Atopic dermatitis (AD) is the most common chronic inflammatory skin disorder worldwide.¹ Its pathogenesis is multifactorial,^2,3 and different AD treatments purport to focus on one or more of the pathogenic mechanisms. The most common treatment target is the disordered immune system and resulting inflammation,² but there is an emerging body of research suggesting that the skin microbiome can be manipulated for AD treatment.

A growing body of literature has reinforced the relationship between AD severity and alterations in the species distribution of the skin microbiome, specifically decreased biodiversity with predominance of Staphylococcus aureus (S. aureus) in AD.^1,4–7 Whether these microorganisms are colonizers, true pathogens, or more nuanced pathobionts is still not fully understood. Some studies suggest that AD skin is permissive to S. aureus colonization and more susceptible to S. aureus virulence.^1,5 In turn, S. aureus disrupts the epidermal barrier and induces Th2 response, both pathogenic mechanisms implicated in AD,⁵ but not all microbes have detrimental effects on the skin. Some coagulase-negative Staphylococcus species, which are deficient in lesional AD skin, produce antimicrobial peptides against S. aureus.^8,9 Other studies have found that specific bacteria such as Roseomonas mucosa,¹⁰ Vitreoscilla filiformis,¹¹ Lactobacillus johnsonii NCC 533,¹² and Streptococcus thermophilus¹³ appear to reduce inflammation and may decrease AD severity as well.

Metabolic pathways provide another potential target in altering the skin microbiome for AD treatment. Metagenomic studies suggest that alterations in the skin microbiome’s metabolism may influence AD severity.^14,15 Prebiotics are substrates in the metabolic pathway that can selectively increase or limit the population of beneficial and detrimental microbes, respectively.¹⁶ Postbiotics are the metabolites produced by the microbiome that can influence microbial diversity and inflammation.^14,15,17 Many studies demonstrate the positive effect of oral prebiotics and postbiotics on the gut microbiome in a variety of conditions and found potential benefit towards AD,^17–20 but few studies have investigated the effect of topical prebiotics on the skin microbiome of AD patients.

While the literature suggests that the skin microbiome plays an important role in the pathogenesis of AD, there have not yet been larger studies analyzing the effects of altering the metabolic pathways of the skin microbiome in the setting of AD. This systematic review aims to assess the current literature investigating the effects of prebiotics and metabolites of the skin microbiome on AD. The present study also evaluates the current literature’s quality to help guide investigators in designing and effectively conducting future studies such that protocols are optimized to reduce bias.

The selection process is outlined in Figure 1. This systematic review was conducted in accordance with PRISMA guidelines. Articles were retrieved from PubMed, Embase, and Web of Science database from the earliest records to March 2022 using the search formula: (“atopic dermatitis” OR eczema) AND ((metabolite* OR metabolom* OR metabonom* OR “metabolic profile” OR “fatty acid metabol*” OR ceramide* OR “fatty acid” OR sphingo* OR prebiotic) AND (“skin microbiome*” OR “skin flora”)) OR ((prebiotic OR oat OR glucan OR oligosaccharide) AND (topical OR emollient OR moisturizer OR lotion OR cream OR ointment)).

Figure 1.

The initial search resulted in 1,124 studies. During the initial screening, 112 duplicate studies were removed, and 906 studies were excluded. The remaining 106 studies were assessed based on the full article, and 100 articles were excluded because they did not meet the inclusion criteria. Five articles were included in this systematic review.

Four out of the five studies were clinical trials investigating the effect of topical prebiotic application on atopic dermatitis (Table 1). Of these studies, 50% (2/4) were double-blinded^22,23 and 50% (2/4) were open-label.^24,25 Only one study included a control.²² All four studies used SCORAD as the primary evaluation method of AD severity.

167896 Table 1.

Table 2 summarizes the study design for trials investigating the utility of topical prebiotics. Exclusion criteria included: severe AD,^24,25 psychological disease,²⁴ immunodeficiencies,²² pregnancy,²² and simultaneous AD treatment.^24,25 Two studies allowed concomitant treatment.^24,25 Antao et al (2017) allowed topical corticosteroid use, but Rigoni et al (2018) was more stringent and did not allow any simultaneous topical or systemic treatment except for oral antihistamines.²⁵ Seité et al (2017) only included patients with mild to moderate AD defined as SCORAD < 40 at the initial visit and excluded seven subjects if they improved less than 25% in the 15 days from screening to initial study visit.²³

167897 Table 2.

The total number of subjects in each study ranged from 22-60. Two studies enrolled subjects from a wide range of ages, including both children and adults,^23,25 but one study focused on children from 3-36 months old.²⁴ One study did not report the age of their subjects.²²

The final included study was a translational study. Gel-patch and swab were used to collect skin samples from 19 subjects with AD and 19 controls and analyzed for the concentration of indole-3-aldehyde (IAId), which is produced by the skin microbiome after metabolizing tryptophan (Trp). To study the effects of topical IAId application, a mouse model for AD was used, and MC903 was painted on mouse ears for 14 days to induce an AD-like dermatitis before treatment with IAId.²⁶

Out of 168 total subjects across four studies, only one subject discontinued the use of a face cream because of an unpleasantly strong fragrance.²⁴ Otherwise, none of the studies reported an adverse effect, and Rigoni et al (2018) actually reported that 77% (20/26) and 33% (6/26) of subjects rated the prebiotic emollient as very good and good, respectively.²⁵

The studies demonstrated varying levels of bias (Figure 2). One study was judged to not have high risk of bias in any criteria but did have moderate bias in two criteria.²² All other studies had high risk of bias in 8% (1/12) to 25% (5/20) of the graded criteria in their respective quality assessment tools.²⁴

Figure 2.

Table 3 summarizes the risk of bias for each criterion. The JBI checklist for RCT was used to assess two publications.^22,23 Both demonstrated moderate risk of selection bias in randomization and allocation concealment. Of the two publications, one also had high risk of attrition bias because they chose to analyze patients because they did not include every randomized subject in their analysis.²³

The JBI checklist for quasi-experimental studies was used to assess two publications.^24,25 Both lacked a control group and were open label. Furthermore, one study demonstrated moderate risk of selection bias in the baseline characteristics of their subjects.

Yu et al (2019) required evaluation based on both the JBI checklist for case-control and SYRCLE’s risk of bias tool because their publication included multiple study designs.²⁶ There was moderate risk of selection bias in subject recruitment, but they appropriately matched subjects in both groups. In the animal study, there was high risk of selection bias in randomization and allocation concealment, high risk of performance bias, and some moderate risk of attrition and reporting bias.

Historically, there has been a debate concerning whether AD follows an “outside-in” hypothesis or “inside-out” hypothesis.²⁷ That is to say, whether immune dysregulation causes skin barrier dysfunction or that skin barrier dysfunction is primary with resultant secondary inflammation. Regardless of which aspect may come first, it is now abundantly clear that AD is characterized by both intrinsic and extrinsic factors, and that the skin microbiome is intimately entwined with these processes. An emerging area of research focuses on the metabolic pathways of the skin microbiome and demonstrates potential treatment avenues.

Although the current literature does not generally suffer from high risk of bias, several components of their study design have the potential to introduce bias. Based on the quality assessment performed, there was a risk of bias in the selection process, determination of treatment efficacy, and statistical analyses.

167898 Table 3.

Most commonly, studies lacked controls against selection bias or did not adequately describe their protocol for selection. In two studies, the authors explicitly stated that their study was double-blind and randomized, but they did not detail how subjects were randomized.^22,23 As a result, it is not possible to determine whether the subjects were randomized and allocated into groups appropriately to prevent bias. In a third study, it is unclear if the mice were randomized at all.²⁶

Yu et al (2019) also did not mention if the investigators were blinded to whether the mice in the AD model were given treatment or not, which could lead to performance and detection bias.²⁶ However, the authors attempted to decrease detection bias by preparing more objective outcome measurements including quantitative measurements of ear and epidermal thickness.

Some studies potentially had attrition bias due to their chosen method of statistical analysis. Seite et al (2017) and Antao et al (2017) followed a per protocol (PP) rather than intention-to-treat (ITT) protocol.^23,24 With this protocol, data analysis only includes subjects who completed the treatment according to the protocol and disregards subjects who were non-adherent or lost to follow-up. Although it is acceptable to use a PP principle for statistical analysis, it does have the potential to overestimate treatment effect.

This study did not use any criteria that excluded studies based on study design as research in this field of study is relatively new and there were few studies found. Consequently, a meta-analysis was not conducted because of the variations in study design, including open label versus double-blind studies along with differences in statistical analysis methods.

The studies included in this systematic review suggest that topical prebiotics may be a valuable therapeutic modality for patients with mild-to-moderate AD. They have demonstrated potential in modifying the skin microbiome to decrease AD severity, but many questions remain about the specific prebiotic or prebiotics, dosing, administration, and optimal patient selection. Current results highlighting the influence of metabolites of the skin microbiome provide the basis for developing multiple new methods of AD treatment. However, both areas of study require more investigation. Future studies on prebiotics should aim to perform more randomized, controlled trials with larger patient populations and double blinding, and further investigation is required on the skin microbiome’s metabolism before potential treatments can be developed.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.