EUS-guided through-the-needle biopsy (TTNB) sampling has been reported to improve diagnostic yield compared with cytology for the evaluation of pancreatic cystic lesions (PCLs). The number of macroscopically visible tissue samples needed to reach an adequate diagnosis is still unknown.

This is a retrospective, single-center study on consecutive patients with PCLs with risk features (cyst >3 cm, thickened wall, cyst growth during follow-up, and mural nodules) who underwent TTNB sampling. The capability of differentiating mucinous versus nonmucinous cysts, ability to obtain a cyst-lining epithelium, definition of the grade of dysplasia, and specific diagnosis of cyst histotype were evaluated for 1, 2, or 3 TTNB macroscopically visible specimens.

Sixty-one patients were evaluated. A 100% histologic adequacy was reached by 2 samples (P = .05 versus 1). Compared with cytology, 1 TTNB specimen improved the possibility of defining cyst histotype (P < .0001), whereas 2 specimens increased all 4 diagnostic categories (P < .003). Two specimens also increased diagnostic yield compared with 1 sample (P < .085). The collection of a third sample did not improve the value of any diagnostic categories. A specific diagnosis was reached in 74% of patients with 2 histologic samples. The diagnostic reliability of TTNB sampling compared with surgical histology was 90%, with a 22.9% rate of adverse events.

Two TTNB macroscopically visible specimens reached 100% histologic adequacy and a specific diagnosis in 74% of patients. The collection of a third specimen did not add any additional information and should be avoided to possibly decrease the risk of adverse events.