Predictive tools and scoring systems for surgical intervention in antenatal hydronephrosis and pelviureteric junction obstruction: An ATLAS based on comprehensive review of literature.

Abstract

Antenatal hydronephrosis (ANH) is detected in up to 5% of pregnancies and is most commonly caused by pelviureteric junction obstruction (PUJO). While many cases resolve spontaneously, subset of patients require surgical intervention. Differentiating between these groups remains a clinical challenge, often leading to unnecessary investigations or delayed treatment. Numerous scoring systems and predictive tools have been developed to support risk stratification, yet none have achieved universal adoption. A comprehensive literature search of MEDLINE and Google Scholar was performed to identify scoring systems, predictive models, and tools designed to predict the need for surgical intervention in patients with ANH or confirmed PUJO. Search terms included variations of "PUJO," "prognosis," "predictor," and "surgery." Included studies described original tools or external validation of models using radiological parameters to stratify risk. Each tool was appraised for input parameters, derivation methodology, outcome definitions, external validation, and clinical applicability. Nine predictive tools were identified, all based on imaging data, with one incorporating machine learning (ML). Eight of nine tools aimed to predict the need for pyeloplasty. Four tools have undergone some form of external validation. Most tools used numerical scores, one applied a visual grading system, and another used a nonpercentage-based ML approach. While several demonstrated high predictive accuracy, limitations included retrospective design, small sample sizes, subjective imaging interpretation, and lack of consistent surgical outcome definitions. Despite growing interest and several promising models, no tool has yet been externally validated in large, diverse prospective cohorts. Further research is needed to develop clinically robust, generalizable tools for early risk stratification in ANH.