Multicenter, randomized controlled trial of EUS-guided fine-needle biopsy using a fork-tip needle with macroscopic or rapid on-site evaluation for pancreatic lesions (H2O trial).

Background and objectives: According to previous reports, EUS-fine-needle biopsy (FNB) with or without rapid on-site evaluation (ROSE) showed the nonsuperiority of EUS-FNB + ROSE over EUS-FNB. However, previous studies included various kinds of FNB needle. This might be a critical limitation due to heterogeneity. Therefore, the aim of the present multicenter, randomized controlled trial was to compare the diagnostic accuracy of the fork-tip needle with or without ROSE for pancreatic lesions.

Methods: In the ROSE group, if an adequate sample was obtained to diagnose by on-site evaluation, EUS-FNB was stopped. If cytological results were insufficient or indeterminate, EUS-FNB was repeated. In the macroscopic on-site evaluation (MOSE) group, if a 4-mm length of visible core tissue was obtained, EUS-FNB was finished. If visible core tissue was not obtained or was less than 4 mm in length, a second puncture was attempted.

Results: One hundred seventy-one patients were randomized, 85 to the ROSE group and 86 to the MOSE group. In the MOSE group, diagnostic sensitivity and accuracy were 94.4% and 91.8%, respectively, for visible core tissue and 80.6% and 70.0%, respectively, for red tissue. Finally, overall diagnostic sensitivity and accuracy were 97.1% and 95.3%, respectively, in the ROSE group and 95.8% and 95.3%, respectively, in the MOSE group. Although there were no significant differences regarding adverse events between groups, the mean number of punctures was significantly lower in the MOSE group than in the ROSE group (1.47 vs. 1.20, P = 0.0171).

Conclusions: EUS-FNB using a fork-tip needle for pancreatic lesions has high diagnostic yield even without ROSE.