Integrated Bulk and Single-Cell Transcriptomic Analyses Nominate IER3 as a Glycolysis-Associated Candidate Marker in Diabetic Foot Ulcers

Background

Diabetic foot ulcers (DFUs) are characterized by chronic inflammation, impaired re-epithelialization, and metabolic dysregulation. However, the landscape of glycolysis-associated transcriptional alterations in DFUs remains incompletely understood.

Objectives

To prioritize glycolysis-associated candidate genes in DFUs and investigate the cellular distribution and inferred pseudotemporal expression patterns of IER3.

Design

Exploratory retrospective analysis of public bulk and single-cell transcriptomic data.

Methods

GSE134431 (13 DFU and 8 diabetic foot skin samples) was used for co-expression and differential-expression analyses. An unsigned network used power 14. Nominal differential-expression screening was supplemented by Benjamini-Hochberg sensitivity analysis. LASSO, random forest, and linear support vector machine analyses used the updated candidate pool. Conditional performance was assessed with 100 repeats of five-fold cross-validation. A fixed four-gene model was refitted using Firth logistic regression.

GSE80178 was used for secondary candidate evaluation, not independent validation. Single-cell transcriptomic analysis was performed using 14 samples from GSE165816.

Results

Two significant modules contained 5,486 genes. Overlap with 5,721 differentially expressed genes and 89 glycolysis-related genes yielded 14 candidates; FDR sensitivity analysis retained 11, including IER3. The six-gene full-data consensus included IER3, but its three-method resampling consensus frequency was 39.0%. Mean conditional out-of-fold LASSO AUC was 0.888. Apparent IER3 AUCs were 0.933 in GSE134431 and 1.000 in GSE80178. The four-gene model had a leave-one-out AUC of 0.933. Single-cell quality control retained 44,887 cells; IER3 expression varied across cell populations and inferred basal-keratinocyte pseudotime.

Conclusion

IER3 represents an exploratory glycolysis-associated candidate in DFUs rather than an established biomarker. These findings provide a foundation for future studies using larger cohorts and experimental approaches to further elucidate the role of IER3 in DFU biology.