Gut microbiota dysbiosis and altered fecal metabolome in patients with age-related cataract

Abstract

Age-related cataract (ARC) is a leading cause of vision loss with incompletely understood mechanisms. Emerging evidence suggests the gut microbiota can influence ocular health, yet the gut–eye connection in cataract remains unexplored. We characterized gut microbial communities and fecal metabolic profiles in 30 ARC patients and 30 healthy controls using 16S rDNA gene sequencing, untargeted LC–MS metabolomics, and targeted GC–MS for short-chain fatty acids (SCFAs). While alpha- and beta-diversity were comparable between groups, ARC patients exhibited significant dysbiosis, including reduced Gut Microbiome Health Index, increased Microbial Dysbiosis Index, higher relative abundance of Bifidobacterium and Klebsiella, and depletion of butyrate-producing taxa (Faecalibacterium, Clostridia). Fecal metabolomic profiles distinctly separated ARC patients from controls, with pathway analysis highlighting disruptions in glycerophospholipid and choline metabolism. Targeted analysis confirmed significant depletion of acetate, propionate, and butyrate in ARC patients (all  P< 0.01), which positively correlated with beneficial genera abundance. This integrative study reveals that ARC patients harbor gut microbial dysbiosis and distinct fecal metabolomic signatures, notably a loss of SCFA-producing bacteria and anti-inflammatory SCFAs. These findings support a novel gut–eye axis in cataract pathogenesis and suggest that gut-derived microbial and metabolic biomarkers may aid in non-invasive risk assessment, while microbiome-targeted interventions could offer new preventive or therapeutic avenues.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-53261-5.

Introduction

Age-related cataract (ARC) is a major ocular disease characterized by reduced lens transparency or color alteration¹, ultimately leading to visual impairment or complete blindness^2,3. Currently, ARC affects approximately 95 million individuals worldwide. Aging is the major risk factor for ARC, but accumulating evidence indicates that systemic factors such as nutrition, metabolism and inflammation also contribute to cataractogenesis^4,5. For example, high-glycemic diets and micronutrient deficiencies have been epidemiologically linked to accelerated cataract formation, highlighting metabolic contributions to lens aging^6,7. Nevertheless, the precise mechanisms by which these systemic influences promote cataract remain poorly defined⁸.

The gut microbiota is increasingly appreciated as a critical regulator of host metabolism and immune homeostasis^9,10. Dysbiosis of the intestinal microbiome can drive chronic inflammation¹¹ and metabolic derangements¹² that affect distant organs. A “gut–eye axis” has been proposed in other ocular diseases: recent studies associate gut microbial changes with uveitis^13,14, diabetic retinopathy^15,16 and age-related macular degeneration^17,18, possibly via circulating microbial metabolites and inflammatory mediators. Of course, numerous studies have also proposed that the use of probiotics, prebiotics, antibiotics, and fecal microbiota transplantation to ameliorate ocular diseases has shown promise in preclinical and early-phase human studies^19,20. However, whether gut microbiota dysregulation plays a role in ARC pathogenesis is unknown.

Here, we applied an integrative multi-omics approach to investigate gut microbiota and fecal metabolites in ARC. We performed 16S rDNA gene sequencing on fecal samples from ARC patients and matched healthy controls, alongside untargeted LC–MS metabolomics and targeted short-chain fatty acid (SCFA) profiling. Our objective was to identify specific microbial and metabolic signatures of ARC. Uncovering such gut-derived factors could shed light on novel mechanisms in cataract development and point toward non-invasive biomarkers or microbiome-based interventions for prevention.

Results

The characteristic composition of the sample

In this study, 30 individuals with age-related cataract (25 females, 5 males) and 30 healthy controls (21 females, 9 males) were enrolled. Participant information collected included age, sex, and cataract severity grading. Sex distribution (P = 0.35) and body mass index (BMI, P = 0.9) showed no statistically significant differences between groups. The cataract group had a higher mean age (62.37 ± 5.62 years) than the healthy controls (57.97 ± 3.47 years). See Table 1 for details.

Table 1.

Main Features of Enrolled Patients and Controls (males, females, mean age, BMI, probiotics, and number of antibiotics are listed for each group).

Features at the Time of Sample Collection	Patients	Controls	P
Male/Female	5/25	9/21	0.35
Average age, y	62.37 ± 5.62	57.97 ± 3.47	< 0.001
BMI	23.16 ± 3.41	23.22 ± 2.06	0.9
C2/C3/C4	6/18/6	-
N2/N3/N4	17/10/3	-
P0/P1/P2/P3	6/21/2/1	-
Probiotics	-	-
Antibiotics	-	-

Numbers of male, female, average age, BMI, probiotics, and antibiotics are listed in the table for each group.

Gut microbial diversity indices are similar between ARC patients and controls, but dysbiosis indices differ significantly

The diversity data from 60 fecal samples were analyzed, resulting in a total of 3,511,418 and 1,437,879,351 bases of optimized sequences, with an average sequence length of 410 base pairs (bp). Alpha diversity, measured by Shannon and Chao1 indices, was modestly lower in ARC patients compared to controls, but the differences were not statistically significant (both indices  P> 0.05) (Fig. 1A-C). Rarefaction and rank-abundance curves indicated sufficient sequencing depth for all samples (online supplemental Fig. S1). Beta diversity by PCoA on Bray-curtis distances showed substantial overlap between the ARC and control groups, with no clear clustering by group ( P> 0.05) (Fig. 1D).

Fig. 1.

Fecal 16 S rDNA analysis revealed the microbial composition of the gut microbiota in the ARC and N groups. A–C The box plots showed alpha diversity comparisons between ARC and N groups with chao index (A), shannon index (B) and coverage index (C). D PCoA using Bray-curtis of beta diversity, indicating no significant difference in distribution between groups. E, F The violin plots showing the gut flora health index (GMHI) (E) and microbial dysbiosis index (MDI) (F) for the ARC and N groups, which were used to assess the relative abundance of microbial species and the degree of gut microbiota disorder in the ARC and N groups. *P < 0.05, **P < 0.01, ***P < 0.001. PCoA, principal coordinate analysis; ARC, age-related cataracts(n = 30); N, healthy control(n = 30).

We next assessed global microbiota health metrics. The Gut Microbiome Health Index (GMHI), a composite measure favoring beneficial taxa, was significantly lower in ARC patients than in controls ( P< 0.01), while the Microbial Dysbiosis Index (MDI), which increases with dysbiotic taxa, was significantly higher in ARC ( P< 0.01) (Fig. 1E–F). When comparing GMHI and Shannon diversity for each sample, GMHI consistently outperformed Shannon diversity in two groups (online supplemental Fig. S2). These results suggest that, despite similar diversity, the ARC group harbors a more dysbiotic microbiome signature.

Bifidobacterium is enriched, and butyrate-producing genera are reduced in ARC patients

We evaluated shared and unique microbial taxa (e.g., Amplicon Sequence Variants, ASVs) between ARC patients and healthy controls to identify potential disease-specific biomarkers for diagnostic modeling. At the bacterial phylum level, our analysis revealed 11 phyla common to both groups, with no unique phylum-level taxa observed exclusively in either the ARC patient group or the control group (Fig. 2A). Further analysis at the genus level revealed 217 genera shared between ARC patients and healthy controls. Additionally, we identified 45 genera unique to healthy controls and 46 genera unique to ARC patients (Fig. 2B). At the phylum level, the microbial community composition showed that Firmicutes, Proteobacteria, Actinobacteria, and Bacteroidetes were the predominant phyla in both ARC patients and healthy controls (Fig. 2C). At the genus level, ARC patients exhibited an increased abundance of Bifidobacterium and Klebsiella, accompanied by a reduced presence of Escherichia–Shigella and the butyrate-producing genus Faecalibacterium compared to healthy controls (Fig. 2D). Correlation heatmaps illustrating microbiota levels in all ARC patients and healthy controls (Fig. 2E).

Fig. 2.

Fecal 16S rDNA analysis revealed the community composition of the gut microbiota in the ARC and N groups. A, B The venn plots showed the number of common and unique species in the ARC and N groups at the bacterial phylum level (A) and at the genus level (B). C The histogram showed that Firmicutes, Proteobacteria, Actinobacteria, and Bacteroidetes (marked with red box) were the most abundant in the phylum level community composition of intestinal bacteria in ARC and N groups. D The histogram showed that there were significant differences in the community composition of Escherichia-Shigella, Bifidobacterium, Faecalibacterium and Klebsiella (marked with red box) at the genus level between ARC and N groups. E The heatmap showed the horizontal distribution of intestinal bacterial phyla in the ARC and N groups. ARC age-related cataracts (n = 30), N healthy control (n = 30).

We applied robust statistical methods to compare species abundance between ARC patients and healthy controls based on microbial community composition. Taxon-specific abundance patterns were assessed using effect size filtering to identify organisms with biologically meaningful differences between groups. Comparative analysis revealed a significantly higher abundance of Bifidobacterium in ARC patients compared to controls (P < 0.05) (Fig. 3A). To further identify dominant taxa, we used linear discriminant analysis effect size (LEfSe), which highlighted key discriminative bacteria across taxonomic levels. Clostridia, Oscillospirales, and Ruminococcaceae were enriched in healthy controls, whereas Bifidobacterium was predominant in the ARC group (Fig. 3B). In total, LEfSe identified four bacterial taxa significantly enriched in ARC patients and ten taxa enriched in healthy individuals (Fig. 3C).

Fig. 3.

Analysis of differences in gut microbial species between ARC and N groups. A The bars showed the difference in the average relative abundance of different species between the ARC and N groups. B LEfSe cladogram showed the difference of gut microbiota between ARC and N groups at different species levels. C The bar chart shows the LDA value of each different species of intestinal flora in the ARC and N groups, which is used to measure the influence of different species on the difference effect. ARC age-related cataracts (n = 30), N healthy control (n = 30).

Fecal metabolomic profiles significantly differ between ARC patients and controls, highlighting altered lipid metabolism

Untargeted LC–MS profiling detected several thousand metabolic features across all fecal samples. In unsupervised PCA, ARC and control samples overlapped extensively (Fig. 4A), suggesting no gross global differences. However, supervised PLS-DA achieved clear separation of the two groups (Fig. 4B). The corresponding OPLS-DA model (cross-validated R² = 0.95, Q²= –0.42) confirmed robust metabolic discrimination (Fig. 4C–D).

Fig. 4.

Untargeted metabolomics revealed differential metabolites in the gut microbiota between ARC and N groups. A, B Through PCA analysis (A) and PLSDA analysis (B), the scatter plots showed the overall differences in gut microbiota between ARC and N groups. C The OPLS-DA scores plot was rotated orthogonally to filter out the information irrelevant to the group and better distinguish the difference between ARC and N groups. D OPLS-DA permutation test. ARC, age-related cataracts (n = 30), N healthy control (n = 30).

Using a combination of VIP scores and statistical testing, we identified 430 metabolites that differed between ARC and controls (122 were upregulated and 308 downregulated in ARC) (Fig. 5A–B). After FDR correction, 39 remained significant ( P< 0.05 FDR; 13 up, 26 down in ARC; Supplementary Tables 1–2). Notably, the majority of differential metabolites fell into two broad chemical classes: lipids and peptides (Fig. 5C). KEGG compound classification similarly showed enrichment of lipid metabolites among ARC-associated changes.

Fig. 5.

Differential metabolite analysis and metabolite set enrichment analysis between ARC and N groups. A, B The volcano plot (A) and histogram (B) showed the significantly different metabolite profiles between the ARC and N groups. C The differentiated metabolites in ARC group were classified according to KEGG compound classification. D KEGG functional pathways were classified among ARC group differential metabolites. E The bubble map shows the pathway in which KEGG is significantly enriched in the differential metabolites of the ARC group. ARC age-related cataracts (n = 30), N healthy control (n = 30).

Pathway enrichment analysis of the significant metabolites revealed that lipid-related pathways were predominantly affected. The most enriched pathways included glycerophospholipid metabolism and choline metabolism (Fig. 5D–E). These pathways are central to cell membrane composition and lipid signaling. The prominence of glycerophospholipid alterations suggests perturbed membrane lipid homeostasis in ARC patients’ gut metabolome. Overall, the metabolomic data indicate that ARC is accompanied by disruptions in intestinal lipid and amino acid metabolism.

Acetate, propionate, and butyrate levels are reduced in ARC patients and positively correlate with specific SCFA-producing bacterial genera

Given the loss of butyrate-producing bacteria in ARC, we quantified fecal SCFAs to assess this metabolic consequence. Targeted GC–MS showed that acetate, propionate, and butyrate were significantly lower in ARC patients than in controls (all  P< 0.01; Fig. 6A–C). Other SCFAs (isobutyrate, valerate, etc.) did not differ significantly. This pattern of reduced major SCFAs suggests impaired microbial fermentation in ARC.

Fig. 6.

The content of eight short-chain fatty acids in the metabolites of ARC and N groups was analyzed by GC-MS, including acetic acid (A), propanoic acid (B), butanoic acid (C), isobutyric acid (D), valeric acid (E), isovaleric acid (F), hexanoic acid (G) and isohexanoic acid (H). The correlations between acetic acid, propionic acid, butyric acid and the top ten bacteria of genus level were analyzed (I). *P < 0.05, **P < 0.01, ***P < 0.001. ARC age-related cataracts (n = 30), N healthy control (n = 30).

Correlation analysis linked these SCFA changes to the altered microbiota. Acetate, propionate and butyrate levels positively correlated with the relative abundances of several beneficial genera (Spearman ρ > 0.5,  P< 0.05), including Faecalibacterium, Subdoligranulum, Eubacterium hallii group and Ruminococcus torques (Fig. 6I). These genera are known butyrate or propionate producers. Thus, the loss of these taxa in ARC likely underlies the observed SCFA deficiency.

Discussion

This study provides the first detailed characterization of the gut microbiota and fecal metabolome in patients with age-related cataract. Although overall microbial diversity was similar between groups, we observed clear signs of dysbiosis and altered metabolism in ARC. Notably, ARC patients exhibited expansion of Bifidobacterium and other non-butyrate-producing bacteria, accompanied by depletion of key commensals that generate short-chain fatty acids (SCFAs). These compositional shifts were paralleled by major differences in fecal metabolites, especially lipids, and a marked reduction in acetate, propionate and butyrate levels. We propose that these microbe–metabolite perturbations may mechanistically contribute to ARC pathogenesis.

The enrichment of Bifidobacterium in ARC is intriguing. Bifidobacteria are typically early colonizers of the infant gut and are often considered beneficial²¹; however, their expansion in adult dysbiosis has been reported in metabolic disorders. Bifidobacteria primarily ferment dietary carbohydrates to acetate and lactate²², but in our data ARC patients nonetheless had lower fecal acetate, suggesting altered metabolic flux or cross-feeding in the community. The concomitant depletion of major Firmicutes (e.g. Faecalibacterium, Clostridia) indicates loss of butyrate-producing capacity. Butyrate is known to strengthen the intestinal barrier and regulate inflammation through multiple mechanisms, including stimulation of T regulatory cells and inhibition of histone deacetylases^23–25. Reduced butyrate availability may therefore allow a more pro-inflammatory intestinal milieu. In other organs, such as the retina, impaired SCFA signaling has been linked to inflammatory eye diseases^23,26. We speculate that chronic low-grade gut inflammation or increased gut permeability in ARC patients could lead to systemic dissemination of inflammatory mediators or oxidative metabolites, ultimately affecting the lens. For instance, systemic inflammation can promote protein oxidation and crosslinking in the lens, accelerating opacification.

The fecal metabolome of ARC patients was dominated by dysregulated lipid metabolism. Glycerophospholipids are fundamental components of cell membranes throughout the body, with lens fiber cell membranes being a notable site of their prominence. The significant alterations we observed in fecal glycerophospholipids are particularly intriguing in light of recent reviews highlighting the disruption of glycerophospholipid metabolism as a central feature in the pathogenesis of age-related cataract within the lens itself²⁷. The enrichment of metabolites related to glycerophospholipid and choline pathways in ARC stool may reflect systemic lipid imbalance²⁸. It is known that aging and senescence are associated with lipid droplet accumulation and changes in membrane lipids²⁹, which can trigger inflammatory pathways (for example via phospholipase A2 activity)^30,31. The ARC-associated metabolite signature thus suggests a link between gut microbiota-induced lipid dysregulation and the senescence-associated secretory phenotype (SASP) that drives tissue aging. Our pathway analysis also highlighted amino-acid and energy metabolism pathways, consistent with a broader metabolic shift in ARC.For example, The literature suggests that taurine deficiency may lead to reduced glutathione activity and an increased demand for NAD+, resulting in tryptophan metabolism via the kynurenine pathway—a known pathway in cataract formation. This process can exacerbate oxidative stress and promote cataract progression, which adds an important new dimension to our proposed gut-eye axis mechanism for cataract³².

Integrating our findings, we propose a model in which gut dysbiosis in ARC leads to altered microbial metabolism that promotes systemic oxidative stress and inflammation. The loss of SCFA-mediated anti-inflammatory signaling, combined with aberrant lipid metabolites, could exacerbate lens protein damage. This “gut–eye axis” concept aligns with emerging evidence in other ocular diseases. While the depletion of SCFAs is unlikely to be the sole driver of cataract progression, it appears to be a pivotal event within the gut-eye axis. This depletion, likely in concert with other metabolic disruptions such as altered taurine and lipid metabolism, may compromise intestinal barrier function and diminish systemic anti-inflammatory and antioxidant defenses, thereby contributing to the pathological environment that promotes lens opacification. Importantly, our data identify specific gut-derived signals—such as decreased fecal acetate/propionate/butyrate and elevated lipid metabolites—as candidate biomarkers for cataract risk. These non-invasive stool markers may complement traditional ocular risk assessments.

Several limitations of this study should be acknowledged. Due to the challenges of recruiting a strictly defined elderly cohort willing to provide fecal samples, our current sample size did not permit stratified analysis by cataract severity (LOCS III grade). Future studies with larger, well-powered cohorts are warranted to investigate whether the degree of gut microbiota dysbiosis correlates with cataract progression.

We acknowledge the significant age difference between our cataract and control groups as a limitation. However, several lines of evidence suggest that this ~ 4.4-year age gap is unlikely to be the primary driver of our key findings: (1) 91.67% of participants could be matched within ± 3 years, indicating substantial age overlap; (2) alpha- and beta-diversity did not differ between groups, consistent with literature showing gradual, non-abrupt age-related microbiota changes^33,34; and (3) most importantly, the key differentially abundant metabolites (acetate, propionate, butyrate) showed no significant correlation with age (online supplemental Fig. S3, Supplementary Table S3). Nevertheless, future studies with stricter prospective age-matching are warranted to validate our findings.

Our design is cross-sectional, so causality cannot be established; longitudinal studies are needed. Additionally, we analyzed fecal samples as a proxy for gut metabolites, but direct measurement of circulating inflammatory markers or lens biochemistry would strengthen mechanistic claims. Finally, 16 S sequencing provides genus-level resolution; future shotgun metagenomic or metabolomic studies could reveal more specific microbial functions.

Conclusions

Our multi-omics analysis demonstrates that patients with age-related cataract exhibit gut microbiota dysbiosis and characteristic fecal metabolic changes. In particular, the loss of commensal SCFA-producing bacteria and corresponding depletion of anti-inflammatory SCFAs may promote chronic inflammation and oxidative stress, contributing to lens aging. These findings support the existence of a gut–eye axis in cataract pathogenesis. Stool-based microbial and metabolic biomarkers identified here could facilitate early detection of cataract risk. Moreover, microbiome-targeted therapies (for example, dietary fiber to restore SCFAs or probiotics to rebalance the gut flora) represent promising novel strategies to prevent or slow the development of ARC. Future research should validate these candidates in larger cohorts and experimental models.

Methods

Patients recruitment and sample collection

Thirty patients aged over 50 years, diagnosed with ARC and undergoing phacoemulsification surgery, were recruited between November 2022 and March 2023 at the University-Town Hospital of Chongqing Medical University. Thirty healthy volunteers matched for age and sex were enrolled as controls (denoted as N in figures). A single physician conducted a comprehensive physical examination and medical history assessment, along with ocular slit lamp and funduscopic examinations. The cataracts were graded using the Lens Opacity Classification System III (LOCS III)³⁵. Ethical approval was obtained with documented informed consent from all participants. Study protocols strictly followed international ethical guidelines (Declaration of Helsinki). Exclusion criteria encompassed: (1) Three-month pre-collection use of pharmacological agents (antibiotics, probiotics, corticosteroids) or gastrointestinal surgical history; (2) Comorbidities including oncological, metabolic (obesity, diabetes), cardiovascular (hypertension), dermatological (psoriasis/psoriatic arthritis), or gastrointestinal (inflammatory bowel disease) disorders; (3) Neuropsychiatric conditions (e.g., depressive disorders, anxiety spectrum disorders, bipolar manifestations). Eligible subjects maintained standardized nutritional intake and demonstrated regular bowel function without chronic diarrheal or constipatory disorders.

Fecal DNA extraction

Total microbial genomic DNA was extracted from approximately 200 mg of each fecal sample using the FastPure Stool DNA Isolation Kit (MJYH, Shanghai, China) according to the manufacturer’s optimized protocol. The procedure strictly followed the manufacturer’s instructions, including optimized steps for pretreatment, enzymatic digestion, and purification to ensure high-quality DNA extraction. DNA integrity was assessed by 1% agarose gel electrophoresis using a Bio-Rad Sub-Cell GT electrophoresis system. The concentration and purity of the extracted DNA were measured using a Thermo Scientific NanoDrop 2000 spectrophotometer.

16S rDNA sequencing

Genomic DNA was used to amplify the V3–V4 hypervariable regions of the 16S rRNA gene with barcoded primers 338 F and 806R described previously³⁶. Each sample underwent triplicate PCR reactions, and amplicons were pooled for each sample. The combined PCR products were purified from 2% agarose gels, and concentrations were measured using a microplate reader. Sequencing libraries were prepared with the NEXTFLEX Rapid DNA-Seq Kit (Bioo Scientific, Austin, Texas, USA), followed by paired-end sequencing (2 × 300 bp) on the Illumina NextSeq 2000 platform (Shanghai Majorbio Bio-pharm Technology Co., Ltd., China).

Sequence analysis

Quality Raw paired-end sequencing data were first quality-filtered using fastp (v0.19.6; https://github.com/OpenGene/fastp)³⁷. S High-quality reads were then merged with FLASH (v1.2.11; http://www.cbcb.umd.edu/software/flash)³⁸. The merged reads were denoised using the DADA2³⁷ within the Qiime2 pipeline³⁹, applying default parameters. Sequences identified as originating from chloroplasts or mitochondria were removed from the dataset. Amplicon sequence variants (ASVs) were taxonomically assigned using a Naive Bayes classifier trained on the Silva 16S rRNA reference database (version 138) within QIIME 2. Functional prediction of the 16S rRNA gene was subsequently performed with PICRUSt2 software (version 2.2.0)⁴⁰.

LC-MS analysis

After sample preparation, intestinal metabolites were analyzed by liquid chromatography–mass spectrometry (LC-MS) at Majorbio Bio-pharm Technology Co., Ltd. (Shanghai, China). Analyses were performed on a Thermo Vanquish UHPLC system coupled to a Q Exactive HF-X mass spectrometer, using an Accucore C30 column (100 mm × 2.1 mm, 2.6 μm) with a 2 µL injection volume. A quality control (QC) sample was run after every 15 experimental samples. Raw LC-MS data were processed in LipidSearch software (Thermo, CA), which included peak detection, retention time alignment, and lipid identification, resulting in a matrix containing lipid names, retention times, m/z values, and peak intensities.

SCFAs extraction

To prepare SCFA calibration standards, 9840 µL of n-butanol was combined with appropriate amounts of eight SCFA standards, thoroughly vortexed, and used as mixed standard stock solution A. For the internal standard, 10 µL of 2-ethylbutyric acid was added to 9990 µL of n-butanol (stock solution B). Serial dilutions of A and B were performed in n-butanol to generate a seven-point calibration curve. All working solutions were transferred to autosampler vials for GC-MS measurement.

For sample extraction, about 25 mg of fecal sample was weighed into a 2 mL grinding tube, and 500 µL of 0.5% phosphoric acid solution was added. After rapid freezing in liquid nitrogen, the sample was homogenized twice at 50 Hz for 3 min each. The homogenate was sonicated for 10 min, then centrifuged at 4 °C, 13,000 × g for 15 min. A 200 µL aliquot of the supernatant was mixed with 200 µL of n-butanol containing 2-ethylbutyric acid (10 µg/mL) as the internal standard. The mixture was vortexed, sonicated in an ice bath for 10 min, and centrifuged for another 5 min at 4 °C, 13,000 × g. The final supernatant was transferred to an autosampler vial for GC-MS analysis.

Gas chromatography–mass spectrometry (GC–MS) analysis

Targeted detection of short-chain fatty acids (SCFAs) was carried out on an Agilent 8890B-5977B GC-MS system. Quality control (QC) samples were analyzed after every 5 to 10 experimental runs to ensure reproducibility, with the relative standard deviation (RSD) for all analytes kept below 15%. Data processing, including peak detection and integration, was performed using MassHunter Quantitation software (Agilent Technologies, USA; version v10.0.707.0), with manual review as needed. SCFA concentrations in the samples were calculated based on external calibration curves to determine their actual levels in the original specimens.

Statistical analyses

Quantitative analyses were performed with SPSS Statistics 29.0. Normally distributed continuous data were expressed as arithmetic mean ± standard deviation. Non-parametric Mann-Whitney U tests were utilized for inter-group comparisons of microbial alpha-diversity indices (Shannon/Chao1). Multivariate beta-diversity patterns were assessed via using Adonis dissimilarity matrices. Metabolomic feature discrimination incorporated dimensionality reduction through principal component modeling (PCA) coupled with orthogonal signal correction-enhanced discriminant analysis (OPLS-DA), with differential compound identification achieved via two-tailed parametric hypothesis testing. To assess the degree of age overlap between groups, we performed a post-hoc age-matching analysis using a tolerance of ± 3 years. The percentage of participants who could be matched across groups was calculated. Additionally, Spearman correlation analysis was performed to evaluate the relationship between age and key differential metabolites (acetate, propionate, butyrate). Statistical significance threshold was set at P < 0.05. Comprehensive methodological details are archived in the Supplemental Experimental Procedures.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Y.L.contributed to collected and analyzed clinical data of patients, organized data and wrote the manuscript. C.S.and H.L.are mainly responsible for making pictures and tables. X. S.is responsible for assisting in the writing of the manuscript. T. Y.is responsible for provide assistance in sample processing. J. L. and Y.G. conceived of the study, participated in its design and coordination, and revised the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by CQMU Program for Youth Innovation in Future Medicine(W015), Natural Science Foundation Project of Chongqing(CSTB2023NSCQ-MSX0194; CSTB2022NSCQ-MSX1274), 2024 Annual Chongqing Postdoctoral Research Special Funding Program, 2025 Annual Chongqing Municipal Education Commission Science and Technology Research Plan Youth Project, and the National Natural Science Foundation of China (NSFC; 32300778).

Data availability

The 16S rRNA sequencing data generated in this study are available in the NCBI Sequence Read Archive (SRA) repository under BioProject accession number PRJNA1310227. The metabolomics data are available in the MetaboLights repository under accession number MTBLS12932.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

This study was approved by the Ethics Committee of University Town Hospital of Chongqing Medical University on August 8, 2022, and the approval number was LL-202215. All patients were informed of the purpose of the study and the content of the examination in advance, and all patients signed an informed consent before enrollment. All experimental data for this study were collected in strict accordance with the principles of the Declaration of Helsinki.