Introduction

Pancreatic neuroendocrine tumors (PNET) are rare neoplasms with an incidence of 2–3 per million, based on data from USA and Norway.1,2 As most PNET are slow-growing, their prognosis is better than most pancreatic malignancies.1,3 Surgery is the only possibility for cure and should always be considered.4 However, surgical treatment of locally advanced and metastatic disease remains controversial.5 There are only a few reports that discuss vascular reconstruction among patients with locally advanced PNET, and none of these discuss the role of vascular reconstruction as such.5–17 The concept of vascular reconstruction has already been discussed widely in the treatment of locally advanced pancreatic adenocarcinoma.18–21 In this group of patients, combined portal vein resection and reconstruction provides acceptable morbidity, mortality and better survival as compared to unresected patients.22 Hence, discussion on the role of vascular reconstruction in locally advanced PNET also seems to be warranted.

The aim of this study was to evaluate the feasibility and outcome of pancreatic surgery with vascular reconstruction in patients with locally advanced PNET, based on experience from a single institution and a review of the literature.

Material and Methods

Single Institution Clinical Experience

All patients who underwent pancreatic surgery with vascular reconstruction for locally advanced PNET at Oslo University Hospital from April 2007 to July 2012 were identified and retrospectively included in the study.

Data were extracted from patient records from our hospital and records from local hospitals if the patient was transferred to such hospital after discharge from our institution. Patient and tumor parameters included demographics, tumor location, tumor diameter, biologic behavior, visceral and vascular tumor infiltration and metastasis. Perioperative variables included operative time, intraoperative blood loss, length of hospital stay, surgical morbidity and postoperative pancreatic fistula (POPF). Surgical morbidity was defined by the revised Accordion Classification.23 Surgical procedures as well as pre- and postoperative non-surgical treatment were listed for each patient. Follow-up was measured in months from the date of surgery. Short-term-outcome was defined by the surgical morbidity, POPF and length of hospital stay. Long-term-outcome was defined by survival or disease stage at last follow-up.

Standard preoperative diagnostic imaging consisted of three phase contrast-enhanced helical computed tomography (CT). Each patient was discussed in a multidisciplinary tumor board where the indication for surgical treatment was decided. To be considered for surgery a Ki-67 below 20 % was mandatory in the pancreatic tumor and/or the liver metastases. In patients where liver metastases were present, a treatment plan for the metastases was decided on prior to operation of the primary tumor. The surgical procedures were planned before operation in close cooperation with an abdominal transplant surgeon with high expertise in vascular reconstructive surgery. However, the final approach was decided intraoperatively based on macroscopic findings during laparotomy.

While performing vascular resection in these patients, the patients received a bolus of 2,500–5,000 IU (depending on weight, renal function and gender) unfractioned heparin intravenously directly before vascular occlusion. Dalteparin (5000 IU) were routinely administered subcutaneously 6 h postoperatively and administered twice daily the first 14 days postoperatively, followed by one daily administration for another 14 days. Moreover, 75 mg acetylsalicylic acid was administered once daily from day 7 postoperatively in patients with arterial reconstruction, and continued life-long.

Patients undergoing arterial reconstruction of hepatic arteries, the superior mesenteric artery or reconstruction of the portal vein were monitored with lactate-measurement every 2–3 h the first 24 h postoperatively. A rise in lactate above 2.5 mmol/L resulted in acute abdominal Doppler examination of the relevant vessel. All patients were routinely controlled with abdominal Doppler of the relevant vessel on the first postoperative day.

Histopathological data were acquired according to the World Health Organization (WHO) 2010-classification for neuroendocrine neoplasms of the gastroenteropancreatic system24 and the European Neuroendocrine Tumor Society (ENETS) TNM-classification.25 In addition, lymph node status and resection status were recorded for each patient. The study was approved by the Data Protection Officer for Research at our institution.

Literature Review

A computerized literature search of the Pubmed database was conducted for the period January 1990 to December 2012 using the keywords “pancreas”, “endocrine”, “neuroendocrine”, “vascular resection”, “arterial resection”, “venous resection”, “pancreatic resection”, and “pancreatectomy” to identify original studies and case reports concerning vascular resection of PNET.

The titles and abstracts of the search hits were assessed and the relevant articles were reviewed. Only studies that specifically reported perioperative and/or long-term results after pancreatectomy with vascular reconstruction in PNET were eligible for inclusion. In order to identify additional studies, the reference lists of all the identified articles were manually searched. The search was restricted to studies published in English. The last search was done on December 15, 2012.

The primary evaluated outcome of the literature review was the feasibility and outcome of pancreatic surgery with vascular reconstruction in patients with locally advanced PNET.

Results

Single Institution Clinical Experience

Seven patients who underwent surgery with vascular reconstruction for locally advanced PNET from April 2007 to July 2012, were included in the study. During the study period, no patients with PNET and preoperatively suspected vascular involvement were found non-resectable intraoperatively. In the same period, a total of 75 pancreatic resections were performed for any PNET. In addition, 57 patients underwent pancreatic resection with vascular reconstruction for other histological variants of pancreatic or periampullary malignancy.

All patients included in this study were male with an average age of 59 (46–72) years and an average body mass index of 25.5 (21.9–32.9) kg/m2 at time of surgery. The primary tumor was located in the pancreatic head in two patients, in the pancreatic body in two patients, in the pancreatic tail in two patients and in the whole length of the pancreas in one patient. Three patients had functioning disease (Patients 2, 5, and 7). The median tumor diameter was 9.0 (3.0–12.0) cm with visceral infiltration in three of the patients (Patients 3, 4, and 7). All seven tumors infiltrated the portal vein (PV), two infiltrated the common hepatic artery (CHA), two infiltrated the celiac trunk (CET), one infiltrated the splenic artery (SA), one infiltrated the left hepatic artery (LHA), one infiltrated the left gastric artery (LGA), one infiltrated the middle colic artery (MCA), one infiltrated the middle colic vein (MCV) and one infiltrated the pancreaticoduodenal artery (PDA). Resection of vessels was done after careful examination of the preoperative CT scans and by assessing the anatomic situation in the operative field. If preoperative scans were indicative of vascular infiltration, a planned resection was performed. Tumor infiltration of the vessels was suspected in all seven patients preoperatively. This was confirmed in the final histopathology reports. Four patients had liver metastases at time of surgery (Patient 2, 4, 5 and 7). Tumor characteristics are shown in Table 1.

Table 1 Patient and tumor characteristics in seven patients with locally advanced pancreatic neuroendocrine tumors
Full size table

Patient 1 had a tumor in the pancreatic body, which infiltrated the PV and was associated with a thrombosis in the splenic vein. He underwent distal pancreatectomy (DP) with splenectomy and vascular reconstruction of the PV. The latter was performed as an end-to-end-anastomosis of the superior mesenteric vein (SMV) to the PV. Patient 2 had a tumor in the pancreatic head, which infiltrated the PV. The patient underwent a pylorus-preserving Whipple procedure with vascular reconstruction of the PV, using an interponate of a venous autograft from the left renal vein. Patient 3 had a large tumor infiltrating all parts of the pancreas, parts of the stomach and the following vessels: PV, CHA, CET, MCV, MCA, and SA. The patient underwent a total pancreato-duodenectomy (TP) with splenectomy, distal gastrectomy and vascular reconstruction of the PV and CHA. In this patient, the right hepatic artery (RHA) came from the superior mesenteric artery (SMA), which gave off the PDA. The PDA was set off due to malignant infiltration. For the same reason the CET and proximal CHA were resected, leaving the distal CHA, including the gastroduodenal artery, left. The PDA and the remaining end of the CHA were then anastomosized end-to-end in order to re-establish perfusion. The PV was reconstructed with an interponate of a venous allograft from an external iliac vein. Patient 4 had a large tumor in the pancreatic tail, which infiltrated parts of the stomach, spleen, left colic flexure, diaphragm, left kidney and left adrenal gland and the following vessels: PV, LHA, and LGA. The patient underwent a DP with splenectomy, distal gastrectomy and resection of the left colic flexure, diaphragm, left kidney, left adrenal gland and simultaneous wedge resection of a solitary liver metastasis with a diameter of 30 mm in liver segment IV. He also underwent vascular reconstruction of the PV, LHA, and LGA. In this patient, the RHA came from the SMA. Due to malignant infiltration, the CET was resected. In order to re-establish perfusion of the left liver, an arteriotomy in the RHA was made before an end-to-side anastomosis with an interponate of a venous allograft from the great saphenous vein (GSV) was performed. The LHA was resected due to malignant infiltration and was anastomosized end-to-end with the venous allograft from the RHA. The LGA was also resected due to malignant infiltration. Another interponate of a venous allograft from the GSV was then anastomosized end-to-end with the LGA and then end-to-side with the LHA in order to re-establish perfusion to the stomach. The PV was finally reconstructed with an interponate of a venous allograft from an external iliac vein. Patient 5 had a tumor in the pancreatic body, which infiltrated the PV and CHA. This patient underwent a TP with splenectomy and vascular reconstruction of the PV and CHA. The reconstruction of the PV was performed as an end-to-end-anastomosis of the SMV to the PV. The CHA was also reconstructed with an end-to-end-anastomosis. Patient 6 had a tumor in the pancreatic head, which infiltrated the portal vein. He underwent a pylorus-preserving Whipple procedure with vascular reconstruction of the PV. The latter was performed as an end-to-end-anastomosis of the SMV to the PV. Patient 7 had a tumor in the pancreatic tail, which infiltrated the spleen and the PV. He underwent a DP with splenectomy and vascular reconstruction of the PV with a venous allograft between the SMV and PV.

The median operative time of the seven patients was 427 (range, 232–718) min, the median intraoperative bleeding 1,375 (range, 500–4,750) ml and the median length of hospital stay 25 (range, 8–64) days. Four patients had complications following surgery (Patients 1, 3, 4, and 6). Patient 1 developed a pneumonia that was treated successfully with antibiotics. Patient 3 developed an upper gastrointestinal bleeding secondary to ischemic necrosis of parts of the stomach. He underwent gastroscopy without any sign of active bleeding and was only treated with blood transfusions. The bleeding was probably due to prior resection of the LGA. Patient 4 underwent simultaneous liver resection and developed a liver abscess in the resection area, which was successfully treated with percutaneous drainage and antibiotics. Postoperative CT revealed a peripancreatic pseudoaneurysm of the PDA without bleeding. This was embolized in percutaneous transluminal technique. Patient 6 developed a subcutaneous infection in the laparotomy wound, which was treated successfully by conservative means. There was no pancreatic fistula formation in any of the patients. The surgical procedures, perioperative variables and short-term-outcome are shown in Table 2.

Table 2 Surgical procedures, perioperative variables and short-term-outcome in seven patients with locally advanced pancreatic neuroendocrine tumors
Full size table

Histopathology revealed a positive lymph node status in four patients and a median peripancreatic lymph node count of 17 (range, 1–37) in the surgical specimen. One patient was graded NET G1 and six patients were graded NET G2. Three patients had received systemic chemotherapy prior to surgery and three patients received palliative non-surgical treatment after the operation.

The four patients with synchronous liver metastases were managed as follows: Patient 2 had multiple bilobar liver metastases and was initially planned for liver transplantation after removal of the primary pancreatic tumor. However, he developed progressive hepatic disease postoperatively and did therefore not undergo transplantation. Patient 4 had a solitary metastasis in liver segment IV and underwent a simultaneous wedge resection during pancreatic surgery. Patient 5 had one metastasis in each of the liver segments III, V/VI, VI, and VII. He was planned for liver resection in a subsequent procedure after undergoing pancreatic surgery. Due to progressive hepatic disease first diagnosed at laparotomy four months after the initial pancreatic surgery, he did not undergo liver resection. Patient 7 had multiple bilobar non-resectable liver metastases, but had long-term stable disease before decision for pancreatic surgery was made.

The overall median follow-up was 21 (range, 3–58) months with one death (Patient 2), three patients with progressive hepatic and/or intraperitoneal disease (Patients 4, 5, and 7) and three patients with disease in remission (Patients 1, 3, and 6). Histopathology, non-surgical treatment and long-term outcome are shown in Table 3. Disease-specific survival is shown in Fig. 1.

Table 3 Tumor stage, non-surgical treatment and long-term outcome in seven patients operated for locally advanced pancreatic neuroendocrine tumors
Full size table
Fig. 1
figure 1

Kaplan–Meier curve for disease-specific survival in seven patients operated for locally advanced pancreatic neuroendocrine tumors

Full size image

Literature Review

The literature search identified 13 publications, which reported perioperative and/or long-term results after pancreatic resections with vascular reconstruction in PNET (Table 4). These studies included altogether 30 patients who underwent pancreatic surgery with vascular reconstruction for locally advanced PNET. The PV/SMV was reconstructed in 28 patients, the SMA in one patient, and the PV in combination with the CET in one patient.

Table 4 Previously reported vascular reconstruction in surgical treatment of locally advanced pancreatic neuroendocrine tumors
Full size table

Discussion

The aim of this study was to evaluate the feasibility and outcome of pancreatic surgery with vascular reconstruction in patients with locally advanced PNET, based on surgical experience from a single institution and a review of the literature. Surgery is the only chance for cure in patients with PNET, and resection rates up to 65 % have been reported.26 To the best of our knowledge, there are no systematic studies discussing the role of vascular reconstruction as such in patients with locally advanced PNET. However, in some institutions, involvement of the SMV, PV, splenic vein or the hepatic arteries is considered a contraindication to surgery in this patient group.

This study shows that pancreatic surgery with vascular reconstruction in a series of seven patients with locally advanced PNET is feasible with acceptable short- and long-term outcomes. All seven patients in our single institution series underwent pancreatic surgery with vascular reconstruction and were discharged home after a median hospital stay of 25 days. As surgery required both extensive visceral and vascular dissection, the amount of intraoperative bleeding was relatively high, and operative time was relatively long. In a recent retrospective study, Norton et al. presented ten patients who underwent pancreatic resection with vascular resection and reconstruction for locally advanced PNET and could demonstrate that aggressive surgery including SMV reconstruction, and liver resection can be done with acceptable morbidity and mortality rates for these patients.11 This is also supported by our study, where four out of seven patients developed surgical complications, but no surgical mortality occurred. Case reports/series represent an observational study that reports on data from a subject group without a comparison population, and represents level IV evidence. Such studies can be prone to bias, particularly selection bias, which limits its generalizability to larger patient populations. However, case reports and case series are often the first data alerting to new treatment opportunity. Based on our experience and the additional 13 publications in the literature, it is highly relevant to address the issue of vascular resection in locally advanced PNET.

In our study, four out of seven patients had liver metastasis at time of surgery. All of these patients developed progressive disease, independently of whether liver resection had been performed or not. The remaining three patients without liver metastasis did not develop recurrent disease throughout the study period, implicating a better long-term survival of patients without synchronous liver metastasis.

The goal for surgical resection in patients with PNET is cure, relief from hormonally active tumors or relief from tumors causing symptoms related to mass effect (biliary obstruction, gastric outlet obstruction, abdominal pain, gastrointestinal hemorrhage). Some patients will experience serious complications related to the unresected primary tumor. Thus, radical resection of the primary may significantly delay or prevent the occurrence of such complications. In a single center retrospective study, Solorzano et al. could show that PNET in the pancreatic head had a tendency of causing excessive local morbidity due to hemorrhage and biliary obstruction.27 Among 20 patients with unresectable PNET, 12 patients died of the disease after a median time of 2.7 years from diagnosis. Ten of these 12 patients had tumors in the head of the pancreas. Before death, these 10 patients had a median of two hospital admissions each for complications such as biliary obstruction, gastric outlet obstruction, gastrointestinal hemorrhage, or treatment-related toxicity. Furthermore, among 81 patients with metastatic unresectable PNET, 7 patients experienced complications related to the primary tumor that required hospitalization. Thus, the authors concluded that symptomatic primary tumors should be resected when technically possible. Accordingly, the current ENETS guidelines conclude that surgery could alleviate mass-reducing symptoms by reducing tumor burden in selected cases of locally advanced PNET.4

The rarity of PNET enables few institutions to collect large patient series. The present case series is among the largest that addresses the issue of pancreatic resection with vascular resection and reconstruction for locally advanced PNET. The literature search identified 13 publications with a total of 30 patients. Although 8 of the 13 publications are single case reports, these show that aggressive surgery can be safe, feasible and associated with long-term-survival. Among the 30 patients reported, 28 underwent reconstruction of the PV/SMV, one underwent reconstruction of the CET and PV, whereas one underwent reconstruction of the SMA, as shown in Table 4. The favorable results of the present case series and the current literature on this topic support an aggressive approach in patients with locally advanced PNET.

In conclusion, this study shows that radical pancreatic surgery with vascular reconstruction in patients with locally advanced PNET is safe and feasible with acceptable outcome. Benefits seem to be more evident in patients without synchronous liver metastasis. Patients with locally advanced or metastatic PNET should be referred to a tertiary high volume centre where multidisciplinary expertise is available.