Introduction

Laparoscopic total gastrectomy (LTG) is a standard, widely accepted surgical procedure for early cancer of the esophagogastric junction or upper part of the stomach, whose incidence has recently begun to increase1,2,3,4. However, in-depth clinical research, the demand for functional preservation and the recommendations of recent guidelines have led to the gradual adoption of laparoscopic proximal gastrectomy (LPG) as a radical function-preserving surgical approach5. LPG can preserve more gastric function than LTG, thereby improving the postoperative nutritional status of patients. However, there is a high incidence of reflux esophagitis after traditional esophagogastric anastomosis, accompanied by an increased risk of anastomotic stenosis6. In 1998, Kamikawa reported a novel method of digestive tract reconstruction using esophagogastric anastomosis with the double-flap technique, which has a good anti-reflux effect. This is known as Kamikawa anastomosis. At our centre, traditional Kamikawa anastomosis has been further modified to reduce operational difficulty and shorten the surgical duration while preserving the anti-reflux effect. While studies have compared the double-tract reconstruction of LPG with traditional LTG, there have been few reports comparing Kamikawa anastomosis in LPG with LTG. Therefore, this study retrospectively analysed the clinical and pathological data of 268 patients with cancer of the esophagogastric junction or upper part of the stomach admitted to our centre between January 2016 and October 2022. We explored the short-term clinical efficacy of modified Kamikawa anastomosis in LPG and Roux-en-Y in LTG.

Materials and methods

Patients

This was a retrospective cohort study of the clinical and pathological data of patients with adenocarcinoma of the esophagogastric junction or upper part of the stomach who were admitted to our centre between January 2016 and October 2022. Among the initial 268 patients identified in our medical records, 26 had undergone LPG with modified Kamikawa anastomosis, and the remaining 242 had undergone LTG with Roux-en-Y anastomosis.

Initially, there were statistically significant differences between the two groups in age, preoperative albumin levels, maximum tumour diameter, degree of tumour differentiation and pathological stage (all P < 0.05) (Table 1). To ameliorate this, we matched patients in the Kamikawa and Roux-en-Y groups at a ratio of 1:1 using propensity score matching. Finally, 16 patients from the Kamikawa group and 16 patients from the Roux-en-Y group were included. The baseline data of these two groups were comparable, as shown in Table 2.

Table 1 Baseline data.
Full size table
Table 2 Baseline data after propensity score matching.
Full size table

Inclusion and exclusion criteria

Inclusion criteria: (1) Patients diagnosed with cancer of the esophagogastric junction or upper part of the stomach by preoperative pathological examination of gastroscopy biopsy samples; (2) Patients at clinical stage cT1-2N0M0 after preoperative enhanced computed tomography (CT) and endoscopic ultrasonography; (3) Patients without preoperative distant metastasis; and (4) Patients with no history of abdominal surgery.

Exclusion criteria: (1) Patients receiving preoperative neoadjuvant therapy; (2) Patients with severe cardiopulmonary dysfunction or poor nutritional status who showed poor tolerance to surgery; (3) Patients with concomitant malignant tumours; and (4) Patients with incomplete clinical and pathological data.

Surgical procedures

The same group of surgeons performed all surgeries in this study. Under intravenous and inhaled general anaesthesia, patients were placed in the supine split-leg position with the head slightly raised. The surgical site was routinely disinfected. The surgeon stood on the left side of the patient, the surgical assistant on the right and the laparoscope holder between the patient's legs. Using the five-port method, a 12 mm trocar was placed under the umbilicus as the observation port from which pneumoperitoneum was established, with pressure maintained at 12–15 mmHg (1 mmHg = 0.133 kPa). Then, a 12 mm trocar was placed 2 cm below the costal margin of the left anterior axillary line, and a 5 mm trocar was placed 2 cm above the level of the umbilicus at the left midclavicular line to establish the operating ports. As additional operating ports, 5 mm trocars were placed at corresponding sites on the right side. Laparoscopic exploration was performed to identify the tumour location, size, degree of infiltration and relationship with surrounding organ tissue.

Roux-en-Y group

With patients in the Roux-en-Y group, we performed total gastric dissociation and D1 + or D2 lymph node dissection. Splenic lymph node dissection was performed when there was splenic lymph node enlargement but was not conducted routinely. We then transected the oesophagus and removed the entire stomach. The jejunum was cut off 15–20 cm from Treitz's ligament, and jejunal oesophageal anastomosis was performed using the distal end of the jejunal stump. A jejunojejunostomy was carried out with the proximal end of the jejunum 30–40 cm from the jejunal oesophageal anastomosis (Fig. 1).

Figure 1
figure 1

The jejunal oesophageal anastomosis.

Full size image

Kamikawa group

With patients in the Kamikawa group, we performed proximal gastric dissociation and D1 + lymph node dissection. Splenic lymph node dissection was performed when there was splenic lymph node enlargement but was not conducted routinely.

The tumour was intraoperatively localised under gastroscopy. After the oesophageal hiatus was opened, the oesophagus was fully dissociated and dissected, and the posterior wall 5 cm away from the oesophageal stump was marked with gentian violet. A small subxiphoid incision was made to pull out the stomach, and the proximal stomach was severed with the linear cutter stapler at a distance of 3 cm from the distal end of the tumour, followed by the interrupted suture of the gastric stump for reinforcement. An 'I'-shaped mark [(2.5–3) cm × 3.5 cm] was marked with gentian violet on the anterior wall of the gastric remnant (approximately 1.5 cm from the upper incisal margin) near the lesser curvature of stomach to match the width of the seromuscular flap with the oesophageal diameter; the upper edge of the seromuscular flap must be parallel to the incisal margin of the upper end of the gastric remnant (Fig. 2). The submucosal layer was dissected and dissociated from the muscular layer along the mark to prepare the seromuscular flap while protecting the integrity of the seromuscular flap and gastric mucosa. At this point, the muscle flap should be vertically pulled upwards by the assistant to create tension, followed by separation by the surgeon using an electrotome to ensure the complete separation of the submucosa and muscle layer (Fig. 3). The submucosal layer and mucosal layer were cut open at the lower edge of the seromuscular flap for anastomosis; the cut width must be equivalent to that of the oesophagus. The gastric remnant was then placed into the abdominal cavity. The lower segment of the oesophagus, marked with gentian violet, was continuously sutured and fixed to the upper edge of the seromuscular flap using a 3–0 5/8 curved endoscopic suture (Fig. 4). After the oesophageal stump was opened with an ultrasound knife, the posterior wall of the oesophageal stump opening and the upper edge of the anastomotic stoma were intermittently sutured with two sutures on the left and middle for fixation (Fig. 5). The next step was the continuous suture of the entire layer of the posterior wall of the oesophageal stump with the gastric mucosa and submucosa at the upper edge of the anastomotic stoma using a 3–0 barbed suture from left to right until the right edge (Fig. 6). The suture needle was simultaneously threaded through the mucosa of the anterior wall of the oesophagus and out of the outer membrane for the subsequent suture of the seromuscular flap. The entire anterior wall of the oesophageal stump was then continuously sutured with the entire stomach layer at the lower edge of the anastomotic stoma using another 3–0 barbed suture from left to right until the right edge (Fig. 7). The reserved barbed suture was threaded through the serosal layer at the lower right corner of the seromuscular flap. The lower ends of bilateral seromuscular flaps were crossed and fixed on the anterior wall of the stomach below the midpoint of the anastomotic stoma. Another suture from the right side to the intersection point of bilateral seromuscular flaps was placed using the reserved barbed suture, followed by sutures for the seromuscular flap at the anastomotic stoma and the left oesophageal 'Y'-shaped edge upwards (Fig. 8). Another barbed suture was used to suture the seromuscular flap at the anastomotic stoma and the right oesophageal 'Y'-shaped edge; while the anastomotic stoma was being covered, the renconstruction of modified Kamikawa anastomsis was completed (Fig. 9). Finally, the condition of the anastomotic stoma was examined under gastroscopy to observe the presence of any stenosis at the stoma.

Figure 2
figure 2

An 'I'-shaped mark [(2.5–3) cm × 3.5 cm] was marked with gentian violet on the anterior wall of the gastric remnant (approximately 1.5 cm from the upper incisal margin) near the lesser curvature of stomach to match the width of the seromuscular flap with the oesophageal diameter; the upper edge of the seromuscular flap must be parallel to the incisal margin of the upper end of the gastric remnant.

Full size image
Figure 3
figure 3

The muscle flap should be vertically pulled upwards by the assistant to create tension, followed by separation by the surgeon using an electrotome to ensure the complete separation of the submucosa and muscle layer.

Full size image
Figure 4
figure 4

The lower segment of the oesophagus, marked with gentian violet, was continuously sutured and fixed to the upper edge of the seromuscular flap using a 3–0 5/8 curved endoscopic suture.

Full size image
Figure 5
figure 5

The posterior wall of the oesophageal stump opening and the upper edge of the anastomotic stoma were intermittently sutured with two sutures on the left and middle for fixation.

Full size image
Figure 6
figure 6

The entire layer of the posterior wall of the oesophageal stump was then continuously sutured with the gastric mucosa and submucosa at the upper edge of the anastomotic stoma using a 3–0 barbed suture from left to right until the right edge.

Full size image
Figure 7
figure 7

The entire anterior wall of the oesophageal stump was then continuously sutured with the entire stomach layer at the lower edge of the anastomotic stoma using another 3–0 barbed suture from left to right until the right edge.

Full size image
Figure 8
figure 8

Another suture from the right side to the intersection point of bilateral seromuscular flaps was placed using the reserved barbed suture, followed by sutures for the seromuscular flap at the anastomotic stoma and the left oesophageal 'Y' -shaped edge upwards.

Full size image
Figure 9
figure 9

While the anastomotic stoma was being covered, the renconstruction of modified Kamikawa anastomsis was completed.

Full size image

Outcome measures and evaluation criteria

The outcome measures were: (1) Intraoperative data, including the surgical condition, total surgery duration, intraoperative blood loss volume and digestive tract reconstruction duration; (2) Postoperative recovery period, measured as the time to first postoperative anal exhaust, the time to first postoperative liquid food intake and the length of the postoperative hospital stay; (3) Postoperative complications including intestinal obstructions, lymphatic stenosis, abdominal bleeding, anastomotic fistula and pulmonary infections; (4) Follow-up data, including follow-up duration, nutritional status after discharge, presence of reflux esophagitis at follow-up appointments and anastomotic condition.

The evaluation criteria were: (1) Nutritional status assessment comprising body mass index (BMI) and Nutritional Risk Screening 2002 (NRS 2002) score7; (2) Oesophageal reflux symptom score on the Gastroesophageal Reflux Disease (GERD) scale8; (3) Reflux esophagitis diagnosis by gastroscopy; (4) The extent of the lesion, evaluated using the Los Angeles grading system9; (5) The condition of the anastomotic site, determined using upper gastrointestinal angiography.

Follow-up

All patients were followed up after discharge to assess their nutritional status, reflux esophagitis and anastomotic site status. Follow-ups were performed at outpatient visits, telephone consultations and/or online using WeChat. Each patient was followed up for at least 12 months. We gathered follow-up data from the records up to October 2023.

Statistical analysis

Normally distributed data were expressed as x ± s and analysed using t-tests for inter-group comparisons. Data with skewed distributions were expressed as M (range), and inter-group comparisons were analysed using Mann–Whitney U tests. Count data were expressed as absolute numbers, and inter-group comparisons were made using X2 or Fisher's exact tests. The P-values of < 0.05 were considered statistically significant.

Ethical approval

This study was reviewed and approved by the Ethics Committee of the Second Affiliated Hospital of Fujian Medical University. The requirement for informed consent was waived by the Ethics Committee considering the retrospective design of the study. All analysis were done in accordance with relevant guidelines and regulations.

Results

Surgical conditions

All of the LPG and LTG performed were successfully completed. As shown in Table 3, the digestive tract reconstruction times in the Kamikawa group and Roux-en-Y group patients were 93.0 (74.0–111.0) min and 39.7 (35.1–46.2) min, respectively. The between-group difference in these times was statistically significant (P < 0.05). The mean surgery durations for the Kamikawa group and the Roux-en-Y group were 195.9 (± 25.0) min and 228.3 (± 59.8) min, respectively, with no significant difference between the two groups. The intraoperative blood loss volumes were 22.0 (16.0–27.0) mL and 23.0 (11.0–48.0) mL, respectively, for the two groups, with no significant difference.

Table 3 Intraoperative condition.
Full size table

Postoperative recovery

The time to first postoperative anal exhaust, time to first postoperative liquid food intake and length of postoperative hospital stay were 2.0 (1.0–3.0) d, 4.0 (3.0–5.0) d and 9.0 (8.0–10.0) d in the Kamikawa group and 2.0 (1.0–5.0) d, 3.5 (2.0–6.0) d and 9.0 (7.0–12.0) d in the Roux-en-Y group (Table 4). No statistically significant differences were found between the groups (all P > 0.05).

Table 4 Postoperative condition.
Full size table

The incidence of postoperative complications in the Kamikawa group included one case of intestinal obstruction, one case of anastomotic stenosis, and no cases of anastomotic bleeding, anastomotic fistula, or pulmonary infection. In contrast, the Roux-en-Y group had one case of postoperative intestinal obstruction and one case of pulmonary infection, with no instances of anastomotic stenosis, anastomotic bleeding, or anastomotic fistula. No statistically significant differences in these indicators between the two groups were detected (all P > 0.05). The patient in the Kamikawa group who developed anastomotic stenosis during the perioperative period improved after one cycle of endoscopic balloon dilation (15 mm, 8ATM). All other postoperative complications in both groups improved after conservative treatment.

Follow-up outcomes

All patients were followed up 6 and 12 months after surgery. Table 5 shows the follow-up data for the two groups. The mean BMI at 6 and 12 months were 22.9 (± 3.0) kg/m2 and 20.8 (± 2.2) kg/m2 in the Kamikawa group and 23.1 (± 3.0) kg/m2 and 20.3 (± 2.2) kg/m2 in the Roux-en-Y group, respectively. The NRS 2002 scores at 6 and 12 months after surgery were 2 (1–2) points and 2 (1–3) points, respectively, in the Kamikawa group. In the Roux-en-Y group, the scores were also 2 (1–2) points at 6 months and 2 (1–3) points at 12 months after surgery. At both follow-up times, the between-group difference in above data was statistically significant (P < 0.05). Furthermore, the GERD scale scores at 6 and 12 months were 3 (2–4) points and 3 (2–4) points, respectively, in the Kamikawa group, and 3 (2–4) points and 3 (2–4) points, respectively, in the Roux-en-Y group. One year after surgery, there was one case of ≥ grade B reflux esophagitis in the Kamikawa group and two cases in the Roux-en-Y group. The difference between the two groups was not statistically significant differences between groups. No anastomotic stenosis was found on postoperative upper gastrointestinal angiography in any of the patients.

Table 5 Follow-up condition.
Full size table

Discussion

A gradual increase in the incidence of gastric cancer in the upper part of the stomach in recent years has led to increased attention to this condition. A national survey in South Korea found that the incidence of proximal gastric cancer increased from 11.2 to 16% in 201410. Until recently, a total gastrectomy was the standard treatment for gastric cancer in the upper part of the stomach. However, with medical developments and improved early gastric cancer detection rates, surgeons are now expected to maximise the preservation of gastric function during radical tumour resection. To meet this need, new methods of proximal gastrectomy that preserve gastric function have been developed. Current proximal gastrectomy techniques for early gastric cancer can effectively improve the postoperative nutritional status of patients without negatively affecting the long-term survival of the patient. However, the surgery may still disrupt the anti-reflux mechanism of the cardia and the rate of reflux esophagitis after traditional esophagogastric anastomosis is between 21.8 and 71.6%. This seriously affects the quality of life of patients after surgery11. Furthermore, based on the experience from bariatric surgery, sleeve gastrectomy leads to the disruption of the lower esophageal sphincter, resulting in de novo GERD12.

In 1998, Kamikawa reported a novel double-flap technique (DFT) for digestive tract reconstruction, now known as Kamikawa anastomosis. The technique is suitable for gastric cancer in the upper part of the stomach in patients with an estimated postoperative residual gastric volume of > 50%. Kamikawa anastomosis is increasingly utilised due to its excellent anti-reflux effect and reduced risk of postoperative anastomotic leakage13,14,15,16. Constructing a seromuscular flap to cover the lower oesophagus and anastomotic site increases the pressure on both sites. Like a reconstructed cardia, this acts as a one-way valve, thereby achieving the desired anti-reflux effect. After digestive tract reconstruction, just one anastomotic orifice is created, which is then covered by a seromuscular flap, reducing the risk of postoperative anastomotic leakage. The procedure also carries a cost-saving benefit as the anastomotic orifice made during Kamikawa anastomosis is sutured manually. However, the procedure is more complicated than conventional digestive tract reconstruction methods, requiring longer surgical durations and greater skill on the part of the surgeon, particularly in the endoscopically-guided suture technique. In addition, improper seromuscular flap preparation can potentially induce ischemia of the flap and stenosis of the anastomotic site. Consequently, the popularity of this surgical technique is somewhat limited. To address some of the issues with this technique, without reducing the anti-reflux effect or the extent of resection, our centre has made certain modifications to Kamikawa anastomosis.

Our centre's approach to Kamikawa anastomosis incorporates the following modifications. Firstly, we dissociate the left outer lobe of the liver for flipping and suspension to reduce obstructions and improve the extent of the operative field, facilitating easier operational access. This modification is simple and easily applied, requires no additional equipment, and does not impose any trauma on the liver. Secondly, the width of the I-shaped seromuscular flap is changed to 2.5–3 cm to match the oesophageal stump opening, reducing the risk of postoperative anastomotic stenosis. Thirdly, the I-shaped upper edge of the seromuscular flap is made parallel to the cutting edge of the upper end of the gastric stump, bringing the flap closer to the lesser curvature of the stomach with good blood supply. This improves the blood flow to the flap and ensures that it performs the required anti-reflux function. Fourthly, the posterior wall of the oesophageal stump opening and the upper edge of the seromuscular flap are intermittently sutured with two stitches, beginning on the right side and the middle for fixation, facilitating subsequent continuous suturing. Fifthly, continuous suturing of the posterior wall of the oesophageal stump, anastomotic site and seromuscular flap using barbed sutures avoids the need for slower intermittent sutures and reduces surgical time. Lastly, the modified technique creates a large false fornix of the gastric fundus on the left side of the oesophagus to achieve a better anti-reflux effect. In cases of excess suture tension of the seromuscular flap, the flap can be directly sutured to the same side of the oesophageal wall to reduce such tension and prevent postoperative anastomotic stenosis.

In the present study, the application of our modified Kamikawa anastomosis reduced surgical durations by approximately 50% compared to the traditional technique. We found that the digestive tract reconstruction time in the Kamikawa group was longer than that in the Roux-en-Y group, which may be explained by the complexity and difficulty of the surgical mode. However, there was no significant difference in total surgical time between the two groups, indicating that LPG does not prolong the surgical duration compared to traditional LTG. We recommend the adoption of this method in patients with early-stage tumours that do not invade the dentate line. In such cases, this technique can reduce the difficulty of surgery and ensure surgical safety. Furthermore, there were no statistically significant differences between the Kamikawa and the Roux-en-Y groups in the time to first postoperative anal exhaust, time to first postoperative liquid food intake, length of postoperative hospital stay or the incidence or types of postoperative complications. This suggests that LPG and traditional LTG have comparative surgical safety and postoperative recovery. In the Kamikawa group, one patient experienced postoperative anastomotic stenosis, but the condition improved after treatment with gastroscopy balloon dilation. This occurrence of anastomotic stenosis was attributable to a smaller I-shaped seromuscular flap that did not match the oesophageal stump opening. The I-shaped seromuscular flaps of all other modified Kamikawa anastomosis patients matched the oesophageal diameter, resulting in no other occurrences of postoperative anastomotic stenosis. Our comparative analysis also revealed better BMI and NRS 2002 scores in the Kamikawa group than the Roux-en-Y group at both 6 months and 1 year postoperatively. This is because the preservation of the distal stomach after LPG allows a certain amount of food intake to be maintained. Moreover, this procedure partially preserves the digestive, secretory and food storage functions of the stomach, resulting in more significant postoperative nutritional improvements after Kamikawa anastomosis17,18. These findings are consistent with research results from studies in Japan and South Korea and further confirm the advantages of LPG for the postoperative nutritional status of patients19,20,21. In addition, similar rates of oesophageal reflux during follow-up were observed in both groups of patients, indicating that the artificial anti-reflux structure constructed during modified Kamikawa anastomosis has an effective anti-reflux effect while allowing continued food intake.

During digestive tract reconstruction in LPG, our centre has attempted various other reconstruction methods, including oesophageal-tubular gastric anastomosis, dual-channel anastomosis and side-overlap anastomosis. In future research, we intend to compare the clinical efficacy of these other digestive tract reconstructions in LPG with both one another and the Roux-en-Y anastomosis method.

Our study had several limitations. Firstly, our sample was relatively small owing to the relatively small number of patients with early gastric cancer in the upper part of the stomach in China and the limited number of patients among this population who are suitable for LPG. Secondly, until recently, our centre routinely used Roux-en-Y anastomosis in LTG for early gastric cancer located in the upper part of the stomach. Since we have only been using the Kamikawa approach for a short while, there was a shorter follow-up period on record for the patients in the Kamikawa group. Longer follow-up is needed in future research to evaluate the long-term clinical efficacy, nutritional status and survival outcomes of the two surgical methods. Lastly, this was a single-centre study, so our findings should be generalised cautiously. Future multi-centre prospective clinical studies are needed to validate our findings and evaluate related surgical approaches for early gastric cancer located in the upper part of the stomach.

Conclusion

To sum up, modified Kamikawa anastomosis in LPG is safe and feasible for application to early gastric cancer located in the upper part of the stomach. Its postoperative efficacy is comparable to that of Roux-en-Y anastomosis in LTG, with the additional advantage of maintaining better postoperative nutritional status in patients.