Metabolomics revealed alterations in serum lipid metabolite levels in childhood atopic dermatitis

Atopic dermatitis (AD) is a chronic inflammatory skin disease that imposes a substantial burden on healthcare systems and diminishes the quality of life for patients. It affects 10%–20% of children, particularly those under the age of 2 years. Children with AD not only experience skin discomfort but also have a high risk of developing asthma and allergic rhinitis.1 Therefore, there is a need to identify biomarkers that can provide insights into the underlying pathogenesis of AD. Metabolites are substrates and end products in diverse biochemical reactions. Their levels are tightly regulated by cellular states, providing comprehensive information about biological states. With advancements in analytical techniques and recently updated human databases, metabolomics has been considered a promising strategy for capturing the dynamic of metabolism.2 Although metabolomics-based approaches are powerful tools for exploring the overall cellular states, most metabolic studies in AD lack an explanation regarding the relationship between altered metabolites and clinical state. This indicates that these biomarkers may possess limited reliability. In this study, we aimed to gain a new perspective on AD and discover potential biomarkers for AD in children. We investigated plasma metabolites in AD children using ultra-performance lipid chromatography with a quadruple time-of-flight mass spectrometry system. We analyzed the correlation between key metabolites and clinical parameters. The study population consisted of 41 children (24 healthy controls and 17 AD children) involved in the Cohort for Childhood Origin of Asthma and Allergic Diseases3 (Table S1). A total of 1260 mass ions (in positive mode) and 1178 mass ions (in negative mode) were identified at specific retention time points. We selected five differential molecules in the plasma of children with AD based on the combined criteria of p-values (adjusted p < .1 by Benjamini–Hochberg correction) and variable importance in projection scores (VIP >1.0) from the PLS-DA model (Figure S1). To control the possible covariate effect of feeding type on the relationship between AD and key metabolites, we conducted an analysis of covariance. Even after adjusting cofounder effects, there were still significant differences in the metabolite levels between the control and AD groups (data not shown). Notably, all key features were annotated as lysophosphatidylcholines (lysoPC), and their levels were significantly reduced in AD children compared with healthy children (Figure 1A). To evaluate the discriminative ability of the key metabolites in diagnosing AD, we conducted a univariate receiver operating characteristic curve analysis (Figure 1B). Differential metabolites exhibited area under the curve (AUC) values above 0.74. Among them, lysoPC(18:1), which is 7.9_566.3463 (retention time_mass-to-charge ratio), showed the highest AUC value of 0.80, along with a sensitivity of 88% and specificity of 71%. Furthermore, we performed Spearman's correlation analysis to examine the relationship between selected metabolites and clinical parameters. The clinical parameters included the severity index of AD (scoring of atopic dermatitis, SCORAD), the total serum IgE (kU/L), specific IgE to D.f (kU/L), and eosinophil (%). We observed a negative correlation between key metabolites and clinical data (Figure 2). Specifically, plasma levels of lysoPC (16:1) showed a negative association with D.f-lgE (Spearman's rho [ρ] = −.39, p = .08) and eosinophil ([ρ] = −.4, p = .08). To sum up, these findings indicate that the levels of differential metabolites were negatively correlated with clinical parameters of allergic inflammation, highlighting their potential as promising biomarkers for AD. Several metabolomics studies have reported metabolic alterations occurring during chronic inflammation induced by AD.4 For example, a study on adults aged 20–59 years with AD observed that plasma levels of lysoPCs and lysoPEs were decreased in the AD compared with healthy controls.5 Additionally, a large-scale clinical study including 629 mother–child pairs for respiratory disease in infants found a negative correlation between plasma levels of lysoPCs and the expressional levels of the inflammasome.6 Our findings about metabolite expression align with these previous reports. In this study, we additionally reported a correlation between lysoPCs and clinical parameters of IgE and eosinophil in AD patients (Figure 2). Since the number of metabolites selected on the basis of combined criteria (FDR <0.1 and VIP >1.0) was too small, we attempted to expand our analysis by including additional features with lower statistical power (uncorrected p-value <.05 and VIP score >1.0). While this approach requires a cautious interpretation of the results, it may provide a broader perspective and help identify a wider range of potential biomarkers. Notably, a significant proportion of newly annotated metabolites belonged to the glycerophospholipid and were annotated as lysoPC (Table S2). The similarity in metabolite types of the newly identified metabolites indicates that lysoPC may be involved in the pathogenesis of AD. Furthermore, we observed that the intensities of all lysoPCs identified in this study were decreased in children with AD compared with healthy controls. These results suggest a potential association between reduced lysoPC levels and the occurrence of AD in children. However, it is important to note that further research is needed to confirm these results and understand the underlying mechanisms behind this observation. LysoPC is produced from PC catalyzed by phospholipase A2. We reported several lysoPCs of the same species because lysoPCs have the potential to exist as positional isomers or stereoisomers, which refer to different arrangements of the fatty acyl chain and variations in the stereochemistry of the molecule.7 LysoPC could regulate a broad range of cellular processes, including cell proliferation, tumor cell invasiveness, and inflammation.8 The immunomodulatory effect of lysoPC depends on the type of fatty acyl group.8 In this study, the majority of identified lysoPCs were unsaturated lysoPCs, and interestingly, these species can exhibit anti-inflammatory properties by inhibiting the production of leukotriene C4, tumor necrosis factor-alpha, and interleukin-6.9 However, saturated or monounsaturated lysoPC, which can induce inflammation, also was decreased in the plasma of AD children (Table S2). These contradictory findings have been previously reported.6, 10 Thus, it may be challenging to define the complex AD pathophysiology based on a single metabolite. Further study is needed to elucidate the relationship between AD pathogenesis and lysoPC levels. Our study had limitations due to its small sample size. Further studies with larger cohorts are needed to validate these findings. Also alternative extraction methods such as MeOH/chloroform could be considered to obtain a comprehensive metabolic profile in the future. Nevertheless, we showed relationships between clinical parameters and metabolites. Furthermore, our subject characteristics had a clear history of AD until at least 5 years of age. Since the subjects have clinically confirmed AD without other allergic diseases, this study provides more direct evidence of AD. Another point to consider is that there was a significant difference in the use history of steroid lotion between the two groups (Table S1). However, since our AD subjects did not use steroid lotion within 2 weeks prior to this analysis, there would have been no acute effects of steroids on the serum. We highlighted the potential role of lysoPC in distinguishing AD children from healthy controls. Our data indicate that specific lysoPC was negatively correlated with clinical parameters of allergic inflammation. It suggests that plasma lysoPCs could serve as useful biomarkers for AD in children. Jinwoo Kim: Writing – original draft; methodology; formal analysis; visualization; project administration; writing – review and editing. Yoon Mee Park: Writing – original draft; conceptualization; formal analysis; writing – review and editing; investigation. So-Yeon Lee: Conceptualization; project administration; investigation. Bong-Soo Kim: Project administration. Yun Kyung Lee: Project administration. Dong-Woo Lee: Project administration. Myung Hee Nam: Writing – original draft; writing – review and editing; funding acquisition; methodology; project administration; investigation. Soo-Jong Hong: Conceptualization; investigation; funding acquisition; writing – original draft; writing – review and editing; project administration. This study was supported by the Bio & Medical Technology Development Program of the National Research Foundation of Korea (NRF) funded by the Korean government (MSIT; NRF-2018M3A9F3056901, 2017M3A9F3043834, and 2022M3H9A2083956) and the Korean Centers for Disease Control and Prevention (2008-E33030-00, 2009-E33033-00, 2011-E33021-00, 2012-E33012-00, 2013-E51003-00, 2014-E51004-00, 2014-E51004-01, 2014-E51004-02, 2017-E67002-00, 2017-E67002-01, 2017-E67002-02, 2020E670200, 2020E670201, and 2020E670202). The authors declare no conflicts of interest in relation to this study. The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/pai.14021. Appendix S1. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.