Abstract: TH-PO0982
Computationally Quantified Lymphocyte Cluster Architecture in Protocol Biopsies Predicts eGFR Decline Independent of Conventional Banff Scoring
Session Information
- Transplantation: Basic - Immune Biology, Tissue Injury, and Emerging Technologies
October 22, 2026 | Location: Exhibit Hall A, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Transplantation
- 2001 Transplantation: Basic
Authors
- Tokuyama, Naoto, Emory University, Atlanta, Georgia, United States
- Krishnamoorthy, Anusha, Emory University, Atlanta, Georgia, United States
- Abe, Yohei, Emory University, Atlanta, Georgia, United States
- Fan, Fan, Emory University, Atlanta, Georgia, United States
- Subramanian, Ajay Rajaraman, Emory University, Atlanta, Georgia, United States
- Dhamdhere, Rohan Narendra, Emory University, Atlanta, Georgia, United States
- Pathak, Tilak B., Emory University, Atlanta, Georgia, United States
- Janowczyk, Andrew, Emory University, Atlanta, Georgia, United States
- Farris, Alton Brad, Emory University, Atlanta, Georgia, United States
- Taoka, Rikiya, Kagawa Daigaku, Takamatsu, Kagawa Prefecture, Japan
- Madabhushi, Anant, Emory University, Atlanta, Georgia, United States
Group or Team Name
- Anant Madabhushi lab
Background
Immune cell infiltration in kidney transplant biopsies is measured using the Banff “i” score, which relies on density-based scoring but does not capture potentially informative patterns of spatial immune cell clustering. We applied computational pathology to quantify spatial immune cell clustering from 3-month H&E-stained protocol biopsies and evaluated association with post-transplant eGFR trajectories.
Methods
We analyzed 149 living-donor kidney transplant recipients at Kagawa University. Patients were stratified into Stable (n=108), Moderate (n=29), and Rapid decline (n=12) groups based on eGFR slope thresholds defined by linear regression of longitudinal data (Stable: ≥1 mL/min/1.73 m2/year; Moderate: -5 to 1; Rapid: <-5). Four lymphocyte clustering features (mean and max cluster size, clustered fraction, and cluster size heterogeneity) were extracted from H&E-stained biopsies and combined into a composite Lymphocyte Cluster Score (LCS), dichotomized at cohort median. Associations of LCS with eGFR slope were assessed using multivariable linear regression adjusting for clinical variables and graft survival by Kaplan-Meier analysis.
Results
High LCS was associated with steeper eGFR slope (p<0.001). In multivariable linear regression, LCS remained independently associated with eGFR slope after adjustment, whereas Banff “i” score was not. High LCS was associated with inferior overall graft survival, but not with death-censored graft survival, suggesting that the association may be mediated by patient mortality rather than primary graft failure. All four clustering features showed significant but modest inverse correlations with eGFR slope (ρ=−0.31 to −0.34, FDR<0.005).
Conclusion
Computationally quantified immune cell clustering provides prognostic information complementary to conventional pathologic assessment and warrants prospective evaluation as a risk stratification tool using routine H&E slides.
Acknowledgment
N.T. is supported by the Winship Postdoctoral Scholar Award from the Winship Cancer Institute of Emory University.