Treatment response to intravesical therapies in non-Hunner interstitial cystitis/bladder pain syndrome (NHIC) is highly variable, and clinicians currently lack tools to estimate individual treatment benefit. This study aimed to develop and internally validate treatment-specific predictive models integrating clinical characteristics, bladder functional parameters, and urinary biomarkers to estimate response to intravesical onabotulinumtoxinA (BoNT-A) and platelet-rich plasma (PRP) therapy in female patients with NHIC.

Females with NHIC who underwent intravesical BoNT-A injection or PRP therapy were retrospectively analyzed. Multivariable logistic regression models were constructed to predict treatment response, defined as a Global Response Assessment score ≥2. Model discrimination was assessed using the area under the receiver operating characteristic (ROC) curve. Nomograms were generated, and internal validation was performed by comparing predicted probabilities with observed outcomes in the derivation cohort.

Although the BoNT-A model demonstrated strong discrimination (area under the ROC curve: 0.789), the PRP model’s discriminatory performance was superior (area under the ROC curve: 0.895) (Table 1). Distinct clinical features and biomarker patterns contributed to each treatment-specific model. Internal validation revealed higher observed response rates when the administered therapy matched the model-predicted higher probability of response, supporting the models’ internal concordance. (Figure 1).

In this study, we developed and internally validated treatment-specific predictive models integrating clinical characteristics, bladder functional parameters, and urinary biomarkers to estimate response to intravesical BoNT-A injection and PRP therapy in females with NHIC. Both models demonstrated acceptable discrimination, although distinct predictor profiles were identified for each treatment. Internal validation results showed concordance between model-predicted probabilities and observed outcomes. Higher response rates (GRA ≥ 2) and lower nonresponse rates (GRA < 2) were observed when the administered therapy matched the treatment associated with the higher predicted probability. Importantly, these results reflect internal model consistency rather than a prespecified clinical decision rule. Collectively, these findings support the relevance of treatment-specific prediction in a disorder characterized by marked clinical heterogeneity.

Treatment-specific predictive models integrating clinical characteristics, bladder functional parameters, and urinary biomarkers were developed and internally validated to estimate response to intravesical BoNT-A injection and PRP therapy in females with NHIC. Identification of distinct predictor profiles for each treatment underscores the biological and clinical heterogeneity of IC/BPS and highlights the limitations of uniform therapeutic strategies. By offering a structured, probability-based framework for response estimation, these models can support future biomarker-informed research in IC/BPS. Nevertheless, external validation in independent cohorts is essential for establishing generalizability and clinical utility before broader implementation.