Introduction

Minimally invasive surgery is being established as the standard surgery for gastric cancer [1,2,3,4,5]. More than half of gastric cancer surgeries (55.1%) were performed with a totally laparoscopic approach in 2019 in Korea [6]. Precise and appropriate lymph node dissection is one of the most important factors for ensuring oncological safety. Since many blood vessels and lymphatics surround the stomach, secured hemostasis is vital in laparoscopy surgery.

Energy-based devices (EBDs) are the most essential and effective instruments for facilitating appropriate tissue dissection and securing hemostasis during laparoscopic surgery. These technologies enable the surgeon to improve procedural efficiency and safety with decreased blood loss and operative duration. However, because collateral heat derived from the energy devices can cause iatrogenic damage, such as pancreas fistula, the surgeon cannot become distracted during the procedure [7, 8]. In contemporary practice, there is a constant expectation for technological improvements to produce high-quality EBDs with more precise vessel sealing quality and reduced thermal injury to the surrounding tissues. Previous studies related to EBD functions have primarily been conducted using animal or human tissue, however [9,10,11,12,13]. Prospective comparative studies on EBDs in real surgery are scarce and have only addressed open thyroidectomy and laparoscopic left colon surgery, not, to the best of our knowledge, gastric cancer surgery [14, 15].

Currently, three different types of EBD are widely used in gastric cancer surgery: ultrasonic shears (US), advanced bipolar (BP) and ultrasonic-bipolar hybrid (HB). All three types of device use different mechanisms to optimally ligate and coagulate the vessels and lymphatics during minimally invasive surgery. US rapidly cut and precisely dissect tissue; however, they reach temperatures of up to 200 °C, which can cause lateral thermal damage to adjacent tissue. BP devices provide reliable vessel sealing with a maximum temperature below 100 °C. However, although it reduces thermal spread up to 1 mm, because a separated cutting system is needed, the tip of this device is relatively blunt. Thunderbeat® (Olympus, Japan) is the first device to integrate both ultrasonic and bipolar energy into one instrument for reliable vessel sealing and coagulation with simultaneous cutting [16, 17]. However, due to the nature of EBDs, some degree of lateral thermal spread is inevitable, so there is a potential risk of thermal damage to the tissue in all three types of device. To reduce such “collateral damage”, US and hybrid devices employ a method of reducing activation by perceiving heat, while BP devices use their own, low-temperature method, as described above; however, previous studies have shown conflicting results on the effectiveness of these methods [11,12,13, 18,19,20].

Our study group previously compared and analyzed different EBDs in robot-assisted gastrectomy [21]. We showed that compared with an ultrasonic device, the EndoWrist® One™ Vessel Sealer (Intuitive Surgical, Inc., Sunnyvale, CA), an articulating bipolar energy device employed in da Vinci® (Intuitive Surgical, Inc., Sunnyvale, CA) robot-assisted gastrectomy, reduced inflammation as represented by the C-reactive protein (CRP) level.

However, the difference between EBDs in laparoscopic gastrectomy has not yet been compared. Therefore, in this study, we aimed to compare the intraoperative inflammatory response and short-term surgical outcome in real laparoscopic distal gastrectomy patients using different types of EBDs.

Materials and methods

Study design and Ethical statement

This is a prospective, multicenter, randomized, single-blind (participant), parallel-group, controlled, comparative clinical trial. The study was conducted in two different high-volume centers in Korea, Seoul National University Hospital (SNUH) and Seoul National University Bundang Hospital (SNUBH), where about 700–1,000 gastric cancer surgeries are performed annually. The recruitment period was from March 2018 to December 2020, followed by the first postoperative outpatient visit from March 2018 to January 2021. Data were recorded in an electronic case report form and collected in a web database.

The protocol was approved by the Institutional Review Board of SNUH (IRB No. 1705-099-856) and SNUBH (IRB No. B-1709-421-401) and registered on Clinical Trials.gov (No. NCT03356626). The study was conducted under the guidance of Good Clinical Practice and in accordance with the Declaration of Helsinki for research involving human studies. Written informed consent was obtained preoperatively from all participants.

Sample size calculation

In our earlier robot-assisted distal gastrectomy study, the average CRP levels following ultrasound and bipolar device use on postoperative day (POD) 2 were 11.5 and 8.0, with standard deviations of 6.6 and 2.8, respectively [21]. The primary purpose of this study was to compare patients treated with US and BP devices (the main experimental groups). HB devices have been exploratively compared since there are no reliable previous data. The sample allocation ratio between groups was decided as US:BP = 1:1, and the same number was decided for the HB device group (exploratory group), for whom the device was expected to have a temperature to the US and BP devices according to its operating features. The risk for type I error was allowed to be 0.05% by considering a P value of 0.05 as statistically significant. The Type 2 error (β) was 0.20, retaining a test power of 80%. Sample size was calculated using Power Analysis and Sample Size software (PASS; http://www.ncss.com). Taking into account a 10% dropout rate, the sample size was determined to be 63 people in each group, for a total of 189 people.

Participants

Participants who met the following criteria were included: (1) age between 20 and 80 years; (2) pathologically proven clinical stage I primary gastric adenocarcinoma [T1N0, T1 and borderline lymphadenopathy, and T2N0 on preoperative gastroscopy and abdominal computed tomography (CT)] according to the 7th edition of the American Joint Committee on Cancer/Union for International Cancer Control (AJCC/UICC) TNM staging system; (3) preoperative high-sensitivity C-reactive protein (hsCRP) < 1.0 mg/dl; (4) expected R0 resection with curative intent by laparoscopic distal gastrectomy; (5) available for 30 days of follow-up in an outpatient clinic after surgery without any communication problems; and (6) voluntary agreement to participate in this clinical trial by written informed consent. The initial study design set the preoperative hsCRP < 0.5 mg/dl. However, despite the absence of clinical inflammatory findings, a CRP level slightly higher than 0.5 mg/dl often leads to screening failure. Even though the exact normal range of CRP is unknown and depends on the laboratory performing the tests, it was modified to < 1.0 mg/dl from the time when the enrollment of 53 participants was completed (from July 2019) with reference to previous studies [22, 23].

The exclusion criteria for the participants were as follows: (1) history of upper abdominal surgery; (2) history of prior therapy for gastric cancer (chemotherapy and/or radiotherapy); (3) the need for combined resection of other organs during the operation (except cholecystectomy); (4) significant ascites on CT; (5) liver cirrhosis; (6) heart disease (uncontrolled hypertension, history of angina, coronary artery disease, or cardiomyopathy, or ejection fraction < 50% at echocardiography); (7) inappropriate renal function (serum creatinine > 1.4 mg/dl or blood urea nitrogen (BUN) > 26 mg/dl); (9) inappropriate coagulation function (prothrombin time international normalized ratio (PT INR) > 1.2 or activated partial thromboplastin time (aPTT) > 45 s); (10) uncontrolled diabetes; (11) inflammatory bowel disease (including Crohn’s disease, ulcerative colitis and intestinal tuberculosis); (12) history of taking aspirin or antiplatelet medication within 5 days before surgery; (13) history of anticoagulants; (14) history of steroid use that needed stress dose inoculation for the surgery; (14) pregnancy; (15) history of gout; (16) history of allergy reaction to iodine when planning to use indocyanine green (ICG) during the operation; and (17) patients with any other clinical conditions that investigators deemed unsuitable for the study.

All eight participating gastrointestinal surgeons were experts in gastric cancer surgery with sufficient experience to follow the surgical procedures of this study. Each of them had performed gastric cancer-related gastrectomy for more than 7 years and performed more than 100 laparoscopic gastric cancer surgeries annually. To reduce the bias of the learning curve of the three instruments, they alternatively used the three EBDs for 3 months before the clinical trial to become accustomed to their use. Even in laparoscopic gastric surgeries that were not registered in the study, the three different EBDs were equally used throughout the whole study period.

Randomization

The centralization method was used to implement the random allocation and conceal the random allocation sequence from the surgeons until the time of operation. The participating surgeons enrolled subjects who met the preoperative eligibility criteria at the outpatient clinic. After the surgeons confirmed the last eligibility criterion intraoperatively by diagnostic laparoscopy, they made a phone call to the designated study coordinator in the operating room. The coordinator ran a web-based computer software randomization program developed by the Medical Research Collaboration Center (MRCC), SNUH. This program automatically assigned each enrolled patient into one of three groups at a 1:1:1 ratio: US─ultrasonic shear (Harmonic® HD 1000i shear, Ethicon, USA), BP─advanced bipolar (LigaSure™ Maryland, Medtronic, USA) and HB─ultrasonic-bipolar hybrid (Thunderbeat®, Olympus, Japan). A stratified block randomization method was used to ensure a balanced distribution among the three groups. Institutes and the presence of an incision during anastomosis (intracorporeal or extracorporeal anastomosis) were employed as stratification factors during randomization. Patients were blinded to the assignment, as they agreed in the informed consent form.

Interventions

Diagnostic laparoscopy was performed under usual general anesthesia in all cases. After randomization, conventional laparoscopy distal gastrectomy with partial omentectomy and D1 + LN dissection (D2 allowed at the surgeon’s discretion) were performed. Intracorporeal/extracorporeal, Billroth-I, Billroth-II, and Roux-en-Y procedures were all allowed for reconstruction.

After randomization, the study proceeded to prospective data collection before, during and after surgery. Intraoperative blood loss (IBL) and blood transfusion volume, operation time, and lymphatic leakage were measured during the operation. Serum laboratory tests, including hsCRP, were performed preoperatively, on POD 2, POD 4, and 14 days after discharge. The amount of Jackson–Pratt (JP) drainage was measured until POD 3. Serum/drain triglyceride and amylase levels were tested on POD 2. Cytokines [interleukin (IL)-6 and IL-10] were sequentially checked preoperatively, 2–3 h after the operation, and on POD 2 and POD 4. For all the enrolled patients, the visual analog scale (VAS) score was recorded once before the operation and daily until POD 4; pain severity scores were compared.

Adverse reactions and complications were postoperatively observed and assessed. Unless follow-up or extension of admission was needed due to adverse reactions, discharge was planned within POD 5–10. Follow-up was performed until POD 30, including the first outpatient visit 2–3 weeks after discharge.

Outcomes

Primary end point

The primary outcome was serum hsCRP (a surrogate marker for complications and inflammation) level on POD 2.

Secondary end points

Secondary outcome measures were as follows: (1) postoperative cytokine (IL-6 and IL-10) and white blood cell (WBC) levels, serum/drainage amylase levels, VAS pain score, and complications within 30 days after surgery to further estimate collateral damage in addition to the primary end point; (2) IBL and postoperative hemoglobin (Hb) level to evaluate the function of blood vessel sealing and cutting; and (3) the amount of ICG in used pieces of gauze (leakage amount) and serum/drainage triglycerides to investigate the lymphatic sealing function of the devices.

In accordance with the latest updated definition by the International Study Group on Pancreatic Fistula, the following case is defined as a postoperative pancreatic fistula: a drain output of any measurable volume of fluid with an amylase level > 3 times the upper limit of institutional normal serum amylase activity (Seoul National University Hospital: 30–118 U/L) and associated with a clinically relevant development/condition related directly to the postoperative pancreatic fistula (persistent drainage > 3 weeks) [24,25,26].

For accurate estimation of IBL, intraabdominal irrigation with saline was not allowed during the operation. All bleeding up to the moment of anastomosis was managed with only suction and gauze. The weight of the gauze was measured before the operation, and then the difference was calculated after the operation. The blood collected in the suction bottle was moved to a 50 cc syringe, and then the amount of blood was measured. The weight of all gauze and amount of blood in the suction bottle were added.

A semiquantitative method using ICG and fluorescence was used to estimate and compare lymphatic leakage during surgery. This procedure was performed only at SNUH. ICG (0.5–1.0 ml, 2.5 mg/ml) was injected intraoperatively into the submucosa at 4 sites [2 at the lesser curvature (antrum, above angle) and 2 at the greater curvature side (antrum, lower body)] to ensure the maximum distribution of IGC throughout the dissected lymph node area and to easily confirm the distribution. Whether ICG was well distributed at each surgery was frequently confirmed through the near-infrared (NIR) camera throughout the surgery (Supplementary Fig. 1). Gauze was placed near each anatomical cutting margin during the operation to absorb the leaked ICG. All gauze used during the operation were collected before stomach transection. The ICG was then extracted by rinsing in 200 ml of 70% ethanol. The ICG fluorescence intensity of the solution was measured using a SPY™ (Stryker, USA) NIR imaging system inside a handmade black box to block out light (Fig. 1). Patients with intraabdominal spillage of ICG due to penetration of the stomach wall during ICG injection were excluded from the final analysis.

Fig. 1
figure 1

Semiquantitative measurement of lymphatic leakage with ICG and the SPY™ (Stryker, USA) system. ICG indocyanine green, LN lymph node

Full size image

Statistical analysis

Continuous and categorical variables are summarized as the means ± standard deviations (SDs) and proportions/percentages, respectively. Demographics, surgical outcomes, and complication rates among the three groups were analyzed using the χ-square test and Fisher’s exact tests for categorical variables and one-way analysis of variance (ANOVA) for continuous variables when parametric test assumptions were met. If the ANOVA test was statistically significant, the Tukey test was used post hoc. A linear mixed model (LMM) was used to analyze repeated measurements, such as laboratory results, cytokines, and VAS scores, adjusted by the preoperative baseline measurements. To evaluate the primary outcome, the postoperative hsCRP level, both one-way ANOVA and LMM were performed. IBM SPSS Statistics ver. 26 (IBM Corp., Armonk, NY, USA) was used for analysis. Statistical significance was set at a P value of < 0.05.

Results

Participant selection and patient follow-up

A total of 189 patients were enrolled and randomized in the study from March 2018 to December 2020. After fifteen patients (7.9%) dropped out from the study after randomization for various reasons, the remaining 174 patients who had completed the 30-day follow-up period were included in the final analysis: n = 57 in the US group, n = 60 in the BP group, and n = 57 in the HB group (Fig. 2). The distribution of females was higher in the BP group (% of males, 78.9:56.7:75.4 for US:BP:HB, p = 0.18) despite randomization. There were no significant differences between the three groups in terms of the other characteristics—age, BMI, American Society of Anesthesiologists (ASA) grade, preoperative diagnosis, preoperative comorbidities, and preoperative lab results (Table 1).

Fig. 2
figure 2

CONSORT flow diagram of patient enrollment and randomization. US ultrasonic shears, BP advanced bipolar, HB ultrasonic-bipolar hybrid

Full size image
Table 1 Patient demographics
Full size table

Outcomes

As Table 1 shows, standard laparoscopic distal gastrectomy with D + 1 or higher lymph node dissection was appropriately performed in all participants. There was no significant difference in terms of operation time, number of retrieved lymph nodes (LNs), JP drainage amount, or postoperative hospitalization among the 3 groups (Table 2).

Table 2 Surgical outcomes
Full size table

Collateral damage

The level of CRP, the primary end point of this study, was analyzed in two ways: one-way ANOVA (Fig. 3A) and an LMM (Fig. 3B). There was a significant difference in serum hsCRP level on POD 2 (p = 0.002); the BP group showed the lowest hsCRP level, followed by the US and HB groups [9.03 ± 5.55 ml vs 11.12 ± 5.02 ml vs 12.67 ± 6.14 ml, respectively; p (BP vs. HB) = 0.002]. A more prominent effect was shown in females [BP, 6.52 ± 4.12 ml; US, 9.85 ± 3.71 ml; HB, 11.27 ± 4.74 ml; p (BP vs. HB) = 0.004]. LMM analysis demonstrated that the change in CRP level after surgery showed a significant time × group interaction effect [p (device × time) < 0.0001]. On POD 2, the change for the BP group was lowest, followed by that for the US and HB groups [p (device) < 0.0001; p (US vs. BP) = 0.004, p (BP vs. HB) < 0.0001), p (US vs. HB) = 0.035]. On POD 4, the CRP level was similar between the US and HB groups, while the BP group had a significantly lower CRP level than the other groups [p (device) = 0.006; p (US vs. BP) = 0.004, p (BP vs. HB) = 0.008), p (US vs. HB) = 0.815].

Fig. 3
figure 3

Change in serum hsCRP level. a hsCRP level on POD 2 (one-way ANOVA). b Change in hsCRP level (linear mixed model). hsCRP high-sensitivity C-reactive protein, LMM linear mixed model, POD postoperative day, ANOVA analysis of variance, US ultrasonic shears, BP advanced bipolar, HB ultrasonic-bipolar hybrid, preop preoperative. Data were reported as mean ± standard deviation (n) for continuous variables. *p < 0.05

Full size image

There was no significant difference in postoperative cytokine (IL-6 and IL-10), serum WBC, albumin, BUN, or creatinine level, VAS pain score, or complication rate among the 3 groups (Table 2 and Fig. 4).

Fig. 4
figure 4

Laboratory result, cytokines and VAS score changes (linear mixed model). Hb hemoglobin, LMM linear mixed model, WBC white blood cell, BUN blood urea nitrogen, Cr creatinine, US ultrasonic shears, BP advanced bipolar, HB ultrasonic-bipolar hybrid, IL interleukin, VAS visual analog scale, n.s not significant

Full size image

Blood vessel sealing and cutting function

As shown in Table 2, there were significant differences in IBL among the groups (p = 0.032). The BP group showed the least IBL, followed by the HB and US groups, and the IBL of the BP group was significantly lower than that of the US group [BP, 26.30 ± 25.32 ml; HB, 34.93 ± 37.04 ml; US, 43.65 ± 42.00 ml; p (US vs. BP) = 0.023]. However, no additional transfusion was performed during the operation in any group. There were no significant differences in postoperative hemoglobin change among the 3 groups [p (device × time) = 0.142].

Lymphatic sealing function

There were no significant differences in the number of retrieved LNs between groups (US, 44.44 ± 18.75; BP, 49.85 ± 20.48 ml; HB, 42.93 ± 17.15 ml; p = 0.115) (Table 2). The intraoperatively measured ICG intensity did not differ significantly among the 3 groups (Table 3). There was no significant difference in the postoperative JP drainage amount or serum triglyceride levels among the 3 groups (Tables 2 and 3). However, there was a significant difference in the JP drainage triglyceride level on POD 2 (p = 0.015): the BP group showed the lowest level, followed by the HB and US groups [US, 84.24 ± 59.03 ml; HB, 71.30 ± 51.44 ml; BP, 53.64 ± 33.71 ml; p (US vs. BP) = 0.011] (Table 3).

Table 3 ICG intensity and serum/JP drainage TG (parameters related to lymphatic sealing function)
Full size table

Discussion

According to the results of previous retrospective studies comparing different types of EBD [16, 27,28,29,30,31], the use of EBDs during surgery should be safe and effective and demonstrate certain advantages depending on the mechanism of each device in terms of intraoperative/postoperative outcomes. However, well-established prospective randomized studies comparing EBDs are scarce, especially in laparoscopy gastrectomy. The challenge is to decide the most appropriate device for a particular surgical situation; this choice usually depends on the surgeon’s preference or experience. The purpose of this randomized trial was to investigate the influence and effect of each different type of EBD in real laparoscopic gastrectomy patients. The findings of the study may provide additional guidance.

The main functions of EBDs are to secure hemostasis and lymphatic sealing and assist in appropriate dissection during the surgery. Because EBDs commonly have relatively high temperatures when performing these functions, there is a potential risk of collateral thermal damage during surgery [18,19,20]. In this study, we first focused on determining which device most effectively minimizes lateral thermal damage during gastric cancer surgery.

The most critical problem related to intraoperative collateral thermal damage is postoperative complications. However, the gastric cancer surgeons participating in this trial are experts in using EBDs during gastrectomy. According to prospectively collected complication data for primary gastric cancer surgery at SNUH in 2021, the critical surgical complication rate was only 10%, and infectious complications such as fluid collection, gastrointestinal leakage, and other fistulas accounted for only 2–3%. Because of the relatively low infectious complication rate, thousands of cases may be required for comparison when the complication rate is the primary end point in assessing lateral thermal damage for each device. In fact, even though the HB group showed the highest complication rate (7%) among the three groups (US = 3.5%, BP = 5.0%) in this study, the difference was not statistically significant (p = 0.771). All cases of stenosis occurred at the anastomosis site, and balloon dilatation and/or stent insertion were performed. Other cases included simple postoperative ileus, which recovered only with conservative management. One fluid collection case was managed by percutaneous drainage. The proportion of patients with high drain amylase was approximately 30%, with no significant difference between groups, and no patient among them was classified as having complications according to the Clavien‒Dindo classification (Supplementary table 1).

The rate of complications after gastric cancer surgery itself is low, which reduces the real-world applicability of the study, so CRP was used as a surrogate marker in this study. Although CRP is a marker of systemic inflammatory reactions, thermal damage applied during surgery is considered one of the most important factors associated with systemic inflammation after laparoscopic surgery. Several studies have shown that CRP can be used as an indicator of inflammation and complications [23, 32,33,34,35,36,37]. Shi et al. and Kim et al. showed that CRP is a reliable predictive marker for diagnosing infective complications following gastric surgery. Tanaka et al. concluded that the ratio of CRP levels on postoperative days 1 and 3 (CRP ratio) could predict the risk of severe complications after laparoscopy gastrectomy. Zhang et al. constructed a CRP-based nomogram for the early detection of complications after laparoscopy-assisted gastrectomy.

While many studies have shown that postoperative CRP concentrations can be a useful predictor of postoperative complications, the suggested CRP concentration threshold varies from study to study. van Winsen M et al. showed a significant increase in CRP from POD 0 (day of the operation) to PODs 2 and 3 in all patients, and the median CRP was significantly higher from POD 2 in patients with complications. Since a local injury sustained during surgery causes the CRP level to slowly increase, we decided to evaluate the CRP level on POD 2, which typically peaks at this time and remains stable for a while.

Previously, we compared BP and US devices in da Vinci robot-assisted gastrectomy [21]. This nonrandomized comparative study showed a significant difference in CRP levels between the articulating BP and US devices on POD 2 (8.06 vs. 11.7, p = 0.002). Consistent with this result, there was a significant difference in serum hsCRP levels on POD 2 in this study (p = 0.002). Interestingly, the actual numerical values of the CRP level in the US and BP groups were similar to those of previous studies [BP, 9.03 ± 5.55 ml; US, 11.12 ± 5.02 ml; HB, 12.67 ± 6.14 ml; p (BP vs. HB) = 0.002]. In the LMM analysis, the CRP value peaked on POD 2, decreased slightly on POD 4, and returned to within the normal range on POD 14 for all three device groups. However, compared with the other groups, the BP group showed the significantly lowest peak value on POD 2 and POD 4. This result may indicate that the BP device is advantageous in reducing lateral thermal damage during the operation over the US and HB device. However, since there was no significant difference between the three groups in overall postoperative complications and the CRP levels returned to normal 2 weeks after surgery, it seems that there were no differences between the three groups in overall perioperative outcomes with respect to collateral damage.

IL-6 and IL-10 are the main acute-phase proteins; their levels are an indicator of the extent of surgical trauma and thus can be considered predictors of morbidities after surgical intervention [38, 39]. They have also been used as parameter representing the extent of inflammation and damage in tissues, and higher levels have been found in cases of open surgery than in laparoscopy. A previous study also showed that IL-6 and IL-10 levels increased and peaked between 3 and 24 h after surgery and returned to baseline after 3–4 days, similar to our findings [40, 41]. However, in this study, there were no significant differences between the groups regarding IL-6 and IL-10 levels.

Our study group previously compared conventional monopolar electrocautery and US in open distal gastrectomy, and the results showed reduced IBL in the US group [41]. Our next question was to determine which of the EBDs had better hemostatic function. Kim et al. retrospectively compared perioperative surgical outcomes between a BP device and US during laparoscopic gastrectomy [28]. The results showed that the estimated blood loss between the two groups was similar, but the operation time was shorter in the BP group than in the US group. They reasoned that because the BP device has better vascular sealing capabilities [16], the surgeon could skip the use of clips to ligate identified small perigastric vessels in the BP group, while in the US group, clips were used in most cases. Similar to this result, our study showed that the measured IBL was significantly lower in the BP than in the US and HB groups, since most identified perigastric vessels were ligated with clips in all three groups. These results indicated that the BP device was advantageous in vessel sealing function over US and the HB device. However, because the absolute IBL was very small in all three groups and there was no additional intraoperative transfusion, rapid postoperative hemoglobin change, or significant postoperative complications related to intraabdominal/intraluminal bleeding in any of the three groups, all three devices can allow a safe surgery in terms of intraoperative hemostasis and postoperative bleeding.

Finally, the lymphatic sealing function of the three devices was investigated. First, the lack of a significant difference in the number of retrieved LNs indicates that all three devices helped properly dissect the LNs. Even though proper lymphatic sealing is a critical function in gastric cancer surgery, evaluation of the lymphatic sealing function is difficult since the lymphatics are invisible to the naked eye. To overcome this problem, we planned two methods to evaluate the lymphatic sealing function of the device. To intuitively check the amount of lymphatic leakage, we used ICG and measured its intensity with an NIR camera. After intraoperative ICG injection in the submucosa, the ICG absorbed by pieces of gauze used during the operation was not well extracted with water, but it was well separated when washed with ethanol (Supplementary Fig. 2). In this study, we used the open-field NIR camera included in the SPY system, which had very high sensitivity that allowed us to evaluate and compare small amounts of ICG (Supplementary Figs. 3 and 4). The maximum intensity shown by the NIR camera is 253. When the ICG injection needle pierced the stomach serosa and the ICG leaked into the abdominal cavity, the measured ICG intensity was far greater than 150, well above the average test value; these patients were excluded from the final analysis. The results showed that there was no difference in ICG intensity between the groups. To further analyze our findings, we measured the JP drainage triglyceride level. According to previous studies, postoperative chylous ascites that arise due to improper sealing of the lymphatics or thoracic duct are usually rich in triglycerides (> 200 mg/dl) [42]. In our study, the JP drainage triglyceride level in the BP group was significantly lower than that in the US group, while the JP drain amount was not statistically different among the 3 groups. Although these results should be interpreted with caution because these levels were only measured in some patients depending on the timing of the JP drainage removal, they indicate that the BP device is advantageous over the other two systems in terms of the lymphatic sealing function. However, there was no significant difference in overall perioperative outcomes in terms of the lymphatic sealing function.

A limitation of this study is that the fundamental consequence, the complication rate, could not be evaluated due to the limited number of cases. A stratified randomization method was chosen in this study to prevent an imbalance of prognostic factors that could affect the results, but the baseline characteristics of the patients in each group were not perfectly balanced. Since gastric surgery may be affected by patient factors, such as sex and BMI, those factors should be considered stratification factors.

However, the strength of this study is that it is the first multicenter randomized trial to compare perioperative outcomes between three different EBDs in laparoscopic distal gastrectomy. Although Kim et al. previously conducted a similar study, it was retrospective in nature, and the operations were performed by one surgeon, both mentioned as study limitations [28]. Eight expert gastrointestinal surgeons in different institutions participated in this study and practiced the use of each device for over 3 months to reduce bias. To measure each device’s proper function, we set appropriate study parameters, such as CRP level, IBL, ICG intensity and JP triglyceride level.

In conclusion, the BP group showed the least blood loss, the lowest serum CRP level on PODs 2 and 4, and the lowest JP drainage triglycerides on POD 2 in laparoscopic distal gastrectomy among Stage I patients. This suggests better hemostasis and sealing function and less collateral damage with the BP device. However, overall perioperative outcomes were not different between the three devices. The results of this study could serve as a reference for surgeons in selecting the most appropriate EBD for gastrectomy and other operations.