Gastrointestinal (GI) tract cytology has high specificity but poor sensitivity for detecting GI tract cancer. Newer methods of slide preparation may improve cytology performance and additionally permit molecular slide-based assays that could improve diagnostic accuracy. A split-sample validation study compared slides prepared using ThinPrep UroCyte filters or a cytocentrifuge method with respect to cellularity, stain quality, and interpretation. In this 15-slide split-sample study, UroCyte slide preparations were judged to be superior to cytocentrifuge preparations, and the method was implemented for GI cytology in December 2006. To assess diagnostic performance for GI cytology, we retrospectively reviewed outcomes for one year before and after implementation of UroCyte filter slide preparation. Sensitivity, specificity, positive predictive value, and negative predictive value for both slide preparations were largely equivalent to one another and compared favorably with values in the literature, but varied greatly depending on how atypical and suspicious-atypical cases were defined for calculations. For biopsied biliary samples, the highest sensitivities were observed when all atypical and suspicious-atypical cases were considered positive for malignancy, but were lower when suspicious-atypical cases were considered positive and atypical cases were considered negative for malignancy. This highlights the difficulty with comparing studies that define atypical classes differently, and points to the need for a well-defined approach to performance evaluation that relates directly to how diagnostic information is used clinically. We conclude that the UroCyte filter slide preparation is valid for evaluation of GI cytology specimens and may simplify adjunct molecular testing such as FISH. This is the first reported use of UroCyte filters for preparation of GI specimens.

No abstract