Introduction
Robotic proctectomy has become an increasingly popular approach for rectal surgery. Compared with traditional laparoscopy, robotic assistance offers high-definition 3D vision, improved surgeon ergonomics, and a greater range of motion with instrument wrist articulation. This theoretically results in more precise and dexterous movements, particularly relevant in pelvic surgery given the deep and narrow working space. Patients associate robotics with smaller incisions, reduced infections, and greater precision [
1].) The demand for robotic surgery continues to increase, and with it, the importance of understanding its impact on outcomes in different patient populations.
In previous studies, robotic proctectomy, as part of either low anterior resection (LAR) or abdominoperineal resection (APR), has been associated with decreased conversion to open surgery and decreased hospital length of stay (LOS) compared to laparoscopy [
2]. However, it has also been associated with increased operative duration and as much as a 31% increased cost compared to open or laparoscopic proctectomy [
3]. Therefore, controversy persists regarding the value of robotics. Given this controversy, it is imperative to study the impact of robotic surgery in specific patient populations to understand its differential impact and derive value through careful patient selection.
Obese patients are a specific population that may particularly benefit from the advantages of the robotic platform given the ability of the robot to offload the cognitive load of exposure from the surgeon to the robot, which would allow the surgeon to direct greater focus to precise dissection, resection, and reconstruction. This is especially important in the pelvis where exposure is challenging with any approach. As a result, robotic proctectomy in obese patients may lead to improved intraoperative and post-operative outcomes, potentially justifying increased costs and operative times. Previous BMI-based subgroup analyses suggested that decreased conversion associated with robotic proctectomy is meaningful in overweight and obese patients, but these analyses have been underpowered to draw definitive conclusions [
4]. Studies that investigated robotic proctectomy in obese rectal cancer patients demonstrated marginal benefits in post-operative morbidity, but these studies were limited by small sample size [
5,
6]. Thus, evidence supporting the role of robotic proctectomy for obese patients remains incomplete. We, therefore, sought to compare outcomes of robotic versus laparoscopic proctectomy, focusing on overweight and obese patients using a large, contemporary, and nationally representative dataset.
Materials and methods
A retrospective analysis of the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) database was conducted using 2016–2018 participant and targeted proctectomy data. NSQIP data are prospectively collected and validated by trained surgical clinical reviewers. We included patients ≥ 18 years old who underwent elective robotic or laparoscopic proctectomy, which included LAR and APR. We included patients who underwent open proctectomy as a single analysis comparing outcomes in this group to those among patients who had been converted from a minimally invasive approach. This study was approved by the Institutional Review Board of the University of California, San Francisco (UCSF): study number 18–26,677.
Preoperative and operative characteristics
Preoperative characteristics included age, sex, race/ethnicity, body mass index (BMI), American Society of Anesthesiology (ASA) class, functional status, and diagnosis [cancer, benign tumor, diverticular disease, inflammatory bowel disease (IBD), vascular, and other]. BMI was categorized as underweight (< 18.5), normal weight (18.5–25), overweight (25–30), obese (30–40), and morbidly obese (> 40) according to Centers for Disease Control and Prevention (CDC) guidelines. Comorbidities included hypertension, diabetes (requiring hypoglycemic medications or insulin), smoking status, congestive heart failure (CHF), chronic obstructive pulmonary disease (COPD), renal failure, bleeding disorder, and chronic steroid use. Proctectomy-specific variables included prior chemotherapy, prior radiation therapy, clinical tumor stage, and location in the rectum for rectal cancer. Complications present before surgery (sepsis/septic shock; superficial, deep, and organ space surgical site infections [SSI]; ventilator dependence; pneumonia; urinary tract infection [UTI]) were included as predictors and used to adjust post-operative complications as defined by NSQIP.
Operative factors included the approach (open, laparoscopic, or robotic), wound class, and whether a blood transfusion (intraoperative or within 72 h of surgery) was received. For rectal cancer patients, clinical T stage and tumor location (upper third [> 10 cm from anal verge], middle third [5–10 cm from anal verge], and lower third [< 5 cm from anal verge]) were also included.
Outcomes
The primary outcome was unplanned conversion to open surgery. Secondary outcomes included case duration, intraoperative or post-operative blood transfusion, LOS in cases without post-operative complications, unplanned readmission, unplanned reoperation, death within 30 days, anastomotic leak, ileus (determined by prolonged NPO status or need for nasogastric tube placement post-operatively), SSI, sepsis/ septic shock, wound dehiscence, pneumonia, reintubation or failure to wean from ventilator > 48 h, pulmonary embolism, deep vein thrombosis (DVT), acute renal insufficiency, UTI, Clostridium difficile infection, stroke, cardiac arrest, and myocardial infarction. All anastomotic leaks were included as organ space SSIs, but they were also separately counted based on positive identification on imaging (presence of extraluminal air–fluid levels or leakage of enteric contrast at the anastomosis) or specific notation by the surgeon. A composite morbidity outcome was created based on the occurrence of any of these complications. Margin status (distal and radial) was assessed for rectal cancer patients.
Secondary outcomes were compared between robotic and laparoscopic proctectomies using intention-to-treat with regard to approach: converted cases were categorized as the intended minimally invasive approach. Additionally, secondary outcomes were compared between patients who underwent either robotic or laparoscopic proctectomy converted to open, collectively termed minimally invasive converted to open, and planned open proctectomy.
Statistical methods
Patient characteristics and post-operative outcomes were compared between the laparoscopic and robotic cohorts using t test for continuous variables and the Chi-square test for categorical data. Multivariable logistic models were created for all categorical outcomes using all available covariates without model selection or training sets. Covariates included year of surgery, primary diagnosis, age category, sex, race/ethnicity, BMI category, smoking status, ASA class, functional status, diabetes, hypertension requiring medications, chronic obstructive pulmonary disease (COPD), congestive heart failure (CHF), renal failure/dialysis, presence of ascites, distant cancer, steroid use, recent weight loss, bleeding disorder, preoperative transfusion, preoperative sepsis or ventilator, tumor location, neoadjuvant chemotherapy or radiation, and wound class. Linear, Gamma log-linked, Poisson, and log-transformed multivariable models to evaluate operative time and LOS were compared to account for right-skewness. Interaction terms between variables of interest were added to models based on clinical context. Subgroup regression analyses were performed by BMI category, diagnosis, and conversion status. Multicollinearity in regression models was assessed using generalized variance inflation factors. Missing categorical data were treated as separate categories. Analysis was performed using R version 4.0.2 (R Foundation for Statistical Computing).
Discussion
Robotic proctectomies represented 30% of our cohort, which aligns more closely with modern practice patterns compared to previously studied cohorts in which only 10% of patients underwent robotic proctectomy.(7) This study is also larger than any other contemporary analysis of robotic proctectomy patients [
2]. Given the rapid uptake and expansion of robotic proctectomy, detailed contemporary analysis of its associated morbidity and value are essential. Furthermore, given the size of our cohort, we were able to evaluate outcomes in BMI-based subgroups.
Our study demonstrates that patients undergoing robotic compared to laparoscopic proctectomy were more likely to be older, male, obese, and have higher ASA class. Consistent with prior studies, robotic proctectomy was associated with increased operative time in our cohort [
2,
8,
9]. However, the mean operative time for robotic proctectomy in our cohort was only 21 min longer than the mean operative time for laparoscopic proctectomy when adjusted for covariates. The decreased difference in mean operative time between minimally invasive approaches using contemporary data suggests that proficiency in performing robotic proctectomies may be increasing and that the difference in operative time compared to the laparoscopic approach may become clinically insignificant over time.
We demonstrated a significant difference in the conversion rate of robotic compared to laparoscopic proctectomy (5.1% vs 12.3%, respectively), an advantage maintained after multivariable adjustment. This reaffirms findings from other proctectomy-specific studies and aligns with recent conversion rates reported [
2,
7,
9,
10]. Of note, the ROLARR randomized trial did not find a significant difference in conversion rates between robotic and laparoscopic surgery for patients with rectal cancer; however, its results likely no longer reflect current practice patterns as enrollment closed in 2014 and participating surgeons had far greater experience using the laparoscopic rather than the robotic platform [
11]. Patients converted to open surgery in our study were more likely to be male, obese, older, and have more comorbidities. As these are the same factors that were more common among patients undergoing the robotic approach, it suggests that surgeons were selecting patients for the robotic approach who were at greatest risk of conversion and that, despite this selection, multivariate analysis still demonstrated that the conversion rate was significantly lower in patients undergoing robotic proctectomy.
In our study, the significantly lower conversion rate of robotic compared to laparoscopic proctectomy was maintained in normal, overweight, and obese subgroups; however, the difference in the conversion rate favoring the robotic approach was greater for the obese subgroup (10.9%) than for either the overweight (6.6%) or normal weight (5.6%) subgroups. This suggests that obese patients derive particular benefit from robotic compared to laparoscopic proctectomy, perhaps due to the weight of an obese abdominal wall being offloaded on the robot rather than on the surgeon. The lower conversion rate conferred by robotic surgery allows more medically complex patients to successfully undergo minimally invasive surgery. It also likely explains the significantly lower LOS in patients undergoing a robotic approach, and may explain the differences in post-operative pneumonia, DVT, and acute renal insufficiency favoring the robotic group.
Surgeons may choose a minimally invasive approach without regard to the likelihood of success, planning to convert to open if needed, with the rationale that conversion to an open operation poses the same risks as a planned open operation. However, we found that conversion has consequences. In comparing patients who underwent conversion from a minimally invasive approach to patients who underwent planned open surgery using an intention-to-treat analysis, converted patients had significantly increased composite morbidity. This increased composite morbidity is multifactorial but appears largely driven by increased rates of organ space surgical site infections, at least in part due to increased anastomotic leak rates. Patients undergoing proctectomy require thoughtful and individualized operative planning with consideration of the comorbid or anatomical factors that may increase risk of conversion and careful selection for planned open surgery. When a minimally invasive approach is considered for obese patients in particular, the robotic platform may increase the likelihood of success.
Use of the robotic approach for rectal cancer operations has continued to grow. Given that laparoscopic proctectomy failed to reach non-inferiority compared to open proctectomy for pathologic outcomes in rectal cancer, as demonstrated by the ACOSOG Z6051 trial [
12], and that robotic proctectomy has not been directly compared to the open approach, ongoing analysis of outcomes for robotic rectal cancer surgery is imperative. Several studies have demonstrated no difference in oncologic outcomes between robotic and laparoscopic proctectomy for rectal cancer, but these did not analyze outcomes by BMI subgroup [
9,
13‐
16].
In our study, when we examined proctectomies done for rectal cancer in obese patients (controlling for T stage), we identified a significantly increased risk of positive radial margins among patients who underwent robotic versus laparoscopic proctectomy. This could be the result of the preferred use of the robotic approach for lower rectal tumors, especially in this cohort of patients. In obese patients, mesorectal adipose tissue obscures natural tissue planes and can make an adequate dissection in the deep pelvis particularly challenging. It may also be that the most difficult cases are performed at tertiary-care referral centers that are likely to have robotic capabilities. As the robotic approach becomes more popular, comparing oncologic outcomes to the gold standard of open proctectomy in obese patients will be essential [
13,
15]. Unfortunately, our analysis is limited to 30-day post-operative outcomes and an oncologic analysis would be best addressed with longer term data not available in NSQIP.
In addition to lack of long-term oncologic data, an important limitation of this study is the selection bias inherent in the nature of retrospective analyses and the clinical heterogeneity of our study population. While NSQIP data are rigorously collected by certified reviewers, it is limited to the specific variables being collected and excludes reporting hospital type, surgeon experience, use of diverting ostomy, and neoadjuvant chemotherapy regimen [
17]. Especially with regard to conversions, the lack of granular data pertaining to the experience of the surgeon in performing laparoscopic or robotic proctectomy and the hospital type (academic, high-volume robotic institution, and community practice) limits interpretation of our results. While all surgeons performing robotic operations must receive training in the form of online modules, time spent on a practice counsel with tissue training, and in-person proctoring from Intuitive Surgical (manufacturer of the Da Vinci robotic system), there is variability in terms of the specialty training of the individual surgeon and prior robotic experience. We were also unable to exclude centers without robotic capabilities, as NSQIP does not indicate which included centers have surgical robots available; as a result, some of the laparoscopic cases were inevitably performed via this approach simply because a robotic approach was not an option. A prospective trial controlling for these variables is warranted to more definitively determine if there is benefit of the robotic approach relative to either the open or laparoscopic approach, particularly in obese patients at higher risk of surgical morbidity.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.