Does Robotic Roux-en-Y Gastric Bypass Provide Outcome Advantages over Standard Laparoscopic Approaches?

This retrospective study was approved by the Institutional Review Board (IRB) of the Cleveland Clinic. Between January 2008 and December 2015, 703 obese patients underwent RYGB in Cleveland Clinic’s Bariatric and Metabolic Institute. All procedures were performed by high-volume (500+ cases/surgeon) bariatric surgeons (TR, MK SB, PS). Study data were reviewed and recorded using an (REDCap) electronic data capture tool hosted by Cleveland Clinic [11]. The exclusion criteria were as follows: history of previous weight loss surgery, any cases with concurrent procedures, or other unusual problems including a diagnosis of gastrointestinal dysmotility (gastroparesis, global bowel dysmotility, colonic inertia). One hundred twenty-five patients were excluded from the study. Five hundred seventy-eight patients were divided into four groups according to technique used for GJA creation: robotic (TRRYGB), 21-mm EEA circular-stapled (LRYGB-21CS), linear-stapled (LRYGB-LS), hand-sewn (LRYGB-HS) (Fig. 1). The patient variables collected included: age, gender, race, initial BMI, and comorbidities (chronic obstructive pulmonary disease, asthma, obstructive sleep apnea, pulmonary hypertension, pulmonary embolism, dyslipidemia, hypertension, diabetes mellitus, coronary artery disease, myocardial infarction, valvular heart disease, cardiomyopathy, cardiac arrhythmia, congestive heart disease, arthritis, metabolic syndrome, tobacco abuse, chronic kidney disease, dialysis, gastroesophageal reflux). In addition, we analyzed preoperative laboratory parameters: creatinine, blood urea nitrogen, glomerular filtration rate, hemoglobin, arterial blood pressure, bilirubin, alanine aminotransferase, and aspartate aminotransferase.

Regardless of surgical approach, construction of jejunojejunal anastomosis was similar in all groups. A 60-mm linear stapler was used for division of the jejunum 50 cm distal to ligament of Treitz (LoT), and was used for stapled anastomosis of a 50-cm bilipancreatic limb (BPL) and a 150-cm Roux limb. Closure for the common opening of the jejunojejunal anastomosis (JJ) was completed with the same stapling device for all laparoscopic approaches. With the robotic approach, the enterotomy was hand-sewn closed. Closure of the mesenteric defect was implemented to prevent internal herniation in the TRRYGB, LRYGB-LS, and LRYGB-HS groups. The greater omentum was divided vertically for the placement of an antecolic antegastric Roux limb; a 15–30 ml pouch was formed with a 60-mm or 3.5-mm height linear stapler. Method of the GJ formation is described below. The diameter of gastrojejunostomy was approximately 15–20 mm regardless of surgical technique used. All cases underwent intraoperatively endoscopy to rule out leaks and to verify the anastomosis diameter.

To make the GJA by linear stapler, the proximal Roux limb was affixed to the posterior wall of the gastric pouch by a running 2-0 absorbable stitch. After making a small gastrotomy and an enterotomy, a blue load Endo GIA linear stapler (Ethicon, Cincinnati, Ohio, USA) was inserted to create the posterior wall of anastomosis. The common opening was closed in a double-layer hand-sewn fashion. Hand-sewn GJA was created by running fashion with 2-0 sutures using Endo-Stitch device. To make a GJA by a circular stapler, we utilized the C21 (Medtronic, Minneapolis, Minnesota, USA) for creation GJA at the corner of the two first staple lines used to create the gastric pouch.

Robotic surgeries were performed with a da Vinci Si robotic platform by a single surgeon (TR). Trocar placement and initial survey of operating field was completed laparoscopically. The robot was then docked. The primary surgeon operated the robot from a single console. An assistant remained at the bedside to aid with retraction and firing of the laparoscopic staplers. The common enterotomy in JJ anastomosis was closed by two layer running sutures. GJA in this group was robot-sewn with a 2-0 absorbable stitch.

Medical records were evaluated for operative time, length of hospital stay, 1-year readmission, and the following adverse events: any intraoperative complications, including intraabdominal bleeding, return to the operating room within 1 year after surgery, and early and intermediate complications including GJA leaks, GJA strictures, marginal ulcers, small bowel obstruction, wound infection, ventral/internal hernia, deep venous thrombosis, pulmonary embolism, urinary tract infection, urinary retention, acute kidney injury, pneumonia, vomiting, dehydration, and abdominal pain. Weight loss was expressed as a percentage total weight loss ((%TWL = 100*[(BMI_current – BMI_baseline)/BMI_baseline])) and a change in BMI (ΔBMI).

The robotic group was matched, on the basis of propensity scores, with the circular stapler group in a 1:1 ratio, with the linear stapler group in a 1:2 ratio and with the hand-sewn group in a 2:1 ratio [12]. These ratios were selected to maximize the matched sample sizes. Propensity scores (i.e., the estimated probability of undergoing robotic surgery on the basis of the preoperative and baseline laboratory values listed above) were estimated for each patient by use of logistic regression. All the pre-specified potential confounding variables were considered; however, variables were excluded from each propensity score model if either the amount of missing data was unacceptable (race and glomerular filtration rate) or comorbidities present in less than five patients per surgery group. After randomly ordering the observations in the dataset, we used a sequential distance matching algorithm to match, each patient in the robotic group to patient(s) undergoing the alternative surgery types, restricting successful matches to those for whom the propensity scores did not differ by more than 0.05 units. The subsets of matched patients were used for all subsequent analyses. Balance on potential confounding variables between the matched robotic group and the EEA stapler, linear stapler or hand-sewn group were evaluated with standard univariable summary statistics and absolute standardized difference scores (absolute value of difference in means or proportions, divided by a combined estimate of standard deviation among the groups being compared). Variables with an absolute standardized difference score of less than 0.20 were considered adequately balanced [13]. Linear mixed models were developed for estimation of the relationships between type of surgery and total operative time, length of hospital stay, and change in percent total weight loss and change in BMI. Generalized estimating equation models were developed for estimation of the relationships between type of surgery and occurrence of intraoperative complication and postoperative complications within 6 months of surgery. All models accounted for the clustering of the propensity score matched patients and assumed an exchangeable correlation structure. A significance level of 0.05 was employed. SAS statistical software version 9.3 (SAS Institute, Cary, NC, USA) for 64-bit Microsoft Windows machine was used for all statistical analyses.

After propensity score matching, the median follow-up was 193 (57, 362) days in TRRYGB and 368 (174, 719) days in LRYGB-21CS group (P < 0.001). The mean age of patients in TRRYGB group (n = 82) was 43.0 ± 11.3 years, and the mean age of patients after LRYGB-21CS (n = 82) was 42.0 ± 11.8 years. The mean BMI in TRRYGB group was 48.5 ± 7.0, whereas the mean BMI in LRYGB-21CS group was 48.0 ± 7.0. There was no statistical difference between groups regarding comorbidities, laboratory results, and smoking status (Supplemental Table 1).

Total operative time was significantly longer in TRRYGB group (204 ± 36 min. versus 139 ± 30, P < 0.001). The rate of intraoperative complications was 0% in TRRYGB and 1% in LRYGB-21CS and included intraabdominal bleeding. The mean length of hospital stay was significantly shorter in TRRYGB group in comparison to LRYGB-21CS group (2.6 ± 1.2 vs. 4.3 ± 5.5 days, P = 0.008). The readmission rate was significantly lower in TRRYGB group (12 vs. 28%, P = 0.009). TRRYGB patients reported less vomiting episodes then those after LRYGB-21CS (15 vs. 33%, P = 0.005). The incidence of anastomosis stricture was significantly lower in TRRYGB patients (2 vs. 17%, P = 0.003). The mean %TWL after 1 year was significantly lower in TRRYGB patients (− 20.7 ± 10.6 vs. − 25.1 ± 10.6, P = 0.02). The mean BMI change after surgery was significantly lower in TRRYGB patients (− 10.3 ± 5.3 vs. − 12.3 ± 5.2, P = 0.01). The incidence of marginal ulcer was not statistically significant (12 vs. 23%, P = 0.09). The incidence of marginal ulcer was not statistically significant, but possibly clinically important (12 vs. 23%, P = 0.09) (Table 1).

After propensity score matching, the median follow-up was 205 (57, 365) days in TRRYGB and 299 (135, 589) days in LTYGB-LS group (P < 0.001). The mean age of patients in TRRYGB group (n = 79) was 43.4 ± 11.9 years, and the mean age those who underwent LRYGB-LS (n = 150) was 44.3 ± 12.0 years. The mean BMI of patients in TRRYGB group was 48.7 ± 7.2, whereas the mean BMI in LRYGB-LS group was 48.5 ± 7.0. There was no statistical difference between groups regarding comorbidities, laboratory results, and smoking status (Supplemental Table 2).

Total operative time was significantly longer in TRRYGB group (206 ± 37 min vs. 158 ± 30 min, P < 0.001). The rate of intraoperative complications was 0% in TRRYGB and 2% in LRYGB-LS. There was no statistical difference in mean length of hospital stay in TRRYGB (2.5 ± 1.1 vs. 2.8 ± 1.0 days, P − 0.06). The readmission rate was comparable between TRRYGB and LRYGB-LS (13 vs. 11%, P = 0.64). No statistical difference was observed in the incidence of anastomosis leak, anastomosis stricture, and other complications. There was no significant difference in %TWL and ΔBMI at the 1-year follow-up (Table 2).

After propensity score matching, the median follow-up was 198 (84, 365) days in TRRYGB and 214 (114, 372) days in LRYGB-HS group (P = 0.11). The mean age of patients in TRRYGB group (n = 41) was 46.0 ± 12.3 years, and the mean age in LRYGB-HS group (n = 21) was 46.5 ± 13.6 years (P = 0.89). The mean BMI of patients in the TRRYGB group was 46.3 ± 7.4, whereas the mean BMI in LRYGB-HS group was 45.4 ± 5.6 (P = 0.66). There were no differences in comorbidities, laboratory results, and smoking status (Supplemental Table 3).

Total operative time was significantly longer in TRRYGB group (210 ± 36 min vs. 167 ± 30 min, P < 0.001). The rate of intraoperative complications was 2% in TRRYGB and 10% in LRYGB-LS but based on small sample sizes (P = 0.32). There was no difference in mean length of hospital stay in TRRYGB (2.9 ± 1.6 vs. 3.6 ± 2.9 days, P = 0.25).The readmission rate was comparable between TRRYGB and LRYGB-HS groups (15 vs. 14%, P = 0.97). No statistical differences were observed in the incidence of anastomosis leak, anastomosis stricture, and other complications. There was no significant difference in %TWL and ΔBMI at the 1-year follow-up (Table 3).