The nonlesional skin surface distinguishes atopic dermatitis with food allergy as a unique endotype

Abstract

Skin barrier dysfunction has been reported in both atopic dermatitis (AD) and food allergy (FA). However, only one-third of patients with AD have FA. The purpose of this study was to use a minimally invasive skin tape strip sampling method and a multiomics approach to determine whether children with AD and FA (AD FA+) have stratum corneum (SC) abnormalities that distinguish them from AD without FA (AD FA-) and nonatopic (NA) controls. Transepidermal water loss was found to be increased in AD FA+. Filaggrin and the proportion of ω-hydroxy fatty acid sphingosine ceramide content in nonlesional skin of children with AD FA+ were substantially lower than in AD FA- and NA skin. These abnormalities correlated with morphologic changes in epidermal lamellar bilayer architecture responsible for barrier homeostasis. Shotgun metagenomic studies revealed that the nonlesional skin of AD FA+ had increased abundance of Staphylococcus aureus compared to NA. Increased expression of keratins 5, 14, and 16 indicative of hyperproliferative keratinocytes was observed in the SC of AD FA+. The skin transcriptome of AD FA+ had increased gene expression for dendritic cells and type 2 immune pathways. A network analysis revealed keratins 5, 14, and 16 were positively correlated with AD FA+, whereas filaggrin breakdown products were negatively correlated with AD FA+. These data suggest that the most superficial compartment of nonlesional skin in AD FA+ has unique properties associated with an immature skin barrier and type 2 immune activation.

Fig. 1.. TEWL AUC and TEWL after sequential STS by group.

The primary end point comparisons between groups were TEWL AUC assessed on nonlesional skin of 62 participants. (A) Boxplots: Solid horizontal line and filled circle within a box represent the median and mean, respectively, the box margins are the interquartile range (50% of the observations), whisker lines extend for 1.5 times the interquartile range, and observations outside the whisker are marked by an open circle. The annotations are the P values from pairwise comparisons between groups obtained from a one-way analysis of variance (ANOVA). (B) TEWL measurements after 5, 10, 15, and 20 STS. The line figure represents means and SEs from a linear mixed model (green line represents NA, purple line represents AD FA−, and orange line represents AD FA+). Colored asterisks represent statistically significant differences: green represents a statistically significant difference between AD FA+ and NA, whereas purple compares AD FA+ and AD FA−.

Fig. 2.. Correlation of TEWL AUC with SCORAD in nonlesional skin for AD FA+ and AD FA− group.

Correlation between SCORAD and TEWL AUC in nonlesional skin stratified by AD FA+ (A) and AD FA− (B). The annotations in each panel are the Pearson correlations [95% confidence interval (CI)] and P values.

Fig. 3.. FLG breakdown products and the proportion of EOS CER content in nonlesional skin.

Comparisons between groups for FLG breakdown products total UCA (A), PCA (B), and EOS CER/NS CER ratio (C) were all assessed at skin tapes 15 and 16 on nonlesional skin. In the boxplot, the solid horizontal line represents the median, and the filled circle represents the mean. The box margins are the interquartile range, and the whiskers extend 1.5 times the interquartile range. Observations outside the whisker are marked by an open circle. The annotations are the P values from pairwise comparisons between groups obtained from a one-way ANOVA.

Fig. 4.. Staphylococcus, S. aureus, and S. hominis correlations with TEWL AUC for AD FA+, AD FA−, and NA.

Correlations between overall Staphylococcal species including S. aureus and S. hominis relative abundance (%, square root transformed) and TEWL AUC assessed on nonlesional skin from AD FA+, AD FA−, and NA. Relative abundance of Staphylococcus species were obtained by mapping the trimmed reads to reference genomes. Correlation between relative abundance and TEWL was obtained using Pearson correlation (95% CI) and P values. Linear regression models were used to estimate the best-fitting line and SE.

Fig. 5.. Keratin expression in nonlesional skin.

Comparisons between groups for markers of keratin expression measured as cumulative reporter ion signal to noise: KRT5 (A), KRT14 (B), and KRT16 (C) all assessed at skin tapes 15 and 16 on nonlesional skin. In the boxplot, the solid horizontal line represents the median, and the filled circle represents the mean. The box margins are the interquartile range, and the whiskers extend 1.5 times the interquartile range. Observations outside the whisker are marked by an open circle. The annotations are the P values from pairwise comparisons between groups obtained from a one-way ANOVA.

Fig. 6.. Dendritic cell and immune activation signatures in nonlesional skin of AD FA+ are similar to lesional AD FA+ and AD FA− skin.

Heat map and dendrogram for the transcriptome dataset (A). In this map, each row represents a transcriptome, and each column represents a participant. The cell color represents normalized levels from high (red) to middle (yellow) to low (blue), and the top column color represents the diagnostic group with appropriate annotations at the bottom (+ for AD FA+ and − for AD FA−). Boxplot for nonlesional skin (B) and lesional skin (C) of the first principal component. In the boxplots, the solid horizontal line represents the median, and the filled circle represents the mean. The box margins are the interquartile range, and the whiskers extend 1.5 times the interquartile range. Observations outside the whisker are marked by an open circle. The annotations are the P values from pairwise comparisons between groups obtained from a one-way ANOVA.

Fig. 7.. Network and relative importance analyses.

Network (A) of the intercorrelations between the measurements of TEWL (TEWL), total UCA (UCA), PCA (PCA), EOS CER/NS CER ratio (EOSNS), AD severity indices (EASI, NESS, and SCORAD), keratin expression (KRT5, KRT14, and KRT16), transcriptome PC1 (TPC1), the relative abundance of S. aureus (SAUR) and S. hominis (SHOM), and an indicator variable for AD FA+ (yes/no) based on peanut wheal size of ≥8 mm. TEWL measurements were done at STS 15; UCA, PCA and EOS CER/NS CER ratio were evaluated at STS layers 15 and 16. The color saturation and the width of the connecting lines correspond to the strength of the Pearson correlation coefficient, and the color of each connecting line indicates a positive (red) or negative (blue) correlation. Connecting lines are only shown for significant Pearson correlations (P < 0.01). Relative importance for prediction of AD FA+ (B) and TEWL STS 15 (C).