Repeat EUS-FNA of pancreatic masses after nondiagnostic or inconclusive results: systematic review and meta-analysis

Background and Aims: EUS-guided FNA (EUS-FNA) is the criterion standard for the diagnosis of solid pancreatic lesions. Several studies assessed the role of repeat EUS-FNA (rEUS-FNA) after an inconclusive examination. Our aim was to evaluate the pooled diagnostic accuracy of rEUS-FNA after a nondiagnostic result. Methods: We conducted systematic research on electronic databases (Medline, PubMed, EMBASE) and a meta-analysis to obtain pooled sensitivity, specificity, positive and negative likelihood ratio, and diagnostic odds ratio. A summary receiver operating characteristic curve was used to calculate area under the curve. Subgroup analysis was used to assess the role of rapid on-site evaluation (ROSE). Results: Twelve studies (505 patients) were included. Sensitivity was 77% (66%-86%), specificity 98% (78%-100%), and positive and negative predictive values 99% (98%-100%) and 61 (60%-63%), respectively. At 73% of disease prevalence (pretest probability), positive rEUS-FNA increased the disease probability to 99%, whereas a negative result decreased the disease probability to 39%. The sensitivity was 83% (64%-93%) and specificity 98% (80%-100%) when ROSE was available and 65% (57%-73%) and 94% (31%-100%) when not available. The number needed to diagnose was 1.2 (1.1-2.3) and 1.7 (1.4-8.3) in ROSE-positive and ROSE-negative cases, respectively. The number of correctly diagnosed cases increased from 6 (1-7) to 8 (4-9) of 10 patients without and with ROSE, respectively. Conclusions: This study objectively substantiated the added value of rEUS-FNA for the diagnosis of solid pancreatic masses in cases of a previous nondiagnostic or inconclusive result. Moreover, our data suggested that ROSE may be beneficial in this setting, because it increased the proportion of definitive diagnoses.