885 Comprehensive gene expression meta-analysis of lesional scalp in alopecia areata

Background: Alopecia areata (AA) is the most common type of non-scarring inflammatory alopecia. Objectives: To identify a molecular signature of lesional scalp in AA by genes expression microarray-derived meta-analysis. Methods: A meta-analytic model was used to integrate gene expression profiles from five independent Gene Expression Omnibus (GEO) datasets (GSE68801, GSE45512, GSE80342, GSE58573, GSE74761) involving 82 AA patients and 50 controls. Both a random effect model meta-analysis and enrichment analyses of pathways were performed with imaGEO and GeneCodis, respectively. Results: We identified 1,366 genes (494 up-, 872 down-regulated) being CXCL9, CCL18, CXCL10, CD8A, and GZMB the top consistent genes (all FDR < 0.05). Transcriptome had a high representation of chemokines (CCL13, CXCL9, CXCL10, CCL18, CCL13, CCR1, XCL1), and markers of cytotoxic T lymphocytes (GZMA, GZMB, GZMH, GZMK) and NK cells (NKG2A, NKG2D). We also found significant downregulated genes that encode the majority of type I (KRT31-35, KRT38) and type II (KRT81-86) keratins and keratin-associated proteins, and other proteins that regulate the structural conformation (PADI3, GPRC5D, DSG4) or functionality (FGF18) of the follicle. Finally, we found previously unreported associations with molecular signatures of 'contact dermatitis' with/without associated atopic dermatitis, 'hypersensitivity to dust mites' and 'delayed sensitivity' to black rubber, fragrance, and nickel (all p < 0.01). Conclusions: Our study shows an integrative lesional skin transcriptome of AA derived from independent microarray gene expression studies. Targeting chemokines could be a promising therapeutic strategy for these patients.