- Email: [email protected]
Robotic Esophagectomy for Cancer: Early Results and Lessons Learned
Author's Accepted Manuscript
Robotic Esophagectomy for Cancer: Early Results and Lessons Learned Robert J Cerfolio MD, MBA, FACS, FCCP, Benjamin Wei MD, Mary T Hawn MD, Douglas J Minnich MD
www.elsevier.com/locate/buildenv
PII: DOI: Reference:
S1043-0679(15)00165-3 http://dx.doi.org/10.1053/j.semtcvs.2015.10.006 YSTCS784
To appear in:
Semin Thoracic Surg
Cite this article as: Robert J Cerfolio MD, MBA, FACS, FCCP, Benjamin Wei MD, Mary T Hawn MD, Douglas J Minnich MD, Robotic Esophagectomy for Cancer: Early Results and Lessons Learned, Semin Thoracic Surg, http://dx.doi.org/10.1053/j.semtcvs.2015.10.006 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Robotic Esophagectomy for Cancer: Early Results and Lessons Learned Robert J Cerfolio, MD, MBA, FACS, FCCP Benjamin Wei MD Mary T Hawn MD Douglas J Minnich MD Address correspondence to: Robert J Cerfolio, MD, MBA Division of Cardiothoracic Surgery Chief of Thoracic Surgery University of Alabama at Birmingham 703 19th St S, ZRB 739 Birmingham, AL 352094 Tel: 205-996-7561 Fax: 205-934-6218 Email : [email protected] From the Division of Thoracic Surgery at University of Alabama at Birmingham
Keywords: esophagus, esophageal cancer, esophageal surgery, robotics, minimally invasive surgery Funding: None Disclosures: Robert J. Cerfolio - proctor for Intuitive Surgical; consultant for Ethicon and Community Health Systems (CHS)
Abbreviations
ASA – American Society of Anesthesiologists classification COPD – chronic obstructive pulmonary disease DLCO – diffusion of carbon monoxide ECOG – Eastern Cancer Oncology Group FES – fiberoptic evaluation of swallowing FEV1 – forced expiratory volume in 1 second MIE – minimally invasive esophagectomy
Abstract Objective: Minimally invasive esophagectomy with intra-thoracic dissection and anastomosis is increasingly performed. Our objectives are to report our operative technique, early results and lessons learned.
Methods: This is a retrospective review of 85 consecutive patients who were scheduled for minimally invasive Ivor Lewis esophagectomy (laparoscopic or robotic abdominal and robotic chest) for esophageal cancer.
Methods: Between 4/2011 and 3/2015, 85 (74 men, median age 63) patients underwent robotic Ivor Lewis esophageal resection. Sixty-four patients (75%) had preoperative chemo-radiotherapy, 99% had esophageal cancer and 99% had an R0 resection. There were no abdominal or thoracic conversions for bleeding. There was one abdominal conversion for the inability to completely staple the gastric conduit. The mean operative time was 6 hours, median blood loss was 35 ml (no intraoperative transfusions), median number of resected lymph nodes was 22 and median length of stay was 8 days. Conduit complications (anastomotic leak or conduit ischemia) occurred in 6 patients. The 30-day and 90-day mortality were
3/85 (3.5%) and 9/85 (10.6%), respectively. Initial poor results led to protocol changes via root cause analysis: longer rehabilitation prior to surgery, liver biopsy in patients with history of suspected cirrhosis, and refinements to conduit preparation and anastomotic technique.
Conclusion: Robotic Ivor Lewis esophagectomy for cancer provides an R0 resection with excellent lymph node resection. Our preferred port placement and operative techniques are described. Disappointingly high thoracic conduit problems and 30 and 90-day mortality led to lessons learned and implementation of change which are shared. Ultra-miniabstract
A review of 85 consecutive patients undergoing robotic Ivor Lewis esophagectomy shows our preferred technique and early results. Root cause analysis led to protocol changes that were implemented in order to improve our 30 and 90-day mortality and other outcomes.
Introduction The use of minimally invasive esophagectomy (MIE) and hybrid esophagectomy has increased recently. The term “minimally invasive” refers to performing both the thoracic and abdominal phases of the operation with laparoscopic, thoracoscopic, or robotic assistance. Hybrid esophagectomy combines minimally invasive with neck approach and open approaches (ex. laparoscopy and thoracotomy, laparotomy and VATS). Recent studies have demonstrated that MIE may be associated with decreased blood loss, chest tube duration, length of stay, and respiratory complications and possibly reduced cost versus open esophagectomy [1] [2],[3].[4],[5]. Previous reports on robotic-assisted Ivor Lewis esophagectomies, have demonstrated feasibility, with perioperative outcomes similar to MIE[6],[7],[8]. Our objective in this study is to report our series and some of the lessons we have learned during the experience.
Methods This is a retrospective cohort study of prospectively collected data evaluating evalua a consecutive series of patients who underwent thoracic robotic minimally invasive Ivor Lewis wis esophagectomy. All patients with a planned Ivor Lewis esophagectomy with a robotic thoracic approach and laparoscopic or robotic abdominal approach were included in the study. Patients with mid or distal esophageal lesions were candidates for Ivor Lewis esophagectomy. Preoperative evaluation included endoscopic ultrasound, esophagogastroduodenoscopy, and integrated PET/CT scan. Patients also typically received pulmonary function testing testing, stress testing and stair climbing. There were no contrain contraindications to offering a robotic approach for Ivor Lewis esophagectomy. Initially, patients underwent planned open abdominal and robotic thoracic approach (“hybrid “approach) for Ivor Lewis esophagectomy during the period of the study study, and were excluded. These patients had a planned open abdominal approach to test feasibility early on in the experience rather than due to patient factors. Root cause analysis (as shown in Figure 1) was used to improve results after adverse events. Information was obtained through hospital databases and medical records. Descriptive statistics were used to estimate the frequency of categorical variables and median of the continuous study variables. Comparisons were done with two two-tailed t tests for continuous variables and either the 2 or the Fisher exact test was used to compare categorical data. The University of Alabama at Birmingham’s Institutional Review Board approved this protocol as well as the prospective database used to collect information for this study. Individual dual consent was waived for inclusion in this study; however it was required and obtained in order to enter patient data in the prospective database.
Operative Details
Laparoscopic Phase with Gastric Conduit Creation The placement of ports for the laparo laparoscopic scopic phase of the operation is shown in Figure 2a. The camera port is located 15 cm inferior to the xiphoid process and 3
cm to the left of midline. The liver retractor may be positioned via a subxiphoid port (grasper or Nathanson retractor) as depicted or a right subcostal port using a Mediflex (Islandia,NY) Positractor with a Lapro-Flex self-forming retractor. A tongue of omentum is preserved during conduit creation order to cover the anastomosis and protect the carina. After division of the short gastric arteries and dissection of the esophagus at the hiatus and into the mediastinum, the left gastric artery is identified, lymph node tissue swept upward, and the vein and artery are divided with a vascular staple load. A Kocher maneuver is not routinely performed. Adequate mobilization is confirmed by demonstrating that the pylorus can reach the diaphragmatic hiatus. Botulinum toxin injection into the pylorus (100 units in 4 ml of saline) with a spinal needle was used as the pyloric drainage procedure. A 4-5 cm gastric conduit is fashioned with multiple gastrointestinal staple loads with the stomach on tension to maximize length of the conduit. The conduit is sutured to the future specimen and positioned into the mediastinum along with a Penrose drain encircling the esophagus. The jejunum is tacked to the abdominal wall with 2-0 vicryl sutures and a 12 French jejunostomy tube is placed with a modified Seldinger technique.
Robotic Thoracic Technique After completion of the abdominal phase, a double lumen endotracheal tube is placed. Patient positioning is shown in Figure 2b. This modified lateral decubitus position allows the lung and blood to fall away from the operative field while avoiding the anesthetic delays of prone positioning. The robot is positioned at a 75 degree angle to the longitudinal axis of the patient’s body, approaching the patient from their back so that the instruments/arms are directed posteriorly.
Thoracic Robotic Port Placement Port placement is shown in Figure 2b. The 8 mm metal trocar for robotic arm 1 is first placed just below the hair bearing part of the axilla in the triangle between the latissimus dorsi and pectoralis major muscles. A paravertebral block using 0.25% bupivicaine and epinephrine is performed under direct vision of a 5
mm VATS camera. A 5 mm metal trocar for robotic arm 3 is inserted next, as posteriorly and inferiorly in the chest as possible. The remaining ports are placed under direct vision with the camera through this posterior port. A 12 mm plastic port for the robotic camera is placed third, located 9 cm away from robotic arm 1 in a line towards the right hip. Robotic arm 2 is placed fourth, 9 cm away from the camera port. It can be either a 5 mm or 8 mm metal trocar. The last port placed is the 12 mm bedside assistant access port, which is triangulated behind the camera port and the port for robotic arm 2 and placed as low as possible to maximize room for the bedside assistant.
Robotic Instruments and Operative Conduct of Thoracic Phase A Cadiere grasper is used for robotic arm 2 (Schertel instead if 5 mm port). A bipolar thoracic dissector (also known as a long curved tipped dissector) is placed in robotic arm 1. The 5 mm thoracic grasper is used for robotic arm 3 and serves mainly as a retractor. The intrathoracic esophagus is mobilized from thoracic inlet to diaphragmatic hiatus. All tissue from the pericardium to the left pleural surface is removed. Both the right and left inferior pulmonary veins are visualized. All lymph nodes are removed using a bipolar instrument that is not hot on either of its outside surfaces. Special care is taken to avoid thermal injury to the airway during the carinal lymph node dissection. We prefer to identify the right main stem bronchus first, then the trachea and then the left main stem bronchus prior to the removal of the entire subcarinal lymph node basin. The azygos vein is divided with a vascular stapler. We do not routinely ligate the thoracic duct. The esophagus along with vagus nerves are divided at the level of the azygos vein for distal esophageal tumors; for mid-esophageal tumors the esophagus is divided higher to ensure a negative margin. The right paratracheal lymph nodes are not routinely removed unless a mid or mid-low squamous cell cancer is the primary lesion. Dissection of the esophagus above the azygos vein is performed close to the esophagus with bi-polar cautery only to avoid injury to the recurrent laryngeal nerve.
Robotic Chest Anastomosis After transection of the esophagus and removal of the specimen, the conduit is tacked posteriorly to the pleura superior to the divided azygos vein and anteriorly to the divided right vagus nerve, which helps line up the anastomosis. A posterior longitudinal gastrotomy is made at least four cm from the tip of the conduit near the greater curvature of the gastric conduit. We now prefer to staple the posterior aspect of the anastomosis using a gastrointestinal 30 mm stapler and complete then anterior part of the anastomosis with continuous 3-0 Vicryl sutures on the inner layer and interrupted 3-0 silk sutures as the outer row. However, if it seems that stapling the posterior wall of the anastomosis would create too much tension or seems unfavorable from an anatomic perspective, a completely hand-sewn anastomosis to a more horizontal gastrotomy can be performed. Once the anastomosis is complete, we buttress the anastomosis anteriorly with a tongue of omentum, interposing it between the conduit and the airway. The conduit is then secured to the diaphragmatic hiatus with silk suture to prevent herniation of abdominal contents. A single 20 French chest tube is placed lying near the anastomosis, and the operation completed.
Postoperative Management Patients are sent to the floor and kept NPO. Tube feeds are started on postoperative day 1. Patients receive a fiber-optic swallow study on post-operative day 4 and if normal a gastrograffin and then barium swallow study are performed later that day. If no anastomotic leak is visualized, clear liquids are started following strict aspiration precautions. Chest tubes are removed once tube feedings are at goal and no chylothorax is observed. All patients are sent home with full nutritional support via the jejunostomy tube.
Results Between 4/2011 and 3/2015, 92 patients underwent consecutive robotic esophagectomy. The first seven patients had their abdominal phase of the operation performed via laparotomy and were excluded from this study. The remaining 85
consecutive patients (74 men, 87%) underwent a minimally invasive esophagectomy. Their preoperative characteristics are shown in Table 1. Mean operating time, defined as the time between first skin incision and closure of last skin incision and including repositioning and docking/undocking of the robot, was 6 hours (360 minutes, range 283-489 minutes). Median blood loss was 35 ml. No patients underwent a blood transfusion during the operation and no patients went back to the operating room for bleeding. The median number of lymph nodes obtained during the surgery was 22. The R0 resection rate was 99% (84/85). One patient (1.2%) required conversion to a thoracotomy because of tumor invasion of the main stem bronchus. Seventy-nine (92.9%) patients underwent laparoscopic gastric conduit mobilization, 5 (5.9%) underwent robotic gastric conduit mobilization, and only 1 patient (1.2%) required conversion from laparoscopy to laparotomy. This occurred after the staple line of the gastric conduit dehisced because the gastric wall was too thick. Median hospital stay was 8 days (interquartile range 7-12 days, range 5-46 days). Morbidity occurred in 31 patients (36.4%) as described in Table 2. Six (7.1%) of patients developed pneumonia, defined as presence of new pulmonary infiltrate on radiograph accompanied by fever, leukocytosis, worsening oxygenation, or cough with productive sputum, or respiratory failure, defined as reintubation and/or requirement for bronchoscopy. Four patients had an anastomotic leak. Of these, the last 3 patients underwent esophageal stent placement. Two patients had gastric conduit necrosis, and were treated with reoperation with revision of the anastomosis and intercostal muscle flap placement. Therefore, a total of 6 patients (7.1%) experienced confirmed anastomotic/conduit complications. One of these 6 patients (16.7%) died within 30-days, and 2 of 6 (33.3%) within 90 days. The overall perioperative mortality rate, defined as mortality within 30 days of surgery, was 3.5% (3 of 85). The in-hospital mortality rate was also 3.5%.. All 3 patients that died had pre-operative chemoradiotherapy. The causes of 30-day mortality were suspected pulmonary embolism, anastomotic leak, and ischemic bowel due to an embolic event. The overall 90 day mortality rate was 10.6% (9 of
85). Of the 6 patients dying between 30 and 90 days, the causes of death was liver failure (2 of 7), pneumonia (2 of 7), urosepsis leading to bacteremia (1 of 7), and unknown(1 of 7). Of the patients who died of liver failure, none of them had issues with postoperative bleeding, sepsis, or leak, and both were readmitted after discharge from the initial postoperative convalescence. We found a 16.6% 30-day mortality and 33.3% 90-day mortality rate in patients who had an anastomotic or conduit complication, which occurred a median of 8 days after the operation (range 4-15 days). The characteristics of patients who died within 90 days after esophagectomy are compared to those of patients who did not die within 90 days in Table 1. Patients who died were more likely to have COPD (p=0.0143) and had a lower FEV1 (p=0.0036). There were no other significant differences in characteristics between the two groups. In our first tercile of esophagectomy patients (n=28), the 30-day mortality rate was 7.1% and 90-day mortality rate was 7.1%. In the next tercile of esophagectomy patients, the 30-day mortality rate was 10.7% and 90-day mortality rate was 17.9%. In the last tercile of patients, the 30-day mortality rate was 0% and 90-day mortality rate was 7.1%. The Kaplan-Meier overall survival curve of patients undergoing robotic esophagectomy is shown in Figure 3. Median follow up time was 27.4 months. Overall survival was 73.1% at 1 year, 59.9% at 2 years, and 54.2% at 3 years.
Discussion Reported series of robotic esophagectomies are shown in Table 3. Our series compares favorably in terms of operative time, intraoperative blood loss, and conversion rate. Our series was similar to prior robotic series in terms of anastomotic/conduit complication rate (7.1%), lymph nodes dissected (median 22), R0 resection rate (99%), median postoperative length of stay (8 days), rate of pneumonia (7.1%), atrial fibrillation (7.1%) and overall major morbidity. When compared to other non-robotic MIE series, our results are fairly similar across the above parameters also [10][9][10][11][12] [13][29]. Our mortality rate is comparable to that from a recent study using the SEER-Medicare database showing
a 6% 30-day mortality and 13.3% 90-day mortality rate in patients undergoing esophagectomy between 2006-2009 [21]. However, our mortality rate is high when compared to Luketich’s series that reported a 30-day mortality of 1.7%[7]. In our initial series of 22 patients, we reported a 0% 30-day and 90-day operative mortality [11]. Subsequently, operative mortality temporarily increased possibly due to the fact that overall patient selection became less stringent for esophagectomy. Ultimately, our overall 30-day mortality rate of 3.5% for roboticassisted Ivor Lewis esophagectomy was comparable to the 30-day mortality rate of 4.5% in our prior open series [20]. The key to any successful process is to evaluate outcomes and modify the process involved to respond to those that are suboptimal. Over the course of our experience, we have learned many technical lessons and incrementally modified our methods as outlined in Table 4. We have also learned lessons about patient selection. We intensified our post-chemoradiation rehabilitation process and no longer assumed that we could get sicker, older, and weaker patients through the operation just because we were employing minimally invasive techniques, which seemed to be true in our robotic pulmonary experience. We use stair climbing as a criteria for surgery. If patients were too weak to stair climb after neoadjuvant therapy we sent them for additional rehabilitation until they could go down and immediately back up 28 steps without stopping because of loss of balance or dyspnea. In addition, because the most common cause of our 90-day mortality was liver failure, we starting to perform a liver biopsy on any patient with any history of significant alcohol use at the start of the operation and aborted the operation in those with cirrhosis. Although laboratory testing and imaging can suggest cirrhosis, we view liver biopsy as the gold standard14. Of six patients we performed liver biopsy on, 2 had cirrhosis leading to discontinuing the operation (neither had evidence of cirrhosis on preoperative CT scan and laboratory testing), and none had adverse events related to the biopsy. Our initial results in terms of anastomotic leaks compared unfavorably to our prior experience with open Ivor Lewis esophagectomy (one anastomotic leak in 221 patients)15. We found a 16.6% 30-day mortality and 33.2% 90-day mortality rate in
the 6 patients who had an anastomotic or conduit complication, which occurred a median of 8 days after the operation (range 4-15 days). Of the 2 patients that died, one patient developed a tracheoesophageal fistula that did not respond to esophageal or tracheal stenting, leading to respiratory failure, and the other succumbed to pneumonia after discharge. We applied root cause analysis to this problem and considered several causes and followed the algorithm shown in Figure 1. The likelihood of potential causes of the problem having an impact was assessed by reviewing patient records, team discussion, seeking opinions from surgeons experienced with esophagectomy at other centers, and appraisal of the surgical literature. One potential cause was a poor anastomotic technique. One root cause considered was that we were hand sewing the anastomosis with the robot and we may have had anastomotic tension that we could not assess. Our corrective action was to staple instead of hand-sew and our other corrective action was to make the conduit longer by stretching it better in the abdomen during creation. For the problem of leak leading to significant morbidity our root cause analysis led to identify one possible cause of problems as delay in making the diagnosis of leak. The root cause was not identifying the subtle signs of leak. We routinely used a drain by the anastomosis but were not checking a white blood cell count daily. Our corrective action was to check a white blood cell count every day until the fiber-optic evaluation of swallowing (FES, vocal chord check) that we routinely perform on postoperative day four prior to the swallow. Other corrective action was to take a more aggressive approach to managing patients experiencing a conduit complication, with source control, early esophagoscopy, nutritional support, and avoiding further pulmonary aspiration. We are more likely to deploy an esophageal stent and if needed re-operate to repair or revise the anastomosis, and perform decortication and drainage of the leak. A final problem we identified using root cause analysis was patient weakness on post-operative day 3-7 that led to complications. Some of the considered causes were insufficient pre-operative nutrition and physical therapy prior to surgery. Root causes considered were too few patients getting a jejunostomy tube or adequate caloric supplementation prior to surgery but the albumin and pre-albumin
data did not support this cause. Another root cause considered was inadequate time to recovery after pre-operative chemo-radiotherapy and not aggressive enough physical therapy prior to esophagectomy. Our corrective action was to lengthen the duration of our pre-operative physical therapy prior to esophagectomy. In spite of these challenges, we believe that minimally invasive esophagectomy provides advantages in terms of pain, recovery time, and cardiopulmonary perioperative complications and thus decided to refine our robotic approach as opposed to reverting back to open. Although it is impossible to statistically prove the impact of our changes we believe that important lessons were learned. We should note that the primary surgeon in this study has had very extensive experience on the robotic platform; in spite of this, we have had the challenges as described above. Surgeons planning on moving towards a robotic platform for esophagectomy should be familiar with less complex robotic procedures prior to attempting esophagectomy. Given the “growing pains” that we encountered, we recommend that only surgeons and institutions with good to excellent perioperative morbidity and mortality following open esophagectomy consider transitioning to a robotic approach. The development of any robotic program requires a step-wise, systematic, team-based approach with specific interventions to solve problems and ensure safety, which we have detailed previously16. A progression from on-line modules, cadaveric models, observation, proctored surgeries, to finally independent operating should be followed. Proper equipment and patient positioning and establishing a time limit after which converting to an open operation should be done are among the processes that contribute to safe robotic thoracic surgery. In this series, only 5 patients underwent robotic gastric conduit mobilization. This was largely due to the fact that, until recently, a separate general surgery team more comfortable with laparoscopic techniques executed the abdominal phase of the operation. That said, although robotic gastric conduit mobilization is feasible, we currently prefer to perform it laparoscopically because 1) there is limited dissection and suture tying to be done that would yield an advantage to a robotic
approach and 2) we are concerned about trauma to the conduit due to the power of the robot. Early survival following robotic-assisted minimally invasive esophagectomy was comparable to survival following non-robotic MIE in the ECOG E2202 study: overall 1-year, 2-year, and 3-year survival of 73.1%, 59.9%, and 54.2% in this study compared to 80.5%, 68.0%, and 58.4% in the ECOG study, though our study had a higher proportion of clinical stage III patients (58% vs 33%)17. Median follow up time in our study was only 27.4 months, which makes it difficult to make conclusions about long-term survival based on our series. Previous reports suggest the oncologic efficacy of MIE is comparable to open esophagectomy[18][19]. Further studies on ways to improve the 30-day, 90-day and long-term survival of patients that undergo esophagectomy are needed.
Disclosures: Robert J. Cerfolio: proctor for Intuitive Surgical, consultant for Ethicon and Community Health Systems. No funding support was used in the conduct of this study.
Figure 1. Schematic and example of root cause analysis
Problem
Possible Cause
Likely Impact
Corrective Action
Possible Cause
Unlikely/uncertain Impact
Unlikely/uncertain Impact
Corrective Action
Likely Impact
Corrective Action
Corrective Action
Conduit Complications
Possible Cause: Too much tension leading to ischemia
Possible Cause: Anastomotic Technique
Possible Cause: Injury to gastroepiploic artery
Possible Cause: Poor Preoperative Nutrition
Likely Impact
Likely Impact
Unlikely Impact
Unlikely Impact
Stretch conduit during creation
Staple anastomosis rather than hand-sew
Figure 2A. Sample port ort placement of lap laparoscopic phase of robotic Ivor Lewis esophagectomy. (LV) Liver retractor, (L) Surgeon left hand, (R) Surgeon right hand, (12 mm) (A) Assistant, nt, (C) Camera, (U) Umbilicus. B..Port .Port placement for thoracic
phase of robotic Ivor Lewis esophagectomy: (C) camera port (1) robotic arm 1 (2) robotic arm 2 (3) robotic arm 3 (A) assistant port. Figure 3. Kaplan-Meier overall survival curve of patients undergoing robotic IvorLewis esophagectomy (n=85)
Table 1. Patient characteristics of the 85 patients undergoing robotic Ivor Lewis esophagectomy All Patients with
Patients with (n=85) mortality day mortality
90-day no 90(n=9)
(n=76) Age (median, y) (range) (57-77) 62 (36-80) Sex Male (78%) 67 (88%) Female (22%) 9 (12%) Neoadjuvant chemoradiation therapy (67%) 58 (76%) Weight loss in 3 months before surgery (median lb) 20)** 0 (0-16)** Comorbidities Hypertension (56%) 43 (57%) Coronary artery disease 13 (17%) Diabetes mellitus (11%) 12 (15.8%) Chronic obstructive pulmonary disease (33%) 3 (3.9%) Prior cardiothoracic surgery 4 (5.3%) Known liver dysfunction 0 (0%) FEV1 (% predicted, median) (55-89)** 99 (89-110)** DLCO (% predicated, median) (57-88)** 84 (71-103)** Hemoglobin (g/dL, median) (11.9-12.3)** 13.3 (12.1-14.3)** ASA class (median) 3
63 (36-80)
66
74 (87%)
7
11 (13%)
2
64 (75%)
6
5 (0-20)**
8 (0-
48 (56%)
5
13 (15%)
0 (0%)
13 (15%)
1
6 (7.1%)
3
4 (4.7%)
0 (0%)
0 (0%)
0 (0%)
99 (83-109)**
70
84 (71-102)**
75
13.1 (12-14.3)**
12.3
3
3
ECOG score (median) 1 Indication Esophageal cancer (100%) 75 (99%) Adenocarcinoma (67%) 66 (88%) Squamous cell carcinoma (33%) 9 (12%) High-grade dysplasia 1 (1.3%) Location of tumor Distal or GE junction (89%) 72 (96%) Mid esophagus (11%) 4 (4%) Clinical staging* High-grade dysplasia 1 (1.3%) Stage I 14 (19%) Stage II (33%) 18 (24%) Stage III (67%) 43 (57%)
1
1
84 (99%)
9
72 (86%)
6
12 (14%)
3
1 (1.2%)
0 (0%)
79 (94%)
8
5 (6%)
1
1 (1.2%)
0 (0%)
14 (17%)
0 (0%)
21 (25%)
3
49 (58%)
6
FEV1 – forced expiratory volume in 1 second; DLCO – diffusion of carbon monoxide; ASA – American Society of Anesthesiologists classification; ECOG – Eastern Cancer Oncology Group; * Staging determined by endoscopic ultrasound (EUS) if available, and CT scan if not staged by EUS, in accordance with the 7th edition AJCC staging system; **Interquartile range; - p value < 0.05 for comparison
Table 2. Morbidity and mortality in 85 patients undergoing robotic Ivor Lewis esophagectomy
ICU admissions From OR During hospital stay
1 (1.2%) 11 (12.9%)
Conversion to thoracotomy
1 (1.2%)
Conversion to laparotomy
1 (1.2%)
Complications Anastomotic leak Conduit ischemia Chylothorax Atrial fibrillation Pneumonia or respiratory failure Reoperation
4 (4.3%) 2 (2.2%) 5 (5.9%) 6 (7.1%) 6 (7.1%) 9 (11%)
30-day mortality In hospital mortality 90-day mortality
3 (3.5%) 3 (3.5%) 9 (11%)
Table 3. Summary of series of robotic esophagectomies, compared to Luketich MIE series. Name, # Lymph Operativ Estimat Operati Leak Overall Mortality year
pt
nodes
e
ed
ve time
s
dissect
approac
blood
(min)
ed
h
loss
rate
major morbid ity
(ml) Cerfolio,
8
2015
5
22
Ivor
35
361
4.3%
36.4%
Lewis
3.5% 30-day 11% 90-day
(lap/rob ot abd , robot chest) Hernandez 5 , 2013[20]
20
2
Ivor
NR
442
3.8%
26.9%
Lewis
0% (“hospital”)
(robot abd/che st) De la
5
Fuente,
0
18.5
Ivor
NR
445
4%
28%
Lewis
2013 [9]
0% (“hospital”)
(robot abd/che st)
Sarkaria,
2
2013 [10]
1
20
Ivor
14%
24%
4.8%
Lewis
(grad
(grade
(“postoperati
(n=17)
e II or
III or
ve”)
and
greate greater
Mckeow
r)
n (n=4), robot abd/che
300
556
)
st Dunn,
4
2013[21]
0
20
Transhia
100
311
25%
NR
2.5% 30-day
200
445
9.1%
36.4%
0%
tal (robot mediasti nal dissectio n)
Weksler,
1
2012 [3]
1
19
Mckeow n (robot
(“hospital”)
abd/che st) Kim, 2010
2
[16]
1
40
Mckeow
150
410
19%
NR
0% 90-day
625
450
21%
NR
6.4%
n (lap/rob ot abd, robot chest)
Boone,
4
2009[22]
7
29
Ivor Lewis
(“postoperati
(lap abd,
ve”)
robot chest) Kernstine,
1
2007 [17]
4
18
Ivor
275
11.2
14%
29%
0% 30-day
Lewis
hours
1 patient
(lap/rob
(total
(7.1%) died
ot abd,
room
at 72 days
robot
time)
chest)
Galvani,
1
2008[23]
8
14
Transhia
54
267
33%
NR
0% 30-day
950
450
14%
NR
5%
tal (robot abd)
Hillegersb
2
erg, 2006
1
20
[13]
Mckeow n (lap
(“hospital”)
abd, robot chest)
Luketich, 2012 [6]
1033 21
Mckeown (n=481), Ivor Lewis (n=530) Lap abd/VATS chest
NR
NR
5% NR (requiring surgery)
Abbreviations: lap – laparoscopic, abd – abdomen, NR – not recorded
1.68% 30day 2.8% 30day or hospital
Table 4. Lessons learned during series of 85 robotic Ivor Lewis esophagectomies. Preoperative Delay operation on patients ECOG 2 or 3 until fully recovered from phase
pre-op chemo/radiation using longer periods of rehabilitation and using 28 step-stair climbing without stopping as criteria.
Laparoscopic
Perform liver biopsy prior to proceeding in patients with a history of
phase
alcohol abuse, to ensure that unsuspected cirrhosis is not present Maximize the length of the conduit by stretching the stomach during the stapling phase of the conduit Preserve a large tongue of well- perfused omentum during gastric conduit creation that is later used to buttress the anastomosis and protect the carina and right and left main stem bronchus.
Thoracic phase
Avoid grasping the conduit with the robotic instruments during mobilization into the chest Use the bedside assistant to help coax the conduit into the chest by pulling on the to-be-resected specimen only and/or the assistant gently grasps the conduit and pulls it up with a non-traumatic lung clamp Perform the gastrotomy on the posterior surface of the gastric conduit at least 4 cm inferior from the end of the gastric conduit’s staple line and make it longitudinal, staying as close to the greater curvature of the conduit and away from the stapled lesser curve of the stomach Suture the conduit under the divided esophagus to the pleura posteriorly and to the transected vagus nerve anteriorly to line up the anastomosis When stapling the posterior aspect of the anastomosis, ensure that the tip of the stapler is at least 2 cm inferior to the end of the staple line on the stomach and lined up with the middle of the conduit After completion of the anastomosis, place two or three interrupted sutures between the gastric conduit and the right hemidiaphragm to help prevent herniation of abdominal contents into the chest
[1]
Lee S, Sudarshan M, Li C et al. Cost-effectiveness of minimally invasive versus open esophagectomy for esophageal cancer. Ann Surg Oncol 2013; 20: 3732-9. [2] Xie MR, Liu CQ, Guo MF et al. Short-term outcomes of minimally invasive Ivor_Lewis esophagectomy for esophageal cancer. Ann Thorac Surg 2014; 97: 1721-7. [3] Biere SS, van Berge Henegouwen MI, Maas KW et al. Minimally invasive versus open oesophagectomy for patients with oesophageal cancer: a multicenter, open-label, randomized controlled trial. Lancet 2012; 379: 1887-92. [4] Weksler B, Sharma P, Moudgill N et al. Robot-assisted minimally invasive esophagectomy is equivalent to thoracoscopic minimally invasive esophagectomy. Dis Esophagus 2012; 25: 403-9. [5] Clark J, Sodergren MH, Purkayastha S et al. The role of robotic assisted laparoscopy for oesophagogastric oncological resection; an appraisal of the literature. Dis Esophagus 2011; 24: 240-50. [6] De La Fuente SG, Weber J, HOffe SE et al. Initial experience from a large referral center with robotic-assisted Ivor Lewis esophagogastrectomy for oncologic purposes. Surg Endosc 2013; 27: 3339-47. [7] Sarkaria IS, Rizk NP, Finley DJ et al. Combined thoracoscopic and laparoscopic roboticassisted minimally invasive esophagectomy using a four-arm platform: experience, technique, and cautions during early procedure development. Eur J Cardiothorac Surg 2013; 43: e107-15. [8] Cerfolio RJ, Bryant AS, Hawn MT. Technical aspects and early results of robotic esophagectomy with chest anastomosis. J Thorac Cardiovasc Surg 2013; 145: 90-6. [9] Kim J, Hyung WJ, Lee CY et al. Thoracoscopic esophagectomy for esophageal cancer: feasibility and safety of robotic assistance in the prone position. J Thorac Cardiovasc Surg 2010; 139: 53-9. [10] Kernstine KH, DeArmond DT, Shamoun DM et al. The first series of completely robotic esophagectomies with three-field lymphadenectomy: initial experience. Surg Endosc 2007; 21: 2285-92. [11] Murthy SC, Law S, Whooley BP et al. Atrial fibrillation after esophagectomy is a marker for postoperative morbidity and mortality. J Thorac Cardiovasc Surg 2003; 126: 1162-7. [12] Vaporciyan AA, Correa AM, Rice DC et al. Risk factors associated with atrial fibrillation after noncardiac thoracic surgery: analysis of 2588 patients. J Thorac Cardiovasc Surg 2004; 127: 779-86. [13] Schoppmann SF, Prager G, Langer FB et al. Open versus minimally invasive esophagectomy: a single-center case controlled study. Surg Endosc 2010; 24: 3044-53. 14 Torruellas C, French SW, Medici V. Diagnosis of alcoholic liver disease. World J Gastroenterol 2014; 20: 11684-99. 15 Cerfolio RJ, Bryant AS, Canon CL et al. Is botulinum toxin injection of the pylorus during Ivor Lewis esophagogastrectomy the optimal drainage strategy? J Thorac Cardiovasc Surg 2009; 137: 565-72. 16 Cerfolio RJ, Bryant AS, Minnich DJ. Starting a robotic program in general thoracic surgery: why, how, and lessons learned. Ann Thorac Surg 2011; 91: 1729-37. 17 Luketich J, Pennathur A, Franchetti Y et al. Minimally invasive esophagectomy: results of a prospective phase II multicenter trial-the Eastern Cooperative Oncology Gropu (E2202) study. Ann Surg 2015; 261: 702-7.
[18]
Dantoc M, Cox MR, Eslick GD. Evidence to support the use of minimally invasive esophagectomy for esophageal cancer: a meta-analysis. Arch Surg 2012; 147: 768-76. [19] Singh RK, Pham TH, Diggs BS et al. Minimally invasive esophagectomy provides equivalent oncologic outcomes to open esophagectomy for locally advanced (stage II or III) esophageal carcinoma. Arch Surg 2011; 146: 711-4. [20] Hernandez JM, Dimou F, Weber J et al. Defining the learning curve for robotic-assisted esophagogastrectomy. J Gastrointest Surg 2013; 17: 1346-51. [21] Dunn DH, Johnson EM, Morphew JA et al. Robot-assisted transhiatal esophagectomy: a 3-year single-center experience. Dis Esophagus 2013; 26: 159-66 [22] Boone J, Schipper ME, Moojen WA et al. Robot-assisted thoracoscopic oesophagectomy for cancer. Br J Surg 2009; 96; 878-86. [23] Galvani CA, Gorodner MV, Moser F et al. Robotically assisted laparoscopic transhiatal esophagectomy. Surg Endosc 2008; 188-95.
Robotic Esophagectomy for Cancer: Early Results and Lessons Learned Robert J Cerfolio MD, MBA, FACS, FCCP, Benjamin Wei MD, Mary T Hawn MD, Douglas J Minnich MD
www.elsevier.com/locate/buildenv
PII: DOI: Reference:
S1043-0679(15)00165-3 http://dx.doi.org/10.1053/j.semtcvs.2015.10.006 YSTCS784
To appear in:
Semin Thoracic Surg
Cite this article as: Robert J Cerfolio MD, MBA, FACS, FCCP, Benjamin Wei MD, Mary T Hawn MD, Douglas J Minnich MD, Robotic Esophagectomy for Cancer: Early Results and Lessons Learned, Semin Thoracic Surg, http://dx.doi.org/10.1053/j.semtcvs.2015.10.006 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Robotic Esophagectomy for Cancer: Early Results and Lessons Learned Robert J Cerfolio, MD, MBA, FACS, FCCP Benjamin Wei MD Mary T Hawn MD Douglas J Minnich MD Address correspondence to: Robert J Cerfolio, MD, MBA Division of Cardiothoracic Surgery Chief of Thoracic Surgery University of Alabama at Birmingham 703 19th St S, ZRB 739 Birmingham, AL 352094 Tel: 205-996-7561 Fax: 205-934-6218 Email : [email protected] From the Division of Thoracic Surgery at University of Alabama at Birmingham
Keywords: esophagus, esophageal cancer, esophageal surgery, robotics, minimally invasive surgery Funding: None Disclosures: Robert J. Cerfolio - proctor for Intuitive Surgical; consultant for Ethicon and Community Health Systems (CHS)
Abbreviations
ASA – American Society of Anesthesiologists classification COPD – chronic obstructive pulmonary disease DLCO – diffusion of carbon monoxide ECOG – Eastern Cancer Oncology Group FES – fiberoptic evaluation of swallowing FEV1 – forced expiratory volume in 1 second MIE – minimally invasive esophagectomy
Abstract Objective: Minimally invasive esophagectomy with intra-thoracic dissection and anastomosis is increasingly performed. Our objectives are to report our operative technique, early results and lessons learned.
Methods: This is a retrospective review of 85 consecutive patients who were scheduled for minimally invasive Ivor Lewis esophagectomy (laparoscopic or robotic abdominal and robotic chest) for esophageal cancer.
Methods: Between 4/2011 and 3/2015, 85 (74 men, median age 63) patients underwent robotic Ivor Lewis esophageal resection. Sixty-four patients (75%) had preoperative chemo-radiotherapy, 99% had esophageal cancer and 99% had an R0 resection. There were no abdominal or thoracic conversions for bleeding. There was one abdominal conversion for the inability to completely staple the gastric conduit. The mean operative time was 6 hours, median blood loss was 35 ml (no intraoperative transfusions), median number of resected lymph nodes was 22 and median length of stay was 8 days. Conduit complications (anastomotic leak or conduit ischemia) occurred in 6 patients. The 30-day and 90-day mortality were
3/85 (3.5%) and 9/85 (10.6%), respectively. Initial poor results led to protocol changes via root cause analysis: longer rehabilitation prior to surgery, liver biopsy in patients with history of suspected cirrhosis, and refinements to conduit preparation and anastomotic technique.
Conclusion: Robotic Ivor Lewis esophagectomy for cancer provides an R0 resection with excellent lymph node resection. Our preferred port placement and operative techniques are described. Disappointingly high thoracic conduit problems and 30 and 90-day mortality led to lessons learned and implementation of change which are shared. Ultra-miniabstract
A review of 85 consecutive patients undergoing robotic Ivor Lewis esophagectomy shows our preferred technique and early results. Root cause analysis led to protocol changes that were implemented in order to improve our 30 and 90-day mortality and other outcomes.
Introduction The use of minimally invasive esophagectomy (MIE) and hybrid esophagectomy has increased recently. The term “minimally invasive” refers to performing both the thoracic and abdominal phases of the operation with laparoscopic, thoracoscopic, or robotic assistance. Hybrid esophagectomy combines minimally invasive with neck approach and open approaches (ex. laparoscopy and thoracotomy, laparotomy and VATS). Recent studies have demonstrated that MIE may be associated with decreased blood loss, chest tube duration, length of stay, and respiratory complications and possibly reduced cost versus open esophagectomy [1] [2],[3].[4],[5]. Previous reports on robotic-assisted Ivor Lewis esophagectomies, have demonstrated feasibility, with perioperative outcomes similar to MIE[6],[7],[8]. Our objective in this study is to report our series and some of the lessons we have learned during the experience.
Methods This is a retrospective cohort study of prospectively collected data evaluating evalua a consecutive series of patients who underwent thoracic robotic minimally invasive Ivor Lewis wis esophagectomy. All patients with a planned Ivor Lewis esophagectomy with a robotic thoracic approach and laparoscopic or robotic abdominal approach were included in the study. Patients with mid or distal esophageal lesions were candidates for Ivor Lewis esophagectomy. Preoperative evaluation included endoscopic ultrasound, esophagogastroduodenoscopy, and integrated PET/CT scan. Patients also typically received pulmonary function testing testing, stress testing and stair climbing. There were no contrain contraindications to offering a robotic approach for Ivor Lewis esophagectomy. Initially, patients underwent planned open abdominal and robotic thoracic approach (“hybrid “approach) for Ivor Lewis esophagectomy during the period of the study study, and were excluded. These patients had a planned open abdominal approach to test feasibility early on in the experience rather than due to patient factors. Root cause analysis (as shown in Figure 1) was used to improve results after adverse events. Information was obtained through hospital databases and medical records. Descriptive statistics were used to estimate the frequency of categorical variables and median of the continuous study variables. Comparisons were done with two two-tailed t tests for continuous variables and either the 2 or the Fisher exact test was used to compare categorical data. The University of Alabama at Birmingham’s Institutional Review Board approved this protocol as well as the prospective database used to collect information for this study. Individual dual consent was waived for inclusion in this study; however it was required and obtained in order to enter patient data in the prospective database.
Operative Details
Laparoscopic Phase with Gastric Conduit Creation The placement of ports for the laparo laparoscopic scopic phase of the operation is shown in Figure 2a. The camera port is located 15 cm inferior to the xiphoid process and 3
cm to the left of midline. The liver retractor may be positioned via a subxiphoid port (grasper or Nathanson retractor) as depicted or a right subcostal port using a Mediflex (Islandia,NY) Positractor with a Lapro-Flex self-forming retractor. A tongue of omentum is preserved during conduit creation order to cover the anastomosis and protect the carina. After division of the short gastric arteries and dissection of the esophagus at the hiatus and into the mediastinum, the left gastric artery is identified, lymph node tissue swept upward, and the vein and artery are divided with a vascular staple load. A Kocher maneuver is not routinely performed. Adequate mobilization is confirmed by demonstrating that the pylorus can reach the diaphragmatic hiatus. Botulinum toxin injection into the pylorus (100 units in 4 ml of saline) with a spinal needle was used as the pyloric drainage procedure. A 4-5 cm gastric conduit is fashioned with multiple gastrointestinal staple loads with the stomach on tension to maximize length of the conduit. The conduit is sutured to the future specimen and positioned into the mediastinum along with a Penrose drain encircling the esophagus. The jejunum is tacked to the abdominal wall with 2-0 vicryl sutures and a 12 French jejunostomy tube is placed with a modified Seldinger technique.
Robotic Thoracic Technique After completion of the abdominal phase, a double lumen endotracheal tube is placed. Patient positioning is shown in Figure 2b. This modified lateral decubitus position allows the lung and blood to fall away from the operative field while avoiding the anesthetic delays of prone positioning. The robot is positioned at a 75 degree angle to the longitudinal axis of the patient’s body, approaching the patient from their back so that the instruments/arms are directed posteriorly.
Thoracic Robotic Port Placement Port placement is shown in Figure 2b. The 8 mm metal trocar for robotic arm 1 is first placed just below the hair bearing part of the axilla in the triangle between the latissimus dorsi and pectoralis major muscles. A paravertebral block using 0.25% bupivicaine and epinephrine is performed under direct vision of a 5
mm VATS camera. A 5 mm metal trocar for robotic arm 3 is inserted next, as posteriorly and inferiorly in the chest as possible. The remaining ports are placed under direct vision with the camera through this posterior port. A 12 mm plastic port for the robotic camera is placed third, located 9 cm away from robotic arm 1 in a line towards the right hip. Robotic arm 2 is placed fourth, 9 cm away from the camera port. It can be either a 5 mm or 8 mm metal trocar. The last port placed is the 12 mm bedside assistant access port, which is triangulated behind the camera port and the port for robotic arm 2 and placed as low as possible to maximize room for the bedside assistant.
Robotic Instruments and Operative Conduct of Thoracic Phase A Cadiere grasper is used for robotic arm 2 (Schertel instead if 5 mm port). A bipolar thoracic dissector (also known as a long curved tipped dissector) is placed in robotic arm 1. The 5 mm thoracic grasper is used for robotic arm 3 and serves mainly as a retractor. The intrathoracic esophagus is mobilized from thoracic inlet to diaphragmatic hiatus. All tissue from the pericardium to the left pleural surface is removed. Both the right and left inferior pulmonary veins are visualized. All lymph nodes are removed using a bipolar instrument that is not hot on either of its outside surfaces. Special care is taken to avoid thermal injury to the airway during the carinal lymph node dissection. We prefer to identify the right main stem bronchus first, then the trachea and then the left main stem bronchus prior to the removal of the entire subcarinal lymph node basin. The azygos vein is divided with a vascular stapler. We do not routinely ligate the thoracic duct. The esophagus along with vagus nerves are divided at the level of the azygos vein for distal esophageal tumors; for mid-esophageal tumors the esophagus is divided higher to ensure a negative margin. The right paratracheal lymph nodes are not routinely removed unless a mid or mid-low squamous cell cancer is the primary lesion. Dissection of the esophagus above the azygos vein is performed close to the esophagus with bi-polar cautery only to avoid injury to the recurrent laryngeal nerve.
Robotic Chest Anastomosis After transection of the esophagus and removal of the specimen, the conduit is tacked posteriorly to the pleura superior to the divided azygos vein and anteriorly to the divided right vagus nerve, which helps line up the anastomosis. A posterior longitudinal gastrotomy is made at least four cm from the tip of the conduit near the greater curvature of the gastric conduit. We now prefer to staple the posterior aspect of the anastomosis using a gastrointestinal 30 mm stapler and complete then anterior part of the anastomosis with continuous 3-0 Vicryl sutures on the inner layer and interrupted 3-0 silk sutures as the outer row. However, if it seems that stapling the posterior wall of the anastomosis would create too much tension or seems unfavorable from an anatomic perspective, a completely hand-sewn anastomosis to a more horizontal gastrotomy can be performed. Once the anastomosis is complete, we buttress the anastomosis anteriorly with a tongue of omentum, interposing it between the conduit and the airway. The conduit is then secured to the diaphragmatic hiatus with silk suture to prevent herniation of abdominal contents. A single 20 French chest tube is placed lying near the anastomosis, and the operation completed.
Postoperative Management Patients are sent to the floor and kept NPO. Tube feeds are started on postoperative day 1. Patients receive a fiber-optic swallow study on post-operative day 4 and if normal a gastrograffin and then barium swallow study are performed later that day. If no anastomotic leak is visualized, clear liquids are started following strict aspiration precautions. Chest tubes are removed once tube feedings are at goal and no chylothorax is observed. All patients are sent home with full nutritional support via the jejunostomy tube.
Results Between 4/2011 and 3/2015, 92 patients underwent consecutive robotic esophagectomy. The first seven patients had their abdominal phase of the operation performed via laparotomy and were excluded from this study. The remaining 85
consecutive patients (74 men, 87%) underwent a minimally invasive esophagectomy. Their preoperative characteristics are shown in Table 1. Mean operating time, defined as the time between first skin incision and closure of last skin incision and including repositioning and docking/undocking of the robot, was 6 hours (360 minutes, range 283-489 minutes). Median blood loss was 35 ml. No patients underwent a blood transfusion during the operation and no patients went back to the operating room for bleeding. The median number of lymph nodes obtained during the surgery was 22. The R0 resection rate was 99% (84/85). One patient (1.2%) required conversion to a thoracotomy because of tumor invasion of the main stem bronchus. Seventy-nine (92.9%) patients underwent laparoscopic gastric conduit mobilization, 5 (5.9%) underwent robotic gastric conduit mobilization, and only 1 patient (1.2%) required conversion from laparoscopy to laparotomy. This occurred after the staple line of the gastric conduit dehisced because the gastric wall was too thick. Median hospital stay was 8 days (interquartile range 7-12 days, range 5-46 days). Morbidity occurred in 31 patients (36.4%) as described in Table 2. Six (7.1%) of patients developed pneumonia, defined as presence of new pulmonary infiltrate on radiograph accompanied by fever, leukocytosis, worsening oxygenation, or cough with productive sputum, or respiratory failure, defined as reintubation and/or requirement for bronchoscopy. Four patients had an anastomotic leak. Of these, the last 3 patients underwent esophageal stent placement. Two patients had gastric conduit necrosis, and were treated with reoperation with revision of the anastomosis and intercostal muscle flap placement. Therefore, a total of 6 patients (7.1%) experienced confirmed anastomotic/conduit complications. One of these 6 patients (16.7%) died within 30-days, and 2 of 6 (33.3%) within 90 days. The overall perioperative mortality rate, defined as mortality within 30 days of surgery, was 3.5% (3 of 85). The in-hospital mortality rate was also 3.5%.. All 3 patients that died had pre-operative chemoradiotherapy. The causes of 30-day mortality were suspected pulmonary embolism, anastomotic leak, and ischemic bowel due to an embolic event. The overall 90 day mortality rate was 10.6% (9 of
85). Of the 6 patients dying between 30 and 90 days, the causes of death was liver failure (2 of 7), pneumonia (2 of 7), urosepsis leading to bacteremia (1 of 7), and unknown(1 of 7). Of the patients who died of liver failure, none of them had issues with postoperative bleeding, sepsis, or leak, and both were readmitted after discharge from the initial postoperative convalescence. We found a 16.6% 30-day mortality and 33.3% 90-day mortality rate in patients who had an anastomotic or conduit complication, which occurred a median of 8 days after the operation (range 4-15 days). The characteristics of patients who died within 90 days after esophagectomy are compared to those of patients who did not die within 90 days in Table 1. Patients who died were more likely to have COPD (p=0.0143) and had a lower FEV1 (p=0.0036). There were no other significant differences in characteristics between the two groups. In our first tercile of esophagectomy patients (n=28), the 30-day mortality rate was 7.1% and 90-day mortality rate was 7.1%. In the next tercile of esophagectomy patients, the 30-day mortality rate was 10.7% and 90-day mortality rate was 17.9%. In the last tercile of patients, the 30-day mortality rate was 0% and 90-day mortality rate was 7.1%. The Kaplan-Meier overall survival curve of patients undergoing robotic esophagectomy is shown in Figure 3. Median follow up time was 27.4 months. Overall survival was 73.1% at 1 year, 59.9% at 2 years, and 54.2% at 3 years.
Discussion Reported series of robotic esophagectomies are shown in Table 3. Our series compares favorably in terms of operative time, intraoperative blood loss, and conversion rate. Our series was similar to prior robotic series in terms of anastomotic/conduit complication rate (7.1%), lymph nodes dissected (median 22), R0 resection rate (99%), median postoperative length of stay (8 days), rate of pneumonia (7.1%), atrial fibrillation (7.1%) and overall major morbidity. When compared to other non-robotic MIE series, our results are fairly similar across the above parameters also [10][9][10][11][12] [13][29]. Our mortality rate is comparable to that from a recent study using the SEER-Medicare database showing
a 6% 30-day mortality and 13.3% 90-day mortality rate in patients undergoing esophagectomy between 2006-2009 [21]. However, our mortality rate is high when compared to Luketich’s series that reported a 30-day mortality of 1.7%[7]. In our initial series of 22 patients, we reported a 0% 30-day and 90-day operative mortality [11]. Subsequently, operative mortality temporarily increased possibly due to the fact that overall patient selection became less stringent for esophagectomy. Ultimately, our overall 30-day mortality rate of 3.5% for roboticassisted Ivor Lewis esophagectomy was comparable to the 30-day mortality rate of 4.5% in our prior open series [20]. The key to any successful process is to evaluate outcomes and modify the process involved to respond to those that are suboptimal. Over the course of our experience, we have learned many technical lessons and incrementally modified our methods as outlined in Table 4. We have also learned lessons about patient selection. We intensified our post-chemoradiation rehabilitation process and no longer assumed that we could get sicker, older, and weaker patients through the operation just because we were employing minimally invasive techniques, which seemed to be true in our robotic pulmonary experience. We use stair climbing as a criteria for surgery. If patients were too weak to stair climb after neoadjuvant therapy we sent them for additional rehabilitation until they could go down and immediately back up 28 steps without stopping because of loss of balance or dyspnea. In addition, because the most common cause of our 90-day mortality was liver failure, we starting to perform a liver biopsy on any patient with any history of significant alcohol use at the start of the operation and aborted the operation in those with cirrhosis. Although laboratory testing and imaging can suggest cirrhosis, we view liver biopsy as the gold standard14. Of six patients we performed liver biopsy on, 2 had cirrhosis leading to discontinuing the operation (neither had evidence of cirrhosis on preoperative CT scan and laboratory testing), and none had adverse events related to the biopsy. Our initial results in terms of anastomotic leaks compared unfavorably to our prior experience with open Ivor Lewis esophagectomy (one anastomotic leak in 221 patients)15. We found a 16.6% 30-day mortality and 33.2% 90-day mortality rate in
the 6 patients who had an anastomotic or conduit complication, which occurred a median of 8 days after the operation (range 4-15 days). Of the 2 patients that died, one patient developed a tracheoesophageal fistula that did not respond to esophageal or tracheal stenting, leading to respiratory failure, and the other succumbed to pneumonia after discharge. We applied root cause analysis to this problem and considered several causes and followed the algorithm shown in Figure 1. The likelihood of potential causes of the problem having an impact was assessed by reviewing patient records, team discussion, seeking opinions from surgeons experienced with esophagectomy at other centers, and appraisal of the surgical literature. One potential cause was a poor anastomotic technique. One root cause considered was that we were hand sewing the anastomosis with the robot and we may have had anastomotic tension that we could not assess. Our corrective action was to staple instead of hand-sew and our other corrective action was to make the conduit longer by stretching it better in the abdomen during creation. For the problem of leak leading to significant morbidity our root cause analysis led to identify one possible cause of problems as delay in making the diagnosis of leak. The root cause was not identifying the subtle signs of leak. We routinely used a drain by the anastomosis but were not checking a white blood cell count daily. Our corrective action was to check a white blood cell count every day until the fiber-optic evaluation of swallowing (FES, vocal chord check) that we routinely perform on postoperative day four prior to the swallow. Other corrective action was to take a more aggressive approach to managing patients experiencing a conduit complication, with source control, early esophagoscopy, nutritional support, and avoiding further pulmonary aspiration. We are more likely to deploy an esophageal stent and if needed re-operate to repair or revise the anastomosis, and perform decortication and drainage of the leak. A final problem we identified using root cause analysis was patient weakness on post-operative day 3-7 that led to complications. Some of the considered causes were insufficient pre-operative nutrition and physical therapy prior to surgery. Root causes considered were too few patients getting a jejunostomy tube or adequate caloric supplementation prior to surgery but the albumin and pre-albumin
data did not support this cause. Another root cause considered was inadequate time to recovery after pre-operative chemo-radiotherapy and not aggressive enough physical therapy prior to esophagectomy. Our corrective action was to lengthen the duration of our pre-operative physical therapy prior to esophagectomy. In spite of these challenges, we believe that minimally invasive esophagectomy provides advantages in terms of pain, recovery time, and cardiopulmonary perioperative complications and thus decided to refine our robotic approach as opposed to reverting back to open. Although it is impossible to statistically prove the impact of our changes we believe that important lessons were learned. We should note that the primary surgeon in this study has had very extensive experience on the robotic platform; in spite of this, we have had the challenges as described above. Surgeons planning on moving towards a robotic platform for esophagectomy should be familiar with less complex robotic procedures prior to attempting esophagectomy. Given the “growing pains” that we encountered, we recommend that only surgeons and institutions with good to excellent perioperative morbidity and mortality following open esophagectomy consider transitioning to a robotic approach. The development of any robotic program requires a step-wise, systematic, team-based approach with specific interventions to solve problems and ensure safety, which we have detailed previously16. A progression from on-line modules, cadaveric models, observation, proctored surgeries, to finally independent operating should be followed. Proper equipment and patient positioning and establishing a time limit after which converting to an open operation should be done are among the processes that contribute to safe robotic thoracic surgery. In this series, only 5 patients underwent robotic gastric conduit mobilization. This was largely due to the fact that, until recently, a separate general surgery team more comfortable with laparoscopic techniques executed the abdominal phase of the operation. That said, although robotic gastric conduit mobilization is feasible, we currently prefer to perform it laparoscopically because 1) there is limited dissection and suture tying to be done that would yield an advantage to a robotic
approach and 2) we are concerned about trauma to the conduit due to the power of the robot. Early survival following robotic-assisted minimally invasive esophagectomy was comparable to survival following non-robotic MIE in the ECOG E2202 study: overall 1-year, 2-year, and 3-year survival of 73.1%, 59.9%, and 54.2% in this study compared to 80.5%, 68.0%, and 58.4% in the ECOG study, though our study had a higher proportion of clinical stage III patients (58% vs 33%)17. Median follow up time in our study was only 27.4 months, which makes it difficult to make conclusions about long-term survival based on our series. Previous reports suggest the oncologic efficacy of MIE is comparable to open esophagectomy[18][19]. Further studies on ways to improve the 30-day, 90-day and long-term survival of patients that undergo esophagectomy are needed.
Disclosures: Robert J. Cerfolio: proctor for Intuitive Surgical, consultant for Ethicon and Community Health Systems. No funding support was used in the conduct of this study.
Figure 1. Schematic and example of root cause analysis
Problem
Possible Cause
Likely Impact
Corrective Action
Possible Cause
Unlikely/uncertain Impact
Unlikely/uncertain Impact
Corrective Action
Likely Impact
Corrective Action
Corrective Action
Conduit Complications
Possible Cause: Too much tension leading to ischemia
Possible Cause: Anastomotic Technique
Possible Cause: Injury to gastroepiploic artery
Possible Cause: Poor Preoperative Nutrition
Likely Impact
Likely Impact
Unlikely Impact
Unlikely Impact
Stretch conduit during creation
Staple anastomosis rather than hand-sew
Figure 2A. Sample port ort placement of lap laparoscopic phase of robotic Ivor Lewis esophagectomy. (LV) Liver retractor, (L) Surgeon left hand, (R) Surgeon right hand, (12 mm) (A) Assistant, nt, (C) Camera, (U) Umbilicus. B..Port .Port placement for thoracic
phase of robotic Ivor Lewis esophagectomy: (C) camera port (1) robotic arm 1 (2) robotic arm 2 (3) robotic arm 3 (A) assistant port. Figure 3. Kaplan-Meier overall survival curve of patients undergoing robotic IvorLewis esophagectomy (n=85)
Table 1. Patient characteristics of the 85 patients undergoing robotic Ivor Lewis esophagectomy All Patients with
Patients with (n=85) mortality day mortality
90-day no 90(n=9)
(n=76) Age (median, y) (range) (57-77) 62 (36-80) Sex Male (78%) 67 (88%) Female (22%) 9 (12%) Neoadjuvant chemoradiation therapy (67%) 58 (76%) Weight loss in 3 months before surgery (median lb) 20)** 0 (0-16)** Comorbidities Hypertension (56%) 43 (57%) Coronary artery disease 13 (17%) Diabetes mellitus (11%) 12 (15.8%) Chronic obstructive pulmonary disease (33%) 3 (3.9%) Prior cardiothoracic surgery 4 (5.3%) Known liver dysfunction 0 (0%) FEV1 (% predicted, median) (55-89)** 99 (89-110)** DLCO (% predicated, median) (57-88)** 84 (71-103)** Hemoglobin (g/dL, median) (11.9-12.3)** 13.3 (12.1-14.3)** ASA class (median) 3
63 (36-80)
66
74 (87%)
7
11 (13%)
2
64 (75%)
6
5 (0-20)**
8 (0-
48 (56%)
5
13 (15%)
0 (0%)
13 (15%)
1
6 (7.1%)
3
4 (4.7%)
0 (0%)
0 (0%)
0 (0%)
99 (83-109)**
70
84 (71-102)**
75
13.1 (12-14.3)**
12.3
3
3
ECOG score (median) 1 Indication Esophageal cancer (100%) 75 (99%) Adenocarcinoma (67%) 66 (88%) Squamous cell carcinoma (33%) 9 (12%) High-grade dysplasia 1 (1.3%) Location of tumor Distal or GE junction (89%) 72 (96%) Mid esophagus (11%) 4 (4%) Clinical staging* High-grade dysplasia 1 (1.3%) Stage I 14 (19%) Stage II (33%) 18 (24%) Stage III (67%) 43 (57%)
1
1
84 (99%)
9
72 (86%)
6
12 (14%)
3
1 (1.2%)
0 (0%)
79 (94%)
8
5 (6%)
1
1 (1.2%)
0 (0%)
14 (17%)
0 (0%)
21 (25%)
3
49 (58%)
6
FEV1 – forced expiratory volume in 1 second; DLCO – diffusion of carbon monoxide; ASA – American Society of Anesthesiologists classification; ECOG – Eastern Cancer Oncology Group; * Staging determined by endoscopic ultrasound (EUS) if available, and CT scan if not staged by EUS, in accordance with the 7th edition AJCC staging system; **Interquartile range; - p value < 0.05 for comparison
Table 2. Morbidity and mortality in 85 patients undergoing robotic Ivor Lewis esophagectomy
ICU admissions From OR During hospital stay
1 (1.2%) 11 (12.9%)
Conversion to thoracotomy
1 (1.2%)
Conversion to laparotomy
1 (1.2%)
Complications Anastomotic leak Conduit ischemia Chylothorax Atrial fibrillation Pneumonia or respiratory failure Reoperation
4 (4.3%) 2 (2.2%) 5 (5.9%) 6 (7.1%) 6 (7.1%) 9 (11%)
30-day mortality In hospital mortality 90-day mortality
3 (3.5%) 3 (3.5%) 9 (11%)
Table 3. Summary of series of robotic esophagectomies, compared to Luketich MIE series. Name, # Lymph Operativ Estimat Operati Leak Overall Mortality year
pt
nodes
e
ed
ve time
s
dissect
approac
blood
(min)
ed
h
loss
rate
major morbid ity
(ml) Cerfolio,
8
2015
5
22
Ivor
35
361
4.3%
36.4%
Lewis
3.5% 30-day 11% 90-day
(lap/rob ot abd , robot chest) Hernandez 5 , 2013[20]
20
2
Ivor
NR
442
3.8%
26.9%
Lewis
0% (“hospital”)
(robot abd/che st) De la
5
Fuente,
0
18.5
Ivor
NR
445
4%
28%
Lewis
2013 [9]
0% (“hospital”)
(robot abd/che st)
Sarkaria,
2
2013 [10]
1
20
Ivor
14%
24%
4.8%
Lewis
(grad
(grade
(“postoperati
(n=17)
e II or
III or
ve”)
and
greate greater
Mckeow
r)
n (n=4), robot abd/che
300
556
)
st Dunn,
4
2013[21]
0
20
Transhia
100
311
25%
NR
2.5% 30-day
200
445
9.1%
36.4%
0%
tal (robot mediasti nal dissectio n)
Weksler,
1
2012 [3]
1
19
Mckeow n (robot
(“hospital”)
abd/che st) Kim, 2010
2
[16]
1
40
Mckeow
150
410
19%
NR
0% 90-day
625
450
21%
NR
6.4%
n (lap/rob ot abd, robot chest)
Boone,
4
2009[22]
7
29
Ivor Lewis
(“postoperati
(lap abd,
ve”)
robot chest) Kernstine,
1
2007 [17]
4
18
Ivor
275
11.2
14%
29%
0% 30-day
Lewis
hours
1 patient
(lap/rob
(total
(7.1%) died
ot abd,
room
at 72 days
robot
time)
chest)
Galvani,
1
2008[23]
8
14
Transhia
54
267
33%
NR
0% 30-day
950
450
14%
NR
5%
tal (robot abd)
Hillegersb
2
erg, 2006
1
20
[13]
Mckeow n (lap
(“hospital”)
abd, robot chest)
Luketich, 2012 [6]
1033 21
Mckeown (n=481), Ivor Lewis (n=530) Lap abd/VATS chest
NR
NR
5% NR (requiring surgery)
Abbreviations: lap – laparoscopic, abd – abdomen, NR – not recorded
1.68% 30day 2.8% 30day or hospital
Table 4. Lessons learned during series of 85 robotic Ivor Lewis esophagectomies. Preoperative Delay operation on patients ECOG 2 or 3 until fully recovered from phase
pre-op chemo/radiation using longer periods of rehabilitation and using 28 step-stair climbing without stopping as criteria.
Laparoscopic
Perform liver biopsy prior to proceeding in patients with a history of
phase
alcohol abuse, to ensure that unsuspected cirrhosis is not present Maximize the length of the conduit by stretching the stomach during the stapling phase of the conduit Preserve a large tongue of well- perfused omentum during gastric conduit creation that is later used to buttress the anastomosis and protect the carina and right and left main stem bronchus.
Thoracic phase
Avoid grasping the conduit with the robotic instruments during mobilization into the chest Use the bedside assistant to help coax the conduit into the chest by pulling on the to-be-resected specimen only and/or the assistant gently grasps the conduit and pulls it up with a non-traumatic lung clamp Perform the gastrotomy on the posterior surface of the gastric conduit at least 4 cm inferior from the end of the gastric conduit’s staple line and make it longitudinal, staying as close to the greater curvature of the conduit and away from the stapled lesser curve of the stomach Suture the conduit under the divided esophagus to the pleura posteriorly and to the transected vagus nerve anteriorly to line up the anastomosis When stapling the posterior aspect of the anastomosis, ensure that the tip of the stapler is at least 2 cm inferior to the end of the staple line on the stomach and lined up with the middle of the conduit After completion of the anastomosis, place two or three interrupted sutures between the gastric conduit and the right hemidiaphragm to help prevent herniation of abdominal contents into the chest
[1]
Lee S, Sudarshan M, Li C et al. Cost-effectiveness of minimally invasive versus open esophagectomy for esophageal cancer. Ann Surg Oncol 2013; 20: 3732-9. [2] Xie MR, Liu CQ, Guo MF et al. Short-term outcomes of minimally invasive Ivor_Lewis esophagectomy for esophageal cancer. Ann Thorac Surg 2014; 97: 1721-7. [3] Biere SS, van Berge Henegouwen MI, Maas KW et al. Minimally invasive versus open oesophagectomy for patients with oesophageal cancer: a multicenter, open-label, randomized controlled trial. Lancet 2012; 379: 1887-92. [4] Weksler B, Sharma P, Moudgill N et al. Robot-assisted minimally invasive esophagectomy is equivalent to thoracoscopic minimally invasive esophagectomy. Dis Esophagus 2012; 25: 403-9. [5] Clark J, Sodergren MH, Purkayastha S et al. The role of robotic assisted laparoscopy for oesophagogastric oncological resection; an appraisal of the literature. Dis Esophagus 2011; 24: 240-50. [6] De La Fuente SG, Weber J, HOffe SE et al. Initial experience from a large referral center with robotic-assisted Ivor Lewis esophagogastrectomy for oncologic purposes. Surg Endosc 2013; 27: 3339-47. [7] Sarkaria IS, Rizk NP, Finley DJ et al. Combined thoracoscopic and laparoscopic roboticassisted minimally invasive esophagectomy using a four-arm platform: experience, technique, and cautions during early procedure development. Eur J Cardiothorac Surg 2013; 43: e107-15. [8] Cerfolio RJ, Bryant AS, Hawn MT. Technical aspects and early results of robotic esophagectomy with chest anastomosis. J Thorac Cardiovasc Surg 2013; 145: 90-6. [9] Kim J, Hyung WJ, Lee CY et al. Thoracoscopic esophagectomy for esophageal cancer: feasibility and safety of robotic assistance in the prone position. J Thorac Cardiovasc Surg 2010; 139: 53-9. [10] Kernstine KH, DeArmond DT, Shamoun DM et al. The first series of completely robotic esophagectomies with three-field lymphadenectomy: initial experience. Surg Endosc 2007; 21: 2285-92. [11] Murthy SC, Law S, Whooley BP et al. Atrial fibrillation after esophagectomy is a marker for postoperative morbidity and mortality. J Thorac Cardiovasc Surg 2003; 126: 1162-7. [12] Vaporciyan AA, Correa AM, Rice DC et al. Risk factors associated with atrial fibrillation after noncardiac thoracic surgery: analysis of 2588 patients. J Thorac Cardiovasc Surg 2004; 127: 779-86. [13] Schoppmann SF, Prager G, Langer FB et al. Open versus minimally invasive esophagectomy: a single-center case controlled study. Surg Endosc 2010; 24: 3044-53. 14 Torruellas C, French SW, Medici V. Diagnosis of alcoholic liver disease. World J Gastroenterol 2014; 20: 11684-99. 15 Cerfolio RJ, Bryant AS, Canon CL et al. Is botulinum toxin injection of the pylorus during Ivor Lewis esophagogastrectomy the optimal drainage strategy? J Thorac Cardiovasc Surg 2009; 137: 565-72. 16 Cerfolio RJ, Bryant AS, Minnich DJ. Starting a robotic program in general thoracic surgery: why, how, and lessons learned. Ann Thorac Surg 2011; 91: 1729-37. 17 Luketich J, Pennathur A, Franchetti Y et al. Minimally invasive esophagectomy: results of a prospective phase II multicenter trial-the Eastern Cooperative Oncology Gropu (E2202) study. Ann Surg 2015; 261: 702-7.
[18]
Dantoc M, Cox MR, Eslick GD. Evidence to support the use of minimally invasive esophagectomy for esophageal cancer: a meta-analysis. Arch Surg 2012; 147: 768-76. [19] Singh RK, Pham TH, Diggs BS et al. Minimally invasive esophagectomy provides equivalent oncologic outcomes to open esophagectomy for locally advanced (stage II or III) esophageal carcinoma. Arch Surg 2011; 146: 711-4. [20] Hernandez JM, Dimou F, Weber J et al. Defining the learning curve for robotic-assisted esophagogastrectomy. J Gastrointest Surg 2013; 17: 1346-51. [21] Dunn DH, Johnson EM, Morphew JA et al. Robot-assisted transhiatal esophagectomy: a 3-year single-center experience. Dis Esophagus 2013; 26: 159-66 [22] Boone J, Schipper ME, Moojen WA et al. Robot-assisted thoracoscopic oesophagectomy for cancer. Br J Surg 2009; 96; 878-86. [23] Galvani CA, Gorodner MV, Moser F et al. Robotically assisted laparoscopic transhiatal esophagectomy. Surg Endosc 2008; 188-95.