Development and Internal Validation of a Nomogram for Predicting Renal Function after Partial Nephrectomy.
Academic Article
Overview
abstract
Loss of renal function can be a clinically impactful event after partial nephrectomy (PN). We aimed to create a model to predict loss of renal function in patients undergoing PN. Data for 1897 consecutive patients who underwent PN with warm ischemia between 2008 and 2017 were extracted from our institutional database. Loss of renal function was defined as upstaging of chronic kidney disease in terms of the estimated glomerular filtration rate (eGFR) at 3 mo after PN. A nomogram was built based on a multivariable model comprising age, sex, body mass index, baseline eGFR, RENAL score, and ischemia time. Interval validation and calibration were performed using data from 676 patients for whom complete data were available. Receiver operator characteristic (ROC) curves with 1000 bootstrap replications were plotted, as well as the observed incidence versus the nomogram-predicted probability. We also applied the extreme training versus test procedure known as leave-one-out cross-validation. After internal validation, the area under the ROC curve was 76%. The model demonstrated excellent calibration. At an upstaging cutoff of 27% probability, upstaging was predicted with a positive predictive value of 86%. PATIENT SUMMARY: In this report, we created a model to predict postoperative loss of renal function after partial nephrectomy for renal tumors. Inputting baseline characteristics and ischemia time into our model allows early identification of patients at higher risk of renal function decline after partial nephrectomy with good predictive power.
