Diagnostic Model Development for IC/BPS and Its Subtypes Using Clinical Indicators, Urinary Biomarkers, and Single‐Cell Transcriptomic Analysis

ABSTRACT

Interstitial cystitis/bladder pain syndrome (IC/BPS) is a chronic, heterogeneous, and often debilitating condition. This condition is often misdiagnosed, as it overlaps significantly with other urinary infections, necessitating the development of new markers and targets to improve diagnosis and therapy. Tumor necrosis factor‐α (TNF‐α) participates in IC/BPS inflammation, but its role remains understudied. This study aimed to develop a diagnostic model for IC/BPS and its subtypes to uncover underlying mechanisms, new biomarkers, and TNF‐α‐targeted treatments. Patients with non‐Hunner IC (NHIC, n  = 196), Hunner IC (HIC, n  = 56), and healthy controls (HC, n  = 230) were enrolled in the study. A diagnostic model was developed based on key features identified by SHapley Additive exPlanations (SHAP) analysis. Molecular analysis using Gene Expression Omnibus (GEO) data and single‐cell RNA sequencing (scRNA‐seq) showed cellular heterogeneity and inflammatory networks. In silico TNF‐α knockout was performed using scTenifoldKnk to investigate the function of TNF‐α. Both NHIC and HIC subtypes present with severe clinical symptoms and dysregulated urinary biomarker profiles, with the HIC subtype showing significantly more severe symptoms and markedly increased levels of specific inflammatory mediators. A total of 127 machine learning (ML) models were constructed, of which most reached an AUC of 1.000 in training and external validation cohorts. Lasso+plsRglm was identified as the optimal diagnostic model. Core diagnostic features included nocturia frequency (NUF), visual analogue scale (VAS), functional bladder wall thickness (BWT), and TNF‐α. Transcriptomics identified inflammation‐related patterns and pathways between IC/BPS subtypes. scRNA‐seq identified 12 cell types and an elevated inflammatory state. We discovered a TNF‐α‐driven inflammatory network influencing intergroup heterogeneity of IC/BPS, identified candidate diagnostic biomarkers to differentiate IC/BPS from HC, and highlighted possible therapeutic targets, thereby laying the groundwork for future clinical translation research.