Chapter 5 – PANCREATIC ISLET FUNCTION TESTS

Ruth S. Weinstock, MD, PhD - Professor of Medicine, SUNY Upstate Medical University, Syracuse, NY

Steven V. Zygmont, MD - Assistant Professor of Medicine, SUNY Upstate Medical University, Syracuse, NY

Updated 10 July 2009

TO OBTAIN A COMPLETE DOWNLOAD OF THIS CHAPTER IN PDF FORMAT,  CLICK HERE

A. Evaluation of glucose tolerance: screening for diabetes mellitus and diagnosing diabetes mellitus

Screening for diabetes

Early detection and treatment of diabetes mellitus is important in preventing the chronic and acute complications of this disease. Individuals with symptoms suggestive of hyperglycemia, such as polyuria, polyphagia, polydipsia, unexplained weight loss, blurred vision, excessive fatigue, or infections or wounds that heal poorly, should be promptly tested.

The American Diabetes Association recommends routinely screening all adults for type 2 diabetes beginning at age 45. In asymptomatic people, testing for type 2 diabetes should be considered in adults of any age if they are overweight or obese (BMI >= 25kg/m2), with repeat screening every three years (1). The American Association of Clinical Endocrinologists recommends screening individuals at risk beginning at age 30 (2). Fasting plasma glucose is the preferred screening test (see below Sections 1 and 2), but the optimal screening methods, cutoff points, and screening intervals remain controversial (3).

Risk Factors for the Development of Type 2 Diabetes

Type 2 diabetes is becoming a growing problem in children and adolescents in high-risk populations. To address this issue, the American Diabetes Association recommends screening children every 3 years, beginning at age 10 or the onset of puberty, if they have 2 or more risk factors listed below and are overweight (1).

Risk Factors for Type 2 Diabetes in Children and Adolescents

Diagnosing diabetes

The diagnosis of diabetes can be made using the fasting plasma glucose, casual plasma glucose, or the oral glucose tolerance test (4). Testing should be performed on 2 separate days using one or more of the above tests, unless evidence for acute metabolic decompensation is present.

The use of the hemoglobin A1c (HbA1c) assay for the diagnosis of diabetes is controversial. HbA1c levels reflect overall glycemic control and correlate with the development of microvascular complications. A recent International Expert Committee suggests that HbA1c > 6.5% on two separate occasions can be used to diagnose diabetes, where as 6.0% to less than 6.5% identifies high risk of developing diabetes (5). The HbA1c assay is discussed in more detail below (see #6).

1. FASTING AND CASUAL PLASMA GLUCOSE

Fasting plasma glucose is the preferred screening test recommended by the American Diabetes Association for the diagnosis of diabetes in children and non pregnant adults (1). If abnormal, the test should be repeated on at least one additional day to confirm the diagnosis of diabetes (4). The test should be performed after an 8 hour fast. For routine clinical practice, fasting plasma glucose is preferred over the oral glucose tolerance test for the diagnosis of diabetes because it is rapid, easier to administer, is more convenient for patients and providers, and has a lower cost (1). The use of the oral glucose tolerance test, however, may be warranted in high risk individuals with normal fasting plasma glucose levels.

  

Fasting Plasma Glucose

Normal glucose tolerance 

<100 mg/dl (5.6 mmol/l)

Impaired fasting glucose 

("pre-diabetes")

100-125 mg/dl (5.6-6.9 mmol/l)

Diabetes mellitus 

>125 mg/dl (7.0 mmol/l)

A casual plasma glucose level, which is obtained at any time of the day regardless of the time of the last meal, can be used in individuals with symptoms of hyperglycemia. A casual plasma glucose level of >200 mg/dl (11.1 mmol/l) is diagnostic of diabetes. In the absence of acute metabolic decompensation, the diagnosis should be confirmed by repeating the casual plasma glucose or by obtaining a fasting plasma glucose level on at least one additional occasion.

For the diagnosis of diabetes, standard venous plasma glucose specimens should be obtained. Specimens should be processed promptly, since glucose is metabolized at room temperature and the process is influenced by storage temperature and storage time. This breakdown is accelerated in the presence of bacteria or leukocytosis. If serum is in contact with cells for a prolonged length of time (>30 min), the addition of a preservative such as sodium fluoride is recommended.

Whole blood glucose specimens obtained with point-of-care devices should not be used for the diagnosis of diabetes because of the inaccuracies associated with these methods. Capillary and venous whole blood glucose concentrations are approximately 15% lower than plasma glucose levels in fasting specimens.

2. ORAL GLUCOSE TOLERANCE TEST (OGTT)

a) OGTTs for the diagnosis of diabetes in non-pregnant individuals

Formal oral glucose tolerance tests are usually not necessary to establish the diagnosis of diabetes mellitus. They are more cumbersome and costly than the fasting plasma glucose test and therefore are not routinely performed in clinical practice. There is concern, however, that the use of only the fasting plasma glucose may not identify a proportion of individuals with impaired glucose tolerance or diabetes. A plasma glucose level 2 hours after a glucose challenge may identify additional individuals with abnormal glucose tolerance who are at risk for microvascular and macrovascular complications, particularly in high risk populations in which postprandial (versus fasting) hyperglycemia is evident early in the disease (6,7).

When using an OGTT, the criteria for the diagnosis of diabetes is a 2 h glucose >200 mg/dl (11.1 mmol/l) after a 75 gram oral glucose load (American Diabetes Association and World Health Organization criteria). The 75 gram glucose load should be administered when the patient has ingested at least 150 grams of carbohydrate for the 3 days preceding the test and after an overnight fast. Dilution of the 75 gram oral glucose load (300-900 ml) may improve acceptability and palatability without compromising reproducibility (8). The patient should not be acutely ill or be taking drugs that affect glucose tolerance at the time of testing, and should abstain from tobacco, coffee, tea, food, alcohol and vigorous exercise during the test.

  

2-h Plasma Glucose (after 75 gram Glucose Load)

Normal glucose tolerance 

<140 mg/dl (7.8 mmol/l)

Impaired glucose tolerance

("pre-diabetes") 

140-199 mg/dl (7.8-11.1 mmol/l)

Diabetes mellitus 

≥200 mg/dl (11.1 mmol/l)

b) OGTTs for the diagnosis of gestational diabetes

It has been estimated that gestational diabetes mellitus is present in approximately 7% of pregnancies, which is >200,000 pregnancies/year (9,10). The prevalence increases with increased number of risk factors, such that 33% of women with 4 or more risk factors have gestational diabetes (9). This condition is important to diagnose early because of the increased perinatal morbidity associated with poor glycemic control. More than one method has been recommended for the screening and diagnosis of gestational diabetes. The criteria for the diagnosis of this condition remain controversial because the glucose thresholds for the development of complications in pregnancies with diabetes remain poorly defined.

To routinely screen for gestational diabetes, the American Diabetes Association recommends either the "one-step" or "two-step" approach (4). For the one-step approach a 100 g OGTT is performed. In the two-step approach, first a 50 gram oral glucose load is administered regardless of the timing of previous meals. The test is considered abnormal if the 1 hour postload glucose level is > 140 mg/dl (7.8 mmol/l), identifying 80% of women with gestational diabetes. If a 1 hour postload glucose level of >130 mg/dl (7.2 mmol/l) is used as the cutoff, approximately 90% of women with gestational diabetes will be identified. Women with abnormal glucose levels with this screening test then undergo a formal 100 g OGTT.

1-h Plasma Glucose (after 50 gram Glucose Load)

Abnormal initial screening test for gestational diabetes 

≥140mg/dl (7.8 mmol/l)

If a woman has a fasting plasma glucose >126 mg/dl (7.0 mmol/l) or a casual plasma glucose >200 mg/dl (11.1 mmol/l), the diagnosis of diabetes is established and an OGTT is unnecessary. If a woman is at high risk, the screening test can be omitted and the diagnostic OGTT performed directly.

Screening for gestational diabetes is performed routinely between 24 and 28 weeks gestation. If the woman is at high risk, however, screening should be performed at an earlier stage.

Risk Factors for the Development of Gestational Diabetes

Low risk women, defined as having all of the following attributes, need not undergo routine screening tests per the latest American Diabetes Association guidelines (4). In one study, it was estimated that this would decrease the number of screenings by 10% and only fail to detect 3% of cases (11). The American College of Endocrinology, however, still recommends screening all pregnant women for diabetes (2).

Low Risk for the Development of Gestational Diabetes

If the screening test is abnormal, the diagnosis of gestational diabetes should be confirmed using a formal OGTT. The OGTT should be performed after an overnight (8-14 h) fast. It is generally recommended that the woman ingest at least 150 grams of carbohydrate/day for the 3 days prior to testing to prevent false positive results. The necessity of this preparatory diet in normally nourished women, however, has been challenged (12).

Risk of Development of Gestational Diabetes 

Time of Initial Testing for Gestational Diabetes

Low risk 

Routine testing not necessary (4)

24-28 weeks gestation (2)

Average risk 

24-28 weeks gestation

High risk 

As soon as feasible; repeat at 24-28 weeks if earlier testing normal



The preferred diagnostic test for gestational diabetes is the 100 gram 3 hour OGTT. The American Diabetes Association recently adapted more stringent cutoff values when compared to the older recommendations from the National Diabetes Data Group (4, 10, 13). The American Diabetes Association, using the original work of O’Sullivan and Mahan and the Carpenter and Coustan modifications, suggests that at least 2 of the following 4 venous plasma glucose levels must be attained or exceeded (4,10):

Diagnosis of Gestational Diabetes

100 g glucose load test 

American Diabetes Association 

National Diabetes Data Group

Fasting glucose 

≥95 mg/dl (5.3 mmol/l) 

≥105 mg/dl (5.8 mmol/l)

1 hour glucose 

≥180 mg/dl (10.0 mmol/l) 

≥190 mg/dl (10.6 mmol/l)

2 hour glucose 

≥155 mg/dl (8.6 mmol/l) 

≥165 mg/dl (9.2 mmol/l)

3 hour glucose 

≥140 mg/dl (7.8 mmol/l) 

≥145 mg/dl (8.1 mmol/l)

If the 3rd hour glucose is omitted, the sensitivity of this test is lowered by 13% (14). This "2 tiered" approach (1 hour 50 gram glucose load screening test followed by the 3 hour 100 gram OGTT in women with abnormal screen results) has been endorsed by the National Diabetes Data Group, the American College of Obstetricians and Gynecologists, and the American Diabetes Association, and has been shown to be cost-effective (15).

The 75 gram OGTT is advocated by the World Health Organization in the "one-tiered" approach but is less well validated than the 100 gram test. The World Health Organization uses cutoffs of fasting plasma glucose > 126 mg/dl (7.0 mmol/l) or 2 hour post load glucose > 140 mg/dl (7.8 mmol/l). The American Diabetes Association, in contrast, requires that at least 2 of the 3 venous plasma glucose levels be attained or exceeded to diagnose gestational diabetes as shown below (4). A large observational study found no difference in adverse pregnancy outcomes when gestational diabetes was diagnosed using the American Diabetes Association or World Health Organization criteria (16).

Diagnosis of Gestational Diabetes

75 g glucose load test 

American Diabetes Association 

World Health Organization

Fasting glucose 

≥95 mg/dl (5.3 mmol/l) 

≥126 mg/dl (7.0 mmol/l)

1 hour glucose 

≥180 mg/dl (10.0 mmol/l)

------------

2 hour glucose 

≥155 mg/dl (8.6 mmol/l) 

≥140 mg/dl (7.8 mmol/l)

c) Postpartum testing of women with gestational diabetes

The incidence of abnormal glucose tolerance one-year after gestational diabetes has been reported in ranges of 7-57% and 3-38% for diagnosed diabetes. Women at the highest risk are those with multiple risk factors, those who had more severe gestational diabetes and those with poorer beta cell function (9).

The American Diabetes Association recommends testing women 6-12 weeks after delivery (1). Recommended studies include a fasting plasma glucose level or a 75 g oral glucose tolerance test (described in Sections 1 and 2 above). Women with normal results should be retested every 3 years or sooner (1). It is recommended that subjects with impaired fasting glucose or impaired glucose tolerance be retested on a yearly basis (4,10).

3. INTRAVENOUS GLUCOSE TOLERANCE TEST

The short intravenous glucose tolerance test (IVGTT) is used in research studies to assess first phase insulin release. This acute insulin secretory response is typically lost early in the development of both type 1 and type 2 diabetes due to reduction of beta cells and islet dysfunction. Abnormal IVGTT results can occur prior to the onset of the diabetes. The test is performed after an overnight 10 h fast, and the patients are instructed to ingest at least 150 grams of carbohydrate for the 3 days preceding the test. A 25 gram glucose bolus (of a 25% glucose solution) is given intravenously, and the acute insulin response calculated from the third to fifth minute after the glucose bolus. The short intravenous glucose tolerance test is sometimes used to assess pancreatic function after pancreatic transplantation.

In the Diabetes Prevention Trial – Type 1, a glucose load was given intravenously (0.5 g/kg body weight up to a maximum of 35 grams) over 3 minutes, and insulin levels at 1 and 3 minutes post-load were used to estimate acute insulin production (17). Individuals with low insulin response (<100 uU/ml) and positive autoantibodies were at high risk of developing type 1 diabetes. Until effective interventions are established, however, the routine use of this test for the detection of early type 1 diabetes is not recommended.

The standard intravenous glucose tolerance test is used in research studies to estimate insulin sensitivity (SI) and glucose effectiveness (SG) using minimal model methodology. The procedure for the standard intravenous glucose tolerance test is to intravenously inject glucose (0.33 g/kg body weight) over 2 minutes and to frequently sample for glucose and insulin over 3-4 hours. Modifications include the addition of a tolbutamide (125 mg/m2) or insulin (20-30 mU/kg) infusion 20-25 minutes after the glucose load. These tests are not used in clinical practice.

4. C-PEPTIDE TESTING

During the processing of proinsulin to insulin in the beta cell of the pancreas, the 31 amino acid connecting peptide which connects the A and B chains, called c-peptide, is enzymatically removed and secreted into the portal vein. C-peptide circulates independently from insulin and is mainly secreted by the kidneys. Levels are elevated in renal failure. Standardization of different c-peptide assays is still suboptimal. The major use of c-peptide measurements is in the evaluation of hypoglycemia.

C-peptide testing has been used in studies to examine insulin secretory reserve. In type 1 diabetes, there is progressive loss of c-peptide with progressive destruction of the beta cells in the islets of the pancreas, until eventually levels are extremely low or undetectable. In type 2 diabetes, there is also a progressive loss of beta cell function over many years, with progressive loss of insulin secretory capacity and decreasing c-peptide levels. Fasting and glucose-stimulated c-peptide levels have used to distinguish type 1 (severe insulin deficiency) from type 2 diabetes with limited success and poor discrimination.

C-peptide stimulation using glucagon or a mixed meal such as Sustacal has also been used to help differentiate between type 1 and type 2 diabetes, and to determine the need for insulin therapy in type 2 diabetes. In the glucagon stimulation test, glucose, insulin and c-peptide levels are measured 6 and 10 min after the intravenous injection of 1 mg of glucagon. Normal stimulation of c-peptide is a 150-300% elevation over basal levels. In the mixed meal tolerance test, Sustacal (6 mg/kg up to a maximum or 360 ml) is ingested over 5 minutes, and glucose and c-peptide are measured 90 min after oral ingestion.

These tests have had limited general clinical utility since they do not reliably discriminate between patients who require insulin therapy. They have been used in research studies and in the evaluation of patients after pancreatectomy and pancreatic transplantation. In the Diabetes Control and Complications Trial, a basal c-peptide value of <0.2 pmol/ml and stimulated level of <0.5 pmol/ml were used to confirm the presence of type 1 diabetes at entry (18).

5. AUTOANTIBODIES

Islet autoantibodies can be detected early in the development of type 1 diabetes, and are considered markers of autoimmune beta cell destruction. They predict progressive beta cell destruction and ultimately beta cell failure. The autoantibodies for which specific immunoassays are available include the 65-KDa isoform of glutamic acid decarboxylase (GAD65), insulin autoantibodies (IAA), islet cell antigen 512 autoantibodies (ICA512) and autoantibodies to the tyrosine phosphatase related antigens IA-2 and IA-2b. ICA512 are autoantibodies to parts of the IA-2 antigen. The presence of high levels of 2 or more antibodies is strongly predictive of type 1 diabetes mellitus. These antibodies may be detected before the onset of type 1 diabetes and at the time of diagnosis, and have been used in screening for type 1 diabetes in research studies related to the early detection, treatment and prevention of type 1 diabetes. These measurements are not recommended for use in general screening programs since effective prevention interventions are yet to be available. In the past these assays have not been standardized and some do not have well-established cutoff values. The Immunology of Diabetes Society and The Centers for Disease Control and Prevention have developed the Diabetes Antibody Standardization Program and have reported the results from their first Assay Proficiency Evaluation (19).

Commercially available assays for autoantibodies are sometimes useful in distinguishing early type 1 diabetes from type 2 diabetes. The absence of detection of these antibodies, however, does not exclude the diagnosis of type 1 diabetes. Since insulin antibodies can form in response to insulin therapy, detection can be the result of insulin injections or autoimmune insulin antibody formation. GAD65 antibodies are frequently observed early in the course of type 1 diabetes. They are also present in the rare neurological disorder, stiff-man syndrome, and in some patients with polyendocrine autoimmune disease. The GAD65 assay is considered more sensitive than the ICA assay for the detection of early type 1 diabetes in adults, whereas IAA are more common in young children who develop type 1 diabetes. Although it has been suggested that GAD65 and IA-2 positivity show high diagnostic specificity for type 1 diabetes and may be helpful in determining which type 2 diabetes patients require insulin therapy (or really have type 1 diabetes), the cost-effectiveness of this approach has not been established.

B. Evaluation of glycemic control in diabetes mellitus

6. HEMOGLOBIN A1C

Glycosylated hemoglobin, or the hemoglobin A1c (HbA1c) assay, is the most widely accepted laboratory test for the measurement of glycemic control and is recommended for routine use in the management of patients with diabetes mellitus. HbA1c levels reflect average blood glucose levels over the preceding 2-3 months. Although the life span of erythrocytes is approximately 120 days, HbA1c levels represent a "weighted" average of blood glucose levels, with youngest red blood cells, reflecting mean glucose levels over the past month, contributing to a greater extent then older ones.

The International Federation of Clinical Chemistry Working Group on HbA1c defines the HbA1c as the hemoglobin A that is irreversibly non-enzymatically glycosylated at one or both N-terminal valines of the beta-chains of the hemoglobin. Multiple methods have been certified to measure HbA1c, including immunoassay, ion-exchange, high performance liquid chromatography (HPLC), boronate affinity HPLC, electrophoresis, isoelectric focusing, colorimetry and enzyme methods. The National Glycohemoglobin Standardization Program, which was started in 1996, issued their 5 year progress report in 2001 (20). They have been largely successful in their goal to standardize HbA1c assays throughout the United States to the HPLC method used in the Diabetes Control and Complications Trial. The process has involved certification and proficiency testing, and long-term monitoring of quality control data. Providers should only use laboratories that are certified by the National Glycohemoglobin Standardization Program. Information concerning certified methods and laboratories can be found on their website www.ifcchba1c.net.

In 2007 a consensus statement on the international standardization of HbA1c assays was issued by the American Diabetes Association, the European Association for the Study of Diabetes, the International Federation of Clinical and Laboratory Medicine and the International Diabetes Federation (21). A1c assays will be calibrated to a new reference method and results reported in a standardized manner [A1c (%); A1c (mmol/mol), and estimated average glucose.

The American Diabetes Association recommends determining HbA1c levels every 3 – 6 months to monitor glycemic control. Reducing the HbA1c level to as close to normal as possible is directly related to the reduction of the chronic complications of diabetes in several studies (18, 22-23). The American Diabetes Association goal HbA1c is <7% but states that "more stringent goals (i.e. a normal A1c <6%) can be considered in individual patients" (1). It also states "less stringent A1c goals may be appropriate for patients with a history of severe hypoglycemia, limited life expectancy, advanced microvascular complications, extensive comorbid conditions and those in whom the general goal is difficult to attain" (1). The American Association of Clinical Endocrinologists lowered their target HbA1c to 6.5% or as close to normal without causing significant hypoglycemia (2).

The international A1C-Derived Average Glucose Study (ADAG) utilized frequent self monitoring of blood glucose and continuous glucose monitoring in adults with type 1 diabetes, type 2 diabetes, and no diabetes. The study described a linear relationship between HbA1C and average glucose level ().


In the ADAG study, there was no significant difference in the regression lines between HbA1C and mean glucose levels among ethnic and racial groups, although there was a trend toward a difference in regression lines between African/African-American and Caucasian adults (24). In children with type 1 diabetes the correlation between HbA1C and mean blood glucose levels was significant but lower then in the ADAG trial (25). The American Diabetes Association and the American Association of Clinical Chemists have determined that this correlation is strong enough to justify reporting HbA1C results in association with estimated average glucose (eAG) results.

Based upon results obtained in the ADAG study, the mean glucose levels can be estimated by the following formula: 1.583 x HbA1c – 2.52 x 18. HbA1c levels correlate with mean blood glucose levels as follows (24):

Hemoglobin A1c (%)

Approximate Mean Plasma Glucose


(mg/dl)

(mmol/l)




6

126

7.0

7

154

8.6

8

183

10.2

9

212

11.8

10

240

13.4

11

269

14.9

12

298

16.5

Depending upon the assay method being used, certain hemoglobinopathies may interfere with results. This problem is highly method-dependent. Inaccurate results may be obtained in the presence of salicylates, chronic alcohol or opiate use, hyperbilirubinemia, liver or renal disease, iron deficiency, vitamin C, vitamin E, hypertriglyceridemia and lead poisoning, and when there are conditions of abnormal red blood cell turnover such as in anemia or hemolysis.

In the past, the American Diabetes Association did not recommend the use of HbA1c testing as a screening test for the diagnosis of diabetes, but this position has been controversial. This decision was due, in part, to the low sensitivity of the test in some studies, problems in the past with standardization of the HbA1c assay, and inaccuracies in the presence of abnormal red blood cell turnover. A benefit of the use of HbA1c for the diagnosis of diabetes is that the test is easy to perform, does not have to be performed in the fasting state, and does not require any special preparation. In one study, HbA1c at a cutoff of 2 SD above the normal mean had moderate sensitivity (63.2%) and high specificity (97.4%) as a screening test for undiagnosed diabetes (26). Higher sensitivity (83.6%) was observed in the Mexican-American population (26). As noted above (see "Diagnosing Diabetes"), an Expert Committee suggests that a HbA1c > 6.5% on two occasions is diagnostic of diabetes and 6.0% to <6.5% identifies individuals at high risk of developing diabetes (5).

7. FRUCTOSAMINE

Assays of glycated serum proteins, which mostly measure glycated serum albumin, can reflect short-term glycemic control. The fructosamine assay is most commonly used. Since albumin has a short half-life (14-20 days), this test indicates average blood glucose levels over the past few weeks, which can be helpful in certain conditions such as pregnancy (27). These tests may be affected by hypertriglyceridemia, hyperbilirubinemia, and hemolysis as well as by low serum protein and albumin levels, but it is unclear if fructosamine results should be corrected for serum albumin or protein levels.

There is a lack of studies demonstrating the usefulness of the fructosamine assay in predicting the development of diabetes-related complications. Since the clinical usefulness is not well established, fructosamine testing is generally recommended in situations where HbA1c testing is expected to be inaccurate, such as in the presence of hemoglobinopathies.

C. Evaluation of hypoglycemia

Symptomatic hypoglycemia is defined clinically using Whipple’s triad, which includes the presence of symptoms (confusion, lightheadedness, loss of consciousness, seizure, aberrant behavior, sweating, palpitations, weakness, blurred vision or hunger) at the time of a low plasma glucose level, and improvement of symptoms with plasma glucose levels returning to normal (28). The physician should determine if the patient truly has hypoglycemia prior to seeking an etiology. A plasma glucose level < 50 mg/dl (2.8 mmol/l) should raise the suspicion for a hypoglycemic disorder and initiate further evaluation, but many authorities rely on a glucose <40 mg/dl (2.2 mmol/l) as being diagnostic (29). Although symptoms are commonly observed when plasma glucose falls to <55 mg/dl (3.1 mmol/l), levels of <45 mg/dl (2.5 mmol/l) are associated with cognitive dysfunction (neuroglycopenia). Capillary glucose determinations should not be used in the evaluation of hypoglycemic disorders due to their poor accuracy in these situations.

The Endocrine Society has published clinical practice guidelines for the evaluation and management of hypoglycemic disorders (30). In persons without diabetes, drugs, critical illnesses, hormone deficiencies and nonislet cell tumors should be considered based on the clinical findings. If the cause of the hypoglycemia is not evident, plasma glucose, insulin, c-peptide, proinsulin, β-hydroxybutyrate, insulin antibodies and screen for oral hypoglycemic drugs should be obtained during an episode of spontaneous hypoglycemia. Glucagon 1 mg IV should then be administered with careful follow of the glucose response. This will help determine if the condition is related to excessive endogenous insulin production. The diagnosis of pancreatic hyperinsulinemic hypoglycemia is supported by the demonstration that insulin secretion is not suppressed normally when the patient develops hypoglycemia. If testing cannot be conducted during an episode of spontaneous hypoglycemia, the prolonged fast or mixed meal test followed by the administration of glucagon is the most useful diagnostic study.

8. PROLONGED FAST

The "gold standard" test in the evaluation of hypoglycemia is the 72-hour supervised fast. The purpose of the fast is twofold. The first is to diagnose hypoglycemia as the cause of the patient’s symptoms. The second is an attempt to determine the etiology of the hypoglycemia. Due to the risk of hypoglycemia, patients should be admitted to the hospital to undergo the fast in a monitored setting. The fast could be initiated in a carefully monitored outpatient facility, with the patient entering the hospital if the fast is not terminated prior to the closing of the site. Baseline bloodwork can also include cortisol, growth hormone, glucagon and catecholamines if deficient counterregulation is suspected.

During the fast, patients are allowed no food but can consume non-caloric caffeine-free beverages for up to 72 hours. Simultaneous insulin, c-peptide and glucose samples are obtained at the beginning of the fast and every 4-6 hours thereafter. Once the plasma glucose falls to <60 mg/dl, specimens should be taken every 1-2 hours under close supervision. Patients should continue activity when they are awake. The fast continues until the plasma glucose falls below 45 mg/dl (2.5 mmol/l) [plasma glucose <55 mg/dl is recommended in the most recent Endocrine Society guidelines (30)] and symptoms of neuroglucopenia develop, at which time, insulin, glucose, c-peptide, sulfonylurea/meglitinide, proinsulin and beta-hydroxybutyrate levels are obtained and the fast is terminated (29). Additional samples for insulin antibodies, anti-insulin receptor antibodies, IGF-1/IGF-2 and plasma cortisol, glucagon or growth hormone can also be obtained at this time if a non-islet cell tumor, autoimmune etiology, or hormone deficiency is suspected.

The diagnosis of insulinoma is likely if the patient, at the conclusion of the fast, has neuroglycopenic symptoms, a fall in plasma glucose to <55 mg/dl, inappropriately elevated beta-cell polypeptides (insulin, proinsulin and c-peptide levels; see below table), beta-hydroxybutyrate level <2.7 mmol/l, and undetectable sulfonylurea/meglitinide levels (30-31).

Diagnostic Criteria for Insulinoma During Symptomatic Hypoglycemia
After a Prolonged Fast

Assay Method 

Insulin 

Proinsulin 

C-Peptide

Radioimmunoassay (RIA)1 

≥6 uU/ml (43 pmol/l)

 

 

Immunochemiluminometric assay (ICMA)2 

≥3 uU/ml 

≥5 pmol/l 

≥200 pmol/l

1RIA: insulin assay sensitivity of 5 uU/ml
2ICMA: insulin assay sensitivity of < 1 uU/ml

Approximately 75% of patients with insulinomas are diagnosed after a 24 hour fast and 90-94% at 48 hours. Although some experts advocate conducting the prolonged fast for only 48 hours (32), others disagree, arguing that prolonging the fast up to 72 hours minimizes misdiagnosis and maximizes the probability of diagnosing an insulinoma (33).

Limitations of the prolonged fast:

9. MIXED MEAL TEST

For patients with reported hypoglycemic symptoms several hours after meals, a mixed meal test may be performed. This test has not been well standardized. Patients are encouraged to eat a meal similar to the meal that usually provokes their symptoms. A serum or plasma glucose level is drawn at the time of symptoms. The patients’ history of hypoglycemia is verified if neuroglucopenic symptoms develop at the time of hypoglycemia.

10. GLUCAGON TOLERANCE TEST

The glucagon tolerance test serves as a supplemental study to aid in the diagnosis of hypoglycemic disorders when results from the prolonged fast are inconclusive. Following an overnight fast (or at the conclusion of the prolonged fast), 1 mg of glucagon is injected intravenously over 2 minutes. Plasma glucose and insulin levels are measured at baseline, 3, 5, 10, 15, 20, and 30 minutes after the glucagon injection. In normal patients, maximum insulin response occurs rapidly and usually does not exceed 100 uU/ml (peak insulin 61+19 uU/ml at 3-15 minutes), and the serum glucose levels peak at 20-30 minutes (140 +24 mg/dl) (34).

Insulinoma patients demonstrate an exaggerated insulin response to glucagon, with values often exceeding 160 uU/ml within 15-30 minutes of the injection (peak insulin 93-343 uU/ml at 15 minutes) (34). In the hypoglycemic patient at the conclusion of the prolonged fast, an increase in plasma glucose of >25 mg/dl (1.4 mmol/l) post-glucagon suggests an insulin-mediated etiology (31).

Patients with malnutrition or hepatic disease may be unable to have a hyperglycemic response to glucagon due to depleted hepatic glycogen stores. Insulin responses in these subjects may be increased but not to the degree seen in subjects with an insulinoma. Drugs such as diazoxide, hydrochlorothiazide and diphenylhydantoin can cause false negative results (34). Patients with non islet cell tumors such as hemangiopericytomas and meningeal sarcomas can have similar glucose elevations (30 mg/dl) as subjects with insulinomas following glucagon injection (35).

Another limitation of the glucagon stimulation test is the failure of some insulinoma patients to hypersecrete insulin following glucagon injection. This problem was reported in 8% of patients with insulinomas in one study (34). In addition, patients with cirrhosis with portocaval anastomosis can have peak insulin levels that are indistinguishable from subjects with insulinomas. Obese subjects and patients with acromegaly can also have exaggerated peak insulin responses, as can patients treated with tolbutamide and aminophylline.

An additional disadvantage of this test is the danger of causing hypoglycemia after 90-180 min (36), as well as inducing nausea and vomiting. Because of the possibility of severe hypoglycemia, a physician needs to be present during the test.

11. ANTI-INSULIN ANTIBODIES

Autoimmune hypoglycemia is a rare condition whereby antibodies, either directed against insulin or against the insulin receptor, are responsible for the hypoglycemia. Autoimmune hypoglycemia should be suspected when the hypoglycemia is associated with high insulin levels (usually >100 uU/ml) and incompletely suppressed C-peptide levels. Insulin levels are rarely >100 uU/ml in the presence of hypoglycemia due to an insulinoma. Although these elevated insulin levels can be observed with exogenous insulin administration, the associated c-peptide levels are usually extremely low. Autoimmune hypoglycemia is more likely to occur in women with other autoimmune conditions.

12. C-PEPTIDE SUPPRESSION TEST

Since c-peptide and insulin are secreted in equimolar concentrations in the pancreas, c-peptide levels are a good marker of endogenous insulin secretion. The c-peptide suppression test can be to test for an insulinoma or to provide supplemental diagnostic information, especially if the results of a supervised fast are not definitive. The c-peptide suppression test must be carefully administered, since the patient is given intravenous insulin to induce hypoglycemia. The advantage of the test is that it is of much shorter duration than the supervised fast.

The c-peptide suppression test is performed following an overnight fast. The procedure is to infuse regular insulin, 0.125 U/kg body weight, intravenously over 60 minutes. Blood samples are obtained from the contralateral arm at 0, 30, 60, 90, and 120 minutes for determination of insulin, c-peptide, and plasma glucose levels. An abnormal result is a lower percentage decrease of c-peptide at 60 minutes compared to normative data appropriately adjusted for the patient’s body mass index and age (31). For example, an abnormal result for a 45 year old with a BMI of 25-29 kg/m2 would be <61% suppression of c-peptide at 60 minutes (37).

Limitations of this test include the fact that some patients with a documented insulinoma have normal c-peptide levels including normal percent decrease in c-peptide levels. There is also the danger of inducing severe hypoglycemia. In addition, little data concerning the reliability, sensitivity and safety of this test are published.