Estimation of nephron number in living humans by combining unenhanced computed tomography with biopsy-based stereology

Abstract

Methods for estimating nephron number in a clinical setting may be useful for predicting renal outcomes. This study aimed to establish such a method using unenhanced computed tomography (CT) and biopsy-based stereology. Patients or living kidney donors simultaneously subjected to enhanced and unenhanced CT examinations were randomly assigned to development and validation groups. The enhanced CT-measured arterial phase and the venous phase images of kidneys were regarded as the true values for cortical volume and parenchymal volume, respectively. Linear multiple regression analysis was used to create models for estimating cortical volume using explanatory variables including unenhanced CT-measured parenchymal volume. Nephron number was determined as the product of cortical volume and the glomerular density in biopsies of donors. Five equations for estimating cortical volume were created and verified. In donors, estimated nephron number by unenhanced CT was consistent with that by enhanced CT, with minimal errors in all models (636–655\u2009±\u2009210–219 vs. 648\u2009±\u2009224\u2009×\u2009103/kidney). Clinical characteristics combined with parenchymal volume did not improve the equation over parenchymal volume alone. These results support the feasibility of estimating nephron number by a combination of unenhanced CT and biopsy-based stereology, with a possible application for renal disease patients who are often not suitable for contrast media.