Endoscopic Ultrasound–guided Fine-needle Biopsy With or Without Rapid On-site Evaluation for Diagnosis of Solid Pancreatic Lesions: A Randomized Controlled Non-Inferiority Trial

Background and AimsThe benefit of rapid on-site evaluation (ROSE) on the diagnostic accuracy of endoscopic ultrasound–guided fine-needle biopsy (EUS-FNB) has never been evaluated in a randomized study. This trial aimed to test the hypothesis that in solid pancreatic lesions (SPLs), diagnostic accuracy of EUS-FNB without ROSE was not inferior to that of EUS-FNB with ROSE.MethodsA noninferiority study (noninferiority margin, 5%) was conducted at 14 centers in 8 countries. Patients with SPLs requiring tissue sampling were randomly assigned (1:1) to undergo EUS-FNB with or without ROSE using new-generation FNB needles. The touch-imprint cytology technique was used to perform ROSE. The primary endpoint was diagnostic accuracy, and secondary endpoints were safety, tissue core procurement, specimen quality, and sampling procedural time.ResultsEight hundred patients were randomized over an 18-month period, and 771 were analyzed (385 with ROSE and 386 without). Comparable diagnostic accuracies were obtained in both arms (96.4% with ROSE and 97.4% without ROSE, P = .396). Noninferiority of EUS-FNB without ROSE was confirmed with an absolute risk difference of 1.0% (1-sided 90% confidence interval, –1.1% to 3.1%; noninferiority P < .001). Safety and sample quality of histologic specimens were similar in both groups. A significantly higher tissue core rate was obtained by EUS-FNB without ROSE (70.7% vs. 78.0%, P = .021), with a significantly shorter mean sampling procedural time (17.9 ± 8.8 vs 11.7 ± 6.0 minutes, P < .0001).ConclusionsEUS-FNB demonstrated high diagnostic accuracy in evaluating SPLs independently on execution of ROSE. When new-generation FNB needles are used, ROSE should not be routinely recommended. (ClinicalTrial.gov number NCT03322592.) The benefit of rapid on-site evaluation (ROSE) on the diagnostic accuracy of endoscopic ultrasound–guided fine-needle biopsy (EUS-FNB) has never been evaluated in a randomized study. This trial aimed to test the hypothesis that in solid pancreatic lesions (SPLs), diagnostic accuracy of EUS-FNB without ROSE was not inferior to that of EUS-FNB with ROSE. A noninferiority study (noninferiority margin, 5%) was conducted at 14 centers in 8 countries. Patients with SPLs requiring tissue sampling were randomly assigned (1:1) to undergo EUS-FNB with or without ROSE using new-generation FNB needles. The touch-imprint cytology technique was used to perform ROSE. The primary endpoint was diagnostic accuracy, and secondary endpoints were safety, tissue core procurement, specimen quality, and sampling procedural time. Eight hundred patients were randomized over an 18-month period, and 771 were analyzed (385 with ROSE and 386 without). Comparable diagnostic accuracies were obtained in both arms (96.4% with ROSE and 97.4% without ROSE, P = .396). Noninferiority of EUS-FNB without ROSE was confirmed with an absolute risk difference of 1.0% (1-sided 90% confidence interval, –1.1% to 3.1%; noninferiority P < .001). Safety and sample quality of histologic specimens were similar in both groups. A significantly higher tissue core rate was obtained by EUS-FNB without ROSE (70.7% vs. 78.0%, P = .021), with a significantly shorter mean sampling procedural time (17.9 ± 8.8 vs 11.7 ± 6.0 minutes, P < .0001). EUS-FNB demonstrated high diagnostic accuracy in evaluating SPLs independently on execution of ROSE. When new-generation FNB needles are used, ROSE should not be routinely recommended. (ClinicalTrial.gov number NCT03322592.)