These data support partial concordance in ulcer-associated epithelial, inflammatory and metabolic expression patterns across diabetic foot ulcer cohorts, but limited cohort coverage, comparator differences and unresolved confounding preclude a universal disease signature, healing predictor or causal interpretation.
Key Findings
Results
A substantial fraction of differentially expressed genes from the primary RNA-seq cohort were replicated in a microarray cohort, with direction and significance concordance.
The primary analysis (13 ulcer vs 8 diabetic foot skin samples) selected 4,704 differentially expressed gene identifiers; 2,943 were represented among the 17,502 gene symbols shared with the microarray cohort.
Of these, 1,254 had concordant directions and nominal microarray P < 0.05.
1,028 of these also had microarray false discovery rate < 0.05.
The microarray cohort comprised six ulcer and three diabetic foot skin samples.
Results
Cross-platform agreement in effect direction was moderate and effect-size rank correlation was modest.
Direction agreement was 70.2%.
Overall effect-rank correlation was modest (Spearman ρ = 0.29).
The authors characterize the findings as partial concordance.
Results
Five Hallmark gene sets met the cross-cohort enrichment criteria.
The criteria were applied across the RNA-seq and microarray cohorts.
A stricter supported gene set was analyzed further for specific functional enrichment.
Results
Functional analysis of the stricter cross-cohort supported gene set highlighted cornified-envelope formation, keratinization and inflammatory-cell migration.
This points to epithelial and inflammatory processes among the concordant signals.
The abstract also describes metabolic expression patterns as partially concordant.
Results
In an additional cohort with a different comparator (healthy foot skin), most represented genes from the supported set agreed in direction.
The additional cohort had 10 ulcer and 11 healthy foot-skin samples.
611 of 823 represented genes from the set agreed in direction (about 74%).
This was a descriptive directional comparison only, and the comparator differed from the primary analysis.
Results
Twelve keratinocyte genes met the cell-type differential-expression threshold in single-cell data.
Single-cell data from 26 participants were analyzed after aggregating gene counts within each patient and cell type (pseudobulk approach).
The authors state that this finding does not establish cell-type exclusivity.
Results
Healing-status comparisons were exploratory and sensitive to covariate adjustment.
Results changed depending on covariate adjustment.
The authors therefore do not support use of the findings as a healing predictor.
Conclusions
The authors conclude the data do not support a universal disease signature, healing predictor, or causal interpretation.
Limitations cited include limited cohort coverage, comparator differences and unresolved confounding.
Small sample sizes, particularly in the microarray cohort (9 samples), are part of the study design.
What This Means
This study asked whether gene activity patterns seen in diabetic foot ulcer skin are consistent across different studies. The researchers compared an RNA sequencing dataset with a separate microarray dataset, and also looked at single-cell data and an additional dataset using healthy foot skin as the comparison.
They found partial agreement. Of 2,943 genes that could be compared, 1,254 changed in the same direction and were statistically supported in the second dataset. Overall, about 70% of genes moved in the same direction, but the correlation in how strongly they changed was modest. The consistent signals pointed to skin-barrier formation (keratinization, cornified envelope), inflammatory-cell movement, and some metabolic processes. A similar directional pattern appeared in the additional healthy-skin comparison, and a small set of keratinocyte genes stood out in single-cell data.
This research suggests some molecular features of diabetic foot ulcers recur across studies, but the sample sizes were small, the comparison groups differed, and confounding could not be ruled out. Results related to healing were not stable when adjusted for other factors. The findings therefore do not establish a universal gene signature, a test to predict healing, or a cause of ulcers.