Spatially Resolved Banff Tubulitis and Glomerulitis Scoring in Kidney Allograft Biopsies via Artificial Intelligent...

Area of Science:

• Nephrology
• Transplant Immunology
• Computational Pathology

Background:

• Tubulitis and glomerulitis are key histological markers for T cell-mediated rejection (TCMR) and antibody-mediated rejection (AMR) in kidney allografts, respectively.
• Current Banff criteria quantification faces interobserver variability, and bulk transcriptomics lack spatial resolution.
• Spatial transcriptomics offers a potential solution for high-resolution analysis of rejection mechanisms.

Purpose of the Study:

• To develop and validate an AI-driven platform (FUSION) for integrated spatial transcriptomics and histology analysis of kidney allograft biopsies.
• To generate spatially informed Banff t- and g-scores for improved diagnosis of TCMR and AMR.
• To assess the concordance of AI-derived scores with pathologist assessments and reduce interobserver variability.

Main Methods:

• Applied the FUSION platform to 8 kidney allograft biopsy cases across acute TCMR, active AMR, chronic active AMR, and control conditions.
• Utilized spatial transcriptomics (10x Genomics Visium v2) and high-resolution whole-slide histology with AI-based segmentation.
• Calculated transcriptomics-derived immune cell proportions within segmented tubular and glomerular regions to generate spatial Banff t- and g-scores.

Main Results:

• AI-based segmentation and spatial transcriptomics successfully generated spatially informed t- and g-scores.
• Derived t-scores showed full concordance with pathologist scores in acute TCMR cases.
• g-scores demonstrated partial concordance in AMR cases, with discrepancies linked to low immune signal near classification boundaries.

Conclusions:

• The AI-driven FUSION platform enables the generation of spatially informed Banff scores, aligning with diagnostic criteria.
• This approach demonstrates feasibility for enhancing diagnostic precision and reducing interobserver variability in kidney allograft rejection assessment.
• The validated spatial transcriptomics-augmented scoring system holds potential for clinical adoption in transplant pathology.