The present study was approved by the Institutional Review Board of our institution (Chonnam National University Hwasun Hospital, Hwasun, Korea). Between January 2012 and December 2017, 283 patients with gastric cancer underwent LDG. All procedures were performed by a single surgeon who had a previous experience of ≥60 gastrectomy cases/year. Before May 2014, 110 patients underwent conventional LDG using 5 abdominal ports (conventional laparoscopic gastrectomy; CPG group). Reduced-port LDG was adopted for every patient in May 2014, and 173 patients subsequently underwent this procedure. Of these, 14 were excluded because they underwent LDG using 4 abdominal ports shortly after reduced-port LDG was introduced. Finally, 159 patients who underwent reduced-port distal gastrectomy were enrolled in the present study (reduced-port gastrectomy; RPG group). We compared the groups with respect to short-term surgical outcomes, including operative results, hospital course, morbidity, and mortality.

Before surgery, all patients underwent gastroduodenoscopy with biopsy, abdominal computed tomography, chest radiography, and laboratory testing. The indication for laparoscopic gastrectomy was clinical stage cT1-2N0. The gastric resection and regional lymph node dissection procedures followed the Japanese gastric cancer treatment guidelines [12]. D1+ lymph node dissection was generally performed, and D2 lymph node dissection was indicated in cases of cT2 tumors or suspected lymph node metastasis. After surgery, patients were managed using standardized care protocols, including early active ambulation, pain control with epidural anesthesia, prophylaxis to prevent deep vein thrombosis, oral nutrition (from postoperative day [POD] 2), and hospital discharge between POD 6 and POD 8, depending on when the patient fulfilled the discharge criteria.

The following clinicopathological data were prospectively collected: age, sex, body mass index (BMI), American Society of Anesthesiologists (ASA) physical status classification, abdominal surgery history, laboratory results, and pathological data (including tumor, node, metastasis [TNM] stage). Operative parameters were also collected, namely the type of operative procedure, operative time, intraoperative blood loss, and the total number of retrieved lymph nodes. The following postoperative outcomes were measured: time to first flatus, time to diet resumption, duration of hospital stay, and postoperative complications, which were classified into local or systemic complications. Postoperative morbidity and mortality were defined as any complications or death occurring within 30 days of the operation or during hospitalization. The severity of the complications was graded using the Clavien-Dindo classification [13], and pathological staging was based on the seventh edition of the Union Internationale Contre le Cancer/American Joint Committee on Cancer TNM classification system [14].

The patients were placed in the reverse Trendelenburg position, with their legs apart. The operator stood on the right side of the patient, and the laparoscopist stood between the legs of the patient. An umbilical port was inserted for the laparoscope, and pneumoperitoneum was established. Two operator ports were then placed in the right subcostal area (5-mm port) and right mid-abdomen (12-mm port) (Fig. 1A). The liver was retracted using a straight needle with nylon sutures. The nylon suture was passed percutaneously, the middle portion was placed in the gastrohepatic ligament with a hemolock, and both ends were fixed on the skin using mosquito forceps (Fig. 1B). All procedures, including omentectomy, lymph node dissection, gastric resection, and reconstruction, were performed using the instruments used for CPG. No special devices were used for retraction or counter-traction of the stomach.

Lymph node dissection at the suprapancreatic area was usually performed without compressing the pancreas (Fig. 1C). However, in some cases that required D2 lymph node dissection, an additional 5-mm assistant port was added to compress the upper border of the pancreas and to better secure the operating field. During suprapancreatic lymph node dissection, the operating field for the suprapancreatic area was secured by compressing the pancreas with the right hand. Starting from the dissection of the soft tissue at the upper border of the pancreas, lymph nodes were removed in the following order: No. 8a, 12a, 11p, 9a, and 7 (Fig. 1D). During the dissection of lymph node No. 11p, a gauze was used to prevent the stomach from descending. In laparoscopy-assisted distal gastrectomy (LADG), resection of the stomach and extracorporeal Billroth-I or -II anastomosis was performed through a 5–6-cm epigastric incision. During RPG, intracorporeal Billroth-II or Roux-en-Y reconstruction was performed for most patients, and the umbilical port was extended to 2–3 cm to extract the specimen. In early RPG, we performed extracorporeal anastomosis with 5–6-cm mini-laparotomy in some patients, which was similar to that performed in conventional LDG. After adoption of intracorporeal anastomosis, most patients in the later period of RPG underwent intracorporeal anastomosis.

In the CPG group, 5 abdominal ports were used: two bilateral subcostal ports, 2 bilateral mid-abdominal ports, and an umbilical port. The method of gastric resection and lymph node dissection was nearly identical to that of RPG.

The CUSUM technique was used for quantitative assessment of the learning curve. The CUSUM is the running total of differences between individual data points and the mean of all data points. The CUSUM of the operative time (CUSUM_RPG) was defined as follows:

where x_i denotes the operative time for each patient and µ, the mean of the operative time in all patients. The slope of the CUSUM curve indicates the trend of learning performance, and the portion where the slope gently stabilizes is regarded as the breakpoint of the learning curve [15]. In the present study, learning curves and CUSUM curves were obtained for the initial 86 patients. Subsequent operations were excluded, because resident doctors assisted in procedures such as liver retraction and reinforcement suture, which might have affected the operative time but not the surgical outcomes.

All data were analyzed using SPSS software version 23.0 (IBM Corp., Armonk, NY, USA). Continuous data were expressed as mean±standard deviation and compared using the Student's t-test. Categorical variables were expressed as number (%) and compared using the chi-square test or Fisher's exact test. Two-sided P-values less than 0.05 were considered statistically significant. A Microsoft Excel 2010 (Microsoft, Los Angeles, CA, USA) line chart was used for the CUSUM method.

There were 110 patients in the CPG group and 159, in the RPG group. Table 1 shows the baseline characteristics of the patients in both the groups. The mean age of the patients (201 men and 68 women) was 63.9±10.6 years and their mean BMI was 23.6±3.2 kg/m², with 148 patients (55%) having underlying comorbidities. In total, 27 patients (10%) had undergone prior abdominal surgery. Patients in the RPG group did not show significant differences from those in the CPG group with respect to age, sex, BMI, ASA classification, or history of abdominal surgery. None of the following histopathological characteristics differed significantly between the groups: number of tumors, tumor location and size, differentiation, and pathological TNM (pTNM).

Table 2 summarizes the operative outcomes in the 2 groups. Reconstruction techniques showed significant differences between the groups. Most patients (84.4%) in the RPG group underwent intracorporeal anastomosis, whereas most of them (99.1%) in the CPG group underwent extracorporeal anastomosis. Accordingly, Billroth-II or Roux-en-Y gastrojejunostomy was performed more frequently in the CPG group. D2 lymph node dissection was performed more frequently in the RPG group than in the CPG group (54.1% vs. 28.2%; P<0.001). Of the 159 patients in the RPG group, 15 (9.4%) required an additional port insertion during the operation. In 12 patients, 1 trocar was inserted as D2 dissection was difficult with only 3 ports. In 2 of the remaining 3 patients, 1 trocar was inserted owing to a protruding pancreatic body, while in the third patient, it was inserted for adhesiolysis. There was no significant difference in the rate of combined resection between the groups.

The operative time in the RPG group did not significantly differ from that in the CPG group. However, intraoperative bleeding was significantly lower in the RPG group (56.3 vs. 48.2 mL; P<0.001). The number of retrieved lymph nodes was greater in the RPG group than in the CPG group (34.2 vs. 28.5; P=0.001).

Table 3 shows the postoperative outcomes in the 2 groups. There were no significant differences between the groups with regard to the first flatus time or length of hospital stay. Patients in the RPG group resumed diet significantly earlier than those in the CPG group did (2.0 vs. 2.6 days; P=0.012). After surgery, 18 patients (16.4%) in the CPG group and 27 patients (17%) in the RPG group developed complications; this difference was not statistically significant. No postoperative deaths occurred in either group. Neither the incidence nor the severity of each complication significantly differed between the groups. Four patients in the CPG group experienced severe complications; 2 patients experienced duodenal stump leakage requiring re-operation, one patient experienced chyle leakage requiring drainage for more than 7 days; and one patient experienced renal failure requiring dialysis. Three patients in the RPG group experienced severe complications; 1 patient each required pigtail insertion for duodenal stump leakage, percutaneous aspiration for intraabdominal infection, and re-operation owing to bleeding.

The CUSUM learning curve and operative times are shown in Fig. 2. The CUSUM learning curve consisted of three phases: phase I (the initial 17 cases), phase II (the middle 15 cases), and phase III (the final 53 cases). From the 1st case to the 17th case (phase I), the CUSUM curve showed a rise, followed by a fall from the 18th case to the 33rd case (phase II). After the 34th case (phase III), the curve showed a relatively gentle and flat shape.