BACKGROUND Several liquid-based cytology (LBC) methods are currently used, but the diagnostic accuracy of each method is not well known. We aimed to compare the diagnostic performance of SurePathTM LBC and conventional smear (CS) cytology in endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) samples of esophageal, gastric, and duodenal lesions.

METHODS As a prospective randomized noninferiority study, patients who needed EUS-FNA due to subepithelial mass in the upper gastrointestinal tract were randomly assigned 1:1 to the LBC and CS groups. Cytologic preparation was carried out using a crossover design where 1 method was used for the first needle-pass sample and another method was used for the second needle-pass sample. The primary outcome was to compare the diagnostic performance between LBC and CS using the final diagnosis as the gold standard.

RESULTS A total of 87 patients were randomized and 60 patients were analyzed. There were no differences between LBC and CS in diagnostic accuracy (91.7% vs 86.7%, P = .380), sensitivity (97.7% vs 90.7%, P = .169), specificity (76.5% vs 76.5%, P > .99), negative predictive value (92.9% vs 76.5%, P = .225), or positive predictive value (91.3% vs 90.7%, P = .921). The background of LBC was less bloody than that of CSs (5.0% vs 53.3%, P < .001) and the sample preparation time of LBC was shorter than that of CSs (29 ± 7 seconds vs 90 ± 17 seconds, P < .001).

CONCLUSION In the EUS-FNA of a subepithelial mass in the upper gastrointestinal tract, the diagnostic performance of LBC was not inferior to that of CS. The field of view was better in LBC, because the background was less bloody and necrotic. As LBC is more convenient to perform and takes shorter time, it is expected that it can replace the CS method for EUS-FNA samples.