Interobserver Agreement for the Polyomavirus Nephropathy (PVN) Classifying Schema in Renal Allografts
A Sar, S Worawichawong, H Benediktsson, J Zhang, S Yilmaz, K Trpkov. Calgary Laboratory Services and University of Calgary, Calgary, Canada
Background: A classification schema for PVN grading was proposed at a 2009 Banff conference on allograft pathology. The schema introduced three stages of PVN: early (stage A), florid (stage B) and late sclerosing (stage C). In addition, histologic reporting of the viral load (VL) was proposed.
Design: To examine the applicability and the interobserver agreement for the PVN schema, we retrieved cases with confirmed PVN by histology and SV40 from our institutional database. We identified 24 allograft biopsies with PVN which fulfilled the adequacy criteria. Four pathologists independently scored the PVN stage (A, B or C), not knowing the clinical history. VL was scored as a numerical percent of tubules with viral replication and the results were assessed using either a 3-tier score (VL1: ≤1%, VL2: >1% to ≤10%, VL3: >10%) or a 4-tier score (VL1: ≤1%, VL2: >1% to ≤5%, VL3: >5% to 15%, VL4: >15%). Reviewer agreement on PVN stage was evaluated by Kappa statistic, whereas the agreement on VL was evaluated by the Kendall's coefficient of concordance. The type of our practice may be a potential study limitation, because cores are divided and processed in plastic and paraffin, with SV40 being performed only on paraffin cores.
Results: Complete agreement for the PVN stage score was found in 12 of 24 (50%) cases and 3 of 4 reviewers agreed in 8 of 24 (33%) cases. Complete disagreement for PVN stage was observed in 4 (17%) cases. The overall kappa score for PVN stage was 0.47 (95% confidence interval 0.35 - 0.60, p<0.001). The highest kappa value was found for stage A (0.66), followed by B (0.43) and C (0.24). Using the 3-tier VL score there was a complete agreement in 13 of 24 (54%) cases and 3 of 4 reviewers agreed in 7 of 24 (29%) cases. Using the 4-tier VL score, there was a complete agreement in 6 of 24 (25%) cases and 3 of 4 reviewers agreed in 10 of 24 (42%) cases. Overall, there was a substantial agreement between the reviewers using both the 3-tier and the 4-tier score for PVN VL (Kendall's concordance coefficients 0.72 and 0.76, respectively; p<0.001 for both).
Conclusions: This is a first attempt to evaluate the proposed PVN classifying schema. Moderate kappa agreement was achieved in scoring the PVN stage; reviewers however did not know the clinical history, which may have reduced the interobserver agreement. Although substantial agreement was found using both the 3-tier and the 4-tier VL score, a better complete agreement was found using the 3-tier score.
Category: Kidney (does not include tumors)
Wednesday, March 24, 2010 1:00 PM
Poster Session VI # 226, Wednesday Afternoon