Abstract: TH-OR070
Digital Pathology and Computer-Aided Quantitative Analysis for Lupus Nephritis
Session Information
- Harnessing Molecular, Machine-Learning, and Genomic Innovations in Pathology
October 25, 2018 | Location: 24A, San Diego Convention Center
Abstract Time: 06:18 PM - 06:30 PM
Category: Pathology and Lab Medicine
- 1502 Pathology and Lab Medicine: Clinical
Authors
- Ragothaman, Srikanth, University of Michigan, Ann Arbor, Michigan, United States
- Chang, Anthony, UChicago Medicine, Chicago, Illinois, United States
- Kahlenberg, J. Michelle, University of Michigan, Ann Arbor, Michigan, United States
- Wylie, Stephanie, University of Michigan, Ann Arbor, Michigan, United States
- Hodgin, Jeffrey B., University of Michigan, Ann Arbor, Michigan, United States
Background
The kidney biopsy has allowed a better understanding of the pathogenesis of renal injury in lupus nephritis (LN) and is routinely used for patients with active LN or previously untreated disease. Currently, the ISN/RPS classification system is used to identify subgroups with different prognoses and response to treatment. However, the ISN/RPS classification suffers from poor reproducibility, raising doubts about its validity and clinical application. Thus novel approaches are required to obtain continuous, quantitative data to improve accuracy, reproducibility, and prognostic utility. We have assembled a novel computer-aided digital pathology image analysis pipeline to improve the utility of the kidney biopsy in LN.
Methods
To facilitate our analyses, we have created the University of Michigan Lupus Nephritis Digital Pathology Image Repository (DPIR). Currently it contains 200+ cases with access to clinical data over many years. Slides are scanned to whole-slide images (WSI) at 40x with Leica Biosystems AT2 scanners. We employ QuPath, a cross-platform open source software for digital pathology and WSI image analysis, for quality control, annotation, and as a framework to run in-house algorithms on WSI. Algorithms to quantitate morphologic features of interest are designed using the image processing package FIJI, a distribution of ImageJ.
Results
Our image analysis pipeline allows supervised WSI annotation (glomeruli, vessels, etc.) and quantitation of glomerular size, glomerular cellularity, mesangial index, interstitial fibrosis and fractional interstitial area, and tubulointerstitial cellularity using PAS and trichrome stained slides. Complete analysis can be performed within one day. Preliminary data comparing glomerular cellularity to ISN/RPS classification groups reveals that cellularity varies widely within each classification group, indicating novel information beyond classification groups are captured.
Conclusion
We have generated a computer-aided, quantitative image analysis pipeline for morphologic features in a lupus nephritis biopsy. The pipeline has a short enough turnaround time suitable for clinical application and generation of an adjunct report wherever cases can be scanned to WSIs. Future studies will seek to improve turnaround time and determine which analyses best predict patient outcome and response to therapy.
Funding
- NIDDK Support