Exploration of the potential pathogenesis of interstitial cystitis and construction of a disease identification model based on the immune microenvironment

There is no known cause of interstitial cystitis (IC), a condition characterized by chronic pelvic floor pain and lower urinary tract symptoms. It is difficult to diagnose IC; diagnosis can only be made by conducting multiple tests and ruling out other similar conditions. In this study, we aimed to build a model for IC diagnosis and provide a new method for IC identification. Gene expression was analyzed in IC and normal bladder tissues to understand the immune infiltration characteristics of IC. Subsequently, 106 differentially expressed immune-related genes (DEIRGs) were identified. We build a protein-protein interaction network and performed enrichment analyses to unveil the functional associations of the identified DEIRGs. Using machine learning, we screened key disease characteristic genes ( S100A8 , S100A12 , FABP5 ) and constructed an effective random forest model to identify IC patients. Immunohistochemical analysis and validation in the extra test cohort are used to prove the effectiveness of the model. Compared with normal tissues, IC shows increased infiltration of immune cells, suggesting that IC is characterized by a high level of immune activation. The strong correlation between S100A8 and S100A12 and neutrophils reiterated the importance of neutrophils in IC. The constructed random forest model based on the identified three characteristic genes S100A8 , S100A12 , and FABP5 is effective in identifying IC (AUC = 1). As expected, in bladder tissue, the levels of S100A8 and S100A12 were significantly increase, while that of FABP5 was significantly decreased in the IC. The extra test cohort also proved the effectiveness of the constructed model (AUC = 0.725).