Urinary cytology has a poor performance for predicting invasive or high-grade upper-tract urothelial carcinoma

OBJECTIVE To evaluate the diagnostic accuracy of urine cytology for detecting aggressive disease in a multi-institutional cohort of patients undergoing extirpative surgery for upper-tract urothelial carcinoma (UTUC). METHODS We reviewed the records of 326 patients with urinary cytology data who underwent a radical nephroureterectomy or distal ureterectomy without concurrent or previous bladder cancer. We assessed the association of cytology (positive, negative and atypical) with final pathology. Sensitivity and positive predictive value (PPV) of a positive (± atypical) cytology for high-grade and muscle-invasive UTUC was calculated. RESULTS On final pathology, 53% of patients had non-muscle invasive disease (pTa, pTis, pT1) and 47% had invasive disease (≥pT2). Low-grade and high-grade cancers were present in 33% and 67% of patients, respectively. Positive, atypical and negative urine cytology was noted in 40%, 40% and 20% of cases. Positive urinary cytology had sensitivity and PPV of 56% and 54% for high-grade and 62% and 44% for muscle-invasive UTUC. Inclusion of atypical cytology with positive cytology improved the sensitivity and PPV for high-grade (74% and 63%) and muscle-invasive (77% and 45%) UTUC. Restricting analysis to patients with selective ureteral cytologies further improved the diagnostic accuracy when compared with bladder specimens (PPV > 85% for high-grade and muscle-invasive UTUC). CONCLUSIONS In this cohort of patients with UTUC treated with radical surgery, urine cytology in isolation lacked performance characteristics to accurately predict muscle-invasive or high-grade disease. Improved surrogate markers for pathological grade and stage are necessary, particularly when considering endoscopic modalities for UTUC.