Simple visual review of pre- to post-operative renal ultrasound images predicts pyeloplasty success equally as well as geometric measurements: A blinded comparison with a gold standard

Abstract

MAG3 diuretic renal scan remains the gold standard for determination of improvement in renal drainage following pyeloplasty for ureteropelvic junction obstruction. We hypothesized that (i) a change in geometric measurements between pre-operative and post-operative renal ultrasound (RUS) images and (ii) blinded simple visual review of images both would predict pyeloplasty success. To determine if simple visual review and/or novel geometric measurement of renal ultrasounds can detect pyeloplasty failure. This study was a retrospective, blinded comparison with a gold standard. Included were children aged ≤18 years undergoing pyeloplasty at our institution from 2009 to 2015. For each kidney, representative pre-operative and post-operative RUS images were chosen. Our standard for pyeloplasty success was improved drainage curve on MAG3 and lack of additional surgery. Measurements for collecting system circularity, roundness, and renal parenchymal to collecting system area ratio (RPCSR) were obtained by three raters (Figure), who were blinded to the outcome of the pyeloplasty. Changes in geometric measurements were analyzed as a diagnostic test for MAG3-defined pyeloplasty success using ROC curve analysis. In addition, six reviewers blinded to pyeloplasty success reviewed pre-operative and post-operative images visually for improved hydronephrosis and categorized pyeloplasty as success or failure based on simple visual review of RUS. Fifty-three repaired renal units were identified (50 children). There were five pyeloplasty failures, four of which underwent revision or nephrectomy. While all geometric measurements could discriminate pyeloplasty failure and success, the geometric measurements that discriminated best between pyeloplasty failure and success were change in collecting system roundness and change in RPCSR. Consensus opinion among six blinded reviewers using simple visual review had a sensitivity of 94% and PPV of 100% with respect to identifying pyeloplasty success (AUC 0.97 (95% CI 0.93-1.0)). This was not significantly different from AUC for change in roundness (p=0.09) or change in RPCSR (p=0.1). Change in collecting system roundness and change in RPCSR were the most accurate geometric measurements in predicting pyeloplasty success. Simple visual review of ultrasound images for pyeloplasty success performed as well or better than geometric measurements. However, geometric measurements remain useful as a research tool or to communicate findings between clinicians. Complex geometric measurements of hydronephrosis or post-operative MAG3 scans are not needed if hydronephrosis is visually significantly improved, as simple visual review is highly sensitive for detecting pyeloplasty failure.