Initial experiences in robot-assisted middle pancreatectomy

Abstract

Objectives

Initial results in robot-assisted middle pancreatectomy (MP) have been encouraging. However, data comparing outcomes of robot-assisted MP with those of open MP are limited. The aim of this study was to compare outcomes in patients undergoing open and robot-assisted MP, respectively.

Methods

Outcomes in an initial experience with seven consecutive patients undergoing robot-assisted MP were compared with those in 36 patients undergoing open MP.

Results

The robot-assisted MP group included five women and two men with a median age of 55 years (range: 30–62 years). Median tumour size, operative time and blood loss were 3.0 cm (range: 0.5–5.0 cm), 210 min (range: 150–330 min) and 200 ml (range: 50–400 ml), respectively. Pancreaticogastrostomy was performed in all patients. No transfusion was given intraoperatively. Pathological examination revealed five serous cystic neoplasms, one mixed-type intraductal papillary mucinous neoplasm and one lipoma. Five patients experienced postoperative pancreatic fistula and one experienced post-pancreatectomy haemorrhage. No operative mortality was noted. Compared with the open MP group, the robot-assisted MP group demonstrated a shorter median length of postoperative gastrointestinal tract recovery [2 days (range: 2–3 days) versus 4 days (range: 2–11 days); P = 0.001].

Conclusions

Robot-assisted MP can be performed safely with satisfactory efficacy; patients experienced faster gastrointestinal tract recovery compared with patients undergoing open surgery.

Introduction

In recent years, there has been growing interest in parenchyma-sparing pancreatic surgeries for benign, borderline or low-grade malignant lesions of the pancreas, especially in young patients with long life expectancies; these surgeries aim to preserve exocrine and endocrine pancreatic function, and facilitate a better quality of life after surgery.^1–5 Middle pancreatectomy (MP), for example, serves as an alternative to extended right or left pancreatic resection for the treatment of benign, borderline or low-grade malignant lesions of the neck or the proximal body of the pancreas.^2,4,5 Since it was first reported by Guillemin and Bessot in 1957,⁶ MP has been studied intensively and is now generally accepted as a safe and effective procedure, especially in terms of pancreatic function preservation.^2,4,5

However, the application of minimally invasive approaches to MP has been disappointing. The first laparoscopic MP was described by Baca and Bokan in 2003,⁷ but few centres have adopted this minimally invasive approach. This is mainly because of the technical difficulty associated with this procedure during the reconstruction phase, although various reports have confirmed the safety and feasibility of the technique.^7–12

The robot-assisted surgical system is an emerging technology which has been designed to overcome the intrinsic limitations of traditional laparoscopic surgery by offering three-dimensional surgical views, the facility for precise and flexible wrist-like movements, lack of tremor, reliable fourth-arm movement and better ergonomics for surgeons.¹³ In fact, the robotic approach has been shown to be safe and feasible for complex and difficult pancreatic resections, including those that require pancreaticoenteric or vascular reconstruction.^13–17

Studies comparing the outcomes of robot-assisted and open MP are rare. Therefore, at a time when the number of procedures performed using the robotic approach is increasing, it is of great significance to evaluate whether such an approach offers any advantages over open surgery. The aim of the current study was to compare outcomes in patients undergoing open and robot-assisted MP, respectively.

Materials and methods

The present authors retrospectively reviewed the medical charts of patients who were initially scheduled to undergo robot-assisted MP using the da Vinci® Surgical System (Intuitive Surgical, Inc., Sunnyvale, CA, USA) between March 2010 and July 2011. Data collected and analysed included gender, age, symptoms, pathologic diagnosis, tumour size, operative time, estimated intraoperative blood loss, transfusion rate, conversion rate, length of postoperative gastrointestinal tract recovery, postoperative hospital length of stay (LoS), postoperative pancreatic fistula, morbidity, mortality, pathology and follow-up. Conversion in the robot-assisted group was defined by the inability to terminate the operation using a robotic approach.

To provide a comparative analysis, the authors reviewed the medical charts of patients who underwent open MP between April 2003 and December 2009. A total of 40 patients were identified. Four patients were found to have had invasive malignant tumours on final pathologic examination and were excluded from the present study. A comparison of perioperative clinicopathologic characteristics between the two groups was conducted. The design of this study was approved by the authors' institutional review board.

Pancreatic fistula was defined according to the guidelines of the International Study Group on Pancreatic Fistula.¹⁸ Post-pancreatectomy haemorrhage was defined according to the guidelines of the International Study Group of Pancreatic Surgery.¹⁹ Operative time was calculated as the time between skin incision and skin closure in the open surgery group, and as the time between skin incision and skin closure of the last port in the robot-assisted surgery group. Complications were defined as those occurring within 60 days of surgery. Postoperative gastrointestinal tract recovery was defined as the time to first flatus after surgery. Mortality was defined as death within the 60 days post-surgery in or out of hospital. Outpatient records combined with telephone interviews were used for follow-up. The follow-up period was defined as the interval between the day of operation and the day of the last follow-up. Follow-up was updated in May 2012. Exocrine deficiency was defined as either new-onset diabetes or deterioration in the metabolic control of previously diagnosed diabetes. Exocrine deficiency was defined as steatorrhea and weight loss requiring pancreatic enzymes supplementation.

The indication for MP during the study period was a lesion located in the neck or proximal body of the pancreas with no evidence of high-grade malignancy that could not be treated with enucleation. To select candidates for this procedure, computed tomography and ultrasonography were routinely performed. Additionally, magnetic resonance imaging, endoscopic ultrasonography and endoscopic retrograde cholangiopancreatography were used at the discretion of the surgeon. Intraoperative frozen-section examination was conducted as part of the surgical protocol to confirm negative margins and frozen pathology was performed in the event of suspected adenocarcinoma. If negative margins were not achieved, subsequent resections and frozen sections were undertaken to achieve negative margins when possible. Negative resection margins were confirmed pathologically in all patients who underwent MP (open and robotic) in the present study.

In the open surgery group, the proximal pancreatic resection was performed with a stapler, electrocautery or knife according to the surgeon's preference. Then the proximal pancreatic stump was closed with continuous or interrupted stitches without special identification and ligation of the main pancreatic duct. The distal pancreatic stump was reconstructed by pancreaticojejunostomy (duct-to-mucosa or invagination) or pancreaticogastrostomy according to the surgeon's preference. Internal stent drainage was carried out as routine practice.

In the robot-assisted surgery group, all patients were operated using a totally robotic approach. The dissection phase was generally similar to that described in the open surgery group. The remnant pancreas was managed by pancreaticogastrostomy with internal stent drainage without a transgastric approach. All robotic and open procedures were performed by the same surgeons (CP and BS).

The operative technique for robot-assisted MP was first described by Giulianotti et al.¹⁴ Modifications were later reported by Addeo et al.¹⁶ and Kang et al.²⁰ The technique used for patients in the current study was as follows. Patient positioning and setting of trocars were generally the same as those described by Giulianotti et al.¹⁴ Four trocars were placed for access by the robotic arms, and an additional 12-mm trocar was placed for the assistant's access (Fig. 1). The da Vinci® surgical arm cart was then docked. The lesser sac was entered by opening the gastrocolic ligament; the posterior gastric wall was lifted and retracted cranially using the fourth robotic arm, exposing the pancreas. The anterior surface of the portal vein was dissected at the superior edge of the pancreatic body (Fig. 2a). The superior mesenteric vein (SMV) was exposed at the inferior edge of the pancreatic neck. A retropancreatic tunnel was then created under the pancreatic neck by gentle dissection with tangential movements in relation to the vascular axis (Fig. 2b). Upon completion of the tunnel, the pancreatic neck was transected using an endoscopic stapler (Fig. 2c) or an ultrasonic scalpel. Interrupted stitches of polypropylene 4–0 were applied to the proximal stump selectively with the aim of achieving satisfactory homeostasis and reducing the risk for postoperative pancreatic fistula. The distal pancreas was then dissected progressively between the pancreas and splenic vessels to ensure a free resection margin. Small branches of the splenic vein and artery, to and from the pancreas, were selectively clipped or ligated and then transected (Fig. 2d). The transection of the pancreatic body was performed on the left side of the lesion using the robotic ultracision device (Fig. 2e). Afterwards, the distal stump was dissected about 2 cm to facilitate ensuing pancreaticogastrostomy (Fig. 2f).

Figure 1.

Ports for middle pancreatectomy. C, 12-mm trocar for camera; R1, 8-mm trocar for robotic arm 1; R2, 8-mm trocar for robotic arm 2; R3, 8-mm trocar for robotic arm 3; A, 12-mm trocar for assistant instruments

Figure 2.

Dissection phase. (a) The anterior surface of the portal vein is exposed at the superior edge of the pancreatic body. (b) A retropancreatic tunnel is created. (c) The pancreatic neck is transected at the right side of the lesion; the common hepatic artery is carefully protected. (d) The pancreatic body is dissected from the splenic vessels. The white arrow shows the management of a small branch of splenic vein. (e) The pancreatic body is transected on the left side of the lesion. (f) The distal stump of the pancreas is prepared for pancreaticogastrostomy. The white arrow shows the management of a small branch of splenic artery. CHA, common hepatic artery; Pan, pancreas; PV, portal vein; SMV, superior mesenteric vein; SpA, splenic artery; SpV, splenic vein

A short stent was inserted into the pancreatic duct without fixation (Fig. 3a). The fourth arm was used to retract the posterior gastric wall in a stable position. Two-layer pancreaticogastrostomy was then carried out. Firstly, a posterior outer layer suture was performed and the anterior wall of the distal pancreatic remnant was anastomosed to the posterior wall of the gastric body using interrupted stitches of 4–0 non-absorbable sutures from the pancreatic parenchyma to the gastric seromuscular layer (Fig. 3b). Once the posterior outer layer suture was finished, a 3–4-cm incision was made at the posterior wall of the gastric body using a monopolar hook. Then, suture of the posterior inner layer was performed using interrupted stitches of 4–0 non-absorbable sutures from the stump of the distal pancreatic remnant to the full layer of the gastric body (Fig. 3c). Subsequently, the distal stump along with a stent was put into the stomach (Fig. 3d). In the same way as the posterior layer suture, an anterior inner layer suture (Fig. 3e) followed by an anterior outer layer suture was performed (Fig. 3f). Two double-lumen drainage tubes were placed near the ends of the proximal pancreatic stump and at the site of anastomosis, respectively.

Figure 3.

Reconstruction phase. (a) The pancreatic duct is stented. (b) The posterior outer layer is sutured. (c) The posterior inner layer is sutured. (d) The distal stump of the pancreas along with a stent is put into the stomach. (e) The anterior inner layer is sutured. (f) The anterior outer layer is sutured. Pan, pancreas; SpA, splenic artery; SpV, splenic vein; stomach, posterior wall of the stomach

Variables are expressed as the median and range, and as numbers. Statistical analysis was performed using spss Version 19.0 for Windows (SPSS, Inc., Chicago, IL, USA). A two-sided P-value of < 0.05 was considered to indicate statistical significance. Comparisons between the two groups were determined using Fisher's exact test for discrete variables and Mann–Whitney U-test for continuous variables.

Results

Data comparing patients undergoing robot-assisted and open MP, respectively, are shown in Table 1. All robotic procedures were successfully completed without conversion. Five patients in the robot-assisted surgery group developed postoperative pancreatic fistulae, all of which were Grade B. Fifteen patients in the open surgery group developed postoperative pancreatic fistulae, of which nine were Grade A and six were Grade B. One patient in the robotic surgery group suffered a post-pancreatectomy haemorrhage (Grade B). No patient required reoperation in either group. Follow-up was complete in all patients in the robot-assisted surgery group and in 30 patients (83.3%) in the open surgery group. Of the six patients lost to follow-up, three died of other disease during follow-up. The median follow-up was 23 months (range: 10–25 months) in the robotic surgery group and 62 months (range: 31–108 months) in the open surgery group. All patients showed no evidence of tumour recurrence. In the robotic surgery group, no patient showed signs of exocrine or endocrine deficiency. In the open surgery group, three patients developed new-onset diabetes, but no diabetic deterioration was observed in the five patients with preoperative diabetes and no patient showed signs of exocrine deficiency.

Table 1.

Comparison between patients undergoing robot-assisted and open middle pancreatectomy

	Robot-assisted surgery group (n = 7)	Open surgery group (n = 36)	P-value
Age, years, median (range)	55.0 (30–62)	51.5 (23–76)	0.856
Female gender, n	5	27	1.000
Symptoms, n	2	16	0.680
Tumour size, cm, median (range)	3.0 (0.5–5.0)	2.1 (0.4–6.2)	0.551
Operation time, min, median (range)	210 (150–300)	226 (100–332)	0.489
Estimated blood loss, ml, median (range)	200 (50–400)	200 (50–1500)	0.432
Blood transfusion, n	0	5	0.572
Remnant reconstruction, PJ/PG, n	0/7	22/14	0.004
Diagnosis, n			0.134a
Serous cystic neoplasm	5	16
Intraductal papillary mucinous neoplasm	1	3
Other	1	17
Pancreatic fistula, n	5	15	0.222
Complications, n	6	18	0.112
GI tract recovery, days, median (range)	2 (2–3)	4 (2–11)	0.001
Postoperative LoS, days, median (range)	21 (13–33)	18 (11–107)	0.587
Mortality, n	0	0	1.000

^aKruskal–Wallis H test.

PJ, pancreaticojejunostomy; PG, pancreaticogastrostomy; GI, gastrointestinal; LoS, length of stay.

Discussion

Recently, parenchyma-sparing pancreatic surgeries for benign, borderline or low-grade malignant lesions of the pancreas have gained increasing attention.^1,3 Several factors may have contributed to this development, including evolving methods and greater confidence in the prevention and treatment of pancreatic fistula,²¹ better knowledge of the natural history of certain pancreatic neoplasms,^22–24 and growth in the numbers of asymptomatic non-invasive lesions diagnosed as a result of the widespread availability of high-resolution imaging techniques.²⁵ Middle pancreatectomy, which is indicated for the treatment of benign, borderline and low-grade malignant lesions of the neck or proximal body of the pancreas in place of extended right or left pancreatic resection, is currently accepted as a rational parenchyma-sparing procedure not only in terms of safety and effectiveness, but also in terms of the preservation of pancreatic function.^2,4,5

Meanwhile, minimally invasive surgical techniques have been increasingly applied in pancreatic surgery in an effort to decrease the morbidity associated with open surgery.^8,10,26 This is particularly true for distal pancreatic resections, which account for most resections currently performed using laparoscopy because they can be carried out with relative ease and do not require reconstruction.²⁶ However, only approximately 20 laparoscopic MP procedures have been reported so far,^7–12 which indicates that this is a demanding procedure, mainly because of the intrinsic limitations of laparoscopy.

In order to compensate for the disadvantages of conventional laparoscopy, robot-assisted surgery has been introduced and is slowly gaining acceptance, even in complex and difficult pancreatic resections that require pancreaticoenteric or vascular reconstruction,^13–17 thereby making minimally invasive techniques in demanding pancreatic procedures such as MP now possible (Table 2). However, information on the advantages of using a robotic approach rather than open surgery in MP is rare. Therefore, a comparison of the outcomes of robot-assisted and open MP was performed.

Table 2.

Reported series of robot-assisted middle pancreatectomy

Year	Study	Patients, n	Operation time, min, median	Blood loss, ml, median	ToR	Blood transfusion, n	Conversion, n	Mortality, n	Morbidity, n	Pancreatic fistula, n	Hospital stay, days, median
2010	Giulianotti et al.14	3	310	199	PG	0	0	0	1	1	9
2010	Zureikat et al.15	4	NA	NA	PJ	0	0	0	NA	3	NA
2011	Addeo et al.16	1	450	Minimal	PG	0	0	0	0	0	15
2010	Kang et al.20	5a	480	200	PG	0	0	0	1	1	12
2010–2011	Current study	7	210	200	PG	0	0	0	6	5	21

^aTwo laparoscopic resections and robotic reconstruction.

ToR, type of reconstruction; PJ, pancreaticojejunostomy; PG, pancreaticogastrostomy; NA, not available.

Since the da Vinci® robot-assisted surgical system became locally available in March 2010, a selective robotic approach in pancreatic surgery has been adopted, especially for parenchyma-sparing resections.²⁷ Over the past 2 years to date, seven consecutive patients have been scheduled to undergo robot-assisted MP, all of whom did so successfully without conversion. All seven of the patients had benign, borderline and low-grade malignant tumours. Four patients with asymptomatic serous cystic neoplasms which met current recommendations for resection²⁸ underwent robot-assisted MP in the present study. Of these, one had a tumour of 5 cm in diameter and one had a tumour with a progressive increase in diameter to 4 cm. In the remaining two patients, the diagnosis of mucinous cystic neoplasm could not be safely excluded preoperatively.

Minimal surgical trauma should represent the initial goal of minimally invasive pancreatic surgery. However, in the current series, the use of robotic surgery did not incur a statistically significant reduction in intraoperative estimated blood loss. By contrast, unlike previous reports of increased operative time in robot-assisted surgery,²⁰ the current experience showed that the operative time required to complete robot-assisted MP was comparable with that required for open MP. These results are encouraging; it would seem that robot-assisted surgery may become associated with operative time that is at least comparable and potentially shorter than that in open surgery as surgeons accumulate more experience.

As Giulianotti et al.¹⁴ reported in a description of the world's first robot-assisted MP in 2010, two phases of this surgical procedure are greatly facilitated by the robotic system; these include the dissection of the pancreatic body from the splenic vessels and the pancreaticoenteric reconstruction. The three-dimensional vision and the flexible wrist-like movements afforded by the robotic system guarantee the easier recognition and dissection of the small tributaries of the splenic vessels directed to and from the pancreatic body¹⁴ (Fig. 2d–f). In addition, the fourth arm of the robotic system provides a stable platform from which to operate.

During the reconstructive phase, pancreaticogastrostomy has so far represented the preferred choice in robot-assisted MP,^14,16,20 despite the ability to perform robotic pancreaticojejunostomy.¹⁵ Pancreaticogastrostomy does not require bowel mobilization and division and thereby avoids any increase in operative time, the interruption of intestinal continuity and potential leaks. The present authors have adopted a two-layer pancreaticogastrostomy with internal stent drainage (Fig. 3), which resembles the procedure carried out in patients undergoing open MP with pancreaticogastrostomy. External stenting was not performed¹⁶ and a transgastric approach to pancreaticogastrostomy²⁰ was not adopted for patients in this series because of the additional time and trauma associated with these methods and the potential risk for gastric leakage.

The most predominant complication and biggest concern associated with cases of MP reported in the literature is the occurrence of pancreatic fistula, which has been reported to occur at frequencies of 8–50%.^2,4,5 The presence of two pancreatic stumps and a soft pancreatic parenchyma may be primarily responsible for this finding. Neither the approach (open, laparoscopic or robotic) nor the type of reconstruction (pancreaticogastrostomy or pancreaticojejunostomy) appear to affect the occurrence of pancreatic fistula.^{2,4,10,12,14,15,20} The incidence of pancreatic fistula in the robotic surgery group in the current study was comparable with those in previous reports of robot-assisted surgery, which ranged from 20% to 75%.^14,15,20 Although the difference between the two groups in the present study was not statistically significant, the rate of pancreatic fistula was higher in the robotic surgery group. However, the postoperative hospital LoS did not increase, which may largely reflect the results of the minimal trauma caused by the robotic procedure. Further, in the present study, median time to postoperative gastrointestinal tract recovery was significantly shorter in the robot-assisted surgery group than in the open surgery group [2 days (range: 2–3 days) versus 4 days (range: 2–11 days); P = 0.001]. Overall, the present authors consider that the potential advantages of the robotic approach might be neutralized by the high rate of postoperative pancreatic fistula identified in the current study. As a result of differences among health care systems, the postoperative hospital LoS was much longer in the present study than at other centres.^14,16,20 Despite the excellent technical improvements in robotic surgical systems, it is presumed that performing the operation as precisely as in open surgery will require the surgeon to progress along a learning curve. Thus, the present authors believe that the accumulation of experience beyond the learning curve phase may decrease the complication rate.

Many authors have cited the costs associated with the use of robotic technology.^13,15,20 This was not assessed in the present series because patients were required to pay an additional sum of money if they chose to be operated using a robotic approach at the time they gave informed consent to the procedure. Thus it was certain that the robotic approach would entail an initial increase in cost. Whether the potential benefits to be derived from robot-assisted surgery may counterbalance the overall costs of surgery remains unknown and requires further investigation.

In conclusion, although the present study is subject to obvious limitations imposed by the small number of patients investigated, the study's retrospective design and a degree of selection bias, the current experience demonstrates that it is safe and feasible to perform MP in selected patients using a robotic system and that patients undergoing robot-assisted MP may benefit from a quicker gastrointestinal tract recovery. Larger series and controlled trials comparing outcomes of robot-assisted and open MP are necessary in order to fully elucidate these potential advantages.

Conflicts of interest

None declared.