Biopsy-based Transcriptomics in Banff 2022 Antibody Mediated Rejection Categories

Biopsy-based transcriptomics (BBT) supports antibody-mediated rejection (AMR) diagnostics, but Banff 2022 probable AMR and donor-specific antibodies (DSA)-/C4d-microvascular inflammation (MVI) remain unvalidated by BBT.

We analyzed 562 kidney allograft biopsies from Prague and Zurich by histology and Molecular Microscope Diagnostic System (MMDx). Biopsies were classified into the following: (i) active AMR-spectrum lesions without transplant glomerulopathy (cg) (active AMR, probable AMR, MVI, DSA-/C4-); (ii) combined active-chronic lesions with cg (chronic-active AMR, probable AMR with cg, MVI, DSA-/C4d- with cg); (iii) chronic AMR, defined by cg without active lesions; and (iv) controls without rejection.

Molecular AMR was detected in 24% of probable AMR and 23% of probable AMR with cg biopsies, 43% of MVI, DSA-/C4d-, 51% of active AMR, 56% of MVI, DSA-/C4d- with cg, and 63% of chronic-active AMR, but was rare in chronic AMR (6%). AMR probability classifier (AMR_prob) scores were higher in probable AMR than in controls and similar between chronic AMR and controls. AMR_prob scores were comparable among active AMR, chronic-active AMR, and MVI, DSA-/C4d- groups, all of which exhibited elevated T cell-mediated rejection (TCMR) _prob scores compared with probable AMR and controls.Least absolute shrinkage and selection operator-penalized Cox regression predicting 3-years graft survival trained in the Zurich cohort showed similar performance of molecular (cg_prob, ct_prob, t_prob, acute kidney injury [AKI]score) and histologic model (Banff cg, interstitial inflammation [ci], tubular atrophy [ct], and tubulitis [t]) (concordance 0.84 vs. 0.82), with the external validation in the Prague cohort (0.78 vs. 0.75).

BBT primarily reflects MVI rather than DSA and is particularly informative in probable AMR and MVI-positive biopsies.