Abstract: PO0907
nPOD-Kidney, a New Tool for Investigating Diabetic Kidney Disease
Session Information
- Diabetic Kidney Disease: Basic Mechanisms
October 22, 2020 | Location: On-Demand
Abstract Time: 10:00 AM - 12:00 PM
Category: Diabetic Kidney Disease
- 601 Diabetic Kidney Disease: Basic
Authors
- Anquetil, Florence, Novo Nordisk Research Center Seattle, Inc., Seattle, Washington, United States
- Das, Vivek, Novo Nordisk Research Center Seattle, Inc., Seattle, Washington, United States
- Ward, Heather Hilary, Novo Nordisk Research Center Seattle, Inc., Seattle, Washington, United States
- Alpers, Charles E., University of Washington School of Medicine, Seattle, Washington, United States
- Zeng, Xu, University of Florida Health Pathology Laboratories, University of Florida Health, Gainesville, Florida, United States
- Wesley, Johnna D., Novo Nordisk Research Center Seattle, Inc., Seattle, Washington, United States
- Karihaloo, Anil K., Novo Nordisk Research Center Seattle, Inc., Seattle, Washington, United States
Background
Diabetic kidney disease (DKD) is a common complication of diabetes, yet it remains poorly understood. The network for Pancreatic Organ donors with Diabetes - Kidney (nPOD-K) project was initiated to assess the feasibility of evaluating human kidneys from organ donors with long-standing diabetes (>8 years), with the long-term goal of improving our understanding of DKD pathogenesis.
Methods
Formalin-fixed paraffin-embedded sections from nPOD-K were stained for specific renal cell and disease markers by multiplex immunofluorescence, followed by periodic acid-Schiff (PAS) staining. Whole sections were imaged using an Axioscan Z1 scanner (20X objective) and quantitative image analyses were performed using Visiopharm software
Results
Tissue integrity and histological stage were independently assessed by two renal pathologists. The majority of cases presented a moderate or severe diagnosis, and 20% of the cohort displayed no overt sign of kidney disease despite long-standing diabetes. Algorithms for automatic segmentation of kidney compartments (e.g. glomeruli, tubules) in the PAS layer were then developed using deep convolution neural network DeepLabV3+. We achieved a DICE coefficient of 0.95 for glomeruli and 0.89 for tubules. Quantification of renal markers was performed using machine learning classification methods. In accordance with published studies, our quantitation demonstrated loss of podocyte marker WT1, endothelial marker CD31, and tubular marker Lotus tetragonolobus lectin correlating with the progression of DKD, concomitantly with increased tubular atrophy and expression of fibrotic markers.
Conclusion
In conclusion, kidneys obtained from organ donors are viable and display all expected features of human DKD at the level of light microscopy. This cohort provides a unique opportunity to better understand DKD pathophysiology through analysis of large, CKD stage-specific regions. Similar to the results from the nPOD pancreas cohort, all histological stages of disease can be detected in affected kidneys, providing a pseudo-timeline of the evolution of DKD and supporting the potential to identify novel therapeutic targets.
Funding
- Commercial Support –