**3.1 Malignant insulinomas**

Malignant insulinomas represent 5-10% of insulinomas (Danforth et al. 1984; Service et al. 1991). The clinical and biological diagnostic criteria of insulinoma do not differ from those of benign insulinomas. On the other hand, there are specific issues in malignant insulinomas, regarding the diagnosis of malignancy, and the therapeutic management.

The histological diagnosis of malignancy is difficult. The only definite basis is the presence of a metastasis.

In clinical practice, a malignant insulinoma is generally a single large (> 4 cm) tumour. In most cases there are synchronous metastases, which allow to make the diagnosis of malignancy. The metastases are generally located in lymph nodes or in the liver. Thus careful radiological evaluation is valuable for the diagnosis of malignancy. Liver metastases of malignant insulinomas may present with a specific aspect on CT-scan (Atwell et al. 2008) and MRI (Sohn et al. 2001) with perilesional steatosis, due to a local effect of insulin. However metastases may be difficult to identify on CT-scan; a combination of pre-contrast, hepatic arterial-dominant phase and portal venous phase imaging improves the sensitivity for detection of liver metastases, since the lesion may be seen on only one of the three phases (Rockall & Reznek 2007), and large lesions may become necrotic and unenhanced by contrast agents. MRI, ultrasound and contrast-enhanced ultrasound may improve the detection of liver metastases. Typical liver metastases appear as low-signal intensity on T1 and high signal intensity on T2-weighted images, and are hypervascular on hepatic arterial dominant post-gadolinium images (Bader et al. 2001; Rockall & Reznek 2007). Nuclear imaging with Octreoscan can also be used, but false negative results are observed in small liver metastases, and MRI is better than Octreoscan scintigraphy for liver metastases detection (Dromain et al. 2005). However, Octreoscan uptake provides a tool to select patients for treatment with radio-labelled somatostatin analogues (Reubi et al. 2005). We did not confirm that Octreoscan uptake was predictive of the efficacy of somatostatin analogues on hypoglycaemia (Vezzosi et al. 2008; Maiza et al. 2011), unlike what had been previously shown in secreting endocrine tumours (Lamberts et al. 1990).

More rarely, the diagnosis of malignancy is suspected on the basis of the histo-pathological findings and it is confirmed later by occurrence of metastases. Apart from metastases and invasion of adjacent organs, suspicion of aggressive tumour behaviour is aroused by a tumour size > 2 cm, angioinvasion, and high proliferative activity (Kloppel 2007; Kloppel et al. 2009). A review of classification systems has been made by Klimstra et al (Klimstra et al. 2010).

The diagnosis of malignancy may be more difficult in some patients when the insulinoma has a benign presentation, even on the histo-pathological findings; in 48 operated insulinoma patients, with a median follow-up of 42 months, 2 patients initially diagnosed to have benign insulinomas were finally found to develop metastases (personal data). Therefore a prolonged follow-up is advisable. Since some patients with previous partial pancreatectomy may not fulfil the typical criteria for the diagnosis of insulinomas, recurrence of even mild clinical symptoms must not be neglected. Tumours of "uncertain behaviour" warrant a long-term clinical, biological and radiological follow-up. The follow-up of benign insulinomas after surgical remission is usually thought to be unnecessary (Arnold et al. 2009) but on the basis of our experience, we would recommend a clinical and biological follow-up and at least a radiological evaluation 1-2 years after surgery.

Regarding the treatment, malignant insulinomas can represent a double therapeutic challenge, i.e. that of the control of tumour progression, and the control of symptomatic

Malignant insulinomas represent 5-10% of insulinomas (Danforth et al. 1984; Service et al. 1991). The clinical and biological diagnostic criteria of insulinoma do not differ from those of benign insulinomas. On the other hand, there are specific issues in malignant insulinomas,

The histological diagnosis of malignancy is difficult. The only definite basis is the presence

In clinical practice, a malignant insulinoma is generally a single large (> 4 cm) tumour. In most cases there are synchronous metastases, which allow to make the diagnosis of malignancy. The metastases are generally located in lymph nodes or in the liver. Thus careful radiological evaluation is valuable for the diagnosis of malignancy. Liver metastases of malignant insulinomas may present with a specific aspect on CT-scan (Atwell et al. 2008) and MRI (Sohn et al. 2001) with perilesional steatosis, due to a local effect of insulin. However metastases may be difficult to identify on CT-scan; a combination of pre-contrast, hepatic arterial-dominant phase and portal venous phase imaging improves the sensitivity for detection of liver metastases, since the lesion may be seen on only one of the three phases (Rockall & Reznek 2007), and large lesions may become necrotic and unenhanced by contrast agents. MRI, ultrasound and contrast-enhanced ultrasound may improve the detection of liver metastases. Typical liver metastases appear as low-signal intensity on T1 and high signal intensity on T2-weighted images, and are hypervascular on hepatic arterial dominant post-gadolinium images (Bader et al. 2001; Rockall & Reznek 2007). Nuclear imaging with Octreoscan can also be used, but false negative results are observed in small liver metastases, and MRI is better than Octreoscan scintigraphy for liver metastases detection (Dromain et al. 2005). However, Octreoscan uptake provides a tool to select patients for treatment with radio-labelled somatostatin analogues (Reubi et al. 2005). We did not confirm that Octreoscan uptake was predictive of the efficacy of somatostatin analogues on hypoglycaemia (Vezzosi et al. 2008; Maiza et al. 2011), unlike what had been previously

More rarely, the diagnosis of malignancy is suspected on the basis of the histo-pathological findings and it is confirmed later by occurrence of metastases. Apart from metastases and invasion of adjacent organs, suspicion of aggressive tumour behaviour is aroused by a tumour size > 2 cm, angioinvasion, and high proliferative activity (Kloppel 2007; Kloppel et al. 2009). A

The diagnosis of malignancy may be more difficult in some patients when the insulinoma has a benign presentation, even on the histo-pathological findings; in 48 operated insulinoma patients, with a median follow-up of 42 months, 2 patients initially diagnosed to have benign insulinomas were finally found to develop metastases (personal data). Therefore a prolonged follow-up is advisable. Since some patients with previous partial pancreatectomy may not fulfil the typical criteria for the diagnosis of insulinomas, recurrence of even mild clinical symptoms must not be neglected. Tumours of "uncertain behaviour" warrant a long-term clinical, biological and radiological follow-up. The follow-up of benign insulinomas after surgical remission is usually thought to be unnecessary (Arnold et al. 2009) but on the basis of our experience, we would recommend a clinical and biological follow-up and at least a

Regarding the treatment, malignant insulinomas can represent a double therapeutic challenge, i.e. that of the control of tumour progression, and the control of symptomatic

review of classification systems has been made by Klimstra et al (Klimstra et al. 2010).

regarding the diagnosis of malignancy, and the therapeutic management.

shown in secreting endocrine tumours (Lamberts et al. 1990).

radiological evaluation 1-2 years after surgery.

**3. Particular conditions 3.1 Malignant insulinomas** 

of a metastasis.

hypoglycaemia. Controlling the hypoglycaemia must not be neglected, since uncontrolled hypoglycaemia results in increased morbidity and mortality.

Regarding the tumour process, 10 year-survival is about 30% in patients with metastatic insulinomas (Service et al. 1976), with very heterogeneous courses. Some patients may present a very slow progression rate, even with a metastatic disease, and have long-term survival, whereas other patients present with rapid tumour progression and poor shortterm survival rate. Predictive factors for rapid tumour progression remain to be established. They could be similar to those found for endocrine bronchial or digestive carcinomas, which comprise the presence of extra-hepatic metastases, the number of liver metastases, the proliferative index Ki67, the spontaneous progression of tumour volume in 3-6 months (Greenberg et al. 1987; Pape et al. 2004; Lepage et al. 2007). The anti-tumour strategy will not be detailed. Since malignant insulinomas are rare, there is no specific prospective study on this particular topic, so that the therapeutic strategy is similar to that of non secreting pancreatic endocrine carcinomas. Whenever it is possible, surgery must aim at totally removing the detectable lesions. However even in selected patients, post-operative complementary treatments are necessary, due to a frequent underestimation of liver metastases. In addition, even when total removal of the tumour has been performed with a R0-resection, recurrence is frequent (about 60% after a follow-up of 3 years) and the recurrence-free median survival is 5 years (Danforth et al. 1984). Treatments other than surgery combine local or regional treatments such as intra-arterial chemotherapy or chemoembolization of liver metastases (Roche et al. 2003) and radiofrequency (Berber et al. 2002), and systemic treatments such as radionuclide systemic administration (Ong et al. 2010), cytotoxic chemotherapy (traditional combination of streptozotocin, doxorubicin and 5-fluoro-uracil and more recently, capecitabine and temozolomide) (Moertel et al. 1992; Strosberg et al. 2008; Strosberg et al. 2011), or targeted therapies (Kulke et al. 2006; Dimou et al. 2010)(Raymond et al. 2011;Yao et al. 2011).

Symptomatic treatment of insulinomas aims at achieving short-term control of the hypoglycaemia while awaiting the effects of anti-tumour treatment or when anti-tumour treatments do not prove to be effective. Surgical removal of the tumour and its metastases is valuable in order to control the hypoglycaemia. Even a reduction of the tumour volume may result in reduction or transient subsiding of symptoms of hypoglycaemia (Sarmiento et al. 2002). A medical treatment is performed if there is no possible surgical cure, or as a complementary therapeutic approach. One should choose as first-line therapies medications that can achieve short-term control of hypoglycaemic spells, and do not jeopardize (by their possible adverse effects) the following use of other treatments. A few recent studies have addressed the issue of the control of severe hypoglycaemic spells in patients with inoperable malignant insulinomas (Bourcier et al. 2009; Ong et al. 2010; Maiza et al. 2011). All the medications used to treat hypoglycaemia in benign insulinomas can be employed, but in most cases, a combination of several hyperglycaemic medications is necessary (Vezzosi et al. 2008). Diazoxide or somatostatin analogues (2-3 subcutaneous injections daily, or long-acting forms, or continuous subcutaneous administration with a portable pump) may result in reduction or disappearance of hypoglycaemic spells in patients with metastatic insulinomas. Radiolabelled somatostatin analogues (especially the Lutetium labelled derivative of octreotide) may prove helpful in long-term control of hypoglycaemia and tumour progression in some patients (Ong et al. 2010). Calcium channel blockers like verapamil gave disappointing results (Stehouwer et al. 1989). Glucocorticoids were found to be effective on hypoglycaemia in malignant insulinomas, and some patients were improved even on doses as low as 2.5 mg of

Diagnosis and Treatment of Insulinomas in the Adults 161

cases there are also secreting or non secreting endocrine tumours that are more than 1 cm in size and are distributed in the whole pancreatic gland. Thus managing MEN-1-related insulinomas is often managing multiple pancreatic endocrine tumours, that may secrete insulin and other hormones (gastrinoma, glucagonoma…) or may be non secreting

The natural course of endocrine disorders in MEN-1 patients was previously thought to start almost always by occurrence of primary hyperparathyroidism (Marx et al. 1998). However others have reported cases of MEN-1 revealed by a secreting pancreatic endocrine tumour; hyperparathyroidism had not been identified prior to the diagnosis of the secreting pancreatic tumour, because it had a mild or asymptomatic presentation (Thakker 2000). Thus insulinoma was the endocrine tumour that revealed MEN-1 in 14-30% of patients (O'Riordain et al. 1994; Cougard et al. 2000). Therefore, apparently sporadic insulinomas in young patients and multiple pancreatic lesions warrant to screen for MEN-1 even if there is

The major issue is that of the treatment, since the multiple lesions increase the difficulty in achieving surgical cure of the disease. There is a mean of 3 pancreatic endocrine tumours (1 to 10) in the reported series of MEN-1 patients with pancreatic lesions who underwent surgery (Bartsch et al. 2005; Triponez et al. 2006). When managing an insulinoma patient with multiple pancreatic tumours, there could be several therapeutic strategies: 1. to perform total pancreatectomy with its associated morbidity and mortality rates; 2. to perform an extended resection of the pancreatic gland with enucleations of the tumours on the remaining pancreatic gland, which is also associated with significant morbidity and adds a risk of tumour recurrence. 3. to perform selective enucleations of the detected tumour(s), which increases the risk of re-operation for other tumours, and also may sometimes result in removing a nonsecreting tumour, and not the insulin-secreting tumour if it has not been detected because of its smaller size; this could warrant a functional study of the lesions (with preoperative transhepatic venous sampling or arterial calcium injection, or intra-operative measurement of insulin) 4.to choose a medical therapeutic approach in responders who do not seem to have malignant lesions. In an operable insulinoma patient, curative surgery is generally attempted for the following two reasons. First, the prevalence of malignant insulinomas is similar in MEN-1 patients and in patients with sporadic insulinomas (Jensen et al. 2008). The prognosis of MEN-1 depends on the progression of malignant pancreatic endocrine tumours (Jensen et al. 2008). Second, in insulinoma patients, even in a patient who is thought to be controlled by a medical treatment, the sudden occurrence of severe symptomatic hypoglycaemia cannot be excluded, with its potentially serious and even fatal consequences. Therefore there is a general agreement regarding the necessity to perform surgery in MEN-1 patients with insulinomas. The extent of surgery remains debated. It mainly depends on the number and the location of the lesions, and the possibility to enucleate the lesion. One of the major advantages of preoperative endoscopic ultrasound could be in the evaluation of MEN-1 patients, since it can visualize even small lesions in the head of the pancreas (Proye et al. 2004). Left-sided pancreatectomy with enucleation of the tumours located in the pancreatic head is generally recommended in patients with MEN-1 related insulinomas, whenever it is possible (Cougard et al. 2000). Tumor enucleation gives acceptable results if the choice of the lowest morbidity rates is made. In those patients who present with a small number of pancreatic tumours, surgery may be less extensive and single or multiple tumour enucleation can be performed, when it is technically possible. The patient must be informed about the risk of recurrence and

endocrine tumours (Cougard et al. 2000).

no hyperparathyroidism.

re-operation.

prednisone per day (Novotny et al. 2005; Starke et al. 2005). m-TOR inhibitors seem now to be very promising regarding the control of the hypoglycaemia, in addition to their antitumour effects (Bourcier et al. 2009; Kulke et al. 2009). They can suppress insulin synthesis and secretion and increase insulin-resistance, thereby resulting in euglycaemia in patients with malignant insulinomas (Di Paolo et al. 2006). Chemoembolization of liver metastases can achieve the control of the hypoglycaemia (Moscetti et al. 2000). Repeated chemoembolization procedures are often necessary. Combining the use of continuous subcutaneous administration of a high dose of octreotide as a first-line treatment, followed if necessary by chemoembolization of liver metastases could achieve control of hypoglycaemic spells in a few inoperable patients (Maiza et al. 2011). Continuous glucose infusion may be necessary in some patients, while awaiting the efficacy of the other treatments. Vitamin B1 supplementation should be systematically implemented for patients with severe recurrent hypoglycaemia requiring large infusions of glucose, in order to prevent Wernicke's encephalopathy (Grunenwald et al. 2009).

#### **3.2 Insulinomas as part of a genetic disorder**

This section will be focused on 4 genetic disorders that are known to be responsible for pancreatic endocrine tumours in adult patients: multiple endocrine neoplasia 1 (MEN-1, OMIM # 131100), Von Hippel Lindau's disease (VHL, OMIM # 193300), tuberous sclerosis (TS) (a disease that is known to involve a dysregulation of the mTOR pathway) (TSC1 and TSC2 genes, OMIM # 191100 and # 191092), and neurofibromatosis type 1 (NF1, OMIM # 162200). Of these 4 disorders, MEN-1 is the most frequent cause of pancreatic endocrine tumours, particularly insulinomas, followed by order of decreasing frequency, by VHL, NF1 and TS (Jensen et al. 2008). It can be noticed that though pancreatic endocrine tumours are found in VHL patients, to our knowledge, there is no case report of a VHL patient with insulinoma. These 4 genetic disorders are caused by mutations in tumour-suppressor genes and are transmitted as autosomal dominant traits. Genetically-determined insulinoma does not differ from sporadic insulinoma regarding the clinical and biological diagnostic criteria. On the other hand, the therapeutic strategy may be different, particularly in MEN-1 insulinomas, since insulinomas are often multiple or associated with other secreting or non secreting endocrine tumours.

MEN-1 is related to a mutation in the menin gene, which maps to chromosome 11q13. Its prevalence is 1-10/100,000. The functional physiological role of menin has not yet been well established. Menin interacts with numerous proteins that play a role in transcriptional regulation, genomic stability, cell division and control of cell cycle (Balogh et al. 2006; Yang & Hua 2007). The major MEN-1 related endocrine disorders are pituitary adenomas, parathyroid adenomas or parathyroid hyperplasia, and secreting or non secreting pancreatic endocrine tumours. Regarding the secreting pancreatic endocrine tumours, the most frequent are gastrinomas (54%, 20-61% of patients) and insulinomas (18%, 7-31% of patients) (Jensen et al. 2008). MEN-1 related insulinomas occur generally in younger patients than sporadic insulinomas, with a greater prevalence in patients aged 25-35 years, but the age of occurrence is very different from patient to patient (5-80 years) (Cougard et al. 2000; Machens et al. 2007). Cystic pancreatic endocrine neoplasms could be more frequent in MEN-1 than in sporadic cases (Bordeianou et al. 2008).

The major characteristic of MEN-1 related insulinomas is that they are generally multiple tumours, so that diffuse microadenomatosis can be observed in the entire pancreatic gland, with multiple infracentimetric lesions of the pancreas (Anlauf et al. 2006). Often in such

prednisone per day (Novotny et al. 2005; Starke et al. 2005). m-TOR inhibitors seem now to be very promising regarding the control of the hypoglycaemia, in addition to their antitumour effects (Bourcier et al. 2009; Kulke et al. 2009). They can suppress insulin synthesis and secretion and increase insulin-resistance, thereby resulting in euglycaemia in patients with malignant insulinomas (Di Paolo et al. 2006). Chemoembolization of liver metastases can achieve the control of the hypoglycaemia (Moscetti et al. 2000). Repeated chemoembolization procedures are often necessary. Combining the use of continuous subcutaneous administration of a high dose of octreotide as a first-line treatment, followed if necessary by chemoembolization of liver metastases could achieve control of hypoglycaemic spells in a few inoperable patients (Maiza et al. 2011). Continuous glucose infusion may be necessary in some patients, while awaiting the efficacy of the other treatments. Vitamin B1 supplementation should be systematically implemented for patients with severe recurrent hypoglycaemia requiring large infusions of glucose, in order to prevent Wernicke's encephalopathy

This section will be focused on 4 genetic disorders that are known to be responsible for pancreatic endocrine tumours in adult patients: multiple endocrine neoplasia 1 (MEN-1, OMIM # 131100), Von Hippel Lindau's disease (VHL, OMIM # 193300), tuberous sclerosis (TS) (a disease that is known to involve a dysregulation of the mTOR pathway) (TSC1 and TSC2 genes, OMIM # 191100 and # 191092), and neurofibromatosis type 1 (NF1, OMIM # 162200). Of these 4 disorders, MEN-1 is the most frequent cause of pancreatic endocrine tumours, particularly insulinomas, followed by order of decreasing frequency, by VHL, NF1 and TS (Jensen et al. 2008). It can be noticed that though pancreatic endocrine tumours are found in VHL patients, to our knowledge, there is no case report of a VHL patient with insulinoma. These 4 genetic disorders are caused by mutations in tumour-suppressor genes and are transmitted as autosomal dominant traits. Genetically-determined insulinoma does not differ from sporadic insulinoma regarding the clinical and biological diagnostic criteria. On the other hand, the therapeutic strategy may be different, particularly in MEN-1 insulinomas, since insulinomas are often multiple or associated with other secreting or non

MEN-1 is related to a mutation in the menin gene, which maps to chromosome 11q13. Its prevalence is 1-10/100,000. The functional physiological role of menin has not yet been well established. Menin interacts with numerous proteins that play a role in transcriptional regulation, genomic stability, cell division and control of cell cycle (Balogh et al. 2006; Yang & Hua 2007). The major MEN-1 related endocrine disorders are pituitary adenomas, parathyroid adenomas or parathyroid hyperplasia, and secreting or non secreting pancreatic endocrine tumours. Regarding the secreting pancreatic endocrine tumours, the most frequent are gastrinomas (54%, 20-61% of patients) and insulinomas (18%, 7-31% of patients) (Jensen et al. 2008). MEN-1 related insulinomas occur generally in younger patients than sporadic insulinomas, with a greater prevalence in patients aged 25-35 years, but the age of occurrence is very different from patient to patient (5-80 years) (Cougard et al. 2000; Machens et al. 2007). Cystic pancreatic endocrine neoplasms could be more frequent in

The major characteristic of MEN-1 related insulinomas is that they are generally multiple tumours, so that diffuse microadenomatosis can be observed in the entire pancreatic gland, with multiple infracentimetric lesions of the pancreas (Anlauf et al. 2006). Often in such

(Grunenwald et al. 2009).

secreting endocrine tumours.

**3.2 Insulinomas as part of a genetic disorder** 

MEN-1 than in sporadic cases (Bordeianou et al. 2008).

cases there are also secreting or non secreting endocrine tumours that are more than 1 cm in size and are distributed in the whole pancreatic gland. Thus managing MEN-1-related insulinomas is often managing multiple pancreatic endocrine tumours, that may secrete insulin and other hormones (gastrinoma, glucagonoma…) or may be non secreting endocrine tumours (Cougard et al. 2000).

The natural course of endocrine disorders in MEN-1 patients was previously thought to start almost always by occurrence of primary hyperparathyroidism (Marx et al. 1998). However others have reported cases of MEN-1 revealed by a secreting pancreatic endocrine tumour; hyperparathyroidism had not been identified prior to the diagnosis of the secreting pancreatic tumour, because it had a mild or asymptomatic presentation (Thakker 2000). Thus insulinoma was the endocrine tumour that revealed MEN-1 in 14-30% of patients (O'Riordain et al. 1994; Cougard et al. 2000). Therefore, apparently sporadic insulinomas in young patients and multiple pancreatic lesions warrant to screen for MEN-1 even if there is no hyperparathyroidism.

The major issue is that of the treatment, since the multiple lesions increase the difficulty in achieving surgical cure of the disease. There is a mean of 3 pancreatic endocrine tumours (1 to 10) in the reported series of MEN-1 patients with pancreatic lesions who underwent surgery (Bartsch et al. 2005; Triponez et al. 2006). When managing an insulinoma patient with multiple pancreatic tumours, there could be several therapeutic strategies: 1. to perform total pancreatectomy with its associated morbidity and mortality rates; 2. to perform an extended resection of the pancreatic gland with enucleations of the tumours on the remaining pancreatic gland, which is also associated with significant morbidity and adds a risk of tumour recurrence. 3. to perform selective enucleations of the detected tumour(s), which increases the risk of re-operation for other tumours, and also may sometimes result in removing a nonsecreting tumour, and not the insulin-secreting tumour if it has not been detected because of its smaller size; this could warrant a functional study of the lesions (with preoperative transhepatic venous sampling or arterial calcium injection, or intra-operative measurement of insulin) 4.to choose a medical therapeutic approach in responders who do not seem to have malignant lesions. In an operable insulinoma patient, curative surgery is generally attempted for the following two reasons. First, the prevalence of malignant insulinomas is similar in MEN-1 patients and in patients with sporadic insulinomas (Jensen et al. 2008). The prognosis of MEN-1 depends on the progression of malignant pancreatic endocrine tumours (Jensen et al. 2008). Second, in insulinoma patients, even in a patient who is thought to be controlled by a medical treatment, the sudden occurrence of severe symptomatic hypoglycaemia cannot be excluded, with its potentially serious and even fatal consequences. Therefore there is a general agreement regarding the necessity to perform surgery in MEN-1 patients with insulinomas. The extent of surgery remains debated. It mainly depends on the number and the location of the lesions, and the possibility to enucleate the lesion. One of the major advantages of preoperative endoscopic ultrasound could be in the evaluation of MEN-1 patients, since it can visualize even small lesions in the head of the pancreas (Proye et al. 2004). Left-sided pancreatectomy with enucleation of the tumours located in the pancreatic head is generally recommended in patients with MEN-1 related insulinomas, whenever it is possible (Cougard et al. 2000). Tumor enucleation gives acceptable results if the choice of the lowest morbidity rates is made. In those patients who present with a small number of pancreatic tumours, surgery may be less extensive and single or multiple tumour enucleation can be performed, when it is technically possible. The patient must be informed about the risk of recurrence and re-operation.

Diagnosis and Treatment of Insulinomas in the Adults 163

reported as "non insulinoma pancreatogenic hypoglycaemia syndrome (NIPHS)", i.e. patients who only suffer from reactive insulin-mediated hypoglycaemia, who have a negative fast test without definite pancreatic endocrine tumour, and without mutations in the ABCC8 and KCNJ11 genes (Service et al. 1999). Though rare, it is not quite exceptional among adult patients with hypoglycaemia related to endogenous hyperinsulinism: 4/128 (3%) patients (Raffel et al. 2007), 15/232 (6%) patients (Anlauf et al. 2005b), and even 5/32 (16%) patients (Witteles et al. 2001). We have made the diagnosis of idiopathic nesidioblastosis only in 4 patients for 20 years (1990-2010), while during the same 20 years 69 patients were diagnosed to have insulinomas in our institution; in addition, two other adult patients who had presented symptoms since the early childhood were found to have

Adult nesidioblastosis is likely to represent a heterogeneous disorder and whether it has specific histological correlates remains debated. The first histological reported features were islet cell enlargement, beta cells budding off ductular epithelium, and islets in apposition to ducts. Such aspects were found in an autopsy series in 36% of subjects without hypoglycaemia (Karnauchow PN 1981), whereas using morphometric criteria, patients with NIPHH could not be distinguished from controls (Goudswaard et al. 1984). In 15 adults with nesidioblastosis, blinded interobserver analysis showed beta-cell hypertrophy with enlarged and hyperchromatic nuclei; the interobserver analysis revealed 100% specificity and 87.7% sensitivity for these criteria (Anlauf et al. 2005b). The degree and extent of these features vary much more from patient to patient than in newborns (Kloppel et al. 2008). The interpretation of histological findings is not unequivocal, as shown by the divergent conclusions regarding 6 patients with reactive insulin-mediated hypoglycaemia after bypass surgery (Service et al. 2005; Meier et al. 2006). In addition, Meier et al. reported a positive correlation between beta cell nuclear diameter and body mass index, so that the actual meaning of an increased nuclear size in beta cells can be debated. In 4 operated patients who all had transient remission of hypoglycaemia for about one year after left-sided pancreatectomy, the histological conclusions were nesidioblastosis in 2 patients, micronodular islet cell hyperplasia in one case, and presence of some enlarged nuclei in rare pancreatic areas in one patient (who had two heterozygous mutations of the ABCC8 gene)

Nesidioblastosis must not be confused with the recently identified pathological entity called insulinomatosis (Anlauf et al. 2009) which has been described in 14 patients, and is characterized by the synchronous and metachronous occurrence of insulinomas, multiple insulinoma precursor lesions, and rare development of metastases (all but one patient had

Nesidioblastosis may be related to well identified genetic disorders with mutations in genes that regulate beta cell insulin secretion (Flanagan et al. 2011a; Flanagan et al. 2011b; Palladino & Stanley 2011) (see above, differential diagnosis) and in such cases, it may occasionally be diagnosed in adult patients (Christesen et al. 2008). It has also been reported to be an adult-onset disease occurring in patients who had been previously treated by insulin, or a sulfonylurea, and in patients with MEN-1, Zollinger-Ellison and Werner-Morrison syndromes (Service et al. 1999). A higher rate of nesidioblastosis in MEN-1 adult patients has not been confirmed by all studies (Anlauf et al. 2005a). None of our patients with adult nesidioblastosis was found to have MEN-1. Conversion of insulin-dependent diabetes mellitus into nesidioblastosis has also been reported (Raffel et al. 2006). Recently, it has been found in patients after gastric bypass surgery (0.5-8 years after surgery in 5/6

ABCC8 or glutamate dehydrogenase gene mutations.

benign disease), but common recurrent hypoglycaemia.

(personal data).

VHL disease is caused by a mutation in the VHL gene, which maps to chromosome 3p25. Its prevalence is 2 to 3/100,000. It is responsible for occurrence of renal cysts and cancers, hemangioblastomas in the central nervous system, pheochromocytomas and paragangliomas, and pancreatic tumours (Barontini & Dahia 2010). Pancreatic tumours and cysts are reported in 35-77% patients with VHL (Mukhopadhyay et al. 2002), but to date, to our knowledge, there is no case report of VHL-related insulinoma. The mean age of VHL patients who present with a pancreatic lesion varies between 29 and 38 years; unlike what is found MEN-1 patients with pancreatic endocrine tumours, 70% of VHL patients with pancreatic tumours have a single pancreatic tumour (Libutti et al. 2000; Blansfield et al. 2007). Pancreatic endocrine tumours represent only 9-17% of pancreatic lesions in VHL patients (Mukhopadhyay et al. 2002), most of them being non secreting tumours (Libutti et al. 2000). It has also been reported that pancreatic lesions in VHL patients were significantly associated with renal cancers. Since insulinomas are obviously rare in VHL, it seems reasonable to screen for VHL in an insulinoma patient only if the insulinoma is associated with other typical VHL lesions, particularly renal cancer.

Tuberous sclerosis is a phakomatosis characterized by cutaneous (facial angiofibromas also termed "adenoma sebaceum", hypomelanic macules also termed "ash leaf spots", fibrous raised discoloured forehead plaques, and periungual fibromas), neurological (mental retardation, seizures, subependymal nodules and cortical/subcortical tubers), and ocular (hamartomas, also termed "phakomas") manifestations. The patients also present with renal cysts and angiomyolipomas and cardiac rhabdomyomas. TS is caused by mutations either in the TSC 1 gene (that encodes hamartin, and maps to chromosome 9q34) or in the TSC2 gene (that encodes tuberin, and maps to chromosome 16p13.3) (Lodish & Stratakis 2010). Its prevalence is about 1/10,000 (Osborne et al. 1991). The occurrence of insulinomas in TS patients has been reported very rarely (Davoren & Epstein 1992; Kim et al. 1995; Boubaddi et al. 1997; Eledrisi et al. 2002; Dworakowska & Grossman 2009). It is generally found in patients with mutations in the TSC2 gene (Merritt et al. 2006). In all reported cases, the insulinoma was found in patients aged more than 20, many years after that the diagnosis of TS had been made. Thus insulinoma is never the first manifestation of TS. Consequently, there is no reason to screen for TS in insulinoma patients. On the other hand, in a patient who is known to have TS, the re-occurrence of neurological or psychiatric symptoms warrants measurement of plasma glucose concentration.

Neurofibromatosis type 1 (NF1) ("Von Recklinghausen's disease") is caused by a mutation in the NF1 gene which maps to chromosome 17q11.2 and encodes a protein named fibromin. The main clinical signs are cutaneous "café-au-lait" spots, neurofibromas, Lisch nodules on iris surface, optic gliomas and bone dysplasia. Its prevalence is 1 in 4000 to 5000. The most frequent endocrine tumour is pheochromocytoma. Pancreatic endocrine tumours are rare (0- 10%). Digestive tumours in NF1 are gastrointestinal stromal tumours (GIST), and duodenal somatostatinomas that characteristically cause biliary dilatation (Miettinen et al. 2006). Insulinomas are exceptional in NF1 (Perren et al. 2006).

#### **3.3 Nesidioblastosis**

There is no general agreement regarding the definition of adult nesidioblastosis. Nesidioblastosis represents a state of diffuse beta cell hyperfunction (though not necessarily uniform in all cases), also termed "non-insulinoma persistent hyperinsulinemic hypoglycaemia (NIPHH)" (Christesen et al. 2008). Such patients have insulin-mediated hypoglycaemia, either in the fasting or in the postprandial state. It also comprises cases

VHL disease is caused by a mutation in the VHL gene, which maps to chromosome 3p25. Its prevalence is 2 to 3/100,000. It is responsible for occurrence of renal cysts and cancers, hemangioblastomas in the central nervous system, pheochromocytomas and paragangliomas, and pancreatic tumours (Barontini & Dahia 2010). Pancreatic tumours and cysts are reported in 35-77% patients with VHL (Mukhopadhyay et al. 2002), but to date, to our knowledge, there is no case report of VHL-related insulinoma. The mean age of VHL patients who present with a pancreatic lesion varies between 29 and 38 years; unlike what is found MEN-1 patients with pancreatic endocrine tumours, 70% of VHL patients with pancreatic tumours have a single pancreatic tumour (Libutti et al. 2000; Blansfield et al. 2007). Pancreatic endocrine tumours represent only 9-17% of pancreatic lesions in VHL patients (Mukhopadhyay et al. 2002), most of them being non secreting tumours (Libutti et al. 2000). It has also been reported that pancreatic lesions in VHL patients were significantly associated with renal cancers. Since insulinomas are obviously rare in VHL, it seems reasonable to screen for VHL in an insulinoma patient only if the insulinoma is associated

Tuberous sclerosis is a phakomatosis characterized by cutaneous (facial angiofibromas also termed "adenoma sebaceum", hypomelanic macules also termed "ash leaf spots", fibrous raised discoloured forehead plaques, and periungual fibromas), neurological (mental retardation, seizures, subependymal nodules and cortical/subcortical tubers), and ocular (hamartomas, also termed "phakomas") manifestations. The patients also present with renal cysts and angiomyolipomas and cardiac rhabdomyomas. TS is caused by mutations either in the TSC 1 gene (that encodes hamartin, and maps to chromosome 9q34) or in the TSC2 gene (that encodes tuberin, and maps to chromosome 16p13.3) (Lodish & Stratakis 2010). Its prevalence is about 1/10,000 (Osborne et al. 1991). The occurrence of insulinomas in TS patients has been reported very rarely (Davoren & Epstein 1992; Kim et al. 1995; Boubaddi et al. 1997; Eledrisi et al. 2002; Dworakowska & Grossman 2009). It is generally found in patients with mutations in the TSC2 gene (Merritt et al. 2006). In all reported cases, the insulinoma was found in patients aged more than 20, many years after that the diagnosis of TS had been made. Thus insulinoma is never the first manifestation of TS. Consequently, there is no reason to screen for TS in insulinoma patients. On the other hand, in a patient who is known to have TS, the re-occurrence of neurological or psychiatric symptoms

Neurofibromatosis type 1 (NF1) ("Von Recklinghausen's disease") is caused by a mutation in the NF1 gene which maps to chromosome 17q11.2 and encodes a protein named fibromin. The main clinical signs are cutaneous "café-au-lait" spots, neurofibromas, Lisch nodules on iris surface, optic gliomas and bone dysplasia. Its prevalence is 1 in 4000 to 5000. The most frequent endocrine tumour is pheochromocytoma. Pancreatic endocrine tumours are rare (0- 10%). Digestive tumours in NF1 are gastrointestinal stromal tumours (GIST), and duodenal somatostatinomas that characteristically cause biliary dilatation (Miettinen et al. 2006).

There is no general agreement regarding the definition of adult nesidioblastosis. Nesidioblastosis represents a state of diffuse beta cell hyperfunction (though not necessarily uniform in all cases), also termed "non-insulinoma persistent hyperinsulinemic hypoglycaemia (NIPHH)" (Christesen et al. 2008). Such patients have insulin-mediated hypoglycaemia, either in the fasting or in the postprandial state. It also comprises cases

with other typical VHL lesions, particularly renal cancer.

warrants measurement of plasma glucose concentration.

Insulinomas are exceptional in NF1 (Perren et al. 2006).

**3.3 Nesidioblastosis** 

reported as "non insulinoma pancreatogenic hypoglycaemia syndrome (NIPHS)", i.e. patients who only suffer from reactive insulin-mediated hypoglycaemia, who have a negative fast test without definite pancreatic endocrine tumour, and without mutations in the ABCC8 and KCNJ11 genes (Service et al. 1999). Though rare, it is not quite exceptional among adult patients with hypoglycaemia related to endogenous hyperinsulinism: 4/128 (3%) patients (Raffel et al. 2007), 15/232 (6%) patients (Anlauf et al. 2005b), and even 5/32 (16%) patients (Witteles et al. 2001). We have made the diagnosis of idiopathic nesidioblastosis only in 4 patients for 20 years (1990-2010), while during the same 20 years 69 patients were diagnosed to have insulinomas in our institution; in addition, two other adult patients who had presented symptoms since the early childhood were found to have ABCC8 or glutamate dehydrogenase gene mutations.

Adult nesidioblastosis is likely to represent a heterogeneous disorder and whether it has specific histological correlates remains debated. The first histological reported features were islet cell enlargement, beta cells budding off ductular epithelium, and islets in apposition to ducts. Such aspects were found in an autopsy series in 36% of subjects without hypoglycaemia (Karnauchow PN 1981), whereas using morphometric criteria, patients with NIPHH could not be distinguished from controls (Goudswaard et al. 1984). In 15 adults with nesidioblastosis, blinded interobserver analysis showed beta-cell hypertrophy with enlarged and hyperchromatic nuclei; the interobserver analysis revealed 100% specificity and 87.7% sensitivity for these criteria (Anlauf et al. 2005b). The degree and extent of these features vary much more from patient to patient than in newborns (Kloppel et al. 2008). The interpretation of histological findings is not unequivocal, as shown by the divergent conclusions regarding 6 patients with reactive insulin-mediated hypoglycaemia after bypass surgery (Service et al. 2005; Meier et al. 2006). In addition, Meier et al. reported a positive correlation between beta cell nuclear diameter and body mass index, so that the actual meaning of an increased nuclear size in beta cells can be debated. In 4 operated patients who all had transient remission of hypoglycaemia for about one year after left-sided pancreatectomy, the histological conclusions were nesidioblastosis in 2 patients, micronodular islet cell hyperplasia in one case, and presence of some enlarged nuclei in rare pancreatic areas in one patient (who had two heterozygous mutations of the ABCC8 gene) (personal data).

Nesidioblastosis must not be confused with the recently identified pathological entity called insulinomatosis (Anlauf et al. 2009) which has been described in 14 patients, and is characterized by the synchronous and metachronous occurrence of insulinomas, multiple insulinoma precursor lesions, and rare development of metastases (all but one patient had benign disease), but common recurrent hypoglycaemia.

Nesidioblastosis may be related to well identified genetic disorders with mutations in genes that regulate beta cell insulin secretion (Flanagan et al. 2011a; Flanagan et al. 2011b; Palladino & Stanley 2011) (see above, differential diagnosis) and in such cases, it may occasionally be diagnosed in adult patients (Christesen et al. 2008). It has also been reported to be an adult-onset disease occurring in patients who had been previously treated by insulin, or a sulfonylurea, and in patients with MEN-1, Zollinger-Ellison and Werner-Morrison syndromes (Service et al. 1999). A higher rate of nesidioblastosis in MEN-1 adult patients has not been confirmed by all studies (Anlauf et al. 2005a). None of our patients with adult nesidioblastosis was found to have MEN-1. Conversion of insulin-dependent diabetes mellitus into nesidioblastosis has also been reported (Raffel et al. 2006). Recently, it has been found in patients after gastric bypass surgery (0.5-8 years after surgery in 5/6

Diagnosis and Treatment of Insulinomas in the Adults 165

origin for excess insulin secretion (Starke et al. 2006). Four of our patients with nesidioblastosis were treated by left-sided pancreatectomy; all of them had transient remission of hypoglycaemia for about one year, then acceptable control was observed with diazoxide or octreotide, which proved to be more effective to control hypoglycaemia after partial pancreatectomy. 70%-distal pancreatectomy was performed in 5 patients with nesidioblastosis (Witteles et al. 2001) resulting in long-term remission in 3 patients, and some recurrences of hypoglycaemia in 2 patients, successfully treated by calcium channel blockers, with a follow-up of 1.5-21 years. Thus 60-75% resection of the pancreatic gland, guided by insulin venous gradient or insulin stimulation by selective intraarterial calcium injection, and followed if necessary by a medical therapy, appears to be a reasonable treatment for adult nesidioblastosis when enough efficacy cannot be achieved by medical

Agostino, N., Chinchilli V. M., et al. (2010). "Effect of the tyrosine kinase inhibitors (sunitinib,

Ahn, J. E., Byun J. H., et al. (2007). "Case report: neuroendocrine carcinoma of the gallbladder causing hyperinsulinaemic hypoglycaemia." *Clinical Radiology* 62(4): 391-394. Altszuler, N., Moraru E., et al. (1977). "On the mechanism of diazoxide-induced

An, L., Li W., et al. (2010). "Assessment of contrast-enhanced ultrasonography in diagnosis and

Anlauf, M., Perren A., et al. (2005a). "Precursor lesions in patients with multiple endocrine

Anlauf, M., Wieben D., et al. (2005b). "Persistent hyperinsulinemic hypoglycemia in 15 adults

Anlauf, M., Bauersfeld J. et al. (2009) Insulinomatosis : a multicentric insulinoma disease that

Arao, T., Okada Y., et al. (2006). "A rare case of adult-onset nesidioblastosis treated

Ardengh, J. C., Rosenbaum P., et al. (2000). "Role of EUS in the preoperative localization of insulinomas compared with spiral CT." *Gastrointestinal Endoscopy* 51(5): 552-555. Arem, R. (1989). "Hypoglycemia associated with renal failure." *Endocrinology and Metabolism* 

Arnold, R., Chen Y.J., et al. (2009). ENETS Consensus Guidelines for the standards of care in

neuroendocrine tumors: follow-up and documentation. *Neuroendocrinology* 90(2): 227-

beta-cell changes." *American Journal of Surgical Pathology* 29(4): 524-533. Anlauf, M., Schlenger R., et al. (2006). "Microadenomatosis of the endocrine pancreas in

sorafenib, dasatinib, and imatinib) on blood glucose levels in diabetic and nondiabetic patients in general clinical practice." *Journal of Oncology Pharmacy Practice* 2010 Aug 14

preoperative localization of insulinoma." *European Journal of Radiology*. 2010 Oct 19.

neoplasia type 1-associated duodenal gastrinomas." *Gastroenterology* 128(5): 1187-

with diffuse nesidioblastosis: diagnostic criteria, incidence, and characterization of

patients with and without the multiple endocrine neoplasia type 1 syndrome."

frequently causes early recurrent hyperinsulinemic hypoglycemia. *American Journal of* 

therapy alone.

**4. References** 

ahead of print.

1198.

233.

hyperglycemia." *Diabetes* 26(10): 931-935.

ahead of print doi:10.1016/j.ejrad.2010.09.014

*American Journal of Surgical Pathology* 30(5): 560-574.

successfully with diazoxide." *Endocrine Journal* 53(1): 95-100.

*Surgical Pathology* 33(3): 339-346.

*Clinics of North America* 18(1): 103-121.

patients) but whether such patients have actual structural pancreatic abnormalities is still debated (see differential diagnosis, dumping syndrome). Most reported patients with adultonset nesidioblastosis do not have any particular medical history and do not present with mutations known to be associated with hypoglycaemia in young children.

The diagnosis of nesidioblastosis is made in patients with definite biological evidence of hypoglycaemia related to endogenous hyperinsulinism (as in insulinoma patients, see biological diagnosis) in the absence of insulin secretagogues, autoimmune hypoglycaemia, and detectable pancreatic tumour. There is neither clinical nor biological basis to make a distinction between insulinoma and nesidioblastosis. Patients with nesidioblastosis may have a positive fast test (as in all our patients), though patients with NIPHS have negative fast tests and only reactive insulin-mediated hypoglycaemia; on the other hand, very rare insulinoma patients may present with a negative fast test, and some of them have only reactive insulin-mediated hypoglycaemia. The diagnostic specificity of multi-site insulin stimulation by intra-arterial calcium injection is not fully established: such insulin stimulation after intraarterial calcium injection was found in normal pancreatic areas for unknown reasons (Wiesli et al. 2004a; Guettier et al. 2009), and has been reported in patients with factitious hypoglycaemia related to insulin secretagogues (Hirshberg et al. 2001). In addition, intra-arterial calcium stimulation may provide a pseudo-insulinoma response pattern in patients with adult nesidioblastosis, maybe because nesidioblastosis is not evenly distributed in the pancreas. One must also be reminded that nesidioblastosis has been found in patients who had concurrent insulinoma (Bright et al. 2008) (Service et al. 1999). Slight focal pancreatic abnormalities have been reported with radionuclide imaging (Kauhanen et al. 2007).

The natural course of adult-onset nesidioblastosis is not well established. Most patients require a medical and/or surgical treatment. In an obese patient with idiopathic nesidioblastosis who had been treated by left-sided pancreatectomy and high doses of octreotide, hypoglycaemia finally subsided during the perimenopausal years, and changed to mild type-2 diabetes mellitus. No similar change was observed in other patients with nesidioblastosis, though the dose of diazoxide could be decreased in one patient during the perimenopausal years (unpublished personal data).

A medical treatment is the first-line therapeutic approach for nesidioblastosis, but it may not be sufficient to control the hypoglycaemia, and pancreatic surgery must often be performed. Diazoxide remains the reference medication for nesidioblastosis, and is often effective to control the hypoglycaemia, at least after partial pancreatectomy (Arao et al. 2006). Longterm treatment with somatostatin analogues has also been reported to be effective, without the side effects of diazoxide (Vezzosi et al. 2008). One of our patients was treated with continuous subcutaneous administration of a high dose of octreotide throughout her pregnancy, without any detectable maternal or foetal adverse effect (Boulanger et al. 2004). Calcium channel blockers have been employed (Witteles et al. 2001) (Vella & Service 2007). Pancreatic surgery, guided by the results of transhepatic venous sampling or those of insulin stimulation after selective intraarterial calcium injection, is often performed. The extent of surgery is controversial, in order to achieve a total control or a significant improvement of hypoglycaemia, without pancreatic insufficiency (Raffel et al. 2007). Selective intra-arterial calcium injection usually helps in choosing the extent of pancreatectomy (Toyomasu et al. 2009). Total pancreatectomy cannot be recommended, since complete or acceptable control of the hypoglycaemia can be achieved in most patients by partial pancreatectomy followed if necessary by a medical treatment. Spleen-preserving distal pancreatectomy is often performed unless the results suggest a proximal pancreatic origin for excess insulin secretion (Starke et al. 2006). Four of our patients with nesidioblastosis were treated by left-sided pancreatectomy; all of them had transient remission of hypoglycaemia for about one year, then acceptable control was observed with diazoxide or octreotide, which proved to be more effective to control hypoglycaemia after partial pancreatectomy. 70%-distal pancreatectomy was performed in 5 patients with nesidioblastosis (Witteles et al. 2001) resulting in long-term remission in 3 patients, and some recurrences of hypoglycaemia in 2 patients, successfully treated by calcium channel blockers, with a follow-up of 1.5-21 years. Thus 60-75% resection of the pancreatic gland, guided by insulin venous gradient or insulin stimulation by selective intraarterial calcium injection, and followed if necessary by a medical therapy, appears to be a reasonable treatment for adult nesidioblastosis when enough efficacy cannot be achieved by medical therapy alone.
