P0093improved Interobserver Agreement in Banff Scoring of Renal Allograft Biopsies When Routinely Using Cd3/Pas Stain

Abstract

Background and Aims The BANFF Classification of renal allograft rejection consists of scores for cellular and antibody mediated rejection. The scores for tubulitis (t), interstitial inflammation (i), glomerulitis (g), total inflammation (ti), and endotheliatis (v) have direct prognostic and therapeutic significance as they contribute to a diagnosis of renal allograft rejection. The original BANFF scoring was developed using hematoxylin and eosin (H&E) stain only. This study examines the level of interobserver agreement using H&E and a combination CD3/PAS stain that highlights T lymphocytes and basement membranes. Method 50 consecutive renal allograft biopsies were retrieved and on each case H&E and CD3/PAS stains (Ventana/Roche Tissue Diagnostics clone 2GV6 for CD3) were performed. CD3/PAS and H&E slides were matched and blinded for two reviewers [reviewer (R1) and reviewer 2 (R2)]. Each case was graded in an identical manner following criteria designated for (t),(i),(g) and (ti) as previously described in the Banff Classification for Renal Allograft Biopsies. Cases were not graded for (v) as there were too few cases with endothelialitis to allow for statistical evaluation. The matched scores were analyzed using weighted Kappa statistics for interobserver agreement within each group. Agreement levels were characterized using Landis and Koch descriptions for level of agreement. All p values were two-tailed, values of <0.05 were considered statistically significant. Statistical analyses was performed using SAS Version 9.4, Cary NC. Results Mean age was 44 years with a range of 9 -75 years. Standard deviation (SD) for age was 18 years. Gender distribution was 20% female and 80% male. Agreement between R1 and R2 for (t) scores was moderate (Weighted Kappa 0.5276) as opposed to fair for H&E group (Weighted Kappa 0.4189). Similarly, interobserver agreement was moderate for scores of (i) and (ti) in the CD3/PAS group and fair in the H&E group. For both CD3/PAS and H&E interrater evaluations, agreement was fair to poor for (g) with weighted Kappa of 0.3928 and 0.3559 respectively. Conclusion This study showed increased levels of interobserver agreement when using a CD3/PAS combination stain to score allograft biopsies for (t), (i) and (ti). Since (t), (i) and (ti) scores directly affect a diagnosis of rejection, we propose routine use of CD3/PAS in addition to H&E when reviewing renal allograft biopsies. Glomerulitis (g) shows poor interobserver agreement regardless of the stain used and further studies to refine the criteria for (g) in Banff revisions of the classification may be of value.