Surgical complications of pancreatectomy for persistent hyperinsulinaemic hypoglycaemia of infancy☆

Materials and methods 

A retrospective review was carried out of the surgical complications in patients undergoing subtotal pancreatectomy for PHHI by the senior author over the 13-year period (1990 through 2002). Fifty-four patients were identified with records available in 48. The operative technique has been described previously, with the exception that diathermy coagulation rather than suture ligation of the pancreatic vessels now is performed.2

Results 

The age at operation ranged from 10 days to 30 months (median, 8 weeks). Weight at operation ranged from 1.97 to 11.4 kg (median, 5.2 kg). The smallest infant, born at 33 weeks, was operated on at 7 weeks of age.

There were no deaths. Intraoperative complications comprised bleeding in 7, which was severe in 2: one from the splenic vein the other from the portal vein, both of which were repaired (Table 1).

Table 1. Surgical complications

	Intraoperative
	Haemorrhage	7
	Splenic injury	1
	Small intestinal injury	1
	Bile duct injury	2
	Postoperative
	Bile leak	2
	Bile duct stricture	3
	Wound infection	3
	Ileus	1
	Wound leakage	1
	Adhesion obstruction	1

Bile duct injury occurred in 7 patients and was related directly to the extent of the pancreatic resection. None of the 6 infants undergoing a subtotal pancreatectomy (<80%) sustained an injury to the bile duct. Five bile duct injuries were encountered among 42 patients undergoing “95%” pancreatectomy as the primary procedure (11.9%), whereas 2 bile ducts were damaged in the 9 infants undergoing redo or extended pancreatectomy after initial subtotal resection (22.2%). Two of the injuries to the bile duct were recognised intraoperatively; in one the duct was repaired, in the other, an end-to-side choledochoduodenostomy was performed. Five children underwent later choledochoduodenostomy, 2 for biliary leak and 3 for delayed bile duct stricture. Of these 3, one had a subsequent bile leak that was managed by percutaneous drainage and another later had portal fibrosis.

Other postoperative complications included wound infection in 3, prolonged ileus in one, serous wound leakage in one, and adhesion obstruction in one. One child had renal failure caused by acute tubular necrosis.

Histopathologic examination of resected specimens identified focal lesions in 15 patients, 9 of whom had undergone a curative 95% resection. A focal lesion was diagnosed intraoperatively in 6 patients (5 on frozen section); all underwent resection of the body and tall of the pancreas. Three of these required a second resection for lesions in the pancreatic head. Six of 33 patients with diffuse disease required a second resection, one of whom still requires medical treatment for hypoglycaemia, consent having been refused for a third resection. Two further patients with diffuse disease continue to require medication for hyperinsulinism. Twenty patients required insulin, and 13 required pancreatic enzyme replacement at the time of last review (Fig 1).

View Large Image Download to PowerPoint

Fig. 1. Outcome of hyperglycaemia.

Discussion 

Hypoglycaemia with hyperinsulism in infancy frequently is transient: the majority of cases can be managed medically without resorting to surgery.11, 12 During the period of stabilisation and diagnosis secure venous access is essential. These6 infants usually are obese, making peripheral cannulation difficult. Early central venous catheter insertion will allow continuous controlled administration of glucose, reducing the risk of hypoglycaemia and subsequent brain damage as well as providing access for repeat blood sampling. Initial medical management includes increasing the carbohydrate intake, accompanied by medication to reduce insulin secretion and promote glycogen mobilisation; these include diazoxide, chlorothiazide, octreotide, and glucagon.13 Once the diagnosis has been established, infants with PHHI that cannot be controlled safely with frequent feeding and oral medication should undergo elective partial pancreatectomy as soon as possible.

The extent of pancreatectomy remains controversial. Inadequate resection will result in continued hyperinsulinism, whereas near-total pancreatectomy will result in almost all patients eventually requiring insulin therapy.8 Differentiation of focal from diffuse disease may allow for more limited resection in approximately one third of patients whose disease is focal.7 Pancreatic venous sampling (PVS) and intraarterial calcium stimulation tests are used to identify and localise focal disease. Our initial experience with these techniques have been disappointing with less than 50% successful diagnosis of focal disease. This compares with published results of 60% to 80% of focal cases correctly identified by PVS.10, 14 Intraoperative biopsy and frozen section are used in conjunction with PVS to identify focal lesions. A biopsy showing normal pancreatic tissue promotes exploration for a focal abnormality. In our experience, focal lesions are not identified easily macroscopically, and patients with suspected focal disease underwent 70% to 80% distal pancreatectomy at the initial operation. Focal lesions were present in 15 patients, 9 had undergone 95% resection (8 before the practice of frozen section assessment). Six underwent limited pancreatic resection, 3 of whom required a second resection for lesions in the pancreatic head. Thus, correct pre- or intraoperative localisation of the lesion may have permitted more limited resection in 4 of our 48 patients.

Haemorrhage is a major risk in pancreatic resection because of the rich vascular supply and the close proximity of the pancreas to the splenic and portal vein and superior mesenteric vessels. This complication is not reported elsewhere, although this may be because of underreporting of a complication common to many surgical procedures. Some investigators do report splenic conservation despite ligation of the splenic vein, presumably to control bleeding.3, 5, 15 In most cases, bleeding from the splenic vein can be controlled by direct suture while maintaining the venous blood flow from the spleen.

Intraoperative bile duct injury and late bile duct strictures presumed secondary to ischaemia are reported in most of the large series.7, 8, 16 Ninety-five percent or greater resection requires resection of pancreatic tissue in close proximity to the common bile duct and adjacent to the duodenum. Technically, this is the most difficult part of the resection, and positive identification of the bile duct before dissection combined with a careful search for injury after dissection are essential.

Long-term follow up data have shown that almost all children undergoing subtotal pancreatectomy eventually will need insulin replacement therapy.8 We anticipate that with longer follow-up the number of patients in our series requiring insulin therapy will increase. We also expect pancreatic exocrine deficiency to develop particularly in children undergoing near-total resection and, thus, actively investigate these children for pancreatic enzyme deficiency.

Partial pancreatectomy brought about resolution of hyperinsulinism in 44 of 48 patients. However, only 16 patients required no further surgery or medication. Thus, although pancreatectomy for PHHI, successfully controlled hyperinsulinism in most of the infants, it was associated with major intra- and postoperative morbidity in over half. Pancreatic resection currently is the only option for patients with PHHI. However, the risk of surgical complications combined with the likelihood of pancreatic insufficiency mean that this answer to the problem of hypersecretion of insulin is far from satisfactory.