Multiple endocrine neoplasia type 1: duodenopancreatic tumours

Transcription

1 Journal of Internal Medicine 2003; 253: MINISYMPOSIUM Multiple endocrine neoplasia type 1: duodenopancreatic tumours G. M. DOHERTY & N. W. THOMPSON Division of Endocrine Surgery, Department of Surgery, University of Michigan, MI, USA Abstract. Doherty GM, Thompson NW (University of Michigan, MI, USA). Multiple endocrine neoplasia type 1: duodenopancreatic tumours (Minisymposium). J Intern Med 2003; 253: The pancreaticoduodenal disease in Multiple endocrine neoplasia type 1 (MEN1) is the most frequent cause of death due to the syndrome, and the most controversial with regard to its management. This article discusses the current data and recommendations with respect to disease screening, functional tumour diagnosis, natural history, preoperative imaging, operative strategy and follow-up. Keywords: gastrinoma, islet cell tumour, MEN1, pancreas. Definitions Multiple endocrine neoplasia type 1 (MEN1) is an autosomal dominant inherited syndrome, with significant variability in its clinical expression. It is classically characterized by tumours of the parathyroid glands, pancreatic islets and anterior pituitary gland; with detailed study, it is now known that other tissues are also often involved, including thymus and bronchi (carcinoids), subcutaneous fat (lipomas), thyroid gland, adrenal glands and skin. The disease is characterized by near-complete penetrance and variable expressivity. Larsson and co-workers used genetic linkage to establish in 1988 the affected chromosomal region-11q13 [1]. Recently the gene that is mutated in patients with MEN1 has been identified, and several groups now do direct DNA testing to identify carriers [2 4]. Screening Screening in MEN1 can include either screening for the MEN1 mutation in a family with a known mutation, to identify gene carriers, or screening for the components of the disease in known gene carriers. This discussion will be limited to the screening for pancreatic endocrine tumours in patients with the MEN1 mutation. The options for screening include (i) hormone measurements, (ii) cross-sectional imaging, (iii) somatostatinreceptor scintigraphy (SRS) and (iv) endoscopic ultrasonography (EUS). Biochemical screening Hormone measurements have been useful to screen for pancreatic tumours in MEN1 because of the frequent functional nature of the tumours [5 7]. Efficient screening programmes have been designed for clinical recognition of the lesions. Patients can be screened biochemically, either with or without a standard meal or other provocative agent, such as calcium or secretin. The results of the biochemical screening can then be used to select patients for imaging examinations. Biochemical testing is useful 590 Ó 2003 Blackwell Publishing Ltd

2 MINISYMPOSIUM: MEN1 AND DUODENOPANCREATIC TUMOURS 591 for annual or semiannual screening of asymptomatic gene carriers. Patients with specific symptoms require investigation by other means. The advantage of biochemical testing is that the patients can be identified as harbouring functional pancreatic endocrine tumours in a presymptomatic state. The disadvantages of the biochemical testing are that some tumours make insufficient hormone to be identified, and some patients harbour tumours that are identifiable biochemically, but not imageable at that point [7]. This can lead to extensive testing and possibly operation that is unlikely to benefit the patient clinically. A standard annual biochemical screening programme has been advocated by Dr Skogseid and her colleagues from the University of Uppsala, Sweden (Table 1). In their programme, within their annual visit that also includes evaluation for parathyroid and pituitary disease, patients have baseline levels of insulin, proinsulin, glucagon, pancreatic polypeptide (PP) and gastrin, followed by meal-stimulated levels of gastrin and pancreatic polypeptide. This approach identified pancreatic tumours as the initial presenting abnormality in several patients and identified tumours at an early stage. However, the biochemical screening, and particularly the meal test, can generate some false-positive results. As a rule, mild elevations in the biochemical screening may be best repeated after an interval to establish confidence that it reflects pancreatic endocrine neoplasia [8]. Efforts to replace the standard meal test with a provocative test using calcium and secretin have been stalled by the unavailability of secretin, and the uncomfortable nature of the calcium infusion [9]. However, the infusion of calcium (2 mg/kg) and secretin (2 U/kg) in 10 known MEN1 carriers did demonstrate marked differences in induced (2-min peak) serum levels of pancreatic polypeptide, gastrin and insulin compared with normal volunteers. The elevation of hormone levels was also correlated with the presence of imageable tumours. Cross-sectional imaging and somatostatin-receptor scintigraphy screening Screening with cross-sectional imaging, such as computed tomography (CT) scan, can demonstrate tumours in the pancreas, regional lymph nodes and hepatic metastases. This modality has the advantage of identifying tumours anatomically, which can provide information useful for clinical decisions about resection (Fig. 1). In addition, by definition, CT scan will not identify tumours that cannot then be found. The disadvantage of CT scan is that it identifies lesions, without regard to their histology or function. Thus, lesions identified in the pancreas, or more commonly, in adjacent nodes, may be irrelevant findings independent of the MEN1 syndrome. Prospective studies that have evaluated the sensitivity of CT imaging to detect pancreatic endocrine tumours, have found that it is significantly less sensitive than biochemical testing in identifying early disease [7]. Thus, for screening, CT is less useful, but for establishing the anatomical extent of disease after biochemical diagnosis, CT is critical. Table 1 Biochemical screening programme for MEN1 Serum calcium Serum glucose Serum parathyroid hormone Serum prolactin Serum growth hormone Serum insulin Serum proinsulin Serum C-peptide Serum pancreatic polypeptide Serum gastrin Serum calcitonin Plasma glucagon Plasma somatostatin Plasma VIP Serum hcg a and b Meal stimulation test hcg, human chorionic gonadotropin; VIP, vasoactive intestinal polypeptide. Source: Skogseid B et al. Multiple endocrine neoplasia type 1: a 10-year prospective screening study in four kindreds. J Clin Endocrinol Metab 1991;73: Fig. 1 Computed tomographic scan demonstrating a large nonfunctioning tumour in the head of the pancreas of a patient with MEN1. The patient had subsequent pancreaticoduodenectomy.

3 592 G. M. DOHERTY & N. W. THOMPSON of 29) of patients. In previously operated patients, SRS found tumour in 40% (four of 10) of patients, again with both new positives and false-negatives compared with other imaging. SRS also had three important false-positive results, including one patient who had laparotomy with no tumour identified. Thus, SRS was useful in identifying otherwise occult neuroendocrine tumours in MEN1 patients and substantially altered management. However, SRS also has significant false-positive and false-negative results which demand correlation with other studies. Endoscopic ultrasound screening Fig. 2 Somatostatin-receptor scintigraphy demonstrating a metastatic lesion in the left lateral segment of the liver of a patient with MEN1 and gastrinoma. The patient has subsequent resection of his duodenal gastrinoma and isolated liver metastasis. The SRS can identify lesions in and around the pancreas and liver, as well as throughout the remainder of the body (Fig. 2). SRS has the advantage of identifying tumours functionally, based on the size of the tumour mass and the concentration of somatostatin receptors [10, 11]. The disadvantage of SRS lies in its lack of anatomical specificity. The nuclear medicine images, even with single-photon emission computed tomography (SPECT) crosssectional imaging, provide only general anatomical definition with respect to surrounding organs. SRS must always, therefore, be combined with CT scan or other anatomical imaging techniques to provide anatomical correlates to the scintigraphic findings. In addition, SRS has the advantage of identifying other potential sites of disease, including intrathoracic disease from either metastases of the pancreatic primary, or carcinoid tumours of the thymus or bronchus. In a prospective study of SRS in MEN1 patients, 37 SRS studies were performed in 29 MEN1 patients [10]. SRS identified occult tumour in 36% (four of 11) of patients with only biochemical evidence of neuroendocrine tumour; two patients went on to resection. SRS showed tumour in 79% (15 of 19) of patients with CT-demonstrated tumour; 30% (six of 20) of the SRS lesions were occult on CT. Conversely, 55% (16 of 29) of CT-identified lesions were occult on SRS. SRS found distant disease in 21% (six The EUS is the most sensitive imaging modality for small pancreatic endocrine tumours. Its advantage lies in the ability to image the parenchyma of the pancreas and the immediately peri-pancreatic lymph nodes with exquisite precision. The disadvantage of the EUS is that it cannot image any of the other potential sites of disease, such as the liver or chest. In a retrospective review of EUS in MEN1 patients screened at the University of Michigan, 14 of 15 asymptomatic patients were found to have pancreatic tumours (12 with multiple lesions) with a median size of 1.5 cm (range cm) [12]. Many of these lesions are at or below the level of reliable detection of other imaging modalities, such as CT scan, and the EUS also provided detailed anatomical information useful for operative planning. The EUS was not useful for identifying duodenal wall gastrinomas, which are frequent in MEN1. Because of its tremendous sensitivity in the pancreas, but limited anatomical scope, the EUS may be useful for a specific role in MEN1 screening, to identify otherwise occult tumours in the asymptomatic screening situation. A unified screening plan for pancreatic endocrine tumours in MEN1 should take advantage of the strengths of each of these modalities. Biochemical assessment of hormones can lead to detailed investigation, but should probably not be the sole screening method. As the most sensitive test, biochemical screening can reassure clinicians and patients, and allow for other examinations at greater intervals if the biochemistry remains normal from year to year. For imaging examinations, a combination of CT scan and SRS provides the best coverage of the abdomen and thorax, to identify

4 MINISYMPOSIUM: MEN1 AND DUODENOPANCREATIC TUMOURS 593 anatomical correlates to the functional lesions revealed by biochemical testing, and should be performed at some interval (every 2 3 years) even in the presence of normal biochemistry, to evaluate for nonfunctional lesions. Finally, the EUS appears to be useful in providing a very sensitive assessment of the pancreas, and may be especially useful in patients with biochemical evidence of tumour, but no imageable disease on SRS or CT scan. This may also be of particular value in patients who have had a previous partial pancreatectomy, which can complicate the interpretation of the CT scan. Functional tumours Pancreatic or duodenal neuroendocrine tumours in MEN1 become symptomatic most frequently in the fourth or fifth decade, however the biochemical abnormalities often develop in the third decade [7]. Once symptomatic, the clinical presentation is typically dependent on the increased hormone levels. Patients may present early when tumours are small, for example with duodenal gastrinomas, which are undetectable on preoperative imaging, but that cause significant peptic ulcer disease (PUD), oesophageal reflux symptoms and diarrhoea. Symptoms of local enlargement or infiltration including back pain and abdominal mass, left-sided portal hypertension, jaundice, or metastatic disease (cachexia, hepatosplenomegaly) may rarely be present at presentation, more frequently in older patients with nonfunctional tumours. Nonfunctional neuroendocrine tumours, of which three-quarters produce pancreatic polypeptide but no syndrome, are the most common tumours overall. Gastrinomas are the most common functional gastrointestinal neuroendocrine tumours, presenting in up to 54% of MEN1 patients. Other functional enteropancreatic neuroendocrine tumours occur occasionally; the most frequent of these is insulinoma, whilst glucagonoma, VIP-oma, growth hormone releasing factor (GRF)-oma, and somatostatinoma are less common [5]. More detailed discussion of each tumour diagnosis follows. Gastrinoma Zollinger Ellison Syndrome (ZE) was initially described as non-b islet cell tumour of the pancreas, with gastric acid hypersecretion and severe PUD, which is less common now because of effective acid suppression therapy. The majority occur in the region of the duodenum or head of the pancreas. Gastrinomas may occur in the pancreas of patients with MEN1, but it is now evident that gastrinomas are very frequently in the duodenum, regardless of their presence in the pancreas [13 19]. In addition, most patients have multiple, small submucosal lesions of the duodenum. The tumours are usually malignant, and regional lymph node metastases are found in at least 50% of patients at the time of pancreatic surgery. The diagnosis of ZE depends on proving elevated gastrin levels in the presence of gastric acid. Therefore, a fasting serum gastrin is typically the initial test, but a simultaneous gastric acid output or ph measurement is necessary to establish the diagnosis of ZE. Approximately, onethird of patients with ZE can be diagnosed by a serum gastrin > 1000 pg ml )1 with a gastric ph < 3.0. The other two-thirds of the patients require provocative tests, which have become more difficult as the secretin supply is limited. Insulinoma Less than 10% of patients with MEN1 have an insulinoma, and less than 10% of patients with insulinomas have MEN1. Whilst the insulin-producing tumour might be one of several islet cell tumours in the patient, the tumour that is making the insulin is usually solitary and relatively large, on the order of 2 4 cm [20 22]. Patients usually present with symptoms of neuroglycopenia during fasting hypoglycaemia (<40 mg dl )1 ). The diagnosis is made by documenting hypoglycaemia in association with inappropriately increased plasma levels of insulin and C-peptide during a supervised fast. Other causes of hypoglycaemia include medications (insulin, sulphonylureas), liver dysfunction, renal failure, wasting and growth hormone deficiency. Once the diagnosis is made, preoperative localization of the functional tumour is necessary to plan the surgical approach. In occasional patients, selective angiography with calcium provocation of insulin production and measurement in the hepatic veins is useful to definitively regionalize the insulin-producing tumour [23].

5 594 G. M. DOHERTY & N. W. THOMPSON Glucagonoma Glucagon-producing tumours can result in migratory necrolytic erythema, weight loss, glucose intolerance, hypoaminoacidaemia and normochromic, normocytic anaemia. Other less common features include thromboembolic phenomena, neuropsychiatric disturbances, diarrhoea and nonspecific abdominal pain [24]. Glucagonomas occur in 3% of all MEN1 patients. They are often large at presentation, 5 10 cm, and malignancy is common. Patients present with signs and symptoms of hyperglycaemia, and usually diabetes mellitus precedes the diagnosis of glucagonoma. However, Cushing s Syndrome is a far more common cause of hyperglycaemia in MEN1. The diagnosis is made by a glucagon level greater than 1000 pg ml )1, often in the presence of the characteristic skin rash. VIP-oma VIP-omas usually (>80%) originate in the pancreas. The signs and symptoms include severe profuse watery diarrhoea (>700 ml day )1 and up to 8 L), hypokalaemia, and hypochlorhydria [24]. Approximately 1% of VIPomas occur in patients in MEN1. VIP-omas are almost always solitary, and they are usually greater than 3 cm in size. Approximately three-quarters occur in the pancreatic tail. Greater than 60% of VIP-omas are malignant, and 37 68% of patients have metastases at presentation. The diagnosis depends on a high volume of secretory diarrhoea in the presence of a pancreatic endocrine tumour, ideally with documentation of elevated VIP (normal level <170 pg ml )1 ). The differential diagnosis of secretory diarrhoea includes pseudo-vipoma syndrome, laxative abuse and ZE. GRF-oma A GRF-oma is an endocrine tumour which produces growth hormone releasing factor, which causes acromegaly. It is a rare tumour, and approximately 30% are associated with MEN1 [24]. These tumours typically occur in the lung (53%), pancreas (30%), or small intestine (10%). The tumours are often multiple, large, and metastatic. 65% of GRF-omas are associated with a tumour causing another hormonal syndrome, such as ZE or Cushing s Syndrome. The diagnosis of a GRF-oma is suspected in patients with a pancreatic endocrine tumour and acromegaly, especially in a patient with ZE, an adrenocorticotropic hormone (ACTH)-producing pancreatic tumour, or MEN1. However, it is a rare cause of acromegaly. The diagnosis is confirmed by a GRF level greater than 300 pg ml )1. Somatostatinoma Somatostatinomas typically occur in the pancreas or small bowel. The mean age of onset is >50 years. Most are solitary, large (average size 5 cm), and have metastases to lymph nodes or liver [24]. This hormonal syndrome typically includes mild diabetes, gallbladder disease, weight loss, anaemia, diarrhoea, steatorrhoea and hypochlorhydria. The clinical effects are more common in tumours located in the pancreas compared with tumours in the duodenum. These tumours are typically first noticed incidentally during intra-abdominal imaging. The diagnosis is then made by an increase in somatostatin-like immunoreactivity in plasma and/or an increase in the number of D cells defined by immunohistochemical stains of the tumour. PP-oma These tumours are clinically silent, however they are the most common neuroendocrine tumour in MEN1 [25]. The true incidence is unknown. They occur throughout the pancreas and may be large at presentation. Malignancy is common, and occurs in 64 92% of patients. The diagnosis is made by an increase in the PP level. Other causes for increase in PP levels include older age, inflammatory conditions, bowel resection, alcohol abuse, chronic renal failure and diabetes mellitus. Natural history and the decision to intervene The natural history of pancreatic disease in MEN1 has been difficult to define in the past because of the rarity of the disease and significant variability in the virulence of the pancreatic malignancy. However, three recent studies have identified significant mortality associated with the pancreatic islet cell tumours, a natural history that demands intervention in that group of patients with virulent malignancy. The studies evaluated patients from

6 MINISYMPOSIUM: MEN1 AND DUODENOPANCREATIC TUMOURS 595 Tasmania, and those followed at Washington University and the Mayo Clinic [26 29]. In the Washington University study, the database contained 34 distinct kindreds with 1838 members. Reliable death data were available for 103 people and survival curves of MEN1 patients who died from causes related to MEN1, were compared with MEN1 carriers who died from a nonendocrine cause, and to unaffected kindred members [26]. We also compared ages of death between affected and unaffected members of MEN1 kindreds. Of 59 MEN1 affected patients, 27 died directly of MEN1-specific illness, and 32 died of non-men1 causes. The MEN1-specific deaths occurred younger (median 47 years) than either MEN1 patients whose death was from some nonendocrine cause (median 60 years, P < 0.02), or than all kindred members who did not die of MEN1 disease (median 55 years, P < 0.05) (Fig. 3). The causes of death in the MEN1 patients included islet cell tumour (12 patients), ulcer disease (six patients), hypercalcaemia/uraemia (three patients), carcinoid tumour (six patients), and nonendocrine malignancies (nine patients). MEN1 carriers did not have a difference in survival compared with unaffected kindred members. Thus, of the Washington University MEN1 patients, 46% died from causes related to their endocrine tumours after a median of 47 years, which was younger than family members who did not die of these tumours, and the most frequent cause of death, especially in the more recent years of the study, was pancreatic islet cell Proportion surviving Men1- specific deaths Non-Men1- specific deaths in all kindred members P < Age (years) Fig. 3 Survival of patients who died of an MEN1-related cause compared with family members who did not. From: Doherty GM et al. Lethality of multiple endocrine neoplasia type 1. World J Surg 1998;22: tumours. In addition, the metastatic islet cell tumours accounted for some of the youngest deaths in the study median age 46 years (range years). Similar data were found in the Tasman and Mayo Clinic series. In the Tasman series, there were 46 deaths in patients with MEN1, 20 of which were caused by components of the disease (44%) [28]. Of these, only three were due to metastatic islet cell tumour; however each of these three were in the most recent decade of the study, when the patients were best evaluated and characterized. The Mayo Clinic series included 60 deaths, of which only 17 were clearly related to MEN1 (28%), however 10 of these were due to metastatic islet cell tumour [29]. The combined series includes 64 deaths clearly related to MEN1, 43 of which were due directly to malignant neoplasms, and 25 of which were malignant islet cell tumours. Thus the natural history of MEN1 justifies an aggressive screening programme with early therapeutic intervention when a tumour is identified. The most controversial issue in MEN1 management is probably the decision to intervene in the pancreatic disease. In our opinion all macroscopic tumours of the MEN1 pancreas should be regarded as potentially malignant, and neither tumour size nor radiological findings or peptide production can safely be used as markers of malignancy (Tables 2 and 3) [30]. It has been shown that pancreatic endocrine lesions develop during the third decade, and time may be an important risk factor for mutagenetic events essential for malignant transformation [31]. In addition, 50% of middle-aged patients display metastases. In our opinion, the prevention of death from metastatic islet cell tumour Table 2 Metastatic potential related to hormone production of tumours Number of patients with metastases to: %of n Lymph nodes Liver patients Gastrinoma Insulinoma Nonfunctional VIP-oma Source: Lowney JK et al. Pancreatic islet cell tumour metastasis in multiple endocrine neoplasia type 1: correlation with primary tumour size. Surgery 1999;125:

7 596 G. M. DOHERTY & N. W. THOMPSON Table 3 Preoperative assessment of largest primary tumour size Size of largest primary tumour (max. diam.) Number of patients (n ¼ 43)* requires pancreatic surgery before frank clinical syndromes occur, in relatively young MEN1 patients, based on active screening programmes. Preoperative imaging Number of patients with metastases No tumour Imageable <1 cm cm >2 cm Percentage of patients with metastases *Includes all patients with documentation of preoperative studies. Source: Lowney JK et al. Pancreatic islet cell tumour metastasis in multiple endocrine neoplasia type 1: correlation with primary tumour size. Surgery 1999;125: The goals of operative resection of pancreatic endocrine tumours are complete removal of the gross tumour burden, and preservation of pancreatic function. The preoperative imaging of the pancreatic disease, then, is selected to define the extent of resection, and to identify tumour that would be outside the scope of resection. The available modalities include those noted above in the discussion of screening. The minimum preoperative imaging should include CT scan of the abdomen and pelvis, and SRS [10]. Other imaging techniques should be applied to address specific questions. For example, questions regarding abnormalities in the liver on CT or SRS can often be resolved by MR scan, and possible lesions in the pancreas can be delineated by endoscopic ultrasound. The advantages and disadvantages of each modality are discussed above, however, the importance of the complementary nature of SRS and cross-sectional imaging (typically CT scan) bears repeating. Each of these modalities has significant false-negatives and false-positive studies, but the combination of the two decreases both types of inaccuracy. Thorough preoperative imaging can help to accurately plan the procedure; however, the final plan for the resection cannot be made until the time of intraoperative assessment of the true tumour burden, including manual exploration of the abdomen, intraoperative ultrasound of the pancreas and, usually, duodenotomy to search for intraduodenal tumours. Operative management The operative management must be individualized for each patient based on their pattern of disease. The principles are complete tumour resection and preservation of pancreatic function, by preserving as much grossly normal pancreas as possible, whilst minimizing the morbidity of the procedure [13, 32, 33]. In practice, this often results in the subtotal resection of the distal pancreas, and enucleation of tumours in the head of the pancreas and duodenum (Fig. 4). This operation removes the gross disease in the distal pancreas, whilst preserving most of the pancreatic mass in the head, and avoiding the need for a pancreatic anastamosis. For patients with bulky disease in the head of the pancreas, however, which may not be amenable to enucleation, then a better option may be pancreaticoduodenectomy, and enucleation of any tumours in the tail. In all patients with gastrinoma, the duodenum should be opened and submucosal tumours resected (Fig. 5). Occasional patients may have disease that can be completely removed without formal pancreatic resection (Fig. 6). Total pancreatectomy is rarely indicated, or necessary, to meet the goal of complete tumour resection. A contemporary series of pancreatectomy in MEN1 patients catalogues the operations and their outcomes at Washington University between 1993 and 1999 [32]. Twenty-one consecutive patients with tumours identified through an integrated disease screening programme underwent exploration, with the twin goals of complete tumour resection and preservation of grossly normal pancreas. In this series of patients, five required pancreaticoduodenectomy, 11 had nonwhipple pancreatic resections and five Duodental gastrinoma Stomach Duodenum Pancreatic PPoma Fig. 4 Resection specimen from a patient with the most common pattern of resection in MEN1 pancreas operation. The operation includes resection of the distal pancreas, duodenal gastrinomas and peripancreatic lymph nodes.

8 MINISYMPOSIUM: MEN1 AND DUODENOPANCREATIC TUMOURS 597 Fig. 5 Submucosal duodenal gastrinomas can often only be identified at the time of duodenotomy, with eversion of the duodenal mucosa and palpation between digits. Here the arrow denotes the small duodenal primary. Fig. 6 Postresection photograph of a patient who has had resection of duodenal gastrinomas (note the transverse duodenal closure, white arrow) and enucleation of tumour from the pancreatic neck (black arrow). No other tumours were evident in the pancreas by palpation or intraoperative ultrasound. underwent enucleation of limited neuroendocrine tumours. The morbidity of these operations was significant, and two patients died in the perioperative period. One patient developed fatal viral encephalitis 6 weeks after an otherwise uneventful recovery. The other mortality was in an elderly man with prohibitive pulmonary function [forced expiratory volume (FEV 1 ) < 700 ml] and uncontrollable hypoglycaemia from an insulinoma, who died of pulmonary failure following resection. The morbidity of operation included a 14% rate of anastomotic leak and abscess and a 10% rate of wound infection. In these patients, the operative strategy was able to render them disease-free in the short term, but follow-up to date is insufficient to demonstrate efficacy in this series. In a similar study with more extended follow-up, the University of Uppsala group detailed their approach in 20 patients, 12 of whom were symptomatic and eight patients identified by screening [31]. They followed the same guidelines of complete tumour resection with preservation of functional pancreatic mass. Till date, none of the patients whose tumours were identified by screening in their study have developed metastatic disease. This currently appears to be the most viable strategy, whilst further follow-up of both this cohort and the Washington University series is necessary to determine whether this strategy will have a long-term benefit in reducing the occurrence of death from malignant islet cell tumours. Follow-up Continued surveillance for components of MEN1 is necessary for all patients. This is a syndrome that one can manage, but not cure, and thus it requires continued attention. Patients should be assessed at least annually and more frequently if indicated by the tempo of their disease. Ongoing surveillance follows the same pattern and uses the same modalities as the screening in unoperated patients. The useful modalities include biochemical hormone screening, and imaging with CT scan, SRS and possibly EUS. Conclusions Pancreatic endocrine tumours in MEN1 are the most likely cause of death in gene carriers. Strategies for identifying disease and managing it effectively include integrated screening programmes using biochemical and image-based approaches. Operative resection should be considered for any patient with imageable tumour, with the main goals of the operation being the complete resection of grossly evident tumour, and the preservation of pancreatic function. Further study is necessary to demonstrate the overall success of this strategy. Conflict of interest statement No conflict of interest was declared.

9 598 G. M. DOHERTY & N. W. THOMPSON References 1 Larsson C, Skogseid B, Öberg K, Nakamura Y, Nordenskjôld M. Multiple endocrine neoplasia type 1 gene maps to chromosome 11 and is lost in insulinoma. Nature 1988; 332: Chandrasekharappa SC, Guru Sc, Manikam P et al. Positional cloning of the gene for multiple endocrine neoplasia-type 1. Science 1997; 276: Skogseid B, Doherty GM. Multiple endocrine neoplasia type 1: clinical and genetic features. Digestive & Liver Disease (Suppl. 2): S Brandi ML, Gagel RF, Angeli A et al. Guidelines for diagnosis and therapy of MEN type 1 and type 2. J Clin Endocrinol Metabol, 2001; 86: Skogseid BS, Rastad J, Oberg K. Multiple endocrine neoplasia type 1: Clinical features and screening. Endocrinol Metab Clin North Am 1994; 23: Vasen HF, Lamers CBH, Lips CJM. Screening for the multiple endocrine neoplasia syndrome type 1. Arch Intern Med 1989; 149: Skogseid BS, Eriksson B, Lundqvist G et al. Multiple endocrine neoplasia type 1: A 10-year prospective screening study in four kindreds. J Clin Endocrinol Metab 1991; 73: Skogseid BS, Oberg K, Benson L et al. A standardized meal stimulation test of the endocrine pancreas for early detection of pancreatic endocrine tumors in multiple endocrine neoplasia type 1 syndrome: five years experience. J Clin Endocrinol Metab 1987; 64: Olson JA, DeBenedetti M, Wells SA, Doherty GM. Islet cell neoplasia in multiple endocrine neoplasia, type 1: Development of a provocative screening test. Surg Forum 1996; 47: Yim JH, Siegel BA, DeBenedetti MK, Norton JA, Lairmore TC, Doherty GM. Prospective study of the utility of somatostatin receptor scintigraphy in the evaluation of patients with multiple endocrine neoplasia type I. Surgery 1998; 124: Meko JB, Doherty GM, Siegel BA, Norton JA. Evaluation of somatostatin-receptor scintigraphy for detecting neuroendocrine tumors. Surgery 1996; 120: Gauger PG, Scheiman JM, Wamsteker E-J, Richards ML, Doherty GM, Thompson NW. An aggressive strategy of early anatomic screening and operative treatment of pancreatic endocrine tumors in asymptomatic patients with multiple endocrine neoplasia type 1. Br J Surg 2003, in press. 13 Thompson NW. Surgical treatment of the endocrine pancreas and Zollinger-Ellison syndrome in the MEN I syndrome. Henry Ford Hosp Med J 1992; 40: Melvin WS, Johnson JA, Sparks J, Innes JT, Ellison EC. Longterm prognosis of Zollinger-Ellison syndrome in multiple endocrine neoplasia. Surgery 1993; 114: Cherner JA, Sawyers JL. Benefit of resection of metastatic gastrinoma in multiple endocrine neoplasia type I. Gastroenterology 1992; 102: Pipeleers-Marichal M, Somers G, Willems G etal. Gastrinomas in the duodenums of patients with multiple endocrine neoplasic type 1 and the Zollinger-Ellison syndrome. N Engl J Med 1990; 322: Pipeleers-Marichal M, Donow C, Heitz PU, Kloppel G. Pathologic aspects of gastrinomas in patients with Zollinger-Ellison syndrome with and without multiple endocrine neoplasia type I. World J Surg, 1993; 17: Sheppard BC, Norton JA, Doppman JL, Maton PN, Gardner JD, Jensen RT. Management of islet cell tumors in patients with multiple endocrine neoplasia: a prospective study. Surgery, 1989; 106: Grama D, Skogseid B, Wilander E et al. Pancreatic tumors in multiple endocrine neoplasia type 1: clinical presentation and surgical treatment. World J Surg 1992; 16: Jadoul M, Koppeschaar HPF, Bax MA et al. Insulinomas in MEN1 patients: early detection and treatment of insulinomas in patients with the multiple endocrine neoplasia syndrome type-1. Neth J Med 1990; 37: Mignon M, Ruszniewski P, Podevin P et al. Current approach to the management of gastrinoma and insulinoma in adults with multiple endocrine neoplasia type I. World J Surg 1993; 17: O Riordain DS, O Brien T, VanHeerden JA, Service FJ, Grant CS. Surgical management of insulinoma associated with multiple endocrine neoplasia type 1., World J Surg 1994; 18: Cohen MS, Picus D, Lairmore TC, Strasberg SM, Doherty GM, Norton JA. Prospective study of provocative angiograms to localize functional islet cell tumors of the pancreas. Surgery 1997; 122: Doherty GM, Skogseid BM. Surgical Endocrinology. Philadelphia, PA: Lippincott, Williams and Wilkins, Mutch MG, Frisella MM, DeBenedetti MK et al. Pancreatic polypeptide is a useful plasma marker for radiographically evident pancreatic islet cell tumors in patients with multiple endocrine neoplasia type 1. Surgery 1997; 122: Doherty GM, Olson JA, Frisella MM, Lairmore TC, Wells SA, Norton JA. Lethality of multiple endocrine neoplasia type 1. World J Surg 1998; 22: Shepherd JJ. The natural history of multiple endocrine neoplasia type 1.. Arch Surg 1991; 126: Wilkinson S, Teh BT, Davey KR, McArdle JP, Young M, Shepherd JJ. Cause of death in multiple endocrine neoplasia type 1. Arch Surg 1993; 128: Dean PG, van Heerden JA, Farley DR et al. Are patients with multiple endocrine neoplasia type 1 prone to premature death? World J Surg 2000; 24: Lowney JK, Frisella MM, Lairmore TC, Doherty GM. Islet cell tumor metastasis in multiple endocrine neoplasia type I: correlation with primary tumor size. Surgery 1998; 124: Skogseid B, Oberg K, Eriksson B et al. Surgery for asymptomatic pancreatic lesion in multiple endocrine neoplasia type 1. World J Surg 1996; 20: Lairmore TC, Chen VY, DeBenedetti MK, Gillanders WE, Norton JA, Doherty GM. Duodenopancreatic resections in patients with multiple endocrine neoplasia type 1 (MEN 1). Ann Surg 2000; 231: Skogseid B, Grama D, Rastad J et al. Operative tumour yields obviates preoperative pancreatic tumor localization in multiple endocrine neoplasia type 1. J Intern Med 1995; 238: Received 12 March 2003; accepted 20 March 2003 Correspondence: Gerard M. Doherty, M.D., NW Thompson Professor of Surgery, Head, Section of General Surgery, University of Michigan, 2920 Taubman Center Ann Arbor, MI 48109, , USA (fax: ; gerardd@ umich.edu).