Prognostic accuracy of individual uropathologists in noninvasive urinary bladder carcinoma: a multicentre study comparing the 1973 and 2004 World Health Organisation classifications.

BACKGROUND: Grading of noninvasive papillary urinary bladder carcinoma (PUC) is routinely performed in clinical oncologic practice; however, reports regarding diagnostic and prognostic accuracy are contradictory.
OBJECTIVE: To compare the 1973 and 2004 World Health Organisation (WHO) classifications in terms of interobserver variability and prognostic implications. DESIGN, SETTING, AND PARTICIPANTS: Two hundred PUC were retrospectively reviewed by four independent expert genitourinary pathologists blinded with respect to patient identity and clinical outcome. Tumour grading was assigned according to the 1973 and 2004 WHO classifications. Surveying a mean postsurgical follow-up of 71.8 mo (range: 18-163 mo), clinical outcome in terms of recurrence-free and progression-free survival was recorded for all patients. INTERVENTION: All of the patients underwent transurethral resection of the bladder. MEASUREMENTS: The generalised κ (kappa statistic) for interobserver variability was calculated, and Kaplan-Meier analysis as well as univariate regression analysis were performed to evaluate prognostic implications in terms of recurrence and progression rates. RESULTS AND LIMITATIONS: During the follow-up, a total of 84 (42%) patients experienced recurrence, whereas another 18 (9%) patients featured disease progression. Owing to the rare presence of papillary urothelial neoplasms of low malignant potential (PUNLMP) in our cohort (0-3.5%), the 2004 WHO classification approached a two-tier system (low and high grade), which showed less interobserver variability than the 1973 classification (κ: 0.30-0.52 vs 0-0.37, respectively). In comparing the power of both classifications to separate indolent from aggressive PUC, striking pathologist-dependent differences became apparent.
CONCLUSIONS: Both WHO classifications for grading of PUC suffer from substantial interobserver variability, with the 2004 WHO classification showing less interobserver variability. Stark differences in the prognostic power of the individual grading approaches were also found. These significant differences in the individual interpretation of the WHO grading schemes for noninvasive PUC highlight the necessity of better-defined criteria for conventional tumour grading; otherwise, the subdivision into prognostically different groups by conventional histomorphology might remain of limited value.