LAUSR

We would like to thank the debaters from India for their careful reading of our article on randomized controlled trial (RCT) regarding pancreatic stump closure during distal pancreatectomy and their constrictive comments. Our study concluded that pancreaticojejunostomy (PJ) of the pancreatic stump during distal pancreatectomy (DP) does not reduce pancreatic fistula compared with stapler closure in a multicenter RCT. Actually in this study, all grades of the pancreatic fistula occurred in 38.7% of the patients who underwent DP with PJ, although we presume the risk reduction rate of 5% when the pancreatic stump was anastomosed with jejunal limb. First, the debaters comment that this discrepancy between estimated rate and actual rate of pancreatic fistula would be due to an element of inexperience for PJ. However, this comment from the debaters is not true and the debaters have to find our efforts to uniform the surgical level regarding PJ before the starting of this RCT. Indeed, in participating institutions, pancreatic surgeons participating in this trial have visited to the principal institution, Wakayama Medical University, to learn how to perform the anastomosis and how to perform the stump closure during December 2010 to June 2011(6 months). Moreover, the institutions in this RCT are definitely high volume centers of pancreatic surgery and the experiences of PJ have been maturated in Whipple operation. Therefore, we absolutely deny the comment of the debaters that the result of PJ in this study was due to an element of inexperience. The second comment is that current size of this RCT may be small. We have to refer other previous literatures to calculate sample size of RCT. Therefore, when pancreatic fistula rate in PJ of the pancreatic stump ranges from 0% to 8.6% in other previous literatures, 5% was expected as pancreatic fistula rate in the PJ group at the design of this protocol. Thus, pancreatic fistula rate was estimated to reduce from 25% to 5% by PJ of the pancreatic stump. On the basis of this hypothesis, the value of 124 patients as sample size was calculated to show a difference between the 2 groups at a power of 80% with a significance level of 0.05. The incidence of pancreatic fistula in other previous literatures might be incidental rate in underpowered studies. However, sample size was correctly calculated on the basis of the hypothesis. In addition, the authors also refer that the result of pancreatic fistula rate in our previous pilot study regarding PJ for pancreatic stump should be used to calculate sample size. Our previous study reported that PF occurred in 23.1% of the patients who underwent DP with PJ. However, all of 13 consecutive patients in the study prospectively underwent DP with en bloc celiac axis resection (DPCAR) for locally advanced pancreatic cancer. The tumors indicated for DP-CAR often require the pancreatic transection on the right side of the portal vein. To transect the pancreas using a stapler device on the right side of the portal vein would increase the risk of pancreatic fistula. Therefore, the previous prospective study evaluated whether PJ in patients with DP-CAR decrease pancreatic fistula. The background of our previous study was completely different from that of this RCT, and the result in our previous study was not referred for this RCT, as DP-CAR often increase morbidity or mortality compared with standard DP. The debaters suggest that dunking or invagination technique for PJ is better than duct-to-mucosa for nondilated duct at the resection site of the pancreas. We never recommend the dunking method for prevention of PF, and we strongly recommend PJ with duct-to-mucosa fashion, because decompression of pancreatic ductal back pressure by duct-to-mucosa PJ might prevent leakage of pancreatic juice from pancreatic stump. Moreover, 1 meta-analysis reported that invagination technique for PJ during pancreaticoduodenectomy does not decrease pancreatic fistula compared with duct-tomucosa. In addition, the latest RCT demonstrated that duct-to-mucosa for PJ during pancreaticoduodenectomy significantly decrease clinically relevant pancreatic fistula compared with invagination technique (3.1% vs 17.6%, P1⁄4 0.004). As the debaters point out, 3 of 4 patients who were switched to stapler closure or hand-sewn suture from the PJ group were due to an invisible main pancreatic duct at the resection site of the pancreas. However, pancreatic duct size did not affect the incidence of pancreatic fistula in PJ group of this RCT. Average pancreatic duct size in patients with pancreatic fistula was 2 mm 1.2 mm, and that in patients without pancreatic fistula was 2 mm 1.1 mm. There was no significant difference between patients with and without pancreatic fistula regarding pancreatic duct size. On the contrary, the frequency of transection line in neck, body, and tail of the pancreas was 20 (17%), 92 (77%), and 7 cases (6%), respectively. The incidence of pancreatic fistula regarding transection line of the pancreas was 4 of 20 (20%) in neck, 39 of 92 (42%) in body, and 4 of 7 (57%) in tail. Transection line of the pancreas also did not affect the incidence of pancreatic fistula in both groups. Last comment is regarding timing of randomization. The debaters suggest that randomization should be performed after start of surgery to exclude dissemination and ascertain the position of tumor. This RCT was performed by multicenter and central randomization and registration system for every participating institution was applied. Randomization after start of surgery was likely to disrupt the progress of surgery. Thus, we decided that randomization was done preoperatively in protocol committee. In addition, we calculated that this study required 124 patients (62 in each group) to show a difference between the 2 groups. Furthermore, calculating an estimated intraoperative withdrawal rate due to peritoneal dissemination, metastasis, or the change of procedure, of about 10%, it was decided to enroll a total of 136 patients (68 in each group) to meet the primary endpoint of this study at the setting of sample size. In fact, 123 patients (61 patients in stapler group and 62 patients in PJ group) were included for final analysis in Consort diagram for the trial. No problem due to timing of randomization occurred regarding sample size in this RCT. Once again, we thank the debaters for their comments on our article, which provide some useful suggestions. We will conduct furthermore high-quality RCTs to improve better outcomes of pancreatic surgery.