Value of reflectance confocal microscopy for the monitoring of rosacea during treatment with topical ivermectin

Abstract

Background

Reflectance confocal microscopy (RCM) enables noninvasive Demodex mite detection in rosacea. Objective scoring of rosacea severity is currently lacking.

Objectives

To determine the value of RCM for monitoring Demodex, inflammation and vascular parameters in rosacea during treatment.

Methods

In 20 rosacea patients, clinical and RCM examination were performed before, during, and 12 weeks after a 16-week treatment course with topical ivermectin. Using RCM, number of mites and inflammatory cells, epidermal thickness, and vascular density and diameter were measured. RCM features were correlated with clinical assessment.

Results

Treatment resulted in clinical reduction of inflammatory lesions. Mites were detected in 80% of patients at baseline, 30% at week 16, and 63% at week 28. The number of mites reduced significantly during treatment, but no changes in inflammatory cells, epidermal thickness or vascular parameters were observed. Correlation between number of inflammatory lesions and mites was low. None of the RCM variables were significant predictors for clinical success.

Conclusions

RCM enables anti-inflammatory effect monitoring of topical ivermectin by determining mite presence. Quantifying exact mite number, and inflammatory and vascular characteristics is challenging due to device limitations. In its current form, RCM seems of limited value for noninvasive follow-up of rosacea in clinical practice.

Keywords:

• Reflectance confocal microscopy
• rosacea
• Demodex
• inflammation
• ivermectin

Introduction

Reflectance confocal microscopy (RCM) is a noninvasive imaging technique for in vivo visualization of the superficial skin layers with a resolution comparable to conventional histology (1,2). Unlike invasive tools such as skin biopsies, RCM enables temporal monitoring of the skin without causing irritation, damage or alteration. Historically, RCM has been mainly used for the diagnosis of (non-)melanoma skin cancer (3–6), but it also has promising potential in the evaluation of inflammatory skin diseases (7). Some knowledge has already been gained with the detection and quantification of Demodex mites in rosacea (8–13).

Rosacea is a chronic inflammatory skin disorder of uncertain etiology, characterized by among others papules, pustules, erythema, telangiectasia, and flushing (14–16). Demodex mites have been implicated in rosacea etiology (14,17–19), and studies show that the Demodex population is markedly increased in patients with rosacea compared to healthy controls (20–25). With RCM, mites are visible as roundish or cone-shaped, bright structures inside hair follicles (8,9,26,27). Moreover, structural changes in Demodex mites have been observed after ivermectin treatment using RCM (28). However, no RCM studies have yet assessed inflammatory and vascular parameters in rosacea (29,30).

Due to the variety in clinical features among individual rosacea patients and the presence of multiple subjective, often not validated, classification systems, there is a need for methods quantifying rosacea severity in an objective manner (31–35). It is hypothesized that clinical effects of rosacea treatment can be quantified and ideally also be predicted using RCM monitoring. Topical ivermectin, an antiparasitic preparation that has dual anti-inflammatory and acaricidal effects, has been widely studied as a safe, easy to use, and effective treatment to reduce papules and pustules in rosacea (36). This makes ivermectin a suitable intervention to monitor rosacea treatment effects.

The aim of this study was to determine the value of RCM to monitor Demodex mites, inflammation and vascular parameters in rosacea patients during treatment with topical ivermectin. To do so, RCM features were correlated to clinical scores, and possible prognostic features correlating to successful clinical response were determined.

Materials and methods

Patients

This study was approved by the ethics committee. Patients were recruited between January 2018 and April 2019 at the Department of Dermatology, Radboud University Medical Center, Nijmegen, The Netherlands. Eligible participants were ≥18 years old, and had a clinical diagnosis of moderate to severe facial rosacea (investigator’s global assessment [IGA] ≥ 3, >15 papules or pustules, and/or erythema scale ≥3 plus telangiectasia scale ≥2). Patients currently using ivermectin cream, having other facial dermatological conditions or underlying diseases requiring use of topical or systemic therapies interfering with rosacea diagnosis or assessment, or the inability to adhere to the washout period for these interfering therapies (Supplementary Table 1) were excluded. Participants were instructed to avoid known rosacea aggravating environmental factors and foods, and to abstain from sunbathing or using tanning beds throughout the study.

Clinical and RCM procedures

Written consent was obtained from all participants. After an acclimatization period of at least 15 min, facial clinical scores were assessed including IGA scale, IGA success scale (0 or 1), papules and pustules scale, erythema scale, telangiectasia scale, and inflammatory lesion count (Supplementary Table 2). Next, participants’ cheek with the highest clinical scores (left cheek: n = 13; right cheek: n = 7) was examined using RCM. The cheek was chosen because highest Demodex densities have been found here, and the RCM ring can be attached firmly on the relatively flat surface (37,38). After examination, participants were instructed to apply topical ivermectin 1% once daily in the evening for 16 consecutive weeks. Clinical and RCM examination were performed at five time points: week 0 (baseline), 6, 12, 16 (during treatment), and week 28 (follow-up). The same cheek was examined with RCM during consecutive visits. Therapy compliance was monitored using patient diaries and by weighing the used ivermectin tubes at week 6, 12, and 16. All examinations were performed in the same room by the same clinician (J. G. M. L.), and room temperature was kept constant between 19 and 24 °C.

RCM device and image analysis

RCM analysis was performed using a Vivascope 1500® confocal microscope (Mavig, Munich, Germany) (39). Four surface-parallel horizontal areas of 8 × 8 mm² were scanned with the Vivablock function at the level of the stratum corneum/granulosum, stratum spinosum, dermo-epidermal junction (DEJ), and the superficial dermis. Next, four vertical mappings (Vivastacks) of 0.5 × 0.5 mm² were obtained, starting from the skin surface up to a depth of 150 µm with interval steps of 3 µm. Then, three movies of 0.5 × 0.5 mm² of approximately 5 s each were made in the superficial dermis to visualize dermal blood flow. Analysis of RCM images was performed by one investigator (J. G. M. L.). Using all Vivablocks, Vivastacks and movies, the number of Demodex mites, inflammatory cells, degree of spongiosis, epidermal thickness, and vascular diameter and density were calculated.

Calculating Demodex mites

After selecting one 8 × 8 mm² Vivablock at z = 20–80 µm (i.e. stratum granulosum/spinosum; mite quantification is best performed at this level (8,11)), mites were noted as present or absent. Then, the number of mites, hair follicles, infested follicles, and number of mites per individual follicle were counted; a separate category doubt was added if the follicle content was not well defined. Mean number of mites per follicle and mean number of mites per infested follicle were calculated.

Calculating inflammatory cells and spongiosis

The number of inflammatory cells, visible as round-to-polygonal bright cells (40), was counted in one 8 × 8 mm² Vivablock at the level of the DEJ. A group of >40 inflammatory cells was defined as one inflammatory nest. Dendritic cells were noted as present or absent.

To determine degree of spongiosis, three single images in a Vivablock with a clear honeycombed pattern from the stratum granulosum were selected. In these images, a number of keratinocytes in a square with known surface area without hair follicles were counted manually using ImageJ^® freeware (41), and extrapolated into number of keratinocytes/mm². Lower values mean more intercellular edema, indicating increased spongiosis. The mean score of the three images was calculated.

Calculating epidermal thickness

Thickness of the stratum corneum (SC) and the viable epidermis, without SC, were determined (in µm) from the four Vivastacks with a standardized protocol as described in our earlier work (39). The mean thickness of the SC and the viable epidermis of the four Vivastacks were calculated.

Scoring vascularity

Vascular diameter and density were determined separately from the three movies of the superficial dermis, and classified using a five-point scale: 1 (normal, i.e. similar to healthy facial skin), 2 (mildly increased), 3 (moderately increased), 4 (severely increased), and 5 (very severely increased). Example movies of each score can be found in Supplementary Table 3. The mean score of three movies for both parameters was calculated.

Statistical analysis

Descriptive statistics were used to explore baseline characteristics and categorical data. Differences in the same continuous variable between various time points were evaluated with Wilcoxon signed-rank tests for matched pairs. No correction for multiple comparisons was applied, because of the exploratory character of this study. Correlating clinical scores and RCM variables using measurements in the same patients at various time points can lead to biased estimates if the clustering within patients is ignored. Therefore, to get an overall estimate of correlation between number of inflammatory lesions and number of mites, a meta-analysis was constructed to aggregate the Spearman correlations (R) at the five consecutive time points. Additionally, possible interactions between RCM features and achieving IGA 0 or 1 (=clinical success) were measured in term of an odds ratio in the 95% confidence interval by using mixed model logistic regression analysis. Statistical analysis was performed using GraphPad Prism 5.03 (GraphPad Software, Inc., San Diego, CA) and R 3.6.0 (Foundation for Statistical Computing, Vienna, Austria). For all tests, p < .05 was considered statistically significant. Missing values were excluded from the analyses.

Results

Patient characteristics

Twenty-one patients were included in this study. One patient was excluded because of starting tetracycline during the study. Therefore, data from 20 participants were used for statistical analysis. In one patient, the week 28 visit was not performed due to the patients’ inability to travel to the hospital within the study visit window (2 weeks). The RCM data of the week 12 visit of one patient were unexpectedly not saved at the computer, and were therefore missing.

Demographic and baseline characteristics are displayed in Table 1. Most patients had IGA 3 (80%), erythema 3 (65%), and telangiectasia 2 or 3 (80%) scores. Median facial inflammatory lesion count at baseline was 36.5 (range 0–144). All patients applied ivermectin daily and according to prescription; mean daily quantity applied was 0.36 g.

Table 1. Demographic and baseline clinical characteristics.

	Rosacea patients (n = 20)
Age, years
Mean ± SD (min–max)	52.8 ± 14.5 (24–81)
Gender, n (%)
Female	11 (55)
Male	9 (45)
Skin photo type
I	2 (10%)
II	15 (75%)
III	3 (15%)
Investigator’s global assessment
3 = Moderate	16 (80%)
4 = Severe	4 (20%)
Inflammatory lesion count
Median (range)	36.5 (0–144)
Erythema scale
1 = Almost clear	2 (10%)
2 = Mild	3 (15%)
3 = Moderate	13 (65%)
4 = Severe	2 (10%)
Telangiectasia scale
1 = None	4 (20%)
2 = Mild	8 (40%)
3 = Moderate	8 (40%)

Eleven patients reported one or more dermatologically related adverse event (AE); stinging/burning (n = 7), increased facial erythema (n = 4), itching (n = 3), and facial skin dryness (n = 1). All AEs occurred during the 16-week treatment period, and were mild and self-limiting.

Clinical assessment

Figure 1 presents the clinical scores at the five time points, and representative clinical pictures are shown in Figure 2. Treatment resulted in clinical success (IGA 0 or 1) in 45% and 47% of patients at week 16 and at week 28, respectively. Median number of facial inflammatory lesions significantly decreased from baseline to week 12 (8.5, range 0–39; p < .001) and week 16 (2.0, range 0–44; p < .001), with no further changes at week 28 compared to week 16 (3.0, range 0.0–30; p = .19). Some improvement in erythema scores was noticed, while telangiectasias remained mainly unaffected.

Figure 1. Clinical scores of all patients during 16 weeks of treatment with topical ivermectin and at week 28 (follow-up). (a–e) IGA scale, IGA success scale (IGA 0 or 1), papules and pustules scale, erythema scale, and telangiectasia scale. (f) Number of facial inflammatory lesions, expressed as median with range. IGA: investigator’s global assessment.

Figure 2. Rosacea in a 56-year old man, (a) at baseline, (b) at week 16 of treatment with topical ivermectin 1% once daily, and (c) 12 weeks post-treatment. Notice the disappearance of papules and pustules at the forehead and prominent erythema at the cheeks, while telangiectasias and residual erythema at both cheeks are persisting. The green dots serve as a tool to assess clinical erythema.

RCM assessment

Demodex mites

Mites were detected in 80% of participants at baseline, 30% at week 12 and 16, and 63% at week 28 (Table 2 and Figure 3). Median number of mites significantly decreased from baseline (13, range 0.0–203) to week 12 (0.0, range 0.0–10; p < .001) and week 16 (0.0, range 0.0–14; p < .001), and remained low at week 28 (1.0, range 0.0–21; p = .51 compared to week 16). Median number of follicles was comparable before, during, and after treatment. The median number of infested follicles and mites per follicle and mites per infested follicle were all significantly decreased after 16 weeks of treatment (p < .001, Table 2), while only number of mites per infested follicle increased slightly at follow-up compared to week 16. Most follicles contained zero mites. The percentage of follicles with unclear content (doubt) was higher than follicles with clearly visible mites, which is visualized in Figure 4(a,b).

Figure 3. Demodex mites during 16 weeks of treatment with topical ivermectin and at week 28 (follow-up), measured with RCM. (a) Mite presence per 8x8 mm², expressed as present or absent. (b) Number of mites per 8 × 8 mm^2, expressed as median with range.

Figure 4. Confocal images of the cheek (0.5 × 0.5 mm²). (a) Follicle at the level of the stratum granulosum containing seven Demodex mites, clearly visible as bright, round contours. (b) Follicle at the level of the stratum granulosum with unclear content due to vague, white-roundish structures, included in mite category doubt. (c) Multiple bright, round to oval or fusiform cells with dendritic processes at the level of the dermo-epidermal junction, possibly melanocytes or Langerhans cells. (d) Multiple bright, round cells at the level of the dermo-epidermal junction, possibly inflammatory cells.

Table 2. Mite and follicle numbers per 8 × 8 mm².

	Baseline	Week 6	Week 12	Week 16	Week 28	Baseline versus week 16 p Value	Baseline versus week 28 p Value	Week 16 versus week 28 p Value
Mite presence
n patients present (%)	16 (80%)	10 (50%)	6 (30%)	6 (30%)	12 (63%)	–	–	–
Mites, n	13	0.5	0.0	0.0	1.0	<.001	.001	.13
Median (range)	(0.0–203)	(0.0–29)	(0.0–10)	(0.0–14)	(0.0–21)
Follicles, n	262	266	263	260	262	.21	>.99	.14
Median (range)	(162–331)	(213–355)	(129–402)	(183–386)	(162–384)
Infested follicles, n	6.0	0.5	0.0	0.0	1.0	<.001	<.001	.20
Median (range)	(0.0–73)	(0.0–9.0)	(0.0–6.0)	(0.0–6.0)	(0.0–7.0)
Mites/follicle, n	0.052	0.002	0.000	0.000	0.004	<.001	<.001	.15
Median (range)	(0.00–0.712)	(0.00–0.109)	(0.00–0.039)	(0.00–0.067)	(0.00–0.072)
Mites/infested follicle, n	2.0	0.5	0.0	0.0	1.0	<.001	.33	.036
Median (range)	(0.0–4.5)	(0.0–4.1)	(0.0–3.0)	(0.0–2.3)	(0.0–4.0)
Mites/follicle
Median % of follicles
0	91	96	97	97	96	–	–	–
1	1.2	0.0	0.0	0.0	0.0	–	–	–
≥2	1.6	0.0	0.0	0.0	0.0	–	–	–
Doubt^a	5.8	3.1	2.8	2.6	3.2	–	–	–

^a Unclear follicle content, not clearly distinguishable as mites.

Inflammation, epidermal thickness and vascularity

Median number of inflammatory cells, inflammatory nests, and thickness of the SC and viable epidermis did not change significantly at week 16 or at follow-up compared to baseline (Table 3). Dendritic-like cells were seen before, during and after treatment in all patients (Figure 4(c)). Interestingly, median number of keratinocytes/mm² slightly decreased from baseline to week 16 (p = .004) and week 28 (p = .031), implying an increase in spongiosis. As shown in Figure 5, vascular diameter decreased only slightly after topical ivermectin application, while vascular density barely changed.

Figure 5. (a) Vascular diameter and (b) vascular density scores during 16 weeks of treatment with topical ivermectin and at week 28 (follow-up), determined with RCM.

Table 3. Inflammatory parameters and epidermal thickness.

	Baseline	Week 6	Week 12	Week 16	Week 28	Baseline versus week 16 p Value	Baseline versus week 28 p Value	Week 16 versus week 28 p Value
Inflammatory cells, n	41	31	39	27	51	.089	.73	.21
Median (range)	(2.0–494)	(0.0–425)	(0.0–426)	(2.0–370)	(1.0–563)
Inflammatory nests, n	0.0 (0.0–2.0)	0.0 (0.0–4.0)	0.0 (0.0–2.0)	0.0 (0.0–1.0)	0.0 (0.0–1.0)	.17	.089	>.99
Median (range)
Spongiosis, n keratinocytes/mm^2	2519	2517	2378	2268	2377	.004	.031	.48
Median (range)	(1933–3435)	(1668–3186)	(1601–3131)	(1513–2602)	(1723–3108)
Thickness SC, µm	10.2	9.28	11.6	10.8	8.50	.58	.21	.48
Median (range)	(6.18–18.6)	(6.18–13.9)	(6.96–20.6)	(5.41–27.8)	(6.18–17.0)
Thickness viable epidermis, µm	55.3	54.9	54.9	55.7	54.9	.32	.30	.51
Median (range)	(44.8–70.9)	(47.9–70.4)	(43.3–68.8)	(44.1–66.5)	(42.5–67.3)

SC: stratum corneum.

Correlating clinical scores with RCM parameters

As the correlation estimates in the meta-analysis were homogeneous over time, standard averaging with inversed variance weighting was used in this method. A low correlation existed between number of inflammatory lesions and number of mites (R = 0.28, 95% CI 0.08–0.48). Mixed model logistic regression revealed that none of the RCM variables were significant predictors for clinical success. Despite this, mite presence was the most promising parameter (Table 4).

Table 4. Mixed model logistic regression models determining RCM features associated with clinical success (IGA 0/1).

	IGA score 0 or 1	p Value
Mite presence, present/absent	2.10 (0.82–6.07)	.14
Demodex mites, n	1.11 (1.01–1.29)	.10
Infested follicles, n	1.28 (1.03–1.82)	.088
Mites/follicle, n	3.61^12 (55.5–2.49^31)^a	.095
Mites/infested follicle, n	1.49 (0.97–2.49)	.090
Inflammatory cells, n	0.9996 (0.9959–1.0039)	.85
Inflammatory nests, n	1.70 (0.65–9.76)	.41
Spongiosis, n keratinocytes/mm^2	1.0002 (0.9990–1.0017)	.64
SC thickness, µm	1.13 (0.94–1.44)	.22
Thickness viable epidermis, µm	1.03 (0.95–1.12)	.50
Vascular diameter, scale 1–5	0.84 (0.43–1.74)	.61
Vascular density, scale 1–5	0.77 (0.33–1.71)	.52

Values are given as odds ratios (95% confidence intervals). IGA: investigator’s global assessment.

^a Large OR + 95% CI due to low median value and highly skewed distribution of this variable (also see Table 2).

Discussion

This study confirmed the well-known efficacy of topical ivermectin to reduce inflammatory lesions in rosacea (36). To the best of our knowledge, this is the first study combining a broad spectrum of RCM and clinical variables to monitor rosacea during treatment. Using RCM, we were able to monitor the anti-inflammatory effect of topical ivermectin by determination of Demodex mite presence, number of infested follicles and number of mites per follicle. However, quantifying exact mite number, inflammatory parameters and vascular characteristics was challenging due to various device limitations.

To guarantee adequate evaluation of rosacea severity, clinical assessment was performed using multiple scales, based on the newly developed phenotype-based classification system for rosacea (31,36,42). Topical ivermectin treatment reduced the number of facial inflammatory lesions and erythema, with the greatest reduction at week 16. We found no changes in telangiectasias, corresponding to previous results (36), supporting the hypothesis that telangiectasias are non-inflammatory features. A significant decrease in mite number was measured with RCM, with lowest counts after 12 weeks of treatment, preceding clinical outcomes; these findings suggest that topical ivermectin treatment should be continued for at least 12 weeks. Ivermectin blocks signal transduction in the neuromuscular system of Demodex mites, causing paralysis and death (28,43). Our results support the notion that a reduction of Demodex may be causative for the observed decrease in inflammation. Various mechanisms for immune system triggering by Demodex have been proposed, such as hair follicle blockage, foreign-body reactions, and secretion of bacteria and waste (11,44–46). On the other hand, Demodex was absent in 20% of rosacea patients at baseline while they responded well to ivermectin treatment, and correlation between number of inflammatory lesions and mites was low. This suggests that the pathogenesis of rosacea cannot be entirely explained by Demodex, and mites could be one of multiple factors inducing follicular inflammation.

Compared to earlier RCM studies calculating Demodex density in rosacea (8,9,11–13,47), we measured low mite numbers, while follicle amounts were very similar (8,11). These differences could be explained by the fact that we excluded follicles with unclear content from the mite count (Figure 4(b)). These follicles could contain sebum, sludge, hair shafts, or (disintegrating) mites (28,48). Additionally, counting exact mite number was challenging; mites’ contours were often vaguely debordered (49,50). Thirdly, despite the used washout periods (Supplementary Table 1), we cannot exclude an influence of previous rosacea treatments applied such as topical metronidazole; however, metronidazole does not directly kill Demodex mites (11,51). RCM has some other limitations for mite analysis. First, Demodex folliculorum is detectable, but not Demodex brevis which lives beneath the RCM maximum visualization depth, namely near the sebaceous glands (45,50). Second, mite distribution within the follicular canal varies, giving false-negative results when counting mites at only one level (52). Lastly, RCM does not enable evaluation of mite viability or life stage (23,49,53,54).

In histological sections, the inflammatory infiltrate in rosacea can be observed in both superficial and deeper skin layers (29,30). In the current study, the DEJ was used as a representative level to score inflammation, because deeper layers impair detailed vision due to reduced image resolution and co-presence of bright collagen bundles (2). No changes in inflammatory cell number at the DEJ were measured, which may be due to the difficulty of differentiating inflammatory cells from pigmented keratinocytes and melanocytes, which all are visible as polygonal, bright structures (Figure 4(c,d)) (2,55). Furthermore, RCM cannot differentiate within leukocyte type (29,30).

Both vasodilation and increased vascular density are thought to be involved in the pathophysiology of facial erythema in rosacea (56). We expected to measure a decrease in vasodilation in multiple participants, as this feature is most linked to inflammation, but we did not. Capillary diameter can be easily measured with RCM in vessels inside dermal papillae (57–60), but as papillae are hardly visible in facial skin (55), this method is not suitable in rosacea. Therefore, we tried quantifying vascular density and diameter using MATLAB by calculating differences in blood vessel movement between two consecutive movie frames. Unfortunately, these differences were not only caused by blood flux, but also by movement of the vessel surroundings, namely the patient and the RCM device. Consequently, we used Likert scales to grade vascular density and diameter, which are unfortunately still subjective.

Our study has several strengths. All examinations were performed in the same room by the same researcher, reducing light- and interobserver variability. Missing values were minimal due to high visit and therapy compliance. Both genders were equally represented, and age range was large, providing decent generalizability of our findings.

The advantage of RCM is its noninvasiveness without causing discomfort or damage, providing the ability to monitor the same facial location over time (27). Limitations of our study are due to RCM limitations. Imaging of convex areas is challenging, impeding the generation of a perfectly horizontal Vivablock necessary for analysis in one skin level; image analysis takes time and requires an experienced investigator; RCM images are two-dimensional and in black-and-white only; and the apparatus is bulky, not wireless, and requires startup of approximately 10 min. Taking these factors into account, the RCM in its current form seems unsuitable for quick monitoring of rosacea patients in daily clinical practice. On the other hand, in research setting, RCM provides valuable additional information about mite presence that could otherwise not have been obtained in a noninvasive way.

In summary, we showed that RCM can be used for monitoring of Demodex presence in rosacea patients during successful anti-inflammatory treatment with topical ivermectin. Quantifying exact mite number and inflammatory and vascular parameters is challenging due to various device limitations. In its current form, RCM seems of limited value for noninvasive follow-up of rosacea treatment in daily clinical practice. However, as imaging techniques become more compact, intelligent (i.e. enabling automatic counting), and affordable, RCM’s future widespread application in research and possibly also clinical practice for rosacea monitoring can be expected.

Abbreviations
AE	=	adverse event;
CI	=	confidence interval;
DEJ	=	dermo-epidermal junction;
IGA	=	investigator’s global assessment;
OR	=	odds ratio;
RCM	=	reflectance confocal microscopy;
SC	=	stratum corneum

Disclosure statement

J. G. M. L. has received a research grant from Galderma. She carried out clinical trials for Abbvie, Novartis, Janssen and LEO Pharma. She has received reimbursement for attending meetings from Abbvie. R. J. B. D. has received a research grant from Galderma. She carried out clinical trials for Cutanea Life Sciences, Galderma, Abbvie, Novartis and Janssen. She has received reimbursement for attending meetings from Abbvie and Galderma. She has served as a consultant for Abbvie, Galderma and Novartis. Fees were paid directly to the institution. M. P., P. E. J. E., E. M. G. J. J. and K. P. N. have no conflicts of interest to declare.

Additional information

Funding

This investigator-initiated study was conducted with financial support from Galderma. The funding source had no influence on study design, data collection and analysis, or the content of the manuscript.