Diagnostic Performance of The Paris System 2022 in Urinary Cytology: A Single-Institution Retrospective Study Comparing Voided and Instrumented Specimens

Abstract

Background: Urinary cytology is a widely used, noninvasive tool for detecting high-grade urothelial carcinoma (HGUC). The Paris System for Reporting Urinary Cytology (TPS) was updated in 2022 (TPS 2.0) to improve diagnostic accuracy and reproducibility. This study evaluated TPS 2.0 performance and compared voided versus instrumented specimens using histopathological follow-up as the reference standard. Methods: We retrospectively reviewed 2,315 urinary cytology samples processed at Gazi University Faculty of Medicine in the years 2018 and 2019, identifying 371 cases with histopathological follow-up within 6 months. Cytology specimens were reclassified according to TPS 2022 categories and correlated with biopsy results. Risk of high-grade malignancy (RoHM), sensitivity, specificity, positive predictive value, and negative predictive value were calculated overall and stratified by specimen type (voided vs. instrumented). Results: Of 371 cases, 93 (25.1%) were HGUC on histopathology. RoHM increased progressively across categories: negative for HGUC 4.4%, atypical urothelial cell (AUC) 55%, suspicious for high-grade urothelial carcinoma (SHGUC) 72.7%, and HGUC 96%. Using HGUC alone as the positive category yielded high specificity (99.3%; 95% CI: 97.4-99.8) but limited sensitivity (51.6%; 95% CI: 41.6-61.5). Expanding thresholds to HGUC+SHGUC increased sensitivity to 68.8% (95% CI: 58.8-77.3) while including AUC maximized sensitivity (80.6%; 95% CI: 71.5-87.4) with reduced specificity (93.9%; 95% CI: 90.4-96.1). Instrumented specimens outperformed voided urine, achieving higher sensitivity (88.7%; 95% CI: 77.5-94.9) and specificity (94.3%; 95% CI: 90.2-96.7). Conclusion: TPS 2022 provides reliable risk stratification for urinary cytology, with a predictable increase in RoHM across categories. Instrumented specimens demonstrate superior diagnostic performance, supporting TPS 2.0 as a robust framework for routine practice.