Natural moisturizing factor as a biomarker for filaggrin mutation status in a multi‐ethnic paediatric atopic dermatitis cohort

Atopic dermatitis (AD) is a common inflammatory skin disease among children with increasing prevalence in the past decades. The strongest and most widely replicated genetic risk factor for AD is a null mutation in the filaggrin gene (FLG) located on chromosome 1q21. FLG encodes the protein filaggrin, which is involved in the formation and homeostasis of the skin barrier.1 Previous research showed that AD patients with a mutation in FLG have a different phenotype, characterized by early onset disease with persistence into adulthood, increased severity and increased risk of asthma and allergic sensitization.1 Additionally, it has been suggested that patients with a mutation in FLG respond differently to immunosuppressive treatment compared to wild-type patients.


Natural moisturizing factor as a biomarker for filaggrin mutation status in a multi-ethnic paediatric atopic dermatitis cohort
To the Editor, Atopic dermatitis (AD) is a common inflammatory skin disease among children with increasing prevalence in the past decades. The strongest and most widely replicated genetic risk factor for AD is a null mutation in the filaggrin gene (FLG) located on chromosome 1q21. FLG encodes the protein filaggrin, which is involved in the formation and homeostasis of the skin barrier. 1 Previous research showed that AD patients with a mutation in FLG have a different phenotype, characterized by early onset disease with persistence into adulthood, increased severity and increased risk of asthma and allergic sensitization. 1 Additionally, it has been suggested that patients with a mutation in FLG respond differently to immunosuppressive treatment compared to wild-type patients. 1 This suggests that FLG mutation profiling could be used to stratify patients in terms of clinical course and to develop personalized treatment strategies.
However, genotyping is time consuming and expensive, and DNA collection poses ethical considerations, which hampers its use in daily practice. Recent literature suggests that decreased concentrations of filaggrin-derived components of the natural moisturizing factor (NMF) are a proxy for the presence of FLG-null mutations. 2,3 During the terminal differentiation of keratinocytes, profilaggrin is dephosphorylated and enzymatically degraded into a highly hygroscopic mixture of amino acids and amino acid derivatives, including pyrrolidone carboxylic acid (PCA), histidine and its metabolite urocanic acid (UCA). 2 The free amino acids and their derivatives constitute the majority of NMF in the stratum corneum (SC). 2 A previous study in a selected Irish paediatric population of AD patients showed that NMF could discriminate between FLG mutation carriers and wildtype with a sensitivity of 98.73% and a specificity of 86.89% using 1.07 arbitrary units (a.u.) as the cut-off value. 4 It has not been investigated whether this cut-off value could be applied to assess FLG mutation status in different clinical cohorts.
The aim of our study was to screen the entire encoding region of FLG for potential mutations and to validate whether NMF could be used as a biomarker for FLG genotype in an unselected multiethnic clinical cohort of children with mild-to-severe disease. We FLG mutations were determined on DNA isolated from buccal swabs using Isohelix SK-1S swabs (Cell Projects Ltd). Single-molecule molecular inversion probes (smMIPS) and barcoded next-generation sequencing (NGS) were performed to screen the entire encoding region of FLG for all mutations resulting in premature protein termination as previously described. 5 All patients with a mutation in FLG were referred to as patients with ≥1 mutation(s) (FLG − ). Wild-type patients were referred to as FLG + . The NMF content was measured in the SC of the thenar eminence (nonlesional skin) using confocal Raman spectroscopy (gen2-SCA Skin Composition Analyzer; RiverD International B.V.) as described previously. 4 All investigators were blinded for FLG genotype until the end of the study.
Acute AD severity was assessed using the Eczema Area and Severity Index (EASI) score on the same day as the NMF measurements. A receiver operating characteristic (ROC) curve was constructed to measure the diagnostic ability of NMF to predict FLG mutation status (wild-type versus ≥1 mutation) by mapping the sensitivity versus 1-specificity for all possible values of the cut-off point. The optimal cut-off point was determined by maximizing the Youden function. A multivariate linear regression model was used to examine the association between EASI (as independent variable) and NMF content, corrected for FLG mutation status, age and sex.
A total of 101 patients were included in this study ( Figure S1).
We identified 12 different FLG mutations, corresponding to 30 (30%) patients with ≥1 mutation in FLG (Table S1). This included 25 heterozygous mutation carriers and 5 patients with more than one mutation in FLG. There were no differences in age, sex, ethnicity and disease severity between both groups (Table S2) Furthermore, the NMF content in patients with a single mutation in FLG was significantly different compared to both wild-type patients and patients with 2 mutations in FLG (p < .01, Figure S2). ROC curves were constructed to test the diagnostic ability of NMF measured using Raman spectroscopy. The optimal cut-off value for NMF to distinguish between FLG − and FLG + patients was 1.03 a.u. with an area under the curve (AUC) of 0.93 (95% CI 0.87-0.99) (Figure 2). This resulted in a sensitivity of 97%, specificity of 87%, positive predictive value of 76% and negative predictive value of 98% (Table S3). We did not find a significant association between the EASI score and NMF measured at the thenar eminence (corrected for FLG mutation status, age and sex) (beta −0.555 (95% C.I. −0.005-0.003), p = .58, Table S4). Our results showed no correlation between EASI score and NMF value, making the nonlesional skin of the thenar eminence a suitable location to predict FLG mutation status without direct interference of acute disease severity. Previous studies have shown that filaggrin degradation products in the SC of the nonlesional skin of the forearm are affected by both a mutation in FLG and disease severity. 6 The effect of disease severity can be attributed to the activated immune system in both lesional and nonlesional skin. Previous research, using a tape stripping technique, showed an up-regulation of markers for AD severity including T helper 2 (Th2)-skewed markers (interleukin [IL]-13, CCL17, CCL22, IL-5) in the nonlesional skin of the forearm, which was associated with a reduced NMF content. 6 The current results did not show a correlation between acute disease severity and NMF measurement on the thenar eminence. This might be explained by the thicker SC with a slower turnover time as compared to the SC on the forearm, making the thenar eminence less susceptible to an acute up-regulation of the immune response. Determining both cytokines and NMF values in the SC of thenar eminence and correlating this to disease severity could support our hypothesis.
Major strengths of this study are the application of the noninvasive clinically compatible NMF measurement in an unselected multi-ethnic patient cohort and the use of a novel technique to detect all mutations leading to a premature protein termination in FLG.
Especially in this multi-ethnic population, screening of only the most common mutations is not sufficient and will lead to an underrepresentation of FLG mutations. A limitation to mention is that patients were instructed not to apply any topical therapies on the thenar eminence 24 h before the measurement, but information on the use of these ointments during the rest of the week prior to the visit was missing. Previous research has suggested that topical steroid treatment could decrease the level of NMF in the SC of both mice and humans. 7 Future research should focus on other factors, next to FLG-null variants, leading to a reduction in the NMF. Furthermore, it

Key message
• Raman spectroscopy is a rapid and non-invasive technique to measure natural moisturizing factor content in AD • Natural moisturizing factor can predict FLG genotype with a sensitivity of 97% and specificity of 87% • The thenar eminence's nonlesional skin is a suitable location to predict FLG genotype is of interest to validate the cut-off value of 1.03 a.u. in other clinical cohorts and to evaluate its use as a predictor for treatment outcomes to enable personalized treatment.
In conclusion, we validated that the NMF content in the SC of the thenar eminence can be used as biomarker for FLG status in an unselected clinical cohort of children with AD from different ethnical backgrounds. This will be useful to identify patients with high risk for a more severe phenotype and to stratify patients in future clinical studies to evaluate treatment response.

ACK N OWLED G EM ENTS
This project is part of the European Reference Network (ERN)-SKIN. We

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.