had a sensitivity of 84.6% and specificity of 92.3% (AUC = 0.911), while combining Fn/Bb and Fn/Fp for diagnosing stage I CRC demonstrated a specificity of 60.0% and sensitivity of 90.0% (AUC = 0.804).

Building on early research that highlighted the objective and effective application of gut microbiota in gastrointestinal cancer diagnosis, researchers in the gynecological field have also recognized the potential utility of gut microbiota as biomarkers for diagnosing ovarian cancer, cervical cancer, and endometrial carcinoma (EC). Kang et al. [83] identified a fecal microbiota model based on Prevotella, Peptostreptococcus, Finegolida, Ruminococcus, Clostridium, Pseudomonas, and Turibacter that predicted early invasive cervical cancer with a mean AUC value of 0.913. By applying random forest (RF) to the raw dataset and examining overlapping features, the model exhibited high predictive accuracy (training AUC = 0.91, validation AUC = 0.88). In the study of gut microbiota for diagnosing EC, the focus has primarily been on describing the gut microbiome features, with relatively limited research on constructing diagnostic models or validating the stability of associated biomarkers in large cohorts. To date, the only study by Zhao [18] has assessed the diagnostic value of the gut microbiome in patients with endometrioid adenocarcinoma. The results showed that a combined prediction model based on Ruminococcus sp. N15.MGS-57 strain, metabolites C16:1 and C20:2, and lipid markers such as triglycerides (TG) and low-density lipoprotein (LDL) had an AUC of 0.902 in the training set and 0.905 in the validation set, demonstrating high diagnostic efficacy. However, since the model features a close relationship between microbiome characteristics, metabolites, and lipid markers, its diagnostic results may be significantly influenced by the lipid levels of patients in clinical applications. Additionally, this study did not thoroughly validate the effectiveness of individual microbiota as independent diagnostic biomarkers, limiting its independence and applicability in various clinical settings.

Moreover, gut microbiota has shown considerable potential in predicting therapeutic efficacy and adverse reactions in gynecological malignancy treatment. For example, in cervical cancer and EC patients, specific changes in the gut microbiota can predict the efficacy of pembrolizumab treatment and potential adverse reactions [84]. The risk of grade 2 radiation enteritis (RE) following internal radiation therapy can also be effectively assessed through the relevant features in the patients’ gut microbiota [85]. Gut microbiota characteristics not only provide novel perspectives for gynecological tumor diagnosis and precision treatment but also help optimize treatment plans and reduce the occurrence of adverse reactions.

Limitations of the study

This study provides preliminary evidence of the potential association between gut microbiota and AEH as well as EC. However, several limitations remain: First, the sample size in this study is relatively small, particularly in the AEH group, which included only 7 samples, potentially limiting the generalizability of the results. Future studies should expand the sample size through multi-center, cross-regional research to validate these findings.Although the study identified gut microbiota signatures and lipid metabolites associated with EC, the expression of key enzymes in the glycerophospholipid metabolism pathway and the specific mechanisms through which gut microbiota structure influences the pathogenesis of EC remain insufficiently explored. Future studies should integrate techniques such as immunohistochemistry and animal models to further investigate the underlying mechanisms.Moreover, the differences in gut microbiota among individuals are influenced by factors such as gut type, body mass index, lifestyle, exercise frequency, ethnicity, and dietary and cultural habits. The heterogeneity of gut microbiota is a common phenomenon across individuals, and future research should employ stratified sampling or mixed-effects models to quantify the impact of these factors on gut microbiota.

Conclusion

This study highlights the changes in gut microbiota at different stages of endometrial lesions and explores its potential role in the development of EC. We observed that as AEH progresses to EC, the abundance of Rothia, Butyrivibrio, Klebsiella, and Luoshenia increases, with Klebsiella showing the most notable changes during disease progression. The abundance shifts of dominant microbiota (such as Megamonas, Escherichia, etc.) in both EC and AEH groups are closely linked to tumor progression, suggesting that these microbiota could serve as novel biomarkers for the early diagnosis of EC. Furthermore, metabolic analyses revealed that lipid metabolism pathways, particularly glycerophospholipid metabolism, are significantly correlated with the occurrence of EC, indicating that the gut microbiota may play a critical role in the pathogenesis of EC through the modulation of lipid metabolism and immune responses. These findings provide new insights into the mechanisms by which gut microbiota influences EC and lay the foundation for developing early diagnostic tools and personalized treatment strategies based on gut microbiota.

Authors’ contributions

Conceptualization, Z.W. and H.P.; methodology, Z.W.; validation, Z.W. and Y.S.; formal analysis, H.P., Y.S. and J.C.; investigation, H.P., J.C.and M.C.; resources, J.L.,C.H.,H.W.,X.C.,H.Y. and J.Z.; data curation, H.P., J.S.and J.C.; writing—original draft preparation, H.P. and J.C; writing—review and editing, Z.W. and Y.S.; supervision, Z.W.; project administration, Z.W.; funding acquisition, Z.W. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by grants from the Natural Science Foundation of Gansu Province (22JR5RA718 and 24JRRA621).

Data availability

The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive in National Genomics Data Center (Nucleic Acids Res 2024), China National Center for Bioinformation / Beijing Institute of Genomics, Chinese Academy of Sciences (GSA: PRJCA034591) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Gansu Provincial Maternity and Child-Care Hospital (Approval No. 2022-040-KYSL). All data have been anonymized and are accessible exclusively to the research team on a need-to-know basis. Informed consent was obtained from all participants in this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.