In the 1960s, bursectomy is seen as an essential component of radical surgery for serosa-involved gastric adenocarcinomas in Japan. However, surgical safety and oncologic benefits are necessary factors in order to make sure bursectomy as a potential useful therapeutic procedure in the gastric cancer surgery according to the viewpoints today. By the results of the previous study, safety of bursectomy with D2 lymphadenectomy strongly hinges on the experience of surgeons . With regard to the long-term survival outcomes, there was only one RCT that suggested that bursectomy had some survival benefits among the serosa-positive (pT3–T4) patients and without significant difference, the 3-year overall survival rate was 69.8 % for the bursectomy patients, in contrast to 50.2 % for the non-bursectomy group , and the 5-years follow-up results of this study existed similar results . On the other hand, other studies had totally opposite results, and they did not find survival benefits of bursectomy when compared with non-bursectomy surgery [10–12]. Moreover, in China, more than half of gastric cancer patients were with advanced stage tumors. Therefore, we assessed the outcomes of advanced gastric cancer patients who underwent D2 gastrectomy with bursectomy or who are not in a single institute of China. In this study, we found that the postoperative complications rate was comparable between patients with or without bursectomy, and D2 gastrectomy with bursectomy had benefits in short-term overall survival outcomes when compared with non-bursectomy gastrectomy, especially for patients who underwent total gastrectomy.
Bursectomy is a complicated and technique-dependent procedure, which may increase the surgical duration and account for more blood loss during the operations. The previous Japanese randomized controlled trial found that bursectomy was associated with an additional 27 min surgical duration and an additional 125 ml intraoperative blood loss compared with non-bursectomy . And another cohort study found that bursectomy procedure was associated with an additional 41 min operation times and an additional 65 ml intraoperative blood loss . Also in our study, increasing of the surgical duration can be discovered in bursectomy group than the non-bursectomy group. The extra time consuming of the operations was mainly because of the dissection of the anterior of mesocolon of transverse and capsule of the pancreas. However, long surgical duration and high intraoperative blood loss do not mean the unsafety of the surgical bursectomy procedures although with the potential injury of the vessels of the mesocolon of transverse. Blouhos et al.  found that the postoperative morbidity rate was 19.4 % for patients with bursectomy, and Imamura et al. , in their study, presented that the overall morbidity rate was 14.3 % for both the bursectomy and non-bursectomy. In our study, the incidence of postoperative complications was comparable with the two groups (bursectomy group 23.3 % vs. non-bursectomy group 17.8 %, p = 0.179). Meanwhile, the grade of complications according to the Clavien–Dindo surgical complications classification  between the two groups was also comparable (p = 0.759). Therefore, despite the fact that the bursectomy is a time-consuming procedure, the D2 lymphadenectomy with gastrectomy plus bursectomy can be performed safely in high-volume experienced centers or by experienced surgeons .
Of the postoperative complications, in details, gastrointestinal surgeons are most concerned about the possible damage of the pancreas and potential trend increase the incidence of pancreatic fistula formation and postoperative intestinal obstruction. Injury to the pancreatic parenchyma may occur when dissecting the pancreatic capsule and lead to the probable incidence of pancreatic fistula. Previous study reported that subclinical pancreatic fistula could occur in up to 10 % of the patients with the resection of the pancreatic capsule . And on the other hand, Imamura et al., in their study, observed that there was no difference in the incidence of pancreatic fistulas and amylase levels in the postoperative drainage fluid between the bursectomy group and non-bursectomy group . They concluded that pancreatic fistula may not be caused by the dissection of the pancreatic capsule but owing to the lymphadenectomy of those lymph nodes adjacent to the pancreas parenchyma. Blouhos et al. reported that the incidence rate of pancreatic fistula is only 4.2 % (3/72) . Generally, the resection of the pancreatic capsule is experienced cumulative procedures, and the experienced surgeons in an experienced center can rarely induce injury to the pancreas and reduce the potential incidence of pancreatic fistula .
Next, concerned postoperative adverse events are the possibility of intra-abdominal adhesion formation and the potential intra-abdominal intestinal adhesive obstruction. For the gastrectomy with bursectomy, adhesions to the mesocolon and pancreas may cause specific symptoms, such as delayed gastric emptying, afferent loop syndrome, or intestinal obstruction . The early stage postoperative intestinal obstruction usually occurs in about 1–2 weeks after the operation. However, postoperative intestinal obstruction can occur at any point in time during the postoperative period. Therefore, short-term postoperative observation is not enough to fully evaluate the incidence of postoperative intestinal obstruction. Long-term follow-up and careful observation in a large sample size cohort study is expected to draw the conclusion about postoperative intestinal obstruction of bursectomy.
The most important purpose of total bursectomy is to eliminate the cancer cells for the prevention of potential peritoneal relapse and improve the survival outcomes. However, evidence of survival outcomes decides whether “bigger surgery is better.” For the tumor that penetrates the serosa of the stomach, some surgeons hold the attitude that bursectomy cannot eliminate all disseminated free cancer cells because the bursa omentalis is not a closed space and bursectomy is unlikely to improve overall survival in patients with subserosa or serosa invasive cancers . Previous studies found that there was no statistical difference between the two groups in overall survival, and the bursectomy procedure was not recommended as a routinely procedure [5, 10–12, 24]. However, we also noticed that the interim analyses and the final reports of Osaka trial found that bursectomy surgery had some survival advantages in stage pT3-4 patients when compared with non-bursectomy surgery regarding the recurrence rate, recurrence type (peritoneal seeding recurrence) and the survival rates [9, 21]. Meanwhile, we noticed that there is no statistical difference between the recurrence rate and recurrence type between the two groups (Additional file 1: Table S1). Recurrence type is another important factor to evaluate the survival benefits for different resection strategies. These results maybe due to the short-term follow-up duration, which may not completely present the potential recurrence between two groups. There is no difference of survival outcomes between the two groups, but the bursectomy group had better curves than the non-bursectomy group (Fig. 1). The following may be the reasons for these results: the bursectomy may have some benefits in terms of prolonging the patients’ survival outcomes, but because of the sample size and limited follow-up duration result, the statistical difference did not appear. In view of these results, besides Japan and Korea, advanced stage gastric cancer account for about half of total gastric cancer patients [25–27], especially in China . Therefore, a well-designed large sample RCTs with long-term study to find out the details operation indications of the bursectomy for advanced gastric cancer patients are expected.
Interestingly, we noticed that the number of harvested lymph nodes was significantly higher in the bursectomy group than the non-bursectomy group (40.6 ± 17.5 vs. 25.4 ± 9.9, respectively, p < 0.001). Meanwhile, the number of positive lymph nodes was also higher in the bursectomy group than the non-bursectomy group (7.5 ± 8.7 vs. 5.9 ± 6.4, respectively, p = 0.045). Lymph node dissection strategy is one of the hottest topics regarding gastric cancer surgery. On one hand, because of the limitation of the retrospective study, the potential selection bias may be one reason, which results to the different lymph node results between the two groups. Although, we only include those patients who underwent surgical treatment by surgeons of gastric cancer surgical treatment team in our department, which aims to dimension the potential bias coming from operators. On the other hand, Greece scholar Blouhos et al. presented right side bursectomy as an access plane for completely en bloc dissection of lymph nodes . This may be another reason for the different lymph node result between the two groups. It is certain that both the Western and Eastern guidelines recommended at least 16 lymph nodes examined for gastric cancer surgery in the N stage determination . In our study, the non-bursectomy dissected 25.4 ± 9.9 lymph nodes, which is higher than the recommendation of the guidelines. Also, some study reported that the number of lymph node dissection is closely related with the prognosis of patients [31–34]. Previous study identified that for stage N2–N3 gastric cancer patients, harvesting at least 25 lymph nodes may represent a superior cutoff for radical gastrectomy and could yield better survival outcomes . Therefore, regardless of the fact that there existed a significant difference in the number of lymph nodes between the two groups, the number of examined lymph nodes in non-bursectomy is higher than the requirements of the present gastric cancer classification.
Meanwhile, different from the previous study, we included those patients who underwent total gastrectomy. Although, the results of this study found that there was no significant difference in the short-term survival outcomes between bursectomy and non-bursectomy groups. For those patients who underwent total gastrectomy, the survival benefit of the bursectomy has appeared. Note that despite that there is no significant difference in survival outcomes for distal gastrectomy patients or pT4 stage patients, the survival curves had demonstrated the separated trending between the two surgical strategies. As the previous description, sample size and short-term follow-up duration may be the reasons for these survival results. Also, upper third gastric cancers usually had poor tumor characteristics and survival outcomes than lower third gastric cancers . Besides, those lower third gastric cancers that underwent total gastrectomy are generally due to the poor tumor characteristics. These may be the main reasons why there exists a statistical difference between the two groups in total gastrectomy and not in distal gastrectomy. Despite the fact that this study set ups subgroup analysis in the resection patterns and conducts the multivariate survival analysis which identified that bursectomy or not is one of the independent prognostic risk factors, the potential selection bias of the retrospective study was existing and which must be emphasized here again. Regardless of the fact that previous and present evidence did not strongly prove that the bursectomy should be routinely performed, this technique-dependent procedure may be existing survival benefits in gastric cancer patients with specific characteristics, such as T stages, tumor locations, or other characteristics. Thus, well-designed RCTs are expected to clear these.