Do Classic Preparations of Gastrointestinal Cytology Perform Differently Than ThinPrep Cases? Observations from the College of American Pathologists Interlaboratory Comparison Program in Nongynecologic Cytology (CAP NGC)
J Bentz, M Schwartz, R Soures, A Clayton, B Crothers, R Laucirica, B Chmara, K Clary, P Wasserman, A Moriarty. University of Utah, Salt Lake City, UT; Methodist Hospital, Houston, TX; College of American Pathologists, Northfield, IL; Mayo Clinic, Rochester, MN; Walter Reed Army Medical Center, Washington, DC; Baylor College of Medicine, Houston, TX; Rochester General Hospital, Rochester, NY; Long Island Jewish Medical Center, New Hyde Park, NY; Ameripath Indiana, Indianapolis, IN
Background: Liquid-based preparations (LBP) differ morphologically from traditional slide preparations and their use is becoming more common in examination of gastrointestinal (GI) cytology specimens. Discordant participant responses from challenges distributed in the College of American Pathologist's Interlaboratory Comparison Program in Nongynecologic Cytology (CAP NGC) were analyzed.
Design: Malignant ThinPrep challenges distributed between 2000-07 were compared to classical preparations (smears, cytospins) for discordant responses (negative or unsatisfactory). Differences between preparation type, participant type, and diagnosis were analyzed.
Results: Classic preparations comprised 93% (n=11588) of the GI challenges while 7% (n=912) were ThinPrep preparations. An exact match to the reference diagnosis of positive-malignant was seen in 88.5% of conventional preparations and 95.9% of ThinPrep (P<.001) challenges. These results were statistically significant when the specific reference diagnosis was adenocarcinoma (P<.001) but no difference for squamous cell carcinoma (P>.99). Cytotechnologist overall performance was not different compared to pathologists (89.2% v 89.0%; P=.75). For specific reference diagnosis, there was statistically better performance by cytotechnologists for cases of squamous cell carcinoma (96.3% vs. 92.6%; P<.001) and better performance by pathologists for cases of spindle cell neoplasm (79.7% vs. 42.9%; P<.001). No performance difference between participant types existed for adenocarcinoma (88.4% vs. 88.0%; P=0.637), carcinoid (83.7% vs. 67.3%; P=.06) or carcinoma NOS (52.9% vs. 66.7%; P=.19).
Conclusions: ThinPrep performed significantly better than classic preparations of GI cytology specimens. Performance varied by reference interpretation, with adenocarcinoma performing best. Cytotechnologists and pathologists perform the same overall.
Wednesday, March 11, 2009 1:00 PM
Poster Session VI # 60, Wednesday Afternoon