Diagnostic yield of brush cytology for biliary stenosis during ERCP.

AIM To evaluate the diagnostic yield of brush cytology for biliary strictures detected on ERCP when a systematic approach is used. PATIENTS AND METHODS Data on 62 consecutive patients with a biliary stricture on ERCP were collected. Cytological samples were processed immediately after brushing in the endoscopy room, and all were analyzed by the same pathologist. For the statistical analysis specimens were classified as positive, negative, suspicious for malignancy (presence of atypias), and unsatisfactory for evaluation. Final diagnosis was based on either histopathologic (surgery or biopsies by other techniques) or clinical/radiographic diagnosis. RESULTS A total of 71 cytological specimens were included. Definite diagnosis was malignancy in 49 samples, and benign stricture in 22. Three samples were excluded because of insufficient material or processing artifacts. The sensitivity of biliary brushing was 62% (95% CI 0.47-0.77), specificity was 100%, positive predictive value was 100%, and negative predictive value was 58% (95% CI 0.43-0.75). When suspicious samples were included as malignant, sensitivity was 67% (95% CI 0.54-0.81) without changes in the remaining parameters. Eight patients underwent more than one ERCP. Repeated brush cytology exams in these patients yielded a definitive diagnosis in every case. CONCLUSIONS Brush cytology has intermediate sensitivity with a high specificity. A systematic approach with a dedicated pathologist and the inclusion of significant atypias as malignant results improves sensitivity. Due to its simplicity brush cytology should be performed in all cases of biliary strictures detected on ERCP, and in case of repeated ERCPs additional cytology brushings are recommended.