Introduction

Pancreatic neuroendocrine tumors (PNET) represent about 2–4 % of all clinically detected pancreatic tumors [1], with an incidence of 2–3 per million based on data from the US and Norway [2, 3]. The incidence is even higher in autoptic material [4] and has increased over the last 40 years [2, 5].

Since the first laparoscopic pancreatic resection was performed in 1992 [6], several studies have shown that laparoscopic pancreatic surgery (LS) can be performed safely [7–13]. The laparoscopic approach has also been beneficial compared with open surgery, with advantages such as shorter hospital stay, better cosmesis, and fewer complications [14–17].

Because most PNET are relatively small and solitary, they may be considered well suited for removal by a minimally invasive approach. However, there are few large series that describe laparoscopic surgery of PNET [18–20]. Throughout the last decade tumors of the pancreatic body and tail have been resected primarily by LS at our institution. Exceptions have been primarily pancreatic head tumors and locally advanced tumors with vascular involvement, which could be radically resected only with open surgery and vascular reconstruction. The latter has been the case in less than ten patients since 1997. Therefore, the primary aim of this study was to describe the feasibility, outcome, and histopathology associated with LS of PNET in a large series. A secondary aim was to evaluate the prognostic value of the World Health Organization (WHO) 2010 classification, European Neuroendocrine Tumor Society (ENETS) TNM classification, and resection status.

Methods

All patients with PNET who underwent laparoscopic surgery at our department from March 1997 to April 2011 were included retrospectively. Patients were grouped according to the laparoscopic procedure performed and the functional activity of the tumor. Data were extracted from patient journals from our hospital and journals from local hospitals if the patient was transferred to such hospital before discharge. Parameters included demographics, operating time, intraoperative blood loss, rate of postoperative pancreatic fistula (POPF), conversion rate, and length of hospital stay.

When multiple tumors were found in the pancreas, the diameter of the largest tumor was registered. Surgical morbidity was defined by the revised Accordion Classification, but excluding POPF, which was seen as a separate complication [21]. Mortality was defined as death prior to discharge. Indication for surgery was based on patient history, clinical symptoms, and biochemical, radiologic, and endoscopic findings. Computer tomography (CT), magnetic resonance imaging (MRI), and/or ultrasonography were performed regularly in preoperative diagnostics. Preoperative octreotide scintigraphy and assessment of chromogranin A in serum were performed when neuroendocrine disease was suspected and specific serum hormones were quantified according to the suspected functioning disease. Endoscopic ultrasound and endoscopic fine-needle aspiration were generally not part of the preoperative evaluation. Preoperative percutaneous biopsy of solid pancreatic lesions was generally avoided because of the risk of tumor dissemination [22]. The study was approved by the Hospital Review Board.

Surgical procedures

Patients were operated on under general anesthesia, preferably by total intravenous administration. The majority of patients were placed in a modified supine position with the left side raised as described previously [7, 8]. The procedures were planned before surgery, but the final approach was decided on during surgery based on a combination of macroscopic and ultrasonographic findings, frozen-section histopathological analyses of the suspected PNET itself, and the nature of the tissue itself. Intraoperative ultrasound (IOUS) was performed on all patients for affirmation and exact location of the tumor and its exact relationship to the pancreatic duct. Trocar position, exploration, and exposure of the body and tail of the pancreas were performed as earlier reported, as was the case for distal pancreatectomy with and without splenectomy [7].

From 2005 to 2010, the resection margin of the remaining pancreas (in enucleations and in distal resections) was sealed by a fibrinogen/thrombin-coated collagen sponge (TachoSil®; Nycomed, Roskilde, Denmark) [23]. From 2010, the resection margin was sutured with a knotless wound closure device (V-Loc®; Covidien, Dublin, Ireland). At the end of the procedure, a drain routinely was placed near the resection line of the pancreas. All patients in whom splenectomy was performed were immunized against pneumococcal infection before discharge or transfer to another hospital.

Enucleation

Lesions smaller than 3 cm located in the pancreatic head, body, or tail and were considered noninvasive and not affecting the main pancreatic duct, were removed by enucleation. If possible, the remaining parenchymal surface was adapted with suture. At the end of the procedure, a drain routinely was placed near the enucleation area of the pancreas. Preoperative stent inlay in the pancreatic duct before enucleation of tumors in the pancreatic head was performed when lesions were located very close to the main pancreatic duct in order to minimize the risk of damage to the pancreatic duct and subsequent pancreatic leakage.

Laparoscopic exploration and conversion to laparotomy

Procedures were limited to laparoscopic exploration in cases of inoperable disease, which was first discovered intraoperatively, or when tissue affected by recent pancreatitis did not allow for laparoscopic removal. For cases in which resection and identification of tumor tissue and severe bleeding could not be managed in laparoscopic technique, conversion to laparotomy was performed.

Perioperative patient care

The operative drain was removed when amylase levels in drainage fluid were less than three times the serum levels. Pancreatic fistulae were diagnosed and graded according to the International Study Group on Pancreatic Fistula (ISGPF) definition [24]. An abscess was defined as a fluid remnant related to the operation area, diagnosed with CT, MRI, and/or ultrasonography, with simultaneous infectious clinical signs and symptoms, and an increase of infection parameters in the blood (CRP, leucocytes, and/or procalcitonin). If patients developed POPF, they were not classified according to the revised Accordion Classification to prevent double counting. Short-acting somatostatin analogs were used only in cases of POPF.

Histopathological data

Histopathological data were acquired according to the WHO 2010 classification for neuroendocrine neoplasms of the gastroenteropancreatic system [25] and the ENETS TNM classification [26]. In addition, resection status, biological behavior, tumor location, and median tumor diameter were recorded.

Statistical analysis

Continuous data were presented as median (range) and analyzed using the Kruskal–Wallis test. Statistical comparisons were carried out between four major groups of laparoscopic pancreatic procedures [distal pancreatic resection with splenectomy (DPS), distal pancreatic resection without splenectomy (DP), enucleation from the pancreatic head (ENH), and enucleation from the pancreatic body or tail (ENBT)]. Post-hoc analysis was performed using the Tukey’s test after applying the Bonferroni method for correction of multiple comparisons. Disease-specific survival was estimated using the Kaplan–Meier method with 95 % confidence intervals. The log-rank test was used to compare differences in survival among subgroups. p values <0.05 were considered statistically significant. Statistical analysis was performed using PASW Statistics ver. 18.0.3 (SPSS, Chicago, IL, USA).

Results

At our institution there has been a steady increase in LS since the first operation was performed in 1997. Of 241 laparoscopic procedures performed on the pancreas until April 2011, 75 have been performed for PNET. Today, practically no suspicious PNET of the pancreatic body or tail undergoes open surgery at our institution.

Seventy-two patients with PNET underwent laparoscopic surgery in the period from March 1997 to April 2011. The median age of all patients at the time of surgery was 57 (21–87) years and the median BMI was 26 (18–39) kg/m2. Thirty-eight patients were female. The overall median American Society of Anesthesiologists (ASA) grade was 2 (1–3) (Table 1). There was an increasing number of laparoscopic procedures for PNET performed throughout this period, with three operations performed in 1997 and a maximum of 11 in 2009. Of the 75 laparoscopic procedures, 65 were pancreatic resections or enucleations, 4 were conversions to laparotomy, 2 were exploratory laparoscopies with pancreatic biopsies, and 4 were exploratory laparoscopies only. Most of the patients underwent a distal pancreatic resection with splenectomy (DPS) (n = 28), followed by distal pancreatic resection without splenectomy (DP) (n = 23), enucleation from the pancreatic head (ENH) (n = 7), and enucleation from the pancreatic body or tail (ENBT) (n = 7). A single-port laparoscopic procedure was performed in two patients, both of whom underwent distal pancreatic resection without splenectomy. Two patients underwent repetitive laparoscopic surgery. One patient with an insulinoma in the pancreatic head first had exploratory laparoscopy that revealed tissue affected by recent pancreatitis, which did not allow for surgery, and later underwent laparoscopic ENH. Another patient with an insulinoma in the pancreatic head first underwent exploratory laparoscopy with IOUS, which could not confirm any pancreatic lesion. The patient did later undergo two additional laparoscopies with use of endoscopic transduodenal ultrasound. Biopsies from suspicious lesions in the pancreatic head were taken in each of the latter operations. However, pathology could not confirm neuroendocrine tissue. The patient did later undergo open enucleation of the insulinoma.

Table 1 Characteristics of 65 patients who underwent laparoscopic removal of pancreatic neuroendocrine tumors
Full size table

Combined resections of adjacent organs were performed in eight patients. A concomitant adrenalectomy was performed in three patients, an adrenalectomy with resection of grossly enlarged retroperitoneal lymph nodes was performed in two patients, an adrenalectomy with liver resection was done in one patient, a partial gastrectomy and segmental colon resection was performed in one patient, and a hysterectomy with bilateral salpingo-oophorectomy was performed in one patient. Palliative resections were performed in ten patients, all with synchronous inoperable liver metastases.

The most frequent surgical complication was POPF, which occurred in 14 of the 65 patients who underwent laparoscopic removal of PNET (Table 1). There was 1 grade A fistula, 13 grade B fistulae, and no grade C fistula. POPF developed in 7 of 14 patients who underwent enucleation and in 7 of 51 patients who underwent resection. The difference in rate of POPF between the groups was significant (p < 0.05). However, post-hoc subgroup analysis could not confirm this.

When considering other surgical complications according to the revised Accordion Classification, grade 1 was present in five, grade 2 in six, grade 3 in one, grade 4 in two, and no patients had grades 5 and 6 (Table 1). There were no significant differences in overall surgical morbidity between the groups (p = 0.446).

In four patients a laparoscopic procedure was converted to laparotomy: one was converted due to intraoperative bleeding that could not be controlled laparoscopically, one was converted due to a tumor in the pancreatic head that was not available for laparoscopic enucleation, one had an insulinoma and was converted because a tumor in the pancreatic head was not found with laparoscopy and IOUS but was seen after laparotomy upon which an open enucleation followed, and one had recurrence of a PNET in the pancreatic tail and was converted due to intraoperative bleeding from the remaining end of the splenic artery that could not be controlled laparoscopically. Five patients were reoperated on due to complications: one patient had laparoscopy-assisted drainage of an abscess because of the danger of injury to intra-abdominal organs if a percutaneous drainage was performed, one patient was resutured for a postoperative periumbilical wound rupture, one patient had a laparoscopic revision for postoperative hemorrhage from the resection surface of the pancreas, one patient underwent laparoscopy for postoperative hemorrhage from the great omentum and one of the trocar wounds, and one patient had a laparotomy for postoperative hemorrhage from the inferior pancreaticoduodenal artery.

Of the four patients who underwent exploratory laparoscopy only, one was intraoperatively found to be inoperable due to advanced disease, two were explored without finding any tumor, and one had signs of pancreatitis, which did not allow for surgical removal of the tumor.

Demographics and operative variables of the 65 patients who underwent laparoscopic removal of PNET are given in Table 1. The median operative time of all patients who underwent laparoscopic pancreatic resection or enucleation was 175 (60–520) min. The median operative time was longest when performing DP (n = 23) at 190 (88–337) min. The median operative time was shortest when performing ENBT at 111 (60–244) min. Overall median blood loss was 300 (5–2,700) ml. In the subgroups, the highest median blood loss was seen when performing ENH with 500 (100–800) ml, followed by DPS with 325 (10–2,700) ml, ENBT with 300 (10–300) ml, and DP with 275 (5–1,040) ml. Overall median length of hospital stay, defined as number of days in the hospital from the day of surgery until discharge, was 7 (2–27) days. Eight patients were transferred from our department to a local hospital before discharge.

In total, six patients had multiple endocrine neoplasia type 1 (MEN-1) syndrome. The patients with MEN-1 were excluded from the presentation of histopathological data as the classification system used was not seen as appropriate for this patient group. Among the rest, 28 patients had functioning disease. Most patients with functioning disease who underwent pancreatic surgery were diagnosed with insulinoma (n = 20) followed by nonspecific functioning disease (n = 4), gastrinoma (n = 2), glucagonoma (n = 1), and vipoma (n = 1). Of the 14 enucleated tumors, 9 were insulinomas, 4 were nonfunctioning tumors, and 1 was a gastrinoma.

The histopathological data are given in Table 2. The overall median tumor diameter was 2.2 (0.5–9.5) cm. The median tumor diameter was larger in the nonfunctioning group at 2.5 (0.5–9.5) cm than in the functioning group at 2.1 (0.8–9.0) cm. Most tumors were located in the body/tail of the pancreas (n = 52). According to the WHO classification of tumors of the digestive system [25], 41 patients were graded NET G1, 17 were graded NET G2 in 17, and no patient was NEC G3. In one pathological report, the grading, defined by the Ki-67 index, was not reported. Most patients were staged T1 (n = 28), followed 20 staged T2, 9 staged T3, and 2 staged T4. Twelve patients had an N0 status, 6 patients had an N1 status, and 41 patients had an Nx status, indicating that lymph nodes were not found by the pathologist. Thirty-nine patients had an R0 status, 4 patients had an R1 status, 9 patients had an R2 status, and 7 patients had an Rx status, indicating that the resection status could not be defined by the pathologist.

Table 2 Histopathological data of 59 laparoscopically removed neuroendocrine tumor specimens
Full size table

The median follow-up of the patients who underwent laparoscopic removal of PNET (n = 65) was 51 (6–178) months. Thirty patients were followed for at least 5 years. Based on these, the 5-year disease-specific survival rate was 90 %. Survival after laparoscopic removal of PNET was significantly influenced by the T-stage and the resection status (p < 0.001 and p < 0.005) (Figs. 1, 2) Survival after laparoscopic removal of PNET was not significantly influenced by the WHO 2010 classification (p = 0.805) (Fig. 3) but was significantly influenced when using a Ki-67 cutoff value of 5 % (p < 0.05) (Fig. 4).

Fig. 1
figure 1

Five-year disease-specific survival associated with stage of primary tumor in 65 patients [median follow-up = 51 (6–178) months] who underwent laparoscopic removal of pancreatic neuroendocrine tumors. p < 0.001 (log rank test)

Full size image
Fig. 2
figure 2

Five-year disease-specific survival associated with resection status after laparoscopic removal of pancreatic neuroendocrine tumors in 65 patients [median follow-up = 51 (6–178) months]. p < 0.05 (log rank test)

Full size image
Fig. 3
figure 3

Five-year disease-specific survival associated with WHO 2010 classification (grading) after laparoscopic removal of pancreatic neuroendocrine tumors in 65 patients [median follow-up = 51 (6–178) months]. p = 0.805 (log rank test)

Full size image
Fig. 4
figure 4

Five-year disease-specific survival associated with a Ki-67 cutoff value of 5 % after laparoscopic removal of pancreatic neuroendocrine tumors in 65 patients [median follow-up = 51 (6–178) months]. p < 0.05 (log rank test)

Full size image

Discussion

To the best of our knowledge, this study is the largest reported single center series using laparoscopic surgery for PNET. It demonstrated that minimally invasive surgery of PNET is feasible with acceptable surgical morbidity and good overall disease-specific long-term prognosis. As some PNETs are large and infiltrate adjacent organs with the need of additional organ resection, experience in both pancreatic surgery and advanced laparoscopy is a prerequisite to perform these procedures [7].

In this study, the overall surgical morbidity rate after laparoscopic removal of PNET was 42 % (surgical morbidity 21 % and POPF 21 %) with no perioperative mortality (Table 1). This is comparable with a series of 45 patients who underwent laparoscopic removal of PNET in which the overall surgical morbidity rate was 31 %, including a rate of POPF of 22 % [20]. The relatively high rate of surgical morbidity, defined by the revised Accordion Classification system, was caused primarily by mild complications (grade 1 and 2) as found in Table 1. A direct comparison of surgical morbidity between the studies is of limited value due to the different complication classification systems used. Of the 14 patients who underwent enucleation, 7 developed POPF. Pancreatic fistula is the most common complication after enucleation of pancreatic lesions, ranging from 13 to 50 % [18, 27–29]. Our findings confirmed this.

Fourteen small solitary lesions, with a median diameter of 1.8 (0.8–2.8) cm and considered noninvasive and not affecting the main pancreatic duct, in our study were removed by enucleation as reported earlier [7]. Enucleation can avoid loss of parenchyma, prevent development of diabetes mellitus, and avoid more invasive surgery like resection (body/tail) or a Whipple procedure (head) [12, 20, 30–32]. However, laparoscopic enucleations of PNET were associated with a higher rate of POPF than laparoscopic resections (Table 1). One reason for this may in part be due to the difficulty in adapting the pancreatic parenchyma with suture after performing enucleations.

Seven solitary tumors in the pancreatic head with a median diameter of 1.4 (0.8–2.8) cm were enucleated laparoscopically, and in four of these a pancreatic fistula (grade A in 1 and grade B in 3) developed. Despite the relatively high rate of POPF after enucleation, our recommendation is that focal tumors of the pancreatic head that are considered noninvasive should be enucleated laparoscopically. This spares the patient from more invasive surgery (Whipple’s procedure). We do not follow a wait-and-see approach at our department when deciding on the indication for surgical removal of small suspicious PNET because to date there is no consensus on the diameter cutoff and no clear evidence of better outcome when waiting to perform pancreatic surgery in this patient group [33]. Another argument is that the size of the primary tumor correlates with the risk of malignancy, indicating that early resection may prevent a higher risk of malignancy [34]. At our institution resection/enucleation of a nonfunctioning primary tumor in metastatic disease is performed in selected patients, as suggested by others [34–36].

When performing an ENH, it is especially important to prevent and detect damage to the main pancreatic duct intraoperatively. In this study a plastic stent was placed in the pancreatic duct preoperatively by means of ERCP in three patients who were planned for ENH. This resulted in post-ERCP pancreatitis in two patients. Two of the patients with preoperative stent inlay developed a pancreatic fistula postoperatively. Based on these findings we do not recommend the use of preoperative stent inlay when planning an ENH.

The histopathological findings of the laparoscopically removed PNET in our study revealed a relatively high fraction of surgical specimen with an Nx status, indicating that lymph nodes were not found by the pathologist. Theoretically, this can be explained by either the surgical techniques, which do not include peripancreatic tissue with lymph nodes, or the pathologic assessment, which fails to detect lymph nodes [37]. It is more likely that there is a lack of lymph nodes in the surgical specimen after ENH and DP than after DPS as the en-bloc resection with splenectomy includes peripancreatic tissue and lymph node stations in the splenic hilus [17]. This was confirmed by our findings, which showed an Nx status in all 14 enucleation specimens (ENH and ENBT), in 21 of 23 DP specimens, and in 10 of 28 DPS specimen. Pomianowska et al. [38] showed that the lymph node status of surgically removed PNET is a prognostic factor. This may be a limitation of laparoscopic procedures and should be further evaluated in future prospective trials.

The disease-specific 5-year survival of the patients who underwent laparoscopic removal of PNET in this study was 90 %. Phan et al. [39] reported an overall 5-year survival of 65 % in a series of 125 patients who underwent open surgical treatment for PNET. Our findings show a better overall 5-year survival. However, there are limitations of such a comparison between two retrospective studies.

In this study we showed that the stage of the primary tumor and resection margin status after LS for PNET predict 5-year survival. This is consistent with comparable studies [25, 33, 34, 38]. Interestingly, the WHO 2010 classification for neuroendocrine neoplasms of the gastroenteropancreatic system with a Ki-67 index of 2 % as a cutoff value between NET G1 and NET G2 did not predict 5-year survival. Boninsegna et al. [40] had a similar observation in 57 patients with PNET. This may indicate the need for a different cutoff value between NET G1 and NET G2 in PNET.

In conclusion, this study shows that laparoscopic surgery of PNET is feasible with acceptable morbidity and a good overall disease-specific long-term prognosis. Laparoscopic enucleation is associated with a higher rate of POPF after LS than after laparoscopic resection. This study indicates that laparoscopic enucleation of focal PNET in the pancreatic head may be favored over alternative surgery (Whipple’s procedure), whereas laparoscopic resection of PNET in the pancreatic body and/or tail may be favored over enucleation. The study also demonstrates that there may be a need for a different cutoff value between NET G1 and NET G2 in PNET.