Clinical and Endosonographic Predictors of the Number of EUS-FNA Passes of Solid Pancreatic Masses: A Prospective Multivariable Analysis in 300 Consecutive Cases

Abstract

Background: Factors affecting the number of passes needed to achieve a diagnosis are poorly understood. Our objective was to determine clinical and endosonographic predictors of the number of passes required for tissue diagnosis in patients with solid pancreatic masses. Methods: All patients presenting for the evaluation of solid pancreatic masses were prospectively evaluated by a single endosonographer. On site attending cytopathologist was present for sample interpretation. Data were collected on the clinical presentation, location of the pancreatic mass, size, presence or absence of EUS pancreatitis features (CP), and the number of passes performed. Both univariate and multivariable analysis were performed to determine factors predicting the number of passes to achieve a diagnosis. The number of passes was classified as 5 vs 1 to 4. The procedures were divided into three groups of 100 procedures each to adjust for the effect of experience on the number of passes (group1=first 100 cases). Results: Over a 3-year period, 300 consecutive pancreatic EUS-FNA were performed. (Mean age 62.6 years, male 64%). The median number of passes was 3 (range 1-12). Malignant masses were significantly less likely to require 5 passes as compared to benign masses (OR = 0.3; 95% CI=0.1-0.4) while suspicious/atypical masses were more likely to require 5 passes as compared to benign masses (OR = 3.6; 95% CI=1.4-9.5). In addition, subjects with changes of CP were significantly more likely to have more than four passes as compared to those without changes (OR=3.1; 95% CI=1.8-5.2). Subjects with prior tissue diagnosis attempt (p=0.10) and women (p=0.06) were more likely to have 5 passes. Also, being in group1 was associated with 5 passes as compared to group2 (p=0.02) and group3 (p=0.004). In the pre-EUS-FNA multivariable model, EUS finding of CP (p=0.0001) and prior tissue attempt (p=0.06) were independent predictors of >5 passes, while being in the third group (p=0.03) was an independent predictor of less number of passes required. When the EUS-FNA reading (benign, versus malignant, suspicious) was included in the model, EUS-FNA interpretation and being in the third group were independent predictors for the number of passes. Conclusion: More passes are possibly required in patients evaluated for solid pancreatic masses with a prior attempt at diagnosis and EUS features of pancreatitis. Experience seems to influence the number of passes overtime.