The da Vinci Surgical System (Intuitive Surgical Inc, Sunnyvale, Calif) in the operating room at Misericordia Hospital, Grosseto, Italy.
Arrangement of the operating room and the robotic cart for cholecystectomy (A), left pulmonary lobectomy (B), gastrectomy or Nissen fundoplication (C), and left pancreatectomy (D).
The working axis and the direction of the robotic cart.
Robotic cholecystectomy. A, Overall operative time. The line indicates the boundary between the first and second periods of experience. B, A comparison of the operative times between the first period (mean [SD], 96.5 [42.2] minutes; median, 90 minutes) and the second period (mean [SD], 66.7 [20.8] minutes; median, 62.5 minutes) (t test; P = .002).
Robotic fundoplication. A, Overall operative time. The line indicates the boundary between the first and second periods of experience. B, A comparison of the operative times between the first period (mean [SD], 132.8 [80.2] minutes; median, 90 minutes) and the second period (mean [SD], 92.0 [34.1] minutes; median, 90 minutes) (t test; P = .04).
Pier Cristoforo Giulianotti, Andrea Coratti, Marta Angelini, Fabio Sbrana, Simone Cecconi, Tommaso Balestracci, Giuseppe Caravaglios. Robotics in General SurgeryPersonal Experience in a Large Community Hospital. Arch Surg. 2003;138(7):777–784. doi:10.1001/archsurg.138.7.777
Robotic technology is the most advanced development of minimally invasive surgery, but there are still some unresolved issues concerning its use in a clinical setting.
The study describes the clinical experience of the Department of General Surgery, Misericordia Hospital, Grosseto, Italy, in robot-assisted surgery using the da Vinci Surgical System.
Between October 2000 and November 2002, 193 patients underwent a minimally invasive robotic procedure (74 men and 119 women; mean age, 55.9 years [range, 16-91 years]). A total of 207 robotic surgical operations, including abdominal, thoracic and vascular procedures, were performed; 179 were single procedures, and 14 were double (2 operations on the same patient). There were 4 conversions to open surgery and 3 to conventional laparoscopy (conversion rate, 3.6%; 7 of 193 patients). The perioperative morbidity rate was 9.3% (18 of 193 patients), and 6 patients (3.1%) required a reoperation. The postoperative mortality rate was 1.5% (3 of 193 patients).
Our preliminary experience at a large community hospital suggests that robotic surgery is feasible in a clinical setting. Its daily use is safe and easily managed, and it expands the applications of minimally invasive surgery. However, the best indications still have to be defined, and the cost-benefit ratio must be evaluated. This report could serve as a basis for a future prospective, randomized trial.
HUMAN ROBOTIC surgery was introduced by Cadière and colleagues in March 1997 when the first telesurgical laparoscopic cholecystectomy was performed.1 Robot-assisted, minimally invasive surgery is now a reality and will probably become the surgical procedure of the future.2 There are several limitations and drawbacks to conventional laparoscopy, including limited movement, the inability to perform high-precision sutures, unnatural positions for the surgeon, and flat vision. Robotic surgery may overcome these limitations and allows extension of minimally invasive surgery to an increasing number of patients. However, many issues are still not resolved, such as the clinical feasibility and safety of robotic surgery in a general surgical context, the length and difficulty of the learning curve, and clinical applications and drawbacks.
The surgical robot used in the Department of General Surgery, Misericordia Hospital, Grosseto, Italy, is the da Vinci Surgical System (Intuitive Surgical Inc, Sunnyvale, Calif). It is a telesurgery system that consists of a robotic cart with 3 operative arms controlled by the surgeon in a remote console (Figure 1). The surgical team (assistant surgeon, nurse, and anesthetist) participate in the operation and watch the procedure on a video-endoscopic unit.
To facilitate coordination between the surgeon and the surgical team, we placed the operative console in a corner of the operating room. The robotic cart and the video-endoscopic unit are the mobile components of the system, and they are arranged in the operating room according to the specific surgical procedure.
Before starting surgery, system setup includes 2 phases: robot setup and optic system setup. During robot setup, the console is connected to the robotic cart via electric cables and optic fibers, and the system is switched on; then, the robot undergoes a self-test, an automatic checkup to verify links and robotic arm functions that takes approximately 3 minutes; next, the robotic arms are wrapped in disposable nylon covers (the tips of the arms contain the microchips used for the connection to the endoscopic surgical instruments); and finally, mechanical supports for trocars on the robotic arms are fixed. During optic system setup, the frontal or inclined position of the scope (0°-30°) is set; 2-dimensional or 3-dimensional vision is selected; the image on the monitor for the remote console is centered following insertion of the scope in its special sight (0°-30°); and white balancing of the robotic camera is performed.
After induction of pneumoperitoneum, the laparoscopic ports are positioned, and the robotic cart is installed (Figure 2). Correct positioning of the robotic cart is fundamental because its axis must coincide with the working axis, coming from the opposite site (Figure 3). The patient is placed in surgical position, and the robotic arms are connected to optic and operating ports; to change the patient's position, the robotic arms must be detached from the ports. Accurate placement of ports and the robotic cart is necessary to avoid collisions between the mechanical arms and to optimize the procedure. Extra ports can be placed for accessory surgical instruments controlled by the assistant surgeon.
The scope and the endoscopic surgical instruments are specific to this robot. The scope is connected to a double camera, which allows 3-dimensional vision. The instruments include a hook, scissors, forceps, and a needle holder. Recently, an ultrasound dissector, specific to the da Vinci, has also been introduced. Each instrument can be used for a maximum of 10 procedures.
Our experience with the da Vinci system began in October 2000. After initial training with cholecystectomy and Nissen fundoplication, we extended the indications to other procedures, including major procedures such as abdominal, thoracic, and vascular surgery. The patients were selected according to criteria currently used for conventional laparoscopic surgery. The operative time reported for robotic procedures includes all phases of setup, induction of pneumoperitoneum, placement of ports, and the overall surgical time.
For a retrospective comparison, we analyzed results from patients who underwent open and laparoscopic procedures at our hospital during the same period. Operative times, postoperative morbidity and mortality, and length of hospital stay were compared for cholecystectomies, Nissen fundoplications, and pancreatic and gastric resections.
Between October 2000 and November 2002, 193 patients underwent a minimally invasive, robot-assisted procedure (74 men and 119 women; mean age, 55.9 years [range, 16-91 years]). A total of 207 robotic surgical operations, including abdominal, thoracic, and vascular procedures, were performed; 179 were single procedures, and 14 were double (2 operations on the same patient) (Table 1 and Table 2).
The first and most frequently performed procedure was cholecystectomy for gallstones (n = 66 cases, 52 single and 14 associated with another operation). This procedure was chosen for the initial learning curve. In the first 26 single procedures, the mean operative time was 96.5 minutes (median, 90 minutes) vs 66.7 minutes (median, 62.5 minutes) for the second 26 single procedures (t test; P = .002) (Figure 4). Moreover, the latter group included 7 cases of acute cholecystitis. The robotic technical approach was similar to that of conventional laparoscopy; the cystic artery and duct were closed by intracorporeal ligation, rather than by clips.
Table 3 presents a retrospective comparison between robotic and laparoscopic cholecystectomies. The mean operative time was 85 minutes (range, 20-210 minutes) for robotic cholecystectomy and 65 minutes (range, 35-160 minutes) for laparoscopic cholecystectomy. After 20 robotic cholecystectomies, the mean time decreased from 96.5 to 66.4 minutes, similar to the mean operative time for laparoscopic cholecystectomy. The conversion rate was higher with the laparoscopic technique (3.5% vs 1.9%). The mean postoperative stay and morbidity rate were similar for the robotic (2 days; 1.9%) and laparoscopic (3 days; 2.2%) groups. There was no mortality in either group.
A double surgical procedure was performed on 14 patients. All underwent cholecystectomy, associated with Nissen fundoplication in 10 cases and choledocholithotomy for bile duct stones, hepatic wedge resection for liver neoplasm, hepatic cyst marsupialization, and subtotal gastrectomy for gastric carcinoma in the other 4 cases, respectively.
Fundoplication for gastroesophageal reflux disease was the second most frequently performed procedure (n = 51 operations, 41 single and 10 associated with cholecystectomy). Nissen fundoplication was done in 49 cases. Only 2 patients received Dor (180°) and Toupet (270°) fundoplication, respectively.
The mean operative time was 110 minutes (range, 40-300 minutes) for robotic fundoplication and 120 minutes (range, 60-280 minutes) for laparoscopic fundoplication. In the first period (n = 21 cases), the mean operative time was 132.8 minutes (median, 90 minutes) vs 92 minutes (median, 90 minutes) for the second period (n = 20 cases) (t test; P = .04) (Figure 5). The conversion rate was similar in both groups. The morbidity rate and mean postoperative hospital stay for robotic and laparoscopic procedures were 4.8% and 4 days and 11.4% and 6 days, respectively. We did not observe mortality in either group (Table 3).
This group included 5 esophageal myotomies with Dor fundoplication for achalasia, 5 esophagectomies for carcinoma, 1 resection of esophageal diverticulum, and 1 resection of esophageal leiomyoma. In 5 cases of esophagectomy, the robot was used for the thoracic esophageal dissection. Abdominal gastric tubulization and cervical esophagogastric anastomosis were performed using conventional laparoscopy and an open cervical incision.
Twenty-one patients underwent robotic gastric surgery; 18 patients with gastric carcinoma underwent D2 subtotal (n = 8 cases) or D2 total (n = 10 cases) gastrectomy. Two patients with unresponsive benign gastric ulcers received partial gastrectomies. The last patient underwent a wedge resection of the gastric fundus for a carcinoid tumor. After subtotal or partial gastrectomy, an intracorporeal robotic suture was made for the gastrojejunal anastomosis. In total gastrectomy, a mechanical Roux-en-Y esophagojejunal anastomosis was performed. The stapler was introduced through a minilaparotomy, which was also used for specimen retrieval.
The mean operative time was 350 minutes (range, 250-420 minutes) for robotic total gastrectomy and 365 minutes (range, 270-480 minutes) for subtotal gastrectomy; in the open gastrectomy group, the mean operative time was 185 minutes (range, 140-310 minutes) for total gastrectomy and 135 minutes (range, 100-220 minutes) for subtotal gastrectomy. The morbidity and mortality rates were 30.0% and 0.0%, respectively, for robotic total gastrectomy and 9.1% and 9.1% for robotic subtotal gastrectomy. In the group that underwent open gastrectomy, the morbidity and mortality rates were 12.5% and 2.5%, respectively, for total gastrectomy and 7.8% and 0.0% for subtotal gastrectomy. The mean postoperative hospital stay was similar in the robotic and open gastrectomy groups (Table 3).
We performed 13 robotic surgical procedures for pancreatic diseases; 8 pancreatoduodenectomies (5 Whipple and 3 Traverso-Longmire) were performed for 2 pancreatic mucinous cystoadenomas, 3 exocrine pancreatic carcinomas, 2 choledochal carcinomas, and 1 ampullary carcinoma. In 6 cases, we used a hybrid technique: conventional laparoscopy for pancreatic resection and robotic surgery for digestive reconstruction (intracorporeal biliojejunal and gastrojejunal anastomoses). In all cases, the Wirsung duct was closed by injecting surgical glue, and the pancreatic stump was sutured. The last 2 pancreatoduodenectomies were performed using a full robotic technique for resection and reconstruction. Five patients underwent a totally robotic left pancreatectomy; 2 of them received a spleen-preserving resection (1 for insulinoma and 1 for a benign cystic lesion), and 3 underwent a splenopancreatectomy (for carcinoma). In 1 patient with advanced pancreatic carcinoma, we performed a palliative procedure, consisting of a robotic gastrojejunostomy.
The mean operative time in the robotic group was 490 minutes (range, 360-660 minutes) for pancreatoduodenectomy and 270 minutes (range, 210-360 minutes) for left pancreatectomy; in the open surgery group, the mean operative time was 250 minutes (range, 180-480 minutes) for pancreatoduodenectomy and 170 minutes (range, 135-250 minutes) for left pancreatectomy (Table 3). The morbidity and mortality rates for the robotic group were 37.5% and 12.5%, respectively, for pancreatoduodenectomy and 20.0% and 0.0% for left pancreatectomy. In the open surgery group, the morbidity and mortality rates were 32.1% and 5.6%, respectively, for pancreatoduodenectomy and 28.5% and 0.0% for left pancreatectomy. The mean postoperative hospital stay was similar in the robotic and open groups.
This group included 2 hepatic resections (V-VI segmentectomy and V segmentectomy for hepatocellular carcinoma and hepatic adenoma, respectively), 1 marsupialization of a giant hepatic cyst, and 1 resection of a common bile duct for carcinoma (reconstruction by hepatojejunal anastomosis).
Our series also included 16 colorectal resections for malignant or benign disease, as reported in Table 1 (5 right hemicolectomies, 2 ileocecal resections, 1 sigmoid resection, 6 rectal anterior resections, and 2 abdominoperineal resections). The dissection of the colon was performed using a robotic technique, whereas the resection was carried out extracorporeally via a minilaparotomy. The anastomosis was performed by means of an extracorporeal suture after right resections and with the Knight-Griffen technique after sigmoidectomy or rectal anterior resection.
Thoracic procedures included 7 pulmonary resections: 1 pneumonectomy (for bronchial carcinoma), 5 lobectomies (1 for bronchial carcinoma, 1 for metastasis, 2 for bronchiectasias, and 1 for pulmonary sequestration), and 1 wedge resection (for an emphysematous bulla).
Our series also included 3 right adrenalectomies (for adenoma); 2 splenectomies (for lymphoma and recurrence of ovarian carcinoma) and 1 partial resection of the spleen (for a posttraumatic hematoma); 1 jejunal resection (jejunal carcinoma); 1 case of fallopian tube anastomosis (for a prior tubal ligation for contraception); 2 aneurysmectomies of the splenic artery and 1 aneurysmectomy of the renal artery; and 1 repeated iliac-femoral bypass (for obstruction of the previous bypass caused by dysplastic arterial stenosis).
The mean operative times and the number of ports used are reported in Table 1 and Table 2. After the initial learning curve, the actual surgical time for some procedures, such as cholecystectomy and Nissen fundoplication, decreased progressively. In some cases, the operative time was the same or less than for conventional laparoscopy.3,4 Intraoperative bleeding was always minimal and was similar to that seen with laparoscopic procedures.
In 193 patients, 4 procedures (2.1%) required conversion to open surgery; 1 Nissen fundoplication was converted because of a traumatic hepatic lesion caused by the retractor, 1 pulmonary lobectomy because of pleural adhesions, 1 pancreatoduodenectomy because of neoplastic infiltration of the portal vein, and 1 total gastrectomy because of neoplastic infiltration of the pancreas. Three other procedures (1.5%) were converted to conventional laparoscopy; a cholecystectomy and a splenic aneurysmectomy were converted because of robotic technical problems, and a Nissen fundoplication was converted because of peritoneal adhesions. The total conversion rate was 3.6% (7 of 193 patients).
Intraoperative morbidity consisted of 2 iatrogenic lesions (1.0%; 2 of 193 patients). A hepatic tear, caused by a blind retraction, occurred during a Nissen fundoplication; the patient required open surgery for liver hemostasis. The second case was a splenic lesion that occurred during total gastrectomy because of tear by traction on adhesions; splenectomy was performed with the robot as well.
The postoperative morbidity rate was 8.3% (16 of 193 patients). We observed 4 low-output pancreatic fistulas, 3 of which occurred after pancreatoduodenectomy (the pancreatic stump was always injected with surgical glue) and 1 after left pancreatectomy. Three fistulas healed spontaneously in a few days, and 1 case needed prolonged drainage for 3 weeks.
One patient who underwent a Whipple procedure had a rare spontaneous esophageal rupture, or Boerhaave syndrome, caused by acute retching on the seventh postoperative day. This patient was reoperated on laparoscopically and underwent esophageal suture and jejunostomy but died owing to septic complications.
In 4 cases, we observed postoperative abdominal bleeding after a Nissen fundoplication, a partial resection of the spleen, a jejunal resection, and a subtotal gastrectomy, respectively. The first and second cases were treated conservatively, but the third and fourth cases required reoperation.
Two cases of severe leakage were observed after a total esophagectomy (cervical esophageal anastomosis) and a total gastrectomy (esophagojejunal anastomosis), respectively. The first patient died owing to septic complications; the second underwent an open reoperation, and anastomosis repair was performed. Another patient had benign leakage after total gastrectomy (esophagojejunal anastomosis) and was treated conservatively. One patient had a splenic infarction after aneurysmectomy with ligature of the splenic artery and required percutaneous ultrasound-guided drainage of the spleen.
Only 1 patient with acute cholecystitis experienced a port site infection after cholecystectomy. Two other patients underwent open reoperations: 1 for a mediastinal abscess after resection of an esophageal diverticulum and 1 for leakage of a colorectal anastomosis after anterior resection.
The total reoperation rate was 3.1% (6 of 193 patients). Considering both intraoperative and postoperative complications, the total perioperative morbidity rate was 9.3% (18 of 193 patients). The total postoperative mortality rate was 1.5% (3 of 193 patients). Two patients died owing to septic complications caused by anastomotic leakage and mediastinitis after total esophagectomy and Boerhaave syndrome after pancreatoduodenectomy, respectively. The third patient, who underwent subtotal gastrectomy, died owing to respiratory failure after a reoperation for hemoperitoneum.
To date, few accounts of clinical experience in robotic surgery have been published. The first important robotic application was in cardiac surgery, for beating-heart coronary artery bypass grafts.5- 8 Since that time other experiences with robotic systems in general surgery,9- 13 thoracic surgery,14,15 urology,16- 19 gynecologic surgery,20 and telesurgery21- 24 have been reported.
Our series represents, to our knowledge, the largest single-institution report of clinical experience in robotic general surgery. Robotic technology may have an interesting role even in the context of a large community hospital; it must not be considered experimental equipment to be used only at selected referral centers.
Daily use of the da Vinci system is relatively simple; the robot is a safe and sturdy machine. It can be set up while preparing the patient and inducing anesthesia, and setup time decreases as the experience of the operating team increases.3,4 Malfunctions are a rare occurrence, and only 2 (1.0%) of our 193 cases required conversion.
The learning curve at the console is relatively short, even for an inexperienced surgeon. It does not take long to learn how to do perfect knots and suturing and to have full control of robotic movements, but to perform advanced procedures, full training in open and laparoscopic surgery is mandatory. Cholecystectomy and Nissen fundoplication are the basic training models, and 20 operations were necessary to complete the learning phase in both procedures (Figure 4 and Figure 5). After completion of the learning phase, the operative times for robotic cholecystectomies and Nissen fundoplications were similar to those for traditional laparoscopy (Table 3). The conversion rate, morbidity rate, and mean postoperative stay for robotic surgery compares favorably with conventional laparoscopy, perhaps reflecting an even more accurate and precise dissection.
It is much more difficult to analyze the impact of robotics on major, complex operations such as gastrectomies, pancreatic resections, or pneumonectomies. The robot allows the operation to be technically feasible in a minimally invasive fashion; but, owing to the still limited experience, it is impossible to evaluate the learning curve and the benefits to patients. Technical tricks are quite important in robotic surgery, but it is equally important to define a standard for each procedure, including case selection, patient and port positioning, cart installation, and mechanical arm setting on the operating field. Accessing the patient from the wrong direction (ie, the cart coming from the right flank instead of the right shoulder for liver surgery) can cause external collisions of the arms and it limits the internal movements of the tools.
In addition, preoperative patient selection must be accurate. Severe cardiorespiratory failure or other associated diseases with a high anesthesiological risk (American Society of Anesthesiologists categories 3-4) may be a contraindication for major procedures, such as esophagectomy, gastrectomy, and pancreatoduodenectomy, that still require long surgical times and pneumoperitoneum.19 In our series, 1 patient died owing to respiratory failure after subtotal gastrectomy; perhaps the prolonged anesthesia and pneumoperitoneum were predisposing factors.
Patient positioning is critical because postural changes during different steps of the operation may be required. The robotic cart position and arm settings have to be selected to avoid collisions and to allow assistants to use complementary instruments. The ports must be positioned as in traditional laparoscopy, but usually further apart from each other.
The most convincing indications for robotic surgery are procedures that involve a small, deep, fixed operating field (lymphadenectomy for D2 gastrectomy and Nissen fundoplication) or when minimally invasive surgery requires extreme accuracy (pulmonary hilum dissection and splenic artery dissection), fine dissection (nerve sparing in total mesorectal excision and biliary tract surgery), and endoscopic suturing and microanastomosis (fallopian tube anastomosis, iliac-femoral bypass, splenic artery anastomosis, renal artery bypass, and biliojejunal anastomosis). In some cases, as in renal artery microreconstruction after aneurysmectomy, the robotic technique made the difference, allowing the performance of an operation otherwise not feasible in minimally invasive surgery.
Difficulties using the robot can arise during procedures with large operative fields, such as in colorectal surgery, because they require changes of patient and instrument position for retraction or exposure. An anterior resection of the rectum needs at least 2 major changes in patient and cart positioning; the former for the splenic flexure mobilization, with the cart coming from the left shoulder and the operating table in an anti-Trendelenburg position, and the latter for the pelvic dissection, with the cart between the legs and the patient in a forced Trendelenburg position. Hemostasis must be very accurate during the whole procedure because, to date, we lack specific devices for washing and suctioning. Because these functions are not controlled from the console, additional maneuvers by the assistant surgeon are required to keep the operative field clear. In the future, these problems will be overcome by technological improvements such as more specific or multifunctional tools and additional mechanical arms. Hybrid procedures, or a robotic technique used during conventional laparoscopy, are a very interesting application of robotic technology. If necessary, it is possible to install the robot at any time during a standard laparoscopic procedure. The 7-mm robotic trocar can be inserted through the 12-mm trocar (ie, the "trocar inside trocar" technique), and the operation can be accomplished with robotic technology. We used the robot for reconstruction after some laparoscopic pancreatoduodenectomies to perform biliojejunal and gastrojejunal anastomosis. In laparoscopic rectal resection, the robot was useful to perform pelvic dissection and total mesorectal excision with a nerve-sparing technique while the splenic flexure was mobilized in conventional laparoscopy.
One of the main targets of robotic technology should be to perform so-called solo surgery. For simple procedures such as cholecystectomy, this is feasible, and the fourth arm prototype of the system will improve this ability. For complex operations such as pancreatoduodenectomy, the role of the assistant surgeon is more sophisticated than in conventional laparoscopy because robotic procedures involve less room for complementary working, risk of collision, need of more accurate suctioning, managing the setup of the arms, and instrument changes.
The lack of tactile feedback does not seem to be a real problem because the surgeon very soon learns to compensate with the 3-dimensional vision. We did not experience any intraoperative complications due to the lack of tactile feedback.
The main drawbacks of the system seem to be the shortage of tools and the bulky setup of the cart and the arms. This hampers the complementary actions on the operative field and sometimes makes anesthesiological maneuvers more difficult.
The end point of this study was not an evaluation of the economical costs of robotic surgery. The equipment is expensive, but costs will decrease as the market expands, and some forms of financial support may be found to purchase the system. Daily operating costs for single procedures such as cholecystectomy, considering instruments, assistance, operative time, anesthesia, and hospital stay, are just a little higher than for standard laparoscopy performed with disposable tools.
At the moment, we cannot evaluate the real cost-benefit ratio for each indication. However, this work could be the basis for a prospective randomized study.
This is one of the major reports of clinical experiences with the da Vinci robotic system in general surgery. Our experience shows that this kind of surgery is feasible, safe, and easily managed in a big community hospital, and it expands the indications of minimally invasive surgery. The learning curve is relatively short, and all surgical staff can be well trained and involved. Hybrid procedures are an interesting application of robotic techniques. The robotic approach requires new operative strategies and modifications of the pattern of port placement. We need more experience to evaluate the best indications and the cost-benefit ratio for patients and institutions, but this work may be the basis for a future, prospective, randomized study.
Corresponding author and reprints: Pier Cristoforo Giulianotti, MD, UO di Chirurgia Generale, Ospedale della Misericordia, Via Senese, 161-58100, Grosseto, Italy (e-mail: email@example.com).
Accepted for publication January 12, 2003.