INTRODUCTION — The diagnosis of adrenal insufficiency can be made by the finding of low morning serum cortisol concentration and a subnormal response to synthetic corticotropin (ACTH [1-24]). In very rare circumstances, however, other tests, such as insulin-induced hypoglycemia, can be utilized to evaluate the response to stress, particularly when ACTH deficiency is suspected as the cause. However, when compared with other tests for the diagnosis of adrenal insufficiency, it is more difficult to perform, has significant risk, and is much more expensive. As a result, it is infrequently performed in clinical practice for the assessment of pituitary-adrenal function. This test also has been used to establish a diagnosis of growth hormone deficiency (although we do not recommend its use in that setting).
Details of the insulin-induced hypoglycemia test (also referred to as the "insulin tolerance test" or "ITT") in adults will be reviewed here. Additional information on the diagnoses of adrenal insufficiency, the physiologic response to hypoglycemia, and diagnostic testing for adult growth hormone deficiency are reviewed separately. (See "Diagnosis of adrenal insufficiency in adults" and "Physiologic response to hypoglycemia in healthy individuals and patients with diabetes mellitus" and "Growth hormone deficiency in adults".)
CLINICAL USES — The rationale for this test is that hypoglycemia induced by insulin administration is a sufficient stress to stimulate corticotropin-releasing hormone (CRH), corticotropin (ACTH), and therefore cortisol secretion. The advantage of this test is that the results also correlate relatively well with the serum cortisol response to surgical stress. Although this is a valid test of hypothalamic-pituitary-adrenal response to stress, in most settings, the ACTH stimulation test provides nearly the same information, is less difficult and less expensive to perform, and is without risk to the patient. (See "Initial testing for adrenal insufficiency: Basal cortisol and the ACTH stimulation test" and "Diagnosis of adrenal insufficiency in adults".)
Recent pituitary surgery — Recent pituitary surgery is the main clinical scenario when the ACTH stimulation test may not be as useful. If ACTH deficiency occurs as a result of transsphenoidal surgery, the cortisol response to exogenous ACTH stimulation may by normal (in the first month or two after surgery) until the volume of the adrenal cortex has decreased significantly (see "Treatment of gonadotroph and other clinically nonfunctioning adenomas"). In this setting, insulin-induced hypoglycemia may be a better test to diagnose ACTH deficiency because the ACTH response to hypoglycemia may be lost [1].
This is suggested by the discrepancy in responses to both the 250 and 1 mcg ACTH stimulation test compared with insulin hypoglycemia. Available data suggest that ACTH stimulation tests are not ideal to test for adrenal insufficiency in the immediate postoperative period and that at one month after surgery, the 250 mcg test may underdiagnose, while the 1 mcg test may overdiagnose adrenal insufficiency. Although the overnight metyrapone test is rarely performed in the United States, its findings six weeks postoperatively correlate well with the cortisol response to insulin-induced hypoglycemia [2]. (See "Metyrapone stimulation tests".)
The limitations of ACTH stimulation testing and its discrepancies with the ITT are illustrated by the following results:
●Of 19 patients after transsphenoidal surgery, 6 had an abnormal response to insulin within 11 days of surgery, but 2 of these had normal responses to the 1 mcg and 5 had normal responses to the 250 mcg ACTH stimulation test [1].
●In another study at three to five weeks after surgery, six patients with an abnormal cortisol response to the 1 mcg ACTH stimulation test had normal responses to insulin, while three with normal responses to the 250 mcg test had an abnormal response to insulin [3].
●A third study showed that 23 of 62 patients with initial normal responses to 250 mcg ACTH had an abnormal response when tested at one or three months [4].
However, an abnormal cortisol response to insulin-induced hypoglycemia test in the immediate postoperative period may not predict long-term dysfunction. In one study of 36 patients with an abnormal response to insulin at three months after surgery, 20 patients had a normal response at 12 months [5]. In another, all six patients with an abnormal response to insulin in the immediate postoperative period had a normal response at three months [1]. By contrast, very few patients with an initially normal response to insulin or ACTH progress to an abnormal function [4-6].
The overnight metyrapone test, when performed six weeks after transsphenoidal surgery, correlates well with the cortisol response to insulin-induced hypoglycemia [2]. However, both tests are now performed infrequently in the United States, and we do not suggest their routine use. (See "Metyrapone stimulation tests".)
The basal cortisol level after surgery may be a useful indicator of adrenal function. A morning value >9 to 11 mcg/dL (250 to 300 nmol/L) within eight days of surgery predicts a normal response to insulin [3,7-9]. Conversely, basal values of 3 to 4 mcg/dL (80 to 111 nmol/L) predict an abnormal response. Intermediate values may represent normal or abnormal function. Clinicians should be aware of the assay method utilized and the relevant cutoffs in their hospital's clinical laboratory, as newer, more specific cortisol platform immunoassays and mass spectrometry methods have lower serum cortisol cutoffs [10,11].
The assessment of pituitary-adrenal function in patients undergoing transsphenoidal surgery is reviewed in detail separately. Patients with abnormal function are unlikely to regain function and should receive perioperative glucocorticoid coverage and postoperative replacement therapy. Other details of postoperative testing and glucocorticoid coverage are reviewed separately. (See "Treatment of gonadotroph and other clinically nonfunctioning adenomas".)
Growth hormone deficiency: Not recommended — We do not suggest the insulin-induced hypoglycemia test for the evaluation of growth hormone deficiency. It was more commonly used in the past, but its disadvantages are that it requires constant supervision by a clinician and the symptoms of hypoglycemia can be very unpleasant for the patient. In fact, none of the tests for secondary adrenal insufficiency (ACTH or CRH stimulation; insulin-induced hypoglycemia) seem to perform well in patients with growth hormone deficiency. This issue is discussed in more detail separately [12]. (See "Growth hormone deficiency in adults", section on 'Insulin-induced hypoglycemia'.)
PRACTICAL ASPECTS
Contraindications — This test is contraindicated in patients with coronary heart disease, cerebrovascular disease or epilepsy; if coronary disease is suspected, appropriate evaluation must be performed before considering the insulin tolerance test. (See "Screening for coronary heart disease".)
Procedure — The patient fasts for at least eight hours before the test and must remain supine during the procedure. An experienced clinician must be present at all times to monitor for complications related to hypoglycemia (chest pain, tachycardia, and/or neuroglycopenic symptoms [confusion, seizures]). A syringe containing 50 percent glucose solution should be at the bedside. An intravenous line is established, and insulin is injected intravenously. The usual dose is 0.15 units/kg of regular insulin, but different doses may be indicated in certain patients [13]:
●In patients thought to have hypopituitarism or primary adrenal insufficiency, the insulin dose is decreased to 0.1 units/kg because these conditions may be associated with decreased release of other counterregulatory hormones such as epinephrine and growth hormone.
●In patients with obesity, diabetes mellitus, suspected acromegaly or Cushing's syndrome, the dose is increased to 0.2 units/kg. However, coronary disease must be ruled out before performing the test, particularly at this higher dose.
●In premenopausal women, the test can be performed at any phase of the menstrual cycle, as there are no cycle effects on the hypothalamic-pituitary-adrenal axis response to insulin-induced hypoglycemia [14].
Blood is obtained for bedside measurement of serum glucose and for laboratory measurement of serum glucose and cortisol assays (and for plasma corticotropin [ACTH], if indicated) immediately before insulin is injected and every 15 minutes thereafter. The last sample is obtained when adequate hypoglycemia, defined as 35 mg/dL (1.9 mmol/L) or less, is achieved. At that time, patients should have some symptoms of hypoglycemia, either of sympathetic discharge or of central nervous system glucose deprivation, such as simply falling asleep. Some use 40 mg/dL (2.2 mmol/L) as the cutoff for serum or blood glucose [2,15].
●Almost all patients have some degree of perspiration. If the patient does not perspire, the adequacy of the stress stimulus must remain suspect regardless of the serum glucose concentration.
●Most patients also have a hyperactive precordium (but not tachycardia or hypotension, because they are supine), and feelings of hunger, drowsiness, detachment, or anxiety. The last is common and sometimes severe, and many patients find this an unpleasant experience.
Hypoglycemia usually occurs 30 to 45 minutes after insulin injection. If adequate hypoglycemia is not achieved, a second similar dose of regular insulin should be injected intravenously. Adequate hypoglycemia should be achieved within the subsequent 30 to 45 minutes.
Ideally, an automated glucose oxidase analyzer or calibrated and validated point of care device should be available at the bedside. Unfortunately, most glucose oxidase strips, whether read visually or with a glucose meter, are inaccurate at low serum glucose concentrations, tending to underestimate them, leading to premature termination of the test. The final, definitive blood sample for measurement of ACTH and cortisol should be obtained 5 to 10 minutes after the patient begins to perspire or, if it can reliably be measured, when the serum glucose falls below 35 mg/dL (1.9 mmol/L). As noted, there is no clear consensus about the glucose nadir necessary for optimal test performance. While many use 35 mg/dL (1.9 mmol/L), others use 40 or 45 mg/dL (2.2 or 2.4 mmol/L) [2,15].
Patients with primary or secondary adrenal insufficiency or longstanding diabetes mellitus have an impaired compensatory response to hypoglycemia. (See "Physiologic response to hypoglycemia in healthy individuals and patients with diabetes mellitus".)
●Therefore, the test should be stopped when the serum glucose concentration decreases to or below 35 mg/dL (1.9 mmol/L). In addition to stopping the test, sweetened orange juice or cola should be given by mouth.
●If hypoglycemic symptoms are severe, administration of 25 g of 50 percent glucose should be administered intravenously (see "Hypoglycemia in adults with diabetes mellitus", section on 'With IV access'). The solution should be infused over a period of one minute (slowly, because of its hypertonicity) if seizure, chest pain, confusion, disorientation, or other potentially serious complications occur; this will result in hyperglycemia within 30 seconds.
●Blood should be drawn for measurement of cortisol when glucose is given, and 15 minutes later, as levels continue to increase despite increasing serum glucose concentrations [16].
Normal values — The criteria for a normal serum cortisol response ranged from 18 to approximately 22 mcg/dL (500 to 600 nmol/L) in multiple studies using different assays and various methods to establish a normal reference range. The newer, more specific cortisol immunoassays and tandem mass spectrometry (LC-MS/MS) methods give significantly lower cortisol results (as much as 25 percent lower). Because of this, serum cortisol cutoffs for any stimulation test must be individualized to the method used in the clinical laboratory [10,11,17,18]. The highest serum cortisol is used to interpret the response, rather than the cortisol increment [16].
The serum cortisol increase parallels that of plasma ACTH up to a serum cortisol concentration of approximately 22 mcg/dL (607 nmol/L), corresponding to a plasma ACTH concentration of approximately 75 pg/mL (16 pmol/L), which is the acute maximally stimulating plasma concentration of ACTH.
Plasma ACTH can also be measured. The normal plasma ACTH response has not been carefully defined, but the peak concentration should exceed 150 pg/mL (33 pmol/L) [19-23].
There is increasing evidence that the salivary cortisol response to insulin hypoglycemia can provide useful information, although definitive cutoffs have yet to be established [24-26].
Interpretation — An inadequate cortisol response can be due to hypopituitarism of any etiology, including hypothalamic corticotropin-releasing hormone (CRH) deficiency, isolated ACTH deficiency, partial or panhypopituitarism, and acute or chronic administration of synthetic glucocorticoids; it can also be due to primary adrenal insufficiency when serum cortisol rather than plasma ACTH values are used as the endpoints of the test. However, false-negative results in patients who have partial adrenal insufficiency can occur because hypoglycemia is such a potent stimulus that it may obscure a partial defect.
●Hypoglycemia is a stronger stimulus of ACTH secretion than is hypocortisolemia (as induced by metyrapone). Consequently, patients may have a normal response to hypoglycemia, but an inadequate response to metyrapone, which can detect subtle defects in ACTH secretion. The reverse is almost never true.
●The low-dose (1 mcg) one-hour ACTH stimulation test does not result in maximally stimulating plasma ACTH concentrations. Thus, as with metyrapone, patients may have a normal serum cortisol response to hypoglycemia (or to the high-dose cosyntropin test), but an inadequate response to low-dose cosyntropin.
Comparison with other tests — As a stress test, insulin-induced hypoglycemia has certain advantages:
●This test evaluates the integrity of the full hypothalamic-pituitary-adrenal axis as hypoglycemia acts centrally to stimulate hypothalamic CRH release and, therefore, ACTH release
●The degree of hypoglycemia can be easily quantified
●The test is relatively safe (if an experienced clinician is present) in patients with no history of seizures or cardiovascular or cerebrovascular disease
●The hypoglycemia can be corrected within seconds by intravenous infusion of glucose
However, compared with other tests for the diagnosis of adrenal insufficiency, it is more difficult to perform, has significant risk, and is much more expensive. As a result, it is rarely performed in clinical practice for the assessment of pituitary-adrenal function [27,28].
The magnitude of the response depends upon the degree of hypoglycemia. Criteria (ie, rate and magnitude of the decrease in serum glucose) that are adequate to elicit a maximal serum growth hormone response are inadequate to ensure a maximal plasma ACTH response. Thus, since the purpose of the test is to increase ACTH release, more stringent criteria of adequate hypoglycemia must be applied.
SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Adrenal insufficiency" and "Society guideline links: Growth hormone deficiency and other growth disorders".)
SUMMARY — Although insulin-induced hypoglycemia is a valid, and perhaps the most rational, test of hypothalamic-pituitary-adrenal response to stress, there is little, if any, reason for performing the test in clinical practice except in patients with suspected recent corticotropin (ACTH) deficiency (eg, recent pituitary surgery). (See 'Clinical uses' above.)
●The test is relatively safe, if a clinician is present, in patients with no history of seizures or cardiovascular or cerebrovascular disease, but otherwise should not be performed. (See 'Practical aspects' above.)
•Insulin (usually at a dose of 0.15 units/kg [in patients with low basal cortisol levels, the dose should be reduced to 0.1 units/kg]) is given with the aim to achieve hypoglycemia of 35 to 45 mg/dL (1.9 to 2.5 mmol/L) or less. (See 'Procedure' above.)
•Cortisol concentrations are measured at 0, 30, and 45 minutes after insulin is given and, when appropriate, 15 minutes after glucose has been given to reduce symptoms of hypoglycemia. It may also be useful to measure the plasma ACTH response. (See 'Normal values' above.)
●The criteria for a normal serum cortisol response range from 18 to approximately 22 mcg/dL (500 to 600 nmol/L) in studies using different assays and various methods to establish a normal reference range. Newer, more specific cortisol assays have cutoffs approximately 25 percent lower. Failure to reach this level is indicative of an inadequate response only if the serum glucose fell to 35 to 45 mg/dL (1.9 to 2.5 mmol/L) or less. If this degree of hypoglycemia was not achieved and the cortisol is inadequate, the stimulus was inadequate and the test must be repeated. It is the peak serum cortisol concentration that is achieved rather than the increment that is important. (See 'Interpretation' above.)
●An inadequate cortisol response can be due to hypopituitarism of any etiology, including hypothalamic corticotropin-releasing hormone (CRH) deficiency, isolated ACTH deficiency, partial or panhypopituitarism, and acute or chronic administration of synthetic glucocorticoids; it can also be due to primary adrenal insufficiency. However, false-negative results in patients who have partial adrenal insufficiency can occur because hypoglycemia is such a potent stimulus that it may obscure a partial defect. (See 'Interpretation' above.)
DISCLOSURE — The views expressed in this topic are those of the author(s) and do not reflect the official views or policy of the United States Government or its components.