Premium
A Probabilistic Approach to Histologic Diagnosis of Antibody‐Mediated Rejection in Kidney Transplant Biopsies
Author(s) -
Halloran P. F.,
Famulski K. S.,
Chang J.
Publication year - 2017
Publication title -
american journal of transplantation
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.89
H-Index - 188
eISSN - 1600-6143
pISSN - 1600-6135
DOI - 10.1111/ajt.13934
Subject(s) - medicine , lesion , radiology , pathology , kidney transplantation , cardiology , kidney
Histologic diagnosis of antibody‐mediated rejection ( ABMR ) in kidney transplant biopsies uses lesion score cutoffs such as 0 versus >0 rather than actual scores and requires donor‐specific antibody ( DSA ); however, cutoffs lose information, and DSA is not always reliable. Using microarray‐derived molecular ABMR scores as a histology‐independent estimate of ABMR in 703 biopsies, we reassessed criteria for ABMR to determine relative importance of various lesions, the utility of equations using actual scores rather than cutoffs, and the potential for diagnosing ABMR when DSA is unknown or negative. We confirmed that the important features for ABMR diagnosis were peritubular capillaritis (ptc), glomerulitis (g), glomerular double contours, DSA and C4d staining, but we questioned some features: arterial fibrosis, vasculitis, acute tubular injury, and sum of ptc+g scores. Regression equations using lesion scores predicted molecular ABMR more accurately than score cutoffs (area under the curve 0.85–0.86 vs. 0.75). DSA positivity improved accuracy, but regression equations predicted ABMR with moderate accuracy when DSA was unknown. Some biopsies without detectable DSA had high probability of ABMR by regression, although most had HLA antibody. We concluded that regression equations using lesion scores plus DSA maximized diagnostic accuracy and can estimate probable ABMR when DSA is unknown or undetectable.