The diagnostic criteria for low-grade urothelial lesions that have been described in the past were based on urinary specimens prepared by the cytospin method. Recognizing the recent popularity of the ThinPrep methodology and the cytologic alterations it introduces to the cellular features, we sought to evaluate the reproducibility of these criteria in ThinPrep urinary samples. One hundred twenty-six ThinPrep urinary specimens with a tissue diagnosis of low-grade urothelial carcinoma (LGUC) and 45 negative controls were evaluated. Three pathologists blindly reviewed the slides separately and the consensus on each feature was used in the study. Logistic regression analysis was used to determine which criteria in combination were most predictive of low-grade urothelial carcinoma. All specimens were evaluated for the following 18 features: nucleus/cytoplasm ratio, irregular nuclear border, cytoplasm homogeneity, cell clusters, high cellularity, prominent nucleoli, granular nuclear chromatin, hyperchromasia, acute inflammation, vesicular chromatin, nuclear molding, nuclear eccentricity, elongated nuclei, necrosis, anisonucleosis, irregular bordered fragments, absent cytoplasmic collar, and peripheral palisading. High nucleus-to-cytoplasm ratio, irregular nuclear borders, and homogeneous cytoplasm (combination sensitivity of 59% and specificity of 100%) were the best predictive features for LGUC. Minor predictive criteria were eccentric nuclei and nuclear molding. ThinPrep provides well preserved, cleaner specimens without significantly altering the morphology. The three key criteria applied in cytospin specimens to diagnose LGUC were reproducible in ThinPrep specimens.
Copyright 2003 Wiley-Liss, Inc.