[420] Accuracy of Grading of Urothelial Carcinoma (UCA) on Urine Cytology: A Measure of Interobserver (INTEROA) and Intraobserver Agreement (INTRAOA).

Michelle D Reid, Adeboye O Osunkoya, Momin T Siddiqui, Stephen W Looney. Emory University Hospital, Atlanta, GA; Medical College of Georgia, Augusta

Background: The accuracy and reproducibility of histologic grading of UCA is well established however cytologic grading of UCA using current World Health Organization (WHO) criteria is not. We determined the accuracy and reproducibility of UCA grading in urine by measuring INTEROA and INTRAOA among 3 pathologists with and without expertise in cytopathology, compared with the gold standard biopsy or resection.
Design: Forty-four UCA-positive urines with tissue confirmation were blindly and independently graded twice (using WHO criteria) by 2 cytopathologists (MR, MS) and 1 genitourinary (GUP) pathologist (AO), with a 1-week-interval between grading. Pathologists were blinded to histologic findings. Coefficient kappa was used to measure INTEROA and INTRAOA. Accuracy was measured by percentage agreement with gold standard, for each pathologist on each occasion and for all pathologists and occasions combined.
Results: Histologic diagnoses included high-grade UCA (32/44, 73%), low-grade UCA (3/44, 7%) and carcinoma in-situ (CIS) (9/44, 20%). Grading accuracy ranged from 64%-89% for the 3 pathologists. AO was more accurate than MR (76% vs 60%) and MS (76% vs 67%) on both occasions combined. Overall accuracy for pathologists and occasions combined was 77% (95% C.I., 72% – 82%). INTEROA was unacceptably low [Table 1].

Coefficient Kappa for INTEROA
ObserverOccasionKappa95% C.I.p-Value
MS vs AO10.05-0.26, 0.361.000
MS vs MR10.22-0.08, 0.510.235
AO vs MR10.370.09, 0.660.013*
MS vs AO20.15-0.19, 0.500.573
MS vs MR20.570.30, 0.84<0.001*
AO vs MR20.29-0.01, 0.590.053
* Significantly different from zero

Only 2 kappa values were significantly different from 0 and only one 95% C.I. included the value 0.75 (minimally acceptable for a reliable clinical measurement). No pathologist differed significantly in INTRAOA.
Conclusions: Although INTRAOA in cytologic grading of UCA was acceptable, INTEROA was unacceptably low. Additionally overall accuracy of grading was minimally acceptable at best. Because of unacceptable INTEROA, marginally acceptable INTRAOA and limited accuracy, cytologic grading of UCA is clinically unreliable and is not recommended for routine practice.The fact that GUP was the most precise suggests that accuracy is independent of expertise in cytopathology.
Category: Cytopathology

Wednesday, March 2, 2011 1:00 PM

Poster Session VI # 67, Wednesday Afternoon


Close Window