通过机器学习和三种皮肤损伤中的动态免疫浸润发现银屑病的生物标志物。

Discovery of biomarkers in the psoriasis through machine learning and dynamic immune infiltration in three types of skin lesions.

机构信息

Department of Dermatology, The Second Xiangya Hospital of Central South University, Changsha, Hunan, China.

School of Mathematics and Statistics, Central South University, Changsha, Hunan, China.

出版信息

Front Immunol. 2024 May 13;15:1388690. doi: 10.3389/fimmu.2024.1388690. eCollection 2024.

DOI:10.3389/fimmu.2024.1388690

PMID:38803495

原文链接:https://pmc.ncbi.nlm.nih.gov/articles/PMC11128609/

Abstract

INTRODUCTION

Psoriasis is a chronic skin disease characterized by unique scaling plaques. However, during the acute phase, psoriatic lesions exhibit eczematous changes, making them difficult to distinguish from atopic dermatitis, which poses challenges for the selection of biological agents. This study aimed to identify potential diagnostic genes in psoriatic lesions and investigate their clinical significance.

METHODS

GSE182740 datasets from the GEO database were analyzed for differential analysis; machine learning algorithms (SVM-RFE and LASSO regression models) are used to screen for diagnostic markers; CIBERSORTx is used to determine the dynamic changes of 22 different immune cell components in normal skin lesions, psoriatic non-lesional skin, and psoriatic lesional skin, as well as the expression of the diagnostic genes in 10 major immune cells, and real-time quantitative polymerase chain reaction (RT-qPCR) and immunohistochemistry are used to validate results.

RESULTS

We obtained 580 differentially expressed genes (DEGs) in the skin lesion and non-lesion of psoriasis patients, 813 DEGs in mixed patients between non-lesions and lesions, and 96 DEGs in the skin lesion and non-lesion of atopic dermatitis, respectively. Then 144 specific DEGs in psoriasis via a Veen diagram were identified. Ultimately, UGGT1, CCNE1, MMP9 and ARHGEF28 are identified for potential diagnostic genes from these 144 specific DEGs. The value of the selected diagnostic genes was verified by receiver operating characteristic (ROC) curves with expanded samples. The the area under the ROC curve (AUC) exceeded 0.7 for the four diagnosis genes. RT-qPCR results showed that compared to normal human epidermis, the expression of UGGT1, CCNE1, and MMP9 was significantly increased in patients with psoriasis, while ARHGEF28 expression was significantly decreased. Notably, the results of CIBERSORTx showed that CCNE1 was highly expressed in CD4+ T cells and neutrophils, ARHGEF28 was also expressed in mast cells. Additionally, CCNE1 was strongly correlated with IL-17/CXCL8/9/10 and CCL20. Immunohistochemical results showed increased nuclear expression of CCNE1 in psoriatic epidermal cells relative to normal.

CONCLUSION

Based on the performance of the four genes in ROC curves and their expression in immune cells from patients with psoriasis, we suggest that CCNE1 possess higher diagnostic value.

摘要

简介

银屑病是一种以独特的鳞屑斑块为特征的慢性皮肤病。然而，在急性阶段，银屑病皮损表现出湿疹样改变，使其难以与特应性皮炎区分，这给生物制剂的选择带来了挑战。本研究旨在鉴定银屑病皮损中的潜在诊断基因，并探讨其临床意义。

方法

分析 GEO 数据库中的 GSE182740 数据集进行差异分析；机器学习算法（SVM-RFE 和 LASSO 回归模型）用于筛选诊断标志物；CIBERSORTx 用于确定正常皮肤损伤、银屑病非损伤皮肤和银屑病损伤皮肤中 22 种不同免疫细胞成分的动态变化，以及 10 种主要免疫细胞中诊断基因的表达，并通过实时定量聚合酶链反应（RT-qPCR）和免疫组织化学进行验证。

结果

我们在银屑病患者皮损和非皮损中获得了 580 个差异表达基因（DEGs），在混合患者非皮损和皮损中获得了 813 个 DEGs，在特应性皮炎皮损和非皮损中获得了 96 个 DEGs。然后通过 Veen 图鉴定出银屑病的 144 个特定 DEGs。最终，从这些 144 个特异性 DEGs 中鉴定出 UGGT1、CCNE1、MMP9 和 ARHGEF28 作为潜在的诊断基因。通过扩展样本的接收者操作特征（ROC）曲线验证了所选诊断基因的价值。四个诊断基因的 ROC 曲线下面积（AUC）均超过 0.7。RT-qPCR 结果显示，与正常人表皮相比，银屑病患者 UGGT1、CCNE1 和 MMP9 的表达明显增加，而 ARHGEF28 的表达明显降低。值得注意的是，CIBERSORTx 的结果表明，CCNE1 在 CD4+T 细胞和中性粒细胞中高表达，ARHGEF28 也在肥大细胞中表达。此外，CCNE1 与 IL-17/CXCL8/9/10 和 CCL20 呈强相关性。免疫组织化学结果显示，与正常相比，银屑病表皮细胞中 CCNE1 的核表达增加。