Background
Since the technique of laparoscopy-assisted Billroth-I gastrectomy was first reported by Kitano and colleagues in 1994[
1], laparoscopic gastrectomy for cancer (LGC) has been gaining increasing popularity worldwide because it is associated with earlier patient recovery compared with open surgery[
2‐
4]. A national survey conducted by the Japan Society of Endoscopic Surgery (JSES) every 2 years has shown increasing use of laparoscopic procedures for gastric cancer in Japan[
5]. According to the 10th JSES survey, more than 7,300 patients underwent LGC in 2009, which equated to 25.9% of 28,600 patients with gastric cancer who underwent open gastrectomy, LGC or endoscopic treatment such as endoscopic mucosal resection or and endoscopic sub-mucosal resection in the same institutions. Laparoscopic distal gastrectomy (LDG) was the most commonly performed type of LGC (75.7% of operation). The survey also reported the incidence of post-operative complications in a total of 10,355 LDGs performed in 2008 and 2009[
6], with the most frequent being stomal stenosis (2.0%), followed by pancreatitis or pancreatic fistula formation (1.3%), anastomotic leakage (1.1%), wound infection, peritoneal abscess, bleeding, pneumonia, and ileus. This suggests that anastomosis-related complications are the most common complication subsequent to LDG. In the context of gastrointestinal reconstruction, the majority of anastomoses following LDG were performed by laparoscopy-assisted procedures through a mini-laparotomy incision of 60 to 70 mm in length made on the epigastrium. In such a laparoscopy-assisted distal gastrectomy (LADG)[
1] procedure, gastrointestinal reconstructive anastomosis is extracorporeally conducted in a limited working space with restricted vision, and it is possible that this may lead to increased risk of anastomotic leakage.
Although we originally began performing LADG, in the hope of overcoming the drawbacks of cumbersome reconstruction, we introduced intracorporeal (IC) stapled gastroduodenostomy and gastrojejunostomy in association with LDG in 2004[
7,
8]. We then developed two methods for IC reconstructive anastomosis following LDG: 1) IC Billroth I anastomosis (B-I) was used when there was no tension expected at the gastroduodenal anastomosis, and 2) IC Roux-en-Y anastomosis (R-Y) was used when there were some concerns about strain.
We report our experience and results of three kinds of gastrointestinal reconstructive anastomosis: LADG followed by extracorporeal hand-sewn anastomosis (LADG + EC), LDG + B-I , and LDG + R-Y.
Results
LGC was performed on 769 patients: 449 (58.4%) underwent LADG or pure LDG, while the remainder consisted of 160 (20.8%) laparoscopic pylorus-preserving gastrectomies, 48 (6.2%) laparoscopic proximal gastrectomies, 47 (6.1%) laparoscopic segmental gastrectomies, 39 (5.1%) laparoscopic total gastrectomies, and 26 (3.4%) laparoscopic wedge resections. In the group of 449 patients with LADG or pure LDG, 73 had LADG with EC anastomosis, and 376 had pure LDG with IC anastomosis (248 B-I and 128 R-Y). The methods of EC anastomosis for the 73 LADG operations were 56 B-I, 9 R-Y, and 8 B-II.
Patients’ characteristics of each reconstruction group
There were no significant differences in age, gender distribution, or body mass index (BMI) between the three reconstruction-method groups (Table
1). The mean age (mid-sixties), male:female ratio (around 15) and BMI (22 ± 3.1) were almost comparable for the three groups. The tumor locations were defined using the Japanese Classification of Gastric Carcinoma[
14] as being in the upper (U), middle (M) or lower (L) third of the stomach. Regarding both IC techniques, there were 0 U, 121 M and 127 L in the LDG + B-I group, and 20 U, 89 M and 19 L in the LDG + R-Y group; that is, there were significantly more tumors located distally in the LDG + B-I than in the LDG + R-Y group (
P < 0.01). However, for LADG, there was no significant difference in tumor location between the B-I (n = 56) and R-Y (n = 9) subgroups, probably because the number of operative cases of R-Y reconstruction in the LADG group was too small.
Table 1
Patient’s characteristics for three kinds of reconstruction
Age, years | 64 ± 11 | 65 ± 11 | 66 ± 10 |
Male : female | 48 : 25 | 147 : 101 | 90 : 38 |
BMI, kg/m2 | 22.3 ± 3.2 | 22.0 ± 3.1 | 22.6 ± 3.0 |
| Type of anastomosis | | |
B-I | R-Y | B-II |
Tumor location | | | | | |
Upper | 0 | 1 | 2 | 0 | 20 |
Middle | 24 | 4 | 6 | 121 | 89 |
Lower | 32 | 4 | 0 | 127 | 19 |
Intraoperative variables
The mean operating time was 261 ± 63 minutes for the LDG + B-I group, 333 ± 79 minutes for the LDG + R-Y group, and 339 ± 83 minutes for the LADG + EC group, respectively (Table
2). Operating time was significantly shorter for the LDG + B-I group than for the LDG + RY and LADG + EC groups (
P < 0.01), whereas there was no significant difference in operating time between these latter two groups.
Table 2
Intraoperative variables for three groups
Operation time, min | 339 ± 83d | 261 ± 63e | 333 ± 79d |
Blood loss, ml | 119 ± 108 | 47 ± 48e | 76 ± 80f |
The intraoperative blood loss was 47 ± 48 ml for the LDG + B-I group, 76 ± 80 ml for the LDG + R-Y group, and 119 ± 108 ml for the LADG + EC group. Blood loss was significantly less for the LDG + B-I group than for the other two groups (P < 0.01), and it was also significantly less for the LDG + R-Y group than for the LADG + EC group (P < 0.05).
Post-operative morbidity
Anastomotic leakage rates were 1.2% (3/248) for the LAD + B-I group, 1.6% (2/128) for the LAD + R-Y group, and 6.8% (5/73) for the LADG + EC group (7.1% (4/56) for B-I, 11.1% (1/9) for R-Y group with LADG and none (0/8) for B-II) (Table
3). Three of these patients required reoperation, and seven patients had collections that were treated by percutaneous drainage.
Table 3
Post-operative complications for three groups
Anastomotic leakage,%d | 5 (6.8) | 3 (1.2) | 2 (1.6) | 0.04 |
Anastomotic stenosis,% | 2 (2.7) | 1 (0.4) | 0 | 0.32 |
Gastric stasis,% | 5(6.8) | 3 (1.2) | 3 (2.3) | 0.06 |
Pancreatic fistula,% | 2 (2.7) | 12 (4.8) | 2 (1.6) | 0.40 |
Anastomotic stenosis was encountered in 0.4% (1/248) of the LDG + B-I group, none (0/128) of the LDG + R-Y group and 2.7% (2/73) of the LADG group (3.6% (2/56) for B-I, and 0% (0/17) for the other reconstruction types). All of the anastomotic strictures were successfully treated by endoscopic balloon dilatation.
Gastric stasis occurred in 1.2% (3/248) of the LDG + B-I group and in 2.3% (3/128) of the LDG + R-Y group, but was much higher at 6.8% (5/73) in the LADG + EC group (7.1% (4/56) for the B-I subgroup, 11.1% (1/9) for the R-Y subgroup and none for the B-II subgroup). Each case with gastric stasis improved spontaneously in 1 to 2 weeks.
Formation of PF (grades B and C of the ISGPF grading system) was found in 4.8% (12/248) of the LDG + B-I group, 1.6% (2/128) of the LDG + R-Y group and 2.7% (2/73) of the LADG + EC group. Although two of these patients had post-operative hemorrhage associated with pancreatic inflammation and secondary pseudoaneurysm of the gastroduodenal artery, both cases were successfully treated by endovascular coiling.
The length of post-operative hospital stay was significantly shorter for the LDG + B-I group (11.2 ± 0.7 days) than for the LDG + R-Y (17 ± 1.3 days) or LADG + EC (21.6 ± 11.2 days) groups (P < 0.05), whereas there was no significant difference between the latter two groups. This trend was similar to those for the operating time and intra-operative blood loss, suggesting that the surgical stresses represented by those parameters may be closely associated with the length of post-operative hospital stay.
Discussion
Laparoscopic surgery is being been increasingly used for the treatment of gastric cancer over the past two decades, especially in East Asian countries such as Japan, Korea, China ,and Taiwan. In general, LGC can be divided into laparoscopy-assisted and pure laparoscopic techniques. With laparoscopy-assisted gastrectomy, lymph-node dissection is performed laparoscopically, but the transection of the stomach and the anastomosis are performed thorough an epigastric mini-laparotomy. Performing the anastomosis in this narrow and restricted space is often difficult, especially on obese patients with thick abdominal walls or on patients with a small remnant stomach.
Although we initially began performing LADG in the hope of overcoming the drawbacks of cumbersome reconstruction, we introduced the use of LDG followed by IC stapled gastroduodenostomy and gastrojejunostomy in 2004[
11]. To investigate the feasibility and benefits of IC anastomosis, we compared our experience of using LADG and pure LDG in the current study, using a consecutive series of patients in our institution. We found that the rates for anastomotic complications, including leakage, stenosis, and gastric stasis, were significantly lower for either method of IC mechanical anastomosis (B-I or R-Y) after LDG than for LADG follwed by hand-sewn anastomosis (using any of the three methods). The rate of anastomotic leakage in the IC groups (1.3%) seems to be within the permissible level compared with other accounts using these techniques[
15,
16]. Similarly, the rate of of 6.8% for anastomotic leakage after LADG (7.1% for B-I, 11.1% for R-Y and 0% for B-II ) in the current study was comparable with other reports using LADG, with a rate of 7.8% (8.1% (7/87) for B-I and 0% (0/3) for R-Y) reported by Fujiwara
et al.[
17] and 5.3% (4/76 for B-I with no case for R-Y reconstruction) by Shimizu
et al.[
18]. The relatively high incidences of anastomotic leakage were possibly associated with the results obtained from initial experience of LADG in each institution.
In the current study, the LADG + EC group had the largest blood loss and longest operating time of the three groups tested. Because patients underwent LADG until May 2004, when we changed our strategy to pure LDG plus IC anastomosis[
8,
11], our LADG results might have been subject to some degree of learning-curve effect while we gained experience of laparoscopy-assisted surgical techniques. However, although this dataset does therefore have the drawback of different time periods when each surgical procedure was performed, the reduced blood loss seen with pure LDG may be also reflective of the small wound length required and avoidance of cumbersome anastomosis through a minilaparotomy.
The current study also indicates that pure LDG + B-I resulted in a significantly smaller volume of blood loss and shorter operating time than did pure LDG + R-Y. The consequent reduction in surgical stresses, including operating time and blood loss, and the lower incidence of post-operative complications seemed to be associated with the fact that the LDG + B-I group also had the shortest length of post-operative hospital stay. It should be noted that the mean length of hospital stay in the current study was rather longer than reported elsewhere[
19]; however, because the Japanese medical insurance system is structured differently from those in other countries, it is difficult to estimate the correct length of hospital stay based solely on surgical aspects.
In our institution, B-I reconstruction using the delta-shape method with linear staplers[
7] was our first choice of reconstruction after LDG, with R-Y reconstruction (also with linear staplers)[
8,
12] reserved for selected cases, including those for which tension would be expected on the reconstruction, those with reflux esophagitis or a hiatus hernia, and those with elderly or high-risk patients. As indicated by the numbers of each reconstruction method used (248 B-I reconstructions and 128 R-Y reconstructions in the current consecutive series), B-I was used for the majority of patients who underwent distal gastrectomy.
With regard to post-operative nutritional status, we already confirmed that decreases in body weight and food intake at 12 months post-operatively compared with pre-operative values were significantly less for the LDG + B-I group than for the LDG + R-Y group. In addition, there were fewer subjective reports from patients about their small stomach in the LDG + B-I group than in the LDG + R-Y group[
20]. Certainly, bile reflux into the remnant stomach and lower esophagus is one of the drawbacks for B-I reconstruction compared with R-Y reconstruction. It may be closely associated with the development of remnant gastritis, mainly caused by bile reflux[
21]. However, as we recently suggested, based on another nationwide survey, the risk of the development of remnant gastric cancer does not appear to be directly associated with the reconstruction method[
22].
The development of our strategy for digestive reconstruction after LDG is indicative of the trend toward IC mechanical reconstruction, which offers advantages in wound length and avoidance of tension through a mini-laparotomy during cumbersome anastomosis. For LDG + R-Y reconstruction, we were able to successfully use IC jejunojejunostomy, which saved a further 20 mm length for the umbilical wound[
12]. This also allowed for reduced manipulation of the bowel, and was particularly useful for obese patients, for whom access through a mini-laparotomy can be limited. Reconstruction was performed under continuous laparoscopic guidance, and the disorientating and time-consuming switch to open surgery was thus avoided. In addition, totally IC laparoscopic gastrectomy has been shown to lead to earlier bowel function recovery compared with laparoscopy-assisted and open resections[
23]. As indicated by others[
15,
23], IC anastomosis is more costly than EC anastomosis, because it requires three to four applications of endoscopic linear stapler cartridges for the anastomosis. We are currently trying to lower the cost by closing the entry hole using an IC hand-sewn technique instead of stapling[
24].
Competing interests
None of the authors have any conflicts of interest or financial ties to disclose.
Authors’ contributions
SL designed and conducted the study, analyzed the data, and helped to write the manuscript. NT helped to design the study, conducted surgical operations, and helped to write the manuscript. EN conducted surgical operations and helped to write the manuscript. TT, MK, KY, MH, and JO helped to design the study and helped to write the manuscript. KU is the principal investigator, and designed the study, assisted in writing, revising and editing the manuscript. All authors approved the final manuscript.