The Skin Microbiome of Patients With Atopic Dermatitis Normalizes Gradually During Treatment

Abstract

Background: Atopic dermatitis (AD) is characterized by an altered skin microbiome dominantly colonized by S. aureus. Standard treatment includes emollients, anti-inflammatory medications and antiseptics.

Objectives: To characterize changes in the skin microbiome during treatment for AD.

Methods: The skin microbiomes of children with moderate-to-severe AD and healthy children were investigated in a longitudinal prospective study. Patients with AD were randomized to receive either standard treatment with emollients and topical corticosteroids or standard treatment with the addition of dilute bleach baths (DBB) and sampled at four visits over a three-month period. At each visit, severity of AD was measured, swabs were taken from four body sites and the composition of the microbiome at those sites was assessed using 16S rRNA amplification.

Results: We included 14 healthy controls and 28 patients. We found high relative abundances of S. aureus in patients, which correlated with AD severity and reduced apparent alpha diversity. As disease severity improved with treatment, the abundance of S. aureus decreased, gradually becoming more similar to the microbiomes of healthy controls. After treatment, patients who received DBB had a significantly lower abundance of S. aureus than those who received only standard treatment.

Conclusions: There are clear differences in the skin microbiome of healthy controls and AD patients that diminish with treatment. After three months, the addition of DBB to standard treatment had significantly decreased the S. aureus burden, supporting its use as a therapeutic option. Further study in double-blinded trials is needed.

Keywords: atopic dermatitis (AD); microbiome & dysbiosis; microbiota (16S); skin; therapeutics.

Figure 1

Longitudinal unblinded study of standard treatment vs standard treatment + dilute bleach baths (DBB) in children with Atopic Dermatitis (AD). 28 children with AD were sampled at four sites (right anterior forearm, right antecubital and popliteal fossa and another lesional site chosen to represent the area of worst active AD on that patient) across four visits, spanning three months. The forearm was determined to be a likely non-lesional site on AD patients. 14 healthy controls were also recruited and sampled similarly (both antecubital fossae, popliteal fossa and forearm) at a single visit. This figure was created with BioRender.com.

Figure 2

Atopic Dermatitis (AD) sites are dominated by Staphylococcus aureus. (A) At baseline (visit 1), AD patients have a significantly higher relative abundance of S. aureus across all sites (29.56%) when compared to all sites on healthy controls (2.08%, p=1.8 x 10^-6). Within AD patients, actively lesional sites had a significantly higher relative abundance of S. aureus (36.35%) than non-lesional sites (7.20%, p=0.01). (B, C) Each datapoint represents an average across all likely lesional and lesional sites from a subject at a given visit (excludes right forearm as a non-lesional site). (B) A higher relative abundance of S. aureus correlates with higher SCORAD in AD patients (Spearman’s rho = 0.545, p=2.7 x10^-08). (C) S. aureus relative abundance inversely correlates with Shannon diversity (rho: -0.307, p=0.003). (D) The skin microbiomes of patients with high SCORAD often have a correspondingly higher relative abundance of S. aureus and smaller amounts of beneficial Staphylococcus species, though this is not always the case. Top: the relative abundances of S. aureus, S. capitis, S. epidermidis and other staphylococci. Bottom: Corresponding SCORAD values for visits. A dashed line at SCORAD = 25 indicates the cutoff above which disease is considered moderate-severe.

Figure 3

Treatment gradually shifts microbiomes of children with AD towards healthy-like microbiota. (A) The relative abundance of S. aureus decreases significantly (p=1.4x10^-5) with treatment from visit 1 (31.83%) to visit 4 (1.90%) across likely lesional sites on AD patients. The decrease in relative abundance of S. aureus is significant across treatment groups (+DBB, p<0.001, -DBB, p=0.014, Supplementary Figure 1 ). (B) By visit 4 (cyan), the composition of microbiota of likely lesional sites on AD patients is much more similar to healthy controls (dark grey) than at visit 1 (yellow), although there is variation in the severity of disease at both timepoints (PCoA on Bray-Curtis distance). (C) Severity of disease as measured by SCORAD decreases with treatment across visits.

Figure 4

S. aureus abundance but not SCORAD is lower in patients after DBB treatment. (A) The relative abundance of S. aureus decreased a significantly larger amount for patients treated with standard treatment + DBB (cyan) than those receiving standard treatment alone (yellow) after 3 months from initial visit, visit 4 (p=0.01), even though both treatment groups had similar baseline visit relative abundances of S. aureus (+DBB 25.6%; -DBB 39.13%, p=0.25). (B) Individual patient trajectories (patients are represented by grey circles, lines indicate SCORAD trajectory and gradient indicates S. aureus relative abundance) highlight that patients treated with DBB had larger decreases in the relative abundance of S. aureus than the standard treatment group, despite having similar values for both SCORAD and S. aureus abundance at the initial visit. In both groups, SCORAD exhibited a decrease over time, as evaluated by a correlation of SCORAD and timepoint (+DBB Spearman’s rho = -0.63, p=1.6 x 10^-6; –DBB Spearman’s rho = -0.59, p=4x10^-5). (C) The number of actively lesional sites per patient decreases across both treatment groups over time. Both treatment groups display a similar number of actively lesional sites at baseline visit (p=0.74), and have similarly decreased numbers of lesional sites by visit 4, reflecting improvement in condition as indicated by SCORAD. (D) In both groups, treatment results in improved patient condition, as indicated by SCORAD. At baseline, both treatment groups present with similar SCORAD values (mean +DBB: 46.6; mean -DBB: 49.6; p=0.49), which decrease with treatment by visit 4 (mean +DBB: 19.5; mean –DBB: 23.6, p=0.38).