Applying the Paris system for reporting urine cytology to challenging cytology cases

BACKGROUND: Implementing the Paris system for reporting urine cytology (TPS) can substantiate atypical diagnosis while improving standardization and risk stratification. This study evaluates its performance and reproducibility in challenging cases and examines whether focused education of morphological features can improve outcomes. METHODS: In our prior study, urine cytology cases diagnosed as "atypical" with surgical follow-up were used. Cases showing poor agreement in that study were collected for this one. Representative photographs of each case were taken and distributed via online questionnaires. Participants were asked to render an initial diagnosis and evaluate the presence of several morphological features. Educational material was distributed, followed by additional questionnaires. RESULTS: Three participants evaluated 40 cases before and after educational materials. TPS diagnoses were significantly more specific (0.23 vs 0.59, P = 0.004) and more accurate (0.43 vs 0.66, P = 0.0125) than diagnoses made with our institutional system. Fewer overall cases were diagnosed as "atypical" with TPS. TPS education resulted in slightly, though not significantly, more specific diagnoses (0.25 vs 0.59, P = 0.083). Interobserver agreement decreased for nuclear-to-cytoplasmic (N/C) ratio, TPS diagnoses and initial diagnoses, and increased for all other features. TPS resulted in downgrading of cases with biopsy-proven low grade urothelial neoplasm (LGUN) from "atypical" to negative for high grade urothelial carcinoma (NHGUC) (P = 0.018). CONCLUSIONS: Use of TPS in challenging urine cytology cases can improve specificity, risk stratification, and diagnostic accuracy while decreasing the number of "atypical" diagnoses. Though training can help cytopathologists better apply these criteria, it is unclear how to effectively improve evaluation of N/C ratio.