Machine learning reveals distinct gene signature profiles in lesional and nonlesional regions of inflammatory skin diseases

Abstract

Analysis of gene expression from cutaneous lupus erythematosus, psoriasis, atopic dermatitis, and systemic sclerosis using gene set variation analysis (GSVA) revealed that lesional samples from each condition had unique features, but all four diseases displayed common enrichment in multiple inflammatory signatures. These findings were confirmed by both classification and regression tree analysis and machine learning (ML) models. Nonlesional samples from each disease also differed from normal samples and each other by ML. Notably, the features used in classification of nonlesional disease were more distinct than their lesional counterparts, and GSVA confirmed unique features of nonlesional disease. These data show that lesional and nonlesional skin samples from inflammatory skin diseases have unique profiles of gene expression abnormalities, especially in nonlesional skin, and suggest a model in which disease-specific abnormalities in “prelesional” skin may permit environmental stimuli to trigger inflammatory responses leading to both the unique and shared manifestations of each disease.

The interferon-rich skin environment regulates Langerhans cell ADAM17 to promote photosensitivity in lupus

“…The IFN-I pathway was comprised of several upregulated IFN-stimulated transcription factors, including IRF 1, 2, 7, 8, and 9 along with multiple IRF targets (ISG15, IFIT1, IFIT3, MX1, MX2) (Figure 1C-D, S1A). Notably, CD207 , the gene encoding Langerin and a lineage marker of LCs, was downregulated (Figure 1C), consistent with accumulating data on LC dysfunction in lupus (Billi et al, 2022; Martínez et al, 2022; Shipman et al, 2018). A supervised approach to analyzing normal control versus non-lesional DLE skin gene expression data, then, yielded results consistent with other cohorts showing upregulation of IFN pathways and LC dysfunction.…”

“…Recent studies have suggested a high IFN-I environment in non-lesional skin of lupus patients based on single cell RNAseq of keratinocytes from SLE patients and skin from CLE patients (Billi et al, 2022; Der et al, 2017; Der et al, 2019), CLE keratinocyte IFN-κ expression (Stannard et al, 2017), and upregulation of IFN-stimulated genes (ISGs) such as MX1 on tissue sections in SLE and incomplete SLE patients (Lambers et al, 2019; Reefman et al, 2008). Consistent with these findings, our recent analysis of publicly available bulk transcriptomic datasets from non-lesional DLE skin using Gene Set Variation Analysis (GSVA), an unsupervised gene set analysis approach, showed a proportion of samples with upregulated IFN gene set expression when compared to healthy controls, although this effect was modest when compared to the upregulation in lesional skin (Martínez et al, 2022). Here, we examined an additional non-lesional skin dataset from a CLE cohort with the discoid form of CLE (DLE) and also non-lesional skin from multiple lupus models to understand shared pathways.…”

“…We next performed GSVA using previously defined lupus-relevant gene sets (Martínez et al, 2022) (Table S3). Analysis of the non-lesional samples along with the previously analyzed lesional and psoriatic samples (Jabbari et al, 2014) reiterated the GSVA results of multiple datasets.…”

“…Gene sets used as input for GSVA are listed in Tables S3 and S7. The human gene sets were previously published (Martínez et al, 2022). The mouse plasma cell, T cell, myeloid cell, neutrophil, pDC, dendritic cell, and keratinocyte gene sets were derived from Mouse CellScan, a tool for the identification of cellular origin of mouse gene expression datasets.…”

Hepatitis B virus pathogenesis relevant immunosignals uncovering amino acids utilization related risk factors guide artificial intelligence-based precision medicine

“…Among these tools, GSVA has been widely used in tumor (Charoentong et al, 2017;Deng et al, 2019;Shen et al, 2019;Xiao et al, 2020;Gong et al, 2021;Zhuang et al, 2021) and nontumor researches (Hu et al, 2021;Shen et al, 2021;Yu et al, 2021) for core module identification at the gene-set level. AIbased models constructed using low-dimensional biological pathway data generated by GSVA as inputs have become popular and demonstrate promising effects (Chawla et al, 2022;Martinez et al, 2022). However, the application of GSVA-derived core immunosignals with even lower dimensionality for efficient feature selection, which benefits machine learning and deep learning in precision oncology, has not been researched.…”