ASN's Mission

To create a world without kidney diseases, the ASN Alliance for Kidney Health elevates care by educating and informing, driving breakthroughs and innovation, and advocating for policies that create transformative changes in kidney medicine throughout the world.

learn more

Contact ASN

1401 H St, NW, Ste 900, Washington, DC 20005

email@asn-online.org

202-640-4660

The Latest on X

Kidney Week

Abstract: FR-OR86

Computational Characterization of Lymphocyte Topology in FSGS/Minimal Change Disease

Session Information

Category: Pathology and Lab Medicine

  • 1800 Pathology and Lab Medicine

Authors

  • Li, Xiang, Duke University, Durham, North Carolina, United States
  • Shah, Manav P., Georgia Institute of Technology, Atlanta, Georgia, United States
  • Sotolongo, Gina, Duke University, Durham, North Carolina, United States
  • Liu, Qian, The Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, United States
  • Zhou, Jin, Duke University, Durham, North Carolina, United States
  • Hodgin, Jeffrey B., University of Michigan, Ann Arbor, Michigan, United States
  • Mariani, Laura H., University of Michigan, Ann Arbor, Michigan, United States
  • Holzman, Lawrence B., University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, United States
  • Janowczyk, Andrew, Emory University, Atlanta, Georgia, United States
  • Zee, Jarcy, University of Pennsylvania Perelman School of Medicine, Philadelphia, Pennsylvania, United States
  • Barisoni, Laura, Duke University, Durham, North Carolina, United States
  • Lafata, Kyle Jon, Duke University, Durham, North Carolina, United States
Background

The semi-quantitative visual scoring of inflammation does not capture the distribution of inflammatory cells in the kidney tissue. This study computationally quantified the topology of inflammation and evaluates their clinical relevance.

Methods

H&E-stained whole slide images (WSIs) of renal biopsies from N=190 NEPTUNE (104 FSGS and 86 MCD) and N=141 CureGN (75 FSGS and 66 MCD) participants were used. Deep learning models were developed to segment renal cortex and lymphocytes. Graph modeling was applied, where nodes were defined as lymphocytes and edges as the spatial connections between lymphocytes. We developed a novel graph-based clustering algorithm to capture dense vs. sparse lymphocytic regions. 30 pathomic features were extracted to capture cell density, connectivity, clustering, and centrality. The prognostic value of the features was evaluated by modeling their association with disease progression (time from biopsy to a 40% decline in eGFR or kidney replacement therapy) in the NEPTUNE dataset. To identify the most informative features, lasso regularization was applied to a Cox model. Next, L2-regularized Cox models including clinical features with and without the selected features were constructed. The selected features were then re-fitted in the CureGN dataset for validation.

Results

The selected pathomic features include topology features from both sparsely (n=8) and densely (n=11) inflamed regions. The addition of pathomic features to clinical feature increased the Concordance index from 0.71 to 0.78 in NEPTUNE and from 0.76 to 0.79 in CureGN.

Conclusion

We developed a computational approach to quantify topological characteristics of lymphocytic inflammation on WSIs. These digital signatures have potential as biomarkers of disease progression in FSGS/MCD and increase our ability to predict clinical outcome above and beyond current approaches.

Funding

  • NIDDK Support