BACKGROUND: Endoscopic ultrasound-guided fine-needle biopsy (EUS-FNB) using larger, next-generation cutting needles is a minimally invasive method for the diagnosis of pancreatic lesions. Rapid on-site evaluation (ROSE) is employed to render preliminary diagnoses, ensure specimen adequacy, and triage tissue for ancillary testing and can be performed on FNB cores. Given the difficulty of pancreatic cytology and the novelty of ROSE with these larger cutting needles, this study was performed to evaluate discrepancies between ROSE and the final diagnosis to uncover challenging diagnostic areas.
METHODS: Final reports from pancreatic FNBs with ROSE between January 2019 and December 2021 were reviewed, and the ROSE and final diagnoses were compared. Cases were categorized into nondiagnostic (ND), negative for malignancy (NEG), atypical, neoplastic (NEO), suspicious for malignancy (SFM), and positive for malignant cells (POS). A major discrepancy was defined as an ND/NEG versus NEO/SFM/POS interpretation.
RESULTS: A total of 454 cases were identified. The ROSE versus final diagnosis breakdown was as follows: ND/NEG 18.7% versus 16.3%, atypical 6.4% versus 5.1%, NEO 10.8% versus 11.9%, SFM 4.4% versus 2.0%, and POS 59.7% versus 64.8%. The concordance rate was high at 96.9% with only 14 (3.1%) major discrepancies, which included 6 due to interpretive error, 3 due to sampling error, and 5 due to a combination of both. While the majority of lesions in the cohort were conventional ductal adenocarcinomas (76%), there was an over-representation of non-ductal tumors constituting major discrepancies (6/14; 42.9%).
CONCLUSIONS: ROSE using pancreatic EUS-FNB is possible and provides an accurate interpretation in most cases. Diagnostic challenges remain with non-ductal tumors.
