The role of rapid on-site evaluation on diagnostic accuracy of endoscopic ultrasound fine needle aspiration for pancreatic, submucosal upper gastrointestinal tract and adjacent lesions.

Background And Aim: Our aim was to assess adequacy and diagnostic accuracy of endoscopic ultrasound-fine needle aspiration (EUS-FNA) specimens with or without rapid on-site evaluation (ROSE) from pancreatic, upper gastrointestinal tract (UGIT) and adjacent masses.

Method: A retrospective cohort study based on patients' files who underwent EUS-FNA in Galilee Medical Center in a 4 years period. Number of needle passes, repeated EUS and ROSE effect on tissue adequacy and diagnostic accuracy were reported.

Results: One-hundred sixty-one patients were included. Ninety-three patients (57.7%) underwent EUS-FNA without ROSE (group A) compared to 68 patients (42.3%) with ROSE (group B). The most common location was in the pancreas (55% in group A vs 81% in group B). Addition of ROSE yielded a significantly higher specimen adequacy (65% in group A vs 92.6% in group B (Chi-Square < 0.0001, OR 6.72, 95% CI 2.45-18.38). The matching rate (accuracy) between ROSE diagnosis and final histopathological diagnosis was noticed in 61 out of 68 patients (89.7%, 95% CI 0.7993-0.9576). The Kappa coefficient correlations of matching rate between ROSE and final histopathological diagnosis of all lesion and in pancreatic lesions were 0.7558, (95% CI 0.625-0.887) and 0.7814, (95% CI 0.639-0.924), respectively.

Conclusions: EUS-FNA with ROSE significantly improve specimen adequacy and was associated with high diagnostic accuracy.