Laparoscopic versus open approach for solitary insulinoma
- First Online:
- Cite this article as:
- Cunha, A.S., Beau, C., Rault, A. et al. Surg Endosc (2007) 21: 103. doi:10.1007/s00464-006-0021-8
- 126 Views
In recent years, advances in laparoscopic techniques have allowed surgeons to treat pancreatic lesions laparoscopically. Insulinoma, the most prevalent pancreatic endocrine tumor, is mostly benign and curable with surgical resection. This study aimed to assess the results from laparoscopic resection (LG) of insulinomas and to compare them with the results from open surgery (OG).
From September 1999 to December 2005, 56 laparoscopic pancreatic resections were performed for selected patients, including 12 laparoscopic resections of insulinomas. The results were compared with those of patients who underwent open resection of insulinomas selected from the authors’ pancreatic database.
Three conversions to the open approach were required because of inability to identify the tumor. There were no deaths in either group, and the morbidity rates were 25% (3/12) for LG and 55% (5/9) for OG (nonsignificant difference). The pancreatic fistula rate after laparoscopic enucleation was statistically lower than after open enucleation (14% vs 100%; p = 0.015). The mean postoperative hospital stay was 13 ± 5.9 days for LG and 17.6 ± 7.5 days for OG (nonsignificant difference). After exclusion of the patients who underwent conversion to laparotomy, the mean postoperative hospital stay was 11.5 ± 5.8 days for LG and 17.6 ± 7.5 days for OG (p = 0.04).
This study demonstrates the feasibility and safety of laparoscopic resection of insulinomas. The laparoscopic approach was associated with a decrease in hospital stay and pancreatic fistula after enucleation. Preoperative localization tests and laparoscopic ultrasonography seem necessary to prevent conversion.
KeywordsLaparoscopic enucleationLaparoscopic pancreatic resectionSolitary insulinoma
The laparoscopic surgical approach has been widely developed to become the standard technique for several procedures [1–3]. In recent years, advances in laparoscopic techniques have allowed surgeons to approach the pancreas and treat pancreatic lesions laparoscopically [4–10]. However, laparoscopic pancreatic surgery still is uncommon due to the anatomic location of the pancreas, technical difficulties of pancreatic resection, the relative rarity of surgical pancreatic disorders, and the requirement of highly experienced laparoscopic and pancreatic surgeons.
Insulinomas represent up to between 70% and 80% of clinically symptomatic islet cell tumors, with an incidence of four insulinomas per million person-years. They occur in all age groups, with a peak incidence during the third to fifth decades of life. Surgical resection of insulinomas is the most effective approach for avoiding symptoms of hypoglycemia [11, 12].
Insulinomas usually are benign, solitary, and intrapancreatic. Because of the characteristic clinical presentation of hypoglycemia, they usually are diagnosed when they still are small (<20 mm) and resectable . Otherwise, two-thirds of these lesions are located in the body or tail of the pancreas. These favorable features make insulinomas suitable for the laparoscopic approach. However, only a few authors have reported laparoscopic resection of pancreatic insulinomas [13–19].
This study aimed to assess the results from laparoscopic treatment of insulinomas and to compare them with the results from open treatment of insulinomas.
In 1999, we started a prospective evaluation of laparoscopic pancreatic resection. By December 2005, the evaluation included 56 patients, 12 of whom had undergone laparoscopic resection of insulinoma and constituted the laparoscopic group (LG). All the patients presented with neuroglycopenic symptoms that developed after fasting or exertion and improved after glucose intake. The diagnosis was confirmed with a supervised 72-h fasting test. Furthermore, circulating concentrations of C-peptide were assessed and elevated in all the patients. Preoperative localization was assessed by computed tomography (CT) scan, magnetic resonance imaging (MRI), endoscopic ultrasonography (EUS), and octreotide scanning scintigraphy.
The dissection is performed using a 5-mm bipolar electrocoagulation instrument, with cautery between the normal parenchyma and the tumor itself. The vessels of the tumor are secured with clips (Fig. 2).
Spleen-preserving distal pancreatectomy without preservation of splenic vessels (Warshaw technique )
Initially, the origin of splenic artery is identified at the superior edge of the pancreas, then clipped and divided. Next, the inferior edge of the pancreas is dissected. A tunnel is created between the portal or splenic vein and the pancreas. The pancreas then is transected with a 30-mm linear stapler. The dissection is continued laterally to the splenic hilum. The splenic artery and vein are clipped and divided as they emerge from the pancreatic tail to enter the splenic hilum. The spleen is kept vascularized from the short gastric vessels and the left gastroepiploic vessels (Fig. 2). The left pancreas then is lifted up and mobilized posteriorly with the splenic artery and vein. The last step is ligation and section of the splenic vein on the left side of the portal vein.
With either technique, the specimen is extracted in an endoscopic plastic bag via an enlarged trocar site. A silicon drain is left in the pancreatic bed left of the pancreatic stump. This drain is removed on day 10. Surgical drainage output is recorded daily. The serum amylase level and the amylase level of drainage fluid are monitored on postoperative days 5 and 10.
From 1995 to 1999, nine patients underwent surgery through a laparotomy according the previously described technique, except for surgical access, and constitute the control group (OG).
Patient characteristics, tumor size, tumor localization, operative data, histologic examination, specific morbidity, hospital stay, and outcome were analyzed. Pancreatic fistula was defined by at least one of the following criteria: amylase level in the surgical drainage fluid higher than five times the serum amylase level after postoperative day 5 and fluid collection shown on the CT scan.
The results are expressed as mean ± 1 standard deviation. Comparisons between groups were analysed using the chi-square test with Yates correction, and the Wilcoxon test for qualitative or quantitative variables as appropriate. Significance was defined as a p value less than 0.05.
Clinical and tumor Features
Laparoscopic Group (n = 12)
Open Group (n = 9)
Gender Ratio, F/M
Age Mean ± SD (y)
47.5 ± 18
41.4 ± 14.3
Tumor size (mm)
13.25 ± 5.5
20.7 ± 9.8
Preoperative localization tests
In LG, 11 patients (91%) underwent computed tomography, and 9 patients (75%) underwent endoscopic ultrasonography. Eight patients (66.6%) underwent MRI, and four patients (33.3%) underwent octreotide scanning scintigraphy.
In OG, all the patients underwent MRI. Four patients (44%) underwent EUS; three patients (33%) underwent CT scan; and two patients (22%) underwent octreotide scanning scintigraphy. The mean number of preoperative localization tests per patient was 2.6 in LG and 2 in OG. The sensitivity was 78% for CT scan, 94% for MRI, 77% for EUS, and 33% for octreotide scanning scintigraphy.
In LG, the operation had to be converted in thee cases (25%). In these three cases, the reason for conversion was inability to locate the tumor during the laparoscopic procedure. In two cases, laparoscopic ultrasound (LUS) was not available and thus was not performed. In one case, the tumor was not localized preoperatively and could not be localized intraoperatively despite LUS examination.
Laparoscopic Group (n = 12)
Open Group (n = 9)
Surgical time (min)
172 ± 14.3
148.8 ± 114.5
Peroperative blood loss (ml)
100 ± 56
115 ± 67
There was no death in either group. One LG patient underwent reoperation on postoperative day 2 for bleeding. A postoperative complication was observed in three LG patients (morbidity 25%) and five OG patients (55%) (nonsignificant difference). Specific morbidity developed in three LG patients (25%): intraabdominal collection (n = 1) and pancreaticocutaneous fistula (n = 2). Pancreaticocutaneous fistula developed in five OG patients. This difference is not statistically significant. None of the patients with specific morbidity was either radiologically drained or surgically retreated. The pancreatic fistula rate after laparoscopic enucleation was statistically lower than after open enucleation (14% vs 100%; p = 0.015).
The mean postoperative hospital stay was 13 ± 5.9 days for LG and 17.6 ± 7.5 days for OG (nonsignificant difference). After exclusion of the patients who had conversion to laparotomy, the mean postoperative hospital stay was 11.5 ± 5.8 days for LG and 17.6 ± 7.5 days for OG (p = 0.04).
The mean follow-up time was 49 ± 29 months for LG and 91.75 ± 19.8 months for OG. No patient was lost for follow-up evaluation. At this writing, 20 patients (95%) are free of symptoms and 1 LG patient still reports hypoglycemia. In this case, the insulinoma could not be localized during the laparoscopic exploration, so an open distal pancreatectomy finally was performed according to preoperative localization tests. The pathologic findings showed nesidioblastosis without any tumor.
Advancing technology and increasing experience with laparoscopic surgery have led surgeons to perform laparoscopic pancreatic resections [4–10]. Benign insulinomas generally are solitary, small, and located in the body and tail of the pancreas. These favorable features make insulinomas amenable to the laparoscopic approach [13–19]. To date, however, the laparoscopic approach for solitary insulinoma has not been compared with the open approach. Our study is the first to focus on this subject. Our findings showed that laparoscopic treatment of insulinomas was at least as safe as open treatment, with a shorter hospital stay.
Increased surgical time has been reported for other complex laparoscopic approaches [21–24]. This difference could be explained by the learning curve and the laparoscopic procedure itself. In our series, the mean surgical time was not statistically different between LG and OG. It was similar to the surgical time reported for other series [13, 14]. This result could be explained by our experience with pancreatic and laparoscopic surgery. We have performed more than 50 laparoscopic pancreatic resections.
In the literature, the estimated blood loss of patients undergoing laparoscopic resection of insulinomas ranges from less than 100 ml to 360 ml [13, 19]. In our study, the mean blood loss after laparoscopic resection of insulinomas was 100 ml. The intraoperative blood loss was not statistically different between LG and OG.
In our study, the morbidity rate was 25% for LG and 55% for OG. However, this difference is not statistically significant. Our morbidity rate compares favorably with the complication rates reported after insulinoma resection by either laparotomy or laparoscopy [13, 14, 16, 17, 19, 25–27].
The pancreatic fistula continues to represent the main major morbidity after pancreatic resection, whether laparoscopic or open. In the literature, enucleations are considered to be associated with a risk for pancreatic fistula . In our study, the incidence of pancreatic fistula after laparoscopic enucleation was statistically lower than after open enucleation. This improvement in pancreatic duct control could be explained by the slower and more meticulous pancreatic enucleation.
In our laparoscopic left pancreatectomies, the pancreas was transected with a stapler. Some authors argue that this could be a cause of pancreatic fistula. A metaanalysis evaluating the technique of pancreatic stump closure after distal pancreatectomy  cannot determine the optimal surgical technique for pancreatic stump closure. Nevertheless, there is a trend in favor of the stapling technique. In our series, the rate of pancreatic fistulas was not statistically different between LG and OG after pancreatic resection.
The use of somatostatin analogues to prevent fistula formation or to accelerate closure has been studied in 10 randomized clinical trials. Despite this, there has not been agreement on the role of these drugs. Only five studies showed a positive effect. Two studies showed a selective effect, and three showed no benefit [29–38]. However a recent metaanalysis has shown that somatostatin analogues reduced morbidity and pancreas-specific complications . In our experience, somatostin analogues have never been used preoperatively. Nevertheless, the course of the fistulas has allowed a conservative treatment in every case. The hospital stay was not notably different between the two groups. However, after exclusion of the patients with conversion to laparotomy, the mean postoperative hospital stay was statistically shorter for LG than for OG. In the study of Ayav et al. , the mean hospital stay after laparoscopy was similar to the mean hospital stay after laparotomy (11 days).
During open surgery, it has been shown that manual palpation and intraoperative ultrasonography allows clinicians to localize nearly 100% of the insulinomas . According to these results, preoperative localization of insulinomas has been regarded as unnecessary [26, 27, 41].
In contrast to open surgery, during laparoscopic surgery, manual palpation lacks and laparoscopic ultrasonography is not always available. However, preoperative localization is essential for patient positioning. For these reasons, in all series of laparoscopic insulinoma treatment, preoperative localization tests are used. Preoperative localization is the key to allowing a laparoscopic approach for insulinomas.
There is continuing debate in the literature regarding the ideal imaging method for preoperative detection of insulinomas. Whereas invasive techniques such as percutaneous transhepatic portal venous sampling, angiography, and arterial simulation venous sampling are progressively being abandoned, gadolinium-enhanced dynamic MRI and endoscopic ultrasound have emerged as potentially competing or complementary techniques to CT scan.
A study by Gouya et al.  found that biphasic CT scan with thin sections had a sensitivity of 94.4%. The sensitivity of biphasic CT scan without thin sections was 57%, and the sensitivity of sequential CT scan was only 28.6%. The sensitivity of endoscopic ultrasonography was 93.8%, with no statistically significant difference between dual-phase thin-section CT and endoscopic sonography. These authors concluded that the overall diagnostic sensitivity for combined biphasic thin-section CT scan and endoscopic sonography was 100%, but that additional studies were required to determine the value of MRI for preoperative evaluation of insulinomas .
Recent MRI studies have shown excellent results, with 85% of lesions detected on 1.5-tesla systems . Furthermore, MRI is particularly accurate in delineating the anatomic relationship of the tumor with the pancreatic duct, and can aid in the surgical decision making as to the appropriate surgical procedure (enucleation vs distal pancreatectomy).
Despite this preoperative localization, conversion rates in the literature vary from 20% to 33%, and the reasons for conversion are either inability to localize the lesion or technical difficulties in performing the pancreatic resection [6, 13, 14, 16]. In our series, despite a preoperative localization rate of 91%, the conversion rate was 25%, and the reason for conversion, in all cases, was inability to localize the lesion.
During open surgery, the most effective method for localizing an insulinoma is intraoperative ultrasound . In the literature, LUS seems to provide information similar to that obtained by open intraoperative ultrasound [16–18]. In our series, LUS was performed for six patients (50%). In two of them (33%), LUS failed to localize the insulinoma. In another case, LUS showed a 4-mm tumor in the body, but the pathologic findings showed nesidioblastosis without any tumor. For the three patients who underwent conversion to laparotomy, LUS performed in two cases failed to localize the insulinoma. These results suggest that LUS is less accurate for localizing insulinoma than reported in other series, and preoperative localization of insulinoma was required before the laparoscopic approach. In our experience, LUS was not required for patients in whom preoperative localization tests showed insulinoma close to pancreatic duct, and for whom a pancreatic resection was planned. In the same way, prominent insulinomas were easily identified during laparoscopic exploration, making the use of LUS unnecessary. In contrast, LUS was required for invisible intrapancreatic insulinomas in patients for whom an enucleation was planned according to preoperative localization tests. In these cases, LUS was the only way to localize the tumor during laparoscopy and to confirm the possibility of tumor enucleation.
In conclusion, as indicated by the results of this series, laparoscopic treatment of insulinomas was at least as safe as open treatment. The laparoscopic approach was associated with a decrease in hospital length of stay and pancreatic fistula after enucleation. Consequently, we have abandoned the open approach for resection of insulinoma, especially as conversion to the open approach did not worsen the postoperative course. Preoperative localization is required for decision making concerning the most appropriate surgical procedure. This perioperative workup probably is the key to reducing the conversion rate.