Minimally invasive procedures are increasingly being recognized as standard treatments for many intra-abdominal pathologies, such as colorectal and gastric malignancies [1, 2]. However, it has been more difficult to define the safety and efficacy of minimally invasive surgery for both benign and malignant pancreatic disease [3–7]. Minimally invasive techniques for the management of pancreatic cancer have been performed since the 1960s; diagnostic laparoscopy has been used to detect peritoneal metastases [8]. Since the 1990s, laparoscopic enucleations and distal pancreatectomies have been performed with comparable and in some instances improved outcomes compared with open techniques [9–11]. For the treatment of benign tumors, minimally invasive distal pancreatectomies are arguably considered the “gold standard” [12, 13].

In 1992, Gagner and Pomp [14] successfully performed the first minimally invasive pancreatoduodenectomy (MIPD). This was done for the treatment of a patient with chronic pancreatitis. Since that first report, multiple international centers have performed this procedure and have reported their results in the literature [15, 16]. Unfortunately, only a few high-volume centers have published case reports in the double digits and even fewer of these adequately discuss the issues pertaining to suitability for the minimally invasive approach in the treatment of malignancies in the head of the pancreas [17–20].

Robotically assisted laparoscopic surgery has been available since the 1990s and began with robotically controlled laparoscope holders [21]. Because of the inability for surgeons to ergonomically perform minimally invasive procedures in the pelvis, complete surgical systems (DaVinci, Intuitive Surgical, Inc., Sunnyvale, CA) have thrived in gynecologic, urologic, and rectal surgery. However, for procedures in the upper abdomen, where the patient can be placed in the low lithotomy (“French”) position and the operating surgeons can stand between the patient’s legs and operate ergonomically, the complete robotic system has not been as popular [22]. This also may be due to the increased potential for massive hemorrhage when dissecting around the portal vein and inferior vena cava (IVC) in the absence of haptics, and because of the fact that surgeons at the robotic console are not in contact with the actual patient or sterile, resulting in a delay if urgent conversion to an open approach is needed [23]. In addition, the operating surgeon cannot utilize a hand-assisted approach when indicated, causing many robotic surgeons to simply convert to open approaches instead of continuing in a laparoscopic hand-assisted fashion [24]. Because of this, a review of two international centers using a robotically assisted laparoscopic approach to pancreatic pathologies that also enables the operating surgeon to feel was undertaken to see if the results would be comparable to other reports of minimally invasive pancreatectomy.

Methods

From March 1994 to June 2011, a total of 200 laparoscopic pancreatic procedures were performed by 2 surgeons, from North America and Western Europe. Until 2006, all procedures were performed with a nonsterilizeable robotically controlled camera holder (AESOP, Computer Motion, Inc., Goleta, CA); in Western Europe a second, sterilizeable robot (ViKY, Endocontrol, Grenoble, France) became available in 2006, followed by North America in 2007. All laparoscopic procedures begun with the intention of completing the procedure minimally invasively were included in this study. At our institutions, all pancreatectomies are approached laparoscopically unless patients have absolute contraindications to pneumoperitoneum, such as closed angle glaucoma, intracranial hypertension, or bullous emphysema. Patients who were deemed to require a portal vein reconstruction preoperatively also were begun via an open incision. Due to the high variability of procedures performed and the complexity of the vascular dissection, analysis of results was limited to patients who underwent duodenopancreatectomy and left-sided pancreatectomy. Morbidity was classified by using the Clavien System.

A review of published reports of minimally invasive distal pancreatectomy (MIDP) from 1997 until 2010 was undertaken and included the search words “laparoscopic” and ”distal pancreatectomy.” Robotic reports using the full robotics system (DaVinci, Intuitive Surgical) and/or hand-assisted procedures also were included. Studies with volumes <15 were excluded, except one report by Gumbs et al. because of its high percentage of cases done for cancer. Due to the decreased frequency of laparoscopic duodenopancreatectomy, all published reports were analyzed. Duplicate and imprecise data were excluded. The following variables from articles found on Pubmed were studied: number of cases, conversion rate, mean operating time, mean estimated blood loss (EBL), mean length of stay, malignancy rates, margin status, and morbidity and mortality rates. A weighted average adjustment form statistical analysis was used to obtain the statistical sum of all the means for the rest of the variables:

$$ W_{\text{a}} = (X_{1} Y_{1} + X_{2} Y_{2} + \cdots + X_{n} Y_{n} ) \, / \, (X_{1} + X_{2} + \cdots + X_{n} ) $$

where X is the number of cases in a report and Y is the mean for the variable studied in that report.

Operating room setup of robotically controlled laparoscope holder

The robotically controlled laparoscope holder is placed to the left if the nonsterilizeable laparoscope holder (AESOP, Computer Motion Inc., Santa Barbara, CA) is used regardless of the location of the pancreatic tumor (Fig. 1), but at the level of the right axilla for right-sided lesions if the sterilizeable robot (ViKY, Endocontrol, Grenoble, France) is used and on the left for left-sided lesions (Fig. 2) [21]. Both robots can be controlled with foot pedals or voice activation (Fig. 3).

Fig. 1
figure 1

Operating room setup with a nonsterilizeable robotically controlled laparoscope holder (AESOP, Computer Motion, Inc., Goleta, CA, USA) to the patient’s left. The patient is in the low lithotomy or “French” position. Unlike complete robotic systems where the operating surgeon is not in contact with the actual patient, high-definition monitors (M), laparoscopic ultrasound devices (UD), laparoscopic ultrasonic shears (USS), and an autostatic liver retractor enable the operating laparoscopic surgeon to control all aspects of the case

Full size image
Fig. 2
figure 2

Operating room setup for a laparoscopic distal pancreatectomy with a sterilizeable robotically controlled laparoscope holder (ViKY, Endocontrol, Grenoble, France). As opposed to AESOP, this device is completely sterilizeable and fits directly on the patient

Full size image
Fig. 3
figure 3

The sterilizeable robotically controlled laparoscope holder (ViKY, Endocontrol, Grenoble, France) with foot pedal

Full size image

Laparoscopic distal pancreatectomy

The steps necessary to perform laparoscopic duodenopancreatectomy and distal pancreatectomy have been extensively described in multiple previous publications [4, 17, 25, 26]. When embarking on minimally invasive pancreatic surgery, as with open surgery, we recommend beginning with distal pancreatectomy. Splenic preservation can be attempted in patients without concerns for cancer. A medial to lateral approach is favored, but a lateral to medial approach may be necessary for larger tumors.

Laparoscopic duodenopancreatectomy

The choledochojejunostomy is created laparoscopically before the pancreatic anastomosis is fashioned. It is created in an end-to-side fashion, the posterior layer is done with running suture and the anterior layer either running or with interrupted stitches. A 5-Fr pediatric feeding tube is used as an internal stent for common bile ducts less than 1 cm in diameter. The pancreatic anastomosis is then fashioned laparoscopically either in one or two layers to the stomach or jejunum depending on the patient’s anatomy and the surgeon’s preference. All ducts less than 1 cm are similarly stented with a 5-Fr pediatric feeding tube. For tubes smaller than this, angio-catheters may be necessary. The anastomosis to the stomach or duodenum can then be performed laparoscopically or via the extraction site. Two closed suction drains are left in the area of the biliary and pancreatic anastomoses, respectively. Drain amylase levels are checked on postoperative day #3 and removed if patients are tolerating a low-fat diet and levels are not elevated greater than three times the upper limit of the normal serum value.

Results

A total of 72 duodenopancreatectomies, 67 distal pancreatectomies, 23 enucleations, 20 pancreatic cyst drainage procedures, 5 necrosectomies, 5 atypical pancreatic resections, 4 total pancreatectomies, and 4 central pancreatectomies were performed.

The mean age of patients undergoing distal pancreatectomy is 57 years. The mean estimated blood is 100 mL (range 0–1,500 mL) and the mean operating time 203 min (range 120–460 min). Major complications occurred in 14 (21 %) patients, and pancreatic leaks occurred in 13 (19 %) patients (Type A = 5, Type B = 4, Type C = 4). Ten (15 %) patients required conversion to either a lap-assisted or open approach, and six (9 %) patients required reoperation. One patient died on postoperative day 35 due to hemorrhage for a mortality rate of 0 % at 30 days and 1.5 % at 90 days. The average length of stay was 6 days after distal pancreatectomy (range 2–43).

Thirty-six (54 %) of the laparoscopic distal pancreatectomies were for malignancy: 20 adenocarcinoma (including 3 mucinous cystadenocarcinoma and 3 arising from an IPMN), 10 neuroendocrine, 4 renal cell metastases, 1 pulmonary metastasis, and 1 due to a gastric cancer invading into the tail of the pancreas. The remaining benign pathology includes 11 mucinous cystadenomas, 9 benign neuroendocrine tumors (including 2 borderline lesions), 4 serous cystic adenomas, 4 IPMN, 2 with chronic pancreatitis, and 1 epithelial cyst. In the patients with adenocarcinoma, there was one patient with stage 0 disease, two with stage IA, five with stage IB, three with stage IIA, six with stage IIB, and three with stage 3. Lymph node retrieval average was 10 (range 3–17). In all resections for malignancy, only one patent had an R1 resection for a positive margin rate of 3 %. Follow-up ranges from 8 to 28 months and averages 18 months; 16 (80 %) patients are currently alive, 13 (65 %) of them without evidence of disease.

The mean age for patients undergoing duodenopancreatectomy is 65 years. The mean EBL is 400 mL (range 0–200 mL), and the mean operating room time is 436 min (range 255–660 min). A total of 24 (33 %) patients suffered a major complication. A total of 14 (19 %) patients required conversion to an open approach and 8 (12 %) to a hand-assisted approach. Sixteen (22 %) patients developed a pancreatic leak and 19 patients required reoperation (26 %). Most reoperations were for postoperative hemorrhage [10] or intra-abdominal infection [7]. However, one reoperation was to rule out bowel ischemia, and a second was due to an obstruction at the transverse colon mesentery, both of these were managed laparoscopically. In total, 58 % of patients reoperated on were managed laparoscopically. One patient died on postoperative day 10 due to hemorrhage and infected ascites, for a mortality rate of 1.4 % at 30 and 90 months. Length of stay ranged from 4 to 38 days and averaged 9 days.

Sixty (83 %) of the laparoscopic duodenopancreatectomies were for malignancies: 27 adenocarcinomas, 18 ampullary cancers, 8 distal cholangiocarcinomas, 4 duodenal cancers, and 3 malignant neuroendocrine tumors. The remaining pathologies were benign and included seven intraductal papillary mucinous neoplasms (of which 4 had borderline pathology), two chronic pancreatitis, one pseudopapillary tumor, one osteoclastic tumor, and one duodenal polyp in a patient with familial adenomatous polyposis. Staging was done according to the AJCC guidelines and included all patients with adenocarcinoma and periampullary tumors and included 3 with stage 0 disease, 5 with stage IA, 4 stage IB, 9 stage IIA, 23 stage IIB, 6 stage 3, and 3 stage 4. Fifty-seven patients had an R0 resection and three patients had an R1 resection for a 95 % negative margin rate. An average of 16 (range 6–24) lymph nodes was retrieved. Follow-up ranges from 6 to 30 months and averages 19 months; 36 (60 %) patients are currently alive, 26 (43 %) of them without evidence of disease.

Discussion

This paper describes our experience with laparoscopic pancreatectomy using a robotically controlled laparoscope holder called ViKY (VideoendosKopY) made by Endocontrol, not the complete robotic system, DaVinci, by Intuitive. ViKY costs approximately $60,000 and is sterilizeable and completely reusable and there are no disposable parts. Dr. Zeh’s group from the University of Pittsburgh reported that robotic distal pancreatectomy using the complete surgical system may be superior to laparoscopic distal pancreatectomy in conversion rates to open (0 vs. 16 %) [27]. Interestingly, this was in a cohort with 43 % malignancy in the robotic group compared with only 15 % in the laparoscopic group. The robotic cases also had statistically significant negative margin rates and improved lymph node retrieval [27]. These findings could be confounded by the possibility that these surgeons are simply more comfortable using the complete robotic system. Larger series with the complete surgical system are needed to confirm or refute these findings.

Minimally invasive distal pancreatectomy

To date 1,141 cases of MIDP have been reported in the literature (Table 1). The adjusted mean EBL in these cases is 250 mL (range 100–667 mL), the mean adjusted operating time is 215 min (range 145–308 min), the adjusted mean LOS is 7 days, and mean follow-up is 23 months. Conversions occurred in 12 % of cases because of uncontrollable hemorrhage, unclear margin status, inability to localize the lesion, and adhesions (Table 2). The reported resection status was reported in only 208 cases (18 %); margins were found to be positive in 11 % of cases for an R0 resection rate of 89 %. Reported lymph node retrieval ranged from 5 to 15.

Table 1 Summary of LDP procedures in the literature
Full size table
Table 2 Number of conversions to open procedures per study
Full size table

Complications occurred in 23.4 % (268 patients) of cases. As in the open experience, pancreatic fistula was the most common cause of morbidity and occurred in 14.9 % of cases (171 patients; Table 3). Other reported morbidity includes delayed gastric emptying, postoperative hemorrhage, infection, deep venous thrombosis, and anemia found in 97 cases (8.5 %). Mortality was reported in 8 cases (<1 %).

Table 3 Complications of LDP procedures
Full size table

Initial concerns regarding MIDP included concerns regarding the safety of the approach and its cost-effectiveness [12, 28–32]. Nonetheless, numerous studies became available that demonstrated the safety, feasibility, and cost-neutral aspect of this technique [12, 28–44]. Unfortunately, there are still no randomized, controlled trials comparing MIDP to the open approach. Operating room times also do not appear significantly different when compared to the open literature. Morbidity and mortality rates appear similar to the open literature (Tables 1, 3) [12, 28–44]. As with open distal pancreatectomy, pancreatic fistula remains the most prevalent complication after MIDP and depends on numerous parameters including presence or absence of soft pancreatic parenchyma and the malignancy status of the pancreatic disease. Although essentially all maneuvers and modalities attempted to reduce the pancreatic leak rate after distal pancreatectomy have also been attempted in minimally invasive approaches, there is still no “gold standard” for pancreatic parenchymal transection [26, 45].

Minimally invasive pancreaticoduodenectomy

Since 1992, 436 MIPD procedures have been reported in the literature (Table 4), series with multiple publications. The adjusted mean EBL was reported in 409 (94 %) of the 436 published cases and was 267 mL (range 50–1,200 mL). The adjusted mean operating time was 396 min (range 263–750 min); interestingly, in series with at least 20 patients the mean operating time ranges decrease to 287–435 min. The mean LOS was 14 days (range 7–37 days) and was reported in 391 (90 %) cases. The adjusted mean follow-up was 23 months (range 3–51 months) and was only reported in 251 (58 %) cases. In these series, 10.5 % of the cases required conversion to an open approach (Table 5). Conversions occurred for the following reasons: invasion into the portomesenteric confluence, uncontrollable intraoperative hemorrhage, positive margin on intraoperative frozen section, planned conversion for the reconstruction, and robotic malfunction.

Table 4 Summary of MIPD procedures in the literature
Full size table
Table 5 Number of conversions to open procedures per study
Full size table

Morbidity occurred in 151 (38 %) of the 395 cases that reported complication rates. Pancreatic and or biliary fistula occurred in 71 (18 %) patients (Table 6). Other complications included: intraperitoneal hemorrhage 17 (4 %) cases, delayed gastric emptying 17 (4 %) cases, deep venous thrombosis/pulmonary embolism in 9 (2 %) cases, wound infection in 7 (1.8 %) cases, intra-abdominal abscess in 6 (2 %) cases, postoperative ileus in 6 (2 %) cases, gastrointestinal hemorrhage in 6 (2 %) cases, and anemia in 5 (1 %) cases. Remaining reported complications included pneumonia, urinary tract infections, and colitis. The mortality rate was reported in 92 % of cases (399 patients) and is 2 % (7 patients).

Table 6 Complications of MIPD procedures
Full size table

In cases of malignancy margins were reported in 255 cases, 243 patients had an R0 resection for a positive margin rate of 5 % (Table 7). A mean of 15 (range 5–54) lymph nodes were retrieved. The majority of tumors removed were for adenocarcinoma and periampullary tumors: 154 (50 %) adenocarcinoma and 82 (27 %) periampullary (Table 8). Chronic pancreatitis is the most common nonmalignant disease treated and accounts for 8 % (29 patients) of all cases.

Table 7 Summary of malignant cases with retrieved mean number of lymph nodes and rate of R0 resections per study
Full size table
Table 8 Pathologic findings in MIPD procedures
Full size table

Gagner’s review of minimally invasive duodenopancreatectomy, published in 2009, reviewed the experience with 146 cases and noted a conversion rate of 46 % [46]. In our review, the conversion rate has decreased to 11 % in the literature; in our series it is 19 % probably due to our high rate of malignancy (Table 5). Morbidity rates in high-volume pancreatic centers have been reported to be as high as 54 %, which is comparable to the rate reported in the minimally invasive literature (38 %) and in our series [47]. Our relatively high rate of reoperation (26 %) is more due to a culture of aggressive reoperation early in our experience and is tempered by our relatively low rate of major complications dealt with by our interventional radiologists (7 %). Some of the early cases of delayed hemorrhage were due to the use of only clips on large vessels. Now, large vessels are either stapled with laparoscopic vascular staplers or clips are oversewn with prolene suture. Currently, most of our major complications are dealt with by interventional radiology. Unlike most series, our series has a high malignancy rate, implying that with adequate experience both benign and malignant tumors and not just periampullary tumors can be approached safely and successfully laparoscopically. This is supported by the fact that mortality rates after minimally invasive duodenopancreatectomy are extremely low in both the literature review and in our series.

The vast majority of reoperations occurred early in our experience with laparoscopic duodenopancreatectomy and was precisely due to the fact that the minimally invasive Whipple procedure is such an experimental and controversial procedure. In fact, 80 % (15 patients) of our reoperations occurred in our first 25 patients. Since then, our reoperation rate has decreased to 11 % (4 patients). As our experience grows, it is hopeful that we will continue to enjoy a decrease in reoperation rate. In addition, due to the fact that patients were done laparoscopically, perhaps our threshold to reenter the abdomen was lower because a diagnostic laparoscopy may be easier tolerated than a second-look laparotomy. Compared to The John’s Hopkins experience of 650 consecutive pancreaticoduodenectomies, they reported a reoperation rate of only 3.5 % [47]. Nonetheless, we are effectively at the equivalent of their first 10 % of cases (62 vs. 650). When we have an experience of 650 laparoscopic duodenopancreatectomies, a comparison of reoperation rates may be more useful. Interestingly, after only 1/10th of their experience, our mortality rates are already equivalent: 1.4 vs. 1.4 % [47].

Wound infections have been noted in 10 % of patients after open Whipple procedures, which is markedly elevated compared with the 2 % incidence after laparoscopic duodenopancreatectomies (Table 6). Delayed gastric emptying also has been found in almost 20 % of cases after open duodenopancreatectomies compared with only 4 % in the minimally invasive literature (Table 6). All of these factors contribute to length of hospitalization; however, due to the high degree of variability in policies regarding discharge criteria among different countries, it is difficult to meaningfully comment on differences in length of stay between the modalities [16].

As with all procedures as surgeon’s get through their learning curves, operating room times decrease. Kendrick et al. noted a decrease in operating time from 7.7 to 5.3 h when comparing his first 10 cases with his last 10 cases in a series of 62 patients [20]. Because of this, it is still recommended that minimally invasive surgeon’s began with tumors less than 2 cm, without involvement of the portomesenteric vessels and with pancreatic ducts larger than 4 mm [3].

Conclusions

In reviewing the literature and this series of MIDP and duodenopancreatectomy with a robotically controlled laparoscope holder, it becomes clear that this approach has similar morbidity and mortality rates compared with other minimally invasive techniques, specifically laparoscopy, hand-assisted approaches, and use of a complete robotic system. Furthermore, adequate R0 resection rates and lymph node retrieval can be attained. Patients who required conversion to an open approach more commonly had malignant disease and an increased body mass index (BMI) [48]. Because of this, patients with these criteria should be more vigorously considered for an open or hand-assisted approach from the onset, especially early in a surgeon’s learning curve [6, 49]. As mentioned, the hand-assisted approach has potential benefits but is not possible for the operating surgeon when using the complete surgical system.

Perhaps the greatest concern for the use of complete robotics in minimally invasive pancreatic surgery is the absence of haptics and inability of the operating surgeon to maintain contact with the patient when using the complete robotic system. In the future, hand-held and smaller robotically controlled laparoscopic instruments used in conjunction with a robotically controlled laparoscope holder may enable pancreatic surgeons to enjoy the benefits of robotics and maintain the sense of touch (Fig. 4).

Fig. 4
figure 4

Prototype of hand-held robotically controlled laparoscopic instruments (JAIMY, Endoconrol, Grenoble, France)

Full size image