Impact of a robotic system on intra‐abdominal infectious complications after minimally invasive gastrectomy in patients with gastric cancer: A propensity score matching analysis regarding visceral obesity

Abstract Background The efficacy of robotic gastrectomy (RG) on postoperative complications in patients with gastric cancer (GC) is unclear in terms of the volume of intra‐abdominal fat tissue. Patients and Methods We enrolled 403 patients with GC who had minimally invasive surgery between January 2015 and July 2022. During this time, 197 RG and 206 laparoscopic gastrectomies (LG) were performed on the study participants. According to the computed tomography scan, patients were defined as having or not having visceral obesity based on the visceral fatty area (VFA). In each high and low VFA group, we compared short‐term outcomes between the RG group and LG group. Results After PS matching for patients with high VFA, the two groups were well matched, with 71 cases in both the RG and LG groups. The median surgical time in the RG group was significantly longer (420 vs. 365 min, p < 0.001). However, the RG group had a significantly lower rate of severe intra‐abdominal infectious complications (IAIC), such as anastomotic leakage, pancreatic fistula, and intra‐abdominal abscess (1.4% vs. 15.4%, p = 0.004). However, among the 77 patients with low VFA values, we found no significant difference in the rate of severe IAIC between the two groups (1.1% in the RG group vs. 2.6% in the LG group, p = 1.00). Conclusion RG may be a viable alternative to LG because of the lower postoperative IAIC for patients with visceral obesity and GC. However, RG may not benefit non‐obese patients.


| INTRODUC TI ON
Gastric cancer (GC) ranks third and fifth in terms of the annual number of cancer deaths and cancer incidence worldwide. 1The cornerstone of treatment for localized GC is surgical resection of the primary tumor and regional lymph nodes.Recently, two types of minimally invasive surgical approaches, including laparoscopic gastrectomy (LG) and robotic gastrectomy (RG), are used to complete GC resection.Some surgeons 2,3 report that the robotic system has many advantages over laparoscopic surgery, such as articulated instruments, magnified clear view via a 3D camera, and tremor filtering.These advantages enable complete endoscopic resection of GC without increasing surgeon burden. 4However, the clear clinical benefits of RG compared to LG are unclear.Some surgeons espouse the merits of RG-including less blood loss, more retrieved lymph nodes, fewer intra-abdominal complications, and shorter hospital stays compared with LG. [5][6][7][8] Other surgeon's feel that RG has no significant short-or long-term advantages over LG. 9,10 RG may be helpful during more technically demanding cases such as surgery in patients with obesity or more-advanced disease.However, there is little evidence to support the clinical advantages of RG for those technically demanding cases.
Intra-abdominal excess fatty tissues tear easily and result in easy bleeding during surgery.Excess fat around the pancreas makes it difficult to recognize the pancreatic parenchyma border.There are some reports in which viscerally obese patients are closely associated with more bleeding, longer operation time, and more postoperative complications, especially pancreas-related complications. 11,12cently the number of obese patients has increased globally, 13,14 as has the number of patients with visceral obesity and GC requiring surgery.We hypothesized that RG has clinical benefits over LG for patients with obesity because of the robot's mechanical advantage.We compared short-term outcomes between RG and LG for patients with and without visceral obesity.LG surgeries.Among the MIS cases, RG was mainly performed for GC from April 2018 to July 2022 because Japanese National Insurance approved financial support for RG in April 2018.RG and LG surgical procedures at our institution have been previously reported. 15ree surgeons performed all RG and LG procedures in this study.

| Patients
All three surgeons have experienced more than 50 LG cases and are certified by the Japanese Endoscopic Surgical Society (JESA), and also possess RG certificates issued by an Intuitive Corporation.
Regarding RG types, two surgeons performed Ultrasonic Based Robotic Gastrectomy reported by Hyung et al. 2 The other surgeon performed Maryland Bipolar-Based Robotic Gastrectomy, as reported by Uyama et al. 3

| VFA measurement
VFA is determined by measuring the intra-abdominal fat density area of an axial slice of an abdominal CT image at the height of the patient's umbilicus.VFA is calculated by setting the attenuation level within the range of −200 and −50 threshold unit using the image analyzing system SYNAPSE VINCENT® (FUJI film corporation) reported in a previous study. 11To determine the optimal cutoff point of the VFA values, receiver operating characteristic (ROC) curve were generated using the presence of severe intraabdominal infectious complications (IAICs) as the endpoint.The maximum Youden index, which was calculated by the ROC curve, was determined to be the optimal cutoff point for VFA.The patients were placed in either the high or low VFA group according to the cutoff point.

| Outcomes
We measured surgical time, estimated surgical blood loss, the number of retrieved LNs, postoperative complications, the severity of complications, the rate of readmission and re-operation, and the duration of postoperative stay.
The Clavien-Dindo (CD) classification was followed to evaluate postoperative complication severity. 16We defined overall postoperative complications as Grade II or greater severity as classified by the CD system.Severe complications were those with a severity greater than Grade IIIa.We defined pancreatic fistula, anastomotic leakage, leakage at the duodenal stump, and intra-abdominal abscess as intra-abdominal infectious complications (IAICs).

| Postoperative management
Prophylactic antibiotics were administered intraoperatively every 3 h.Patients generally started a liquid diet on postoperative day (POD) 1 and a soft diet on POD 3. If infectious complications developed after surgery, broad-spectrum antibiotics were administered.If intra-abdominal complications occurred after surgery, such as anastomotic leakage, bowel obstruction, pancreatic fistula, and intra-abdominal abscess, appropriate interventions (i.e., radiographic or surgical interventions and conservative pharmacological treatment), and parenteral venous nutrition and fasting were carried out according to the severity of those complications.Patients were discharged from the hospital around POD 13 if their condition was stable enough to resume daily life at home.Some patients required longer hospital stays due to reduced food intake and poor home environments; the need for an extended stay was determined on a case-by-case basis.
The pathological diagnosis and classifications of GC were made following the JGCA guidelines 17 and the Union for International Cancer Control TNM Classification of Malignant Tumors (8th edition). 18All data were extracted from a prospectively registered database.This study was approved by the Ethics Committee of Osaka City General Hospital (No. 1806031).All patients provided informed consent.

| Propensity score matching
We performed 1:1 PS matching to reduce the heterogeneity of patients' backgrounds between the RG and LG groups.A PS was calculated as the conditional probability of receiving cases from either group using a logistic regression model and included age, sex, ASA-PS score, clinical and pathological oncological stage, BMI, VFA, type of gastrectomy, the extent of lymph node dissection, and surgical experiences of three surgeons.

| Statistical analysis
Continuous variables were compared using Mann-Whitney's U test, and categorical variables were compared using the chi-squared test or Fisher's exact test.Differences were considered significant for p values <0.05.SPSS and EZR (Saitama Medical Center, Jichii Medical University, Saitama, Japan) software were used for data analysis.which was determined to be the VFA cutoff value.All patients were assigned to a "high" (n = 188) or "low" (n = 215) VFA group based on where that patient's VFA was relative to the threshold of 106 cm 2 .

| Association of VFA values with baseline characteristics and surgical outcomes in the whole cohort
The relationships between VFA and patients' characteristics and surgical outcomes are shown in Table 1.The median VFA value in the high and low VFA groups were 153.1 and 60.1 cm 2 , respectively.BMI values in the high and low VFA groups were 24.9 and 21.4,respectively.There were more males, fewer patients classified as ASA-PS category I, and more patients classified as ASA-PS category II in the high VFA group compared to the low VFA group.There were no significant differences in the two groups' mean age, oncological stage, type of gastrectomy, and extent of LN dissection.Regarding surgical outcomes, surgical time was significantly longer for high VFA patients (389 vs. 356 min p < 0.001), and estimated blood loss during surgery was significantly greater (50 vs. 20 mL, p < 0.001).Patients in the high VFA group had significantly fewer retrieved LNs (28 vs. 31, p < 0.01).There were significantly more postoperative overall and severe complications among high VFA patients (CD Grade II or higher; 23.1% vs. 9.8% p < 0.001; CD Grade III or higher; 10.6% vs. 3.3%, p = 0.004).There were significantly more IAICs and moresevere IAICs in the high VFA group (CD Grade II or higher 12.8% vs. 1.9%, p < 0.001; CD Grade III or higher 8.0% vs. 1.4%, p < 0.001).

| Association of surgical approaches with baseline characteristics and surgical outcomes in the low VFA group
Tables 2 and 3 show the patients' backgrounds and surgical outcomes before and after PS matching for the low VFA group.Before PS matching, there were more patients with more-advanced oncological stages and low VFA in the RG group.After PS matching, 77 patients' backgrounds were well balanced with similar BMI and VFA values (BMI; 21.0 vs. 20.9,p = 0.932, VFA; 65.9 vs. 64.7,p = 0.453).
Regarding surgical outcomes, surgical time was significantly longer in the RG group than in the LG group (378 vs. 312 min, p = 0.001), while estimated surgical blood loss was similar between the two F I G U R E 2 Box plot of the VFA values in the whole cohort (A) and receiver operating characteristic (ROC) curve of VFA modulated by presence/absence of severe IAICs (B).The median of VFA values in this study was 98.2 cm 2 .The area under the curve (AUC) for VFA was 0.681.At a VFA of 106 cm 2 , the Youden index was maximized.groups (15.0 vs. 20.0mL, p = 0.167).Overall complication rate was significantly higher in the RG group than in the LG group (18.2% vs. 5.2%, p = 0.022), while severe complication was equivalent between the two groups (5.2% vs. 2.6%, p = 0.681).Overall and severe IAIC were similar between the two groups (CD Grade II or higher IAIC 2.6% vs. 2.6%, p = 1.0;CD Grade III or higher IAIC 1.1% vs. 2.6%, p = 1.0).There were no significant between-group differences for readmission and re-operation rate.Hospital stay after surgery was similar between the two groups (10.0 vs. 10.0 days, p = 0.23).
The rate of total surgical site complications, including gastric delayed emptying, bowel obstruction, and chyle leakage, were significantly higher in the RG group than in the LG group (15.6% vs. 3.9%, p = 0.027), while medical complications were similar between the two groups.b Data were expressed as median (interquartile range).

| Association of surgical approaches with baseline characteristics and surgical outcomes in the high VFA group
Tables 4 and 5 show the patients' backgrounds and surgical outcomes before and after PS matching in the high VFA group.Before PS matching, the RG group had significantly more females than the LG group.After PS matching, each of the 71 patients in the RG group and LG group were selected.12.0 days, p = 0.039).Regarding individual post-surgical morbidities, there were no significant between-group differences (Table 6).
TA B L E 3 Surgical outcomes in patients with the low VFA. a Data were expressed as mean (range).
b Data were expressed as median (interquartile range).
We found that patients with high VFA and GC had longer operation times, more bleeding, and more postoperative complications such as anastomotic leakage, intra-abdominal abscesses, and pancreatic fistulas than patients with low VFA.These results have been previously reported. 11,19,20Generally, increased abdominal cavity fat makes surgery difficult due to the increased likelihood of bleeding, difficulty in recognizing the appropriate dissection layer around the pancreas, and a smaller abdominal cavity surgical field.Laparoscopic gastrectomy for patients with visceral obesity is technically demanding because articulation surgical devices are not used.We investigated the impact of a robotic system on postoperative complication rates in patients with visceral obesity.We found that the robotic system was associated with fewer severe IAIC in patients with visceral obesity.Robotic systems have many advantages, such as an articulated endo-wrist, tremor filtering, and a superior surgical view.These advantages enable us to perform safe and meticulous lymph node dissection around the pancreas, which may result in less severe IAICs in patients with visceral obesity.
Although there have been previous reports on the effects of RG on surgical outcomes in obese patients, few studies have shown the short-term outcome benefits. 21,22Previous studies have used the BMI as a surrogate marker for obesity and investigated the efficacy of BMI in predicting surgical outcomes.Some reports have shown that VFA is a better surrogate obesity marker for IAIC than BMI. 12,23A precisely reflects the amount of intra-abdominal visceral fatty tissue, whereas BMI reflects the muscle and fatty tissue in the whole body.Moreover, females, in particular, can sometimes have a high BMI and normal VFA since females tend to deposit fat subcutaneously and not in the abdominal cavity. 24Therefore, we used VFA as a surrogate obesity marker for IAIC after surgery.To the best of our knowledge, only two studies have examined the impact of RG on surgical outcomes in patients with a high VFA.These two studies yielded different results.One report by Hikage et al. showed that pancreatic fistulas were less common in patients with high VFA in the RG group than in the LG group. 25Another report showed that the surgical outcomes following distal gastrectomy in patients with visceral obesity were similar between the RG and LG groups. 26Our study showed that severe IAICs were significantly less common in RG than in LG.However, we could not detect the exact type of complications among the IAICs.In our study, patients with a high VFA in the RG group tended to have less severe anastomotic leakage, pancreatic fistulas, and intra-abdominal abscesses.However, these results were not statistically significant (anastomotic leakage, 1.4% vs. 8.5%, p = 0.116; pancreatic fistula, 0% vs. 5.6%, p = 0.120; and abscess, 0% vs. 5.6%, p = 0.120    27 reported that a history of abdominal surgery was an independent risk factor for postoperative complications after robotic gastrectomy for gastric cancer.The history of previous abdominal surgery might be associated with a higher incidence of surgical site complications in RG patients in the low VFA group.In this study, the surgical time was significantly longer in the RG group than in the LG group.Many studies have reported longer operation times and higher RG costs than LG. 9,28 According to these reports and ours, RG may not provide sufficient short-term surgical benefits in patients with low VFA and GC. There have been certain reports on the impact of Ultrasonic Shears (US) or bipolar devices as the main energy device for pancreatic thermal injury.Suda et al. reported that pancreatic fistulas after GC surgeries were significantly less common in RG cases using a bipolar device than in LG cases using a US as the main energy devise. 7ey speculated that thermal damage to the pancreatic parenchyma was likely to be less in the bipolar device than in the US.Recently, a randomized controlled study comparing the short-term surgical results of laparoscopic gastrectomy for GC between the groups using US and those using bipolar devices was published in Korea. 29The results showed that postoperative CRP levels were significantly higher in the US group than in the bipolar group.The authors speculated that the higher CRP levels in the US group might be associated with pancreatic thermal damage.In this study, all surgeons in the LG group used a US as the main energy device, whereas in the RG group, two surgeons used a US and one surgeon used a bipolar device as the main energy device.However, in the low VFA group, no pancreatic fistulas were observed in the RG and LG groups, whereas in the high VFA group, pancreatic fistulas were observed in one (1.2%) and five (4.7%) cases in the RG and LG groups, respectively.Moreover, in our previous study on the impact of the main energy devices (US and bipolar) used in the RG on short-term surgical results after GC surgery, no correlation was observed between the surgical devices and pancreatic fistula (0% (0/113) in the US group versus 1.7% (1/58) in the bipolar group; p = 0.339). 30These results suggested that pancreatic fistula was not associated with the type of energy device but with visceral obesity.

| Limitations
First, this was a retrospective, single institution study.Second, we could not precisely quantify the surgeons' RG and LG skills; this could have created between-group differences in short-term outcomes.All of three surgeons in this study qualified endoscopic surgical skill qualification system of the Japanese society of endoscopic surgery and experienced LG and RG equally.Therefore, the surgeons were expected to have comparable surgical skills.Moreover, we used the numbers of patients who underwent RG and LG performed by each surgeon as a covariate in PS matching analysis to reduce the surgeon's technical bias.Lastly, only short-term outcomes after RG and LG were investigated.Thus, the impact of the robotic system on long-term outcomes requires further study.

Figure 1
Figure 1 shows the flow of patient selection in this study.Between January 2015 and July 2022, 561 patients with GC underwent curative surgical resection at Osaka City General Hospital.We excluded 110 patients with a suspected T4 tumor or bulky metastatic LNs who underwent open surgery, five with esophagogastric junction cancer, including Siewert Type I and Type II, five with remnant GC, and three with GC with another synchronous cancer.Open surgery was indicated for patients with suspected T4 tumors or

Figure 2
Figure 2 shows the box plot and the ROC curve of VFA modulated by presence/absence of severe IAICs.The median VFA value in the whole cohort was 98.2 cm 2 .Interquartile range (IQR) was 61.4-147.The area under the curve (AUC) of the ROC curve was 0.686.The maximum Youden index for VFA was 106 cm 2 , Abbreviation: IAIC, intraabdominal infectious complication.a Data were expressed as median (interquartile range).bClavien-Dindo classification Grade II or higher.c Clavien-Dindo classification Grade III or higher.
Patient characteristics and surgical outcomes stratified by VFA value.Patient characteristics in the low VFA group.
Patient characteristics in the high VFA group.
). Large-scale prospective multicenter trials and national databases are required to clarify the exact types TA B L E 5