Identification and optimized detection of two type 2 diabetes-associated enteric pathogens: Klebsiella and EPEC

Abstract

Background

Infections rank as the third leading cause of death among patients with type 2 diabetes (T2D), representing a critical clinical burden that cannot be ignored. The gut, as a major microbial reservoir, is strongly implicated in this heightened susceptibility. However, the specific enteric pathogens responsible remain poorly defined. This study, therefore, aims to identify specific enteric pathogens associated with T2D and translate these findings into an effective detection strategy by developing a robust molecular tool for clinical prevention.

Methods

We first conducted a systematic meta-analysis to identify T2D-associated gut pathogens. Then, targeted next-generation sequencing (tNGS) was performed on carefully matched stool samples to further characterize pathogen enrichment. Subsequently, findings from meta-analysis and tNGS were validated in independent clinical cohorts using qPCR analysis. Finally, we developed a dual-probe fluorescence-based PCR assay and conducted comparative evaluation between droplet digital PCR (ddPCR) and quantitative PCR (qPCR) platforms using clinical fecal samples.

Results

The meta-analysis revealed significantly increased abundance of Klebsiella and Escherichia/Shigella in T2D individuals, findings that were consistently replicated in independent validation cohorts. Subsequent tNGS analysis confirmed elevated abundance and detection rates of enteropathogenic Escherichia coli (EPEC) in the T2D group. Based on these findings, we developed a highly sensitive and specific dual-probe fluorescence-based PCR assay. Methodological comparisons demonstrated that ddPCR substantially outperformed qPCR in direct detection of target bacteria from fecal samples, exhibiting enhanced analytical sensitivity and superior resistance to fecal matrix inhibitors.

Conclusion

This study establishes a significant association between specific enteric pathogens and T2D, underscoring their putative role in diabetes-related complications. The demonstrated superiority of ddPCR enables sensitive, matrix-resistant enteric pathogen detection in T2D patients, providing a robust diagnostic tool that supports targeted diagnosis, risk stratification, and optimized clinical management of at-risk individuals.

Trial registration

National Medical Research Registration and Filing Information System MR4225044248, 13-Jun-2025.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-026-12786-w.

Introduction

Type 2 diabetes (T2D), which constitutes approximately 95% of all diabetes mellitus cases worldwide, presents a major global health challenge that extends far beyond its core metabolic dysfunctions [1, 2]. A significant component of its clinical burden is a heightened susceptibility to a broad spectrum of infectious complications. Robust evidence confirms that individuals with T2D face a significantly elevated risk of numerous common infections, including those of the gastrointestinal tract, skin, and respiratory system [3, 4], as well as specific pathogens such as Mycobacterium tuberculosis [5], Helicobacter pylori [6], hepatitis virus [7], and SARS-CoV2 [8]. This risk is further exacerbated by poorer glycemic control, which is consistently linked to a higher likelihood of infection. The profound public health impact of this relationship is underscored by the estimate that diabetes is responsible for 6% of all infection-related hospitalizations and 12% of infection-related deaths, making infections the third leading underlying cause of death in the T2D population after cardiovascular disease and cancer [3, 9].

The gut constitutes the most complex and densely populated microbial niche of the human body, frequently described as a “microbial organ” or “second brain” due to its immense diversity and metabolic influence [10, 11]. In T2D patients, the gut ecosystem exhibits a moderate degree of dysbiosis, characterized by a decline in beneficial microbes (e.g., short-chain fatty acid (SCFA)-producing bacteria) and an increase in opportunistic pathogens [12, 13]. Pathogen overgrowth may drive disease progression through two distinct pathways: by directly elevating risks of local and systemic infection, and by promoting chronic inflammation and insulin resistance through released bacterial products (e.g., metabolites and lipopolysaccharide) [14, 15]. Therefore, the identification of T2D-associated gut pathogens and the development of rapid diagnostic tools are essential for the prevention and clinical management of both T2D and infectious complications. Although 16S rRNA gene sequencing of fecal samples has revealed broad ecological shifts in the gut microbiome of T2D patients, the associations with specific gut pathogenic bacteria lack a systematic assessment.

In this study, we first conducted a systematic meta-analysis that revealed a significant increase in the abundance of Klebsiella and Escherichia/Shigella in individuals with T2D, which was subsequently validated through quantitative PCR (qPCR) analysis of independent clinical cohorts. Using carefully matched clinical samples, we further confirmed a marked elevation in the abundance and detection rate of enteropathogenic Escherichia coli (EPEC) in the T2D group by targeted next-generation sequencing (tNGS). Building on these findings, we subsequently developed a highly sensitive and specific dual-probe fluorescence-based PCR assay for the simultaneous detection of these two pathogens. Applying this assay, we demonstrated that droplet digital PCR (ddPCR) outperforms qPCR in directly detecting target bacteria in fecal samples. Together, our results establish a clear association between T2D and specific enteric pathogens while providing an optimized detection methodology with potential clinical utility for preventing intestinal infections in diabetic patients.

Materials and methods

Systematic review and meta-analysis

This PRISMA-adherent review (PROSPERO: CRD42025112485) systematically searched PubMed, Embase, Cochrane Library, CNKI, WanFang, and VIP until October 2025 for studies quantifying gut pathogens via 16S rRNA sequencing in type 2 diabetes (T2D) patients and normoglycemic controls. Two reviewers independently screened studies, extracted data, and assessed quality using the RTI Item Bank. Primary outcomes were Klebsiella and pathogenic Escherichia/Shigella abundances. The standardized mean differences (SMD) were pooled using random-effects meta-analysis. Heterogeneity, sensitivity, and publication bias were rigorously evaluated (detailed in Supplementary methods 1).

Targeted next-generation sequencing

A case-control study was approved by Taihe Hospital IRB (2025KS60). We recruited 62 T2D patients and 26 controls, from which 15 rigorously matched T2D-control pairs (n = 30) were selected for tNGS profiling. Fecal DNA was subjected to tNGS using a Gastrointestinal Pathogen Detection Kit (KingCreate) on a KM MiniSeqDx-CN platform. Pathogen detection and abundance were analyzed against a clinical pathogen database (detailed in Supplementary methods 2).

qPCR and ddPCR assay

Key findings were validated using TaqMan qPCR and ddPCR (QX200™ System, Bio-Rad) with specific primers/probes for Klebsiella pneumoniae (khe gene) and EPEC (eae gene). The resistance of each platform to fecal matrix inhibitors was assessed by spiking known DNA into fecal supernatant (detailed in Supplementary methods 3).

Statistical analysis

Analyses used R, SPSS, and GraphPad Prism. Group comparisons employed t-tests, Mann-Whitney U, Chi-square, or McNemar’s tests as appropriate. Correlations were assessed using Spearman’s rank test. A p-value < 0.05 was considered significant (detailed in Supplementary methods).

Results

Association of Escherichia/Shigella and Klebsiella with T2D revealed by meta-analysis

To evaluate the link between gut pathogens and T2D, we conducted a systematic meta-analysis based on the existing 16S rRNA sequencing studies. After literature screening and exclusion, a total of 24 eligible studies were included in the systematic analysis (Fig. 1). The methodological quality of the included studies was generally high, as assessed by the RTI item bank (Figure S1).

Fig. 1.

Study selection flow diagram. This figure illustrates the process of literature screening and study inclusion for the systematic review and meta-analysis, detailing the number of records identified, included, and excluded at each stage

In our keyword search, the term “enteric pathogen” encompasses not only taxa classically defined as obligate pathogens (e.g., Salmonella) but also commensal-derived taxa that can acquire pathogenic potential under permissive host or environmental conditions. While multiple gut pathogens were initially considered, only Escherichia/Shigella and Klebsiella were consistently reported across a sufficient number of studies to permit meta-analysis. Other gut pathogenic genera, including Pseudomonas, Yersinia, Vibrio, Citrobacter, and Salmonella, were rarely observed from 16S rRNA sequencing studies.

Escherichia coli (E. coli) and Klebsiella are typically gut commensals or potential pathogens, yet both are listed as critical priority pathogens by the World Health Organization (2017) [16]. While commensal E. coli is harmless, acquisition of virulence genes can drive diarrheal and extraintestinal diseases [17]. Klebsiella rarely causes primary intestinal infections and is more accurately categorized as an extraintestinal pathogen, with its major threat lying in severe systemic infections (e.g., sepsis, urinary tract infections, pneumonia, meningitis, pyogenic liver abscesses) [18]. Both pathogens also exhibit strong antimicrobial resistance, which complicates treatment and worsens clinical outcomes [19]. Thus, our meta-analysis focused specifically on comparing the relative abundance of these two genera between T2D patients and controls. An initial unweighted study-level mean analysis identified significant enrichment of Escherichia/Shigella and Klebsiella in T2D groups (Figure S2). Subsequent meta-analysis also demonstrated a significant enrichment of both Escherichia/Shigella and Klebsiella in the gut microbiota of individuals with T2D compared to controls. The pooled standardized mean difference (SMD) for Escherichia/Shigella was 0.34 (95% CI: 0.23–0.44, Fig. 2A), and for Klebsiella it was 0.30 (95% CI: 0.16–0.43, Fig. 2B). Both associations exhibited low heterogeneity, suggesting consistent effects across the included studies. Subgroup analyses confirmed that these associations were independent of comorbidity status (Fig. 2). These results show that Escherichia/Shigella and Klebsiella are diabetes-associated enteric pathogens that are significantly enriched in T2D.

Fig. 2.

Meta-analysis of gut pathogenic taxa abundance in type 2 diabetes. (A-B) Forest plots show the standardized mean difference (SMD) in the abundance of (A) Escherichia/Shigella and (B) Klebsiella between T2D patients and normoglycemic controls. T2D cohorts are stratified by comorbidity status: Comorbidity = 0 (T2D alone) and Comorbidity = 1 (T2D with other concurrent diseases; see Table S1B). For each study, the square represents the SMD (size corresponds to study weight), and the horizontal line represents the 95% confidence interval. For each forest plot, the summary estimate is represented by a diamond, with its width indicating the 95% confidence interval. The I² statistic quantifies heterogeneity. The standard deviation was imputed via a pooled-CV method

The robustness of these primary findings was tested through a series of sensitivity analyses. A leave-one-out analysis confirmed that no single study unduly influenced the results, as all recalculated pooled estimates remained positive and statistically significant (SMD range: 0.32–0.38 for Escherichia/Shigella; 0.27–0.33 for Klebsiella; Figure S3). Furthermore, the effect estimates were stable under different strategies for handling missing standard deviation data (Figure S4A). Contour-enhanced funnel plots showed broadly symmetric distributions and no evidence of substantial publication bias or small-study effects in Egger’s regression tests. (Figure S4B). Galbraith plots were used to profile the sources of residual heterogeneity, which identified a single study as the primary contributor to the variation in effect sizes (Figure S4C).

Targeted sequencing identifies EPEC enrichment and a skewed pathobione in T2D

Although 16S rRNA sequencing surveys the global bacterial landscape, its utility in pathogen detection is limited by insufficient sensitivity for low-abundance taxa and an inability to resolve closely related species or serotypes. To overcome these constraints, we employed tNGS, a method that integrates ultra-multiplex PCR with high-throughput sequencing for highly sensitive detection [20]. We applied this tNGS approach to a tightly matched subcohort (n = 15 per group; Fig. 3A and Table S2A) that was well-balanced for all parameters except HbA1c. This enabled the profiling of a comprehensive panel of 53 enteric pathogens (24 bacteria, 14 viruses, and 15 parasites; Table S2B) and the comparison of their presence and abundance in T2D.

Fig. 3.

Identification of the T2D-associated gut pathobione by tNGS. (A) Baseline characteristics of the matched tNGS subcohort (Control, n = 15; T2DM, n = 15). Data are presented as mean ± SD for continuous variables and n (%) for categorical variables. p-values were calculated using paired t-tests (age, height, weight, BMI, HbA1c) and McNemar’s test (sex). The complete dataset is available in Table S2A. (B) Pathogen detection rates (positivity) in matched T2DM-control pairs. Statistical significance was determined by McNemar’s test. An asterisk (*) denotes a significant difference (p < 0.05), as observed for EPEC. (C-D) The load distribution of the gut bacterial pathogen (C) and the gut parasite and virus (D). Violin plots show the absolute abundance of bacterial pathogens on a log₁₀(value+1) scale. The red solid line within each violin represents the median, and blue dashed lines indicate the first and third quartiles. Statistical significance was determined by the wilcoxon signed-rank test. *p < 0.05. Pathogens not detected in the cohort are not shown. See Table S2B for the complete quantitative dataset

Our analysis revealed a markedly higher detection rate and pathogen load of EPEC in the T2DM group compared to matched controls (Fig. 3B–C). Furthermore, several other pathogens, including diffusely adherent E. coli (DAEC), Salmonella, Cronobacter sakazakii (C. sakazakii), Bacillus cereus (B. cereus), and the parasite Blastocystis hominis (B. hominis), exhibited a consistent pattern of right-skewed distribution in the T2D group (Fig. 3C–D). Although their median loads showed no significant difference, the elevation of their upper quartiles (Q75) suggests that microbial overgrowth is primarily confined to a subset of T2DM patients with high pathogen burdens.

Increased susceptibility to severe Klebsiella and EPEC infections in T2D patients

To validate the hypothesis derived from our meta-genomic and tNGS analyses that Klebsiella and EPEC may exacerbate infections in diabetic individuals, we collected additional clinical fecal samples for further assessment. The T2D and control groups were well-matched in terms of age and gender distribution. However, the T2D group exhibited significantly higher body weight, body mass index (BMI), and as anticipated, HbA1c levels. Hematological analysis further revealed a marked systemic inflammatory profile in T2D subjects, characterized by elevated total leukocyte, lymphocyte, and most notably, monocyte counts compared to controls (Fig. 4A). This finding is consistent with previously reported profiles of chronic inflammation in diabetic patients [21].

Fig. 4.

Type 2 diabetes is associated with increased severity of Klebsiella and EPEC infections. (A) Comparison of clinical characteristics and hematological parameters between T2D (n = 62) and control groups (n = 26). Continuous variables are expressed as mean ± standard deviation or median [interquartile range], while categorical variables are presented as number (percentage). Intergroup comparisons were performed using appropriate statistical methods, with significance levels indicated by asterisks: *p < 0.05, **p < 0.01. ns, not significant. The completed and detailed information was listed in Table S3A. (B-C) Comparing bacterial load between T2D and control groups by qPCR. (B) Stacked bar graphs showing the distribution of samples stratified by bacterial load (as determined by Ct values), for Klebsiella (left) and EPEC (right) in T2D and control groups. (C) Violin plots comparing the absolute bacterial loads of Klebsiella (left) and EPEC (right) in positive samples (Ct < 35) from T2D and control groups. Statistical significance was determined by the mann-whitney U test; *p < 0.05, ns, not significant. The measured individual Ct value was listed in Table S3C

We next quantified the bacterial loads via qPCR using DNA extracted from the fecal samples. Our initial optimization focused on two key parameters to ensure robust qPCR performance. First, we validated primer specificity and established a linear dynamic range for quantification using serial dilutions of purified genomic DNA from EPEC and Klebsiella (Figure S5A). Second, to determine the optimal DNA input from inhibitor-prone clinical fecal extracts, we performed a three-fold serial dilution of a tNGS-confirmed EPEC-positive sample (PD029–P1). This revealed that 20–30 ng of total DNA input from fecal extracts provided the highest amplification efficiency (Figure S5B). This optimal input concentration was subsequently applied to all downstream clinical DNA sample testing.

We observed no significant differences in the overall prevalence of Klebsiella (87.1% vs. 84.6%, p = 0.99) or EPEC (22.5% vs. 19.4%, p = 0.77) between the T2D and control groups. However, when samples were stratified by bacterial load, the T2D group showed a significantly higher proportion of “super-high load” infections (Ct < 25) for both pathogens (Fig. 4B). Notably, lower Ct values correlate with higher bacterial abundance, indicating that T2D patients are more likely to harbor high-level colonization of these taxa. To quantitatively assess bacterial abundance, we further analyzed Ct-positive samples (Ct < 35) by converting Ct values to absolute bacterial loads using standard curves. The median bacterial load of Klebsiella was 1.36-fold higher in the T2D group than in controls (p = 0.042). A similar trend was observed for EPEC, with a 2.05-fold increase in median load among T2D patients, although this difference did not reach statistical significance (p = 0.42), likely due to the limited number of EPEC-positive controls (Fig. 4C). Taken together, these findings suggest that type 2 diabetes is associated with elevated pathogen loads and an increased likelihood of harbouring high-abundance populations of these taxa.

Development of a duplex qPCR/ddPCR assay for detecting Klebsiella and EPEC

To provide a simpler and more robust detection method, we developed a duplex PCR system using dual fluorescent probe-based primers (FAM for EPEC, HEX for K. pneumoniae) to enable simultaneous detection of these two T2D-associated pathogens via single-reaction target amplification (Figure S6A). Following optimization of primer ratio and annealing temperature, a 1:1 primer ratio and 60 °C annealing temperature were selected for all subsequent experiments, as these conditions supported efficient and balanced amplification of both targets. (Figure S6B-S6C).

The duplex PCR system could be adapted for both qPCR and ddPCR platforms. Therefore, a systematic comparison was performed to evaluate the two methods in terms of specificity, sensitivity, and quantitative accuracy. Primer and probe sequences were selected based on published literature and pre-experimental in silico analysis, which had already validated their target gene specificity across a panel of bacterial strains [22, 23]. However, interference between the two primer-probe sets required experimental verification. We found that both qPCR and ddPCR amplification plots demonstrate specific and simultaneous detection of Klebsiella and EPEC using the duplex assay, with no cross-reactivity between the two primer-probe sets or non-specific amplification observed, demonstrating the high specificity and reliability of the assay (Fig. 5A–B).

Fig. 5.

Development of a duplex PCR assay for simultaneous detection of Klebsiella and EPEC. (A–B) Specificity validation of the duplex PCR assays. (A) Shown are the qPCR amplification curves and calculated Ct value from 1) negative control (NC), (2) EPEC only, (3) Klebsiella only, and (4) EPEC and Klebsiella combined. The dashed line indicates the amplification threshold. N/D: not detectable. (B) Shown are the droplet scatter plot from ddPCR. The vertical yellow dotted lines separate the four sample groups. The pink horizontal line denotes the manually set threshold for positive/negative droplet classification. (C-D) Analytical sensitivity and linearity of the duplex PCR assays. (C) Standard curves generated by duplex qPCR. The Ct values for EPEC (FAM) and Klebsiella (HEX) are plotted against the log₁₀ of the template DNA input. The Ct threshold (Ct = 35) is used for the limit of detection (LOD) calculation. (D) Standard curves generated by duplex ddPCR. The measured target concentration (copies/μL) is plotted against the log₃ of template DNA input. The threshold for positivity (1 copy/μL) used to determine the LOD. For both panels, the calculated LOD values (in fg and converted genome equivalents (copies/μL)) are provided within the figure. The complete raw and transformed data are available in Tables S3D and S3L. (E-F). Assessing quantification accuracy for EPEC and Klebsiella in artificially spiked samples. Each panel compares the measured concentrations from the duplex assay against the known theoretical concentrations for every template combination. The accompanying relative bias was calculated as [(measured - theoretical)/theoretical] × 100%. Results are shown for (D) duplex qPCR and (E) duplex ddPCR

To evaluate analytical sensitivity, we tested serially diluted aliquots of spiked bacterial genomic DNA. Notably, these measured limits of detection (LOD) represent analytical sensitivity derived from purified bacterial genomic DNA, rather than the clinical LOD validated in complex fecal matrices. Both duplex qPCR and ddPCR exhibited a wide dynamic range and high linearity (R² > 0.99). The LOD for qPCR, defined at Ct = 35, were 178 fg for EPEC and 118 fg for K. pneumoniae, corresponding to genome equivalents of 3.3 and 2.0 copies/μL, respectively (Fig. 5C). For ddPCR, using a threshold of 1 copy/μL, the LODs were 630 fg and 152 fg, corresponding to genome equivalents of 5.75 and 1.25 copies/μL, respectively (Fig. 5D). These results indicate that both methods exhibit comparable sensitivity within one order of magnitude.

To validate the quantification performance of the duplex detection assay for mixed pathogens, spiked samples were prepared with purified genomic DNA (no fecal background) to assess intrinsic assay performance free of matrix interference. Four EPEC-K. pneumoniae concentration combinations (targets differing by several orders of magnitude per mixture) were designed, with theoretical concentrations from single-plex standard curves (Fig. 5C–D). Both duplex qPCR and ddPCR showed low relative bias ( < 15%) across nearly all template combinations, demonstrating minimal template competition and reliable quantification performance in duplex reactions (Fig. 5E–F).

Droplet digital PCR demonstrates superior performance to qPCR in the detection of EPEC and Klebsiella in T2D patient stool samples

Following the establishment of robust linearity and comparable sensitivity for both duplex assays using pure bacterial DNA (Fig. 5), we proceeded to evaluate their performance on clinical stool-derived DNA. While the probe-based qPCR and ddPCR assays both returned positive results for samples previously identified as Klebsiella-positive, their quantitative results showed poor correlation (Fig. 6A), indicating a significant discrepancy in detection efficiency when analyzing complex clinical samples. Furthermore, ddPCR detected EPEC in over 50% of samples deemed negative by qPCR, demonstrating superior sensitivity for low-abundance targets in detecting clinical DNA samples (Fig. 6B).

Fig. 6.

Comparative clinical performance of duplex qPCR and ddPCR in detecting EPEC and Klebsiella. (A) Comparison between ddPCR and qPCR for quantifying Klebsiella in fecal DNA samples. Quantification results for Klebsiella are shown for a series of clinical fecal DNA samples measured by both ddPcR and Taqman qPCR. The dashed horizontal line in the ddPCR plot indicates the positive detection threshold. The measured values were included in Table S3G. (B) Detection of EPEC by ddPCR in fecal DNA samples from T2D patients pre-selected based on a qPCR ct threshold of > 35. Scatter plot shows EPEC concentration measured by duplex ddPCR versus the corresponding pre-screening qPCR ct values for individual patient samples. The dashed horizontal line indicates the positive detection threshold for ddPCR. Samples with undetectable EPEC by qPCR were assigned a maximum Ct value of 45. Data represent mean ± SD of three technical replicates. The completed measured values were included in Table S3H. (C) Impact of stool-derived RM on the detection efficiency of duplex qPCR and ddPCR. Shown are the percentage detection efficiency for EPEC and K. pneumoniae by duplex qPCR and ddPCR with increasing volumes of residual stool matrix spiked into the reaction. Data represent mean ± SD of three technical replicates. Statistical significance was determined by Student’s t-test (**p < 0.01). The completed measured values were included in Table S3I. (D) Comparative clinical detection rates of Klebsiella in stool samples by duplex qPCR and ddPCR. The scatter plot shows the quantitative results for Klebsiella in clinical stool samples measured by both duplex ddPCR (blue dots, left y-axis) and duplex qPCR (red dots, right y-axis), plotted against the original screening qPCR ct values (x-axis) of corresponding DNA samples. The positivity threshold for ddPCR was set at 0.4 log₁₀(copies/μL). for qPCR, samples with Ct values > 35 were considered negative. Data represent mean ± SD of three technical replicates. The complete dataset is available in Table S3J

The presence of PCR inhibitors in stool matrices poses a significant challenge for molecular diagnostics. To evaluate the robustness of our assays, we spiked known quantities of EPEC and Klebsiella DNA into a residual matrix (RM) derived from a negative stool sample. The results demonstrated that while both duplex qPCR and ddPCR were inhibited by the RM, ddPCR exhibited markedly superior tolerance. Specifically, with the addition of 4 μL RM, ddPCR maintained detection efficiencies above 65% for EPEC and over 80% for Klebsiella. In contrast, the efficiency of qPCR was severely compromised, dropping below 50% for EPEC and 5% for Klebsiella under the same conditions (Fig. 6C). This robust performance was further confirmed by direct testing of stool samples from a Klebsiella-positive cohort, where ddPCR achieved a detection rate of 82.6%, vastly outperforming qPCR with a rate of only 17.3% (Fig. 6D). Collectively, these findings unequivocally demonstrate that duplex ddPCR is more robust and reliable than duplex qPCR for the direct detection of these bacterial pathogens in clinical samples.

Discussion

Identification of T2D-associated gut pathogens

Evidence regarding the role of specific gut pathogens in T2D remains limited. Our meta-analysis confirms consistent enrichment of Klebsiella and Escherichia/Shigella in T2D patients. Subsequent tNGS analysis identified enteropathogenic EPEC as a key T2D-associated pathogen and revealed a right-skewed distribution of multiple pathogens (including DAEC and Salmonella). This pattern suggests that gut environment of T2D may generally favor the colonization and overgrowth of this cluster of pathogens, which may partially explain the high incidence of gastrointestinal infections and systemic inflammation in T2D [3].

Notably, tNGS analysis demonstrated significant increases of EPEC in both abundance and prevalence of these pathogens in T2D patients. In contrast, a larger-scale qPCR study revealed no significant difference in prevalence but confirmed a marked elevation in pathogen abundance in the same cohort. This discrepancy can be attributed to two key technical differences: the smaller sample size of the tNGS cohort, and the more stringent sequence matching criteria applied in tNGS analyses. Therefore, this finding requires further validation in larger, well-matched cohorts.

Despite this caveat, our study retains important scientific significance. Given the prevalent immune compromise in T2D patients, overgrowth of these pathogens substantially elevates their translocation risk to the bloodstream or sterile sites, potentially triggering severe infections (e.g., sepsis, urinary tract infections, pneumonia) [24]. Thus, elevated intestinal loads of these pathogens may serve as a potential biomarker for heightened infection risk in T2D.

Detection platform of T2D-associated gut pathogens

Current gut pathogen detection platforms each have notable limitations. 16S rRNA gene sequencing and metagenomic sequencing enable comprehensive microbiota profiling but are susceptible to signal interference from commensal bacteria and host DNA [25, 26]. Targeted NGS improves sensitivity yet may introduce quantification biases due to uneven amplification [27]. Although ddPCR achieves high-precision absolute quantification, its clinical translation is hindered by high costs and technical complexity, a challenge compounded by its reliance on predefined, specific targets as a prerequisite for detection [28].

Our study proposes an integrated framework that combines meta-analysis-driven target selection with optimized detection protocols, aiming to provide a potential practical solution for pathogen screening in clinical settings. For the detection of the identified T2D-associated gut pathogens, ddPCR is particularly advantageous compared with qPCR. It enables reproducible, standard curve-independent absolute quantification. Furthermore, its superior inhibitor tolerance and lower limits of detection in clinical fecal samples make it especially suitable for analyzing complex biological matrices like stool.

Notably, this study quantified the absolute loads of EPEC and K. pneumoniae in stool samples exclusively via fixed-input fecal DNA analysis. Although this approach directly confirms the elevated pathogen burdens in T2D patients, it inherently overlooks confounding factors such as variations in stool water content and total bacterial biomass, leading to weak correlations among fecal mass, Cq values, and clinical severity. This limitation can be addressed by PCR-based relative quantification, which normalizes signals to universal bacterial markers (e.g., 16S rRNA gene) to distinguish pathogen expansion from commensal depletion and clarify whether alterations reflect specific overgrowth or general gut dysbiosis [29–31]. Future studies combining absolute and relative quantification could better elucidate if these load increases coincide with microbiota shifts and their correlations with key T2D clinical features (e.g., glycemic variability, systemic inflammation).

A limitation of our detection approach is its insufficient resolution, which prevents the identification of the exact pathogenic subtypes of K. pneumoniae and EPEC associated with T2D. Our meta-analysis relied on 16S rRNA gene sequencing (V3–V4 region), a method constrained to genus-level resolution that cannot even distinguish genetically similar taxa (e.g., E. coli and Shigella) (Table S1B). For tNGS, we screened a predefined panel of 53 intestinal pathogens to identify T2D-associated species (Table S2B) but performed no further subtyping. Thus, distinguishing subtypes (e.g., tEPEC vs. aEPEC) was beyond our scope. While we cannot define the specific subtypes of enriched Klebsiella and EPEC in T2D, our findings provide a critical correlative foundation for future research. Subsequent studies should deploy targeted PCR to characterize virulence/resistance gene profiles in EPEC-positive (e.g., eae, bfpA, stx) [32] and Klebsiella-positive (e.g., K1, K2 capsular genes, blaKPC, mcr-1 resistance genes) individuals [33], thus enabling precise subtyping and clarifying their pathogenic role in T2D.

Clinical implications

T2D patients have a well-documented elevated risk of intestinal infections, underscoring the necessity of sensitive enteric pathogen detection. Clarifying T2D-related pathogens, combined with our sensitive ddPCR platform, enables early, accurate pathogen identification and timely clinical decision-making. We have proposed several areas highlighting the clinical utility of this research: 1) Etiological diagnosis of infections in T2D patients. This study provides a targeted diagnostic framework for T2D patients with unexplained infection symptoms or persistent systemic inflammation despite standard treatment. 2) Prevention of severe infections in high‑risk scenarios. In clinical settings where T2D patients face elevated infection risks, such as before immunosuppressive therapy or major surgery, our assay enables targeted pathogen screening to help prevent opportunistic bacterial infections and improve patient outcomes. c) Enhanced diagnostic accuracy. The validated ddPCR assay overcomes limitations of conventional PCR, offering a fecal matrix‑resistant platform for more sensitive pathogen detection in complex clinical samples.

Unlike standardized clinical assessments (integrating symptoms, inflammatory markers, and tissue damage), our qPCR-based quantification solely reflects target pathogen abundance in fecal samples. An elevated bacterial load detected by this assay does not equate to pathogenicity or clinical infection severity. Thus, our findings do not support routine screening of asymptomatic T2D patients, nor does detection of these pathogens mandate immediate intervention. Clinical management should be stratified by pathogen load, symptomology, immune status, and differentiation between active infection and colonization. Targeted antimicrobial or microbiota-modulating therapies are only indicated for symptomatic, high-burden cases under clinical supervision, avoiding unnecessary antibiotic use. Future investigations should address this gap by recruiting larger, well-characterized cohorts with comprehensive clinical metadata, including detailed records of T2D duration, glycemic control parameters, and comorbidities.

Limitations

This study has several technical and methodological limitations that warrant consideration. First, the generalizability of our meta-analysis findings may be constrained by the geographic concentration of included studies, as the meta-analysis predominantly featured Chinese cohorts (26 of 31 studies) and a modest total sample size. Second, our clinical cohort was single-center and the ddPCR assay and pathogen prevalence require external, multi-center confirmation. Third, the observational nature of our analyses precludes causal inference-whether the diabetic milieu promotes pathogen overgrowth or vice versa remains unclear and warrants prospective and mechanistic studies. Fourth, while the primers and probes for our PCR assay were based on established targets and their theoretical specificity is supported by prior studies, we did not perform additional multi-faceted experimental validation of specificity (e.g., testing across different bacterial strains, between pathogenic and non-pathogenic E. coli, or against host intestinal epithelial DNA). We therefore acknowledge that potential unknown interferences in experimental conditions or inter-laboratory variations could affect specificity in practice, a limitation that should be considered when applying this assay. Finally, while specific pathogens were identified, their functional virulence profiles and host interactions remain uncharacterized, calling for future multi-omics and in vitro investigations.

Conclusion

This study establishes a critical link between T2D and specific gut pathogens by integrating multi-omics and molecular diagnostics. We not only confirmed the consistent enrichment of Klebsiella and E. coli/Shigella in T2D through a meta-analysis but also identified EPEC as a key associated pathogen by tNGS. Additionally, we developed a highly sensitive dual-probe fluorescent PCR system, showing that ddPCR surpasses qPCR in robustness and sensitivity for detecting these targets in complex fecal samples, providing an optimized detection solution. These findings not only deepen our understanding of gut microbiota dysbiosis associated with T2D but also lay a critical foundation for achieving early warning and precise prevention of T2D-related infectious complications.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

T2D

Type 2 Diabetes

qPCR

Quantitative Polymerase Chain Reaction

tNGS

Targeted Next-Generation Sequencing

ddPCR

Droplet Digital Polymerase Chain Reaction

EPEC

Enteropathogenic Escherichia coli

DAEC

Diffusely Adherent Escherichia coli

SCFA

Short-Chain Fatty Acid

RM

Residual Matrix

BMI

Body Mass Index

LOD

Limit of Detection

Ct

Cycle Threshold

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

SMD

Standardized Mean Difference

Author contributions

K.M. and D.X. were responsible for conceptualization, supervision, project administration, and funding acquisition. K.M., L.W., X.F., and L.S. were responsible for methodology, investigation, data curation, and writing-original draft. P.W., W.C., and J.Y. were responsible for formal analysis, software, and visualization. K.M., Y.D., and Z.C. were responsible for validation, resources, and writing―review & editing. All authors have read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (32200156), the Innovative Research Program for Graduates of Hubei University of Medicine (YC202573), the Hubei Provincial Natural Science Foundation (2025AFD201), and the Advantages Discipline Group (Medicine) Project in Higher Education of Hubei Province (2021-2025) (2025XKQY22).

Data availability

The raw tNGS data generated in this study have been deposited in the Genome Sequence Archive (GSA) of the BIG Data Center, Beijing Institute of Genomics, Chinese Academy of Sciences (https://bigd.big.ac.cn/gsa) under accession number CRA030656. Other supporting data are included in the Supplementary materials or available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Board (IRB) of Taihe Hospital (Ethics Approval No: 2025KS60) and conducted in accordance with the ethical principles of the Declaration of Helsinki. The study protocol was registered at the National Medical Research Registration and Filing Information System (https://www.medicalresearch.org.cn) on 13 June 2025 (Registration ID: MR4225044248). As this study involved this analysis of fully anonymized data using clinical waste samples obtained during standard diagnostic procedures, the requirement for informed consent was waived by the Taihe Hospital IRB. This waiver is in accordance with institutional policies and was granted to protect participant privacy.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.