Background: Chronic kidney disease (CKD) is associated with substantial morbidity and mortality in patients with bladder dysfunctions. In urodynamic study (UDS), only few arbitrary pressure/volume points are selected based on subjective annotations. We aimed to bridge the gap by our novel algorithm to analyze the bladder compliance, enhancing the understanding of the association between bladder function and CKD risk. By leveraging quantitative variants extracted from the compliance curves, we sought to improve risk assessment of CKD, providing a reliable, reproducible, and scalable method for evaluating bladder compliance.
Methods: UDS at a single instruction from 2013 to 2019 were used for algorithm development and validation. The electronic medical record was used to identify patient demographics, clinical characteristics including renal bladder ultrasound (RBUS) and the resulting urinary tract dilatation (UTD), vesicoureteral reflux (VUR), febrile urinary tract infection (fUTI) within 12 months in patients undergoing UDS. We defined the “compliance curve” by utilizing a proprietary algorithm to systematically create a noise-free curve from intravesical pressure (Figure 1). Additional measurements extracted from the compliance curve algorithm included curve tortuosity, arc length, algorithm-derived detrusor maximal pressure (PdetMax) and detrusor leak point pressure (DLPP), segmented area under the compliance curve (AUCC), and mean compliance curve pressure versus % volume. The last UDS prior to the CKD measurement was used to examine the association to CKD. The outcome was defined as stages of CKD were acquired based on age and eGFR following the operation of the UDS. Binary classifications were established as CKD stage 1 (eGFR >=90) versus CKD stages 2-5 (eGFR <90). Multivariate logistic regression model was fitted to assess the association between the CKD stage > 1 and clinical or UDS covariates, Results: 557 patients who had eGFR data available were included. The mean age at UDS was 10.2±6.8 years and 50% were female. CKD Stage >=2, which is equivalent to GFR under 90, was found in 100 patients. In the multivariate model (Table 1), higher CKD risk was significantly associated with higher algorithm-derived AUC over 45 cmH2O with normalized volume (OR=1.20[95CI=1.02-1.40], p=0.022), higher mean of first derivative in the initial third of compliance curve (OR=7.23[95CI=1.36-59.0], p=0.035), and less curvature of the compliance curve (OR=0.92[95CI=0.86-0.98], p=0.011) after adjusting for gender, VUR, age, fUTI within 12 months, UTD, and major GU anatomy (exstrophy, posterior urethral valve, prune belly). Conclusions: Our novel bladder compliance curve algorithm and derived quantitative parameters are significantly associated with the stages of CKD in children. This novel algorithm demonstrates promising performance to provide an objective, reliable, and scalable solution for bladder compliance evaluations.