Objective
To compare surgical outcomes for robotic vs laparotomy staging in obese endometrial cancer patients.
Study Design
This was a retrospective cohort study of patients with body mass index ≥30 kg/m 2 staged in a community gynecologic oncology practice. Patients undergoing robotic staging were compared with historic laparotomy controls.
Results
One hundred twenty-nine patients underwent robotic staging, compared with 110 laparotomy patients. The robotic cohort had fewer abdominal wound complications (13.9% vs 32.7%, P < .001), but more vaginal cuff complications (4.7% vs 0%, P = .032). Blood loss was lower in the robotic group ( P < .001), as was length of stay ( P < .001). Surgical times were longer in the robotic group ( P < .001). There was no difference in terms of percentage of patients undergoing pelvic or paraaortic lymph node dissection.
Conclusion
Robotic staging for endometrial cancer is feasible in obese women, with fewer abdominal wound complications, but more vaginal cuff complications.
Endometrial cancer is the most common gynecologic malignancy in the United States, and is strongly associated with obesity. Comprehensive surgical staging is the standard for initial treatment of endometrial cancer in medically operable patients. However, surgical management of obese patients presents unique technical challenges. Several studies have demonstrated the overall feasibility of robot-assisted surgical staging for endometrial cancer, as well as the benefits of robotics compared with laparotomy: lower estimated blood loss (EBL), shorter length of stay (LOS), fewer postoperative complications, with equivalent or improved lymph node yields. However, only 2 prior studies have directly compared outcomes for robot-assisted surgical staging vs laparotomy in obese women (body mass index [BMI] >30 kg/m 2 ). Both of these studies were performed in academic institutions.
Conventional laparoscopic staging has been associated with decreased surgical complications compared with laparotomy in obese endometrial cancer patients. However, these studies demonstrated relatively high conversion rates. In the Gynecologic Oncology Group (GOG) LAP2 trial, the conversion rate was 25.8% overall, with likelihood of conversion increasing in proportion to BMI. This was attributed to the challenges of operating on obese patients in terms of visualization/exposure and positioning. The stereoscopic optics and wristed instrumentation of robotic surgery offer the benefits of improved surgeon visualization and ergonomics, thereby decreasing the perioperative complications and conversion rate, whereas increasing the chance for a successful minimally invasive surgery in the obese patient.
The primary objective of this study was to compare perioperative complications, particularly wound complications, in robot-assisted vs open staging via laparotomy in obese women with endometrial cancer operated on in a community gynecologic oncology practice. A secondary objective was to evaluate measures of surgical precision, such as lymph node yield, in this population.
Materials and Methods
This was a retrospective cohort study of obese women (BMI ≥30 kg/m 2 ) undergoing surgical staging for endometrial cancer in a single community-based gynecologic oncology practice in Portland, OR, consisting of 6 surgeons. Institutional review board approval was obtained from Legacy and Providence Health Systems in Portland, OR. BMI was calculated as [weight (kg)/ height (m) 2 ] or [weight (lb)/height (in) 2 ] × 703. Height and weight data were taken from the preoperative history and physical, or anesthesia preoperative intake form.
Robotic cases were collected from the time of introduction of robotic surgery into this practice, June 2007 through February 2010. Robotic cases were performed by all 6 gynecologic oncologists in the practice, though individual surgeons began performing robotic surgeries at varying times, ranging from June 2007 to April 2008. The robotic cohort was compared with consecutive historical controls undergoing staging via laparotomy by the same group of surgeons from January 2005 through January 2010. After the introduction of robotic surgery into the practice, an open approach was chosen primarily when uterine size was thought to be too large (width >10 cm by ultrasound or examination) to be removed vaginally without morcellation. Patient BMI did not significantly influence surgical approach after the introduction of robotics. Demographic data were collected and analyzed. Measures included BMI, age, grade, stage, number of medical comorbidities, and prior abdominal surgeries. Inpatient charts from the patients’ index surgery and hospitalization were reviewed, as well as inpatient records for subsequent readmissions or reoperations through a period of 6 months postoperatively. Outpatient records were reviewed through a period of 2 months postoperatively.
The primary outcomes were wound-related complications. Wound complications were divided into abdominal (eg, cellulitis requiring antibiotics, subcutaneous soft tissue infection, pelvic or incisional abscess, wound separation requiring packing, fascial dehiscence, incisional hematoma or seroma, or postoperative intraabdominal bleeding) and vaginal cuff-related (eg, cuff cellulitis, seroma, hematoma, and dehiscence). Other complications collected included thromboembolic events (eg, myocardial infarction, pulmonary embolus, deep venous thrombosis, stroke, or transient ischemic attack), nonwound related infectious complications (eg, pneumonia, sepsis), intraoperative injuries (eg, nerve, vascular, bowel, bladder, or ureteral injuries), and other major complications (eg, ileus, small bowel obstruction, acute renal failure, requirement of blood transfusion, intensive care unit [ICU] admission, hospital readmission within 30 days postoperatively, death within 30 days postoperatively). Minor complications such as urinary tract infection or brief urinary retention requiring catheterization were not counted. Power calculation was based on prior results from Seamon et al, which showed overall complication rates of 11% and 27% (odds ratio [OR], 0.29; 95% confidence interval [CI], 0.13–0.65) for robotic and laparotomy staging, respectively. To detect a similar difference with 80% power, α = .05, our sample size required would be 106 per group, presuming equal samples.
Secondary outcomes data included lymph node yield, measured as the number of pelvic, paraaortic, and total lymph nodes, as well as the percentage of patients within each cohort undergoing pelvic and/or paraaortic lymph node dissection. The decision to perform complete staging with paraaortic sampling was determined by the individual surgeon, based on assessment of disease grade, intraoperative assessment of depth of invasion, and surgical risk. To adjust for the fact that surgeons may vary in performance of paraaortic lymph node dissection in patients with grade 1 disease, we also calculated the above outcomes for patients with grade 2 or 3 histology only with the knowledge that these patients were routinely staged with both paraaortic and pelvic lymph node dissections.
Other data collected included operating times, EBL, LOS defined as number of nights spent in the hospital, and rate of conversion to laparotomy for robotic cases. Operative times collected included total operating room time (total minutes from entering to exiting the operating room), as well as surgical time (total minutes from first skin incision or insertion of uterine manipulator, to completion of skin closure). Robotic conversions were defined as cases where the robot was undocked and laparotomy was performed to complete the procedure. Robotic cases requiring conversion to laparotomy were analyzed as part of the robotic cohort on an intent-to-treat basis.
Surgical procedure
Robotic procedures were all performed on the da Vinci S or Si systems (Intuitive Surgical, Sunnyvale, CA) using 4 arms, with monopolar scissors, fenestrated bipolar forceps, and atraumatic graspers. Uterine manipulators used were the Rumi with Koh Colpotomizer, and the Colpo Probe (Apple Medical, Denison, TX). Colpotomy was performed using the monopolar scissors. The vaginal cuff was closed with 0 or 2-0 vicryl with interrupted figure-of-8 or running stitches. Initially, Lapra-ty absorbable clips (Ethicon Endo-Surgery, Cincinnati, OH) were also commonly used in robotic vaginal cuff closures; however, currently only 1 surgeon uses these clips. All patients received preoperative prophylactic antibiotic coverage with a second-generation cephalosporin, or clindamycin for those with significant penicillin or cephalosporin allergy. Pelvic and paraaortic lymphadenectomy were performed according to GOG guidelines in terms of extent of dissection.
Continuous variables were analyzed using unpaired t test and Wilcoxon rank sum for comparison of means, depending on whether the data was normally distributed. Fisher exact test was used for comparison of proportions. In all cases, P ≤ .05 was used to determine statistical significance. Linear regression analysis was performed to control for confounding variables. Statistical analyses were performed using R version 2.11.1 software ( R Foundation for Statistical Computing, Vienna, Austria).
Results
From June 2007 through February 2010, a total of 129 obese patients underwent robotic surgical staging for endometrial cancer. These were compared with a cohort of 110 patients undergoing staging via laparotomy from January 2005 through January 2010. The groups did not differ significantly in terms of any of the demographic variables: age, BMI, grade, stage, number of medical comorbidities, or prior abdominal surgeries. Data for histologic grade were not provided for 1 laparotomy and 2 robotic patients and were omitted from Table 1 . Mean BMI for the robotic and laparotomy groups were similar – 39.8 (SD 7.9) and 40.3 (SD 8.6) respectively ( P = .63) ( Table 1 ).
| Demographic | Robotic staging (n = 129) | Laparotomy (n = 110) | P value |
|---|---|---|---|
| Age y, mean (SD) | 59.8 (10.6) | 58.5 (9.9) | .33 a |
| BMI kg/m 2 , mean (SD) | 39.8 (7.9) | 40.3 (8.6) | .63 a |
| No. medical comorbidities, % | |||
| 0 | 23 (17.8) | 29 (26.4) | .44 b |
| 1 | 60 (46.5) | 41 (37.3) | |
| 2 | 36 (27.9) | 31 (28.2) | |
| 3 | 9 (7.0) | 7 (6.4) | |
| 4 | 1 (0.8) | 2 (1.8) | |
| Patients with prior abdominal surgeries, % | 76 (58.9) | 59 (53.6) | .43 b |
| Grade, % | |||
| 1 | 78 (60.5) | 53 (48.2) | .11 b |
| 2 | 33 (25.6) | 41 (37.3) | |
| 3 | 16 (12.4) | 15 (13.6) | |
| Stage, % | |||
| I | 110 (85.3) | 90 (81.8) | .77 b |
| II | 10 (7.8) | 10 (9.1) | |
| III | 6 (4.7) | 8 (7.3) | |
| IV | 2 (1.6) | 1 (0.9) |
a P value based on 2-sample t test;
EBL was significantly lower for the robotic group (160 ± 150 cc vs 292 ± 226, P <.001). LOS was also significantly lower in the robotic group (1.5 ± 1.0 days vs 4.1 ± 2.2, P < .001). Operative times were significantly longer for the robotic group (operating room time 237 ± 65 min vs 177 ± 50, surgical time 188 ± 63 vs 128 ± 39, P < .001) ( Table 2 ).
| Variable | Robotic staging (n = 129) | Laparotomy (n = 110) | P value |
|---|---|---|---|
| EBL mL, mean (SD) | 160 (150) | 292 (226) | < .001 a |
| OR time, min, mean (SD) | 237 (65) | 177 (50) | < .001 b |
| Incision time, min, mean (SD) | 188 (63) | 128 (39) | < .001 b |
| LOS, d, mean (SD) | 1.5 (1.0) | 4.1 (2.2) | < .001 a |
a P value based Wilcoxon rank-sum test;
The robotic group had a significantly lower rate of abdominal wound complications (13.9% vs 32.7%, P < .001), but a higher number of vaginal cuff complications (4.7% vs 0%, P = .032), including 2 cuff dehiscences ( Table 3 ). In the 2 cases of cuff dehiscence, the cuff closure was performed using 0-vicryl suture. Interrupted figure-of-8 stitches were used in 1 case, the other operative report did not specify a closure technique.
| Variable | Robotic staging (n = 129) | Laparotomy (n = 110) | P value |
|---|---|---|---|
| Abdominal wound complications, % | 18 (13.9) | 36 (32.7) | < .001 |
| Cellulitis | 7 | 8 | |
| Soft tissue infection | 2 | 7 | |
| Wound separation | 2 | 9 | |
| Abscess | 6 | 6 | |
| Incisional seroma | 1 | 2 | |
| Incisional hematoma | 0 | 1 | |
| Fascial dehiscence/hernia | 0 | 1 | |
| Intraabdominal bleeding | 0 | 2 | |
| Cuff wound complications total | 6 (4.7) | 0 | .032 |
| Cuff cellulitis | 1 | 0 | |
| Cuff seroma | 1 | 0 | |
| Cuff hematoma | 2 | 0 | |
| Cuff dehiscence | 2 | 0 | |
| Thromboembolic events total | 2 (1.5) | 4 (3.6) | .418 |
| Stroke | 1 | 0 | |
| DVT | 0 | 1 | |
| PE | 1 | 3 | |
| Nonwound related infections total | 2 (1.6) | 4 (3.6) | .418 |
| Pneumonia | 1 | 3 | |
| Sepsis | 1 | 1 | |
| Perioperative injuries total | 9 (7.0) | 7 (6.4) | > .99 |
| Genitourinary injury | 3 | 5 | |
| Gastrointestinal injury | 5 | 2 | |
| Vascular injury | 1 | 0 | |
| Other | 17 (13.2) | 40 (36.4) | > .001 |
| Transfusion | 6 | 8 | |
| Renal failure | 1 | 2 | |
| Hospital readmission <30 d postoperative | 5 | 8 | |
| ICU admission | 2 | 12 | |
| Ileus | 2 | 9 | |
| Small bowel obstruction | 1 | 1 |
There was no significant difference in frequency of thromboembolic events, nonwound related infections, or perioperative gastrointestinal or genitourinary injuries. In the robotic arm, there were 3 genitourinary injuries (1 ureteral injury, 1 cystotomy, and 1 ureterovaginal fistula), and 5 gastrointestinal injuries (4 enterotomies, and 1 rectovaginal fistula). In the laparotomy arm, there were 5 genitourinary injuries (3 ureteral injuries, 2 cystotomies), and 2 gastrointestinal injuries (1 enterotomy and 1 rectovaginal fistula). There were also a higher number of other complications in the laparotomy group, including requirement for ICU admission and postoperative ileus. These remained independently significant when using regression analysis to account for potential confounders in the demographics.
There was no difference in terms of percentage of patients undergoing pelvic (75.2% robotic vs 74.5% open, P > .999) or paraaortic (38.8% vs 36.4%, P = .79) lymph node dissection. The number of pelvic lymph nodes collected in the robotic group was higher than that in the laparotomy group in terms of statistical significance, though the absolute difference was small (10.7 ± 8.8 vs 8.7 ± 8.7, P = .03). There was no statistical difference in number of paraaortic (2.4 ± 3.9 vs 2.0 ± 3.4, P = .70), and total lymph nodes obtained (13.0 ± 11.3 vs 10.7 ± 11.3, P = .06). In the patients with grade 2 or 3 disease only, the percentage of patients successfully undergoing any pelvic (91.8% vs 82.1%, P = .16) or paraaortic lymph node dissection (55.1% vs 41.1%, P = .17) was still not significantly different. However, the number of pelvic (13.1 ± 8.3 vs 8.6 ± 7.9, P = .002) and total lymph nodes (16.1 ± 10.2 vs 10.7 ± 10.6, P = .002) was significantly greater in the robotic cohort. The number of paraaortic nodes (2.9 ± 3.8 vs 2.1 ± 3.4, P = .31) was equal between groups ( Table 4 ).
