Background and aims: Our goal was to compare the diagnostic accuracy of FISH in the detection of malignancy compared with other standard diagnostic modalities, including brush cytology and biopsy specimens over a 10-year period of prospective data collection.
Methods: We conducted a review of all consecutive biliary strictures evaluated between 2006 and 2016. Patients with a final pathologic diagnosis or conclusive follow-up were included. We evaluated the performance of FISH polysomy (CEP 3, 7, and 17) and 9p21 deletion as well as cholangioscopic biopsy (CBx) and EUS-FNA. Statistical analysis was performed with the Mann-Whitney U and Fisher's exact tests.
Results: Of 382 patients with indeterminate strictures, 281 met inclusion criteria. Forty-nine percent were malignant. Cytology, FISH polysomy, and FISH polysomy/9p21 showed a specificity of 99.3%. FISH polysomy/9p21 as a single modality was the most sensitive at 56% (p < 0.001). The sensitivity of FISH polysomy/9p21 and cytology was significantly higher than cytology alone at 63 versus 35% (p < 0.05). EUS-FNA for distal strictures and CBx for proximal strictures increased sensitivity from 33 to 93% (p < 0.001) and 48-76% (p = 0.05) in cytology-negative strictures.
Conclusions: The high specificity of FISH polysomy/9p21 suggests that a positive result is sufficient for diagnosing malignancy in indeterminate strictures. The significantly higher sensitivity of FISH polysomy/9p21 compared to cytology supports the use of FISH in all non-diagnostic cases. Although both EUS-FNA and CBx were complimentary, our results suggest that distal strictures should be evaluated by EUS initially. Proximal strictures may be evaluated by FISH first and then by CBx if inconclusive.
Keywords: Biliary strictures; Cholangiocarcinoma; ERCP; EUS-FNA; Fluorescence in situ hybridization; Pancreatic cancer.