1 Introduction

Atopic dermatitis (AD) is a common chronic inflammatory skin disease affecting up to 20% of children worldwide [1]. Its pathogenesis is complex and highly heterogeneous, characterised by a predominant T helper (Th)2-driven inflammation, variable activation of Th22, Th17, and Th1 pathways, and a dysfunctional epidermal barrier [1-3]. The understanding of the underlying local and systemic mechanisms of AD, derived from molecular studies in skin biopsies and serum, has advanced in recent years. This has led to the development of targeted therapies [4].

Dupilumab, a monoclonal antibody blocking interleukin (IL)-4 and IL-13 receptor binding, was the first targeted treatment approved for (moderate to) severe AD, and is currently available for patients aged ≥ 6 months [5-7]. Molecular studies have been valuable in elucidating the modulatory effects of dupilumab on systemic AD-related proteins in paediatric AD patients [5-11]. However, its modulatory effects on local skin proteins in this paediatric population remain poorly characterised. This is primarily due to the limited data available and the challenges associated with skin biopsies, which are invasive, painful, and potentially scarring [12-14].

Tape stripping, a minimally invasive technique that samples the stratum corneum and upper stratum granulosum, is emerging as an alternative to skin biopsies for determining local proteomic signatures [15-17]. Recent studies have shown that tape strips effectively profile key immune and epidermal barrier dysregulations in skin inflammation, including AD and psoriasis, in both lesional and non-lesional skin [13, 17-26]. Moreover, tape strips have proven useful for tracking molecular changes in adult AD patients and a single paediatric cohort (n = 14) treated with dupilumab [14, 18, 20]. While previous studies have primarily focused on direct comparisons between local tape strips and skin biopsies, studies comparing tape strips with serum are limited, particularly in evaluating treatment response [16, 17, 24, 25, 27, 28]. However, such comparisons may provide crucial insights into distinct local and systemic proteomic signatures and changes, and may enhance our understanding of the proteomic changes in response to targeted treatment.

Therefore, this study investigated local and systemic immune-related proteins in tape strips and serum of paediatric atopic dermatitis (AD) patients treated with dupilumab, and explored their correlation with clinical severity.

2 Methods

2.1 Study Design

Twenty paediatric patients (aged < 18 years) with moderate-to-severe AD, eligible for systemic treatment and starting dupilumab treatment, were recruited from the Department of Dermatology at the Wilhelmina Children's Hospital (University Medical Center Utrecht), The Netherlands, between October 2022 and November 2023 (Figure 1). Exclusion criteria were the use of oral corticosteroids, small molecule inhibitors, or cyclosporine A within 1 week, methotrexate within 4 weeks, and biologics within 10 weeks prior to the start of treatment. All patients participated in the BioDay registry, a prospective, multicenter registry of patients treated with new systemic treatments for AD in daily practice. Written informed consent was obtained from all patients (and/or their parents). This study did not fall under the scope of the Medical Research Involving Human Subjects Act (METC 18/249 and 23/014) and has been performed according to the declaration of Helsinki.

3 Results

3.1 Study Population and Clinical Improvement During Dupilumab Treatment

Twenty paediatric patients were included in the study (median age 10.0 years, interquartile range (IQR) 4.3–13.5 years; 70.0% female). Patient characteristics are shown in Table 1. Patients were evenly distributed across the age groups: 2–5 years (n = 6), 6–11 years (n = 7), and 12–17 years (n = 7). None of the patients reported a history of other inflammatory or autoimmune diseases (e.g., psoriasis, rheumatoid arthritis). The majority of tape strips were collected from the upper extremities (Table S2). Non-lesional tape strips were collected at least 4 cm from the lesional site, except in one patient where the distance was 2 cm. Most patients applied concomitant topical corticosteroids prior to sampling: 95.0% at baseline, 85.0% at week 4, and 75.0% at week 16 (median: 1.0 day prior; range 0–7 days) (Table S3). None of the patients applied concomitant topical calcineurin inhibitors to the sampling site.

TABLE 1. Baseline patient characteristics.

Characteristic	N = 20
Sex (female), n (%)	14 (70.0)
Age (years), median (IQR)	10.0 (4.3–13.5)
2–5 years, n (%)	6 (30.0)
6–11 years, n (%)	7 (35.0)
12–17 years, n (%)	7 (35.0)
Age of AD onset (years), median (IQR)	0.0 (0.0–0.0)
Atopic comorbidity, ≥ 1, n (%)	17 (85.0)
Asthmaa	13 (65.0)
Allergic rhinitisa	16 (80.0)
Allergic conjunctivitisb	11 (55.0)
Food allergyc	6 (30.0)
First-degree relatives with atopic diseases,d ≥ 1, n (%)	19 (95.0)
Atopic dermatitis, n (%)	15 (75.0)
Allergic rhinitis, n (%)	12 (60.0)
Asthma, n (%)	10 (50.0)
Food allergy, n (%)	2 (10.0)
Previous use of immunosuppressants, n (%)	15 (75.0)
Oral corticosteroids	4 (20.0)
Cyclosporine A	10 (50.0)
Methotrexate	1 (5.0)
EASI score, median (IQR)	16.0 (9.0–23.7)
IGA score, n (%)
Mild	3 (15.0)
Moderate	12 (60.0)
Severe	5 (25.0)
Very severe	0 (0.0)
NRS itch, median (IQR)	7.0 (5.0–9.0)
Missing, n (%)	1 (5.0)
NRS pain, median (IQR)	6.0 (1.0–8.0)
Total IgE levels (kU/L), median (IQR)	2335.0 (223.5–4777.5)
Missing, n (%)	4 (20.0)
Eosinophil levels (×109/L), median (IQR)	0.6 (0.4–1.0)
Eosinophilia (≥ 0.5 × 109/L), n (%)	11 (55.0)

• Abbreviations: AD, atopic dermatitis; EASI, Eczema Area and Severity Index; IGA, Investigator Global Assessment Scale; IgE, immunoglobulin E; IQR, Interquartile range; NRS, Numeric Rating Scale.
• ^a Diagnosed by pulmonologist.
• ^b Patient reported.
• ^c Confirmed by elevated (≥ 0.35 kU/L) food specific IgE levels and corresponding symptoms.
• ^d Self-reported.

Patients demonstrated significant clinical improvement during dupilumab treatment, with a median percentage improvement in EASI of 64.3% and NRS itch score of 37.5% at week 4, increasing to 67.0% and 46.4%, respectively, at week 16 (all p < 0.001) (Figure 2). At week 16, 80.0% of patients achieved EASI ≤ 7, 68.4% achieved NRS itch ≤ 4, and 57.9% met both criteria.

3.2 Tape Strips and Serum Capture Distinct Protein Expressions at the Start of Treatment

Differentially expressed proteins in tape-stripped lesional skin and serum at baseline are depicted in an unsupervised hierarchical clustering heatmap (Figure 3). Relatively high expression of innate pro-inflammatory cytokines (IL-1α, IL-18) was observed in tape strips of lesional skin, while T cell-derived cytokines were either undetectable (IL-13, IFN-γ) or poorly expressed (IL-4, IL-15, IL-22). In contrast, relatively high levels of T cell-derived cytokines (IL-13, IFN-γ) and Type 2 immune mediators TARC and PARC were observed in serum (Figure 3). Protein expression in non-lesional skin was generally lower than in lesional skin, except for IL-1α (Figure S1). Despite small sample sizes, consistent protein expression was observed across all age groups at baseline (Figure S2).

3.3 Proteomic Changes in Tape Strips and Serum During Dupilumab Treatment

In both lesional skin and serum, the disease-associated Type 2 immune mediators TARC and PARC significantly decreased after 4 and 16 weeks of dupilumab treatment, and significantly decreased in non-lesional skin after 16 weeks of treatment. However, distinct proteomic changes were observed in the skin and serum during treatment. In lesional skin, T cell-recruiting chemokine CTACK, innate pro-inflammatory cytokines (IL-8, IL-18), tissue repair and remodelling proteins (periostin, MMP-1), and Type 1 immune mediator CXCL10 significantly decreased after 16 weeks of treatment, while innate pro-inflammatory cytokine IL-1α significantly increased (Figure 4). Significant reductions in IL-8, periostin, and MMP-1 levels were observed after 4 weeks of treatment. In serum, IL-4 and IL-13 levels significantly increased after 4 and 16 weeks of treatment.

3.4 Tape Strip and Serum Proteins Correlate With Disease Severity

Several proteins were significantly correlated with AD severity as measured by EASI and NRS itch (Tables S4 and S5). In lesional skin, the strongest correlations with EASI and NRS itch were observed for Type 2 immune mediators PARC (r = 0.62 and r = 0.54, respectively), CTACK (r = 0.57 and r = 0.46, respectively), and TARC (r = 0.56 and r = 0.52, respectively). Tissue repair and remodelling proteins MMP-1 (r = 0.56 and r = 0.36, respectively) and periostin (r = 0.55 and r = 0.43, respectively), and innate pro-inflammatory cytokine IL-8 (r = 0.55 and r = 0.41, respectively) and IL-18 (r = 0.45 and r = 0.43, respectively) also showed significant correlations (Figure 5A, Figure S3A). Proteins in non-lesional skin demonstrated correlations with disease severity that were generally comparable to those in lesional skin, though less robust (Figure 5B, Figure S3B).

In serum, EASI and NRS itch were significantly correlated with Th22-related cytokine IL-22 (r = 0.38 and r = 0.51, respectively) and PARC (r = 0.32 and r = 0.60, respectively) (Figure 5C, Figure S3C). In addition, NRS itch was correlated with TARC (r = 0.63), whereas TARC was not significantly correlated with EASI. EASI was inversely correlated with Th2-related cytokine IL-4 (r = −0.53), which could be interpreted as IL-4 inhibition by dupilumab rather than being associated with EASI.

4 Discussion

This study explores the local and systemic immune signatures of paediatric AD patients treated with dupilumab by analysing tape strips and serum samples. By analysing both local and systemic proteins, we captured a partially overlapping yet distinct immune response characterised by robust Type 2-mediated inflammation. Tape strips capture reductions in innate pro-inflammatory cytokines (IL-8, IL-18) and tissue remodelling proteins (periostin, MMP-1) in response to treatment, reflecting improvements in local inflammation and barrier repair. In contrast, serum is more sensitive in detecting T cell-derived cytokines (IL-4, IL-13), reflecting systemic immune modulation. Additionally, the Th2-related chemokine PARC emerged as the strongest protein associated with clinical severity in both skin and serum.

Consistent with our findings at the start of dupilumab treatment, Andersson et al. and McAleer et al. observed elevated levels of innate mediators (IL-18, IL-8, IL-1β) and T cell-recruiting chemokines (TARC, CTACK) in tape-stripped lesional skin of paediatric AD patients, compared to non-lesional skin and healthy controls [23, 25]. In line, Del Duca et al. observed upregulation of innate pro-inflammatory cytokines (IL-8, IL-1β) in adult AD patients. In contrast to our findings, upregulation of IL-13 was observed in lesional tape-stripped skin compared to skin biopsies, possibly due to the higher sensitivity of transcriptomic analysis used [24]. Similar to our findings, Andersson et al. and McAleer et al. demonstrated that levels of T cell-derived cytokines (e.g., IL-4, IL-5, IL-13, IL-15, IL-31) were low or undetectable in tape-stripped skin, but were highly expressed in skin biopsies and plasma of paediatric AD patients [23, 25]. Similarly, Del Duca et al. showed that dermal T-cell cytokines IL-22 and IFN-γ were uniquely upregulated in skin biopsies compared to tape strips in adult AD patients [24]. Accordingly, a consistent pattern emerges wherein tape strips effectively capture local innate and barrier-related proteins but demonstrate limited sensitivity for detecting deeper, T cell-derived cytokines.

Our results demonstrate that tape strips accurately capture a rapid suppression of Type 2 immune mediators TARC and PARC, in both lesional and non-lesional skin, along with innate pro-inflammatory cytokines IL-8 and IL-18, and Th1-related chemokine CXCL10 in lesional skin, aligning with findings in adults [18, 20]. Although not assessed in studies of adult patients treated with dupilumab, we additionally observed a significant suppression of T cell-recruiting chemokine CTACK during treatment in lesional skin. IL-4 and IL-13 promote dendritic cell maturation and activation, which correlate with AD severity and drive Th2 skewing [32, 33]. Our findings suggest that dupilumab, by blocking the IL-4/IL-13 pathway, significantly reduces mediators associated with dendritic cell activity in lesional skin, including TARC, PARC, IL-18, and CXCL10. Although CTACK is not directly associated with dendritic cells, it is induced by pro-inflammatory cytokines [34]. Consequently, by reducing pro-inflammatory proteins, dupilumab may indirectly reduce CTACK levels. Furthermore, CTACK plays a critical role in recruiting CCR10+ T cells to the epidermis, which subsequently promotes the expression of pro-inflammatory proteins that interact with dendritic and other cells [35, 36]. Thus, a reduction of CTACK may in turn suppress the associated pro-inflammatory cascade.

Beyond its effects in the skin, dupilumab treatment resulted in a rapid decrease in serum TARC and PARC levels, supporting previous paediatric studies evaluating its impact on serum proteins [8-11]. While TARC and PARC levels significantly decreased during treatment in both skin and serum, reductions in CTACK were observed only in lesional skin. The lack of detectable changes in serum CTACK levels may reflect the spatial variability of protein expression within the skin. Tape strips primarily sample the stratum corneum, where CTACK is expressed by epidermal keratinocytes [36]. Conversely, TARC and PARC are predominantly produced in deeper dermal layers, which may explain their presence in both skin and serum [37, 38]. These findings highlight the complementary roles of tape strips and serum in capturing treatment-induced proteomic changes in AD.

Notably, we observed a significant increase in IL-1α levels in lesional tape-stripped skin during treatment, trending towards levels observed in non-lesional skin. Consistent with previous studies, IL-1α expression is generally lower in lesional tape-stripped skin compared to non-lesional or healthy skin, hypothesised to result from skin barrier damage [22, 25, 39]. As IL-1α is crucial for local inflammatory responses that maintain skin integrity (i.e., inducing wound repair) and support barrier function, its increasing levels in lesional skin may reflect barrier repair and remodelling processes induced by dupilumab treatment [40].

Dupilumab treatment also significantly increased serum IL-4 and IL-13 levels, likely due to its blocking effect on IL-4Rα, which prevents the binding of IL-4 and IL-13, resulting in elevated unbound circulating levels [8]. The greater increase in serum IL-4 relative to IL-13 suggests that IL-4Rα is more frequently bound by IL-4 in untreated patients. This is supported by findings suggesting that IL-13 predominantly drives peripheral inflammation, as it is more abundant in the skin, while IL-4, which is low or undetectable in the skin, primarily exerts central effects [41-43].

We identified proteins correlating with clinical severity, measured by EASI and NRS itch, with lesional tape strips showing the strongest correlations. Notably, PARC emerged as the strongest correlated protein. Similarly, previous studies have reported positive correlations between skin and serum PARC levels and clinical severity [44]. In some studies, PARC even showed a stronger correlation with clinical severity than TARC, which showed a significant but less pronounced correlation in our study [45]. Guttman et al. reported a positive correlation between PARC in tape strips and the Pruritus Atopic Dermatitis Quickscore, but not with EASI or Scoring Atopic Dermatitis (SCORAD), whereas TARC showed no correlation [26]. Esaki et al. and Ungar et al. reported a positive correlation between PARC in skin biopsies and SCORAD, which was not observed for TARC [28, 46]. Khattri et al. found a slightly stronger correlation between PARC in skin biopsies and SCORAD compared to TARC [47]. Although evidence for the generalisability of PARC as an AD biomarker remains limited due to the small number of studies, these findings suggest its potential and warrant further investigation [45].

Other proteins correlating with clinical severity in lesional and non-lesional skin included innate pro-inflammatory cytokines (IL-8, IL-18), T cell-recruiting chemokine CTACK, and tissue repair and remodelling proteins (periostin, MMP-1). These results align with prior studies investigating tape-stripped skin from paediatric AD patients, demonstrating positive correlations between TARC, CTACK, IL-8, and IL-18 and clinical severity (EASI, SCORAD) [23, 25]. In serum, IL-22 and TARC were notable correlated proteins with NRS itch, consistent with the findings of Kishi et al., demonstrating a positive correlation between IL-22 and TARC levels and Visual Analogue Scale Itch in adult patients. Kishi et al., along with other studies, also observed a positive correlation of these proteins with EASI and/or SCORAD [45, 48-51]. However, while serum TARC strongly correlated with NRS itch in our study, its correlation with EASI was weak and non-significant, possibly due to the small sample size.

With the development of novel targeted treatments for paediatric AD, proteomic profiling is increasingly important for enhancing patient and treatment selection, treatment monitoring, monitoring side effects, and tapering strategies [15, 52, 53]. It may also help predict treatment response and AD flares. In our study, one patient experienced an AD flare after 16 weeks of dupilumab treatment. Interestingly, levels of IL-18 and CTACK were already elevated at week 4 in lesional tape-stripped skin. These proteins, along with IL-8 and PARC, further increased up to the AD flare (week 16). These findings demonstrate the clinical relevance of these proteins and support the potential of tape strips for predicting treatment response and AD flares. However, given the single-patient observation, these findings require cautious interpretation and validation in larger cohorts before clinical implementation.

This study has several limitations, including the relatively small sample size. Furthermore, the real-world nature of our BioDay registry did not allow inclusion of a healthy control group. However, by comparing lesional and non-lesional skin, and pre- and post-treatment samples within the same patient, each patient effectively served as their own control. The real-world clinical setting allowed patients to use concomitant topical therapies, which possibly skewed the immunologic response. Nevertheless, sub-analyses revealed no significant differences in protein levels at week 16 between patients using topical corticosteroids and those who did not (Table S6). Moreover, patients were followed for 16 weeks, while longer follow-up is needed to assess the long-term persistence of dupilumab-induced proteomic changes. Additionally, we eluted tape strips in a buffer containing serum albumin to reduce skin protein loss via adsorption onto plastic surfaces, limiting comparison to other studies that normalise eluted proteins to the total amount of proteins. Lastly, a limited panel of proteins was used, thereby potentially excluding other relevant proteins.

In conclusion, our findings provide insight into the distinct local and systemic proteomic changes in response to dupilumab treatment in paediatric AD patients. These findings underscore the complementary roles of tape strips and serum in profiling immune and epidermal barrier proteins, highlighting the utility of minimally invasive tape stripping for monitoring proteomic responses to targeted therapies. Moreover, several proteins measured in both tape-stripped skin and serum correlated with clinical severity, suggesting their potential utility as biomarkers for disease and treatment response monitoring.

Author Contributions

Lisa P. van der Rijst: conceptualization; investigation; methodology; writing – original draft preparation; visualization; writing – review & editing; formal analysis; data curation. Edward F. Knol:conceptualization; methodology; investigation; writing – review & editing; formal analysis. Nicolaas P.A. Zuithoff: methodology; formal analysis; writing – review & editing. Constance F. den Hartog Jager: investigation; formal analysis; data curation. Femke van Wijk: conceptualization; methodology; project administration; resources; writing – review & editing; supervision. Marjolein S. de Bruin-Weller: conceptualization; funding acquisition; methodology; project administration; resources; writing – review & editing; supervision. Marlies de Graaf: conceptualization; writing – original draft preparation; methodology; project administration; resources; writing - review & editing; supervision.

Ethics Statement

All included patients participated in the Dutch BioDay registry and (parents) provided written informed consent (Medical Ethics Committee, University Medical Center Utrecht; approval 18/239). In addition, all included patients (and/or parents) provided written informed consent for the collection of blood and tape strip samples during the observation period (Medical Ethics Committee, University Medical Center Utrecht; approval 23/014).

Conflicts of Interest

Lisa P. van der Rijst has been a speaker for AbbVie and Novartis; fees were paid directly to the institution. Femke van Wijk has previously served as a speaker and/or consultant for Janssen, Johnson & Johnson, and Takeda and has received grants from Regeneron Pharmaceuticals, LEO Pharma, Sanofi, BMS, Galapagos, and Takeda. Edward F. Knol has previously served as a speaker and consultant for Sanofi, Thermo Fisher Scientific, Astra Zeneca, and GSK. Nicolaas P.A. Zuithoff was involved in studies financed by Eli Lilly; fees were paid directly to the institution. Constance F. den Hartog Jager reports no conflict. Marjolein S. de Bruin-Weller has been a consultant, advisory board member, and/or speaker for AbbVie, Almirall, Amgen, Aslan, Eli Lilly, Galderma, Janssen, LEO Pharma, Pfizer, Regeneron Pharmaceuticals, and Sanofi; fees were paid directly to the institution. Marlies de Graaf has been a consultant, advisory board member and/or speaker for AbbVie, Almirall, Eli Lilly, ALK, Janssen, LEO Pharma, Novartis, Pfizer, Regeneron Pharmaceuticals and Sanofi; fees were paid directly to the institution.