19 G nitinol needle versus 22 G needle for transduodenal endoscopic ultrasound-guided sampling of pancreatic solid masses: a randomized study.

BACKGROUND: The aim of this prospective multicenter study was to compare a flexible 19 G needle with nitinol shaft (19 G Flex) with a standard 22 G needle for transduodenal endoscopic ultrasound (EUS)-guided sampling of pancreatic head tumors. METHODS: Patients with pancreatic head tumors requiring tissue diagnosis were randomized into two arms: puncture with either a 19 G Flex needle or a 22 G needle. The primary end point was diagnostic accuracy for malignancy. The secondary end points were ergonomic scores, sample cytohistological quality, and complications. A 6-month follow-up was performed. RESULTS: 125 patients were randomized and 122 were analyzed: 59 patients in the 19 G Flex arm and 63 patients in the 22 G arm. The final diagnosis was malignancy in 111 patients and benign condition in 11. In intention-to-treat analysis, the diagnostic accuracy for malignancy of the 19 G Flex and 22 G needles was 69.5 % (95 % confidence interval [CI] 56.1 % - 80.8 %) vs. 87.3 % (95 %CI 76.5 % - 94.4 %), respectively (P = 0.02). In per-protocol analysis excluding eight technical failures in the 19 G Flex group, the diagnostic accuracy of the 19 G Flex and 22 G needles was not statistically different: 80.4 % (95 %CI 66.9 % - 90.2 %) vs. 87.3 % (95 %CI 76.5 % - 94.4 %; P = 0.12). Technical success was higher in the 22 G arm than in the 19 G Flex arm: 100 % (95 %CI 94.3 % - 100 %) vs. 86.4 % (95 %CI 75.0 % - 94.0 %), respectively (P = 0.003). Transduodenal EUS-guided sampling was more difficult with the 19 G Flex (odds ratio 0.68, 95 %CI 0.47 - 0.97). CONCLUSION : The 19 G Flex needle was inferior to a standard 22 G needle in diagnosing pancreatic head cancer and more difficult to use in the transduodenal approach.