Gastric cancer (GC) is the fifth most commonly diagnosed malignancy and the third leading cause of cancer death worldwide [1, 2]. However, in Japan, the incidence of early GC (EGC) is increasing due to a mass-screening program initiated by the government, and now accounts for over 50% of patients with GC [3, 4]. Patients with EGC have an excellent survival outcome after surgical treatment, with 5-year survival rates of more than 90% [5,6,7,8,9,10]. Therefore, current research is focused on offering a better postoperative quality of life by developing function-preserving gastrectomy.

Pylorus-preserving gastrectomy (PPG) is a type of function-preserving gastrectomy. According to the Japanese Gastric Cancer Treatment Guidelines [11], PPG is recommended for cT1cN0 EGC in the middle-third of the stomach with the distal tumor border at least 4 cm proximal to the pylorus. Compared with conventional distal gastrectomy with Billroth I fashion reconstruction, PPG has a lower incidence of dumping syndrome, bile reflux, weight loss, and nutritional deficit [12,13,14,15]. PPG is currently laparoscopically performed and laparoscopic PPG (LPPG) has realized less invasive and function-preserving surgery for EGC in the middle stomach [15].

Recently, totally laparoscopic gastrectomy for EGC, which means that both dissection and anastomosis are intracorporeally performed, has been widely applied [16,17,18]. LPPG conventionally involves a small laparotomy through which the dissected stomach is pulled out and transected and an extracorporeal hand-sewing gastrogastrostomy (EHG) is performed, which is called laparoscopy-assisted PPG (LAPPG). However, we have applied an intracorporeally mechanical gastrogastrostomy (IMG) to totally LPPG (TLPPG) and reported the short-term outcomes of the delta-shaped anastomosis, which was initially developed for Billroth I fashion reconstruction after laparoscopic distal gastrectomy [19]. Furthermore, we originally developed a novel method of IMG, named the piercing method, to create an end-to-end anastomosis [20], while the delta-shaped anastomosis is a side-to-side anastomosis. These methods of IMG were safe and feasible in our initial small-scale studies [19, 20].

Although TLPPG, including IMG, is a beneficial procedure for patients, several concerns may be raised. First, IMG is technically demanding. Compared with EHG, which is a basic and easy procedure, IMG includes laparoscopic manipulation of the remnant stomachs, intracorporeal handlings of an endoscopic linear stapler, and mechanical anastomosis of the thick stomach wall, which involves some knacks and pitfalls. Thus, difficulties of IMG may increase anastomotic complications, such as anastomotic leakage. Second, IMG may be associated with gastric stasis, which is a distinctive early postoperative complication of PPG because IMG is a rigid anastomosis made of metal staples, multiple staple lines, and a side-to-side form in the delta-shape anastomosis, which may impair a favorable peristatic movement of the remnant stomach. Previously, we reported the incidence of Grade 2 or higher stasis was 10% in patients who underwent delta-shaped anastomosis, which was slightly higher than that in patients undergoing EHG [19]. Although TLPPG, including IMG, is possibly an ideal procedure for its lower invasiveness, whether IMG is truly safe has been unclear because large-scale studies that included patients undergoing TLPPG have not been conducted. In the largest-scale randomized controlled trial in South Korea, which demonstrated the safety of LPPG for EGC in the middle stomach, only 10% of the patients underwent IMG [21].

To elucidate whether IMG is associated with safe short-term outcomes, we conducted a retrospective non-inferiority study, which compared those between EHG in TLPPG and IMG in LAPPG. The findings obtained from this study will be useful in daily practice and serve as the most reliable reference for the safety of IMG.

Materials and methods

Patients

We collected the data of patients who underwent LPPG between April 2005 and October 2022 from our prospectively developed database (Fig. 1). The indication for LPPG was EGC located in the middle stomach, which had no lymph node metastasis (cT1N0). Diagnosis of EGC was based on the preoperative findings of gastroendoscopy, barium meal study, and endoscopic ultrasonography. Nodal involvement was determined by preoperative enhanced computed tomography. The study also included patients undergoing endoscopic submucosal dissection for EGC that had incurable factors, such as the tumor size, pathological type, pathologically positive margin, submucosal invasion, or tumor involvement of submucosal vessels. The distance between the distal tumor boundary and the pylorus was expected to be 4–5 cm. Tumor staging was evaluated according to the 8th edition of the International Union Against Cancer TNM classification system. Pathological information of the collected patients was obtained by reviewing their medical charts.

Fig. 1
figure 1

Flowchart of collecting patients eligible for this study. LPPG, laparoscopic pylorus-preserving gastrectomy; LAPPG, laparoscopy-assisted pylorus-preserving gastrectomy; TLPPG, totally laparoscopic pylorus-preserving gastrectomy; EHG, extracorporeal hand-sewing gastrogastrostomy; IMG, intracorporeal mechanical gastrogastrostomy

Full size image

Surgical procedures

Surgeons

Seven expert surgeons in gastric surgery, who performed more than 100 laparoscopic gastrectomies annually, participated in LPPG either as an operator or a supervisory assistant role. The six of seven surgeons had performed all types of reconstructions (EHG, the delta-shaped anastomosis, and the piercing method). The remaining one surgeon had not selected the piercing method in TLPPG.

Lymph node dissection

Lymph node dissection was a common procedure in LPPG, including EHG or IMG. Mobilization of the stomach and D1+ lymphadenectomy was performed as previously described [19, 22]. The lymph nodes at station No. 5 (supra-pyloric) were routinely left intact. The infra-pyloric artery and vein were routinely preserved to maintain a sufficient blood supply and drainage the blood to the pyloric cuff, and the root of the right gastric artery was preserved and transected just distal to the first branch.

Transection of the stomach and gastrogastrostomy

EHG in LAPPG

After dissection of the lymph nodes and mobilization of the stomach, the stomach was extracted extracorporeally through a mini-laparotomy at the upper abdomen (Fig. 2a). The stomach was transected 4 cm proximal to the pylorus and at the meeting point of the right and left gastroepiploic vessels using a linear stapler. Prior to transection of the stomach, we touched marking clips that were preoperatively placed proximal and distal to the tumor and confirmed that the clips were within the resected stomach. After transection, an EHG was extracorporeally made. The method of EHG was basically a two-layer anastomosis. Interrupted or continuous sutures for the mucosal layer were initially made followed by interrupted sutures for the seromuscular layer. Detailed differences in methods completely depended on the surgeons’ preferences.

Fig. 2
figure 2

a The stomach was extracted extracorporeally through a mini-laparotomy at the upper abdomen. After confirmation of marking clips by palpating them, the dissected stomach was transected and an EHG in an end-to-end fashion was created. b An IMG was created sequentially following laparoscopic dissection of the stomach. EHG, extracorporeal hand-sewing gastrogastrostomy; IMG, intracorporeal mechanical gastrogastrostomy

Full size image

IMG in TLPPG

IMG was represented (Fig. 2b) by the delta-shaped anastomosis (Fig. 3a–c) or by the piercing method (Fig. 3d–f). After lymph node dissection, the stomach was intracorporeally transected using an endoscopic linear stapler. Distal and proximal transection lines were almost fixed 5 cm proximal to the pylorus and at the meeting point of the right and left gastroepiploic vessels, respectively. Before transection of the stomach, we performed intraoperative gastroendoscopy and confirmed that marking clips that were preoperatively placed proximal and distal to the tumor were within the resected stomach [23]. After transection, an IMG was created. We initially applied the delta-shaped anastomosis, the details of which were previously reported [19]. The method is a side-to-side anastomosis, which results in slight torsion of the proximal and distal remnant stomachs. The piercing method was initiated in August 2014 at our institution [20]. Details of the procedures and instruments used differed among the surgeons and changed frequently during the development period. In this study, all patients undergoing the piercing method in different approaches by different surgeons from the first procedure were included. Whether the surgeons applied the delta-shaped anastomosis or the piercing method depended on the surgeons’ preferences. However, no surgeons routinely performed the two methods.

Fig. 3
figure 3

a A linear stapler was inserted through the gastrostomy positioned on the greater curvature of each gastric remnant, thereby twisting both the proximal and distal remnant stomachs to connect the posterior walls of each. b The staple line within the gastric lumen underwent assessment for defects and hemorrhage. The anastomotic opening was expanded to create an obtuse “V” shape along the staple line, while the stapler entry orifice was momentarily sealed by stay sutures. A linear stapler was employed to fully occlude the stapler entry orifice that had been simply closed. c After accomplishing the anastomosis, the staple lines on the severed ends of the proximal and distal remnant stomachs intersected nearly perpendicular to the staple lines that sealed the gastrostomy. d A linear stapler was passed through the gastrostomy on the greater curvature of each remnant stomach, a small gastrotomy was also made on the distal remnant stomach at the lesser curvature edge, and the linear stapler with a curved tip-type cartridge was advanced along the transection line to pierce the antrum using the gastrotomy in the lesser curvature. In the proximal remnant stomach, a posterior wall anastomosis along the transection line was performed with the linear stapler almost reaching the lesser curvature edge. e Closure of the stapler entry orifice was performed with a linear stapler, making temporal stay sutures of the staple entry orifice. Furthermore, the staple lines of the proximal and the distal remnant stomachs were severed and sealed using a linear stapler, including the small gastrotomy of the lesser curvature of the distal remnant stomach lifted using stay sutures. f The piercing method enabled the creation of an end-to-end gastrogastrostomy, using the full length of the transected edge in the antrum

Full size image

Evaluation

The patients’ baseline characteristics and surgery information were reviewed. Patient factors included age, sex, comorbidity, and body mass index. Tumor histopathological factors included tumor size, tumor location, and clinicopathological stage. Surgical factors included surgical procedure, surgical time, bleeding, postoperative hospital stays, and postoperative complications. The severity of postoperative complications was evaluated according to the Clavien–Dindo classification system [24].

This study was primarily designed as a retrospective non-inferiority test based on comparisons of the incidence of anastomosis-related complications (ARCs) between EHG and IMG. ARCs were defined as complications consisting of Grade 1 or more severe anastomotic leakage and gastric stasis, which are essentially different, but both may need a long hospitalization with fasting. The management of ARCs did not significantly differ by the type of anastomosis. ARCs are the most troublesome complications in PPG. Furthermore, a subgroup analysis to elucidate in which subgroup IMG was inferior to EHG in ARCs was performed. We also evaluated factors that were associated with ARCs by univariate and multivariate analyses. In these analyses, clinicopathological and surgical factors were input and whether there were independent factors that were associated with ARCs was explored.

Statistical analysis

The present study determined the safety of LPPG in terms of the frequency of ARCs. The study’s design considers the anticipated postoperative complications and aimed to assess the non-inferiority of IMG to EHG. On the basis of previous studies [21, 25,26,27], assuming that the incidence proportion of complications was 11% for both groups and setting the non-inferiority margin at 5% and a one-sided alpha level of 5%, the sample size required to achieve a statistical power of 80% for the non-inferiority test was 971 patients. For the incidence proportion of ARCs, point estimates in each group and point estimates of risk differences with confidence intervals are shown. The confidence intervals were calculated using the Farrington–Manning method.

Continuous variables were evaluated by Student’s t test and categorical variables by the Mann–Whitney U test or the chi-squared test, as appropriate. Values are shown as medians (range). A p value of < 0.05 was considered to indicate statistical significance. Analyses were performed using a logistic regression model for each category of patient characteristics, surgery-related variables, and postoperative variables. Odds ratios and 95% confidence intervals were calculated. Statistical analyses were conducted using the SAS software package (JMP Pro16, SAS Institute, Cary, NC, USA). Statistical analysis for non-inferiority was performed using R version 4.3.0 (Foundation for Statistical Computing, Vienna, Austria) (2023-04-21).

Results

Patients’ characteristics

A total of 1,106 patients with EGC underwent LPPG during the study period. After excluding patients who underwent intracorporeal hand-sewing anastomosis (n = 17), extracorporeal mechanical gastrogastrostomy (n = 26), and robotic PPG (n = 42), 1,021 patients were analyzed, comprising 488 and 533 patients who underwent LAPPG including EHG (Extra group) and TLPPG including IMG (Intra group), respectively (Fig. 1). The male-to-female ratio was almost equal. Approximately 10% of the patients in the Extra group and approximately 25% in the Intra group experienced endoscopic submucosal dissection as the first treatment for their gastric cancer (Table 1).

Table 1 Patients’ characteristics
Full size table

Pathologically, approximately 6% in the Extra group exhibited muscular or deeper invasive gastric cancer, while approximately 8% had lymph node metastasis. The Intra group demonstrated similar results. Proximal and distal surgical margins were adequate in the Intra group, with only two patients having positive margins (Table 1).

Surgery information

The surgery time of the Intra group was significantly longer than that of the Extra group. However, the blood loss and hospital stay of the Intra group were significantly less than those of the Extra group (Table 2).

Table 2 Surgical data and postoperative complications
Full size table

Non-inferiority of the intra group to the extra group regarding ARCs

The incidence proportion of ARCs in the Extra and Intra groups was 11.3% and 11.4%, respectively. The point estimate of the risk difference was 0.0017 (90% CI − 0.0313 to 0.0345). On the basis of these results, the 90% confidence upper limit for the risk difference was less than 0.05 (non-inferiority margin); hence, non-inferiority of IMG to EHG in the incidence of ARCs was statistically proven (Fig. 4).

Fig. 4
figure 4

Forest plot showing the risk difference for postoperative ARCs. ARCs, anastomosis-related complications

Full size image

Regarding anastomotic leakage, incidences in the Extra and Intra groups were low. However, the incidence in the Intra group was nearly three times higher than that in the Extra group, which presented no significant difference. For gastric stasis, the incidences in the two groups were very similar.

Subgroup analysis of ARCs

A forest plot demonstrating the risk difference of ARCs according to the pre-specified clinical factors is shown in Fig. 5. The findings revealed that there were no subgroups in which the Intra group was an inferior alternative to the Extra group.

Fig. 5
figure 5

Forest plot containing the risk difference of postoperative anastomosis ARCs according to the pre-specified subgroups. BMI, body mass index

Full size image

Risk factors of ARCs

Multivariate analysis revealed that female sex and diabetes mellitus (DM) significantly influenced the occurrence of ARCs (Table 3).

Table 3 Univariate and multivariate analyses to elucidate risk factors of anastomosis-related complications
Full size table

Incidence of other complications

The overall incidence of early Grade 2 or higher postoperative complications did not exhibit a statistically significant difference between the Extra and Intra groups. Only the incidence of postoperative wound infection in the Intra group was significantly lower than that in the Extra group (Table 2).

Discussion

In this retrospective non-inferiority study of the incidence of ARCs in IMG to EHG, we had three new findings. First, the incidence of ARCs in the Intra group was not inferior to that in the Extra group. Although the incidence of anastomotic leakage in the Intra group was approximately three times higher than that in the Extra group, there was no significant difference between them. The Intra group was not a significantly inferior alternative in any of the subgroups. Second, female sex and DM independently influenced the occurrence of ARCs. Third, the Intra group had less wound infection and was not significantly associated with other complications. Although this study was retrospective, it included over 1000 patients and was designed as a non-inferiority test. Therefore, the results of this study may be highly reliable and will be a good reference for the safety of IMG and the risks of ARCs in LPPG, which may encourage the introduction of TLPPG including IMG in daily practice.

ARCs defined in this study consisted of anastomotic leakage and gastric stasis, which are essentially different complications in cause and frequency. However, they commonly prolong postoperative hospitalization and require fasting for treatment. ARCs impair the nutritional condition and quality of life in patients, although LPPG aims to preserve the gastric function and decrease the impairment of nutrition. We were initially concerned that IMG might increase the incidence of ARCs because of its technical complexity, the slightly unnatural appearance of gastrogastrostomy, and a rigid anastomosis using titanium staples. This study clearly presented the non-inferiority of IMG to EHG in the incidence of ARCs. We confirmed that IMG is safe, similar to EHG, and ARCs after IMG do not result in more patients having a longer hospitalization and fasting period than that after EHG. However, the finding that anastomotic leakage of IMG was nearly three times more frequent than that of EHG should be analyzed further, although the difference was not statistically significant. In fact, the incidence of 1.1% in this study is slightly higher than that in previous studies [25]. Several staple lines or overlapping areas of staple lines on the thick wall of the antrum may be associated with the higher incidence of anastomotic leakage in IMG. However, an end-to-end EHG is a very simple procedure, the blood supply is usually kept even in the resected stomach, and the tension in the gastrogastrostomy is almost free. This discussion may imply the difficulty of intracorporeal mechanical anastomosis, and surgeons should be aware that a certain amount of experience is required.

Although the subgroup analysis did not reveal that IMG was a significantly inferior alternative, EHG seems to have some advantages to IMG in older patients and in patients with comorbidities. IMG forms a slightly complicated anastomosis compared with EHG, which forms a simple end-to-end anastomosis. The bifurcation and overlapping of stapling are fragile parts in gastrogastrostomy made by IMG. Such potentially fragile parts may be even weaker in older or compromised patients. Furthermore, the motor function of the remnant stomach may be associated with the patient’s age or general condition. Older patients generally have more comorbidities than younger patients. Thus, older age and comorbidities may be equivalent in the risk differences of ARCs.

ARCs mainly consist of gastric stasis (approximately 10 to 1), which is a distinctive complication in PPG. Some surgeons may never select PPG for EGC in the middle stomach, disliking such an unavoidable complication. Regarding the incidence of gastric stasis in this study, 10.5% or 10.9% is slightly higher compared with previous studies [25,26,27]. Several studies investigated the risk factors of gastric stasis after PPG [28, 29]. We also previously reported that DM and intraabdominal infection were risk factors of gastric stasis in LPPG [30]. Furthermore, this study included more female than male patients with gastric cancer than generally reported [1], and we found that risk factors of ARCs were female sex and DM. Although the mechanisms of gastric stasis after PPG are fully understood, we speculate that the thin muscular stomach wall of female individuals and the nervous dysfunction of the antral cuff or pylorus in patients with DM may inhibit food from going to the duodenum, leading to accumulation of food residue. We perform PPG to preserve the storage capacity of the stomach. However, if the remnant stomach cannot empty food sufficiently, food residue accumulates and patients suffer from abdominal fullness, nausea, or appetite loss, which corresponds to gastric stasis. Preserved stomach functions may depend on the potential abilities of the original stomach. The stomach-emptying ability of female individuals or patients with DM can be essentially inferior to that of male individuals or patients with glucose tolerance. Given that, gastric stasis may be unavoidable to some extent.

Regarding surgical information, our study showed that IMG required a longer surgery time than EHG but blood loss was minimal. In TLPPG, intraoperative gastroscopy for all patients to confirm marking clip placement and intracorporeal gastric transection are time-consuming procedures [23]. Furthermore, IMG itself, especially the piercing method, sometimes takes longer because the method includes several exclusive procedures, such as piercing the remnant stomach using a linear stapler and cutting multiple stapling lines [20]. Regarding complications other than ARCs, there were no complications of which incidences in IMG were significantly higher than those in EHG, and the incidence of wound infection in IMG was significantly lower than that in EHG. We were initially anxious about the high incidence of intraabdominal abscess in IMG because the remnant stomach is intracorporeally opened in the anastomotic process. However, IMG did not increase the incidence of intraabdominal abscess. Considering these findings, IMG is a beneficial procedure in terms of infection but time-consuming in nature.

Although the greatest strengths of this study are the number of patients and the non-inferiority design, there are several limitations. First, it was a single-institution retrospective study. However, over 1000 patients were included in this study, which represents the largest study to date to investigate PPG. Second, given the comparatively low prevalence of complications, the limited sample size of this study posed a challenge in conducting discrete non-inferiority analyses for each individual complication. To increase the number of events, we newly defined ARCs as combined complications that are associated with anastomosis and prolong hospitalization involving fasting. Third, the data collection period spanned a considerable duration, extending from 2005 to 2022, during which some surgical techniques and tools had notably improved. Additionally, between 2010 and 2014, both IMG and EHG procedures were performed based on surgeon preference, potentially introducing a patient selection bias.

In conclusion, TLPPG including IMG for EGC in the middle stomach is not inferior to LAPPG, including EHG in the incidence of ARCs. Furthermore, IMG reduces intraoperative bleeding, does not increase other postoperative complications, shortens the postoperative hospital stay, and further decreases the incidence of wound infection. These results will encourage surgeons to introduce IMG in TLPPG for patients with early middle gastric cancer. However, IMG requires a longer surgery time and might increase anastomotic leakage. Surgeons should pursue their skills to make the procedures of IMG smoother and to eliminate ARCs, especially anastomotic leakage at least. A further analysis focusing on the long-term functional outcomes of different types of anastomoses may be required to obtain additional information for selecting anastomotic techniques.