Integrative oral and gut microbiome profiling highlights microbial correlates of complications in type 1 diabetes: a cross-sectional analysis Authors: Rong Tang, Mei Shi, Xiaolin Ji, Yan Zhang, Li Fan, Fansu Huang, Xia Li Published: 2025 ABSTRACT AbstractBackground/ObjectiveChronic vascular complications are the primary threat in long-standing type 1 diabetes (T1D) patients. We examined the associations between oral-gut microbiome dysbiosis and these complications, offering novel insights into therapeutic strategies and underlying mechanisms.MethodsThis cross-sectional study enrolled 75 T1D participants (disease duration ≥ 10 years) and 43 healthy controls who underwent comprehensive clinical assessment, including blood glucose, lipid profile, and complication-related examinations. Fecal and oral rinse samples were collected for shotgun metagenomic sequencing. T1D participants were stratified by the presence of microvascular (retinopathy, nephropathy, or neuropathy) or macrovascular complications separately. Microbial differences across groups were assessed.ResultsSignificant differences in oral and gut microbiota compositions were observed between T1D participants with and without complications (both microvascular and macrovascular). A core set of 26 gut and 8 oral microbial species was specifically associated with vascular complications. Butyrate-producing gut bacteria (Blautia wexlerae, Anaerobutyricum hallii, Roseburia inulinivorans, A. soehngenii) and specific oral Neisseria species were enriched in T1D without complications individuals, suggesting protective effects against complications. Mediation analysis indicated associations consistent with partial mediation between certain microbial species and the relationships of glycemic control or insulin resistance (HbA1c, glucose risk index, estimated glucose disposal rate) with complication risk. Moreover, potential oral-gut microbiome interconnections were implicated in complication development. Finally, classification models integrating both oral and gut microbial features significantly outperformed models based on either site alone in distinguishing T1D patients with complications.ConclusionsDistinct oral and gut microbiome features are associated with chronic vascular complications in T1D. These findings highlight the potential of microbiome-targeted strategies for understanding and preventing T1D-related complications.Graphical Abstract Supplementary InformationThe online version contains supplementary material available at 10.1186/s12933-025-03012-z.Keywords: Type 1 diabetes, Microvascular complications, Macrovascular complications, Oral microbiome, Gut microbiome Research insights What is currently known about this topic? Alterations in both gut and oral microbiota have been observed in early type 1 diabetes and linked to disease pathogenesis and glycemic control. Alterations in both gut and oral microbiota have been observed in early type 1 diabetes and linked to disease pathogenesis and glycemic control. Periodontitis, reflecting oral dysbiosis, is epidemiologicallyassociated with diabetic microvascular complications(nephropathy, retinopathy), independent of conventional metabolic risk factors. Periodontitis, reflecting oral dysbiosis, is epidemiologicallyassociated with diabetic microvascular complications(nephropathy, retinopathy), independent of conventional metabolic risk factors. The oral microbiome can influence the gut microbiome ("oral-gut axis"), with implications for systemic diseases. The oral microbiome can influence the gut microbiome ("oral-gut axis"), with implications for systemic diseases. What is the key research question? Do distinct alterations within the interconnected oral–gut microbiome ecosystem associate with chronic vascular complications (both micro- and macrovascular) in long-standing type 1 diabetes, and are these associations consistent with potential mediation of metabolic dysregulation effects? Do distinct alterations within the interconnected oral–gut microbiome ecosystem associate with chronic vascular complications (both micro- and macrovascular) in long-standing type 1 diabetes, and are these associations consistent with potential mediation of metabolic dysregulation effects? What is the new? A core set of 26 gut and 8 oral microbial species was specifically identified as being associated with vascular complications in type 1 diabetes, characterized by depletion of protective butyrate-producing gut bacteria and oral Neisseria species in affected individuals. A core set of 26 gut and 8 oral microbial species was specifically identified as being associated with vascular complications in type 1 diabetes, characterized by depletion of protective butyrate-producing gut bacteria and oral Neisseria species in affected individuals. Specific microbial species showed associations consistent with partial mediation between glycemic control/insulin resistance indices and complication risk, suggesting a potential "metabolism–microbiota–complication" axis. Specific microbial species showed associations consistent with partial mediation between glycemic control/insulin resistance indices and complication risk, suggesting a potential "metabolism–microbiota–complication" axis. Integrated models using both oral and gut microbiome signatures significantly outperform single-site models in distinguishing type 1 diabetes patients with complications. Integrated models using both oral and gut microbiome signatures significantly outperform single-site models in distinguishing type 1 diabetes patients with complications. How might this study influence clinical practice? Oral and gut microbiome signatures, particularly as a combined biomarker profile, hold significant potential for noninvasive risk stratification of vascular complications and could guide the development of novel microbiome-targeted preventive or therapeutic strategies for individuals with type 1 diabetes. Oral and gut microbiome signatures, particularly as a combined biomarker profile, hold significant potential for noninvasive risk stratification of vascular complications and could guide the development of novel microbiome-targeted preventive or therapeutic strategies for individuals with type 1 diabetes. Introduction Type 1 diabetes (T1D) is an autoimmune disease triggered by a combination of genetic predispositions and environmental factors. With the advent of insulin therapy, T1D has shifted from an acutely fatal condition to a manageable chronic disease, significantly extending patient survival. However, chronic complications including microvascular and macrovascular have become the primary contributors to disease burden and mortality. Despite clinical efforts focused on controlling glycemia, lipids, and blood pressure, these metabolic interventions remain insufficient to prevent the onset and progression of complications [1]. This underscores the urgent need to explore additional mechanisms and therapeutic targets underlying T1D-related complications. The gut microbiota, the largest microbial community in the human body, has garnered increasing attention for its role in immune regulation and disease. Studies have demonstrated that the gut microbiota contributes to T1D pathogenesis by modulating immune responses and altering intestinal permeability [2]. Moreover, a study reported associations between gut microbiome composition, glycemic control, and the presence of chronic complications in T1D [3]. These findings suggest that gut microbial dysbiosis may persist throughout the disease course and potentially contribute to complications. In addition to the gut microbiota, the oral microbiota represents the second largest microbial community in the human body and is considered a potential upstream source of the gut microbiota [4]. Dysbiosis of the oral microbiota has been observed at the early stages of the T1D and has been shown to correlate with hemoglobin A1c (HbA1c) levels [5, 6]. Importantly, periodontitis has been significantly associated with both diabetic nephropathy (DN) and retinopathy (DR), and these associations remain robust even after adjusting for conventional metabolic risk factors [7, 8]. These findings suggest that oral microbial dysbiosis may contribute to systemic disease progression through mechanisms such as persistent local inflammation and microbial translocation [9]. Moreover, emerging evidence also suggests that oral microbiota may influence host physiology and disease progression by modulating the composition of the gut microbiome [10, 11]. A recent study revealed that oral microbiota alterations significantly reshaped the gut microbial landscape [12]. These findings highlight the potential importance of oral microbiota and its interactions with gut microbes in the pathophysiology of T1D. Although alterations in the oral and gut microbiota have been observed during the early stages of T1D, it remains unclear whether these microbial communities contribute to the development of vascular complications. The potential role of oral microbiota and its interactions with gut microbes has often been overlooked. Therefore, this study aimed to investigate alterations in the oral and gut microbiota among individuals with long-term T1D and explore their potential associations with chronic diabetic complications. The collection of saliva and stool samples is noninvasive, convenient, and amenable to repeated sampling, making it particularly suitable for large-scale surveillance and personalized disease management. Integrated analysis of the oral and gut microbiota may not only provide insights into the microbial mechanisms underlying T1D complications but also enable risk stratification and individualized prevention strategies. Methods Participants T1D participants meeting the following criteria were included in this cross-sectional study: (1) Diabetes diagnosed according to the 1999 World Health Organization criteria [13]. (2) T1D is clinically diagnosed by endocrinologists as follows: (i) insulin dependence from disease onset. (ii) Positive for at least one of islet autoantibodies for glutamic acid decarboxylase antibody [GADA], insulinoma-associated protein 2 antibody [IA-2A], and zinc transporter 8 antibody [ZnT8A]; or negative for all three islet autoantibodies but diagnosed before age 30 or diabetic ketoacidosis (DKA)/diabetic ketosis (DK) at onset. (3) Duration of diabetes ≥ 10 years. Healthy controls were recruited through a recruitment advertisement at local medical centers. They were tested for fasting blood glucose < 6.1 mmol/L and HbA1c < 6%. We used the method of group matching, and controls were matched for sex and age with T1D participants. To control for factors affecting oral and gut microbiota, the exclusion criteria were as follows: (1) antibiotic or probiotic use within 8 weeks; (2) PPI or metformin use within the last 3 months; (3) presence of oral or gastrointestinal diseases such as dental caries, ulcerative colitis, or acute gastroenteritis; and (4) other systemic conditions, including infections, tumors, systemic lupus erythematosus, or other autoimmune diseases. Potential confounders not controlled by eligibility criteria, including lifestyle factors, oral hygiene practices, and medication regimens, were compared across subgroups and showed no significant differences (Supplementary Table 1). All participants signed informed consent, and this study was approved by the Research Ethics Committee of Second Xiangya Hospital, Central South University and was carried out in accordance with the Declaration of Helsinki guidelines. Evaluation of clinical and laboratory variables Demographic characteristics and metabolic parameters were collected and assessed, including age, sex, duration of disease, age at diagnosis, behavioral factors (smoking and alcohol), and body mass index (BMI). Clinical parameters such as daily insulin dosage (U/kg/day), HbA1c, fasting C-peptide (FCP), advanced glycation end products (AGEs) of lens and forearm skin, and lipid profiles (total cholesterol, TC; triglyceride, TG; high-density lipoprotein-cholesterol, HDL-C; and low-density lipoprotein-cholesterol, LDL-C), systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured. Insulin resistance was assessed via the estimated glucose disposal rate (eGDR) = 24.31–12.22 × waist-to-hip ratio − 3.29 × (hypertension: 1 if present, 0 if absent) − 0.57 × HbA1c mg/kg/min. Continuous glucose monitoring (CGM) was performed over a 14-day sensor wear period for all participants with T1D using the SiJoy GS1 CGM system (Shenzhen Sibionics Technology Co. Ltd.), which has demonstrated satisfactory accuracy and usability [14].The CGM-derived metrics included the coefficient of variation (CV), time in range (TIR; 3.9–10.0 mmol/L), time above range (TAR; > 10.0 mmol/L), and time below range (TBR; < 3.9 mmol/L), all of which are expressed as the percentage of time spent in each glycemic state. In addition, the glucose management indicator (GMI) was calculated. The glucose risk index (GRI) was calculated as follows: GRI = (3.0 × TBR < 3 mmol/L) + (2.4 × TBR 3–3.9 mmol/L) + (1.6 × TAR > 13.9 mmol/L) + (0.8 × TAR 10–13.9 mmol/L) [15]. The hemoglobin glycation index (HGI) was defined as the difference between HbA1c and GMI (HGI = HbA1c–GMI). Definitions of macrovascular and microvascular complications Macrovascular damage was defined as meeting any one of the following criteria: arterial stiffness [carotid-femoral pulse wave velocity (cfPWV) > 10 m/s or brachial-ankle pulse wave velocity (baPWV) > 14 m/s], lower extremity atherosclerosis [ankle–brachial index (ABI) < 0.9], carotid hypertrophy (Carotid intima-media thickness (IMT) > 0.9 mm), or presence of carotid plaques. DR was assessed via retinal photography and defined by the presence of characteristic lesions, such as microaneurysms, exudates, or hemorrhages, in either eye. DN was defined as microalbuminuria [urinary albumin to creatinine ratio (UACR) > 30 mg/g] or as estimated glomerular filtration rate (eGFR) ≤ 60 mL/min/1.73 m2 (CKD-EPI 2021 equation for adults and the Schwartz equation for children and adolescents). Diabetic peripheral neuropathy (DPN) is diagnosed if a patient has neuropathic symptoms plus at least one abnormal test—ankle reflex, vibration, pressure, pinprick, or temperature. If symptoms are absent, at least two of these tests must be abnormal. Sample collection, DNA extraction and metagenomic sequencing Fecal samples and oral rinse samples were collected from participants at the Second Xiangya Hospital of Central South University. For fecal collection, participants were instructed to defecate directly into sterile, sealable 750 mL containers. The samples were immediately transported on dry ice to the laboratory, aliquoted into sterile screw-cap tubes, and stored at − 80 °C until further processing. For the oral rinses of the samples, the participants were asked to brush their teeth the night before removing residual food. The following morning, before any food or drink intake or tooth brushing, they vigorously rinsed their mouths with 10 mL of sterile saline for 60 s. The samples were then stored at − 80 °C until processing [16]. The DNA was extracted from both fecal and oral samples according to the manufacturer’s protocols. Concentration and purity of extracted DNA were determined with Synergy HTX and NanoDrop2000, respectively. DNA quality was checked on 1% agarose gel. Metagenomic sequencing was performed at Shanghai Majorbio Biopharm Technology Co., Ltd. using Illumina platforms. Some fecal samples were missing due to participant constipation, and some of the oral samples could not be used because of a low concentration of microbial DNA. Processing of metagenome sequencing data and taxonomic and functional annotation The raw sequencing reads were trimmed of adapters, and low-quality reads (length < 50 bp or average quality value < 20) were removed by fastp. Human-derived reads were removed by aligning to the human reference genome using the Burrows-Wheeler Aligner. The quality-filtered data were assembled using MEGAHIT. Contigs with a length ≥ 300 bp were selected as the final assembly result. Open reading frames (ORFs) from each assembled contigs were predicted using Prodigal and a length ≥ 100 bp ORFs were retrieved. A non-redundant gene catalog was constructed using CD-HIT with 90% sequence identity and 90% coverage. Gene abundance for a certain sample was estimated by SOAPaligner with 95% identity. Taxonomic annotation of non-redundant genes was performed using DIAMOND (e-value < 1e-5) against the NCBI NR database, following the de-unclassified strategy to assign sequences to the lowest confident taxonomic level. Functional annotation was conducted similarly using the KEGG database. Bioinformatics and statistical analysis Data analysis was performed using the Majorbio Cloud (www.majorbio.com) [17] and R version 4.2.2. Microbial species and pathways with a relative abundance > 0.01% and present in ≥ 10% of samples were retained for analysis. α-diversity was assessed via the Shannon index at species level, with group comparisons conducted using Kruskal–Wallis test and Dunn’s post-hoc analysis. β-Diversity analysis employed Bray–Curtis dissimilarity, visualized through principal coordinate analysis (PCoA), and statistically evaluated by PERMANOVA (999 permutations). Differential taxa and functional pathways were identified using linear discriminant analysis effect size (LEfSe). The linear discriminant analysis (LDA) effect size was calculated to estimate the relative contribution of each taxon to the differentiation between groups. Species abundance differences between groups were tested using two-tailed Wilcoxon rank-sum tests with false discovery rate (FDR) correction. Spearman correlation analysis was performed between the identified microbial species and clinical parameters. Significance was defined by a dual threshold of a P-value < 0.05 and a correlation coefficient |r|> 0.3 to ensure robust biological relevance. The FDR-adjusted p-values (PFDR) were also provided accordingly. Mediation analysis evaluated whether differential microbial features mediated the relationship between metabolic traits and complications. This was tested using a structural equation modeling framework with nonparametric bootstrapping (n = 1000) to estimate the indirect effect. Significance of the indirect effect was determined by 95% bias-corrected bootstrap confidence intervals excluding zero. All models were adjusted for sex, age, and BMI. Microbial co-occurrence networks were constructed separately for gut and oral microbiomes. Briefly, associations between microbial species were calculated using Spearman’s rank correlation coefficient. The resulting p-values were adjusted using the FDR correction. The correlation matrices were then transformed into undirected and unweighted graphs by applying specific thresholds for the correlation coefficient and statistical significance. All networks were constructed and analyzed using the NetworkX package (version 1.11) in Python. Key topological properties, including the number of nodes, the number of edges, and the average degree, were calculated to characterize the network structures. Random forest models with 500 trees were used to rank discriminative taxa by importance. A stratified tenfold cross-validation strategy was employed to tune the mtry hyperparameter and to assess the generalizable classification performance, as measured by the area under the receiver operating characteristic curve (ROC-AUC). Clinical variables were summarized as the mean ± SD, median (Q1, Q3), or percentage (categorical), with group comparisons via chi-squared test, t-test, or Wilcoxon test as appropriate. The graphical abstract was created using the BioGDP.com platform [18]. Study size A retrospective power analysis was conducted using actual metagenomic abundance data from this study. The Dirichlet-multinomial model [19] is applied via the ShinyMB online platform (https://fedematt.shinyapps.io/shinyMB/). By incorporating the actual fold-change values and using the parameters of α < 0.05 with 100 Monte Carlo replicates, the power-sample curve indicated that a minimum of 17 samples per group for gut microbiota and 14 for oral microbiota would provide 80% power. All subgroup comparisons exceeded these thresholds, indicating adequate exploratory power. This exploratory analysis aids in the interpretation of the findings within this study. Results Alterations in oral and gut Microbiota in long-duration T1D The study enrolled 75 participants with T1D and 43 healthy controls. Demographic and clinical characteristics are summarized in Table 1. Among the T1D group with a median duration of 15.6 years, 24 individuals (32%) exhibited macrovascular complications, and 35 (47%) had at least one microvascular complication. Unsupervised clustering of glycemic profiles of T1D revealed two distinct patterns: a “High-Control, High-Risk” phenotype (Cluster 1) and a “Low-Control, Low-Risk” phenotype (Cluster 2) (Supplementary Fig. 1). Table 1Demographic and clinical characteristics of the participantsCharacteristicT1D N = 751HC N = 431P value2Male (n, %)29 (39%)17 (40%)0.926Age (years)31.86 (23.23–39.12)32.21 (25.51, 40.44)0.559BMI (kg/m2)21.58 ± 2.3123.05 ± 3.080.008Smoking (n,%)18 (24%)8 (19%)0.496Drinking (n,%)13 (17%)5 (12%)0.407HbA1c (%)7.86 ± 2.035.51 ± 0.27 < 0.001Duration(years)15.58 (12.47–19.22)Age of onset14.80 (8.34–23.01)FCP (pmol/l)19.10 (16.50–26.20)Insulin dose (U/kg/d)0.62 (0.51–0.74)eGDR (mg/kg/min)9.47 (8.08–10.88)Lens-AGEs (AU)0.27 (0.17–0.33)Skin-AGEs (AU)76.00 (66.20–87.40)Macrovascular complications (n,%)24 (32%)Macrovascular complications (n,%)35 (47%)DR (n,%)28 (37%)DN (n,%)16 (21%)DPN (n,%)18 (24%)Continuous glucose monitoringCV(%)36.70 (33.38, 43.7)TBR (% of time)7.30 (5.00–11.80)TIR(% of time)69.98 (56.99–80.94)TAR (% of time)21.00 (10.90–32.49)GMI (%)6.60 (6.28–7.00)GRI46.08 (30.84–63.68)HGI0.81 (0.01–1.20)Lipid profilesTC (mmol/l)4.83 ± 0.92TG (mmol/l)0.86 (0.63–1.16)HDL-C (mmol/l)1.63 ± 0.40LDL-C (mmol/l)2.64 (2.02–3.14)Blood pressureSBP (mmHg)116.25 ± 12.72DBP(mmHg)69.37 ± 7.871n (%); Mean ± SD or median (Q1, Q3). 2Pearson’s chi-squared test; Wilcoxon test ort-test. The percentages for TIR, TAR, and TBR represent the mean percentage of time. CV is presented as a percentage (a dimensionless measure of glycemic variability)T1D, type 1 diabetes; HC, healthy controls; BMI, body mass index; HbA1c, glycated hemoglobin; FCP, Fasting C-peptide; eGDR, estimated glucose disposal rate; AGEs, Advanced glycation end products; DR, Diabetic retinopathy; DN, Diabetic nephropathy; DPN, Diabetic peripheral neuropathy; CV, Coefficient of variation; TBR, Time below range; TIR, Time in range; TAR, Time above range; GMI, Glucose management indicator; GRI, Glucose risk index; HGI, Hemoglobin glycation index; TC, Total cholesterol; TG, Triglyceride; HDL-C, High-density lipoprotein-cholesterol; LDL-C, Low-density lipoprotein-cholesterol; SBP, Systolic blood pressures; DBP, diastolic blood pressures Demographic and clinical characteristics of the participants 1n (%); Mean ± SD or median (Q1, Q3). 2Pearson’s chi-squared test; Wilcoxon test ort-test. The percentages for TIR, TAR, and TBR represent the mean percentage of time. CV is presented as a percentage (a dimensionless measure of glycemic variability) T1D, type 1 diabetes; HC, healthy controls; BMI, body mass index; HbA1c, glycated hemoglobin; FCP, Fasting C-peptide; eGDR, estimated glucose disposal rate; AGEs, Advanced glycation end products; DR, Diabetic retinopathy; DN, Diabetic nephropathy; DPN, Diabetic peripheral neuropathy; CV, Coefficient of variation; TBR, Time below range; TIR, Time in range; TAR, Time above range; GMI, Glucose management indicator; GRI, Glucose risk index; HGI, Hemoglobin glycation index; TC, Total cholesterol; TG, Triglyceride; HDL-C, High-density lipoprotein-cholesterol; LDL-C, Low-density lipoprotein-cholesterol; SBP, Systolic blood pressures; DBP, diastolic blood pressures Shotgun metagenomic sequencing revealed reduced α-diversity in both oral and gut microbiota of T1D participants compared to healthy controls (Fig. 1a–b). Community composition (Bray–Curtis dissimilarity) also differed significantly between groups (P = 0.001; Fig. 1c–d). At the phylum level, T1D gut communities were depleted in Firmicutes and enriched in Bacteroidota, whereas oral communities were enriched in Firmicutes and depleted in Proteobacteria relative to controls (Fig. 1e–f). Alterations in oral and gut microbiota in long-duration T1D. a, b Lower α-diversity in T1D. Shannon index at the species level was significantly reduced in both gut a and oral b microbiota of participants with T1D (n = 75) compared with healthy controls (HC; n = 43). *P < 0.05, **P < 0.01. c, d Distinct overall community structures. Principal-coordinates analysis (PCoA) based on Bray–Curtis dissimilarity demonstrated clear separation between groups for the gut c and oral d niches. e, f Phylum-level compositional changes. Stacked bar charts summarize the mean relative abundance of dominant bacterial phyla in the gut e and oral f Distinct microbial community structures in T1D with and without complications To investigate microbiota associations with chronic complications in T1D, patients were grouped by macrovascular and microvascular complication status. Given distinct pathophysiological mechanisms, analyses were conducted separately. In the macrovascular complication analysis, T1D participants with macrovascular complications exhibited reduced gut microbial α-diversity compared to those without (oral diversity remained unchanged; Fig. 2a–b), alongside altered relative abundances of dominant bacterial phyla (Fig. 2c–d). PCoA revealed significant separation among healthy controls, T1D without macrovascular complications, and T1D with macrovascular complications in both oral and gut microbiota (P = 0.001; all pairwise PFDR < 0.05; Fig. 2e–f). The microbiota of T1D participants with macrovascular complications showed greatest divergence from healthy controls. Similar patterns of α-diversity reduction, phylum-level alterations, and community separation were observed in the microvascular complication analysis (Fig. 3a–f). Gut and oral microbiota differ according to the presence of macrovascular complications in T1D. a, b α-Diversity decreases with macrovascular complications. Shannon index at the species level in healthy controls (HC), T1D without or with macrovascular complications (noMA, MA) for the gut a and oral b. *P < 0.05, **P < 0.01, ***P < 0.001. c, d Phylum-level compositional shifts. Stacked bars depict the mean relative abundance of major bacterial phyla in the gut c and oral d. e, f Distinct overall community structures. Principal-coordinates analysis (PCoA) based on Bray–Curtis dissimilarity revealed clear separation among HC, noMA and MA groups for gut e and oral f samples. Inset boxplots display groupwise distributions along principal coordinate axes Gut and oral microbiota differ according to the presence of microvascular complications in T1D. a, b α-Diversity declines with microvascular complications. Shannon index at the species level in healthy controls (HC), T1D without microvascular complications (noMI) and T1D with microvascular complications (MI) for the gut a and oral b. *P < 0.05, **P < 0.01, ***P < 0.001. c, d Phylum-level compositional shifts. Stacked bars depict the mean relative abundance of major bacterial phyla in the gut c and oral d. e, f Distinct overall community structures. Principal-coordinates analysis (PCoA) based on Bray–Curtis dissimilarity revealed clear separation among HC, noMI and MI groups for gut e and oral f samples. Inset boxplots display groupwise distributions along principal coordinate axes Furthermore, stratifying T1D participants by microvascular complication burden (0, 1, or ≥ 2 complications) revealing significant differences in β-diversity across severity groups (Supplementary Fig. 2). These findings suggest that microbial alterations associate with microvascular disease severity. Differentially abundant species associated with T1D complications LEfSe analysis revealed taxa differentially abundant between complication groups (Supplementary Fig. 3). In the gut microbiota, Bacteroidales were enriched in complication groups, while Eubacteriales enriched in complication-free individuals (Supplementary Fig. 3a–b). In the oral microbiota, Streptococcus was enriched in complication groups, while Neisseria and Haemophilus were enriched in complication-free individuals (Supplementary Fig. 3c–d). Differentially abundant microbial taxa were identified and ranked based on LDA effect size (Supplementary Fig. 4, the list of all results is in Supplementary Data 1). Species with an LDA score > 3 and a significant difference in relative abundance between groups (PFDR < 0.1, Wilcoxon rank-sum test) were considered discriminatory. Finally, analysis identified 24 gut and 6 oral species associated with macrovascular complications (Fig. 4a–b), and 20 gut and 5 oral species associated with microvascular complications (Fig. 4c–d). The robustness of these discriminatory taxa was confirmed by strong concordance with DESeq2 analysis (PFDR < 0.05; Supplementary Fig. 5). Differentially abundant microbial species in gut and oral microbiota grouped by vascular complication status. a, b Differential relative abundance of microbial species in the gut a and oral b microbiota of T1D participants with (MA) and without (noMA) macrovascular complications. c, d Differential species analysis for the gut c and oral d microbiota of T1D participants with (MI) and without (noMI) microvascular complications. Bars represent the relative abundance of each species in the two groups, with 95% confidence intervals indicated. Right-hand panels display the between-group difference in proportions along with significance annotations. *P < 0.05, **P < 0.01, ***P < 0.001 Cross-group comparisons showed substantial overlap of discriminatory species between macrovascular and microvascular complications. After removing overlapping species, we identified 26 gut and 8 oral species that were specifically associated with vascular complications (Table 2). Importantly, these taxa were not differentially abundant between T1D individuals and healthy controls, indicating those species represent microbial signatures linked to diabetic vascular pathology rather than T1D itself (Fig. 5a–b). Notably, complication-free T1D exhibited enrichment of key butyrate-producing gut bacteria (Blautia wexlerae, Anaerobutyricum hallii, Roseburia inulinivorans, Subdoligranulum variabile, and Fusicatenibacter saccharivorans) and specific oral Neisseria species. This protective microbial profile was markedly depleted in individuals with either micro- or macrovascular complications. Table 2The complication-associated gut and oral taxaComplication-associated gut speciesMacrovascular complications onlyMicrovascular complications onlyBoth(n = 6)(n = 2)(n = 18)Anaerobutyricum_soehngeniiBacteroides_fragilisAnaerobutyricum_halliiAnaerostipes_hadrusBlautia_stercorisBacteroides_ovatusClostridioides_difficileBacteroides_thetaiotaomicronLigilactobacillus_ruminisBacteroides_xylanisolvensPhocaeicola_coprophilusBlautia_obeumTyzzerella_nexiliBlautia_wexleraeCollinsella_aerofaciensCoprococcus_catusCoprococcus_comesDorea_formicigeneransDorea_longicatenaFusicatenibacter_saccharivoransMediterraneibacter_faecisParabacteroides_distasonisPhocaeicola_doreiPhocaeicola_vulgatusRoseburia_inulinivoransSubdoligranulum_variabileComplication-associated oral speciesMacrovascular complications onlyMicrovascular complications onlyBoth(n = 3)(n = 2)(n = 3)Streptococcus_salivariusNeisseria_elongataNeisseria_elongataHaemophilus_influenzaePropionibacterium_acidifaciensPropionibacterium_acidifaciensStreptococcus_thermophilus The complication-associated gut and oral taxa Microbial species abundance and significance across groups. Heatmaps display the relative abundance of representative microbial species in the gut a and oral b microbiota across healthy controls (HC) and T1D groups, with and without macrovascular (MA/noMA) subgroups, with and without microvascular (MI/noMI) complications subgroups. Relative abundance values are log-transformed, with red indicating higher and blue lower abundance. Circles above each column indicate statistical significance for between-group comparisons. Phylum information is shown by color-coded bars below each species name Network analysis characterized microbial signatures specific to each complication type and their co-occurrence patterns (|r|> 0.4 and PFDR < 0.05). Several gut species, such as Parabacteroides distasonis and Subdoligranulum variabile, emerged as central nodes linked to multiple vascular complications. Overall, the gut microbiota harbored more taxa strongly associated with complications than did the oral microbiota (Fig. 6). Microbial species co-occurrence networks associated with diabetic complications. Network analysis of gut and oral microbiota showing species significantly correlated (|r|> 0.4, PFDR < 0.05) with specific diabetic complications, including diabetic retinopathy (DR), nephropathy (DN), peripheral neuropathy (DPN), and macrovascular complications (DMA). Nodes represent microbial species, colored by genus. Edges indicate significant correlations. Larger node size reflects higher centrality. Gut and oral networks are shown separately, with species connected to each complication labeled accordingly Additionally, functional profiling based on KEGG metabolic annotation was performed for both oral and gut microbiomes. LEfSe analysis revealed multiple metabolic pathways (KEGG level 3, LDA score > 3) that were differentially enriched between T1D participants with and without complications (Supplementary Fig. 6). Correlations between microbiota and clinical metabolic parameters Given the role of metabolic dysregulation in diabetic complications, correlation analyses were conducted between microbial abundances and clinical metabolic parameters. Multiple gut species showed significant correlations with glycemic indices (HbA1c, TIR, and HGI). Several taxa were also correlated with eGDR and GRI (Fig. 7a, P < 0.05 and |r|> 0.3; see Supplementary Data 2 for full statistics). The differential oral species exhibited fewer correlations with clinical metabolic indicators, with only several species associated with eGDR (Fig. 7b). Analysis of microbial ecology across glycemic phenotypes revealed a clear divergence in gut microbiome composition (PCoA, R2 = 0.04, P = 0.012), which was not observed in the oral microbiome. Among the complication-associated gut species, 16 of the 26 previously identified species showed significant abundance differences between the two glycemic phenotypes, demonstrating a connection between gut microbial communities and glucose homeostasis patterns. Correlation between microbial species and clinical metabolic indicators. a Gut and b oral heatmaps show Spearman correlations between the relative abundance of representative species and clinical metabolic parameters. Correlation strength, significance, and phylum assignment are encoded by color gradients, asterisks, and side bars, respectively (* denotes significant correlation (|r|> 0.3 and P < 0.05.* P < 0.05; ** P < 0.01; *** P < 0.001). Abbreviation: BMI, body mass index; HbA1c, glycated hemoglobin; AGEs, Advanced glycation end products; CV, Coefficient of variation; TBR, Time below range; TIR, Time in range; TAR, Time above range; GMI, Glucose management indicator; GRI, Glucose risk index; HGI, Hemoglobin glycation index; TC, Total cholesterol; TG, Triglyceride; HDL-C, High-density lipoprotein-cholesterol; LDL-C, Low-density lipoprotein-cholesterol; SBP, Systolic blood pressures; DBP, diastolic blood pressures; eGDR, estimated glucose disposal rate To further explore potential relationships among the microbiota, systemic metabolism, and vascular complications, mediation analysis was conducted (Fig. 8). Several differential gut species demonstrated associations between eGDR, GRI, HbA1c, TIR, HGI, and complication status. Specifically, the abundances of Blautia obeum, Subdoligranulum variabile, Phocaeicola coprophilus, Dorea formicigenerans were associated with both glucose metabolism indicators (HbA1c, TIR, HGI) and vascular complications, showing patterns consistent with an indirect association (Fig. 8a). Similarly, certain oral species, particularly Neisseria, were associated with both eGDR and complication status in a manner supporting their involvement in the observed relationship (Fig. 8b). These findings are consistent with a model in which the long-term disease state of T1D, characterized by hyperglycemia, glycemic variability, and insulin resistance, is associated with dysbiosis of both gut and oral microbiota, which in turn is linked to the presence of vascular complications. Significant microbial mediators linking metabolism to complications. a Gut microbial species and b oral microbial species were included as mediators in models evaluating whether they mediate the effects of metabolic traits on macrovascular (MA, red) and microvascular (MI, blue) complications. Only species with significant mediation effects are shown. The x-axis represents the estimated indirect effect (a × b) with 95% confidence intervals. All models were adjusted for sex, age and BMI Additionally, most gut differential species were associated with baPWV, cfPWV, IMT, and eGFR, while only oral Propionibacterium acidifaciens was correlated with eGFR, baPWV and cfPWV (Supplementary Fig. 7). These associations further support the idea that alterations in specific gut and oral microbes may reflect or be linked to the severity of diabetic complications. Associations between oral and gut microbiota Further analyses were conducted to explore potential interactions between the oral and gut microbiota. Bray–Curtis distances confirmed distinct compositions between the two sites (Fig. 9a). Venn diagram analysis identified 61 species shared between the oral and gut compartments, including 7 that were previously identified as differentially abundant in the complication group (Fig. 9b). Correlation analysis of the top 30 shared species revealed significant intra-species correlations between sites for Enterococcus faecium, Prevotella copri and Streptococcus parasanguinis, suggesting potential oral-gut microbial transmission (Fig. 9c). Correlation analysis was also performed among complication-associated microbes from gut and oral microbiota (Fig. 9d). The results showed that the oral Propionibacterium acidifaciens was negatively correlated with gut Subdoligranulum variabile and Collinsella aerofaciens, while Haemophilus influenzae was positively correlated with Anaerobutyricum hallii. Furthermore, functional analysis suggested that species associated with complications in oral and gut microbiota shared similar functional characteristics (Supplementary Fig. 8). Collectively, these data suggest potential crosstalk between the oral and gut microbiomes in T1D, which may play a role in both the disease itself and the development of complications. Associations between oral and gut microbiota. a PCoA based on Bray–Curtis distances shows a clear separation between gut and oral microbial communities from T1D participants and healthy controls (HC). b Venn diagram showing shared microbial species between gut and oral samples with relative abundance > 0.1%. c Correlation heatmap of the top 30 shared species between gut and oral microbiota. d Correlation heatmap of complication-associated differential species from gut and oral microbiota. e Microbial co-occurrence networks within gut and oral microbiota, shown separately for participants with and without complications. Node colors represent phyla; edge colors indicate correlation direction (|r|> 0.5, P < 0.05) Microbial co-occurrence networks (|r|> 0.5 and P < 0.05) were constructed to examine intra-compartment interactions. Compared to the no-complication group, the gut microbial network in the complication group showed fewer interconnections (Fig. 9e), with a lower average degree (4.8 vs 8.2 edges/node). In contrast, the oral network in the no-complication group exhibited slightly fewer connections than that of the complication group (5.5 vs 7.6 edges/node). These findings suggest that dysbiosis in microbial co-occurrence patterns, particularly within the gut, may be associated with T1D complications. Microbiome-based identification of T1D complications To evaluate the diagnostic potential of integrating gut and oral microbiome signatures, random forest classifiers were constructed using relative abundance profiles of complication-associated microbial species. Predictive models based solely on gut microbial species achieved AUC values of 0.82 (95% CI: 0.70–0.94) for macrovascular complications and 0.77 (95% CI: 0.63–0.90) for microvascular complications. Models using only oral species demonstrated comparable performance, with AUCs of 0.70 (95% CI: 0.57–0.84) and 0.79 (95% CI: 0.68–0.90), respectively. Notably, the integration of both gut and oral microbial features significantly enhanced predictive performance, yielding an AUC of 0.89 (95% CI: 0.77–1.00) for macrovascular complications and 0.91 (95% CI: 0.82–1.00) for microvascular complications, outperforming models based on gut microbiota alone (Fig. 10). Species importance and classification performance for complication status. Receiver operating characteristic (ROC) curves and area under the curve (AUC) values from random forest models using oral, gut, or combined microbiota to classify macrovascular a and microvascular b complications. Bar plots show the top-ranked microbial species contributing to each model, based on feature importance scores. Oral species are colored in blue, gut species in orange Feature importance analysis identified the top 20 microbial contributors in each model (Fig. 10). The sets of top-ranked species differed substantially between the two complication types, suggesting that the impact of specific oral and gut microbes differs in magnitude between microvascular and macrovascular complications. Discussion This study provides evidence that dysbiosis within the integrated oral-gut microbiome ecosystem is significantly associated with the development of chronic vascular complications in individuals with long-duration T1D. Beyond confirming distinct compositional shifts in both microbial niches, we identified a core set of 26 gut and 8 oral species whose profiles correlate with complication status. Critically, the depletion of beneficial butyrate-producing gut bacteria and specific oral Neisseria species in individuals with complications suggests their potential role in maintaining vascular health. The mediation analyses revealed associations consistent with a potential mechanistic link between microbial signatures and key indices of glycemic control or insulin resistance (HbA1c, TIR, HGI, eGDR), in relation to complication risk. The demonstrable superiority of a classifier integrating both oral and gut microbial data underscores the biological relevance of their interconnection in the pathogenesis of T1D complications. These findings offer crucial new insights into the microbial mechanisms underpinning T1D vascular complications and highlight the potential for developing targeted interventions aimed at modulating the oral-gut microbiome axis. The 75% overlap in gut microbial signatures between microvascular and macrovascular complications suggests a common biochemical pathway, likely involving microbial metabolite disruption. A core finding is the depletion of key butyrate-producing taxa, including Anaerobutyricum soehngenii, Anaerobutyricum hallii and Blautia wexlerae. Butyrate is known to suppress endothelial VCAM-1 and ICAM-1 expression via NF-κB inhibition and PPARα activation, reducing leukocyte adhesion and vascular inflammation [20, 21], and sodium butyrate has been shown to ameliorate myocardial fibrosis in animal models [22]. The depletion of Anaerobutyricum soehngenii is notable, as trials in type 2 diabetes and metabolic syndrome report its association with improved glycemic control, insulin sensitivity, and cardiometabolic health [23, 24]. Similarly, Blautia wexlerae exhibits anti-inflammatory properties that mitigate type 2 diabetes and obesity [25]. Furthermore, functional profiling revealed an enrichment of the biosynthesis of cofactors pathway in the gut microbiota of patients without complications. Within this category, vitamin K2 biosynthesis is particularly relevant, as it plays a critical role in vascular health by regulating calcium homeostasis and inhibiting arterial calcification [26]. Vitamin K2 also may exert additional protective effects by inhibiting vascular smooth muscle cell apoptosis through the Gas6/Axl pathway [27]. The co-depletion of butyrate producers and this protective metabolic pathway in individuals with complications underscores a potential multi-faceted loss of microbial support for vascular integrity. Conversely, patients with complications showed higher levels of Bacteroides thetaiotaomicron. This may promote L-tryptophan metabolism toward pro-inflammatory metabolites, such as indoxyl sulfate and kynurenine, which impair endothelial function and trigger inflammation [28, 29]. Indoxyl sulfate has been shown to impair endothelial progenitor cell–mediated neovascularization [30], while kynurenine contributes to cardiomyocyte apoptosis by inducing reactive oxygen species through an aryl hydrocarbon receptor–dependent pathway [31]. Together, loss of protective butyrate-producing species and potential vitamin K2 synthesis, along with the enrichment of Bacteroides thetaiotaomicron, may create a milieu that favors endothelial dysfunction, immune activation, and vascular damage. Interestingly, the enrichment of Dorea formicigenerans and Subdoligranulum variabile in complication-free individuals aligns with reports of their beneficial associations in other autoimmune disease [32, 33]. This consistency across immune-metabolic contexts suggests that these species may confer protection against T1D complications through immunomodulatory mechanisms. There is increasing evidences linking oral health to systemic vascular outcomes [34]. Periodontitis has been shown to be significantly associated with DN, retinopathy, and cardiovascular disease [7, 8]. In the present study, a depletion of commensal Neisseria species (N. sicca, N. mucoa, N. elongatas) was identified in T1D patients with complications. While often noted for their weak opportunistic potential in immunocompromised states, these species are common commensals whose potential beneficial roles in symbiosis are frequently overlooked [35]. Supporting a protective role for Neisseria, a recent study found enrichment of this genus in individuals without coronary atherosclerosis, correlating negatively with HbA1c, arterial plaque thickness [36]. Reduced oral Neisseria abundance was also observed in type 2 diabetes [37]. Mechanistically, commensal Neisseria species in the oral cavity are efficient nitrate-reducing bacteria, contributing to endogenous nitric oxide production, a potent vasodilator and protector of endothelial integrity [38]. Several oral bacteria, including Porphyromonas gingivalis and Actinobacillus actinomycetemcomitans, have been detected in atherosclerotic plaques [39]. Mechanistically, periodontal pathogens, their toxic products, and cytokines from periodontal tissue can be detected by receptors on vascular endothelial cells [40, 41]. This triggers the release of inflammatory mediators (TNF-α, IL-1, IL-6, IL-8), promotes chemokine and adhesion molecule expression, and exacerbates endothelial dysfunction [42, 43]. These processes provide a plausible link between oral dysbiosis and vascular pathology, including platelet aggregation, endothelial injury, and bacterial translocation. This study identified an increased abundance of Streptococcus species in patients with complications. Consistent with this, Streptococcus were increased in atherosclerotic patients and correlated positively with HbA1c and TG [36]. Research has detected Streptococcus mitis in both atherosclerotic and oral plaques of elderly individuals without periodontitis, suggesting a potential role for these bacteria in vascular pathology [44]. Oral streptococci express a family of highly glycosylated serine-rich proteins that bind to platelet glycoprotein Ibα: S. sanguinis (expressing serine-rich protein A) and S. gordonii (hemagglutinin salivary antigen) [45, 46]. Such interactions may further induce platelet activation through FcγRIIA, integrin αIIbβ3, and platelet factor 4, providing additional mechanistic evidence linking oral streptococci to vascular dysfunction [47]. Functional analysis further complements these taxonomic findings, as the oral microbiota of patients with complications showed an enrichment in the HIF-1 signaling pathway. Previous studies have shown that HIF-1 signaling promotes diabetic vascular complications by inducing VEGF-driven neovascularization, and increasing the levels of inflammatory mediators that impair endothelial function [48]. This suggests that the observed oral dysbiosis may be linked to pro-angiogenic and pro-inflammatory processes that characterize advanced complications. Beyond the direct effects of oral dysbiosis, the oral microbiome may be linked to T1D complications through the oral–gut microbiome axis. Evidence implicates oral-gut microbial interactions in systemic diseases such as rheumatoid arthritis [49], type 2 diabetic coronary heart disease [50], and hypertension [51]. Crucially, a key study demonstrated that T1D-associated oral microbiome alterations can reshape the gut microbial community, notably reducing the “mouth-to-gut” transfer of Streptococcus salivarius [12]. Our analysis of shared taxa and interspecies correlations between the oral and gut microbiota supports the plausibility of crosstalk between these microbial niches. This crosstalk may contribute to the development of T1D complications. Furthermore, functional analysis suggested that complication-associated species from both niches were enriched in similar metabolic pathways, particularly glucose and fatty acid metabolism. Disturbances in glucose metabolism can induce de novo production of diacylglycerol, which, together with reactive oxygen species (ROS), activates protein kinase C. In addition, hyperglycemia promotes excessive generation of ROS and reactive nitrogen species via NADPH oxidase, leading to oxidative stress in cardiac and vascular tissues [52]. Lipid metabolic dysregulation contributes to the formation of cholesterol-rich foam cells within the arterial wall, and these lipid-laden foam cells, in concert with persistent ROS and inflammatory mediators, drive the progression of macrovascular complications [53]. Collectively, these findings highlight the importance of integrating both oral and gut microbiota when investigating complications of T1D. However, as our analyses are correlation-based, they cannot establish causality or distinguish microbial transmission from shared ecological pressures. Future work using strain-level tracking, functional assays, and metabolomic profiling will be needed to validate these mechanisms. While chronic hyperglycemia is an established driver of vascular complications in T1D, our data suggest a plausible microbiome-linked pathway. Mediation analyses revealed associations consistent with shifts in microbial taxa such as Collinsella aerofaciens and Coprococcus comes partially accounting for the relationship between glycemic control indices (HbA1c, TIR, GRI, and HGI) and complication risk in T1D. These findings are consistent with previous reports linking these bacteria to metabolic profiles [54], while our integrated analysis extends this evidence to complication-related outcomes. Furthermore, the abundance of complication-associated species correlated with quantitative clinical markers of complications, supporting a link between microbial alterations and disease severity. Previous studies in type 2 diabetes have established associations between oral-gut microbiota and diabetic complications [50]. However, T1D possesses a distinct autoimmune etiology. Emerging evidence suggests that the oral-gut microbiome axis may play a unique role in autoimmune conditions [10] and atherosclerosis [55]. Our study expands upon previous work by integrative oral and gut microbiome profiling with detailed glucose metabolism indicators and formal mediation analysis. This integrated approach provides a more comprehensive view of the interplay between metabolic control, microbial ecosystems, and end-organ damage in T1D. We also developed a diagnostic model based on multimodal microbial features that exhibited encouraging discriminatory performance (AUC 0.89–0.91). Given the non-invasive and accessible nature of microbiota sampling, such a model not only holds promise for early diagnosis but also offers potential targets for personalized therapeutic interventions. There are several limitations to this study. First, the cross-sectional design precludes causal inference, and the directions of the observed associations remain uncertain. The mediation analysis suggesting microbiota as a potential mediator requires validation in longitudinal or interventional studies. Second, although lifestyle and medication variables were considered, potential residual confounding cannot be entirely excluded. Finally, despite the strong performance of our microbiome-based classifier, the results are based on internal validation with a limited sample size and may overestimate predictive accuracy. Validation in independent, larger cohorts is therefore needed to confirm generalizability. The lack of longitudinal follow-up and functional or interventional approaches (e.g., metabolomics, cytokine profiling, probiotics, or fecal microbiota transplantation) also limits mechanistic and therapeutic insight. Future studies should adopt longitudinal and functional designs to validate mechanisms and clarify the clinical implications of targeting the microbiome in T1D complications. In summary, our integrative analysis identified distinct oral and gut microbial signatures associated with vascular complications in T1D, implicating the depletion of protective taxa and the enrichment of pro-inflammatory species. Mediation analyses further proposed that these microbial shifts may serve as a novel pathway linking established metabolic risk factors to end-organ damage. While our study provides a foundational association, addressing the causal relationship through longitudinal and mechanistic studies remains a critical next step. Conclusion This study underscores a significant association between oral-gut microbiome dysbiosis and vascular complications in long-duration T1D. The identified microbial signatures, including the loss of key butyrate-producing gut bacteria and oral Neisseria species, provide new insights into the complex pathophysiology of these complications. The integrated analysis positions the microbiome as a potential biological interface between metabolic factors and vascular health. These findings highlight the potential of targeting the oral-gut axis for developing novel prognostic and therapeutic strategies to alleviate the burden of diabetic complications. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Author contributions XLi contributed to the design and conduct of the study and the interpretation of the data. RT and MS contributed to the analysis and interpretation of data and wrote the manuscript. All authors were responsible for recruiting patients and collecting data. All authors have reviewed and approved the final version. XLi is the guarantor of this work. Funding This work was supported by the Noncommunicable Chronic Diseases-National Science and Technology Major Project (grant 2023ZD0508200 and 2023ZD0508205), National Natural Science Foundation of China (grant 82470871), Hunan Provincial Health High-Level Talent Scientific Research Project (grant R2023001),Sinocare Diabetes Foundation (LYF2022039), and the Graduate Innovation Project of Central South University (2024XQLH049). Data availability The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate This study was performed in line with the principles of the Declaration of Helsinki. Approval was granted by the Ethics Committee of the Second Xiangya Hospital of Central South University (2022–005) and was conducted in strict adherence to the guiding principles of the Helsinki Declaration. This study has been registered at ClinicalTrials.gov (ID: NCT05359796). Informed consent was obtained from all individual participants included in the study. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Rong Tang and Mei Shi have contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.