**2. Single benign sporadic insulinomas**

#### **2.1 Diagnosis of hypoglycaemia related to endogenous hyperinsulinism 2.1.1 Clinical symptoms of insulinomas**

The clinical symptoms of insulinomas are heterogeneous and may differ among the patients. There is no specific clinical symptom. Hypoglycaemia results in autonomic and neuroglycopenic symptoms. Autonomic symptoms comprise adrenergic symptoms, such as palpitations and tremor, and cholinergic symptoms, with sweating, hunger, sometimes paresthesias. Neuroglycopenic symptoms comprise severe weakness, and many psychiatric and neurological manifestations, the most commonly reported being behavioural changes, confusion, agitation or slow reaction patterns, blurred vision, and finally seizures, transient loss of consciousness or hypoglycaemic coma. Hypoglycaemic coma is typically sudden, with agitation, sometimes pyramidal signs, and hypothermia; it may be profound and prolonged and therefore may be a cause of neurologic and cognitive sequelae and may occasionally lead to brain death. The combination of autonomic and moderate neuroglycopenic symptoms strongly suggests the diagnosis of hypoglycaemia. However, in insulinoma patients, autonomic symptoms may often be lacking. Neuroglycopenic symptoms are more specific of hypoglycemia related to insulinoma, but may be mild, or more often misleading. Therefore measurement of plasma glucose concentrations should be recommended in all patients with such symptoms, either in a patient without any known neurological or psychiatric disorder, or in a patient in whom such disorder had been previously controlled. Hypoglycaemic spells typically occur suddenly after a period of fast, several hours after the last meal, or after physical exercise. Some of them have been reported to occur within the hours following a meal or without any relationship with meal time, so that symptoms occurring in non fasting patients cannot rule out the diagnosis of insulinoma. In a series of 214 insulinoma patients with a reliable clinical history, symptoms of hypoglycaemia were reported exclusively in the fasting state in 73%, postprandial state in 6%, and both fasting and postprandial state in 21% (Placzkowski et al. 2009). Weight gain is found only in 25% of patients.

Rapid (or rarely delayed) resolution of symptoms after glucose administration is part of Whipple's triad, which makes the diagnosis of hypoglycaemia when it has been assessed reliably; the triad comprises clinical symptoms of hypoglycaemia, the finding of a low plasma glucose concentration at the time of clinical symptoms, and subsiding of clinical symptoms after glucose administration.

In rare pregnant patients with insulinoma, symptoms may resemble those of emesis in early pregnancy and may subside in late pregnancy as a consequence of insulin resistance, then are unmasked after delivery (Diaz et al. 2008), as reported in a patient who presented with coma and clonic seizures 14 hours after delivery (Christiansen & Vestergaard 2008).

#### **2.1.2 Biological diagnosis of insulinomas**

#### **Plasma glucose concentrations**

Since the clinical symptoms lack diagnostic specificity, measurement of plasma concentration of glucose is mandatory (ideally at the time of occurrence of clinical symptoms) and must be performed in venous blood samples with a reliable laboratory method. In order to avoid pseudohypoglycaemia, i.e. a consequence of glucose metabolism by blood cells in the tube containing the blood sample, the blood sample should be collected in a tube with a glycolysis inhibitor (NaF). Glucose concentrations are measured using reference method with hexokinase.

The clinical symptoms of insulinomas are heterogeneous and may differ among the patients. There is no specific clinical symptom. Hypoglycaemia results in autonomic and neuroglycopenic symptoms. Autonomic symptoms comprise adrenergic symptoms, such as palpitations and tremor, and cholinergic symptoms, with sweating, hunger, sometimes paresthesias. Neuroglycopenic symptoms comprise severe weakness, and many psychiatric and neurological manifestations, the most commonly reported being behavioural changes, confusion, agitation or slow reaction patterns, blurred vision, and finally seizures, transient loss of consciousness or hypoglycaemic coma. Hypoglycaemic coma is typically sudden, with agitation, sometimes pyramidal signs, and hypothermia; it may be profound and prolonged and therefore may be a cause of neurologic and cognitive sequelae and may occasionally lead to brain death. The combination of autonomic and moderate neuroglycopenic symptoms strongly suggests the diagnosis of hypoglycaemia. However, in insulinoma patients, autonomic symptoms may often be lacking. Neuroglycopenic symptoms are more specific of hypoglycemia related to insulinoma, but may be mild, or more often misleading. Therefore measurement of plasma glucose concentrations should be recommended in all patients with such symptoms, either in a patient without any known neurological or psychiatric disorder, or in a patient in whom such disorder had been previously controlled. Hypoglycaemic spells typically occur suddenly after a period of fast, several hours after the last meal, or after physical exercise. Some of them have been reported to occur within the hours following a meal or without any relationship with meal time, so that symptoms occurring in non fasting patients cannot rule out the diagnosis of insulinoma. In a series of 214 insulinoma patients with a reliable clinical history, symptoms of hypoglycaemia were reported exclusively in the fasting state in 73%, postprandial state in 6%, and both fasting and postprandial state in 21%

**2.1 Diagnosis of hypoglycaemia related to endogenous hyperinsulinism** 

(Placzkowski et al. 2009). Weight gain is found only in 25% of patients.

symptoms after glucose administration.

**2.1.2 Biological diagnosis of insulinomas** 

**Plasma glucose concentrations** 

reference method with hexokinase.

Rapid (or rarely delayed) resolution of symptoms after glucose administration is part of Whipple's triad, which makes the diagnosis of hypoglycaemia when it has been assessed reliably; the triad comprises clinical symptoms of hypoglycaemia, the finding of a low plasma glucose concentration at the time of clinical symptoms, and subsiding of clinical

In rare pregnant patients with insulinoma, symptoms may resemble those of emesis in early pregnancy and may subside in late pregnancy as a consequence of insulin resistance, then are unmasked after delivery (Diaz et al. 2008), as reported in a patient who presented with

Since the clinical symptoms lack diagnostic specificity, measurement of plasma concentration of glucose is mandatory (ideally at the time of occurrence of clinical symptoms) and must be performed in venous blood samples with a reliable laboratory method. In order to avoid pseudohypoglycaemia, i.e. a consequence of glucose metabolism by blood cells in the tube containing the blood sample, the blood sample should be collected in a tube with a glycolysis inhibitor (NaF). Glucose concentrations are measured using

coma and clonic seizures 14 hours after delivery (Christiansen & Vestergaard 2008).

**2. Single benign sporadic insulinomas** 

**2.1.1 Clinical symptoms of insulinomas** 

The threshold to define hypoglycemia remains controversial. According to recent expert recommendations (Cryer et al. 2009) a 0.55 g/L threshold must be selected. However, the glycaemic threshold that can cause clinical symptoms is very different among individuals, and a plasma concentration of glucose below 0.7 g/L at the time of clinical symptoms is thought be enough, if the other criteria of Whipple's triad are fulfilled, to warrant further investigation. A plasma glucose concentration above 0.7 g/L during a symptomatic episode indicates that those symptoms are not the result of hypoglycaemia.

Since even values below 0.55 g/L can be found in some healthy individuals, it is also recommended by the expert consensus that such levels be taken into account for further evaluation only in patients who had presented Whipple's triad. On the other hand, very rare insulinoma patients were found to be asymptomatic and did not fulfil Whipple's triad. Among our patients (Vezzosi et al. 2007), the diagnosis of insulinoma was made on the basis of repeated plasma glucose concentrations of 0.40-0.55 g/l after overnight fasts in one patient, though she never described any clinical symptom. In many insulinoma patients, the glycaemic threshold for clinical symptoms is shifted to very low glucose levels, so that they may be asymptomatic at the time of plasma glucose concentrations below 0.55 and even 0.45 g/L; most of these patients experienced some clinical symptoms suggestive of hypoglycaemia, but even if they present with fasting glucose levels below 0.55 g/L, Whipple's triad has not been assessed in most of them, and awaiting Whipple's triad to be fulfilled would make the diagnosis be delayed. The finding of a spontaneous plasma glucose concentration below 0.55 g/L is rare in normal subjects; during a 72-hour fast test, plasma glucose concentrations do not reach values below 0.45 g/L in most controls, and do not drop below 0.4 g/L in controls (Vezzosi et al. 2007). Therefore we think that patients with spontaneous plasma glucose concentrations below 0.55 g/L warrant further investigation, even when they do not fulfil all the criteria of Whipple's triad, and that such evaluation is mandatory in patients with plasma concentrations below 0.45 g/L, provided that plasma glucose concentration has been measured reliably in a venous sample.

Confirmation of hypoglycaemia in a venous sample is the first requirement for the biological diagnosis of insulinoma.

If a venous sample cannot be collected when hypoglycaemia occurs spontaneously, a 72 hour-fast test is to be performed.

A detailed protocol for the fast test has first been described by the Mayo Clinic group (Service 1995)(Service 1999) and more recently by an expert consensus (Cryer et al. 2009). The patient is allowed to drink calorie-free beverages. Samples are to be collected every 6 hours until the plasma glucose concentration is less than 0.6 g/L if the patient remains asymptomatic, then the frequency of sampling is increased (every 1 or 2 hours). Serum insulin, C-peptide, proinsulin and beta-hydroxy-butyrate are to be measured in all the samples taken at the time when plasma glucose concentration drops below 0.6 g/L. At the end of the fast test, a sample is collected to measure oral hypoglycaemic agents; a glucagon test (1.0 mg intravenously) has also been recommended. Insulin antibodies should also be measured.

The criteria used to decide to stop the fast test before 72 hours have been modified recently. According to previous recommendations (Service 1995), the fast test is to be stopped if the patient has symptomatic hypoglycaemia with a plasma glucose concentration of 0.45 g/l or less, or if the plasma glucose concentration drops below 0.4 g/L. According to the recent recommendations (Cryer et al. 2009) the fast test can be stopped: 1) when Whipple's triad is observed; 2) when plasma glucose concentrations drop below 0.55 g/L in a patient who had previously experienced Whipple's triad; 3) if plasma beta-hydroxy-butyrate levels rise above 2.7 mmol/L.

Diagnosis and Treatment of Insulinomas in the Adults 139

According to the expert consensus, the diagnosis of hypoglycaemia related to endogenous hyperinsulinism can be made when serum insulin concentration is equal to or greater than 3 mIU/L, serum C-peptide concentration is equal to or greater than 0.6 ng/mL (0.2 nmol/L), and serum proinsulin concentration is equal to or greater than 5 pmol/L at the time of hypoglycaemia, with venous plasma glucose concentrations less than 0.55 g/L. During a fast test, beta-hydroxy-butyrate levels of 2.7 mmol/L or less and an increase in plasma glucose of at least 0.25 g/L after intravenous glucagon indicate mediation of hypoglycaemia by insulin.

Serum insulin concentrations can be artifactually modified by haemolysis, heterophilic antibodies, and endogenous anti-insulin antibodies. Red cells contain an insulin-degrading enzyme which may lead to underestimation of the actual insulin concentrations on haemolysed samples. Most insulin assays are calibrated against the standard IRP 66/304 preparation, and the new 83/500 standard is not yet used by all insulin assays. As a consequence of the differences between the standards and the methods employed, the conversion factor from mIU/L to pmol/L varies from 6.0 to 7.5. In the 2009 expert

Insulin concentrations had been measured for long with non-specific assays that yielded cross-reactions with proinsulin. The insulin assays that are now in use have negligible crossreaction with proinsulin. Though the 3 mIU/L threshold (which replaces a 6 mIU/l threshold previously established with less specific insulin assays) was established for insulin-specific assays, it is known now that the finding of low insulin levels with such assays cannot rule out the diagnosis of insulinoma (Vezzosi et al. 2003). 11-35% patients with insulinoma were found to have insulin levels below the recommended diagnostic threshold (3 mIU/L) at the time of hypoglycemia when insulin was measured with insulinspecific assays, the percentage of values below 3 mIU/L depending on the assay employed (Vezzosi et al. 2007). On the other hand, while the specificity of the 3 mIU/L threshold was found to be 100% with a plasma glucose concentration threshold of 0.45 g/L for the diagnosis of hypoglycaemia related to endogenous hyperinsulinism, it was lowered to 87 % for glucose levels ranging from 0.46 to 0.54 g/L (personal data). The sensitivity and specificity for serum insulin concentrations with a threshold level of at least 3 mIU/l at the time of plasma glucose concentrations below 0.6 g/L were found to be 93% and 95%,

C-peptide and proinsulin levels at the time of hypoglycemia have better diagnostic accuracy than insulin levels. C-peptide assays are calibrated against the 1st standard IRP 84/510 (WHO). According to the expert consensus (Cryer et al. 2009), serum C-peptide levels of at least 0.6 ng/mL concomitant with plasma glucose concentrations below 0.55 g/l are one of the criteria that make the diagnosis of hypoglycaemia related to endogenous hyperinsulinism. In 33 patients who reached plasma glucose concentrations below 0.45 g/L during a fast test, all insulinoma patients had C-peptide levels above 0.6 ng/mL, and only one patient who had previously been treated by left-sided pancreatectomy for nesidioblastosis and who had recurrent hypoglycaemia did not reach the 0.6 ng/mL threshold at the time of hypoglycaemia below 0.45 g/L (Vezzosi et al. 2007). The diagnostic accuracy is improved when blood glucose levels drop below 0.45 g/L: for such blood glucose levels, the sensitivity

consensus, the threshold of 3 mIU/L is equal to 18 pmol/L.

respectively (Placzkowski et al. 2009).

**C-peptide** 

**Criteria for inappropriate insulin secretion (hypoglycaemia related to endogenous** 

**hyperinsulinism)** 

**Insulin** 

Awaiting that plasma glucose concentrations drop below 0.45 g/L in an asymptomatic patient to stop the fast test, instead of stopping the fast when plasma glucose is 0.46-0.54 g/L, leads to an improvement in diagnostic specificity regarding the diagnosis of a hypoglycaemic disorder. In our own data, out of 67 controls, 31 reached glucose plasma concentrations of 0.46-0.54 g/L during a 72-hour fast test, and only 4 reached values within the 0.40-0.45 g/L range (0.41-0.43 g/L) after 24 hours of fast. No control subject reached plasma glucose concentrations below 0.4 g/L. Among 55 patients who were diagnosed to have insulinoma in our institution between 1997 and 2010, and who underwent a fast test, only one patient reached a plasma glucose concentration below 0.55 g/L (0.53 g/L) without further decrease in glucose concentration, and all the other patients who had a positive fast test reached glucose concentrations below 0.45 g/L; the fast test was negative in one patient. In addition, using a 0.45 g/L glucose level threshold increases the diagnostic specificity of concomitant serum insulin, C-peptide and proinsulin levels for the diagnosis of insulinoma (see below). Therefore we think that if the patient remains asymptomatic, the fast test is to be prolonged at least until reaching plasma glucose concentrations below 0.45 g/L, the diagnostic specificity being about 100% if such levels are reached.

In most insulinoma patients such plasma glucose concentrations are reached within the first 48 hours of the test (Hirshberg et al. 2000; Vezzosi et al. 2007), though the 72 hour-fast test was reported to be necessary to provide a clear diagnostic conclusion in a few patients (Service 1999; Service & Natt 2000). Moreover, additional criteria for the diagnosis of insulinoma using proinsulin and beta-hydroxy-butyrate levels and plasma glucose response to intravenous glucagon were established during a 72-hour (not 48-hour) fast test for patients who did not reach plasma glucose concentrations below 0.45 or 0.55 g/L. Thus in most patients the test to be performed is the classical 72 hour-test. However, 48 hours of fast test are enough to provide diagnostic evidence in almost all insulinoma patients who have a positive fast test.

In the patients who experience symptoms of hypoglycemia only within the hours following a meal, a meal test is to be performed with serial measurement of plasma concentrations of glucose, insulin, C-peptide and proinsulin during the 5 hours following the ingestion of a standardized meal. A mixed meal similar to that which the patient reports to have caused the clinical symptoms is to be used, or a commercial formula mixed meal. Samples should be collected every 30 minutes and those collected for insulin, C-peptide and proinsulin measurement should be sent for analysis only if the sample has been taken at the time when plasma glucose concentration was found to be below 0.6 g/L. Measurement of oral hypoglycaemic agents and anti-insulin antibodies must also be performed if Whipple's triad is demonstrated (Cryer et al. 2009). Post-prandial hypoglycaemia without fasting hypoglycaemia is now known to occur in some insulinoma patients (6% according to Placzkowski et al. 2009) and in patients with noninsulinoma pancreatogenous hypoglycaemia (Service et al. 1999), a syndrome which is part of nesidioblastosis (see below). Others have reported false negative tests in insulinoma patients who are correctly diagnosed after a meal test or oral glucose tolerance test (Sjoberg & Kidd 1992) (Izumiyama et al. 2006; Kar et al. 2006). A mixed meal test was shown to be preferable to the oral glucose tolerance test (Hogan et al. 1983), so that only the meal test was recommended in the 2009 expert consensus.

The next requirement for the diagnosis of insulinoma is to provide evidence for inappropriate insulin secretion at the time of hypoglycaemia.

#### **Criteria for inappropriate insulin secretion (hypoglycaemia related to endogenous hyperinsulinism)**

According to the expert consensus, the diagnosis of hypoglycaemia related to endogenous hyperinsulinism can be made when serum insulin concentration is equal to or greater than 3 mIU/L, serum C-peptide concentration is equal to or greater than 0.6 ng/mL (0.2 nmol/L), and serum proinsulin concentration is equal to or greater than 5 pmol/L at the time of hypoglycaemia, with venous plasma glucose concentrations less than 0.55 g/L. During a fast test, beta-hydroxy-butyrate levels of 2.7 mmol/L or less and an increase in plasma glucose of at least 0.25 g/L after intravenous glucagon indicate mediation of hypoglycaemia by insulin.

#### **Insulin**

138 Basic and Clinical Endocrinology Up-to-Date

Awaiting that plasma glucose concentrations drop below 0.45 g/L in an asymptomatic patient to stop the fast test, instead of stopping the fast when plasma glucose is 0.46-0.54 g/L, leads to an improvement in diagnostic specificity regarding the diagnosis of a hypoglycaemic disorder. In our own data, out of 67 controls, 31 reached glucose plasma concentrations of 0.46-0.54 g/L during a 72-hour fast test, and only 4 reached values within the 0.40-0.45 g/L range (0.41-0.43 g/L) after 24 hours of fast. No control subject reached plasma glucose concentrations below 0.4 g/L. Among 55 patients who were diagnosed to have insulinoma in our institution between 1997 and 2010, and who underwent a fast test, only one patient reached a plasma glucose concentration below 0.55 g/L (0.53 g/L) without further decrease in glucose concentration, and all the other patients who had a positive fast test reached glucose concentrations below 0.45 g/L; the fast test was negative in one patient. In addition, using a 0.45 g/L glucose level threshold increases the diagnostic specificity of concomitant serum insulin, C-peptide and proinsulin levels for the diagnosis of insulinoma (see below). Therefore we think that if the patient remains asymptomatic, the fast test is to be prolonged at least until reaching plasma glucose concentrations below 0.45 g/L, the diagnostic specificity being about

In most insulinoma patients such plasma glucose concentrations are reached within the first 48 hours of the test (Hirshberg et al. 2000; Vezzosi et al. 2007), though the 72 hour-fast test was reported to be necessary to provide a clear diagnostic conclusion in a few patients (Service 1999; Service & Natt 2000). Moreover, additional criteria for the diagnosis of insulinoma using proinsulin and beta-hydroxy-butyrate levels and plasma glucose response to intravenous glucagon were established during a 72-hour (not 48-hour) fast test for patients who did not reach plasma glucose concentrations below 0.45 or 0.55 g/L. Thus in most patients the test to be performed is the classical 72 hour-test. However, 48 hours of fast test are enough to provide diagnostic evidence in almost all insulinoma patients who have a

In the patients who experience symptoms of hypoglycemia only within the hours following a meal, a meal test is to be performed with serial measurement of plasma concentrations of glucose, insulin, C-peptide and proinsulin during the 5 hours following the ingestion of a standardized meal. A mixed meal similar to that which the patient reports to have caused the clinical symptoms is to be used, or a commercial formula mixed meal. Samples should be collected every 30 minutes and those collected for insulin, C-peptide and proinsulin measurement should be sent for analysis only if the sample has been taken at the time when plasma glucose concentration was found to be below 0.6 g/L. Measurement of oral hypoglycaemic agents and anti-insulin antibodies must also be performed if Whipple's triad is demonstrated (Cryer et al. 2009). Post-prandial hypoglycaemia without fasting hypoglycaemia is now known to occur in some insulinoma patients (6% according to Placzkowski et al. 2009) and in patients with noninsulinoma pancreatogenous hypoglycaemia (Service et al. 1999), a syndrome which is part of nesidioblastosis (see below). Others have reported false negative tests in insulinoma patients who are correctly diagnosed after a meal test or oral glucose tolerance test (Sjoberg & Kidd 1992) (Izumiyama et al. 2006; Kar et al. 2006). A mixed meal test was shown to be preferable to the oral glucose tolerance test (Hogan et al. 1983), so that only the meal test was recommended in the 2009

The next requirement for the diagnosis of insulinoma is to provide evidence for

inappropriate insulin secretion at the time of hypoglycaemia.

100% if such levels are reached.

positive fast test.

expert consensus.

Serum insulin concentrations can be artifactually modified by haemolysis, heterophilic antibodies, and endogenous anti-insulin antibodies. Red cells contain an insulin-degrading enzyme which may lead to underestimation of the actual insulin concentrations on haemolysed samples. Most insulin assays are calibrated against the standard IRP 66/304 preparation, and the new 83/500 standard is not yet used by all insulin assays. As a consequence of the differences between the standards and the methods employed, the conversion factor from mIU/L to pmol/L varies from 6.0 to 7.5. In the 2009 expert consensus, the threshold of 3 mIU/L is equal to 18 pmol/L.

Insulin concentrations had been measured for long with non-specific assays that yielded cross-reactions with proinsulin. The insulin assays that are now in use have negligible crossreaction with proinsulin. Though the 3 mIU/L threshold (which replaces a 6 mIU/l threshold previously established with less specific insulin assays) was established for insulin-specific assays, it is known now that the finding of low insulin levels with such assays cannot rule out the diagnosis of insulinoma (Vezzosi et al. 2003). 11-35% patients with insulinoma were found to have insulin levels below the recommended diagnostic threshold (3 mIU/L) at the time of hypoglycemia when insulin was measured with insulinspecific assays, the percentage of values below 3 mIU/L depending on the assay employed (Vezzosi et al. 2007). On the other hand, while the specificity of the 3 mIU/L threshold was found to be 100% with a plasma glucose concentration threshold of 0.45 g/L for the diagnosis of hypoglycaemia related to endogenous hyperinsulinism, it was lowered to 87 % for glucose levels ranging from 0.46 to 0.54 g/L (personal data). The sensitivity and specificity for serum insulin concentrations with a threshold level of at least 3 mIU/l at the time of plasma glucose concentrations below 0.6 g/L were found to be 93% and 95%, respectively (Placzkowski et al. 2009).

#### **C-peptide**

C-peptide and proinsulin levels at the time of hypoglycemia have better diagnostic accuracy than insulin levels. C-peptide assays are calibrated against the 1st standard IRP 84/510 (WHO). According to the expert consensus (Cryer et al. 2009), serum C-peptide levels of at least 0.6 ng/mL concomitant with plasma glucose concentrations below 0.55 g/l are one of the criteria that make the diagnosis of hypoglycaemia related to endogenous hyperinsulinism. In 33 patients who reached plasma glucose concentrations below 0.45 g/L during a fast test, all insulinoma patients had C-peptide levels above 0.6 ng/mL, and only one patient who had previously been treated by left-sided pancreatectomy for nesidioblastosis and who had recurrent hypoglycaemia did not reach the 0.6 ng/mL threshold at the time of hypoglycaemia below 0.45 g/L (Vezzosi et al. 2007). The diagnostic accuracy is improved when blood glucose levels drop below 0.45 g/L: for such blood glucose levels, the sensitivity

Diagnosis and Treatment of Insulinomas in the Adults 141

concentrations (Vezzosi et al. 2007). This 22 pmol/L (0.2 ng/mL) threshold for proinsulin had been previously reported by Hirshberg et al. (Hirshberg et al. 2000). In obese or insulinresistant subjects, proinsulin levels after an overnight fast exceeding 22 pmol/L do not make the diagnosis of insulinoma: in a group of 110 obese subjects who did not present with glycaemic disorders, 23 (21%) had proinsulin levels above 22 pmol/L after an overnight fast (Vezzosi et al. 2007). In conclusion, serum proinsulin levels after an overnight fast below 22 pmol/L do not rule out the diagnosis of insulinoma, while if they are above 22 pmol/L in a non-obese, non-insulin-resistant subject they should lead to a high suspicion of insulinoma, but to date, this can be used as a diagnostic criterion only if plasma glucose levels do not

There are some limitations regarding the insulin, C-peptide and proinsulin threshold levels

The recommended diagnostic thresholds for serum insulin, C-peptide and proinsulin levels are known to be valid only in patients with normal liver and renal function, and there is no study regarding their validity in patients who have concurrent endocrine disorders liable to modify insulin resistance or sensitivity and glucose counter-regulation. In addition, there is no specific study regarding their validity for the diagnosis of recurrence of insulinoma after

In 33 patients with hypoglycaemia related to endogenous hyperinsulinism, the only patient who did not present with a C-peptide level of at least 0.6 ng/mL at the time of hypoglycaemia below 0.45 g/l during a fast test was a patient who had recurrence of nesidioblastosis one year after undergoing left-sided pancreatectomy; we speculated that a lower glucagon production could explain why severe hypoglycaemia was associated with lower insulin and C-peptide levels than in the other patients (Vezzosi et al. 2007). In addition, three of our patients with malignant insulinomas, in whom total remission had been initially achieved by partial pancreatectomy, did not fulfil the criteria for C-peptide levels when they were found to have recurrence of tumour-induced hypoglycaemia

Insulin clearance takes place in the liver (50% of portal insulin is cleared by the liver) and in the kidney (50% of peripheral insulin removal) both by diffusion and receptor-mediated transport, the major route of clearance being receptor-mediated degradation by the renal tubular epithelial cells; the kidney also removes 70% of C-peptide by renal filtration. Therefore, measuring insulin and C-peptide levels in patients with severe renal insufficiency is not helpful; in such patients it has been recommended to use beta-hydroxy-butyrate levels and glucose response to glucagon at the end of a fast test to provide evidence for insulin-

Another valuable tool for the diagnosis of insulinomas is measurement of plasma betahydroxy-butyrate level during the fast test. Insulin is known to suppress ketogenesis. According to the 2009 consensus (Cryer et al. 2009), during a fast test, beta-hydroxy-butyrate

A progressive rise in beta-hydroxy-butyrate levels during the fast test was found to rule out insulinomas (Service & O'Brien 2005). A plasma level of beta-hydroxy-butyrate exceeding 2.7 mmol/L was found to rule out the diagnosis of insulinoma when blood glucose levels were 0.50-0.60 g/L during a 72-hour fast test with both sensitivity and specificity close to

levels of 2.7 mmol/L or less indicate mediation of hypoglycaemia by insulin.

for the diagnosis of hypoglycaemia related to endogenous hyperinsulinism.

exceed 0.6 g/L.

partial pancreatectomy.

(personal data).

**Beta-hydroxy-butyrate** 

mediated hypoglycaemia (Basu et al. 2002).

and specificity of C-peptide (with a threshold of 0.6 ng/mL) are close to 100%, whereas we observed a false positive rate of 9% for C-peptide for concomitant plasma glucose concentrations of 0.46-0.54 g/L (personal data). It had been previously shown that for the 0.50- 0.60 g/L (O'Brien et al.1993) or 0.45-0.60 g/L (Vezzosi et al. 2007) range for plasma glucose concentrations, there was a considerable overlap between patients and controls regarding Cpeptide levels. In the study by Placzowski et al. the sensitivity of the 0.6 ng/mL threshold for the diagnosis of insulinoma is 100 % if concomitant plasma glucose is below 0.6 g/L, while its specificity is 60% for concomitant glucose levels below 0.6 g/L and 78% for concomitant glucose levels of 0.5 g/l or less (Placzkowski et al. 2009). This argues for prolonging the fast test until a glucose level of 0.45 g/L or less is reached.

#### **Proinsulin**

A high proportion of proinsulin is known to be secreted by insulinomas. Intact proinsulin enzymatic processing leads to split 32,33 proinsulin or split 65,66 proinsulin, then to des-31,32-proinsulin (the molecule with the highest serum concentration) and des-64,65 proinsulin and finally to insulin and C-peptide. Proinsulin assays may measure either intact proinsulin (most of them measuring both intact proinsulin and des-64,65-proinsulin) or "total proinsulin" (i.e. intact proinsulin and the other molecules, mostly des-31,32 proinsulin). Therefore the results depend strongly on the method employed. In addition, proinsulin should be measured in serum samples, since results observed in plasma samples may be different (and higher).

According to the 2009 consensus (Cryer et al. 2009), proinsulin levels of at least 5 pmol/L concomitant with plasma glucose concentrations below 0.55 g/L are one of the criteria that make the diagnosis of hypoglycaemia related to endogenous hyperinsulinism. Whether this 5 pmol/L threshold refers to assays measuring "total proinsulin" or "intact proinsulin", or to a specific assay, has not been specified. We used the Human Proinsulin RIA kit (Linco Research), which measures "total proinsulin" (in fact intact proinsulin and des-31,32-proinsulin, the cross-reactions with des-64,65-proinsulin, insulin and C-peptide being < 0.1%), to measure proinsulin in 33 patients with hypoglycaemia related to endogenous hyperinsulinism and 67 controls who underwent a 72-hour fast test. 100% sensitivity and specificity were found using a 5 pmol/L threshold for serum proinsulin and a 0.45 g/l threshold for plasma glucose concentration (Vezzosi et al. 2007). The specificity would have been much lowered if the 5 pmol/L threshold had been applied to the 0.46-0.54 g/L range for plasma glucose concentrations: 9 out of 31 control subjects (29%) who reached this range of plasma glucose concentrations during the prolonged fast were found to have proinsulin levels above 5 pmol/L (personal data). For this threshold of 5 pmol/L, Placzowski et al reported a 100%-sensitivity for the diagnosis with plasma glucose concentrations below 0.6 g/L, whereas the specificity was only 68% and 78% with glucose levels below 0.6 g/L and below 0.5 g/L, respectively (Placzkowski et al. 2009). This also argues for prolonging the fast test until a glucose level of 0.45 g/L or less is reached.

In addition, serum proinsulin levels above 22 pmol/L can make the diagnosis of hypoglycaemia related to endogenous hyperinsulinism when concomitant plasma glucose concentration is below 0.6 g/L with 74% sensitivity and 100% specificity. Moreover, in nonobese subjects (body mass index less than 30 kg/m2) the diagnostic specificity of a 22 pmol/L for serum proinsulin levels measured after an overnight fast was found to be 100%, while its sensitivity was 70%, with concomitant low or normal plasma glucose

and specificity of C-peptide (with a threshold of 0.6 ng/mL) are close to 100%, whereas we observed a false positive rate of 9% for C-peptide for concomitant plasma glucose concentrations of 0.46-0.54 g/L (personal data). It had been previously shown that for the 0.50- 0.60 g/L (O'Brien et al.1993) or 0.45-0.60 g/L (Vezzosi et al. 2007) range for plasma glucose concentrations, there was a considerable overlap between patients and controls regarding Cpeptide levels. In the study by Placzowski et al. the sensitivity of the 0.6 ng/mL threshold for the diagnosis of insulinoma is 100 % if concomitant plasma glucose is below 0.6 g/L, while its specificity is 60% for concomitant glucose levels below 0.6 g/L and 78% for concomitant glucose levels of 0.5 g/l or less (Placzkowski et al. 2009). This argues for prolonging the fast

A high proportion of proinsulin is known to be secreted by insulinomas. Intact proinsulin enzymatic processing leads to split 32,33 proinsulin or split 65,66 proinsulin, then to des-31,32-proinsulin (the molecule with the highest serum concentration) and des-64,65 proinsulin and finally to insulin and C-peptide. Proinsulin assays may measure either intact proinsulin (most of them measuring both intact proinsulin and des-64,65-proinsulin) or "total proinsulin" (i.e. intact proinsulin and the other molecules, mostly des-31,32 proinsulin). Therefore the results depend strongly on the method employed. In addition, proinsulin should be measured in serum samples, since results observed in plasma samples

According to the 2009 consensus (Cryer et al. 2009), proinsulin levels of at least 5 pmol/L concomitant with plasma glucose concentrations below 0.55 g/L are one of the criteria that make the diagnosis of hypoglycaemia related to endogenous hyperinsulinism. Whether this 5 pmol/L threshold refers to assays measuring "total proinsulin" or "intact proinsulin", or to a specific assay, has not been specified. We used the Human Proinsulin RIA kit (Linco Research), which measures "total proinsulin" (in fact intact proinsulin and des-31,32-proinsulin, the cross-reactions with des-64,65-proinsulin, insulin and C-peptide being < 0.1%), to measure proinsulin in 33 patients with hypoglycaemia related to endogenous hyperinsulinism and 67 controls who underwent a 72-hour fast test. 100% sensitivity and specificity were found using a 5 pmol/L threshold for serum proinsulin and a 0.45 g/l threshold for plasma glucose concentration (Vezzosi et al. 2007). The specificity would have been much lowered if the 5 pmol/L threshold had been applied to the 0.46-0.54 g/L range for plasma glucose concentrations: 9 out of 31 control subjects (29%) who reached this range of plasma glucose concentrations during the prolonged fast were found to have proinsulin levels above 5 pmol/L (personal data). For this threshold of 5 pmol/L, Placzowski et al reported a 100%-sensitivity for the diagnosis with plasma glucose concentrations below 0.6 g/L, whereas the specificity was only 68% and 78% with glucose levels below 0.6 g/L and below 0.5 g/L, respectively (Placzkowski et al. 2009). This also argues for prolonging the fast test until a glucose level of 0.45 g/L or less is

In addition, serum proinsulin levels above 22 pmol/L can make the diagnosis of hypoglycaemia related to endogenous hyperinsulinism when concomitant plasma glucose concentration is below 0.6 g/L with 74% sensitivity and 100% specificity. Moreover, in nonobese subjects (body mass index less than 30 kg/m2) the diagnostic specificity of a 22 pmol/L for serum proinsulin levels measured after an overnight fast was found to be 100%, while its sensitivity was 70%, with concomitant low or normal plasma glucose

test until a glucose level of 0.45 g/L or less is reached.

**Proinsulin** 

reached.

may be different (and higher).

concentrations (Vezzosi et al. 2007). This 22 pmol/L (0.2 ng/mL) threshold for proinsulin had been previously reported by Hirshberg et al. (Hirshberg et al. 2000). In obese or insulinresistant subjects, proinsulin levels after an overnight fast exceeding 22 pmol/L do not make the diagnosis of insulinoma: in a group of 110 obese subjects who did not present with glycaemic disorders, 23 (21%) had proinsulin levels above 22 pmol/L after an overnight fast (Vezzosi et al. 2007). In conclusion, serum proinsulin levels after an overnight fast below 22 pmol/L do not rule out the diagnosis of insulinoma, while if they are above 22 pmol/L in a non-obese, non-insulin-resistant subject they should lead to a high suspicion of insulinoma, but to date, this can be used as a diagnostic criterion only if plasma glucose levels do not exceed 0.6 g/L.

There are some limitations regarding the insulin, C-peptide and proinsulin threshold levels for the diagnosis of hypoglycaemia related to endogenous hyperinsulinism.

The recommended diagnostic thresholds for serum insulin, C-peptide and proinsulin levels are known to be valid only in patients with normal liver and renal function, and there is no study regarding their validity in patients who have concurrent endocrine disorders liable to modify insulin resistance or sensitivity and glucose counter-regulation. In addition, there is no specific study regarding their validity for the diagnosis of recurrence of insulinoma after partial pancreatectomy.

In 33 patients with hypoglycaemia related to endogenous hyperinsulinism, the only patient who did not present with a C-peptide level of at least 0.6 ng/mL at the time of hypoglycaemia below 0.45 g/l during a fast test was a patient who had recurrence of nesidioblastosis one year after undergoing left-sided pancreatectomy; we speculated that a lower glucagon production could explain why severe hypoglycaemia was associated with lower insulin and C-peptide levels than in the other patients (Vezzosi et al. 2007). In addition, three of our patients with malignant insulinomas, in whom total remission had been initially achieved by partial pancreatectomy, did not fulfil the criteria for C-peptide levels when they were found to have recurrence of tumour-induced hypoglycaemia (personal data).

Insulin clearance takes place in the liver (50% of portal insulin is cleared by the liver) and in the kidney (50% of peripheral insulin removal) both by diffusion and receptor-mediated transport, the major route of clearance being receptor-mediated degradation by the renal tubular epithelial cells; the kidney also removes 70% of C-peptide by renal filtration. Therefore, measuring insulin and C-peptide levels in patients with severe renal insufficiency is not helpful; in such patients it has been recommended to use beta-hydroxy-butyrate levels and glucose response to glucagon at the end of a fast test to provide evidence for insulinmediated hypoglycaemia (Basu et al. 2002).

#### **Beta-hydroxy-butyrate**

Another valuable tool for the diagnosis of insulinomas is measurement of plasma betahydroxy-butyrate level during the fast test. Insulin is known to suppress ketogenesis. According to the 2009 consensus (Cryer et al. 2009), during a fast test, beta-hydroxy-butyrate levels of 2.7 mmol/L or less indicate mediation of hypoglycaemia by insulin.

A progressive rise in beta-hydroxy-butyrate levels during the fast test was found to rule out insulinomas (Service & O'Brien 2005). A plasma level of beta-hydroxy-butyrate exceeding 2.7 mmol/L was found to rule out the diagnosis of insulinoma when blood glucose levels were 0.50-0.60 g/L during a 72-hour fast test with both sensitivity and specificity close to

Diagnosis and Treatment of Insulinomas in the Adults 143

plasma glucose concentrations of 0.45 g/L (2.5 mmol/L) or less, or after 72 hours of fast: plasma glucose responses to the intravenous injection of 1 mg glucagon are measured at 10 min-intervals for 30 min. Under these conditions, plasma glucose response to glucagon was found to be higher in insulinoma patients than in controls with a range of maximal increase in plasma glucose concentrations of 1.4-5.4 mmol/L in insulinoma patients and of 0.1-1.3 mmol/L in controls, as a consequence of a lesser depletion of hepatic glycogen in insulinoma patients during the fast, which is related to the higher insulin secretion in the patients. This allowed a clear distinction between patients and controls who has reached plasma glucose

On the other hand, the glucagon test has long been known to be a provocative test in insulinoma patients: glucagon may trigger severe hypoglycaemia related to insulin secretion in insulinoma patients. This provocative test was found to be helpful for the diagnosis of

\* Regarding the various tests employed in the evaluation of a patient with suspected hypoglycaemic disorder, only the fast test, with a glucagon test at its termination, in patients with symptoms of fasting hypoglycaemia, and the mixed meal test in patients with postprandial hypoglycaemia are recommended by the 2009 expert consensus (Cryer et al.
