A preoperative nomogram predicting the pseudocapsule status in localized renal cell carcinoma

Abstract

Background: Tumor enucleation (TE) surgery for localized renal cell carcinoma (RCC) relies on a complete peritumoral pseudocapsule (PC). Study objective was to develop a preoperative model to predict PC status. Methods: The prediction model was developed in a cohort that consisted of 170 patients with localized RCC, and data was gathered from 2010 to 2015. Multivariable logistic regression analysis and R were used to generate this prediction model. The statistical performance was assessed with respect to the calibration, discrimination, and clinical usefulness. Results: The prediction model incorporated the systemic inflammatory markers [neutrophil-lymphocyte ratio (NLR); albumin-globulin ratio (AGR)], CT imaging features (tumor size and necrosis), and clinical risk factors (BMI). The model showed good discrimination, with a C-index of 0.85 (0.78-0.91), and good calibration (P=0.60). The sensitivity and specificity were 62% and 94% respectively. Decision curves and clinical impact curve demonstrated that the current model was clinically useful. Conclusions: We constructed a model that incorporated both the systematic inflammatory markers and clinical risk factors. It can be conveniently used to preoperatively predict the individualized risk of PC invasion and identify the best candidates to receive TE surgery.