Effects of Two Different Basic Skincare Regimens on Children and Adult Skin Microbiota: An Exploratory Randomized Controlled Crossover Trial Open Access

Atopic dermatitis (AD) is a chronic inflammatory skin condition marked by dry skin, intense itching, and inflammatory lesions, with symptoms that can fluctuate between alternating periods of recurrent flares, dry skin, and symptom-free intervals [1‒5]. It affects up to 20% of children and 2–10% of adults, making it one of the most common inflammatory skin diseases [6]. The pathogenesis involves genetic and environmental factors, skin barrier dysfunction, microbial imbalance, and immune dysregulation [4, 5, 7]. AD therapy focuses on anti-inflammatory treatment and strengthening the skin barrier [4].

In recent years, the importance of microbial colonization and the microbiome in both healthy and diseased skin has gained prominence. Specifically, increased colonization by Staphylococcus aureus has been correlated with more severe cases of AD [8]. However, it is not yet clear whether changes in the skin microbiome trigger inflammation or are a consequence of it [9]. The application of leave-on skincare products, including emollients and moisturizers, is common in basic AD therapy. Products are intended to strengthen the skin barrier, but they may also influence the skin microbiome [8, 10‒12]. However, product-specific effects on the skin microbiome are unclear. “Basiscreme DAC” (DAC) is a widely used petrolatum-based lipid-rich emulsion as an interval therapy alternating with topical products containing active agents in patients with AD [13]. “Bepanthen SensiDaily^®” (BSD) (Bayer Consumer Care AG, Basel, Switzerland) is a dexpanthenol-containing leave-on product developed for sensitive and dry and AD skin. Previous research suggests that BSD may increase stratum corneum hydration, strengthen the skin barrier, and be helpful as maintenance therapy for infants with mild AD [14, 15].

A single-center, prospective, randomized, controlled, exploratory crossover trial clinical study was conducted to answer the following questions about mild AD in children and adults: Does using topical leave-on products affect skin microbial colonization and diversity? Are these changes product-specific? Are there differences in clinical signs between treatments? How are changes in microbial colonization, skin barrier function, and clinical parameters related?

Participants were aged 2–12 and 18–65 years. Parents/guardians and adults provided informed consent. AD diagnosis followed Hanifin and Rajka criteria, with mild cases defined as an Eczema Area and Severity Index (EASI) of less than seven [16]. Participants were required to be in good health, confirmed by medical history and physical examination, and to avoid intensive sun exposure and tanning beds during the study. For detailed in-/exclusion criteria, see online supplementary material (for all online suppl. material, see https://doi.org/10.1159/000545433).

Participants were randomized into two groups (shown in Fig. 1): Group A used BSD for 3 months, followed by DAC for 3 months or Group B used DAC for 3 months, followed by BSD for 3 months with visits at baseline, month 3, and month 6. Blinding of subjects and parents/caregivers was not possible due to the nature of the interventions. However, investigators and outcome assessors, including for microbiome analysis, were blinded throughout the study. The data manager remained blinded during data entry, verification, and until database closure.

The product was applied to the whole body of all participants once or twice a day. On the measurement areas, the center of the inner sides of the forearms and the center of the calves’ subjects were asked to apply twice a day. A two-minute wait time before dressing was prescribed to prevent immediate wear-off. Usual skin cleansing routines, including bathing and showering products, remained unchanged throughout the study.

At all visits, microbial swab samples and skin physiology measurements were performed. The latter included transepidermal water loss (TEWL) using the Tewameter TM 300 (g/m²/h), stratum corneum hydration (SCH) using the Corneometer CM 825 (arbitrary units), and skin surface pH using the Skin-pH-Meter^® PH 905, all using equipment by Courage + Khazaka, Cologne, Germany.

Skin swab samples were collected from dry skin on the forearm or lower leg (5 × 5 cm) using sterile, DNA-free water-moistened swabs for 30 s with constant pressure. Control swabs without any skin contact were also included. Samples were stored at 4°C until transported for analysis. DNA extraction, 16S rRNA gene PCR amplification, and microbiome analysis via next-generation sequencing (NGS) were performed. Briefly, the first 500 base pairs of the 16S rRNA gene, including variable regions V1–V3, were amplified using pan-bacterial primers and NGS performed at a commercial sequencing facility (Microsynth, Switzerland). Sequencing data were analyzed using the Centroid database of the SmartGene program [17].

At each visit, subjects were actively asked for undesirable effects (UE), meaning an adverse reaction to human health attributable to the normal or reasonably foreseeable use of a cosmetic product representing a skin or systemic reaction [18]. These were documented, with severity and product relationship assessed by the investigator.

The study planned to enroll 50 subjects with mild AD, anticipating a 10% loss to follow-up, aiming for about 45 subjects to complete the study. No prior sample size calculation was conducted due to the study’s exploratory nature. Separate randomization lists were created for children and adults. Block randomization was utilized with a 1:1 allocation, and block lengths were set at six for both age groups. A block size of six was chosen, to protect against predicting the treatment sequence but to achieve a balanced allocation. An independent team prepared sequentially numbered, sealed, opaque envelopes. After informed consent and baseline measurements, these envelopes were opened by a research assistant to assign subjects to either Group A (BSD-DAC) or Group B (DAC-BSD).

Participant demographic characteristics were described using frequencies, means and standard deviations (SD) for continuous outcomes, while ordinal scaled outcomes were presented using medians and interquartile ranges. The differences between the two skincare products were described using means, standard deviations and two-sided 95% confidence intervals. Because of the exploratory design, all group comparisons are descriptive. In this exploratory study, no correction for multiple statistical testing was made. Therefore, p values should be interpreted with caution, and only changes with p values clearly below 0.05 were considered meaningful indicators of potential treatment effects. For the analysis of the microbiome data, principal coordinate analysis (PCoA) of Bray-Curtis dissimilarities and permutation analysis of variance (PERMANOVA) were used to assess differences in β-diversity. This analysis was performed with the R programming language (version 4.2.3) using the ggplot2 and vegan packages. Spearman correlation analyzed the relationship between microbial changes and skin parameters on the right forearm after 3 months.

Compliance was calculated as the percentage of actual applications performed out of the total possible applications, documented by participants in a diary. Diaries were checked, and relevant information entered into the database by the study assistant during visits 2–5.

The first participant was enrolled in December 2020, and the last participant completed the study in December 2022. Recruitment paused from December 16, 2020, to February 28, 2021, due to COVID-19 restrictions. The study predominantly occurred during the pandemic, with recruitment challenges stemming from intermittent university screening restrictions and cautious participant behavior avoiding contact.

Fifty-three participants underwent screening, with three excluded for not meeting inclusion criteria. Subsequently, 50 participants were randomized: 24 to Group A and 26 to Group B. All received planned interventions. The study flow is detailed in online supplementary material 2. In Group A, four subjects discontinued during the first treatment period: two due to treatment-related UE and two due to consent withdrawal. In Group B, five subjects did not complete the first treatment period: three due to UE, one due to consent withdrawal, and one lost to follow-up. Group A had 19 subjects completing both treatment periods, with one withdrawal during the second period due to UE. Group B had no withdrawal in the second period and 21 subjects completed both periods and were analyzed accordingly.

Table 1 presents demographic and baseline characteristics. Functional skin parameters and EASI scores were similar between subjects who completed the study and those who withdrew prematurely. Demographic characteristics and skin parameters at study onset were comparable across all age groups and study terminations. Online supplementary material 3 shows demographic and baseline characteristics by group, showing similar sex, ethnic status, and skin phototype characteristics, as well as comparable baseline TEWL, SCH, and pH values on the forearms and legs. Group A started with a higher median baseline EASI of 2.9 compared to Group B with a median of 1.7.

Table 2 displays skin functional parameter results for adults and children, showing similar overall skin barrier characteristics and EASI scores between groups. In children, SCH was higher after BSD treatment (36.2 [SD 7.0] a.u.) compared to DAC treatment (32.9 [SD 6.9] a.u.) on the forearms (P 0.03). On the lower leg, TEWL was similar between treatment groups in children, whereas in adults, TEWL was lower after BSD treatment (9.2 [SD 6.7] g/m²/h) compared to DAC treatment (12.4 [SD 10.8] g/m²/h).

Overall treatment compliance was very high, with 97% in Group A (minimum 81%) and 99% in Group B (minimum 96%). Nineteen random control swab samples out of 56 were analyzed by PCR amplification, giving negative results except one. Samples from this sampling day were excluded from further evaluation. Control samples underwent NGS to establish the background microbiome from kit reagents and handling. PCR results were obtained for all patients except one, though not all swabs were positive, resulting in 61 microbiomes from 35 patients.

Baseline microbiomes significantly differed between children and adults (p < 0.001), but not between males and females (p = 0.639), different body regions (p = 0.608), or treatment groups (BSD-DAC and DAC-BSD) (p = 0.157) (shown in Fig. 2). PCoAs also showed no significant differences in overall skin microbiomes between BSD and DAC after treatment for adults and children. Additionally, comparing BSD or DAC treatments to baseline microbiomes in adults and children revealed no significant differences (online suppl. material 4A–F). Alpha diversity, ranging from 0.95 to 3.6, was similar in groups A and B and after BSD and DAC treatments, indicating comparable diversity of features or species across these groups.

The analysis of microbial genera and species on the forearm and lower leg was conducted separately for children and adults for Paracoccus sp., Cutibacterium sp., Roseomonas sp., Staphylococcus sp., and S. aureus. The analysis showed shifts from baseline and between treatments (shown in online suppl. material 5, 6). In children, Paracoccus sp. was more frequent after BSD or DAC treatment for 3 months compared to baseline. Cutibacterium sp. was more common in adults than children. Staphylococcus sp. was less prevalent in children than in adults under both treatments. S. aureus was found at very low levels at baseline and post-treatment in both groups (online suppl. material 6).

Correlations between microbial colonization and clinical parameters, as well as median changes in bacterial genus percentages after 3 months, are shown in Supplementary material 5 (children) and 6 (adults) only for the lower forearm. Varying correlations between bacterial genus and clinical parameters under BSD and DAC in both children (n = 11) and adults (n = 15) were observed. In children, BSD showed positive correlations with EASI for Brachybacterium, Nocardioides, and some environmental bacteria (Massilia, Microbacterium, Oxalobacteraceae). Using DAC, there was a positive correlation with Corynebacteriaceae. Adults exhibited fewer correlations with skin parameters after 3 months of DAC and BSD compared to children. Median percentage reads after 3 months differed by intervention and age group (see online suppl. material 7 + 8).

In Group A, 14 UEs occurred in 11 subjects, while in Group B, there were 10 UEs in eight subjects. In total, there were 24 UEs in 19 subjects (see online suppl. material 9). The main symptoms reported by the subjects were burning in three subjects using BSD compared to two subjects using DAC. In addition, pruritus alone, as well as generalized and eczema were reported by the subjects or seen in the physical examination. Pruritus and eczema were also the reasons for two subjects using BSD and three subjects using DAC to end the study early.

In this study, sample characteristics in both groups (A+B) as well as in children and adults were comparable, showing a mean EASI of 2.3 at baseline, indicating mild AD [19]. The baseline TEWL as well as SCH values were similar to those in other studies in children and adults with mild AD or xerosis cutis on non-lesional skin [20]. Similar to previous research in subjects with dry skin, we also demonstrated a lower SCH on the lower legs compared to the arms [21, 22]. Mean pH values on the forearms and on the lower legs were also similar to previous studies [22].

After 3 months’ application of either study products, there was an increase in SCH, both in the forearms and the lower leg, indicating skin barrier-enhancing properties of both lipophilic leave-on products. The size of the group differences was similar to previous studies using leave-on products to treat dry skin [23]. The TEWL did not show major changes after 3 months of product application. As the measured inner sides of the forearms and lower legs were not eczematous in most of the subjects, these results are not surprising. The improved EASI in children and adults after 3 months shows that both lipophilic leave-on products had a positive therapeutic effect on AD.

Few group differences after 3 months of treatment were observed. SCH on the forearm in children was higher after 3 months of BSD application compared to DAC. Similarly, TEWL on the lower leg in adults was higher after 3 months of BSD compared to DAC treatment. This improvement in hydration by BSD compared to DAC may be due to the properties of BSD as an emollient plus compared to petrolatum-based DAC. In studies with other emollient plus products compared to DAC, an improved effectiveness of the former in skin barrier function could be demonstrated [24]. Regarding all other TEWL, SCH parameters, as well as pH and EASI treatment group differences were minor.

Microbiome analysis following treatment with BSD or DAC suggests no significant changes in the overall composition of the skin microbiome. This could be due to the stability of the pH value, which impacts bacterial growth, remaining within a physiological range. This aligns with Hülpüsch et al. [25] where S. aureus abundance correlated with skin pH values between 5.7 and 6.2. A constant pH may have contributed to stable microbial communities on the skin [26]. Although an increase in microbiome “richness” after treatments was expected, it was not observed, possibly due to the diversity already present at baseline [17].

There were no signs of typical microbial dysbiosis at baseline, characterized by reduced diversity or overrepresentation of S. aureus [27]. Roseomonas sp. and Cutibacterium sp., known to inhibit S. aureus and act as “probiotics” for the skin in AD, were already present at baseline. The microbial diversity of the skin was maintained under both interventions, reflected in stable or decreasing EASI values, indicating no deterioration in AD. In patients with active AD, microbial dysbiosis and reduced diversity with overrepresentation of S. aureus correlate with severity [27]. In our study, S. aureus occurrence was low at baseline and after 3 months, likely due to mild forms of AD.

The type of product used could influence skin microbiota growth [28]. The effects of BSD and DAC after 3 months on different bacterial genera and species vary depending on the swab localization (forearm only or forearm/lower leg combined) and also on children compared to adults. The resilience of the skin microbiota could explain why no significant changes were observed despite the treatments [29]. This resilience may be a natural protective mechanism, maintaining the stability of microbial communities on the skin.

The incidence of UEs appears relatively high for both products, but that is probably due to the fact that all subjects were explicitly asked about local intolerances from the first visit and asked to report if they noticed any symptoms such as burning, itching or rash, even if these were minor. The number of UEs was comparable to previous studies, and the types of UE are typical for the use of cosmetic products [5]. Furthermore, since most of the subjects perform basic skincare at home, which they tolerate well, they possibly decided faster, to stop the study and to continue with their established skincare regimen.

Due to the nature of the interventions, blinding of the participants and/or parents/caregivers was not possible. Despite this challenge, blinding of the assessors and evaluators, including the analysis of microbiome samples, was maintained throughout the entire study. The data manager remained blinded until the closure of the database, ensuring the integrity of data input and verification.

Another limitation, especially related to the microbiome analysis, concerns the small sample size due to technical challenges. Due to the nature of skin sampling, PCR analysis yielded negative results for some study participants, especially in children. Consequently, subsequent NGS analyses could not be conducted for all subjects. This may be attributed to a portion of the skin microbiome located beneath the skin surface and around hair follicles. This issue affected the interpretation of microbial colonization and changes in clinical skin parameters after 3 months of intervention (online suppl. material 5, 6).

A notable limitation of the study is the potential impact of social isolation during the COVID-19 pandemic on the microbiome, which could act as a confounding factor. Social isolation and changes in lifestyle associated with the pandemic may have influenced participants’ microbiomes in ways not accounted for in this study. This factor could affect the interpretation of changes in microbial colonization and clinical skin parameters over the intervention period.

The observation of higher TEWL values in adults at the study end compared to baseline should be interpreted with a focus on between-group comparisons rather than baseline comparisons. The important comparison is between the treatment groups (DAC-BDS versus BDS-DAC) as these provide insights into the relative efficacy of the emollients regardless of baseline values. Different baseline TEWL measurements were observed depending on treatment sequence (DAC-BDS or BDS-DAC groups), making direct before-after comparisons less informative than between-group analyses. Seasonal effects may also play a role in the observed TEWL changes as the study period spanned different seasons, which can influence skin barrier function. Additionally, as we acknowledged in the limitations, detecting significant differences in mild AD cases (EASI of 7) is inherently challenging compared to more severe forms of the condition. A notable limitation of this study is the low severity of AD, particularly in adults at baseline for group B where the lower quartile range was below 1. This very mild disease severity might explain why no treatment effects on the microbiome were found as changes may be more readily detectable in moderate to severe AD cases.

Results of this exploratory clinical trial indicate that both leave-on products improved SCH and AD severity after 3 months of treatment. Differences between products were minor. The application of two different products did not lead to significant alterations in the overall composition of the skin microbiome. Microbial diversity remained relatively stable, emphasizing the resilience of the skin microbiome to interventions. Although there were shifts in the frequency of certain microbial genera, these changes were not consistent between age groups, swab localization, and treatments.

We would like to express our gratitude to Bayer Consumer Care AG in Basel, Switzerland, for their financial support, which made this research project possible. We are also very grateful for the support of Dimitra Aikaterini Lintzeri and Andria Constantinou in their work as study physicians and of Michelle Lisy and Marion Feuereisen in conducting visits and supporting participants.

The study protocol was reviewed and approved by the Ethics Committee of the Charité Universitätsmedizin Berlin on 3 September 2020 (EA2/186/20) and registered at the German Register of clinical trials (DRKS00028746) and conducted at the Clinical Research Center for Hair and Skin Science, Charité – Universitätsmedizin Berlin. Written informed consent was obtained from adult participants and, in the case of children, from their parent/legal guardian to participate in the study.

Kathrin Hillmann is an employee of Charité Universitätsmedizin Berlin. Tsenka Tomova-Simitchieva is an employee of Charité Universitätsmedizin Berlin. Pauline Sophia Pinta is a PhD student at Charité Universitätsmedizin Berlin. Zhile Xiong is a PhD student at Charité Universitätsmedizin Berlin. Annette Moter co-founded and holds shares of MoKi Analytics GmbH, a start-up from Charité Universitätsmedizin Berlin. Varvara Kanti-Schmidt is an employee of Johannes Wesling Klinikum Minden. Jan Kottner is an employee of Charité Universitätsmedizin Berlin. Ulrike Blume-Peytavi is an employee of Charité Universitätsmedizin Berlin.

This study was an investigator-initiated trial conducted at the Charité – Universitaetsmedizin Berlin. Germany. The project was financially supported by Bayer Consumer Care AG in Basel, Switzerland, by an unrestricted grant. The funder had no role in the design, data collection, data analysis, and reporting of this study.

Kathrin Hillmann: project administration, investigation, and writing – original draft. Tsenka Tomova-Simitchieva, Pauline Sophia Pinta, Zhile Xiong, Annette Moter, and Varvara Kanti-Schmidt: investigation and writing – review and editing. Jan Kottner: formal analysis, conceptualization, methodology, and writing – review and editing. Ulrike Blume-Peytavi: conceptualization, funding acquisition, methodology, resources, supervision, and writing – review and editing.

All data generated or analyzed during this study are included in this article and its Supplementary material files. Further inquiries can be directed to the corresponding author.