Development of an interstitial cystitis risk score for bladder permeability

Abstract

Background: Interstitial cystitis/bladder pain syndrome (IC) is a multifactorial syndrome of severe pelvic and genitalia pain and compromised urinary function; a subset of IC patients present with Hunner's lesions or ulcers on their bladder walls (UIC). UIC is diagnosed by cystoscopy, which may be quite painful. The objective of this study was to determine if a calculated Bladder Permeability Defect Risk Score (BP-RS) based on non-invasive urinary cytokines could discriminate UIC patients from controls and IC patients without Hunner's ulcers.

Methods: A national crowdsourcing effort targeted IC patients and age-matched controls to provide urine samples. Urinary cytokine levels for GRO, IL-6, and IL-8 were determined using a Luminex assay.

Results: We collected 448 urine samples from 46 states consisting of 153 IC patients (147 female, 6 male), of which 54 UIC patients (50 females, 4 male), 159 female controls, and 136 male controls. A defined BP-RS was calculated to classify UIC, or a bladder permeability defect etiology, with 89% validity.

Conclusions: The BP-RS Score quantifies UIC risk, indicative of a bladder permeability defect etiology in a subset of IC patients. The Bladder Permeability Defect Risk Score is the first validated urine biomarker assay for interstitial cystitis/bladder pain syndrome.

Fig 1. Consolidated standards of reporting trials flow diagram for study numbers.

UIC = IC with Hunner’s lesions; NUIC = IC without Hunner’s lesions.

Fig 2. Urinary cytokine levels by group.

A) Mean and SEM values for urine cytokines GRO, IL-6, and IL-8. B-G) Urinary levels of cytokines GRO, IL-6, and IL-8 in groups Control and NUIC compared to UIC (B-D) or control compared to NUIC compared to UIC (E-F). Error bars are SEM. Statistical differences were determined using Mann-Whitney test (B-G) or Kruskal-Wallis ANOVA (E-G). * indicates p<0.05, ** indicates p<0.01, *** indicates p<0.001, **** indicates p<0.0001.

Fig 3. Average BP-RS probabilities in IP4IC training set and of P3 validation set.

The BP-RS is a score that gives the probability that a patient has UIC. Each of the validation data points were fed into the trained Random Forest Classifier (RFC), and the BP-RS (probability of UIC) was calculated and plotted for each patient. The bars on the boxes describe the minimum and maximum points, and the horizontal lines of the box show the 25%, 50% (median), and 75% quartiles. Points outside of the bars are outliers. The individual probabilities are also plotted (black points). Because we used a binary classifier, all of the points above 0.5 (dotted line) are predicted to have UIC, and all points below are classified as control or NUIC. In the validation set, two patients from each of the control and NUIC groups were incorrectly classified as having UIC, and two points from the UIC group were incorrectly classified as not having UIC.

Fig 4. Receiver operating characteristic (ROC) curves.

ROC curves for the IP4IC (training set) and P3 (validation set) using the BP-RS. The AUC for the IP4IC dataset = 0.971 and P3 dataset = 0.919.