Introduction

In recent decades, the incidence of gastric cancer in the upper third of the stomach has steadily increased worldwide [1, 2]. Function-preserving surgery for the treatment of early gastric cancer has mainly focused on minimizing postgastrectomy problems and improving the quality of life (QOL) after gastrectomy [3–5]. Proximal gastrectomy (PG) is occasionally performed to preserve the physiological function of the remnant stomach with the aim of maintaining a gastric reservoir for patients with early upper-third gastric cancer [6, 7]. For cT1cN0 tumors, the Japanese gastric cancer treatment guidelines [8] indicate PG and D1+(D1) lymphadenectomy for proximal tumors to preserve more than half of the distal stomach. However, after PG, the loss of the lower esophageal sphincter and the acute angle of His occasionally leads to reflux esophagitis through acid reflux and regurgitation, which impair postoperative QOL.

Many reconstructive procedures after PG have so far been reported, including esophagogastrostomy (EG) [6, 7], jejunal interposition (JI) [9, 10], jejunal pouch interposition (JPI) [11, 12], and double tract (DT) [13, 14]. Historically, EG was widely used for treating early upper-third gastric cancer because EG is a simple and easy reconstruction method [15]. However, because this procedure often leads to severe reflux esophagitis [16, 17], some surgeons instead perform total gastrectomy, while others select other reconstruction methods that do not cause severe reflux esophagitis, such as JI, JPI, and DT [9–14]. Recently, with the increasingly widespread application of laparoscopic gastrectomy as a less-invasive treatment [18, 19], EG has become more common as a reconstruction method after PG. Some studies have reported that an optimal additional procedure could prevent reflux esophagitis in patients undergoing EG [20–23]. However, some investigations have shown that JI, JPI, and DT reconstruction methods are advantageous with regard to their prevention of reflux esophagitis [9, 10, 13]. Although JPI reconstruction was found to be effective for preserving the gastric function after PG [11, 24], some studies have reported dilatation and stasis of the jejunal pouch [25, 26].

Thus, the most effective surgical reconstruction method after PG remains controversial, and the optimal procedure is currently unclear. Although many groups have studied reconstructive procedures for PG, to the best of our knowledge, no reviews have summarized the findings from all large studies. We believe that a summary of these results is necessary to establish the efficacy of PG for gastric cancer.

We herein review the short-term and long-term outcomes of PG for gastric cancer, focusing on the surgical outcomes, postoperative complications, endoscopic findings, and QOL according to the reconstructive procedures. This article is intended to review the current clinical data on the different types of reconstructive procedures following PG and to examine the effective and safe reconstructive procedures for PG.

Methods

An English literature search was performed using the PubMed database with the terms “proximal gastrectomy” and “gastric cancer” along with their synonyms or abbreviations for the years 2000 through 2014. Comparative studies including randomized controlled trials (RCTs) and case series were reviewed. Comparative studies of different reconstructive procedures including more than 10 patients in a single group and retrospective series including more than 10 patients published in peer-reviewed journals were selected. We excluded comparative studies between PG and total gastrectomy. The length of surgery, intraoperative blood loss, morbidity, reflux esophagitis, anastomotic stricture, gastric stasis, nutrition status, and QOL were examined. As a result, eight comparative studies and 15 case series were included in this review.

Surgical procedures for PG

Vagal nerve preservation

Preservation of the vagal nerve system was reported in 11 studies. Among these reports, five documented the preservation of the hepatic branch and pyloric branch of the vagal nerve, and six documented the preservation of the celiac branch as well (Tables 1, 3, 4, 5).

Table 1 Literature review of proximal gastrectomy for gastric cancer (EG alone)
Full size table

Pyloric drainage procedure

Pyloric drainage procedure was reported in 6 studies. Among these reports, one performed pyloromyotomy and the others performed pyloroplasty, such as the finger fracture method (Tables 1, 3, 4, 5).

Esophagogastrostomy

Esophagogastrostomy is a simple reconstruction method compared with the other reconstruction methods because it includes only one anastomotic site. The results of EG were reported in 17 studies (Tables 1, 2, 5, 6). An esophagogastric anastomosis was performed using circular stapling in 11 studies, linear stapling in three studies, and the hand-sewn method in two studies.

Table 2 Literature review of proximal gastrectomy for gastric cancer (EG alone)
Full size table

Zhang et al. [27] compared three types of EG: esophagogastric anterior wall end-to-side anastomosis (EA), esophagogastric posterior wall end-to-side anastomosis (EP), and esophagogastric end-to-end anastomosis (EE). They found that the EA procedure seemed to confer clinical benefits in terms of the postoperative QOL, specifically in the form of an improved meal intake, reduced gastroesophageal reflux, and improved body weight.

Recently, with the increasingly widespread application of laparoscopic gastrectomy as a less-invasive treatment [18], simplicity and ease of surgery are required for reconstruction methods. Kong et al. [28] and Hiki et al. [29] reported a fast and feasible double-stapling anastomotic technique, which is particularly useful for laparoscopic gastric surgery. Tsujimoto et al. [30] showed that a side-to-side (so-called “overlap”) anastomosis using a linear stapler after LPG was safe and feasible and did not require additional minilaparotomy; this procedure may result in less pain and favorable cosmetic outcomes.

Optimal additional procedures were reported for the prevention of reflux esophagitis in patients undergoing EG. Some studies reported PG with a subtotal or semicircular wrap of the abdominal esophagus by the residual stomach, similar to a Toupet fundoplication and showed that the fundoplication was useful for preventing reflux esophagitis after EG [21, 30–33].

Additionally, Ronellenfitsch et al. [34] and Aihara et al. [22] showed that EG using a narrow (3–4 cm) gastric conduit to limit its receptive capacity and to reduce postoperative reflux was capable of preventing reflux esophagitis. Moreover, Hosogi et al. [35] reported preparing a gastric tube of 35 mm width and making an esophagogastric tube anastomosis with pseudo-fornix with a no-knife linear stapler to prevent postoperative reflux esophagitis. Okabe et al. [36] reported their experience of PG with a hand-sewn esophagogastric anastomosis using a knifeless endoscopic linear stapler to contribute to an easier hand-sewn anastomosis and completion of the fundoplication under laparoscopy. Yasuda et al. [37] reported laparoscopy-assisted PG with EG with a reliable angle of His by placing a gastric tube in the lower mediastinum and showed its simplicity and low incidence of residual food and bile reflux.

Moreover, Chen et al. [38] reported PG followed by EG plus gastrojejunostomy to prevent bile reflux esophagitis and showed that this procedure reduced the incidence of reflux esophagitis, most likely through resolving the problem of delayed gastric emptying.

Jejunal interposition

Ten studies reported the results of JI reconstruction, which is generally performed as follows (Tables 3, 5, 6). A jejunal limb is brought up via either the antecolic or the retrocolic route and anastomosed with the esophagus and the remaining stomach. The length of the jejunal limb is 10–20 cm. An esophagojejunal anastomosis is performed with an end-to-side anastomosis, and a jejunogastric anastomosis is made with an end-to-side or side-to-side anastomosis on the anterior wall of the remnant stomach.

Table 3 Literature review of proximal gastrectomy for gastric cancer (JI alone)
Full size table

Tokunaga et al. [17] reported that JI with a 10-cm or shorter length jejunal limb should be performed after PG because it could prevent reflux esophagitis and was advantageous in evaluating the remnant stomach. Katai et al. [9, 10] reported that PG with JI was well tolerated, with excellent outcomes in the postoperative complications and lower mortality for patients with suspected early gastric cancer. Moreover, Kinoshita et al. [39] reported that LPG with JI had equivalent safety and curability, which may lead to faster recovery, better cosmesis, and improved QOL in the short-term compared to OPG with JI.

Jejunal pouch interposition

The results of JPI were reported in two previous studies (Tables 5, 6). JPI reconstruction is generally performed as follows. To construct a 10- to 15-cm-long reverse U-shaped jejunal pouch, a 25- to 35-cm jejunal limb is brought up via the retrocolic route and anastomosed side-to-end with the esophagus and end-to-side with the remaining stomach.

One RCT reported more favorable short-term and mid-term outcomes following JPI compared with JI after PG and significantly more frequent postoperative short-term morbidity with JI than JPI [11]. Moreover, the incidence of gastrointestinal complaints was more frequent in the JI group until 6 months after surgery. By contrast, the caloric intake was more favorable in the JPI group until 1 year after surgery. This RCT showed that JPI reconstruction may reduce the occurrence of gastrointestinal symptoms in the early postsurgery phase and provide a satisfactory volume of oral intake in the stable postsurgery phase.

Double tract

The results of DT were reported in three previous studies (Tables 4, 5, 6). DT reconstruction is generally performed as follows. A Roux-en-Y esophagojejunostomy is performed with a circular stapler, and the jejunal stump is closed with a linear stapler. Next, side-to-side gastrojejunostomy, 10 cm below the esophagojejunostomy, is performed using linear staplers. Finally, end-to-side jejunojejunostomy, 20 cm below the gastrojejunostomy, is performed with hand-sewn sutures.

Table 4 Literature review of proximal gastrectomy for gastric cancer (DT alone)
Full size table
Table 5 Comparative studies of different reconstructive procedures (EG vs JI, JI vs JPI, and EG vs DT)
Full size table
Table 6 Comparative studies of different reconstructive procedures (EG vs JI, JI vs JPI, and EG vs DT)
Full size table

One prospective study reported a significantly higher postoperative/preoperative body weight ratio in the JI group than in the DT group and a reflux esophagitis incidence of 10 % in both groups [14]. With respect to postprandial symptoms, no significant differences were observed between the two groups. Ahn et al. [13] reported that DT was a feasible, simple, and novel reconstruction method with exceptional postoperative outcomes in terms of preventing reflux symptoms.

Surgical outcomes

Length of surgery

The duration of surgery has been previously reported in 20 studies. In the reports on EG, the surgery lasted from 100 to 452 min, the median value ranged from 156.5 to 292 min, and the mean value ranged from 149.7 to 315 min (Tables 1, 5). In the reports on JI, the length of surgery ranged from 125 to 321 min, the median value ranged from 201 to 233 min, and the mean value ranged from 256.5 to 308 min (Tables 3, 5). In the reports on JPI, the mean length of surgery ranged from 311 to 335 min (Table 5). In the reports on DT, the length of surgery ranged from 115 to 357 min (Tables 4, 5).

Four previous studies compared EG and JI and reported that the operative time was shorter for EG than for JI [17, 33, 40, 41].

Blood loss

Blood loss results were previously reported in 19 studies. In the reports on EG, the amount of blood loss during surgery ranged from 0 to 2500 mL, the median value ranged from 119 to 280 mL, and the mean value ranged from 26 to 345.8 mL (Tables 1, 5). In the reports on JI, the amount of blood loss ranged from 0 to 776 mL, the median value ranged from 20 to 331 mL, and the mean value ranged from 287 to 456 mL (Tables 3, 5). In the reports on JPI, the mean blood loss during surgery ranged from 287 to 402 mL (Table 5). In the reports on DT, the amount of blood loss ranged from 12 to 357 mL (Tables 4, 5).

Our study compared EG and JI or JPI and reported that the intraoperative blood loss (mL) was significantly lower in the EG group than in the other groups (EG, JI, JPI: 179 ± 158, 393 ± 338, 402 ± 385, respectively, p < 0.05; EG vs JI: p = 0.0009; EG vs JPI: p = 0.0001) [33].

Morbidity

Morbidity was reported in 18 previous studies; 13 reports on EG (incidence range 0–35.7 %; Tables 2, 6), eight reports on JI (incidence range 9.4–32 %; Tables 3, 6), two reports on JPI (incidence range 5.3–25 %; Table 6), and two reports on DT (incidence range 11.6–20 %; Tables 4, 6). Morbidity was calculated using these described data. In three comparative studies, no significant difference was found in the early or late postoperative complications among the procedures [17, 37, 40]. However, one RCT reported that the postoperative morbidity was significantly more frequent in the JI compared with the JIP group, particularly with respect to complications affecting surgical anastomoses (p = 0.036) [11]. Our previous study reported that the JI and the JPI groups had significantly more early complications than the EG group, and 8.0 % of the patients in the JI group and 16.7 % of the patients in the JPI group developed intestinal obstruction during the follow-up [33]. This may reflect the complexity of the procedures using the jejunum. Overall, these findings imply that the JI and JPI reconstruction procedures are technically difficult and complex surgeries.

The results of anastomotic leakage were reported in 18 previous studies. Anastomotic leakage was documented in 14 reports on EG (incidence range 0–8.8 %; Tables 2, 6), six reports on JI (incidence range 0–15.8 %; Tables 3, 6), two reports on JPI (incidence range 0–5.3 %; Table 6), and two reports on DT (incidence rate 0 %; Tables 4, 6). Anastomotic leakage was calculated using these described data. In three comparative studies, no significant difference was found in anastomotic leakage among the procedures [33, 37, 40]. Interestingly, the number of anastomoses did not significantly affect the occurrence of anastomotic leakage.

Two comparative studies reported the survival data in this review, and there was no difference in the 5-year survival rate between the EG group and the JI group [17, 40].

Reflux esophagitis

Reflux esophagitis was reported and evaluated endoscopically in 16 previous studies. Reflux esophagitis was documented in 10 reports on EG (incidence range 9.1–35.3 %; Tables 2, 6), eight reports on JI (incidence range 0–15.8 %; Tables 3, 6), two reports on JPI (incidence range 8.3–15.8 %; Table 6), and three reports on DT (incidence range 0–25 %; Tables 4, 6). The incidence of reflux esophagitis was calculated using these described data. JI, JPI, and DT were associated with a relatively low frequency of reflux esophagitis. In two comparative studies, the frequency of reflux esophagitis in simple EG without additional procedures was higher than in other procedures [17, 33]. Some additional procedures, such as the fundoplication, use of a narrow gastric conduit, addition of gastrojejunostomy, and placement of a gastric tube in the lower mediastinum on EG, were performed to prevent reflux esophagitis. Three comparative studies reported that the frequency of reflux esophagitis in EG with additional procedures was similar to other procedures or that the degree of reflux esophagitis in EG with additional procedures was mild or moderate [21, 33, 37]. Our previous study showed that the frequency of reflux esophagitis (Los Angeles classification grades B and C) in patients with a more than 180-degree wrap of the remnant stomach around the esophagus during EG was 3.6 %, and reflux esophagitis could be prevented when the fundoplication was performed adequately [33]. Additional procedures such as the fundoplication could decrease the frequency of reflux esophagitis.

Anastomotic stricture

Anastomotic stricture was reported in 18 previous studies. Anastomotic stricture was documented in 13 reports on EG (incidence range 0–28.6 %; Tables 2, 6), seven reports on JI (incidence range 3.1–31.8 %; Tables 3, 6), two reports on JPI (incidence range 8.3–10.6 %; Table 6), and three reports on DT (incidence range 4.65–10 %; Tables 4, 6). Anastomotic stricture was calculated using these described data.

Esophagogastric anastomosis on EG was performed using circular stapling in 11 studies and linear stapling in three studies. The incidence range of anastomotic stricture using circular stapling was 2.6–28.6 % and that using linear stapling was 0–20 % (Tables 1, 2, 5, 6).

In six comparative studies, no significant difference occurred in anastomotic stricture among the procedures [11, 14, 21, 33, 37, 40]. Our previous study showed that anastomotic stricture occurred in 22.2 % of the patients with a less than 180-degree wrap and in 21.4 % of the patients with fundoplication of a more than 180-degree wrap after EG. No significant relationships existed between the degree of the wrap and the development of anastomotic stricture [33].

Food residues

With respect to complications arising after pouch reconstruction, some studies have reported pouch stasis resulting from dilatation of the jejunal pouch [25, 26].

Results pertaining to food residues were reported in five previous studies. Food residues were documented in two reports on EG (incidence range 18.2–21.8 %; Table 6), four reports on JI (incidence range 8.5–58.8 %; Tables 3, 6), two reports on JPI (incidence range 21.2–91.7 %; Table 6), and one report on DT (incidence rate: 48.9 %; Table 4). Food residues were calculated using these described data. EG was associated with a relatively low frequency of food residues. Yasuda et al. [37] reported that the frequency of food residues in JI was significantly higher than in EG. Takagawa et al. [11] reported that there was no significant difference in food residues between JI and JPI. They introduced the use of a 10-cm jejunal pouch, and this technique preserved the vagus nerve, helping to maintain motility in the stomach and prevent pouch stasis. However, in our previous study, although the use of a 10- to 15-cm jejunal pouch and preservation of the vagus nerves were similarly performed, residual food was found in 92 % of the patients in the JPI group, 22 % of those in the EG group, and 32 % of those in the JI group; it was significantly more frequently detected in the JPI group than in the other 2 groups (p < 0.0001) [33]. Moreover, Tokunaga et al. [41] collected and analyzed the subjective symptoms after PG using a questionnaire survey, and they reported statistically significant differences in the symptoms, indicating delayed emptying syndrome in the JI group compared with the EG group. They concluded that an interposed segment could disturb the food passage. These results suggest that some patients that undergo JP or JPI may have a low QOL due to food residues after gastrectomy.

Changes in body weight

Body weight loss is a defining characteristic of postgastrectomy syndrome and leads to impaired postoperative QOL. Changes in body weight were reported in 12 previous studies and documented in six comparative studies (Tables 2, 3, 4, 6). Body weight was calculated using these described data. Four comparative studies reported that postsurgical weight loss did not differ between the EG group and the JI group [40, 41], between the EG group and the DT group [21], or between the JI group and the JPI group [11]. However, our study reported that the rate of body weight loss was significantly lower in the EG group than in the JI and JPI groups at 6, 12, and 36 months postoperatively (p < 0.05) [33]. Nomura et al. [14] reported a significantly higher postoperative/preoperative body weight ratio in the JI group than in the DT group (p < 0.05).

Reflux symptoms

With respect to gastrectomy-associated symptoms, reflux symptoms such as heartburn and regurgitation considerably influence the QOL after PG. Reflux symptoms were reported in 18 previous studies and evaluated by the frequency of heartburn or regurgitation using QOL questionnaires or medical records. Reflux symptoms were documented in 12 reports on EG (incidence range 0–33.3 %; Tables 2, 6), eight reports on JI (incidence range 0–15.6 %; Tables 3, 6), one report on JPI (incidence rate 5.3 %; Table 6), and three reports on DT (incidence range 4.6–12.5 %; Tables 4, 6). Reflux symptoms were calculated using these described data. All six comparative studies reported that the frequency of reflux symptoms did not differ between the EG group and the JI group [37, 40, 41], between the EG group and the DT group [21], between the JI group and the JPI group [11], or between the DT group and the JI group [14]. Moreover, some studies on EG with additional procedures reported that the reflux symptoms were none or mild [31, 34–36]. Therefore, it was suggested that EG with additional procedures could prevent reflux symptoms.

Treatment with proton pump inhibitors following gastrectomy was reported in eight previous studies [17, 21, 33–37, 41]. Sakuramoto et al. [21] showed that reflux symptoms could be controlled by treatment with proton pump inhibitors.

Conclusion

We herein reviewed the short-term and long-term outcomes of PG for gastric cancer and found no significant difference in anastomotic leakage and anastomotic stricture among the procedures. The number of anastomoses did not significantly affect the occurrence of anastomotic complications. Moreover, although EG is a simple technique, the frequency of reflux esophagitis was higher in simple EG. Additional procedures such as the fundoplication, use of a narrow gastric conduit, and placement of a gastric tube in the lower mediastinum on EG could decrease the frequency of reflux esophagitis and reflux symptoms. Therefore, those additional procedures may be essential for preventing reflux esophagitis and preserving the QOL. However, all studies were small and could not adequately compare the reconstructive procedures. Therefore, prospective RCTs that involve a longer trial period and more institutions are needed to clarify the optimal reconstructive procedures after PG.