Deciphering peri-implantitis: Unraveling signature genes and immune cell associations through bioinformatics and machine learning

Abstract

Early diagnosis of peri-implantitis (PI) is crucial to understand its pathological progression and prevention. This study is committed to investigating the signature genes, relevant signaling pathways and their associations with immune cells in PI. We analyzed differentially expressed genes (DEGs) from a PI dataset in the gene expression omnibus database. Functional enrichment analysis was conducted for these DEGs. Weighted Gene Co-expression Network Analysis was used to identify specific modules. Least absolute shrinkage and selection operator and support vector machine recursive feature elimination were ultimately applied to identify the signature genes. These genes were subsequently validated in an external dataset. And the immune cells infiltration was classified using CIBERSORT. A total of 180 DEGs were screened from GSE33774. Weighted Gene Co-expression Network Analysis revealed a significant association between the MEturquoise module and PI (cor = 0.6, P < .0001). Least absolute shrinkage and selection operator and support vector machine recursive feature elimination algorithms were applied to select the signature genes, containing myeloid-epithelial-reproductive tyrosine kinase, microfibrillar-associated protein 5, membrane-spanning 4A 4A, tribbles homolog 1. In the validation on the external dataset GSE106090, all these genes achieved area under curve values exceeding 0.95. GSEA analysis showed that these genes were correlated with the NOD-like receptor signaling pathway, metabolism of xenobiotics by cytochrome P450, and arachidonic acid metabolism. CIBERSORT revealed elevated levels of macrophage M2 and activated mast cells in PI. This study provides novel insights into understanding the molecular mechanisms of PI and contributes to advancements in its early diagnosis and prevention.