Recent investigations have reported that DGE should be considered a warning of IACs, such as PF or IAA [6, 8,14-16]. In fact, DGE is most often secondary to the presence of IACs (secondary DGE), and very seldom a primary event (primary DGE) [6, 7, 17]. Similarly, in the present study, most DGE patients (n=20, 87%) had IACs, which confirmed this association (p<0.001), whereas primary DGE was rare (n=3, 13%). As to surgical procedure, pylorus-preserving pancreaticoduodenectomy (PPPD)  and a postcolic route for the GE reconstruction  have been reported as typical risk factors for DGE. In our hospital, resection of pylorus ring with preservation of nearly the entire stomach (pylorus-resecting pancreaticoduodenectomy; PrPD), antecolic route have been adopted for GE reconstruction. Similarly, Kawai et al. reported that PrPD in which 95% of stomach was preserved resulted in very low incidence of DGE (18) . On the other hand, the reconstruction method of the pancreatic stump is closely associated with IACs, which have been frequently reported as risk factors for DGE [3,6]. Considering this relationship, a close association between the procedure of pancreatic reconstruction and DGE occurrence is expected. Although several resent meta-analyses reported DGE occurrence was comparable between PJ and PG, there seemed to be numerous variations in the performance of PG in these meta-analyses [1,9,11,12]. In the present study, both PJ and PG (invagination procedure) were performed uniformly in all patients and thus any reconstruction bias that might have influenced DGE was almost eliminated. Moreover, PJ was exclusively adopted for pancreatic reconstruction in SSPPD instead of PG since 2009. There was no bias in the selection of patients between PG and PJ even though the present study was retrospective, which adds to the validity of the present analysis. As expected, and this was the most striking result, DGE occurred more frequently in the PG group.
The frequent occurrence of DGE in PG would suggest that the surgical procedure of PG itself affected DGE occurrence. It is possible that the PG resulted in fixation of the subtotal stomach to the posterior wall, thereby disturbing gastric peristalsis.
In particular, PG in the present study was performed with an invagination technique, which contained more anatomical destruction than PG with duct-to-mucosa anastomosis and could lead to more severe disturbance of gastric peristalsis, in pancreatic fistula or other intraabdominal complications.
Next, as to DGE with IACs, safer pancreatic reconstruction with fewer IACs should decrease this type of DGE. Some surgeons prefer PG to PJ for patients at high risk of PF because some observational clinical studies (OCSs) reported a lower incidence of PF with PG [8,9]. However, these reports showed high heterogeneity, and, moreover, no high-quality, randomized, controlled trials (RCTs) have yet to provide adequate evidence of greater safety with PG than with PJ . Similarly, in the present study, there were no significant differences in the frequencies of PF and IACs between the PG and PJ groups, but DGE with IACs occurred more frequently in the PG group. As suspected, the present data suggested that the PG group was more vulnerable than the PJ group to the impact of DGE occurring by IACs. Especially with regard to PF, a strong association with DGE was shown in the PG group, but not in the PJ group. In the PJ group, 3 of 8 patients (37.5%) with PF had DGE, whereas all PG patients with PF were affected by DGE (P＝0.007). It is probable that, in patients with PJ, the increasing distance from the pancreatic anastomosis would reduce gastric paresis due to PF or peripancreatic inflammation.
Although pancreatic anastomosis should not be chosen based on prevention of DGE alone, PG appeared to have a tendency to induce DGE, and thereby resulted in patient frustration, the need for nutritional support, and prolonged hospital stay. When PG is adopted, surgeons should take care to prevent disturbance of gastric peristalsis, including the anastomotic procedure, by omitting the incision on the anterior gastric wall, or choosing a vertical incision instead of a horizontal incision .
Although the ISGPS criteria have enabled comparisons of DGE between investigators using the general definition, interpretation of DGE is sometimes confusing. For example, patients with IACs, including chylorrhea, postoperative hemorrhage, enterocolitis, and gastric ulcer bleeding, may require withdrawal of oral intake despite lacking gastroparesis. DGE was originally described as gastroparesis following PPPD ; therefore, in our opinion, these patients should not be categorized as having DGE, whereas some investigators might have interpreted such cases as having DGE. This confusion occurred because the ISGPS criteria did not mention the presence or absence of co-existing complications, exclusion criteria, and the method for evaluating gastroparesis, although the criteria are simple, objective, and clearly measurable. Refinement of the definition is warranted for further analysis of the etiology of DGE.
The limitation of the current study is that the number of patients was small and there was a slight difference in surgical devises or drugs between PJ and PG cohorts depending on the period. Accumulation of additional cases with minimal variations is needed to definitively characterize the risk for DGE between PJ and PG cohorts in the future.
In conclusion, the occurrence of DGE and its association with IACs was compared between different pancreatic reconstruction techniques. Intra-abdominal complications including PF were strongly associated with DGE. As to the pancreatic reconstruction, DGE developed more frequently in the PG than in the PJ. We speculate that PG itself predisposed patients to DGE by the fixation to the posterior wall and intra-abdominal complications affected patients with PG more, these resulted in frequent occurrence of DGE in PG.