Genome-resolved gut metagenomics identified a three-MAG signature—one Escherichia coli MAG enriched and two Collinsella MAGs depleted in Wagner 4 gangrenous diabetic foot ulcers—that discriminates advanced from less severe disease with an out-of-bag AUC of 0.703.
Key Findings
Results
Community-level gut microbiome diversity and dominant-taxon composition did not separate Wagner 4 gangrenous DFU from Wagner 1-3 disease.
Study included 43 patients with type 2 diabetes mellitus and active DFU: 30 with Wagner grades 1-3 and 13 with Wagner 4 gangrenous disease.
Shotgun metagenomic sequencing of stool samples was performed.
Authors concluded that 'advanced disease was not reflected by broad ecological restructuring.'
This finding motivated feature-level rather than community-level analysis.
Methods
De novo assembly and binning recovered 440 dereplicated metagenome-assembled genomes (MAGs) meeting quality thresholds.
MAGs met medium-quality or high-completeness/low-contamination thresholds.
These 440 dereplicated MAGs formed the basis for subsequent feature-level classification and differential abundance analyses.
Genome-resolved metagenomics was used rather than marker-gene or taxonomy-based approaches.
Results
Random forest stability selection identified 24 MAGs with reproducibly high classification importance for distinguishing Wagner 4 from Wagner 1-3 disease.
RF stability selection identified MAGs 'with reproducibly high classification importance across resampled folds.'
Two complementary approaches were combined: random forest stability selection and covariate-adjusted MaAsLin2 differential-abundance testing.
The intersection of both methods was used to prioritize the most robust candidate MAGs.
Results
A three-MAG signature—one Escherichia coli MAG enriched and two Collinsella MAGs depleted in Wagner 4—was identified as the priority discriminatory signal.
The three MAGs were 'supported by each method' (both RF stability selection and MaAsLin2 differential-abundance testing).
The E. coli MAG was enriched in Wagner 4; the two Collinsella MAGs were depleted in Wagner 4.
Abundance directions were 'concordant' across classifier interpretation, differential-abundance testing, and per-MAG abundance distributions.
This three-MAG signature achieved an out-of-bag AUC of 0.703.
The three-MAG signature 'retained much of the discriminatory information captured by the broader 24-MAG RF classifier.'
Results
The Wagner 4-enriched Escherichia coli MAG carried antibiotic-resistance and virulence-factor signals and encoded respiratory metabolic capacity.
Functional annotation 'separated the Wagner 4-enriched Escherichia coli from the Collinsella MAGs.'
The E. coli MAG carried 'antibiotic-resistance and virulence-factor signals.'
The E. coli MAG 'encoded respiratory metabolic capacity.'
These functional features were identified through annotation of the genome-resolved MAG.
Results
The two Collinsella MAGs depleted in Wagner 4 lacked detectable antibiotic resistance and virulence hits and showed metabolically compact profiles.
Collinsella MAGs 'lacked detectable resistance and virulence hits.'
Collinsella MAGs 'showed metabolically compact profiles,' contrasting with the E. coli MAG.
Both Collinsella MAGs were independently identified as depleted in Wagner 4 by both analytical methods.
Results
The E. coli-Collinsella abundance score was associated with longer DFU duration in exploratory clinical association analysis.
The association was described as 'exploratory' in nature.
The finding is 'consistent with a gut microbial correlate of chronic or advanced disease burden.'
Specific statistical values for this clinical association were not reported in the abstract.
Authors noted this links the microbial signature to duration rather than simply severity grade.
What This Means
This research suggests that the gut microbiome contains specific bacterial signatures that differ between patients with mild-to-moderate diabetic foot ulcers and those with the most severe form (Wagner grade 4, or gangrenous wounds). The researchers analyzed stool samples from 43 diabetic patients with active foot ulcers and reconstructed hundreds of individual bacterial genomes from the sequencing data. Rather than finding broad differences in overall gut microbial community structure, they discovered a specific pattern: a strain of Escherichia coli was more abundant in patients with severe gangrenous ulcers, while two strains of a bacterium called Collinsella were less abundant. This three-organism signature could distinguish severe from less severe ulcers with moderate accuracy (AUC of 0.703). The E. coli strain also carried genes associated with antibiotic resistance and the ability to cause harm to the host, while the Collinsella strains did not.
This research suggests that the gut—not just the wound itself—may harbor microbial signals that reflect how advanced a diabetic foot ulcer has become. The E. coli enriched in severe cases was functionally distinct, carrying traits associated with pathogenicity, which raises questions about whether gut bacteria contribute to or reflect systemic disease processes in diabetic foot complications. Additionally, the microbial signature appeared to correlate with how long patients had been living with their foot ulcers, hinting that it may track chronic disease burden over time.
The practical implications are still preliminary, as this was a relatively small cross-sectional study. However, the findings open the door to future research exploring whether monitoring gut bacteria could help identify patients at risk of ulcer progression, or whether changes in the gut microbiome could be used to track treatment responses. The authors themselves call for larger longitudinal studies to determine whether this gut microbial pattern changes as ulcers progress, heal, or respond to treatment.
Li L, Luo Y, Liu S, Liang C, Ruan H, Peng P, et al.. (2026). Genome-resolved gut metagenomics identifies an Escherichia coli-Collinsella signature associated with Wagner 4 gangrenous diabetic foot ulcers.. Frontiers in immunology. https://doi.org/10.3389/fimmu.2026.1893357