Confirmation of Hypoglycemia

The presence of hypoglycemia is often suspected on the basis of a history compatible with neurogenic or neuroglycopenic symptoms. Although sometimes compelling, the history is more often only suggestive, since such symptoms are common in a variety of other disorders. Thus the diagnosis of hypoglycemia is suspected much more often than it is confirmed. Symptoms of hypoglycemia usually occur more than several hours after meals, often after exercise. However, as noted above, patients with disorders that cause fasting hypoglycemia, may also develop hypoglycemia within several hours of a meal. Occasionally, a hypoglycemic disorder is first suspected because of a low glucose level in a plasma sample obtained for other reasons.

Once suspected, the diagnosis of hypoglycemia is most convincingly established when it is based on Whipple’s triad: symptoms consistent with hypoglycemia, a concomitant low plasma glucose concentration, and relief of symptoms after the glucose level is raised to normal. The diagnosis of a hypoglycemic disorder is seldom tenable in the absence of consistent symptoms.

Patients with fasting hypoglycemia may have plasma glucose levels below 45 mg/dl (2.5 mmol/L) after a 10- to 12-hour overnight fast, especially with repeated measurements. Thus the first diagnostic step in suspected fasting hypoglycemia is to measure the plasma glucose concentration after an overnight fast, on several days if necessary. A value less than 45 mg/dl documents the presence of fasting hypoglycemia. If plasma glucose level is greater than 45 mg/dl after an overnight fast, the fast should be prolonged until symptomatic hypoglycemia with a plasma glucose level less than 45 mg/dl occurs, or for a maximum of 72 hours. Plasma glucose should be measured at least every 4 hours (more frequently as plasma glucose level falls) and when symptoms occur. Bedside glucose monitoring devices may be used for rapid estimation but are not accurate enough for definitive diagnosis; laboratory measurements should be performed. Symptomatic hypoglycemia usually occurs within the first 24 hours of a diagnostic fast in affected patients. A period of exercise at the end of a prolonged fast may precipitate hypoglycemia in an affected patient, in contrast to the normal stability of plasma glucose during exercise.

When symptomatic hypoglycemia with a plasma glucose level less than 45 mg/dl occurs, certain tests should be performed to distinguish among possible causes. Plasma insulin and C-peptide should be measured, preferably on several samples, before the fast is ended, and a plasma or urine assay for sulfonylureas should be performed.

Failure of the plasma glucose concentration to fall below 50 mg/dl during a prolonged fast, particularly if a period of exercise is included, excludes the diagnosis of fasting hypoglycemia. In men, fasting plasma glucose values below 50 mg/dl document the presence of fasting hypoglycemia. The same statement cannot be made for women or children, in whom plasma glucose concentrations commonly fall below 50 mg/dl in the absence of symptoms during a prolonged fast. If values below 50 mg/dl are associated with unequivocal symptoms in a woman or a child, fasting hypoglycemia has been documented. If not, convincing biochemical evidence is required for diagnosis of a fasting hypoglycemic disorder (see later discussion).

The differential diagnosis of fasting hypoglycemia can be narrowed rapidly with standard clinical data. A history of the use of insulin or other offending drugs may be obtained. Surreptitious use of hypoglycemic drugs, however, may be difficult to detect and requires screening of urine or plasma for such agents. Sulfonylureas may be inadvertently substituted for another drug, so the identity of each of the patient’s medications should be confirmed by inspection. Critical organ dysfunction severe enough to cause hypoglycemia becomes apparent on clinical examination and with routine laboratory tests. Extrapancreatic tumors associated with hypoglycemia generally are large and clinically evident. The hypoglycemias of childhood are self-limited except in patients with congenital enzymatic defects in whom associated findings suggest the diagnosis. In the absence of these causes the differential diagnosis is limited to excessive insulin secretion or deficient glucose counterregulatory hormone secretion.

Hypoglycemia stimulates the secretion of cortisol, growth hormone, glucagon, and epinephrine. Thus the finding of elevated plasma concentrations of these hormones during spontaneous (or insulin-induced) hypoglycemia excludes deficiencies. It is conventional to assess the adequacy of growth hormone and cortisol secretion in patients with fasting hypoglycemia. Because deficiencies of glucagon or epinephrine rarely if ever cause hypoglycemia except in patients with insulin-dependent diabetes, these hormones usually are not measured.

Diagnosis of hyperinsulinism requires the demonstration of an inappropriately elevated plasma insulin level during hypoglycemia. Plasma insulin concentrations are often not elevated above the fasting reference range. Because normal insulin secretion nearly ceases when the plasma glucose concentration falls to less than 45 mg/dl, plasma insulin concentrations should be measured when the fasting plasma glucose level falls below this level. In such samples a plasma insulin concentration greater than 6 uU/ml (36 pmol/L) is diagnostic of hyperinsulinism.

Measurement of plasma C-peptide (the connecting peptide cleaved from proinsulin during conversion to insulin) can distinguish endogenous hyperinsulinism from surreptitious insulin administration. When plasma glucose level is less than 45 mg/dl, a plasma C-peptide level greater than 0.6 ng/ml (0.2 nmol/L) confirms that hyperinsulinism is of endogenous origin. Because C-peptide levels are also elevated in factitious hypoglycemia attributable to sulfonylureas, plasma or urine should be tested for these drugs. C-peptide levels are suppressed below these values in patients with exogenous hyperinsulinism (except in the presence of insulin antibodies; see later discussion). In an adult, fasting hypoglycemia that is due to hyperinsulinism implies the presence of an insulinoma if factitious hypoglycemia has been excluded.

Most patients with pancreatic beta-cell neoplasms have increased plasma levels of proinsulin during hypoglycemia. Specific proinsulin immunoassays have a high degree of sensitivity for insulinomas and may be useful when insulin and C-peptide levels during a diagnostic fast are equivocal. Other conditions, however, produce elevated proinsulin levels, so the diagnosis cannot be made on this basis alone.

The presence of circulating antibodies to insulin in patients with no history of insulin use suggests surreptitious insulin injection or autoimmune hypoglycemia. Such antibodies produce artifactually high values for plasma insulin in double-antibody immunoassays, and by binding endogenous proinsulin, which contains the C-peptide sequence, they also elevate plasma total C-peptide immunoreactivity. Methods are available for measuring free insulin and C-peptide levels in the presence of insulin antibodies. Injection of human insulin produces a much smaller immune response than animal insulins, and surreptitious use of human insulin may produce hypoglycemia in the absence of detectable antibodies to insulin. This diagnosis is still evident, however, from the combination of elevated plasma insulin and suppressed plasma C-peptide levels.

Evaluation of a patient with fasting hypoglycemia should include a diagnostic fast, with measurement of plasma insulin and C-peptide when the plasma glucose is less than 45 mg/dl (and preferably several such measurements), along with assay for sulfonylureas and antibodies to insulin. The differential diagnosis of hyperinsulinism is summarized in

Table 304-1.

The preceding discussion assumes accurate measurement of the plasma glucose concentration. Artifactual lowering of the measured glucose level (pseudohypoglycemia) can occur if separation of the plasma from the formed elements of the blood is delayed for several hours, especially if glucose utilization by the formed elements is excessive (e.g., marked leukocytosis). This can be avoided by prompt centrifugation of the blood sample or by the use of special sampling tubes containing an inhibitor of glycolysis.

As noted earlier, patients with disorders that cause fasting hypoglycemia (such as insulinoma) may also have postprandial hypoglycemia. If hypoglycemia is suspected in a patient with symptoms that regularly occur within a few hours of a meal, plasma glucose should be measured during fasting as described earlier. If fasting does not provoke hypoglycemia, a hypoglycemic disorder is very unlikely. If strong clinical suspicion persists, plasma glucose should be measured frequently after ordinary mixed meals, while the patient records all symptoms and their time of occurrence. Unless typical symptoms occur and both coincide with low plasma glucose values and abate as plasma glucose levels rise (Whipple’s triad), the diagnosis of reactive hypoglycemia can be confidently excluded. Blood glucose self-monitoring devices are not sufficiently reliable for such testing and should not replace plasma glucose measurements by a clinical laboratory. The oral glucose tolerance test should not be used for diagnosis of suspected hypoglycemia, since plasma glucose concentrations reach nadirs less than 50 mg/dl after glucose ingestion in 10% of normal asymptomatic persons.