LAUSR

To the Editor: I n their randomized study of 206 patients undergoing pancreaticoduodenectomy (PD), Perinel et al report that nasojejunal early enteral nutrition (NJEEN) is associated with an overall increase in postoperative complications, compared with total parenteral nutrition (TPN). Because this study is of importance in a field where beliefs are stronger than proofs, we wish to raise some concerns related to methods and results, to be sure that conclusions are supported by data. First, we are surprised by a rather low inclusion rate in this study, with only a few patients screened over a 40-month inclusion period in 9 French university hospitals. Considering that such centers each might perform an average of around 30 PD per year, and as no highly restrictive exclusion criteria are reported, significant inclusion bias may hamper adequate interpretation of the results. Second, subjects were randomized during surgery, after termination of the anastomoses. Such timing for randomization is, in our opinion, a major bias in this study as it may have influenced patient selection. We would appreciate if the investigators could justify such a design, and discuss why randomization was not performed before surgery? Were any measures applied to ensure randomization of unselected patients? In addition, we wish to stress out a baseline imbalance between the 2 groups with regards to the risk of postoperative pancreatic fistula (POPF). Although some missing data might limit our interpretation of the results reported in Table 2 (the pancreas texture is described for only 196 patients), data from patients with normal and soft pancreas textures were grouped and then compared with those from patients with hard (ie, chronic) pancreatitis. The NJEEN group included 51 (52.5%) patients with ‘‘nonhard’’ pancreas and the TPN group only 40 (41.2%). The same kind of bias may exist because of more frequent small Wirsung ducts in the NJEEN group, compared with the TPN group (40.6 vs 33.7%). As the main risk factors for POPF are ‘‘nonhard’’ pancreas and a small Wirsung duct, we assume that postoperative complications might be directly explained by inhomogeneous groups, despite randomization. Adjustments for baseline imbalances in covariates influencing major outcomes between study groups could have been of interest. Another way to take into account these imbalances could have been a multivariate analysis of related data. Third, we are surprised that the authors do not report the use of somatostatin analogs and antibiotics in the early postoperative period, especially because their use, in unselected patients or in patients with specific risk factors, has proven beneficial in the prophylaxis of postoperative complications. Rates of postoperative drainages are also lacking in the manuscript, and could influence postoperative outcome especially in an era of enhanced recovery after surgery (ERAS) programs. Indeed, a recent multicenter randomized trial found beneficial effects of prophylactic active drainage on the severity of complications associated with pancreatic fistula, albeit such results should now be validated by future studies. Fourth, detailed information concerning postoperative complications (eg, types of infection), their timing and management (medical or surgical) would be highly informative to better evaluate benefits and risks of each intervention and thus possibly avoid misinterpretations. Moreover, as the use of prokinetic agents might also impact interpretation of the results on postoperative complication rates, we wonder whether the study protocol included specific management of gastroparesis and postoperative ileus or not. Finally, a subgroup analysis according to surgical techniques including the type of anastomoses (pancreaticojejunal or pancreaticogastric) would also be informative as this question remains unresolved. We hope that these comments could lead to further information in the field of early nutritional route post-PD.