This original study investigates the clinical synergy between computed tomography (CT)-derived bladder wall thickness (BWT) and urinary biomarkers, a combination that remains unexplored for predicting therapeutic outcomes. Interstitial cystitis/bladder pain syndrome (IC/BPS) is highly heterogeneous1. We hypothesize that a novel two-step predictive model, integrating morphological BWT classification with specific urinary biomarkers, can effectively sub-phenotype IC/BPS and predict treatment outcomes (Global Response Assessment, GRA). This study aims to shift the focus from static diagnosis to dynamic prognostic prediction, utilizing receiver operating characteristic (ROC) analysis to establish high-accuracy prognostic models for both Hunner's IC (HIC) and non-Hunner's IC (NHIC) patients.

This retrospective study analyzed 149 patients with cystoscopy-proven IC/BPS (123 NHIC and 26 HIC). Before treatment, all patients underwent pelvic CT scans to be categorized into two morphological subgroups: smooth and non-smooth (focal or diffuse thickening)2. Baseline urine samples were analyzed for 14 biomarkers: inflammatory cytokines and chemokines (IL-2, IL-6, IL-8, CXCL10, MCP-1, MIP-1β, RANTES, TNF-α, Eotaxin), neurotrophic factors (NGF, BDNF), oxidative stress markers (8-OHdG, 8-isoprostane), and prostaglandin E2 (PGE2). Treatment outcomes were evaluated using the GRA scale, comparing favorable responders (GRA > 1) versus poor responders (GRA ≤ 1) within BWT subgroups. ROC curves were generated to establish optimal predictive cut-off values and evaluate diagnostic performance, specifically within the pure NHIC non-smooth cohort.

Morphologically, smooth BWT was observed exclusively in NHIC patients (n=71/123, 100% of smooth group), whereas all HIC patients exhibited non-smooth BWT (n=26/26). Urinary biomarkers successfully predicted outcomes only when stratified by BWT. In the non-smooth group (n=63), favorable responders exhibited significantly higher baseline IL-8 (p=0.004), MCP-1 (p=0.017), and PGE2 (p=0.017). Conversely, in the smooth group, all biomarkers failed to differentiate treatment responses (all p > 0.3). In the pure NHIC non-smooth subgroup (excluding HIC, n=45), ROC analysis revealed that lower BDNF and higher PGE2 were significant predictors of success (GRA > 1). BDNF yielded an AUC of 0.683 (cut-off ≤ 0.530, sensitivity 76.5%, specificity 64.3%), while PGE2 had an AUC of 0.727 (cut-off ≥ 46.61, sensitivity 70.6%, specificity 71.4%). The combined model (BDNF+PGE2) demonstrated the highest predictive value with an AUC of 0.758 (cut-off ≥ -0.678, sensitivity 76.5%, specificity 75.0%).

BWT morphology serves as a critical first-step filter. In the pure NHIC cohort, the significant association of BDNF and PGE2 with outcomes reveals distinct underlying mechanisms. A low BDNF level implies a lower degree of central sensitization, while a high PGE2 level indicates active local bladder inflammation. Together, these markers define a patient window highly responsive to bladder-directed therapies. Conversely, the inability of biomarkers to predict outcomes in smooth BWT patients suggests symptoms driven by central or systemic mechanisms.

We propose a novel, non-invasive two-step predictive model utilizing CT-derived BWT and urinary biomarker analysis. By integrating ROC-validated thresholds, this model demonstrates that non-smooth BWT patients with high PGE2 and low BDNF exhibit excellent responses to bladder-directed treatments. This provides a robust, quantitative framework for guiding personalized, phenotype-directed therapy in clinical practice.

1.	Chen YC, Jiang YH, Jhang JF, Kuo HC. Using Urine Biomarkers to Differentiate Bladder Dysfunctions in Women with Sensory Bladder Disorders. Int J Mol Sci 2024;25:9359.
2.	Yu WR, Jiang YH, Jhang JF, Kuo HC. Focal or diffuse bladder wall thickness on bladder computed tomography indicates more severe bladder wall inflammation in patients with interstitial cystitis. World J Urol 2025;43:100.