Integrated bioinformatics and machine learning identify early diagnostic biomarkers for MAFLD with comorbid psoriasis.
Psoriasis and metabolic dysfunction-associated fatty liver disease (MAFLD) share pathological features such as chronic inflammation, immune dysregulation, and metabolic disturbance. Increasing evidence suggests biological crosstalk between the two conditions, offering new insights into their shared mechanisms and comanagement. Early-stage MAFLD, characterized by hepatic steatosis without evident inflammation or fibrosis, provides a crucial window for intervention. This study aimed to identify early diagnostic biomarkers linking psoriasis and MAFLD.
Transcriptomic datasets of psoriasis and MAFLD were retrieved from the Gene Expression Omnibus (GEO) database. Weighted gene co-expression network analysis (WGCNA) was used to identify MAFLD-related modules. Shared genes were obtained by intersecting module genes with differentially expressed genes (DEGs) from psoriasis datasets. Machine learning algorithms, including random forest (RF), least absolute shrinkage and selection operator (LASSO), and support vector machine (SVM), were applied to identify hub genes. Diagnostic performance was evaluated using receiver operating characteristic (ROC) analysis, immune infiltration assessment, Spearman correlation, and experimental validation in psoriasis and MAFLD mouse models.
Twenty-nine shared genes were identified and found to be enriched in immune and metabolic pathways. Six hub genes-ADRB2, WNT5A, S100A9, FAM110C, S100A12, and TUBB6-were selected through integrated machine learning analysis and experimental validation. These genes exhibited high diagnostic accuracy and significant correlations with disease severity and immune cell infiltration.
This study identified six hub genes-ADRB2, WNT5A, S100A9, FAM110C, S100A12, and TUBB6-as potential cross-disease biomarkers for the comorbidity of psoriasis and MAFLD, and these genes are significantly associated with disease severity. These findings provide new targets for early diagnosis and potential treatment strategies for the comorbidity.
Transcriptomic datasets of psoriasis and MAFLD were retrieved from the Gene Expression Omnibus (GEO) database. Weighted gene co-expression network analysis (WGCNA) was used to identify MAFLD-related modules. Shared genes were obtained by intersecting module genes with differentially expressed genes (DEGs) from psoriasis datasets. Machine learning algorithms, including random forest (RF), least absolute shrinkage and selection operator (LASSO), and support vector machine (SVM), were applied to identify hub genes. Diagnostic performance was evaluated using receiver operating characteristic (ROC) analysis, immune infiltration assessment, Spearman correlation, and experimental validation in psoriasis and MAFLD mouse models.
Twenty-nine shared genes were identified and found to be enriched in immune and metabolic pathways. Six hub genes-ADRB2, WNT5A, S100A9, FAM110C, S100A12, and TUBB6-were selected through integrated machine learning analysis and experimental validation. These genes exhibited high diagnostic accuracy and significant correlations with disease severity and immune cell infiltration.
This study identified six hub genes-ADRB2, WNT5A, S100A9, FAM110C, S100A12, and TUBB6-as potential cross-disease biomarkers for the comorbidity of psoriasis and MAFLD, and these genes are significantly associated with disease severity. These findings provide new targets for early diagnosis and potential treatment strategies for the comorbidity.
Authors
Wang Wang, Chen Chen, Yang Yang, Zhang Zhang, Liu Liu, Zhang Zhang, Yang Yang, Wei Wei, Qiu Qiu, Huang Huang, Xiao Xiao
View on Pubmed