OBJECTIVES: Endoscopic ultrasound through-the-needle biopsy (EUS-TTNB) is a useful tool for differential diagnosis among pancreatic cystic lesions (PCLs). Cystic fluid cytology (CFC) is recommended by guidelines, but its diagnostic accuracy is about 50%. The aim of this meta-analysis is to assess the clinical impact of EUS-TTNB in terms of technical success (TS), histological accuracy (HA) and diagnostic yield (DY).
METHODS: Original studies in English language on EUS-TTNB were searched in MEDLINE and EMBASE until October 2019. Diagnostic accuracy of EUS-TTNB for identification of mucinous PCLs was calculated using individual diagnostic data of patients who underwent CFC and surgery.
RESULTS: Nine studies, including 454 patients who underwent EUS-TTNB, met the inclusion criteria for the meta-analysis. TS and HA of EUS-TTNB were, respectively, 98.5% (95% Confidence Interval [CI] 97.3%-99.6%) and 86.7% (95%CI 80.1-93.4). DY was 69.5% (95%CI 59.2-79.7) for EUS-TTNB and 28.7% (95%CI 15.7-41.6) for CFC. Heterogeneity persisted significantly high in most of subgroup analyses. In the multivariate meta-regression, cyst size was independently associated with higher DY. Sensitivity and specificity for mucinous PCLs were 88.6 and 94.7% for EUS-TTNB, and 40 and 100% for CFC. Adverse events rate was 8.6% (95%CI 4.0-13.1).
CONCLUSIONS: This meta-analysis shows that EUS-TTNB is a feasible technique that allows a high rate of adequate specimens to be obtained for histology; in about two-thirds of patients a specific histotype diagnosis could be assessed. The number of adverse events is slightly higher respect to standard EUS-FNA, but complications are very rarely severe.