Laparoscopic Enucleation of Insulinomas FREE

Insulinoma is the most common pancreatic endocrine neoplasm, occurring in 1 per 1 million persons each year. They are typically well encapsulated and less than 2 cm in diameter. Ninety percent are solitary and benign, and the remaining 10% are malignant. Multifocal insulinomas are more common in patients with multiple endocrine neoplasia type 1. The diagnosis of insulinoma is established by measuring insulin and C-peptide levels during a 72-hour fast and by calculating the ratio of insulin-to-glucose in the presence of hypoglycemia. The presence of sulfonylureas must be excluded by blood testing. Treatment is by excision. In the prelaparoscopic era, enucleation was performed via a laparotomy. Since then, surgeons have described their experience with laparoscopic enucleation and distal pancreatectomy.^1- 5 When enucleation is technically feasible, pancreatectomy provides no oncologic advantage, and it often requires an otherwise unnecessary splenectomy. However, few cases of successful laparoscopic enucleation have been reported,^1- 5 and its efficacy has not been established, to our knowledge.

We attempted laparoscopic enucleations in 9 consecutive patients with insulinomas localized before surgery at the University of California, San Francisco Medical Center. The clinical information was analyzed retrospectively. Paper and electronic medical records and pathology reports were reviewed.

There were 3 men and 6 women (mean age, 39 years [age range, 18-56 years]). The median tumor size was 1.6 cm (range, 1.1-5.0 cm). One tumor was located in the neck of the gland, 5 were located in the body, and 3 were located in the tail.

All patients who were discharged on the first or second postoperative day went home with a closed-suction surgical drain in place. Patients were seen in the clinic on the seventh postoperative day for their first follow-up visit, and the drains were removed if the output was less than 30 mL/d. For this study, pancreatic fistula was defined as persistent drainage of more than 30 mL/d after 7 days. Clinically significant fistulas were those that required percutaneous intervention. The study protocol was reviewed and approved by the University of California, San Francisco Committee on Human Research.

Fourteen patients with hyperinsulinism were operated on at the University of California, San Francisco Medical Center from July 1, 2000, to June 30, 2005. Of these patients, 9 underwent attempted laparoscopic enucleation of an insulinoma. The remaining 5 patients had successful open enucleations. Three of these 5 had lesions in the head of the pancreas that were not thought to be amenable to laparoscopic enucleation. The other 2 patients had lesions in the tail of the pancreas, but a laparoscopic approach was not attempted because both patients had undergone a previous pancreatic operation. Therefore, laparoscopic enucleation was confined to patients with solitary lesions of the neck, body, or tail of the pancreas who had not had a previous pancreatic operation. Laparoscopic enucleation was successful in 7 of 9 patients in whom it was attempted. The abdomen was opened in 2 patients to perform a distal pancreatectomy. All operations were curative.

Pancreatic protocol computed tomography (CT) was performed in all patients. Five of 9 showed a discrete mass lesion within the pancreas. Among 4 patients with nondiagnostic CT scans, endoscopic ultrasonography (EUS) was used in 3, and calcium-stimulated venous sampling was used in 1. Overall, 4 patients underwent EUS. It correctly pinpointed the lesion in 3 patients and was a false-positive finding in 1 patient who was found to have islet cell hyperplasia, not insulinoma (Figure).

Laparoscopic ultrasonography was used in 6 of 9 patients. Its principal value was to confirm the CT findings and to show the relationship between the lesion and the pancreatic duct and blood vessels. It was omitted in 1 patient who ultimately required conversion to a laparotomy for a lesion deep in the pancreatic body, as well as in 2 patients with pedunculated tumors of the body and tail of the pancreas.

The hospital course of 7 patients who underwent laparoscopic enucleation was as follows: 5 were discharged on the first postoperative day, 1 on the second postoperative day, and 1 on the ninth postoperative day. The 2 patients whose operations were converted to a laparotomy stayed 9 and 15 days.

The most frequent complication was pancreatic fistula, which occurred in 7 of 9 patients. The median duration of drainage was 21 days (range, 7-70 days). It was well controlled with drains placed at the time of operation in 5 patients. Four patients had complications necessitating intervention in the radiology suite. This group included both patients who had distal pancreatectomies and 2 patients who had laparoscopic enucleations. These 4 patients needed a total of 6 radiology-guided procedures to control the fistulas. Therefore, we had a 44% incidence (4 of 9) of complications from pancreatic fistulas; all were managed without surgery. Clinically significant pancreatic fistulas requiring percutaneous drainage occurred in 2 of 7 patients undergoing successful laparoscopic enucleation.

Two operations were converted from laparoscopic to open procedures. The first patient was a 47-year-old man with nondiagnostic preoperative imaging studies (CT, magnetic resonance imaging, and OctreoScan [Mallinckrodt Medical, St Louis, Missouri]) and positive calcium-stimulated venous sampling in the splenic artery distribution. Endoscopic ultrasonography was not performed. The pancreas was mobilized, but no lesion could be found. A hand-assisted laparoscopic device (HandPort; Smith Nephew Surgical, Andover, Massachusetts) was placed so the pancreas could be felt, and intraoperative ultrasonography was performed. The lesion was found deep in the midbody of the gland, where it was not amenable to enucleation. The abdomen was opened, and a distal pancreatectomy was performed. The specimen contained a 1.4-cm insulinoma. Had it been attempted, the distal pancreatectomy could probably have been performed laparoscopically.

The second patient was a 56-year-old woman who had undergone a previous laparoscopic gastric bypass. Computed tomography showed no mass in the pancreas, and EUS was interpreted as showing a focal lesion in the pancreatic body. At laparoscopy, we were unable to identify a lesion visually or with laparoscopic ultrasonography. The abdomen was opened, and the pancreas was examined. No abnormality was evident, so a distal pancreatectomy was performed. The pathology diagnosis was diffuse hypertrophy of the islets.⁶ Her glucose level returned to normal after surgery and has remained so for the past year.

These findings show that laparoscopic enucleation of insulinomas is safe and effective. Pancreatic fistulas were common, but they caused little morbidity. Our 29% (2 of 7 patients) incidence of complications from the fistulas following laparoscopic enucleation was consistent with other published series. The reported rate of fistula formation or peripancreatic fluid collection ranges from 18% to 33%.^1- 5 Despite this, hospitalization was brief (only 1 day in 5 of 7 patients who had enucleation). Most patients tolerated closed-suction drainage as an outpatient without difficulty.

The principal reason for laparotomy (22% [2 of 9]) was an inability to identify the lesion that had been seen on preoperative imaging. In the first of 2 cases that were converted to an open operation, laparoscopic distal pancreatectomy may have been possible. Such an operation is preferred when it can be performed safely.

The appropriate number of preoperative imaging studies to localize an insulinoma has been the subject of debate. Contrary to previous reports,^4,7 we found that CT based on a pancreatic protocol was the study that best predicted success of laparoscopic enucleation. While only 5 of 9 scans were diagnostic, pancreatic protocol CT rendered EUS unnecessary in 4 of 7 successful laparoscopic enucleations. Furthermore, both open operations were required in patients with nondiagnostic CT scans. Endoscopic ultrasonography identified 2 lesions not visible on CT, both of which were removed laparoscopically. Although these tumors can be localized laparoscopically, patients with equivocal preoperative imaging studies are more likely to require conversion to an open operation in our experience.

All patients who underwent attempted laparoscopic enucleation had at least 1 positive preoperative localization study finding. Preoperative localization is important in planning port placement and in guiding laparoscopic ultrasonography and dissection. Additional imaging tests are usually unnecessary. Spiral CT based on a pancreatic protocol should be sufficient for most cases. Endoscopic ultrasonography may be used to confirm the CT findings in patients with equivocal or nondiagnostic CT scans. For patients with equivocal or negative findings, calcium-stimulated venous sampling can be used to confirm the diagnosis and to direct the intraoperative search for a tumor. If no tumor is found, calcium-stimulated venous sampling may help guide a blind pancreatic resection.

Our preoperative evaluation consists of spiral CT. If the scan is nondiagnostic, the patient is referred for EUS. Calcium-stimulated venous sampling is reserved for patients with equivocal EUS findings. We counsel patients that the closed-suction drains will remain for several weeks following the operation and that they should plan to return home on the first or second postoperative day.

These findings suggest that most insulinomas can be successfully treated by laparoscopic enucleation. An occasional patient may be best managed by laparoscopic distal pancreatectomy. A few patients will require enucleation or partial pancreatectomy via a laparotomy. Therefore, the surgeon or group of surgeons managing a patient with an insulinoma should optimally be capable of performing each of these procedures.

Correspondence: Quan-Yang Duh, MD, Surgical Service, Veterans Affairs Medical Center, 4150 Clement St, San Francisco, CA 94121 (quan-yang.duh@med.va.gov).

Accepted for Publication: July 25, 2006.

Author Contributions:Study concept and design: Sweet, Izumisato, Way, Clark, and Duh. Acquisition of data: Sweet, Izumisato, and Duh. Analysis and interpretation of data: Sweet, Izumisato, Masharani, and Duh. Drafting of the manuscript: Sweet, Izumisato, and Duh. Critical revision of the manuscript for important intellectual content: Sweet, Way, Clark, Masharani, and Duh. Administrative, technical, and material support: Duh. Study supervision: Way and Duh.

Financial Disclosure: None reported.