1773 INTEROBSERVER REPRODUCIBILITY OF THE 1973 WHO AND 2004 WHO/ISUP NON-INVASIVE BLADDER CANCER CLASSIFICATIONS AMONG GENERAL AND GENITOURINARY PATHOLOGISTS

Abstract

INTRODUCTION AND OBJECTIVES: Pathological grade is the most important prognostic indicator used to risk stratify and select treatment for non-invasive urothelial carcinoma (UC). The 1973 World Health Organization (WHO) grading classification has been most widely used, but suffers from unacceptably high interobserver variability. The 2004 WHO/International Society of Urological Pathology (ISUP) revision was developed to reduce variability by providing welldefined histological criteria for each grade. We compared interobserver reproducibility between the 1973 and 2004 systems and between general versus expert genitourinary pathologists. METHODS: 1 general pathologist (GP) and 2 genitourinary pathology specialists (GU1 and GU2) were blinded to patient identity and graded 98 consecutive UC specimens retrospectively using the 1973 and 2004 classifications. Kappa statistics ( ) were used to measure interobserver reproducibility. By convention, values from 0.21 – 0.4 represent “fair” agreement, from 0.41 0.6 represent “moderate” agreement, and 0.6 represent “substantial” agreement. As an example, values for prostate cancer Gleason score among genitourinary pathologists have been reported to be 0.6. RESULTS: Overall agreement was only fair for both systems and while higher for the 2004 system than the 1973, this was not significant ( : 0.38 versus 0.26, respectively). Agreement between the two specialists versus agreement between each specialist and the generalist was higher using the 1973 system ( : 0.37 vs 0.33 and 0.18) but lower using the 2004 system ( : 0.31 vs 0.46 and 0.39). The generalist had better agreement with both specialists using the 2004 system versus the 1973 system ( : 0.46 and 0.39 versus 0.33 and 0.18). These findings were not statistically significant. CONCLUSIONS: Reproducibility is not improved by the 2004 WHO/ISUP revision to the 1973 WHO grading system, even among expert genitourinary pathologists. Since management of non-invasive UC is primarily determined by its grade, this limits a clinician’s ability to risk stratify tumor and potentially alters treatment. Our findings underscore the need to identify molecular markers that can provide a more reliable risk stratification system.