Abstract
BACKGROUND: Bilateral Wilms tumor (BWT) is a relatively rare malignant renal tumor in children. Nephron-sparing surgery (NSS) is the preferred surgical approach for treating BWT, but lacks uniform surgical indications worldwide. This study aimed to summarize the clinical and imaging features of BWT children, establish a radiomics nomogram, and predict the feasibility of NSS for improving outcomes. METHODS: A 12-year retrospective single-center review was conducted on clinical data and preoperative imaging features of BWT patients. The tumor kidneys were divided into NSS and non-NSS groups. Logistic regression analysis was performed to identify independent predictors and develop a prediction model of the feasibility of NSS in BWT patients. A radiomics nomogram was constructed and internally validated by the parametric bootstrapping method. RESULTS: A total of 58 BWT patients (115 renal units) were included in this study. After evaluations based on preoperative imaging and clinical data, 94 renal units underwent NSS with negative resection margins and were included in the NSS group, whereas 16 renal units with positive resection margins, macroscopic residual, or total nephrectomies were included in the non-NSS group. Tumor size [odds ratio (OR): 0.540, 95% confidence interval (CI): 0.308-0.945], relationship with the collecting system (OR: 0.013, 95% CI: 0.0004-0.370), and remaining renal parenchyma (RRP) proportion (OR: 71.23, 95% CI: 1.632-3108.8) were identified as independent predictors for NSS. A nomogram was constructed based on these factors, which demonstrated great consistency between the predicted and observed feasibility of NSS. The model presented with good discriminative ability [area under the curve (AUC), 0.982]. The decision curve analysis (DCA) revealed the clinical usefulness of the model. CONCLUSIONS: This study analyzed the clinical and preoperative imaging data of BWT patients and identified three independent predictors for the feasibility of NSS, including tumor size, relationship with the collecting system, and residual renal parenchyma proportion. The radiomics nomogram established in this study can provide individualized predictions to assist clinicians in making better decisions and improving patient outcomes.