LAUSR

BACKGROUND AND OBJECTIVES Periodontitis, which is a chronic inflammatory periodontal disease resulting in destroyed periodontal tissue, is the leading cause of tooth loss in adults. Many studies have found that inflammatory immune responses are involved in the risk of periodontal tissue damage. Therefore, we analyzed the association between immunity and periodontitis using bioinformatics methods to further understand this disease. MATERIALS AND METHODS First, the expression profiles of periodontitis and healthy samples were downloaded from the GEO database, including a training dataset GSE16134 and an external validation dataset GSE10334. Then, differentially expressed genes were identified using the limma package. Subsequently, immune cell infiltration was calculated by using the CIBERSORT algorithm. We further identified genes linking periodontitis and immunity from the ImmPort and DisGeNet databases. In addition, some of them were selected to construct a diagnostic model via a logistic stepwise regression analysis. RESULTS AND CONCLUSIONS Two hundred sixty differentially expressed genes were identified and found to be involved in responses to bacterial and immune-related processes. Subsequently, immune cell infiltration analysis demonstrates significant differences in the abundance of most immune cells between periodontitis and healthy samples, especially in plasma cells. These results suggested that immunity doses play a non-negligible role in periodontitis. Twenty-one genes linking periodontitis and immunity were further identified. And nine hub genes of them were identified that may be key genes involved in the development of periodontitis. Gene ontology analyses showed that these genes are involved in response to molecules of bacterial origin, cell chemotaxis, and response to chemokines. In addition, three genes of them were selected to construct a diagnostic model. And its good diagnostic performance was demonstrated by the receiver operating characteristic curves, with an area under the curve of 0.9424 for the training dataset and 0.9244 for the external validation dataset.