[738] Biliary Strictures: Methods for Improving Diagnostic Accuracy

Marcela Salomao, Tamas A Gonda, Vasco Eguia, Helen Remotti, John Poneros, Amrita Sethi, Anjali Saqi. Columbia University, New York, NY

Background: Biliary strictures may be caused by inflammatory or neoplastic processes. Differences in management emphasize the need to establish the underlying causes. Biliary brushing is the initial intervention to assess the etiology of strictures, and often concurrent biopsies and FISH studies are also performed. Prior studies have established that brushings have a high specificity but low sensitivity. The aim of this study was to determine the etiology for the low cytological yield and to assess which sampling method(s) (cytology, biopsy and/or FISH) improve the diagnostic yield of pancreatobiliary tumors.
Design: A total of 48 patients with biliary strictures including 29 cases of histology-proven pancreatobiliary adenocarcinomas (PBCa) and 19 'control' cases (benign strictures on resection and/or 17-40 months of clinical follow-up) were retrieved. Brush cytology specimens, FISH analysis (UroVysion probes for chromosomes 3, 7, and 17, and locus 9p21) and bile duct biopsies were re-reviewed by 2 pathologists. Cytology specimens were evaluated for overall cellularity and presence of the following features: large atypical cells with foamy cytoplasm (LACF), drunken honeycomb (DH) and loosely cohesive clusters of round cells (LCCRC). Biopsy specimens were examined for presence or absence of stromal invasion (SI).
Results:

Brushing vs. FISH vs. biopsy

	PBCa* (n=29)	Control* (n=19)	Sensitivity	Specificity	PPV	NPV
Brushing#	13/29 (44.8)	1/19 (5.3)	0.448	0.947	0.929	0.529
FISH	21/29 (72.4)	2/19 (10.5)	0.7	0.895	0.913	0.654
Forceps biopsy	16/21 (76.2)	0/7 (0)	0.786	0.938	0.971	0.625
FISH and brush or biopsy	19/29 (65.5)	1/19 (5.3)	0.655	0.947	0.95	0.643

* positive/total (%); # positive defined as suspicious or positive for adenocarcinoma on cytology

Scant cellularity was present in 55.2% of PBCas and 63.2% of control cases and resulted in 7 of 8 cases of non-diagnostic or false negative cytology results. LACF, DH and LCCRC were identified in 90.5%, 66.7% and 57.1% of PBCas vs. 12.5%, 12.5% and 0% of controls, respectively) (p values 0.0002, 0.007 and 0.001, respectively). The presence of SI aided in diagnosis of carcinoma in biopsies (11 of 16 positive).
Conclusions: Our study suggests that higher specimen cellularity and the presence of LACF, DH, and LCCRC increase the sensitivity in cytology specimens. In addition, sampling of stromal tissue may facilitate the diagnosis of invasive PBCa. Concurrent biopsies and FISH studies are helpful in increasing the diagnostic yield of pancreatobiliary tumors.
Category: Gastrointestinal

Tuesday, March 20, 2012 9:30 AM

Poster Session III # 140, Tuesday Morning