Reproducibility Assigning Grade in Noninvasive Papillary Urothelial Neoplasms among Dedicated Genitourinary Pathologists: A Single-Institution Study Based on the 2004 WHO Classification
Rohit Mehra, Ali Amin, Lan L Gellert, Anuradha Gopalan, Ying-Bei Chen, Samson W Fine, Satish K Tickoo, Victor E Reuter, Hikmat Al-Ahmadie. Memorial Sloan-Kettering Cancer Center, New York, NY
Background: The 2004 WHO classification is the most commonly used system to diagnose and grade bladder lesions. Inter- and intra-observer variability in assessing tumor grade in non-invasive papillary urothelial neoplasms (PUN) is well known. Despite its inherent subjectivity, as subspecialization becomes more prevalent, greater degree of concordance among "experts" and decreased inter-observer variability is expected if pathologists routinely consulted with each other for tumor grading. This study attempts to determine the inter-observer variability in grading PUN among dedicated genitourinary (GU) pathologists practicing at a single institution.
Design: 50 cases of PUN were prospectively selected by a senior member of the team, 4 years prior to the study. Cases were independently reviewed by 6 GU pathologists blinded to patient identity, prior diagnosis and clinical outcome. Based on the 2004 WHO classification, lesions were classified as hyperplasia, papilloma, papillary urothelial neoplasm of low malignant potential (PUNLMP), low grade (LGPUC) or high grade papillary urothelial carcinoma (HGPUC). Consensus was defined as agreement among at least 4 of 6 pathologists. Free-marginal kappa for interobserver variability was calculated.
Results: Overall, there was good agreement among the 6 GU pathologists (κ=0.64). Based on our definition, consensus was achieved in 84% (42/50) of cases. Unanimous agreement (UA) or agreement among 5 of 6 pathologists was seen in 42% (21/50) of cases, each.
Of the 8 cases without consensus, discrepant cases included LGPUC vs HGPUC (5) and PUNLMP vs LGPUC or HGPUC (3). UA was observed in 11 LGPUCs, 8 HGPUCs and 2 urothelial papillomas. A diagnosis of PUNLMP was rendered by at least 1 pathologist 6 times but consensus was reached only once for this category.
Conclusions: In spite of subspecialization and consultation among GU pathologists, considerable inter-observer variability exists in the grading of PUN based on the 2004 WHO classification. This is particularly true in evaluating low grade lesions, mostly PUNLMP. It is evident that even experienced pathologists are reluctant to assign a papillary urothelial neoplasm to a non-malignant category. Novel molecular markers might help in reducing diagnostic variability as well as reflect the true progression/recurrence of these lesions. We hypothesize that grade plus marker will lead to increased interobserver concordance and risk stratification. The search and subsequent evaluation of these markers is ongoing.
Category: Genitourinary (including renal tumors)
Wednesday, March 21, 2012 1:00 PM
Poster Session VI # 148, Wednesday Afternoon