Impact of intravesical therapy for non-muscle invasive bladder cancer on the accuracy of urine cytology

Abstract

Purpose: Urine cytology remains an essential diagnostic tool in the surveillance of patients with non-muscle invasive bladder cancer (NMIBC). The correlation of urine cytology with biopsy specimens to determine its accuracy following induction intravesical therapy has not been investigated.

Methods: A retrospective review was performed of patients who underwent intravesical therapy for biopsy-proven non-muscle invasive disease between 2013 and 2016 at our institution. All patients uniformly underwent cytology and systematic bladder biopsies in the operating room within 12 weeks following intravesical therapy. The accuracy of urinary cytology in predicting high-grade disease recurrence following intravesical therapy was confirmed by correlating cytology results to post-treatment systematic biopsies, regardless of endoscopic findings. Only patients with complete information regarding urine cytology and pathologic biopsy results, both pre- and post-intravesical therapy, were included.

Results: 90 cytology samples following intravesical therapy were analyzed from 76 patients who met inclusion criteria. 72 (80.0%) and 18 (20.0%) of the samples were collected from patients initially treated for high- and low-grade disease, respectively. Fifty-six (62.2%) specimens were obtained from patients following induction of bacillus Calmette-Guerin (BCG) therapy; the remainder were from patients treated with intravesical gemcitabine/docetaxel, mitomycin, or BCG/interferon. For patients treated with BCG, cytology was positive for high-grade disease in 8/15 patients with high-grade pathology on follow-up biopsy, thus demonstrating a sensitivity of 53% (95% CI 27-79%), specificity of 95% (95% CI 84-99%), positive predictive value of 80% (95% CI 44-98%), and negative predictive value of 85% (95% CI 71-94%). If cytologic interpretation was broadened to include high-grade and "suspicious for high-grade" findings, sensitivity increased to 67% (95% CI 38-88%) and specificity decreased to 88% (95% CI 74-96%).

Conclusions: While urinary cytology maintains a high specificity following intravesical therapy, it demonstrates a low sensitivity for potentially aggressive high-grade urothelial carcinoma. Further evaluation of more effective, clinic-based enhanced cystoscopy techniques and biomarkers is warranted to better identify patients at risk for disease recurrence following BCG therapy.

Keywords: Bacillus Calmette–Guerin; Urinary bladder neoplasms; Urine cytology; Urothelial carcinoma.