All subjects in this study were recruited from the Heping Hospital Affiliated to Changzhi Medical College. Based on the abdominal visceral fat area being ≥ 100 cm², subjects were divided into three groups: non-abdominal obesity with normal glucose metabolism (NAO/NG), abdominal obesity with normal glucose metabolism (AO/NG), and abdominal obesity with newly diagnosed T2DM (AO/T2DM) (Table 1). The T2DM diagnosis made using World Health Organization (WHO) guidelines published in 1999, with first-time diagnosis, no prior lifestyle changes, antihyperglycemic drugs, or insulin intervention, and no acute diabetes-related complications or severe liver and kidney dysfunction. The normal glucose metabolism group met the WHO 1999 criteria for normal glucose tolerance, had no family history of diabetes or history of hypertension, and had no abnormal liver or kidney function. This study was limited to individuals aged 18–60 years. Exclusion criteria for this study included:

This study was approved by The Ethics Committee of Heping Hospital Affiliated to Changzhi Medical College (Ethics approval number: (2024) 073), and all experiments adhered to the ethical principles of the Helsinki Declaration. All research participants signed informed consent forms.

This study consisted of two phases: a screening phase and a validation phase. In the screening phase, 4 gender- and age-matched subjects from each group (NAO/NG, AO/NG, and AO/T2DM) were selected for miRNA expression profile analysis using high-throughput sequencing (Table 1). In the validation phase, 18 age- and gender-matched subjects from the NAO/NG group, 14 from the AO/NG group, and 17 from the AO/T2DM group (Table 2) were included for candidate miRNA sequence verification using reverse transcription quantitative real-time PCR (RT-qPCR).

Clinical data, including name, gender, age, family history, medication history, and past medical history, were collected by fixed clinical physicians. Anthropometric parameters, including height, weight, and waist circumference, were measured following the Guideline for diagnosis and treatment of obesity (2024 edition) (15). The BMI was calculated as weight (kg) divided by height (m) squared, and the Waist-to-Height Ratio (WHtR) was calculated as the ratio of waist circumference to height. Blood pressure was recorded in mmHg while the participant was at rest, and the average of three measurements was used. Visceral Fat Area (VFA) was measured using a visceral fat analyzer (Omron HDS-2000, Japan) in a fasting, resting state. Biochemical indicators, including blood glucose, glycated hemoglobin (HbA1c), fasting insulin (FINS), triglycerides (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), aspartate aminotransferase (AST), alanine aminotransferase (ALT), creatinine (CREA), and blood urea nitrogen (BUN), were detected in the laboratory of the Affiliated Peace Hospital of Changzhi Medical College. The Homeostasis Model Assessment of Insulin Resistance (HOMA-IR) was calculated as: HOMR−IR=FINSμIU/ml×FBGmmol/L/22.5; and the Homeostasis Model Assessment of Insulin Secretion (HOMA-β) was calculated as: HOMA−β=20×FINSμIU/ml/FBGmmol/L−3.5

Fasting plasma samples were collected from patients, and exosomes were extracted using ultracentrifugation, following the method of Ye et al. (16) Western blot, Transmission electron microscopy (TEM), and nanoparticle tracking analysis (NTA) were used to verify the exosomes. The total protein of exosomes were extracted by using RIPA buffer. Protein quantification was performed using a BCA assay kit (Beyotime, China) according to the manufacturer’s instructions. Western blot analysis was performed to determine the expression of CD9 (Wuhan Sanying, China, 1:2000)、CD81 (Wuhan Sanying, China, 1:1000)、Tsg101 (Wuhan Sanying, China, 1:2000)、Calnexin (Wuhan Sanying, China, 1:5000)、FABP4 (Santa Cruz, America, 1:500)、β-actin (Wuhan Sanying, China, 1:4000). For transmission electron microscopy (TEM) analysis, the exosomes were fixed with 2% paraformaldehyde and loaded onto Formvar- and carbon-coated copper grids. The grids were then placed on 50 μl of 1% glutaraldehyde droplets for 5 min. After rinsing with ddH2O, the grids were placed on 50 μl of uranium dioxyoxalate droplets. After 5 min, the grids were placed on 50 μl of methyl cellulose droplets for 10 min. After drying for 5 min, the grids were visualized using the HT7700 transmission electron microscope (Hitachi, Japan). Furthermore, the particle size distribution of exosomes was measured by NTA using ZetaView_Particle Metrix (Particle Metrix,PMX-120, Germany) according to the instruments.

Total RNA was extracted from plasma exosomes using the Exosomal RNA Isolation Kit (Norgen Biotek, Canada). The extraction procedure was strictly followed according to the kit instructions. RNA concentration was determined using a NanoDrop 2000 spectrophotometer (Thermo Scientific, USA). The extracted RNA was used as input material to construct the sequencing library using the Small RNA Sample Prep Kit. Index codes were added, and after passing the quality check of the library, sequencing was performed using the Illumina HiSeq2500 SE500 single-end sequencing.

Raw read data generated by Illumina HiSeq2500 SE500 single-end sequencing were processed using the ACGT101-miR software. High-quality sequences (clean reads), with the 3’ adapter removed, were obtained, and length filtering was performed to retain sequences ranging from 18 to 26 nucleotides. These remaining sequences were then aligned with sequences from various RNA databases, such as the Rfam database (containing rRNA, tRNA, snRNA, snoRNA, etc.) and the Repbase database (repetitive sequence database), resulting in valid data. To exclude non-miRNA sequences, a comparison was made with the Rfam database (version 12.0) and the Repbase database (version 22.07, repetitive sequence database). Sequences matching human miRNAs were preferentially retained based on miRNA conservation, prioritizing species, with human hsa sequences being prioritized first, against the miRBase database (version 22.0). Subsequently, secondary validation was performed using the GENCODE database to ensure that all included sequences matched human genome annotations. All miRNA data from non-human species (e.g., mmu, oga, bta) were excluded to prevent cross-species sequence contamination. The specific experimental process was assisted by Hangzhou Lianchuan Biotechnology Co., Ltd.

Five candidate miRNAs were selected based on |log₂fold change (FC)| > 1, a p – value < 0.05, and relevant literature: hsa-let-7g-5p, PC-3p-13356, hsa-miR-6505-5p, hsa-miR-1229-3p, and hsa-miR-4750-5p. RNA was reverse transcribed to cDNA using a miRNA first-strand DNA synthesis kit (Sangon Biotech (Shanghai) Co., Ltd., China). Subsequently, the RT-qPCR fluorescence quantification kit (Novogene, Nanjing, China) was used to measure the expression levels of the target miRNAs. The reaction conditions were as follows: 95°C for 5 min; [95°C for 10 s, 60°C for 30 s] repeated for 40 cycles; 95°C for 15 s, 60°C for 1 min; 95°C for 15 s, 60°C for 15 s. The primers for miRNA and the internal reference U6 were synthesized by GeneWiz (Supplementary Table 1). Relative quantification of candidate miRNA expression was performed using the 2^−ΔΔCt method. Each sample was tested in triplicate.

Subsequently, we selected miRNAs consistent with the screening and validation results to predict their target genes and associated pathways. Target gene prediction for exosomal miRNAs was performed using the TargetScan (v5.5) and miRanda (v3.3a) databases. The analysis was conducted through the OmicStudio tool (https://www.omicstudio.cn/analysis/targetGene?id=4). Subsequently, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed to explore the potential biological functions and pathways associated with the target genes.

To better investigate the interactions among the target genes of differentially expressed miRNAs, we selected validated three miRNAs (hsa-let-7g-5p, PC-3p-13356, and hsa-miR-6505-5p) to construct a Protein-Protein Interaction (PPI) network. STRING database was used for the analysis, and Cytoscape software was employed for visualization. Hub gene analysis was performed using the Cytohubba plugin in Cytoscape software to identify key genes involved in the network.

Statistical analysis was conducted using SPSS 26.0 software. Normally distributed continuous data were expressed as mean ± standard deviation (Mean ± SD), and one-way analysis of variance (ANOVA) was used for comparisons among the three groups. Post-hoc pairwise comparisons were performed using LSD (Least Significant Difference) test. Non-normally distributed data were presented as median and interquartile range (P25~P75), and non-parametric Kruskal-Wallis test was used for pairwise group comparisons. Categorical data were expressed as frequencies, and chi-square (χ²) tests were applied for comparisons. Spearman correlation analysis was used to assess the relationships between variables. ROC curve analysis was performed to validate the diagnostic value of differentially expressed miRNAs. A p - value of < 0.05 was considered statistically significant.

TEM showed that exosomes from all three groups exhibited a “tea-tray” shaped, bilayer lipid membrane structure with intact morphology (Figure 1A). NTA revealed that the particle size distribution of exosomes from all three groups ranged from 50 to 200 nm, with an average size of approximately 90 nm, consistent with the characteristics of exosomes (Figure 1B). Western blot analysis comparatively examined the protein composition in whole cell lysates (CL) and plasma-derived exosomes. The samples of exosomes were found to be positive for the exosome markers, CD9, CD81, and TSG101, and negative for the endoplasmic reticulum marker Calnexin (Figure 1C). These results confirm the presence and proper preparation of plasma-derived exosomes.

To determine the potential origin of the exosomes, we further assessed the expression of the adipose tissue-specific protein-Fatty Acid Binding Protein 4 (FABP4) in the exosomes. The results showed that exosomes from all three groups contained FABP4 (Figure 1D), suggesting that the exosomes extracted in this study may originate from adipose tissue.

In the screening phase, a total of 12 plasma-derived exosome samples from the NAO/NG, AO/NG, and AO/T2DM groups were analyzed. Using |log₂(FC)| > 1 and a p - value < 0.05 as selection criteria, we found significant differential expression of 23 exosomal miRNAs in the AO/NG group compared to the NAO/NG group. Of these, 20 miRNAs were upregulated, and 3 were downregulated. In the AO/T2DM group, 8 exosomal miRNAs were significantly differentially expressed compared to the NAO/NG group, with 4 upregulated and 4 downregulated. Compared to the AO/NG group, 14 exosomal miRNAs were significantly differentially expressed in the AO/T2DM group, with 2 upregulated and 12 downregulated.

Based on this, we followed the method described by Kim et al. (14) (Supplementary Figure 1) to further screen for exosomal miRNAs specifically expressed in the AO/NG group. The results revealed 16 exosomal miRNAs that were specifically expressed in the AO/NG group, with 13 upregulated and 3 downregulated (Supplementary Table 2). Similarly, we used the same approach to screen for exosomal miRNAs specifically expressed in the AO/T2DM group (Supplementary Figure 2). We found 7 exosomal miRNAs that were specifically expressed in the AO/T2DM group, with 3 upregulated and 4 downregulated (Supplementary Table 3).

Finally, we analyzed the exosomal miRNAs that were sequentially upregulated or downregulated across the NAO/NG, AO/NG, and AO/T2DM groups. The results showed that there were 5 exosomal miRNAs with differential expression across the three groups, with 4 miRNAs being sequentially upregulated and 1 miRNAs being sequentially downregulated (Supplementary Table 4).

Based on the criteria of |log₂ (FC)| > 1, a p - value < 0.05, and related literature, we selected five miRNAs for sequence validation: hsa-let-7g-5p, PC-3p-13356, hsa-miR-6505-5p, hsa-miR-1229-3p, and hsa-miR-4750-5p. The results showed that the expression of plasma exosomal hsa-let-7g-5p and PC-3p-13356 sequentially increased from the NAO/NG group to the AO/NG group and then to the AO/T2DM group, with statistically significant differences between the groups (p < 0.05). The expression of plasma exosomal hsa-miR-6505-5p was significantly higher in the AO/T2DM group than in the AO/NG and NAO/NG groups (p < 0.01), while no statistically significant difference was observed between the AO/NG and NAO/NG groups (p > 0.05). The expression of plasma exosomal hsa-miR-1229-3p was significantly downregulated in the AO/NG group compared to the NAO/NG group (p < 0.01), but was significantly upregulated in the AO/T2DM group compared to the AO/NG group (p < 0.01). However, no statistically significant difference was observed between the AO/T2DM and NAO/NG groups (p > 0.05). In Addition, there were no differences in the expression of plasma exosomal hsa-miR-4750-5p among the three groups (p > 0.05). The expression changes of hsa-let-7g-5p, PC-3p-13356, and hsa-miR-6505-5p were consistent with the previous sequencing results (Figure 2, Supplementary Table 5).

The correlation analysis between plasma exosomal miRNAs and clinical parameters revealed the following results (Supplementary Figure 4): Exosomal PC-3p-13356 showed a positive correlation with WHtR, VFA, FBG, HbA1c, HOMA-IR, TG, and LDL-C, and a negative correlation with HDL-C. Exosomal hsa-let-7g-5p exhibited a positive correlation with WHtR, VFA, FBG, HbA1c, HOMA-IR, and LDL-C, and a negative correlation with HDL-C and HOMA-β. Exosomal hsa-miR-6505-5p was positively correlated with WHtR, VFA, FBG, HbA1c, HOMA-IR, and LDL-C, and negatively correlated with HOMA-β.

We further evaluated the potential of plasma exosomal miRNAs, namely hsa-let-7g-5p, PC-3p-13356, and hsa-miR-6505-5p, to distinguish between the AO/NG (abdominal obesity with normal glucose) group and the AO/T2DM (abdominal obesity with newly diagnosed T2DM) patients. The AUC (Area Under the Curve) values for exosomal miRNAs were as follows: hsa-let-7g-5p: 0.857, PC-3p-13356: 0.786, hsa-miR-6505-5p: 0.878. When combined in diagnostic models, the AUC values were improved: PC-3p-13356 + hsa-let-7g-5p: AUC = 0.857; PC-3p-13356 + hsa-miR-6505-5p: AUC = 0.874; hsa-miR-6505-5p + hsa-let-7g-5p: AUC = 0.882. The highest AUC of 0.891 was achieved in the model combining all three miRNAs (Figure 3A). Therefore, we propose that the quantification of plasma exosomal hsa-let-7g-5p, PC-3p-13356, and hsa-miR-6505-5p can be used as a primary or at least an auxiliary criterion for the diagnosis of Abdominal Obesity-Related Type 2 Diabetes.

Then we performed GO (Gene Ontology) and KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway analyses to better understand the biological significance of the identified miRNAs. GO analysis showed that the predicted target genes were enriched in: Biological processes: Transcriptional regulation by RNA polymerase II, signal transduction, lipid metabolism, etc; Cellular components: Cell membrane, cell nucleus, cytoplasm, exosome, etc; Molecular functions: Protein binding, RNA binding, RNA polymerase II-specific DNA-binding transcription factor activity, etc. (Figure 3B). KEGG pathway analysis revealed that the predicted target genes were significantly enriched in pathways related to T2DM, including: Autophagy, cAMP signaling pathway and ErbB signaling pathway (Figure 3C). The top 5 most significant pathways identified were: Axon guidance; Autophagy; Proteoglycans in cancer; cAMP signaling pathway; ErbB signaling pathway (Figure 3D).

Additionally, we constructed a Protein-Protein Interaction (PPI) network for the target genes of hsa-let-7g-5p, PC-3p-13356, and hsa-miR-6505-5p using the STRING database (Figure 3E). The hub gene analysis of the predicted target genes was conducted following the method described by Szydełko et al. (17), and revealed the following top 10 hub genes: PTEN; AGO2; SKP1; SMAD2; EZH2; TRRAP; PIK3CA; SUMO2; GCC1; MMUT. Among these, PTEN had the highest score (Figure 3F).