In a recent article published in Nature Communications, researchers investigated endotype-phenotype associations in atopic dermatitis (AD) beyond its general clinical severity.
Study: Multifaceted analysis of cross-tissue transcriptomes reveals phenotype–endotype associations in atopic dermatitis. Image Credit: Kmpzzz/Shutterstock.com
Background
AD is highly heterogeneous in its clinical manifestations and molecular profiles; moreover, scientists have acknowledged that it is a systemic disease with pathophysiology spanning the whole body. Repeated exacerbation and remission cycles of AD vary across patients. Previous omics-based studies of AD have not been able to account for these heterogeneous disease trajectories.
AD pathophysiology also involves crosstalk between the damaged organ (skin) and the circulatory system, which leads to concurrent biological alterations frequently causing cutaneous inflammation and malignancies.
Novel AD antibody therapeutics, such as an anti-interleukin-4 receptor subunit α (IL-4Rα) and an anti-IL-31Rα are now available; however, there is no consensus among clinicians concerning their use for individual patients given several endotypes linked to disease course. Moreover, studies have revealed how systemic treatment with these immunosuppressants only focuses on alterations in a specific part of the body, even though a complex disease like AD affects the whole body.
Overall, it is crucial to capture endophenotypes of individual AD patients using omics approaches, including transcriptomic and proteomics.
About the study
In the present study, researchers first performed a cross-sectional analysis using integrated ribonucleic acid (RNA)-sequencing on skin tissue and peripheral blood mononuclear cell (PBMC) samples from 151 AD patients and 19 healthy controls.
Of all the study participants, 30 AD patients and five healthy controls were female, with a mean age of 41.3 years. They linked this dataset to the patient's clinical data to capture molecular signatures of clinical profiles of this Japanese AD population.
Further, the team incorporated skin and PBMC transcriptome data of 115 AD patients and 14 healthy controls (cross-sectional dataset) into regression models to develop interpretable transcriptome modules.
Applying weighted gene co-expression network analysis (WGCNA) to the entire dataset identified 21 and 15 skin and PBMC transcriptional modules (sModus and pModus), respectively. Each comprised 51 to 774 genes that worked synchronously in a tissue, suggesting their biological relevance to each other.
Next, the team applied the analysis of the cross-sectional dataset to a time series dataset (total 360 time points) consisting of PBMC transcriptome, blood tests, and clinical severity scores from 30 AD patients for the longitudinal analysis.
It helped them monitor personalized AD progression and examine inter-patient heterogeneity in longitudinal features. Post-quality control, longitudinal analysis of the study encompassed 280 data points.
Further, the team performed a meta-analysis of clinical severity scores using 1424 data points obtained between November 2016 and July 2021. The elastic net regression model tested the longitudinal dataset to predict general disease severity.
The team measured the extent and severity of AD using the Eczema Area and Severity Index (EASI), which graded the severity of four signs of eczema, i.e., erythema, papulation, excoriation, and lichenification over the patient's trunk, head and neck, upper and lower extremities separately.
The median severity of each eczema sign over these regions was assigned a score between zero and three, indicating none, mild, moderate, and severe severity, respectively. They also performed multidimensional scaling (MDS) to capture (visually) the relationship between four components of eczema severity.
Results
The integrative analysis of transcriptome data from diseased tissue (skin) and circulatory system (blood) shed light on AD patient's endotype-phenotype associations. Endotypes in AD, i.e., biological subtypes defined based on tissue transcriptome analysis, were closely correlated to AD phenotypes depicted by visual skin evaluations. Indeed, AD-affected individuals have many pathophysiological subtypes.
The longitudinal analysis identified three patient clusters varying in disease course and medication history. Patient stratification based on longitudinal features in AD could be the first big step toward personalized medicine for this complex disease.
The team also noted that pModu07, pModu09, and neutrophil count were the top three factors contributing to the observed patient clustering. This finding suggested that the dynamics of innate immunity drove instability in longitudinal disease course.
It is worth noting that such assessment(s) appear unfeasible for routine clinical examinations because collecting biospecimens (other than blood) requires invasive sampling. Future work should identify biomarkers to predict system-level pathology in individual patients for these examinations.
Factors correlated to disease severity in individual patients were pModu01 and pModu04, in addition to previously recognized Ad biomarkers, serum thymus and activation-regulated chemokine (TARC), lactate dehydrogenase (LDH), and eosinophil counts.
Conclusions
The current study highlighted inter- and intra-patient heterogeneity in AD via multifaceted analyses of cross-tissue, cross-sectional, and longitudinal transcriptomes of AD-associated phenotypes and endotypes. This approach laid the groundwork for holistic clinical investigations of the pathophysiology of AD, which might lead to personalized AD treatment in the future.
Gut health and diet: Keys to treating acne, alopecia, and dermatitis