Evaluation of algorithms for predicting new baseline renal function after complex partial nephrectomy for multifocal renal tumors.

INTRODUCTION

Partial nephrectomy (PN) for localized renal cancer is associated with favorable survival outcomes due to preservation of renal function. Approaches for predicting new-baseline-GFR (NBGFR) after PN have been proposed, including multivariable algorithms and a formula presuming that 90% of the renal parenchyma is preserved during PN. However, these models were developed from cohorts with sporadic single-tumor renal cancer. We evaluate these predictive models in patients undergoing PN for multifocal disease. Accurate prediction of NBGFR has major implications in these patients, as they may require multiple PN for metachronous kidney cancer.

METHODS

Patients who underwent PN for removal of ≥2 tumors from 2006 to 2024 and had available measurements of preop/postoperative GFR were considered. The final cohort included 453 patients, of whom 184 underwent multiple PN. Four predictive models were evaluated. Model performance was assessed using correlation-coefficient(r) and mean-squared-error (MSE).

RESULTS

Approximately 30% of patients had sporadic, multifocal renal tumors, and 70% had hereditary renal cancer. The 90% formula had a higher r and lower MSE values compared to the multivariable models, and also showed improved accuracy, bias, and precision. In patients undergoing multiple PN, 85% to 91% of the GFR was preserved in each PN.

CONCLUSIONS

Models for predicting NBGFR after PN showed strong performance in our cohort with multifocal renal tumors. The conceptually simple formula presuming 90% renal parenchymal preservation, and thus 90% GFR preservation, most accurately predicted NBGFR. The %GFR preserved with each PN remained stable across sequential PN. These findings suggest that preservation of renal parenchyma is a relevant determinant of NBGFR after PN for multifocal tumors.