External Validation of a Postoperative Renal Function Predictive Model Following Partial Nephrectomy

A predictive model from the Veterans Affairs National Health System estimates glomerular filtration rate (GFR) 1 year after partial nephrectomy (PN). The model demonstrated a coefficient of determination (R²) of 0.66, with 82% of patients having a postoperative estimated GFR within 30% of the predicted GFR (P30). The purpose of this study was to assess its performance in patients enrolled in the renal hypothermia trial.

The model was externally validated using data from the renal hypothermia randomized clinical trial where GFR was measured using ^99mTc-DTPA (technetium-99m-Tc-diethylenetriaminepenta-acetic acid) plasma clearance pre-PN and 1 year post PN. Model performance was evaluated using R², calibration slope, calibration plot, and precision (P10: within 10%; P20: within 20%; and P30: within 30%).

The trial cohort included 175 patients with pre-PN and post-PN complete data. The R² was 0.67. The model was reasonably precise with 73 patients within 10% of predicted GFR (P10 = 41.7%), 129 within 20% of predicted GFR (P20 = 73.7%), and 156 within 30% of predicted GFR (P30 = 89.1%). The calibration slope was 1.04, with good calibration across a wide range of baseline renal function. Limitations include modest sample size, predominantly Caucasian patients, and restriction to open PN, limiting generalizability.

A recently developed model to predict postoperative GFR performed well when applied to clinical trial patients who had GFR measured using ^99mTc-DTPA plasma clearance. This model can be used as a component of informed decision-making when counseling patients on expected outcomes following partial nephrectomy.