How do we prevent severe intra‐abdominal infectious complications following minimally invasive gastrectomy for cancer? The usefulness of a novel marker using computed tomography images (minimum umbilicus–vertebra diameter) and robotic surgery

Abstract Background Intra‐abdominal infectious complications (IAICs) following minimally invasive gastrectomy (MIG) for cancer sometimes worsen short‐ and long‐term outcomes. In this study, we focused on the minimum umbilicus–vertebra diameter (MUVD) in preoperative computed tomography (CT) images and robotic surgery to prevent severe IAIC occurrence. Patients and Methods A total of 400 patients with gastric cancer who underwent 204 laparoscopic gastrectomy (LG) and 196 robotic gastrectomy (RG) procedures were enrolled in this study. We retrospectively investigated the significance of the MUVD and robotic surgery for preventing severe IAICs following MIG using multivariate and propensity score matching analysis. Results The MUVD cutoff value was 84 mm by receiver operating characteristic (ROC) curve using severe IAICs as the end point. The MUVD and visceral fat area (VFA) had significantly higher area under the curve (AUC) than BMI (MUVD vs. BMI, p = 0.032; VFA vs. BMI, p < 0.01). In the multivariate analysis, high MUVD (HR, 9.46; p = 0.026) and laparoscopic surgery (HR, 3.35; p = 0.042) were independent risk factors for severe IAIC occurrence. In the propensity matching analysis between robotic and laparoscopic surgery in the high MUVD group, the RG group tended to have a lower severe IAIC rate than the LG group (0% vs. 9.8%, p = 0.056). Conclusion The MUVD was a novel and easy‐measuring predictor of severe IAICs following MIG. Robotic surgery should be considered first in patients with gastric cancer having an MUVD value of 84 mm or higher from the perspective of severe IAIC occurrence.


| INTRODUC TI ON
Among various cancers worldwide, gastric cancer ranks fifth and third in incidence and mortality rates, respectively. 1The mainstay for localized gastric cancer is surgical treatment.Recently, minimally invasive gastrectomies (MIGs) including laparoscopic gastrectomy (LG) and robotic gastrectomy (RG) have been focused as more useful surgical approaches than open surgery.However, among the complications following MIG, intra-abdominal infectious complications (IAICs) including pancreatic fistula, anastomotic leakage, and intra-abdominal abscess are the main causes of increased length of hospital stay and perioperative mortality.3][4] Therefore, prediction and countermeasures for IAICs following MIG are significant.Various risk factors for IAIC development following gastric cancer surgery have been previously reported, including male sex, 5 obesity, 6,7 pancreatic thickness, 8 laparoscopic surgery compared with open surgery 9 or robotic surgery, 10 and total gastrectomy compared with distal gastrectomy. 7Among these predictive factors, obesity is one of most significant factors for postoperative IAIC occurrence as the number of patients with obesity has been increasing globally in recent years, 11 and it is expected that there will be more opportunities to perform surgery on patients with obesity with gastric cancer in Japan in the near future.It has been previously reported that some obesity-related body parameters measured using preoperative computed tomography (CT) images such as anterior-posterior diameter (APD) were simple and easy-measuring indicators for IAICs following surgery for patients with gastric cancer. 12,13However, no established consensus on its usefulness in predicting IAICs has been observed.Therefore, in this study, we focused on the novel and original obesity-related parameter, which is the minimum umbilicus-vertebra diameter (MUVD) measured using axial CT image.We hypothesized that the high MUVD is closely associated with severe IAIC occurrence following MIG.This study aimed to clarify the impact of the MUVD on severe IAIC occurrence and compare the MUVD with previously reported obesity-related indexes, such as BMI, visceral fat area (VFA), and APD, regarding severe IAIC occurrence following MIG.Additionally, we evaluated the usefulness of robotic system in terms of preventing severe IAICs following MIG for patients with high MUVD.

| Patient selection
Inclusion criteria were as follows: histologically confirmed gastric carcinoma (cT1-4a, cN0-3) and minimally invasive gastrectomy associated with radical lymphadenectomy, including RG and LG.A total Malignant Tumors (8 th edition). 14All data were extracted from a database where patients were prospectively registered.This study was approved by the Ethics Committee of Osaka City General Hospital (No. 1806031).All patients provided their informed consent.

| Surgical procedure and postoperative treatment
Surgery was performed with robotic assistance or laparoscopically.
Regarding robotic surgery, da Vinci Si or Xi (Intuitive Surgical, Inc.,) was used, and the energy device was mainly an ultrasonic coagulation and cutting device reported by Hyung et al. 15 or a bipolar device reported by Uyama et al. 16 The details of the surgery are described in the previous paper. 17In addition, lymph node dissection was performed in accordance with the Japanese Gastric Cancer Society guidelines, with D1+ lymph node dissection for early cancer and D2 lymph node dissection for advanced cancer. 18After gastrectomy, B-I or B-II or Roux-en-Y reconstruction was performed, and a drainage tube was inserted into the abdominal cavity.After surgery, patients were treated according to the clinical pathways, and were allowed to start drinking fluids from the next day after surgery and start a soft diet from the third day.Antibiotics were used only during surgery.When postoperative complications occurred, their severity was determined and appropriate interventions (i.e.,

K E Y W O R D S
CT image, gastric cancer, intra-abdominal infectious complications, robotic gastrectomy radiographic or surgical interventions, and conservative pharmacological treatment) were planned.The drain was removed, and patients were able to take sufficient oral intake and to carry out daily activities without any problems.It was determined whether patients were allowed to leave the hospital.

| Short-term surgical outcomes
Postoperative complications were classified according to the Clavien-Dindo (CD) classification. 19CD class II or higher was defined as overall complication, and CD class III or higher was defined as a severe complication.In addition, anastomotic leakage, intra-abdominal abscess, and pancreatic fistulas were defined as intra-abdominal infectious complications (IAICs).The following parameters were assessed: surgical time, estimated surgical blood loss, number of retrieved lymph nodes.The short-term outcomes compared between the two groups included surgical time, estimated blood loss, postoperative overall, severe complication rates, details of postoperative complication, and duration of hospital stay.

| Body parameters using CT imaging
The following distances were measured in the axial section at the umbilical level in the preoperative CT images to be obtained within 1 month preoperatively (Figure 1A,B).MUVD was defined as the minimum distance from the deepest point of the umbilicus to the vertebra (Figure 1A).APD was defined as the distance of the anterior abdominal skin to the backside skin at the umbilical level (Figure 1B).Moreover, VFA was calculated from axial CT images at the umbilical level. 12It was calculated by setting the attenuation level within the range of −200 and −50 threshold units using the image analyzing system SYNAPSE VINCENT® (FUJI Film Corporation) reported in a previous study. 20

| ROC analysis and determination of the best marker for severe IAIC occurrence
ROC curves for the above measured parameters were constructed using the presence or absence of severe IAICs as the end point, and F I G U R E 1 Schema of body parameters using axial computed tomography (CT) images (A,B) and ROC curves of body parameters using the presence or absence of severe IAICs as the end point (C-F).Minimum umbilicus-vertebra diameter (MUVD) is defined as the minimum distance from the deepest point of the umbilicus to the vertebra (Figure 1A).APD is defined as the distance of the anterior abdominal skin to the backside skin at the umbilical level (Figure 1B).The ROC curves of MUVD, APD, BMI, and VFA are shown in Figure 1C-F, respectively.The AUCs of the ROC curves regarding the MUVD and APD are 0.668 and 0.629, respectively; the AUC of the MUVD tends to be higher than that of the APD (p = 0.097).The AUCs of the ROC curves regarding BMI and VFA are 0.557 and 0.692, respectively.The DeLong test shows that the MUVD and VFA have significantly higher AUCs than BMI (MUVD vs. BMI, p = 0.032; VFA vs. BMI, p < 0.01), whereas APD is similar to BMI (APD vs. BMI, p = 0.104).Furthermore, the AUCs of the MUVD and VFA are similar (p = 0.268).The MUVD cutoff value is determined to be 84 mm from the ROC curve.Sensitivity and specificity for predicting severe IAIC occurrence are 61.9% and 71.7%, respectively.
AUC values were measured.The Youden index of each parameter was also calculated from each ROC curve and used as each cutoff point.Among these two parameters, including the MUVD and APD, the most useful marker for IAIC occurrence was determined by the DeLong test using the AUC value of ROC curves, 21 and all patients were divided into two groups on the basis of the cutoff value of the best marker; subsequently, patient characteristics and short-term postoperative results were compared between the high and low groups.Furthermore, the ROC curves of the best marker, BMI, and VFA were compared for accuracy as a test for IAIC occurrence using the DeLong test.To investigate the association between the best marker and VFA or BMI, the Pearson correlation analysis was introduced.If the Pearson correlation coefficient (r) value was 0.7 or more, two parameters were determined as very high correlation, whereas if the r value was between 0.5 and 0.7, two parameters were determined as high correlation.If the r value was between 0.3 and 0.5, two parameters were determined as medium correlation, whereas if the r value was between 0.3 and 0.1, or 0.1 and 0, two parameters were determined as low or no correlation, respectively.

| Multivariate logistic regression analyses of risk factors for severe IAICs
Multivariate analyses of risk factors for severe IAICs were performed using logistic regression analyses.The cutoff values of each examined factor were determined using the Youden index from ROC curves.

| Propensity score (PS) matching analysis between the RG and LG groups in terms of short-term outcomes following MIG for patients with high MUVD
To reduce the heterogeneity of patients' backgrounds in patients with high MUVD, we performed 1:1 PS matching between the RG and LG groups.The 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, MUVD, gastrectomy type, and the extent of lymph node dissection.Subsequently, we compared the short-term surgical results, including severe IAICs following surgery between the RG and LG groups.

| Statistical analysis
Continuous variables were compared using Mann-Whitney's U test, and categorized variables were compared using the chi-squared test or the Fisher's exact test.A p value of <0.05 or less was considered significant.SPSS software and EZR software were used for data analysis.0.629, respectively.The AUC of ROC curve for MUVD tend to be higher than that of APD (p = 0.092).The AUC of BMI and VFA were 0.557 and 0.692, respectively, and the DeLong test showed that MUVD and VFA had significantly higher AUC than BMI (MUVD vs.

| The Pearson correlation analysis between MUVD, BMI, and VFA
When examining the correlation between MUVD, BMI, and VFA, the Pearson correlation coefficient between MUVD and VFA, and between MUVD and BMI were 0.879 and 0.53, respectively, indicating that MUVD has a very high correlation with VFA, while it has a high correlation with BMI (Figure 2).

| Association of surgical approaches (RG or LG) with baseline characteristics and surgical outcomes in patients with high MUVD using PS matching analysis
The backgrounds and surgical outcomes of patients who underwent RG or LG before and after PS matching in the high MUVD group are presented in Tables 3 and 4. Before PS matching, the RG group had more patients with grade III for the ASA-PS criteria (p = 0.049).The RG group tended to have more males, more-advanced oncological stage, and higher BMI than the LG group (p = 0.094, p = 0.098, and p = 0.073, respectively).Following PS matching, 51 patients in each RG and LG groups were selected.After PS matching, the severe IAIC rate tended to be lower in the RG group than that in the LG group (0% vs. 9.8%, p = 0.056).
Regarding individual overall IAICs, no significant between-group differences were noted (pancreatic fistula, 2.0% in the RG group vs. 5.9% in the LG group, p = 0.617; anastomotic leakage, 2.0% in the RG group vs. 3.9% in the LG group, p = 1.00; intra-abdominal abscess, 2.0% in the RG group vs. 9.8% in the LG group, p = 0.205).
The RG group had a significantly shorter hospital stay following surgery than the LG group (11.0 vs. 12.0 days, p = 0.037).

| DISCUSS ION
In this study, we observed that the high MUVD group was associated with significantly more postoperative IAICs, particularly pancreatic fistula, anastomotic failure, and intra-abdominal abscess, than the low MUVD group.Furthermore, multivariate analysis showed that the MUVD was an independent risk factor for severe IAIC occurrence.VFA has been previously reported to be a significant marker of IAICs. 20,22In this study, the MUVD was noted to be highly correlated with VFA according to the Pearson correlation analysis.It has been reported that patients with a significant amount of visceral fat are more likely to have pancreatic leakage and intra-abdominal abscess because the border between the pancreatic parenchyma and fat tissue is difficult to distinguish. 23 previously reported that VFA is a more significant marker than BMI for IAIC occurrence following gastric cancer surgery, 24,22 which was consistent with this study.
Previous studies have reported that various measurements using axial CT sections affect short-term surgical outcomes following gastrectomy.Ojima et al. reported that in male patients with gastric cancer who underwent LG, the APD was associated with the surgical time but not with the occurrence of postoperative complications. 12Lee et al. reported that female patients who experienced postoperative complications following open subtotal gastrectomy had higher APD values than those who did not. 13In this study, the AUC of the ROC curve for the MUVD tended to be higher than that of the APD (0.668 vs. 0.629, p = 0.092).Additionally, the multivariate analysis of risk factors for severe IAIC occurrence showed that the MUVD was an independent risk factor for severe IAICs, whereas APD was not.These findings suggested that the MUVD was a more reliable predictive marker for severe IAIC occurrence following MIG than BMI and APD.Yamamoto et al. reported that in cases of total gastrectomy and splenectomy, the distance from the epigastric skin to the root of the celiac artery at the upper abdominal body is involved in the occurrence of postoperative pancreatic fistula. 25They reported that when the pancreas is located deep,

| Limitations
This report is a single-center retrospective review.In addition, it has been reported that postoperative complications are correlated with the skill and experience of the surgeon, thus results may vary depending on these factors.In this study, both robotic and laparoscopic surgery were performed by well-experienced surgeons who are certified by the Japanese Society of Endoscopic Surgery.
Because patients who do not qualify were excluded from this study, it is assumed that bias in technique and experience among surgeons was minimized.Additionally, since this study targeted only Japanese patients, this MUVD cutoff value may be suitable only for Japanese patients.It may be necessary to consider them individually in Europe, the United States, and other Asian countries.
In addition, under special circumstances such as ascites retention, intestinal obstruction, or multiple renal cysts, MUVD may increase even though visceral fat is low.Such cases need to be excluded in this study.
TA B L E 4 Surgical outcomes in patients who underwent RG and LG in the high MUVD group using PS matching analysis.b Data were expressed as number (%).
c Clavien-Dindo classification Grade II or higher.
d Clavien-Dindo classification Grade III or higher.
of 574 patients with local gastric cancer (GC) were surgically treated with curative intent at the Department of Gastroenterological Surgery of Osaka City General Hospital from January 2015 to August 2022.Patients with suspected T4 tumors or bulky metastatic lymph nodes who underwent open surgery (n = 122), or esophagogastric junction cancer, including Siewert type I and type II (n = 8), or remnant GC (n = 7), or GC with another synchronous cancer (n = 5) were excluded from this study.In our surgical strategy for localized GC, open surgery was indicated for patients with suspected T4 tumor or bulky metastatic lymph nodes, and minimally invasive surgery (MIS), such as LG or RG, was introduced for patients without highly advanced GC.Furthermore, this study focused on postoperative complications, which are important to maintain the technical safety and quality of surgery; therefore, the surgeons in this study were an experienced four surgeons who held both a technical certification from the Japan Society of Endoscopic Surgery and a da Vinci robotic surgery certificate.Furthermore, the surgeon performed more than 50 laparoscopic surgeries, and met certain technical standards.A total of 32 cases operated by junior surgeons were excluded from this study.Ultimately, 400 gastric cancer patients were enrolled in this study.The pathological diagnosis and classifications of gastric carcinoma were made in accordance with the JGCA guidelines and the Union for International Cancer Control TNM Classification of
lymph node dissection around the pancreas becomes challenging, and pancreatic fluid leakage may occur more easily.Conversely, our study measured the distance from the umbilicus to the vertebral body in a CT slice at the umbilical level, which is presumed to reflect the amount of visceral fat rather than the pancreatic depth.To our knowledge, no previous studies have reported the relationship between the MUVD and severe IAICs.The multivariate analysis of severe IAICs risks showed that in addition to the TA B L E 3 Characteristics of patients who underwent robotic gastrectomy (RG) and laparoscopic gastrectomy (LG) in the high MUVD group using propensity score matching analysis.

Table 1
shows the comparison of patient characteristics and the short-term postoperative results between the high and low MUVD group in all patients.The high MUVD group (n = 159) had significantly lower age (69.0 vs. 72.0,p = 0.039) and higher BMI (25.2 vs. 21.5, p < 0.01) and VFA (154.1 vs. 72.9cm2, p < 0.01), and was more likely to be male (75.5% vs. 53.1%,p < 0.01) than the low MUVD group (n = 241).The high MUVD had more grade II and less grade I (p = 0.012) for the ASA-PS criteria.Additionally, the high MUVD group had a higher rate of laparoscopic surgery (LG) and a lower rate of robotic surgery (RG) (p = 0.003).The patients were comparable in tumor stage, extent of gastric resection, and extent of lymph-node dissection.Short-term postoperative results showed that the high MUVD group had significantly longer operative time (376 vs. 362 min, p < 0.001) and greater blood loss (50.0 vs. 25.0 mL, p < 0.001).Postoperatively, there were signifi-
Multivariate analysis of risk factors for severe intraabdominal infectious complication following MIG for cancer.
IAICs before surgery in daily clinical practice.Conversely, although a high BMI is a marker of obesity that can be easily measured, it was not a significant risk factor for severe IAICs in the multivariate analysis.As BMI reflects the whole-body muscle mass and fat mass compared with VFA and MUVD, it is assumed that BMI does not play a role in IAIC occurrence in gastric cancer surgery.It has beenTA B L E 2