Related Experiment Videos
Serum-Proteomic Profiling Reveals Distinct Atopic Dermatitis Severity-Linked Signatures
Jag S Lally1, Takeshi Yoshida1, Shan Gao2
1Department of Dermatology, University of Rochester Medical Center, Rochester, New York, USA.
Allergy
|July 31, 2026
Summary
This study identifies nine key serum proteins that accurately distinguish severe atopic dermatitis (AD) from mild AD. These biomarkers, found through proteomic and machine learning analysis, offer potential for improved AD severity assessment.
Area of Science:
- Dermatology and immunology research.
- Biomarker discovery using proteomics and machine learning.
Background:
- Atopic dermatitis (AD) is a chronic inflammatory skin condition with varied severity.
- Identifying reliable biomarkers to differentiate AD severity is crucial for effective management.
Purpose of the Study:
- To identify novel serum protein biomarkers distinguishing mild from severe atopic dermatitis (AD).
- To leverage a serum-proteomic approach integrated with machine learning (ML) for biomarker discovery.
Main Methods:
- Serum samples from 67 adults with mild or severe AD were analyzed using Olink Explore 3072.
- Differentially expressed proteins (DEPs) were identified, and correlations with clinical variables and Th-pathways were assessed.
- Machine learning models (TMLE/SuperLearner, Boruta, MUVR) were employed to pinpoint top severity-discriminating biomarkers.
Main Results:
- 469 DEPs were found between severe and mild AD, enriched in epithelial-associated proteins.
- Subsets of DEPs correlated with lactate dehydrogenase (LDH) and Th2/Th22 pathway markers.
- Nine serum proteins (CCL17, CCL22, DEFB4A/B, EZR, GPR15L, IL22, PRSS53, SERPINB8, SETMAR) were identified as top biomarkers with high cross-validated accuracy (AUC=0.989).
Conclusions:
- Severe AD shows a distinct proteomic signature linked to epithelial proteins, tissue injury (LDH), and Th2/Th22 inflammation.
- Machine learning effectively identified robust biomarkers for discriminating AD severity in this cohort.
- Further validation in longitudinal and external cohorts, including healthy controls, is necessary to confirm biomarker utility.
