The Pathologist's Mean Grade Is Constant and Individualizes the Prognostic Value of Bladder Cancer Grading
Background: A new grading system for bladder cancer (BCa) was adopted in 2004 to reduce observer variability and provide better prognostic information. Objective: We compared the World Health Organization (WHO) 1973 and 2004 systems for observer variability and prognosis. Design, setting, and participants: Slides of 173 primary non-muscle-invasive BCa were reviewed two times by four pathologists. Measurements: Intra- and interobserver variability were assessed using κ statistics. We determined the mean grade (eg, G1/low malignant potential is 1 grade point, G2/low grade is 2 grade points) of the pathologists per grading cycle. Kaplan-Meier analyses were applied for prediction of recurrence and progression. Results and limitations: For WHO 2004 and 1973 grading, the agreement between the pathologists was 39-74% (κ: 0.14-0.58) and 39-64% (κ: 0.15-0.41), respectively. The intraobserver agreement varied from 71% to 88% (κ: 0.55-0.81). The mean grade of a pathologist was constant (difference below 0.1 grade point) irrespective of the grading system. Conversely, mean-grade differences among the pathologists were high, up to 0.7 grade point. The mean grades for the WHO 2004 system were 0.3-0.5 grade point higher than those of WHO 1973. Mean grade distinguished low and high graders among the pathologists and was strongly linked with risk of progression in each grade category. Conclusions: The variation in mean grade among individual pathologists exceeded the grade shift caused by WHO 2004 grading. Knowledge of the pathologist's mean grade allows a better assessment of the prognostic value of grading. Mean grade has the potential to become a tool for quality assurance in pathology.