Discussion

This study aimed to identify biomarker patterns corresponding to the disease and the extent of clinical improvement upon systemic anti-IL4Rα treatment. We report on known (CCL17), but also somewhat novel (CCL13, E-selectin, CCL22, and CCL27) proteomic biomarkers in severe AD patients. Exploring these biomarkers upon systemic dupilumab treatment CCL17, CCL13, E-selectin CCL22, and CCL27 protein levels decreased in our patients, whereas the BDNF increased. Skin microbial composition showed a decrease in S. aureus colonization upon treatment, following the clinical improvement. By contrast, miRNAs could not be classified as potential biomarkers of therapy response.
Our data has led us to propose a panel of composite biomarkers which may assist in predicting clinical responses, defined as a decrease of less than 75% in initial SCORAD values and body surface area affected by atopic dermatitis. Despite the overall positive clinical efficacy of dupilumab, as demonstrated by the high percentage of patients who experienced substantial improvement, a subpopulation of non-responders who may benefit from other treatment options remains.
A report by Hamilton et al. also described changes in biomarkers after initiation of dupilumab systemic therapy in various atopic diseases and found that dupilumab suppressed multiple Th2 mediators such as CCL17, total IgE, periostin, and eotaxin-3, consistent with our findings20. Our study adds to the literature by identifying additional biomarker candidates through proteomic screening.
Observed changes of CCL17, CCL13, CCL27, and CCL22 after dupilumab therapy reflect a Th2-inflammation-dependent response. We also observed an increase in the levels of BDNF after 6 months of therapy. BDNF is a protein that is not directly dependent on Th2-pathway and is secreted from eosinophils in patients with atopic dermatitis21. Our study found that levels of BDNF were higher in patients with atopic dermatitis compared to non-atopic individuals, which is consistent with previous reports in both children22 and adults21. The observed increase in BDNF after dupilumab treatment may be a consequence of transient dupilumab-dependent blood eosinophilia but its relevance remains elusive23.
E-selectin was identified as a potential candidate for a good response to dupilumab therapy in our proteomic screening. E-selectin has been previously reported to be increased in atopic dermatitis and correlated with disease severity24. The decrease in both CCL17 and E-selectin may be a result of decreased inflammation in the skin, as these molecules are known to be coexpressed in dermal microvessels25. Like CCL17, E-selectin plays an important role in leukocyte migration into inflamed skin. Soluble E-selectin may result from the proteolytic cleavage of its membrane-bound form, and its presence in the serum could reflect the general state of skin inflammation24.
It is important to note that the observed changes in biomarker levels in our study may be the result of the decrease in clinical severity rather than a direct effect of dupilumab therapy. Other systemic therapeutics have also been shown to decrease cytokines downstream in the inflammatory cascade. For example, previous studies have reported reductions in CCL17, IL13, and IL22 levels after treatment with cyclosporine A (CyA)26, tralokinumab27, or fezakinumab28 in patients with atopic dermatitis. The identification of biomarkers specific to a particular therapy may offer insights into the mechanisms of the disease. However, it is arguably more important to identify biomarkers that specifically represent the severity of the condition for effective clinical management.
Many groups demonstrated a correlation between CCL17 levels and the SCORAD score29–31. However, atopic dermatitis is a complex and heterogeneous disease, with multiple contributing factors such as intrinsic and extrinsic pathways, skin barrier defects, and genetic background, involving multiple cell types. This heterogeneity makes it difficult to rely on a single biomarker as an objective indicator of disease severity.
Limitations have been identified with commonly used clinical scores such as SCORAD and EASI32 for assessing atopic dermatitis, which undermines their use in clinical studies33,34. These scores are often semi-quantitative and subjective, leading to a lack of objective indicators for disease severity, making it challenging to monitor the disease in a longitudinal manner35,36. Therefore, a set of biomarkers (serological and/or histological) may be needed to objectively address the severity of the disease, as it might be more suitable as a baseline truth than widely used subjective clinical scoring systems37.
A recently published study by Nakahara et al. underlined the difficulties to identify a biomarker that correlated with the severity of AD as measured by the EASI score during a 16-week therapy period38. Notably, the study suggested that two biomarkers, CCL17 and CD25s, have the potential to predict response to dupilumab therapy as measured by POEM and pruritus NRS scales. These findings align with the results of our study, in which CCL17 and CD25s were included in our composite classification model (along with Notch1 and E-selectin) and performed as the best predictors of a good response to therapy.
We did not observe significant changes in miRNA expression after treatment with dupilumab. miRNA levels in the blood change quickly in acute conditions14 and may reach a state of equilibrium in a relatively short time, which could explain observed similar levels of miRNA before and after therapy. Alternatively, the similarities of miRNome before and after treatment may reflect the fact that dupilumab is managing the downstream effects of Th2 inflammation which does not play a major role in miRNA expression. Nevertheless, we were able to identify differences in miRNA levels between atopic dermatitis, psoriasis, and healthy individuals that may be disease dependent.
In a recent study, miRNA expression levels in the skin and serum of AD and psoriasis patients were compared to healthy individuals39. The authors reported differentially expressed levels of (among others): hsa-miR-378a-3p, hsa-miR-146b, and hsa-miR-25-3p in skin and hsa-miR-122-5p in serum. These results are in line with our findings, although the authors used different analytical methods (Affymetrix arrays) and different biomaterials. However, the authors report significantly more differentially expressed miRNA from skin samples than from serum, suggesting that skin biopsies are a more suitable material for identifying mechanistically relevant miRNA biomarkers.
In our study, we identified a treatment-related shift in the composition of microorganisms colonizing the skin of AD patients, which was associated with the severity of the disease (IGA, SCORAD, and BSA). The use of the microbiome as a biomarker for AD severity has been proposed in recent literature11, but there is currently a lack of consensus on standardized methodology and a need for further large-scale, longitudinal clinical studies to establish its usefulness. A previous study reported that dupilumab treatment decreased skin colonization with S. aureus and increased skin microbial diversity in children with AD40. Our findings extend these observations to adults. However, we did not find a clear correlation between baseline skin microbial composition and treatment outcomes.
Taken together we identified several proteomic and microbial, but no miRNA, biomarkers in AD patients receiving systemic IL-4Ra targeted treatment. Based on our clinical data, we were able to identify a set of protein-based biomarkers associated with a good clinical response. Nevertheless, these biomarkers need to be validated in a larger cohort during a longer follow-up period. Although the lack of skin based analysis might be regarded as a weakness of our study, the measurement of biomarkers in a patient’s serum is more suitable in daily clinical practice and should be further explored to select the optimal treatment in a currently broadening spectrum of targeted therapies.