Abstract: TH-PO970
A Multivariate Model to Predict Post-Donation Kidney Function and Outcomes in Living Kidney Donors
Session Information
- Live Donor Outcomes and Kidney Transplantation in Pediatric and Ethnic/Racial Groups
November 02, 2017 | Location: Hall H, Morial Convention Center
Abstract Time: 10:00 AM - 10:00 AM
Category: Transplantation
- 1702 Transplantation: Clinical and Translational
Authors
- Limou, Sophie, Center for Research in Transplantation and Immunology, Nantes, France
- Jacquemont, Lola, CHU de Nantes, Nantes, France
- Gardan, Edouard, CHU de Nantes, Nantes, France
- Nusinovici, Simon, CHU de Nantes, Nantes, France
- Hanf, Matthieu, CHU de Nantes, Nantes, France
- Hourmant, Maryvonne, CHU de Nantes, Nantes, France
Background
Predicting kidney function after donation is a major challenge in living kidney donors. The aim of this study was to assess a wide range of demographics and clinical variables as non-invasive preoperative markers of post-donation kidney function.
Methods
110 French living kidney donors who had a 51Cr-EDTA scintigraphy and a measured glomerular filtration rate (mGFR) pre- (D0) and one-year post-donation (Y1) were included. Over 15 characteristics were collected for each subject before and after nephrectomy (e.g. sex, age, hypertension, creatinine and lipids levels). Kidney volume was quantified using three methods: total parenchymal three-dimensional renal volume (3DRV), total parenchymal renal volume contouring (RVCt), and renal cortical volume (RCoV). We tested each variable for association with Y1 mGFR using univariate and multivariate regression models. Finally, we produced receiver operating characteristic (ROC) curves to assess the performance of our model in discriminating chronic kidney disease (mGFR<60mL/min) at Y1.
Results
The mean mGFR was 105.2±17.7 mL/min at D0 and 68.1±12.8 mL/min at Y1. Total parenchymal volume measurements exhibited a high correlation with RCoV (R2=0.92 for 3DRV and 0.84 for RVCt). In univariate models, the correlation between kidney volume and mGFR was the highest for the RCoV measures with R2=0.44 (P=2x10-6) at D0 and R2=0.59 (P=3x10-11) at Y1. Y1 mGFR was also strongly associated with age (R2=-0.62, P=7x10-13) and D0 mGFR (R2=0.68, P=4x10-16). Using stepwise regressions, we developed a model integrating 5 non-invasive preoperative markers (D0 mGFR, age, RCoV, triglycerides level and birth weight) predicting Y1 mGFR with a R2=0.68. Finally, the ROC curve analysis showed that this multivariate model could reliably predict chronic kidney disease at 1-year post donation (area under the curve [AUC]=0.92).
Conclusion
The integration of non-invasive preoperative characteristics in one statistical model can accurately predict post-donation kidney function and outcomes in French living kidney donors. These results warrant validation in an independent population. Our report therefore opens the way for developing a predictive risk score that would be easily implemented in clinics.