Gastrointestinal Neuroendocrine Tumors: Pancreatic Endocrine Tumors

Abstract

Pancreatic endocrine tumors (PETs) have long fascinated clinicians and investigators despite their relative rarity. Their clinical presentation varies depending upon whether the tumor is functional or not and also according to the specific hormonal syndrome produced. Tumors may be sporadic or inherited but little is known about their molecular pathology, especially the sporadic forms. Chromogranin A appears to be the most useful serum marker for diagnosis, staging and monitoring. Initially, therapy should be directed at the hormonal syndrome as this has the major initial impact on the patient's health. Most PETs are relatively indolent but ultimately malignant, except for insulinomas which are predominantly benign. Surgery is the only modality that offers the possibility of cure although it is generally noncurative in patients with Zollinger-Ellison syndrome or nonfunctional PETs with MEN1. Preoperative staging of disease extent is necessary to determine the likelihood of complete resection though debulking surgery is often felt to be useful in unresectable patients. Once metastatic, biotherapy is usually the first modality employed because it is generally well tolerated. Systemic or regional therapies are generally reserved until symptoms occur or tumor growth is rapid. Recently a number of newer agents, as well as receptor-directed radiotherapy, are being evalulated for patients with advanced disease. This review addresses a number of recent advances regarding the molecular pathology, diagnosis, localization and management of PETs including discussion of peptide receptor radionuclide therapy and other novel antitumor approaches. We conclude with a discussion of future directions and unsettled problems in the field.

Introduction

Pancreatic endocrine tumors (PETs) have long fascinated clinicians and investigators because of their unusual and florid symptoms as well as the insights they provide into the actions of gastrointestinal (GI) hormones. PETs share many pathological and biological features with GI carcinoids, but they have important differences which affect treatment as well as having a different pathogenesis^{1, 2}, and thus the two groups of gastrointestinal neuroendocrine tumors(NETs) are best considered separately. There have been a number of recent advances in various aspects of PETs including diagnosis, management, insights into molecular changes, tumor localization, and the treatment of advanced disease. This paper will briefly review a number of these advances as well as their current management. All aspects of PETs will not be covered, because many features are recently covered in reviews or consensus conferences^3-7.

Epidemiology

PETs occur in 0.5-1.5% of autopsies but are functional or symptomatic in <1/1000, resulting in a clinical detection rate of 1:100,000 population, which comprise 1-2% of pancreatic neoplasms⁸. In older studies nonfunctional PETs (NF-PETs), insulinomas and gastrinomas had equal frequency⁹, however in recent studies NF-PETs are twice as frequent^{10, 11}. The relative frequency of PETs varies in surgical or medical series, but most studies suggest a relative order of: NF-PET >insulinoma>gastrinoma >glucagonoma >VIPomas >somatostatinomas>others^{9, 11}. Four inherited disorders have an increased incidence of PETs: Multiple Endocrine Neoplasia-type 1 (MEN1), von Hippel-Lindau disease (VHL), von Recklinghausen's disease (VRD or neurofibromatosis 1[NF-1]) and tuberous sclerosis^{12, 13}. The most important is MEN1 because 80-100% of these patients develop NF-PETs, 50-60% gastrinomas, 20% insulinomas and 3-5% VIPomas or glucagonomas with the result that 20-25% of all gastrinomas and 4% of insulinomas are due to this syndrome^{12, 13}. PETs (primarily NF-PETs) develop in 10-17% of VHL patients, 0-10% of NF-1 patients (primarily duodenal somatostatinomas) and <1% of tuberous sclerosis patients (primarily NF-PETs)^{12, 13}.

Classification/Pathology

PETs are divided clinically into two groups: functional and nonfunctional (NF-PETs). Functional PETs secrete biologically active peptides causing one of nine well-established syndromes (Table 1). NF-PETs are not associated with a specific hormonal syndrome either because no peptide is secreted or the substance secreted does not cause specific symptoms. Most (>70%) NF-PETs are not truly nonfunctional because they secrete substances such as pancreatic polypeptide (PPoma), other peptides (neurotensin, ghrelin, etc), neuron-specific enolase (NSE), chromogranins or human chorionic-gonadotropin subunits, each of which does not cause specific symptoms^{9, 14}(Table 1). In addition to the well-established PET syndromes (Table 1), small numbers of patients are described with PETs producing other biologically active substances and new syndromes have been proposed, although in most cases too few patients have been described to clearly establish this point or its spectrum. GI tumors have been described secreting luteinizing-hormone causing masculinization¹⁵, secreting renin causing erythrocytosis¹⁶ and secreting PYY causing constipation (primarily ovarian tumors)¹⁷.

Table 1.

Pancreatic endocrine tumor syndromes

Name of Tumor [syndrome]	Hormone Causing Syndrome	Signs or Symptoms	Primary Location	Malignant (%)
Gastrinoma	Gastrin	Abdominal Pain	Pancreas-60%	60-90
[Zollinger-Ellison syndrome]		Diarrhea	Duodenum-30%
		Esophageal symptoms	Other-10%
Insulinoma	Insulin	Hypoglycemic symptoms	Pancreas- 99-100%	5-15
Glucagonoma	Glucagon	Rash, anemia	Pancreas- 99-100%	60
		Diabetes/glucose intolerance
		Weight loss
		Thromboembolic disease
VIPoma	Vasoactive intestinal peptide (VIP)	Severe watery diarrhea	Pancreas-90%	80
[Verner-Morrison, Pancreatic cholera WDHA]		Hypokalemia	Other-10% (neural, adrenal, periganglionic tissue)
Somatostatinoma	Somatostatin	Diabetes mellitus	Pancreas-56%	60
		Cholelithiases	Duodenum/jejunum-44%
		Diarrhea
		Steatorrhea
GRFoma	Growth hormone releasing factor (GRF)	Acromegaly	Pancreas-30%	30
			Lung-54%
			Jejunum-7%
			Other-13% (adrenal, foregut, retroperitoneum)
ACTHoma [Cushing's syndrome]	Adrenocorticotropic hormone (ACTH)	Cushing's syndrome	Pancreas - 4-16% all ectopic Cushing's	>90
PET causing the carcinoid syndrome [Carcinoid syndrome]	Serotonin s tachykinins prostaglandins	Diarrhea Flushing	Pancreas-100%	68-88
PET causing hypercalcemia	PTH-RP	Symptoms due to increased calcium	Pancreas-100%	80-90
Nonfunctioning [PPoma, Nonfunctional][PP,CgAAbdomi	None al] [PP, CgA NSE, etc *]	Weight loss, hepatomegaly Abdominal mass Occasionally asymptomatic	Pancreas-100%	60-90

^*but no symptoms due to product hypersecretion; other peptides not causing symptoms include ghrelin, neurotensin, calcitonin, subunits of human chorionic gonadotropin, etc

PETs share pathological features with carcinoids; both are considered to arise from the diffuse neuroendocrine cell system, uncommonly demonstrate mitotic figures, commonly demonstrate electron-dense granules containing various peptides, chromogranins, NSE and synaptophysin, and they many similarities in biological behavior^{14, 18}. The latter properties particularly the presence of chromogranin are widely used to identify GI NETs^{14, 18}. Both functional and NF-PETs frequently (>50%) synthesize more than one peptide^{14, 18}. However, in most cases, these multiple peptides are not associated with specific syndromes. For this reason the diagnosis of a functional syndrome (Table 1) depends not on immunocytochemistry, but is diagnosed clinically^{9, 14, 18}.

A recent standard WHO classification has proposed GI NETs be assigned to one of three categories (well-differentiated tumor, well-differentiated carcinoma, and poorly differentiated carcinoma) based on histology, size and proliferative indices¹⁴. In general histological classifications of PETs have failed to predict growth patterns for a given tumor. However this classification will allow more standardized comparison of results of different studies. For the first time a TNM classification for PETs has also been proposed¹⁹ which is based on the WHO classification of GI NETs and which may provide a more standardized assessment of patients and have important prognostic clinical value.

Molecular pathogenesis

Little is known about the molecular pathogenesis of PETs^{1, 2, 8}. This has occurred in part because alterations in common oncogenes (fos, jun, myc, k-ras, etc) or common tumor suppressor genes (p53, retinoblastoma, etc) are not generally implicated in their pathogenesis^{1, 2, 8}. Some of the most important insights have come from studies of inherited PET syndromes^{1, 2, 8, 20}. Altered genes causing these syndromes are important in some cases of sporadic PETs (i.e. nonfamilial cases)^{1, 2, 8, 20}. MEN1 is caused by mutations in the MEN1 gene on chromosome 11q13 which encodes for a 610 amino acid protein, menin, a nuclear protein that bind to numerous transcription factors^{13, 21, 22}. However, the exact mechanism leading to development of PETs still remains unclear. Sporadic PETs show an acquired loss of heterozygosity (LOH) at this locus in 20-90% and 27-39% have a mutation^{1, 2, 8, 23, 24}. In addition, recent studies show alterations in the p16/MTS1 tumor suppressor gene, the DP64/SMAD4 gene, amplification of the HER-2/neu proto-oncogene, and loss of an unknown tumor suppressor gene on chromosome 1 or 3p could also be important in the molecular pathogenesis of PETs^{1, 2, 8, 20}. Genome-wide allelotyping and comparative genomic hybridization demonstrate that chromosomal gains (especially 7q,17q,17p,20q) and losses (especially 1p,3p,3,6p,22q) frequently occur in PETs and carcinoids, however their frequency varies markedly in these two GI NETs, providing evidence that they have a different pathogenesis^{1, 2, 8, 20, 25}. Gene expression-profiling using microarray analysis has recently identified in PETs numerous additional altered genes^26-29. In comparison to normal islets in one study²⁹, 66 genes were over-expressed [particularly genes for some growth factors (IFGFBP3), cell migration/adhesion molecules (fibronectin) and putative oncogenes (MLLT 10/AF10)] and 119 under-expressed [particularly genes involved in cell cycle regulation (p21^cip1), transcription factors (JunD) and a putative metastasis suppressor gene (NME3)]. In a second study²⁶, when gene expression patterns in NF-PETs were compared to normal islets and three neuroendocrine tumor cell lines, 667 genes were up-regulated (particularly SERPINA10, BIN1, LCK, BST2) and 323 down-regulated. At present a clear concordance amongst studies is still lacking, but this approach is leading to the identification of numerous new candidate genes that may prove important in the pathogenesis of PETs or in determining growth behavior, which may have prognostic implications.

Tumor biology, prognosis and tumor markers

PETs differ in their malignant potential and location (Table 1). Some PETs (insulinomas, glucagonomas,VIPomas in adults) are found almost entirely within the pancreas, whereas others, although still referred to as PETs, are actually extrapancreatic [duodenal gastrinomas (60-80%)^30-32, small intestinal somatostatinomas (40-50%), GRFomas primarily in the lung (>70%)] (Table 1). Insulinomas are malignant in 5-15%, whereas the other PETs are malignant in 50-90%, with metastases usually developing initially in regional lymph nodes, later in the liver and subsequently in distant sites such as bone^{6, 8, 14, 30, 33, 34}. PETs in different patients may show different growth patterns^{33, 35-38}. For example, in patients with gastrinomas, 75% demonstrate no growth/indolent growth whereas 25% demonstrate aggressive growth^{35, 36}. Furthermore, even in patients with liver metastases, aggressive growth occurred in less than one-half of patients³⁷. Therefore, identification of prognostic factors is particularly important in patients with PETs³³. In almost every study, the presence or development of liver metastases, but not lymph node metastases, is a very important prognostic factor^{11, 33, 35, 36, 38-41}. In one study³⁵ the 15-year survival in patients with liver metastases was 26%, whereas without liver metastases it was 96%. The extent or rate of growth of liver metastases, presence of bone metastases, primary tumor size or location (duodenal vs. pancreatic gastrinomas), development of ectopic Cushing's syndrome, various histological features, high tumor marker levels, various flow cytometric features and high proliferative indices (Ki₆₇, mitotic index) are important prognostic factors^{11, 19, 33, 35-38, 42-44}. Survival is related to PET extent such that patients with primary tumors so small they are not found at surgery or with complete resections have survivals of 90-100%, those with incomplete resections 15-75%, and those with diffuse unresectable liver metastases 25-50%^{33, 35, 45-47}. In some studies⁴⁸ but not others^{11, 43, 49} patients with functional PETs have better survivals than those with NF-PETs. Recently, two studies^{50, 51} demonstrated for the first time that complete resection of the primary PET decreases the rate of development of liver metastases and/or improves survival⁵⁰.

In addition to the specific hormone released by a functional PET (Table 1), other putative tumor markers have been proposed which could be useful for diagnosis/prognosis. This is particularly the case for NF-PETs. The marker most widely used is plasma chromogranin A (CgA) (elevated in 88-100%), although also proposed is plasma NSE (elevated in 83-100%), PP, pancreastatin and α or β subunits of human chorionic gonadotropin (elevated in 25-40%)^52-54. Chromogranins (A,B,C) are acidic soluble proteins (MW-49kDa) found in large secretory granules of neuroendocrine cells and assessment of CgA level is now being increasingly used to diagnosis and monitor changes in NF-PETs, carcinoids and other PETs^{52, 54-56}. CgA has an overall diagnostic sensitivity of 60-100% in patients with metastatic disease, but <50% in patients with localized/early disease^56-58. CgA levels reflect tumor burden and it has been used to assess recurrences, tumor growth and changes in tumor size^{52, 55, 58}.

Clinical Features and Diagnosis of PETs

Gastrinoma-clinical features/diagnosis

Gastrinomas secrete gastrin which causes hyperchlorhydria, thereby producing the Zollinger-Ellison syndrome(ZES)^{31, 45, 59, 60}. With a long mean delay (6.1 years) in presentation/diagnosis^{45, 61, 62}, patients generally present with acid-peptic conditions including complicated and uncomplicated ulcers and/or GERD [Table 2 (Top)]. Occasionally other manifestations such as diarrhea, malabsorption or in MEN1/ZES patients, various other endocrine features predominate [Table 2(Top)]^{61, 63, 64}.

Table 2.

Presenting Features of ZES (recent series) and Causes of hypergastrinemia

Data are from60,61,74,177,297-299

Presenting Features of ZES (recent series)

Abdominal pain (75-100%)

Diarrhea (35-73%) (isolated in up to 35%)

Pain and diarrhea (55-60%)

Heartburn (44-64%

Duodenal (and prepyloric) ulcers (71-91%)

Ulcer complications [bleeding (1-17%), perforation (0-5%) or obstruction (0-5%)]

With MEN1 (22-24%)

Causes of hypergastrinemia

Appropriate

Antisecretory therapy (especially proton pump inhibitors)

Atrophic gastritis (autoimmune pernicious anemia)

H. pylori pangastritis with atrophy

Vagotomy

Fundectomy

Chronic renal failure

Inappropriate

Zollinger-Ellison syndrome

Retained antrum syndrome

Antral predominant H. pylori infection (antral G-cell hyperfunction)

Chronic renal failure

Gastric outlet obstruction

Massive intestinal resection

In constrast to the normal circumstance⁶⁵, with gastrinomas, the tumor secretion of gastrin is not physiologically regulated and sustained inappropriate hypergastrinemia occurs.

Basal acid hypersecretion (present in >90% of patients) or after stimulation⁵⁹, is a consequence of the inappropriate hypergastrinemia. Because a fasting serum gastrin (FSG) level is often the initial determination done in the United States in patients suspected of having ZES, it is important to remember that elevated levels can also be due to an appropriate physiological response to hypo/achlorhydria or an inappropriate response in other disease states [Table 2(Bottom)]. With the dramatic increase in proton pump inhibitor (PPI) use in the population, a recent study raises concern⁶⁶ about the impact this is having on the diagnosis/presentation of ZES (Fig. 1). This study⁶⁶ reported a 49% decrease in referrals of patients with possible ZES to two centers in the US and Italy since the widespread use of PPIs, a 40% decrease in the number of patients with ZES diagnosed (Fig. 1) and a 3-fold increase in the number of false positive diagnosis of ZES. This occurred because PPIs can control the symptoms of acid hypersecretion in almost all ZES patients, in contrast to conventional doses of H₂ blockers, and thus mask the diagnosis. The increased false positive rate occurred because treatment with PPIs in non-ZES patients can cause hypergastrinemia to a level seen in 60% of ZES patients^{31, 53, 67, 68}. This delay in diagnosis may lead to more patients with ZES presenting with advanced disease⁶⁶. Diagnosis of ZES requires a typical clinical syndrome together with the demonstration of inappropriate hypergastrinemia^{31, 45, 53, 59, 67, 69}. Fasting hypergastrinemia occurs in 97-99% of patients so this is usually the initial study raising suspicion of the disease^{31, 67}. No absolute level of elevation of FSG alone is diagnostic^{31, 53, 67, 68}. In the 40% of ZES patients with a FSG level >10-fold elevated, the diagnosis can be made with certainty (after excluding retained gastric-antrum syndrome by history) if the gastric pH is <2^{59, 67, 70}. In the 60% of patients with a FSG <10-fold elevated and a gastric pH <2, assessment of BAO and a secretin test should be performed. A BAO>15 mEq/hr with an elevated FSG in the absence of antisecretory therapy and a positive secretin test firmly establishes ZES. A recent study shows that the best criterion for a positive secretin test for ZES is an increase in FSG after subcutaneous secretin injection (0.4ug/kg) of >120 pg/ml above baseline producing a sensitivity of 94% and specificity of 100% (a significantly improved accuracy over the older criterion of >200 pg/ml increase)^{71, 72}. It is important to remember that hypo/achlorhydria can cause a false-positive secretin test. Because of this PPIs need to be stopped to adequately assess for the presence of ZES and because of their long duration of action they generally need to be stopped for at least one week. PPI withdrawal should be done with care by a group familiar with establishing the diagnosis of ZES because abrupt withdrawal in patients with ZES can potentially lead to serious consequences. The diagnosis of ZES in MEN1 can be complicated by the fact that successful treatment of the hyperparathyroidism, which is almost invariably present at the time of the presentation of ZES⁶⁴, can decrease FSG, acid secretion and reverse a previously positive secretin test, thereby masking the disease^73-75.

Figure 1.

Effect of widespread use of PPIs on diagnosis and referral of ZES patients in two centers (Italian-La Sapienza University [Rome, Italy] and NIH [Bethesda, Maryland]). The left panel shows the annual number of referrals of new cases before and after the widespread use of PPIs. The right panel shows the result for diagnosis of ZES at the NIH center. (Modified from⁶⁶)

Insulinoma-clinical features/diagnosis

Insulinomas ectopically secrete insulin resulting in inappropriate hyperinsulinemia which causes hypoglycemic episodes characterized by neuroglycopenic symptoms and sympathetic overdrive [Table 3(Top)]. Symptoms classically develop during periods of relative substrate deficiency (fasting or exercise)^{76, 77}.

Table 3.

Features of the insulinoma and glucagonoma syndromes

Data are from9,47,80-82,76,300-303

Features of the insulinoma syndrome

Neuroglycopenia (90%)

Amnesia or coma (47%)

Confusion (80%)

Visual changes (59%)

Convulsions (17%)

Altered consciousness (38%)

Sympathetic overdrive (60-70%)

Weakness (56%)

Sweating (69%)

Tremors (24%)

Palpitations (12%)

Hyperphagia (14%)

Obesity (<50%)

Features of the glucagonoma syndrome

Migratory necrolytic erythema (70- 90%)

Weight loss (80%)

Glucose intolerance (40-90%)

Normochromic, normocytic anemia (35-90%)

Hypoaminoacidemia (80%)

Diarrhea (25%)

Thromoboembolism (15-25%)

Glossitis, chelitis (15-40%)

Psychiatric symptoms (0-17%)

Similar to ZES, there is a delay in diagnosis (mean 4 yrs)⁷⁶. Elevated serum insulin levels may be appropriate (a consequence of elevated blood glucose levels such as in type 2 diabetes mellitus) or inappropriate (with insulinomas, nesidioblastosis [MEN1-associated or post-bariatric surgery] or exogenous insulin administration. A serum glucose level <2.5 mmol/l (45 mg/dL) with an insulin level >6 uU/ml (43pmol/L by radioimmunoassay [RIA], ≥3 uU/ml by immunochemoluminescent assay [ICMA]) combined with an elevated C-peptide level (≥200 pmol/L) and the absence of sulfonylurea in the plasma, establishes the diagnosis⁷⁶. The gold standard for establishing the diagnosis of insulinoma remains the 72 hour fast⁷⁶. One-third of patients will develop symptoms within 12 hrs, 80% at 24 hrs, 90% at 48 hrs and 100% at 72 hrs⁷⁶. Insulin levels are being increasingly determined using ICMAs or insulin-specific IRMAs that have no cross-reactivity with proinsulin and give lower values, resulting in up to 60 % of patients with insulinomas having plasma insulin levels <6uU/mL^{78, 79}. In one recent study using these specific assays the most sensitive criterion for diagnosing insulinoma was the combination of an elevated proinsulin level with a fasting glucose <45mg/dL⁷⁹.

Glucagonoma-clinical features/diagnosis

Glucagonomas ectopically secrete glucagon resulting in hyperglucagonemia. Glucagonomas cause glucose intolerance, weight loss and a pathognomonic rash called migratory necrolytic erythema (MNE) characterized by erythematous macules that develop into papules, become necrotic and heal with pigmented scarring^{9, 47, 80-82} (Table 3,Bottom). As with gastrinomas and insulinomas, glucagonomas present with a long history of symptoms (mean delay in diagnosis of 7 yrs with reports of up to 18 years) and tumors are commonly large at presentation (mean 6 cm)^{9, 47, 80, 81}.

Despite controversy in the past regarding the specific cause of MNE, recent studies show glucagon infusions can lead directly to MNE^83-85. However, MNE is not specific for glucagonoma occurring also in celiac disease, cirrhosis or pancreatitis^{81, 85, 86}.

Diagnosis of a glucagonoma requires demonstration of an inappropriately elevated serum glucagon level (diagnostic at levels above 500-1000 pg/ml). Lower elevations may be associated with glucagonomas, but can also be caused by cirrhosis, pancreatitis, diabetes mellitus, prolonged fasting, sepsis, burns, renal failure, familial hyperglucagonemia and acromegaly^{9, 47, 80, 81}.

VIPomas-clinical features/diagnosis

VIPomas ectopically secrete vasoactive intestinal polypeptide (VIP) leading to large volume diarrhea (90-100%) (100%>700 mL/day, 70-80%>3L/day), electrolyte disturbances [notably hypokalemia (70-100%)], dehydration (45-95%), hyperglycemia (20-50%), hypercalcemia (25-50%), hypochlorhydria (35-76%) and flushing (15-30%)^{9, 39, 87-89}. The large volume diarrhea often results in dehydration without an osmolar gap because it is secretory in nature^{9, 39, 87-90}. The diagnosis is confirmed by the presence of large volume secretory diarrhea with an elevated serum VIP level together with imaging evidence of a PET (in children the tumor commonly arises in extrapancreatic ganglioneuromas). However, even in the absence of imageable tumor, an elevated serum VIP level (>500pg/ml) in the presence of a documented secretory diarrhea is highly suggestive of VIPoma^{9, 39, 87-89}.

Somatostatinoma-clinical features/diagnosis

Somatostatinomas are somatostatin (SS)-secreting tumors primarily occurring in the duodenum or pancreas which can produce the somatostatinoma syndrome, characterized by diabetes mellitus, gallbladder disease, weight loss, diarrhea, steatorrhea and anemia^{9, 40, 91-93}. In the literature there is no general agreement on the definition of a somatostatinoma with most cases (55-89%) described as a PET with somatostatin present by immunohistochemistry, but with no associated clinical syndrome. It has been proposed that the term somatostatinoma syndrome should be reserved for cases with the specific clinical syndrome only. Duodenal somatostatinomas uncommonly produce the somatostatinoma syndrome (<20%) whereas pancreatic tumors often do (>90%)^{9, 40, 91-93}. Because of the subtle nature of the syndrome, these tumors have an even later presentation than other PETs. They can occur in association with MEN1 (0-1% of all MEN1 patients) or in up to 10% of VRD patients ¹³. The diagnosis is best confirmed by the presence of a pancreatic or duodenal mass together with an elevated serum SS level in a patient with typical symptoms and a tumor staining for SS. However, serum levels should be interpreted with caution in individuals without concomitant masses. Unfortunately, there is no reliable provocative test to confirm the presence of a somatostatinoma in individuals with typical symptoms and no observable mass.

GRFoma-clinical features/diagnosis

GRFomas ectopically secrete growth hormone-releasing factor (GRF) leading to uncontrolled pituitary release of growth hormone resulting in acromegaly^{9, 94-96}. Most cases of acromegaly are due to pituitary tumors and only a small fraction (<2%) to GRFomas. At least 50% of GRFomas arise in the lung (Table 1). Important clues to the presence of a GFRoma producing acromegaly are the absence of a pituitary tumor on imaging, the concomitant presence of MEN1 or the presence of an elevated prolactin level^{9, 94-96}. GRFomas are diagnosed by the presence of an elevated GRF level (>300 pg/ml)^{9, 94-96}. There are no reliable provocative tests for the GRFomas.

Nonfunctional PETs (NF-PET) -clinical features/diagnosis

NF-PETs are not associated with a hormonal syndrome (Table 1). Because of this, they are frequently found by chance and patients generally present late in the disease course with large primaries (70% >5 cm) and advanced disease (>60% have liver metastases)^{9, 97-101}. NF-PETs produce symptoms due to tumor growth/spread [i.e., abdominal pain (40-60%), weight loss (25-50%), or jaundice (30-40%)]. In recent years, NF-PETs are increasingly being identified by chance [up to 35% of patients in one series⁹⁹] as individuals undergo imaging studies for non-specific symptoms. Asymptomatic detection results in lower rates of metastases, increased resectability and improved survival¹⁰²

A NF-PET is suggested by elevated levels of serum chromogranin A (69-100%) or PP (50-100%) or positive somatostatin-receptor scintigraphy (octreoscan) with a pancreatic mass. In the absence of a mass, other potential causes of elevated serum PP levels (e.g., old age, alcoholism, inflammatory conditions, renal failure and bowel resection) need to be considered. A confirmed diagnosis for NF-PET requires histological confirmation^{9, 97-101}.

Tumor Localization/Staging

Imaging studies are essential for the management of patients with PETs. They are needed to localize the primary as well as for staging to guide management, including surgical plans (curative resection, debulking or medical management only), to monitor tumor growth, and for follow-up after therapy^{6, 9, 103-108}.

Conventional cross-sectional imaging studies (MRI, CT, US)

Older studies evaluated various conventional imaging techniques [ultrasonography (US), computed tomographic (CT) scanning, or magnetic resonance imaging (MRI)] for localization/staging of PETs^{104, 105, 107, 109-111}. PET detection with these techniques (which may be suggestive of a PET specifically) is size-dependent with <20% of PETs <1 cm identified, 30-40% 1-3 cm in diameter and >75% of PETs >3 cm^{45, 112}. Most pancreatic VIPomas, glucagonomas and somatostatinomas are large and therefore detectable with conventional studies. However, many gastrinomas, insulinomas and duodenal somatostinomas are frequently <1 cm and will not be detected by these modalities^{104, 105, 109, 110}. For identifying patients with liver metastases, US is the least sensitive (identifies 40% of patients with metastases), whereas CT and MRI are positive in 70-80%^{104, 105, 109, 110}. Figure 2 (Top) shows liver metastases in a patient with gastrinoma by both CT and MRI scanning. As newer generations of scanners are being made available, these sensitivites may change^{105, 107}. At present both high resolution spiral CT and modern MRI are highly effective at identifying liver metastases (sensitivity of up to 94%) but somewhat less effective in identifying primary tumors (sensitivity 55-78%), because the more common functional tumors (insulinomas or gastrinomas) are often small ¹¹³.

Figure 2.

CT, MRI, EUS in patients with PETs. Panel A illustrates CT (top) and MRI(bottom) images of the abdomen in a patient with a metastatic gastrinoma. Liver metastases are indicated by arrowheads. Panel B illustrates an endoscopic ultrasound image of a pancreatic body insulinoma confirmed at subsequent surgery. The tumor is indicated by the black arrowheads.

Endoscopic ultrasonography (EUS)

While standard upper endoscopy is occasionally of value in identifying PETs which arise within the luminal GI tract (gastrinomasm somatostatinomas), EUS with fine needle aspiration (FNA) has become part of the standard armamentarium for evaluating pancreatic masses^114-118. EUS/FNA is useful to distinguish PETs (especially NF-PETs) from adenocarcinomas and also to localize tumors not imaged with conventional studies^117-120. EUS/FNA is reported to have a diagnostic accuracy of 80% for pancreatic adenocarcinoma and 46% for PETs¹¹⁷. FNA is rarely needed with functional PETs (especially insulinomas/gastrinomas) because the diagnosis is made by biochemical/functional testing. EUS is more effective at localizing intrapancreatic PETs than extrapancreatic PETs such as duodenal gastrinomas^{117, 121}. EUS plays an especially important role in localizing primary insulinomas because they are pancreatic, commonly small (<1 cm), frequently missed by conventional studies, and are frequently (>70%) negative on somatostatin receptor scanning (SRS, see below), because of low density or lack of somatostatin receptor subtypes that bind radiolabeled octreotide analogues with high affinity^{106, 122-124}. EUS is able to identify intrapancreatic primary PETs in approximately 90% of cases. Figure 2 (Bottom) shows an EUS image of an insulinoma located in the body of the pancreas.

EUS is playing an increasingly important role in patients with MEN1^{13, 125-128}. MEN1 patients have NF-PETs in 80-100% of cases histologically, although often they are small (<0.5 cm)^{13, 125-128}. EUS is able to detect PETS in MEN1 patients not seen on either SRS or conventional studies, especially in the size range from 0.4-1.1 cm, with the result that 55-100% of asymptomatic patients had NF-PETs identified^{126, 129}. The management of these small asymptomatic NF-PETs is controversial because their natural history is largely unknown¹³. However, because EUS has been shown to have excellent specificity and reproducibility for small NF-PETs (<10 mm), it has been proposed that serial EUS studies could be used to monitor growth and determine when intervention should be considered^{125-127, 130}.

Similarly in patients with VHL, PETs develop in 10-17% and they are almost invariably NF-PETs^{13, 131-135}. Their management is also controversial because these patients are almost invariably asymptomatic, especially if the PET is small (<2 cm). In various studies because no patient with a NF-PET <3 cm had hepatic metastases, it has been recommended that PETs <3 cm not be routinely resected^135-137. EUS is the most accurate method to assess PET size in these patients and could be used for serial studies similar to that proposed above in MEN1.

Angiography and selective hormone sampling

Prior to the development of functional imaging studies (see below), angiography and sampling for hormone gradients were widely used and extremely helpful in patients with PETs^138-141. Originally, selective sampling for hormonal gradients was performed by portal-venous-sampling (PVS) ^{139, 142}. This method was largely replaced by selective-arterial injection of secretin (gastrinomas) or calcium (other functional PETs) with assessment of hepatic venous hormone concentrations, because it can be performed at the time of angiography, has less complications and requires less expertise, but is similarly sensitive to PVS^{139, 140, 143, 144}. This approach can also be utilized to identify liver metastases after selective hepatic artery cannulation¹⁴¹. In recent years, with advancement in other functional tumor localization methods, the utilization of these invasive localization techniques has declined. The three remaining areas in which these studies are still used are: 1) for localizing insulinomas following a negative octreoscan/EUS, 2) for preoperative evaluation of the liver prior to debulking surgery and 3) for localizing a functional PET in MEN1 patients with multiple lesions^{140, 145}. Numerous studies have shown that intra-arterial injection of calcium with hepatic venous insulin sampling is a sensitive method of localizing insulinomas, even in imaging negative cases, being positive in 88-100 %^{139, 140, 145-149}

Functional Imaging (SRS and positron-emission tomography)

Most PETs demonstrate high densities of sst2 or sst5 receptors, two of the 5 somatostatin receptor subtypes (designated sst1-sst5) which have high affinity for the SS analogues: octreotide and lanreotide^150-153. Radiolabeled forms of these synthetic SS analogues with high affinity for sst2/sst5 receptors have proved sensitive and useful for localizing both the primary PET as well as metastases^{104, 151, 154, 155}. [¹¹¹In-DPTA-DPhe¹]-octreotide is approved in the United States. Somatostatin receptor scanning (SRS or octreoscanning) identifies 50-70% of primary PETs but < 25% of insulinomas (which have absent or lower sst2/5 densities)^{104, 122-124, 151, 154, 155}. In one prospective study, SRS was as sensitive as all conventional studies and angiography combined¹⁵⁵. SRS is particularly useful for demonstrating liver metastases with the best sensitivity of any imaging modality (almost 90%)^{104, 155-157}. The imaging results shown in Figure 3 in two patients with ZES demonstrate the greater sensitivity of SRS than conventional studies in localizing both the primary as well as metastatic disease to the liver/lymph nodes. SRS allows whole body scanning and it is therefore also useful to identify tumors beyond the liver (e.g., lungs/bone)^{34, 154, 158}. To achieve high sensitivity it is essential that single photon emission tomography (SPECT imaging) be used to isolate possible lesions from the renal background^{106, 151, 159}. Studies have shown that SRS changes the management in 24-47% of patients with PETs^160-162. Although SRS has high specificity it is important to remember that a number of normal and abnormal tissues express increased densities of sst2/5 receptors that can result in false-positive scans. False positives can occur particularly with thyroid disease, breast disease, lymphoma, cholangiocarcinoma, hemangiomas, sites of inflammation and granulomatous disease^{151, 153, 161}. In one prospective study ¹⁶¹, 12% of SRSs were false-positive for a PET, however when results were interpreted in the clinical context, the false positive rate was only 3%. Detection of PETs by SRS is also size-dependent with appropriately 50% of gastrinomas <1 cm in diameter not detected¹⁶³. Therefore, there is a need for even more sensitive imaging methods^{154, 163}.

Figure 3.

Comparison of conventional imaging (CT, MRI) and SRS to localize a primary gastrinoma (left) or metastatic disease (right) in two patients with ZES. In the left panel the patient had negative preoperative conventional imaging studies (CT, MRI) and angiography, but SRS showed a lesion in the pancreatic head area. At surgery a 2 cm tumor was resected and the patient has remained disease-free. In the right panel neither the MRI nor CT showed recurrent disease in this patient post resection of a gastrinoma, however the fasting gastrin was elevated and the SRS showed extensive metastases in lymph nodes and the liver. Both of these results show the greater sensitivity of SRS for localizing primary PETs as well as metastatic disease.

Positron-emission tomographic scanning is receiving increasing attention for PET localization^{106, 164}. Standard substrates such as ¹⁸F-deoxyglucose (¹⁸FDG) are not useful for most PETs because of their slow glucose turnover and are only useful for the small subset with high proliferative rates and low differentiation¹⁰⁶. ¹¹C-5-hydroxytryptophan or ⁶⁸Gallium-labeled SS analogues have greater sensitivity than SRS or conventional studies^{106, 164-166} and therefore may prove to be clinically useful in the future. Particularly important for the increased use of position-emission tomographic scanning in PET patients is the ability to make ⁶⁸Gallium using a generator, similar to what is now used for ^99mTC in most nuclear medicine departments, rather than requiring a cyclotron as is the case for these other isotopes¹⁰⁶. In a recent study¹⁶⁵ involving 84 patients with various GI NETs (carcinoids, 23 PETs), positron-emission tomographic scanning using ⁶⁸Gallium-DOTA-Tyr³-octreotide had a sensitivity of 97% compared to 55% for SRS and a greater accuracy (96% vs. 58%, p<0.01) with equal specificity for the two techniques. One particular benefit of this scanning is the potential for image fusing (i.e., overlaying CT with PET images). It is likely that such scanning will play an increasing important role in the future for imaging PETs. Figure 4 demonstrates the increased sensitivity of positron-emission tomographic scanning with ¹¹C-5-HTP for detecting liver metastases compared to CT scanning in a patient with a malignant PET.

Figure 4.

Comparison of the extent of liver metastases in a patient with a malignant PET on CT scanning (top panel) and positron emission tomographic scanning (bottom panel). This patient with a malignant PET had a few liver metastases seen on CT scanning (top) and SRS (not shown) but much more extensive disease on positron emission tomographic scanning with ¹¹C-5-HTP demonstrating its greater sensitivity. (Images kindly provided by Prof. Anders Sundin, Department of Radiology, Uppsala University Hospital, Uppsala, Sweden).

Medical Management of the Hormonal Excess-State

Gastrinoma-medical management

In ZES acid hypersecretion is the most important clinical effect^{45, 62, 167, 168}. Because of their potency and long-duration of action, proton pump inhibitors (PPIs) are the agents of choice for management^{31, 45, 53, 167, 169, 170}. Histamine H₂ receptor antagonists or SS analogs are effective, but the former drug class is limited by the need for frequent, high-dose administration^{167, 169}, whereas the latter class is limited by the need for parenteral therapy.

Once or twice daily oral PPIs (i.e., omeprazole (40 mg), lansoprazole (30 mg), rabeprazole (20 mg), pantoprazole (40 mg) or esomeprazolea (40 mg) are effective in virtually all ZES patients^{167, 169, 171-175}. It is important to document control of acid output (i.e., <10mEq/hr in the last hour before the next dose of drug [intact stomachs] or < 5 mEq/hr [prior gastric resections] in patients with uncomplicated disease (i.e., no MEN1, mild GERD, and no prior Billroth 2 resection) rather than to titrate drug dosages to symptoms, since asymptomatic individuals may still have uncontrolled acid hypersecretion^{45, 167, 176}. Patients with complicated disease (i.e., MEN1, moderate-severe GERD, Billroth 2 resection) often need higher doses and are usually best treated with at least BID dosing^177-179. It is recommended that patients with uncomplicated disease be initially started on 40-60 mg of omeprazole (or equivalent) to adequately control acid output acutely¹⁸⁰, however with time the dosage can be decreased in up to 60% of the patients¹⁷⁹. Long-term follow-up of patients receiving PPI's demonstrates no tachyphylaxis and an excellent safety profile^{170-172, 181}, although drug-induced achlorhydria may lead to substrate deficiencies (vitamin B₁₂ is more of a concern than iron)^{181, 182}. Even though in animal studies long-term high dose PPI treatment can lead to the development of gastric carcinoids, there is no evidence of an increased rate of their development with chronic PPI treatment in ZES patients^{167, 183-185}. Almost every ZES patient demonstrates some degree of ECL hyperplasia^{183, 185-187} which is more severe in MEN1 patients^{7, 183, 186}. Patients with MEN1/ZES develop gastric carcinoids in 23-33 % of cases^{183, 185, 186} however the rate in patients with sporadic ZES is <1% ^{183, 185-187} and there is no evidence that PPIs alter this rate in either group.

Intermittent intravenous PPI treatment (with pantoprazole (80 mg), lansoprazole (60 mg) or esomeprazole (80 mg) given two or three times daily effectively substitutes for oral therapy for brief periods in patients who cannot take oral drug^{188, 189}. Three times daily therapy is generally recommended as this more frequent administration precludes the requirement to document effective control of acid in situations when the patients may be quite ill. There is no longer a role for gastric surgery to reduce acid output in ZES patients.

Insulinoma-medical management

Most patients (>85%) have a single small benign insulinoma^{76, 77, 190}, except for those with MEN1 where multiple tumors frequently occur¹³, and therefore they are treated surgically soon after diagnosis with an excellent cure-rate^{76, 77, 190}. However, prior to surgery and for the 5-15% (Table 1) with malignant disease, treatment for the hypoglycemia is needed. In addition to frequent small feedings the initial drug generally used is diazoxide (200-600 mg/day in divided doses), a benzothiadiazide, which directly inhibits insulin release and causes adrenergic stimulation promoting glycogenolysis^{9, 76}. Diazoxide controls hypoglycemia in 50-60% of patients and has been used effectively for >20 years^{76, 77, 190, 191}. Diazoxide frequently results in sodium/fluid retention requiring diuretics, as well as nausea and occasional hirsutism^{76, 77, 190, 191}. Long-acting SS analogues (octreotide, lanreotide) are effective in 35-50% of patients with insulinomas, however they need to be used with care, because in some cases they worsen the hypoglycemia, presumably by inhibiting counter-regulatory mechanisms^{123, 153, 190}. Therapy with other agents such as verapamil, propanolol or phenytoin has also been described though these agents are generally not first-line choices.

Other functional PET tumor syndromes-medical management

Until the availability of octreotide (see below), specific therapy for PETs included blood transfusions; insulin, zinc and amino acid transfusions for glucagonomas; replacement of volume losses and correction of acid-base disturbances for VIPomas; nutritional repletion and insulin administration for the somatostatinoma syndrome; and administration of adrenolytic agents (such as ketoconazole, aminoglutethimide, metyrapone or orthopara-DDD) or adrenalectomy for ectopic ACTH-producing tumors. However, octreotide availability has largely supplanted the need for many of these approaches.

Somatostatin (SS) is a widely distributed 14-amino acid cyclic paracrine peptide which exerts multiple inhibitory effects on secretory and motor functions^{150, 153}. Its effects are mediated by binding to one of 5-receptor subtypes designated sst1-sst5, which are all G protein-coupled receptors¹⁵⁰. SS has a short serum half life of about 2 minutes precluding its use clinically, but its synthetic analog, octreotide, with a serum half life of at least 1 hour has been used successfully to inhibit secretion from a variety of cell types including PETs, which usually exhibit high sst2 receptor densities^{153, 192, 193}.

Octreotide is approved for use in patients with acromegaly, VIPomas and the carcinoid syndrome, but it is also useful off label to lower portal pressure in patients with bleeding from esophageal varices due to portal hypertension, to control diarrhea in patients with AIDS enteropathy and short bowel syndrome, and to control hormonal syndromes in patients with other NETs¹⁵³. Octreotide is usually prescribed at doses ranging from 100-500 ug three times daily by subcutaneous injection initially but this form of administration can then be overlapped with once monthly depot injections of an even longer-acting formulation, octreotide LAR at doses of up to 30 mg/month^{7, 8, 194}. Lanreotide sustained-release or autogel is another depot somatostatin analog available in Europe¹⁹⁵.

In VIPomas, octreotide reduces serum VIP levels in >80% of patients and improves diarrhea in >75% but the response is often short-lived (<1 year) without dose increases. In glucagonomas, octreotide decreases plasma glucagon levels in >80% and improves MNE in 90% (with complete resolution in 30%). There are anecdotal reports of efficacy of octreotide in somatostatinoma syndrome as well as therapy for GRFomas^{7-9, 153}. Octreotide therapy is not recommended for hormonal control of gastrinoma. Octreotide should be used with care in patients with insulinomas (as discussed above). The mean duration of octreotide treatment in studies is one year and frequently tachyphylaxis develops which may be overcome with higher doses⁸

Adverse effects of SS analogs are generally mild and include diarrhea/steatorrhea, flatulence, fluid retention, nausea, gallstones and glucose intolerance. Such side-effects are reported in 50% of patients treated with octreotide, but have rarely been serious enough to stop treatment¹⁵³. In long-term treatment of patients with acromegaly only 5% developed side-effects severe enough to stop treatment^{194, 196}. During long-term treatment concern has been raised about the possibility of an increased rate of gallstone evelopment. This has been particularly well-studied in patients with acromegaly with a mean incidence of 29%, however only 1% develop symptomatic gallbladder disease¹⁹⁴.

Surgical therapy for cure

Surgery is the only treatment-modality with the potential to cure patients with PETs. However, surgery is only likely to be effective in patients without diffuse metastatic disease who are able to tolerate the intervention and, in the case of ZES specifically, only in those with sporadic disease^{13, 113, 197-199}. Negative preoperative localization should not be considered a contraindication to surgery in patients with proven functional PETs as an experienced PET surgeon will very frequently localize the tumor (>95% of insulinomas or gastrinomas)^{76, 113, 198, 200}. On the other hand, preoperative identification of diffuse disease beyond regional lymph nodes precludes attempts at curative surgery, though many authorities favor debulking surgery in cases where ≥90% of identifiable disease is thought resectable (see below). In the 5-15% of patients with limited hepatic metastases, many authorities attempt resection because this approach may result in extended disease free-survival in selected patients^{46, 201-204}. Patients with MEN1 develop potentially curable PETs of various types (insulinomas, VIPomas, somatostatinomas, glucagonomas, GRFomas)^{13, 205-210}, however both the NF-PETs and gastrinomas are invariably multiple arising throughout the pancreas or the proximal duodenum^{30, 127, 211, 212}. At present, most authorities do not recommend subjecting patients with MEN1/ZES to a Whipple's resection or patients with multiple NF-PETs to total pancreatectomy, because these operations are extensive, the long-term consequences are unclear, post-operative morbidity can be significant and the long-term prognosis of these patients without such treatment remains excellent^{121, 127, 198, 206, 209, 213}. In MEN1 patients the surgical treatment of NF-PETs (80-100% of patients) and gastrinomas (40-60% of patients), remains controversial because of multiplicity of primary tumors and failure of enucleation to result in cure^{121, 127, 198, 206, 209, 213}. Potential approaches in these patients include not performing routine surgery, performing surgery with aggressive removal of all larger PETs or only operating in patients with imageable tumors >2cm^{121, 127, 198, 206, 209, 213, 214}. This latter approach stems from a number of studies which demonstrated that patients with MEN1 and NF-PETs or gastrinomas <2cm in diameter have an excellent prognosis (survival equal to patients without PETs or 100% at 15 years) and they rarely develop advanced disease^{127, 197, 198, 206, 215}.

In advance of surgery patients should be vaccinated against encapsulated microorganisms (pneumococcus, H. influenza, meningococcus) in anticipation of a possible splenectomy and they should receive a bowel preparation in anticipation of an expected enterotomy (mandatory in the case of gastrinomas and other hormonal syndromes with a predilection for duodenal primaries)^{198, 216-219}. In general, all PETs (except imaged insulinomas) should be approached by laparotomy to permit an extensive exploration of the entire abdomen^{113, 203, 219-221}. An exception to this rule is surgery for insulinoma in non-MEN1 individuals, because at least 85% of these tumors are benign, there usually is a single primary and if they can be localized preoperatively, laparoscopic resection is successful in 70-100% of cases and its use hastens postoperative recovery^{121, 222-224}. It is also important to examine the entire pancreas which requires complete mobilization of the duodenum and exposure of the pancreatic tail^{32, 198, 216-219}. Surgical exploration is assisted by intraoperative ultrasonography using appropriate transducers for evaluation of the liver (5 MHz) and pancreas (7.5-10MHz). Intraoperative endoscopic transillumination plus duodenotomy is required for tumors with a predilection for the duodenum (GRFomas, somatostatinomas, and especially gastrinomas), because they are frequently small (<0.5 cm), not detected by ultrasound or palapation and are primarily localized in the 1^st and 2nd part of the duodenum^{113, 198, 216-220, 225-227}. Some authorities favor intraoperative hormonal localization as well²²⁸.

The aims of surgical resection for cure are to remove the primary tumor and regional lymph nodes (if affected) with minimal disruption to the underlying anatomy. Enucleation is advised for insulinomas because they are generally benign as well as for localized tumors of the pancreatic head. Duodenal tumors are generally resected unless small and then may be removed endoscopically in some cases, conversely if they are large they may require a duodenectomy^{30, 229}. Tumors in the pancreatic tail are generally resected (with splenic preservation if possible) as opposed to enucleated unless they are insulinomas^{113, 198, 216-220}. MEN1 patients who come to surgery should have a careful exploration of the entire pancreas with enucleation or resection of all dominant masses, realizing that the largest lesion identified may not necessarily be the lesion causing the functional syndrome. In general, blind pancreatectomy in the rare case of no identifiable tumor after a careful exploration of the entire abdomen is not felt to be an acceptable approach.

In appropriate hands, cure rates for insulinomas approach 100%^{76, 230}. For sporadic gastrinomas the figure is 60% immediately postoperatively and 30-40% at five years^{198, 216}. In general, cure rates for other PETs are lower because they are generally larger at presentation, often with metastases. Surgical resection of the primary PET should be attempted whenever possible if the patient does not have another medical problem limiting life-expectancy, substantially increasing surgical risk or diffuse metastatic disease, because studies in patients with ZES show resection of the primary both decreases the rate of development of liver metastases and extends survival by preventing the development of progressive disease^{50, 51}.

Treatment of metastatic disease

General Treatment of metastatic disease

In recent studies the long term outcome in patients with PETs is increasingly dependent on tumor growth. However, even with widespread liver metastases many patients remain relatively well with slow progression, especially early on in the disease process, such that many authorities advocate delaying the introduction of disease modifying agents until there is clear development of enlarging tumor burden or symptoms develop. Furthermore, standard antitumor therapies are not curative and frequently have limited efficacies.

Biotherapy

1. Octreotide/Interferon

Biotherapy with long-acting somatostatin (SS) analogs [octreotide LAR or lanreotide SR (autogel)] is frequently instituted first in patients with enlarging tumor burdens, especially patients with slow-growing tumors without extensive (<50%) liver involvement^{8, 231}. This approach is commonly used even though the results are controversial and there are no studies that have clearly demonstrated it prolongs survival due to inhibition of tumor-related growth^{103, 232}. SS analogues are frequently used first because these agents are well-tolerated and numerous studies suggest they have a tumoristatic effect, causing a decrease or cessation of growth in 30-80% of cases, without tumor regression in most cases (<15%) that demonstrated growth prior to treatment^{103, 206, 232-235}. It is presumed this tumoristatic effect will result in improved survival, but at present this remains unproven. The tumoristatic effect can be prolonged (>2 years) and is more frequently seen in slow-growing tumors with a low proliferative index; therefore some recommend that rapidly growing tumors or tumors with high proliferative indices be treated with other modalities^{8, 206, 231, 233, 236}. The exact mechanism of SS analogue action in PETs is not completely clear, however, they induce apoptosis and in various cells activate phosphatases, suppress release of growth factors, inhibit IGF-1 signaling, have immuno-modulatory effects and inhibit angiogenesis¹⁵⁰

Interferon therapy (human leukocyte/alpha-interferon) is also frequently used for the treatment of metastatic disease but, as with octreotide, its major effect is tumor growth stablization rather than inducing regression (<20% of cases)^{8, 103, 232, 234}. Similar to SS analogues it is hoped that this tumoristatic effect will result in improved survival, but at present this is also unproven²³². The mechanism of interferon's anti-proliferative effect in PETs is not completely known, however it increases tumor expression of bcl-2 resulting in decreased cell proliferation and in other cells inhibits protein and hormone synthesis and angiogenesis and stimulates the immune system⁸. Unfortunately, interferon therapy causes frequent side-effects including flu-like symptoms (which may improve with prolonged therapy), fatigue, weight loss, lipid, thyroid and liver enzyme abnormalities and cytopenias including leucopenia which may persist and interfere with the acceptability of long-term treatment^{232, 233}.

Since both interferon and octreotide therapy are tumoristatic by different mechanisms, combination therapy was felt to have promise. Non-randomized studies were suggestive of additive effects^{232, 237}, but a recent prospective study²³⁸ showed no additivity, however a number of reservations have been raised about this study, primarily methodological issues²³⁹.

2. Peptide receptor radionuclide therapy(PRRT)

PRRT utilizes the fact that PETs almost uniformly overexpress SS receptors and internalize radiolabeled SS agonist analogues thereby facilitating the delivery of cytotoxic doses of localized radiation to the PET^{153, 233, 240-244}. Three different radiolabeled SS analogues have been developed and investigated in patients with malignant NETs including analogues labeled with ¹¹¹Indium (emits conversion and auger electrons, γ-rays), ⁹⁰Yttrium (strongly emits β-particles) and ¹⁷⁷Lutetium (emits β-particles and γ-rays)^240-244. The effect of ¹¹¹In-DPTA-octreotide was examined in two studies^{240, 245} including 52 patients with malignant progressive NETs and complete tumor regression were seen in 0%, partial regression in 0-8% and tumor stabilization in 42-81%. [⁹⁰Y-DOTA,Tyr³]-octreotide, [⁹⁰Y-DOTA]lanreotide or [⁹⁰Y-DOTA-,Tyr³]octreotate were examined in 7 studies involving >280 patients with malignant NETs and complete tumor responses occurred in 0-3%, partial responses in 6-37% % and stabilization in 44-88%^{240, 245}. In one study a longer survival was reported in patients treated with [⁹⁰Y-DOTA-,Tyr³]octreotate than those previously treated with ¹¹¹In-DPTA-octreotide (mean 37 mos vs. 12.5 mos)^{240, 246}. One study reported results with 129 patients with malignant NETs treated with [¹⁷⁷Lu-DOTA,Tyr³]octreotate and found a complete tumor response in 2%, a partial response in 32% and stabilization in 34% ^{240, 247}. To date, no controlled studies have demonstrated that PRRT extends survival. In general PRRT with the different isotopes has been safe with severe side-effects uncommon^{240, 244-246}. Approximately 30% of the patients develop acute side-effects (nausea, pain, vomiting) that are usually mild, can be controlled with symptomatic therapy and do not interfere with continued treatment²⁴⁰. More severe side-effects include hematological toxicity (15%-usually transient, 0.3% develop myelodysplastic syndrome) and renal toxicity (which occurs almost entirely in patients given ⁹⁰Y-labeled SS analogues and can be limited by co-administration with amino acids)^{240, 244, 245}. Although not yet approved for use in any country, the promising results described above have led to PRRT undergoing evaluation in a number of centers in the world to clearly establish its exact utility.

Liver-directed therapy (embolization, chemoembolization)

Most malignant PETs metastasize to the liver where they derive their blood supply from hepatic artery branches (75-80%), in contrast to native liver tissue, which derives the majority of its blood supply from the portal vein^{9, 248, 249}. Recent studies demonstrate that liver metastases demonstrate rapid growth in <50% of patients and up to 30% demonstrate no growth on follow-up^{33, 37, 250}. Consequently, the usual approach to palliative therapy for liver metastases is to delay therapy until symptoms supervene due to the metastases per se, the tumor shows rapid growth, or the patient develops refractory symptoms from a functional PET.

Selective deprivation of blood supply to metastases for the palliative management of metastatic disease can be achieved by surgical ligation, but interventional radiological approaches via intra-arterial catheterization of the iliac/brachial arteries without (hepatic artery embolization [HAE]) or with co-administration of chemotherapeutic agents (HACE) permits a similar result^{9, 248, 249, 251}. Absolute contra-indications to HAE/HACE are portal venous thrombosis, liver failure and biliary reconstruction (Whipple resection), whereas relative contra-indications are hepatic tumor loads >50%, contrast allergy, extensive extrahepatic disease and poor performance status^{249, 252}. There are no randomized studies comparing embolization alone (HAE) to those with embolization combined with chemotherapeutic agents (HACE) such as 5-fluorouracil, cisplatin, mitomycin C or streptozotocin.

The usual approach to HAE/HACE is sequential catheterization of peripheral radicals of the hepatic artery in one liver lobe followed by repeated administration of therapy on the other side about 6-8 weeks later^{248, 249, 253}. In various studies 55-100% of patients with malignant NETs treated by HAE/HACE have symptomatic improvement and 20-80% an objective response with tumor shrinkage^{9, 248, 249, 251, 253-256}. The mean duration of response is 6-42 months^{248, 254-256}. A lower response rate has been reported in patients with >75% of the liver involved and in patients with an intact primary tumor or extrahepatic metastases²⁵⁴.

HAE/HACE is not without side-effects with an overall mortality of <3%, but pain develops in 50-100%, nausea and vomiting in 50-90% and fever/leukocytosis in 30-60%. In 5-15% of patients serious side-effects can occur including hepatic failure, bleeding, gallbladder necrosis, hepatic abscess formation and renal failure^{9, 248, 253}.

At present there is no uniform agreement on when HAE/HACE should be used in patients with malignant PETs. In patients with functional PETs not responding to other therapies or malignant PETs with diffuse hepatic metastases only which are increasing in size or causing local symptoms due to tumor bulk, this procedure may be considered and may be quite helpful in controlling symptoms^{248, 251, 254}.

Surgical debulking (cytoreductive surgery)/radiofrequency ablation of hepatic metastases(RFA)

The role of cytoreductive surgery in patients with malignant PETs with incompletely resectable metastatic disease is controversial. Whereas numerous studies show surgery may help control symptoms in patients with advance metastatic functional PETs and likely prolong life expectancy in patients with malignant PETs, in most studies the patient groups are not strictly comparable and no randomized studies have examined this approach^{9, 46, 201, 257-261}. In an analysis of 63 patients with malignant PETs from five different surgical series who underwent surgical resection, the operative mortality averaged 6%, symptom control was achieved in 85% and 5-year survival was 60-80% ²⁵⁷. The authors of this review, as well as those in a number of other surgical series, concluded that surgical resection should be attempted in patients with malignant PETs whenever it is determined that at least 90% of the visible tumor could likely be removed^{201, 202, 255, 257, 259-262}. In one recent ²⁵⁵ retrospective comparison of results with cytoreduction or embolization in 120 patients with malignant NETs (33-PETs, 87-carcinoids), patients undergoing cytoreductive surgery had longer survival and greater reduction in symptoms.

RFA is being increasing used in patients with PETs with hepatic metastases either alone or in combination with other treatments^{248, 263-266}. RFA can be performed at the time of surgery for isolated hepatic metastases or laparoscopically^{248, 264, 265}. Factor limiting its application include tumor size (usually used in tumors <3.5 cm) and number (usually used in cases with<5 lesions)^{248, 264, 265}. RFA morbidity is low (<15%), although occasional cases of hemorrhage or abscess formation occur. Response rates from 80-95% are reported and responses have lasted up to 3 years^{248, 264-266}. Although RFA has not been shown to extend life, its ability to control local metastases with low morbidity has led to it being increasingly used for the treatment of limited small metastases and it may be particularly helpful for patients with limited metastases from a functional PET, especially at the time of surgery^{232, 263, 266}.

Chemotherapy

Traditional chemotherapeutic approaches

If biotherapy fails or the PET is rapidly growing or poorly-differentiated, chemotherapy is frequently employed^{249, 267, 268}. A large number of regimens have been utilized in patients with metastatic PETs with some success, in contrast to carcinoid tumors, where they have been generally unsuccessful²⁴⁹. Streptozotocin was the first agent shown to have significant benefit in a prospective study as monotherapy for malignant PETs²⁶⁹. However, this approach provided limited benefit with significant renal/hematological toxicity²⁶⁹. Combination therapy with streptozotocin and 5-fluorouracil or doxorubicin was subsequently employed to permit lower doses of streptozotocin to potentially limit side effects without sacrificing efficacy. In the 1992 Eastern Cooperative Oncology Group (ECOG) study of 105 patients who received one of three regimens (streptozotocin-doxorubicin - response rate [RR] 70%, streptozotocin-5-fluorouracil - RR 45% and chlorozotocin monotherapy - RR 30%), the streptozotocin-doxorubicin regimen was shown to improve overall survival with a mean duration of response of 18 months²⁷⁰. Later studies utilizing only imaging assessments and better imaging modalities, have not found this degree of success. In later studies utilizing streptozotocin in various combinations with 5-fluorouracil and doxorubicin, the overall survival was either not impacted at all or only minimally impacted, the response rate was 6-40% with no complete responses and the median response was short (9-18 mos)^{249, 268, 271, 272}. Particularly poor RRs were seen in patients with replacement of >75% of the liver by tumor or in those who had previous received chemotherapy²⁶⁸. Streptozotocin is associated with significant side-effects with 74-100% of patients developing nausea/vomiting, and 20-40% with long-term treatment developing renal toxicity^{249, 268-270}.Studies utilizing other chemotherapeutic agents including etoposide, DTIC (Dacarbazine) and cisplatin or carboplatin alone or in combination have in general also been rather disappointing^{9, 249, 267}. In poorly-differentiated PETs chemotherapy with cisplatin, ectoposide or its derivatives is the recommended treatment with RRs of 40-70% reported, however the RRs are relatively short^{249, 273-275}.

Angiogenesis inhibitors and other new, novel approaches

GI NETs frequently produce multiple growth factors including vascular endothelial-growth factor (VEGF), platelet-derived growth factor (PDGF), insulin-like growth factor (IGF-1), basic fibroblast growth factor (bFGF), and transforming growth factor (TGF) as well as expressing receptors for these (VEGFR, PDGFR, IGF-1R) and other growth factors(epidermal growth factor receptor [EGFR])^276-280. A number of new, novel therapies are now available that are directed at these growth factors or their receptors and are being investigated in GI NETs including a monoclonal antibody to VEGF (bevacizumab) as well as small-molecule inhibitors of the intracellular tyrosine kinase domain of VEGFR or other growth factor receptors (sunitinib [SU11248], sorafenib, vatalanib, imatinib (gleevac), gefitinib)^280-284(Fig. 5). In one study reported in abstract form²⁸⁵ sunitinib, was evaluated in a phase II study of 61 patients with PETs. The treatment was well tolerated and a response occurred in 13%, tumor stabilization in 68% and the median time to tumor progression was 33 weeks. In another phase II trial²⁸⁶ of gefitinib, a tyrosine kinase inhibitor targeting EGFR, in 31 PET patients a tumor response of only 6% was noted. Other novel approaches to the management of metastatic PETs have focused on targeting downstream targets of tyrosine kinase receptor activation (Fig 5). For example, mammalian target of rapamycin (mTor) is a threonine kinase that is involved in the regulation of cell cycle progression and its inhibition has showed promising anti-tumor activity in a number of neoplasms^{280-282, 287}. However, temsirolimus, an mTor inhibitor, when evaluated in a phase II trial of 15 patients with PETs showed a low response rate of 7% ²⁸⁷. Another mTor inhibitor, everolimus (RAD001) yielded a response rate of 15% when administered in combination with octreotide LAR in 13 patients with PETs^280-282.

Figure 5.

Schematic diagram of a theoretical pancreatic endocrine tumor cell, smooth muscle cell (pericyte) or endothelial cell demonstrating the sites and mechanism of action of novel agents for the management of metastatic PETs. These cellular components of PETs all exhibit surface growth factor receptors (e.g., VEGFR, PDGFR, IGF-1R, c-KITR, etc) which when occupied by their respective growth factors (in an autocrine or paracrine manner) lead to autophosphorylation of the intracellular tyrosine kinase component of the receptor. Tyrosine kinase phosphorylation activates the PI3K-AKT-mTOR pathway (amongst others) ultimately promoting protein synthesis, cell cycle progression and cell survival which causes increased cellular proliferation, inhibition of apoptosis, cellular invasion, metastasis and tumor angiogenesis. This pathway can be inhibitied by monoclonal antibodies to growth factor receptors, tyrosine kinase inhibitors with specific activity against various growth factor receptors, or downstream mTOR inhibitors. Whilst mTOR inhibitors are active against both the tumor directly as well as its blood supply, tyrosine kinase inhibitors or antibodies directed against specific growth factors may predominantly effect the tumor itself or secondarily inhibit tumor cell growth by altering its blood supply³⁰⁴-³⁰⁸.

Although response rates in these initial studies are low, these agents represent new approaches to treatment. It is hoped that these novel antitumor agents may play a future role alone or in combination with other agents in the management of patients with metastatic PETs.

Palliative radiotherapy

NET cells are sensitive to standard external beam irradiation. Unfortunately, liver tissue has similar sensitivity such that the therapeutic index for radiation of liver metastases is prohibitive. On the other hand, palliative radiation to bone metastases in the spine and even brain metastases has been shown to be effective^{288, 289}. Proton-beam radiation holds promise for effective palliation of many different types of cancers. To date, no information is available regarding the use of this potentially promising modality in NET patients.

Liver transplantation

In contrast to most other neoplasms, liver transplantation continues to be used for selected patients with metastatic PETs^{9, 290-293}. Conclusions about its potential value or guidelines regarding which patients would most benefit are difficult because the available literature comprises <150 patients with malignant PETs treated with liver transplantation, the individual series are small (largest single center-19 cases) and long-term follow-up data are limited²⁹⁰. In a recent report involving 15 patients with malignant GI NETs (11-PETs) the 5-year disease-free survival was 20% and total survival 90%, which is in contrast, to the results of a review ²⁹³ of 103 patients from multiple small series with NETs (including 48 PETs), which demonstrated 2-and 5-year total survival rates of 60% and 47%. Younger patients (<50 years old), patients without extensive other surgical procedures (cluster operations) and with disease limited to the liver, appeared to fare best^{290, 293}. Recent reviews suggest that liver transplantation should be considered in selected young patient with metastases limited to the liver and a previously resected primary PET who require relief from incapacitating hormonal or tumor symptoms^{290, 291, 293}.

Future directions and unsettled problems

Even though there have been many advances in recent years in the diagnosis/management of PETs, it is not clear that survival in patients with advanced disease has improved. In fact in a recent review ²⁹⁴ of survival for all gastrointestinal NETs (both carcinoids and PETs), no change in survival was reported over a 30 year period. Numerous factors contribute to this including their continued delay in diagnosis (mean-4-6 years), the lack of general availability to most patients of the expertise and experience necessary to diagnose and manage them, the lack of good prognostic factors to stage disease extent and tailor treatment accordingly, and the lack of controlled trials, new treatments and a standardized approach to care so that approaches can be compared in different centers. These problems arise not only because PETs are uncommon, but also because large gaps in our knowledge remain regarding their molecular pathogenesis and there are no widely accepted animal models or PET cell lines that can be used to evaluate innovative treatments. Furthermore, it is difficult for young physicians who may want to acquire the necessary expertise to treat patients with PETs because of a paucity of well-rounded centers that have expertise in all facets of these tumors. Furthermore, comparison of results from study to study is difficult, because of a lack of uniformity in the United States in the pathological classification of these tumors or standardization of the minimum criteria for histological diagnosis. A number of recent consensus conferences statements have been published by the European Neuroendocrine-tumor Network Society (ENETS)^{6, 7, 295} which attempt to begin to standardize the approach to diagnosis/management including, for the first time, a proposed TNM classification^{19, 296}. In addition, the National Cancer Institute recently mandated a summit conference on GI NETs and it has been proposed in another recent consensus conference that centers of excellence should be established dealing with all aspects of the diagnosis, management, and basic /clinical research needs related to PETs²⁹⁴.