The OGTT is more reliable when the patient is ambulatory and does not have other severe acute or chronic illnesses. The test should be preceded by at least 3 days of adequate carbohydrate diet and should be performed in the morning after the patient has fasted at least 10 hours (but no longer than 16 hours). The test dose has been standardized by the NDDG at 75 gm of glucose or dextrose for nonpregnant persons and 100 gm for pregnant women. The dose may be calculated from body weight. Various ready-made commercial preparations can be used, or the dose can be given in up to 300 ml of water, usually flavored with a substance such as lemon juice. The dose should be consumed by the patient within 5 minutes. The test officially begins when the patient begins to drink. The NDDG recommends that the patient should remain seated during the test and should not smoke. One should also beware of medication that could affect test results, such as oral contraceptives, steroids, diuretics, and anticonvulsants.
NDDG test protocol. Blood specimens are taken fasting, then every 30 minutes for 2 hours after the beginning of dextrose ingestion. After ingestion of the test dose, a lag period occurs, after which the blood glucose curve rises sharply to a peak, usually in 15-60 minutes. In one study, 76% had maximal values at 30 minutes and 17% at 1 hour. The curve then falls steadily but more slowly, reaching normal levels at 2 hours. These may be fasting (FBG) values or simply within the blood glucose reference range.
Occasionally, after the fasting level is reached, there may follow a transient dip below the fasting level, usually not great, then a return to fasting values. This relative hypoglycemic phase of the curve (when present) is thought to be due to a lag in the ability of the liver to change from converting glucose to glycogen (in response to previous hyperglycemia) to its other activity of supplying glucose from glycogen. In some cases, residual insulin may also be a factor. This “hypoglycemic phase,” if present, is generally between the second and fourth hours. Several reports indicate that so-called terminal hypoglycemia, which is a somewhat exaggerated form of this phenomenon, occurs in a fairly large number of patients with a GTT response indicative of mild diabetes. They believe that an abnormally marked hypoglycemic dip often appears in mild diabetics 3-5 hours after meals or a test dose of carbohydrates and may be one of the earliest clinical manifestations of the disease.
Flat oral glucose tolerance test. This is an OGTT the peak of which has been defined variously as less than 40 mg, 25 mg, or 20 mg/100 ml above the FBG value. The most commonly used definition is 25 mg/100 ml (1.38 mmol/L). The condition most frequently associated with a flat OGTT is small intestinal carbohydrate malabsorption due to sprue or celiac disease, with a lesser number of cases seen in myxedema and some cases reported in pituitary insufficiency. Flat OGTT results may also occur in clinically normal persons. When the 40 mg definition was used, one study reported a flat OGTT in 90% of patients with sprue, but also in up to 40% of clinically normal persons. When the 25 mg definition was used, another study reported a flat OGTT in about 60% of patients with sprue. Using the 20 or 25 mg definition, several investigators found a flat OGTT result in about 20% of clinically normal persons (range, 7%-25%).
Oral glucose tolerance test interpretation. In the past, criteria for interpretation of the OGTT have varied widely. This situation is brought about because of the absence of a sharp division between diabetics and nondiabetics, variations in methodology, and variations in adjustment for the many conditions that may affect the GTT quite apart from idiopathic diabetes mellitus; some of these factors have been mentioned previously, and others will be discussed later. The NDDG criteria are rapidly replacing previous criteria as world-recognized standards. The NDDG criteria are listed in Table 28-2. Please note that all values from now on in the text discussion will be given in milligrams per 100 ml, using true glucose methods unless otherwise stated.
Table 28-2 National diabetes data group criteria for diagnosis of diabetes mellitus in nonpregnant and pregnant adults*
The NDDG has permitted a diagnosis of diabetes mellitus to be made in any one of three different ways:
1. Sufficient classical symptoms of diabetes mellitus (e.g., polydipsia, polyuria, ketonuria, and weight loss) plus either an unequivocal elevation of the fasting glucose (FBG) level or an elevation of the non-FBG level greater than 200 mg/100 ml (11.1 mmol/L).
2. Elevation of the FBG level (venous serum or plasma) greater than 140 mg/100 ml (7.8 mmol/L) on more than one occasion (assuming no condition is present that falsely increases blood glucose values).
3. A normal FBG level but OGTT peak and 2-hour values both greater than 200 mg/100 ml (11.1 mmol/L) on more than one occasion.
Three points should be noted. First, the diagnosis of diabetes can be made in nonpregnant adults if typical clinical symptoms are present plus a nonfasting serum specimen more than 200 mg/100 ml. Second, the diagnosis can be made without requiring a GTT if the FBG level is sufficiently elevated. Third, when the diagnosis is based predominantly on blood glucose measurement, either the FBG level or the OGTT, diagnosis requires sufficient abnormality on 2 different days rather than only one occasion.
Impaired glucose tolerance. The NDDG recognizes a category of OGTT curve that it calls “impaired” glucose tolerance, which is significantly abnormal values but not sufficiently abnormal to make a diagnosis of diabetes (Table 28-3). This involves an FBG level less than 140 mg/100 ml (7.77 mmol/L) and a single point on the OGTT curve at or above 200 mg/100 ml (11.1 mmol/L); that is, either the peak or the 2-hour value greater than 200 mg/100 ml, but not both.
Table 28-3 National diabetes data group classification of glucose tolerance abnormalities
There are abnormal areas in the OGTT that include FBG between 115-140 mg/100 ml (6.4-7.8 mmol/L) and other points above reference range but less than 200 mg/100 ml (11.1 mmol/L); in other words, intermediate between normal and impaired OGTT. The NDDG calls these “nondiagnostic abnormalities.” World Health Organization (WHO) 1980 criteria for diagnosis of diabetes mellitus and for impaired OGTT are the same as those of the NDDG. However, WHO considers the NDDG category of nondiagnostic abnormalities as being normal, with the exception of a 2-hour value between normal and 200 mg/100 ml (11.1 mmol/L), which WHO includes in the impaired OGTT category.
Oral glucose tolerance test criteria for children. In children, criteria for diagnosis of diabetes mellitus are rather similar to those of nonpregnant adults, but there are a few significant differences. First, if the child has classic symptoms of diabetes, a single random nonfasting serum glucose value at or above 200 mg/100 ml (11.1 mmol/L) is sufficient for diagnosis. Second, the upper limit of the FBG reference range is set at 130 mg/100 ml (7.2 mmol/L) instead of the nonpregnant adult upper limit of 115 mg/100 ml (6.4 mmol/L). However, FBG values necessary for diagnosis of diabetes are the same for children and adults (140 mg/100 ml; 7.8 mmol/L). Second, the glucose dose for children is calculated on the basis of patient weight (1.75 gm/kg of ideal weight to a maximum of 75 gm). Third, elevation of the FBG level alone is not sufficient in children. The FBG level must be more than 140 mg/100 ml (7.8 mmol/L) and either the peak or the 1-hour value must be more than 200 mg/100 ml (11.1 mmol/L) (if the FBG level is normal, both the peak and the 2-hour value must be more than 200 mg/100 ml, which is the same requirement listed for adults).
The NDDG definition of gestational diabetes is abnormal glucose tolerance with onset or recognition during pregnancy but not before. Six weeks or more after the end of pregnancy, the patient should be retested using the standard nonpregnant OGTT with standard nonpregnant NDDG criteria and reclassified into previous abnormal glucose tolerance (if the postpartum standard nonpregnant OGTT result is normal), impaired glucose tolerance (if the standard nonpregnant OGTT result is abnormal but not sufficiently abnormal to fit the NDDG criteria for diabetes), or diabetes mellitus. The American Diabetes Association (1986) recommends that pregnant women who are not known to have abnormal glucose tolerance should have a special screening test between the 24th and 28th week consisting of a test dose of 50 gm of oral glucose (fasting or nonfasting). A single postdose 1-hour venous plasma value of 140 mg/100 ml (7.8 mmol/L) or more suggests the need for the full NDDG gestational OGTT during the pregnancy. The gestational OGTT consists of FBG, 1-hour, 2-hour, and 3-hour specimens following a 100-gm oral glucose dose.
There is general agreement about acceptability of the gestational 50-gm 1-hour screening test. However, one investigator has reported 11% more abnormal results when the 50-gram test was performed at 28 weeks’ gestation than when it was performed on the same patients at 20 weeks, and an additional 8% results were abnormal at 34 weeks than at 28 weeks. There is significantly more controversy regarding the NDDG gestational 100-gm 3-hour diagnostic test. In one study, 17% of patients who had initially normal NDDG 3-hour test results had one abnormal value when a repeat test was done 1-2 weeks later, and 5% had initially abnormal results that were normal when the test was repeated. Overall, it appears that about 25% of initial gestational NDDG 3-hour test results will significantly change when the test is repeated. However, the greatest controversy involves the glucose levels selected as cutoff points between normal and abnormal. This controversy arises because several investigators have found nearly as many newborns with macrosomia (about 30%) in mothers having one significantly elevated time point on the NDDG 3-hour gestational OGTT as in those mothers who had two significantly elevated time points (required for diagnosis of diabetes). This suggested that the gestational NDDG criteria for diabetes were too conservative. Several investigators have proposed revisions of the NDDG gestational 3-hour OGGT criteria; the two most frequently cited in the literature are shown in Table 28-4. One recent report found that about half of study patients with one significantly elevated time point on the NDDG 3-hour OGTT had two significantly elevated time points (therefore, diagnostic of diabetes) using the Carpenter-modified OGGT criteria; about half these mothers had macrosomic infants and half did not.
Table 28-4 Revised cutoff points proposed for abnormality in the 100-gm 3-hour diagnostic test for gestational diabetes
Screening tests for diabetes
Screening tests for diabetes attempt to circumvent the multiple blood glucose determinations required for the GTT. The FBG level and the 2-hour postprandial (2 hours after a meal) blood glucose level have been widely used. Since these essentially are isolated segments of the GTT curve, interpretation of their results must take several problems into consideration in addition to those inherent in the GTT.
An abnormality in the FBG level for example, raises the question of whether a full GTT is needed for confirmation of diabetes. Most authorities, including the NDDG panel, believe that if the FBG level is sufficiently elevated, there is no need to do the full GTT. Whatever the etiology of the abnormal FBG level, the GTT result will also be abnormal. Since it is known in advance that the GTT result will be abnormal, no further information is gained from performing the GTT. Most also agree that a normal FBG value is not reliable in ruling out possible diabetes. In one study, 63% of those with diabetic GTT results had normal FBG values. Others have had similar experiences, although perhaps with less striking figures.
Most investigators believe that of all the postprandial GTT values, the 2-hour level is the most crucial. The 2-hour value alone has therefore been proposed as a screening test. This recommendation is based on the fact that with a normal FBG level, the diagnosis of diabetes mellitus cannot be made with confidence on the basis of an abnormal peak blood glucose level if it is accompanied by a normal 2-hour blood glucose level in the OGTT. The main reason for this lies in the effect of gastric emptying on glucose absorption. It has been fairly well proved that normal gastric emptying does not deliver a saturation dose to the duodenum. Therefore, slow gastric emptying tends to produce a low, or “flat,” GTT curve. On the other hand, either unusually swift gastric transit or delivery of a normal total quantity of glucose to the small intestine within a markedly shortened time span results in abnormally large amounts of glucose absorbed during the initial phases of the tolerance test. Since homeostatic mechanisms are not instantaneous, the peak values of the tolerance curve reach abnormally high figures before the hyperglycemia is brought under control. An extreme example of this situation occurs in the “dumping syndrome” produced by gastrojejunostomy.
If previous time interval specimens are considered unreliable, the question then is justified as to whether the 2-hour value alone is sufficient for diagnosis of diabetes. According to the NDDG criteria, both the peak level and the 2-hour level must be greater than 200 mg/100 ml (11.1 mmol/L), so that a full GTT is necessary, even though it is uncommon to find a 2-hour value more than 200 mg/100 ml and a peak value less than 200 mg/100 ml. The NDDG recommendation may be based on reports that one variant of the normal OGTT drops relatively swiftly to normal at approximately 1.5 hours and then rebounds above 140 mg/100 ml (7.8 mmol/L) by 2 hours. In one report this phenomenon occurred in as many as 5%-10% of patients.
Glucose tolerance in other diseases
Besides the intrinsic and extrinsic factors that modify response to the OGTT, other diseases besides diabetes mellitus regularly produce diabetic-type GTT patterns or curves. Among these are adrenal, thyroid, and pituitary hormone abnormalities that influence liver or tissue response to blood glucose levels. Cushing’s syndrome results from hypersecretion of hydrocortisone. Since this hormone stimulates gluconeogenesis among its other actions, 70%-80% of patients with Cushing’s syndrome exhibit decreased carbohydrate tolerance, including 25% who exhibit overt diabetes. One report suggests that patients with gestational diabetes who are in the third trimester have fasting plasma cortisol levels higher than those of nonpregnant women. Pheochromocytomas of the adrenal medulla (or elsewhere) have been reported to produce hyperglycemia in nearly 60% of affected patients and glucosuria in a lesser number. These tumors produce norepinephrine or epinephrine, either continually or intermittently. The diabetogenic effects of epinephrine were mentioned earlier, and it has been noted that pheochromocytomas that secrete norepinephrine rather than epinephrine are not associated with abnormalities of carbohydrate metabolism. Primary aldosteronism leads to the overproduction of aldosterone, the chief electrolyte-regulating adrenocortical hormone. This increases renal tubular excretion of potassium and retention of sodium. Patients with primary aldosteronism frequently develop decreased carbohydrate tolerance. According to Conn, this is most likely due to potassium depletion, which in some manner adversely affects the ability of pancreatic beta cells to respond normally to a hyperglycemic stimulus. Parenthetically, there may be some analogy in reports that chlorothiazide diuretics may cause added decrease in carbohydrate tolerance in diabetics, thus acting as diabetogenic agents. Some say that no such effect exists without some degree of preexisting glucose tolerance abnormality, but others maintain that it may occur in a few clinically normal persons. Chlorothiazide often leads to potassium depletion as a side effect; indeed, one report indicates that potassium supplements will reverse the diabetogenic effect. However, other mechanisms have been postulated.
Thyroid hormone has several effects on carbohydrate metabolism. First, thyroxine acts in some way on small intestine mucosal cells to increase hexose sugar absorption. In the liver, thyroxine causes increased gluconeogenesis from protein and increased breakdown of glycogen to glucose. The metabolic rate of peripheral tissues is increased, resulting in an increased rate of glucose utilization. Peripheral tissue glycogen is depleted. Nevertheless, the effect of hyperthyroidism on the GTT is variable. Apparently the characteristic hyperthyroid curve is one that peaks at an unusually high value, sometimes with glucosuria, but that returns to normal ranges by 2 hours. However, in one extensive survey, as many as 7% of hyperthyroid patients were reported to have diabetic curves, and another 2% had actual diabetes mellitus. Surprisingly, the type of curve found in any individual patient was not related to the severity of the hyperthyroidism. In myxedema, a flat OGTT curve (defined as a peak rise of <25 mg/100 ml [1.4 mmol/L] above the FBG value) is common. However, since absorption defects vary in degree and are counterposed against decreased tissue metabolism, one investigator reported that 50% of hypothyroid patients tested had a diabetic type of OGTT curve with the FBG value usually normal.
Acromegaly is reported to produce an elevated FBG level in 25% of cases and a diabetic OGTT curve in 50% of cases. Growth hormone (somatotropin) is thought to be able to stimulate gluconeogenesis independently. Actually, the influence of the pituitary on carbohydrate metabolism has been mainly studied in conditions of pituitary hypofunction; in hypopituitarism, a defect in gluconeogenesis was found that was due to a combination of thyroid and adrenocorticosteroid deficiency rather than to deficiency of either agent alone.
In acute pancreatitis, perhaps 25%-50% of patients may have transient hyperglycemia. In chronic pancreatitis, abnormal glucose tolerance or outright diabetes mellitus is extremely common.
A variety of nonendocrine disorders may produce diabetogenic effects on carbohydrate tolerance. Chronic renal disease with azotemia frequently yields a diabetic curve of varying degree, sometimes even to the point of fasting hyperglycemia. The reason is not definitely known. Hyperglycemia with or without glucosuria occurs from time to time in patients with cerebral lesions, including tumors, skull fracture, cerebral infarction, intracerebral hemorrhage, and encephalitis. The mechanism is not known, but experimental evidence suggests some type of center with regulatory influence on glucose metabolism located in the medulla and the hypothalamus, and perhaps elsewhere. Similar reasoning applies to the transient hyperglycemia, sometimes accompanied by glucosuria, seen in severe carbon monoxide poisoning. This is said to appear in 50% of these patients and seems to be due to a direct toxic effect on the cerebral centers responsible for carbohydrate metabolism. Type IV lipoproteinemia is frequently associated with some degree of decreased carbohydrate tolerance, which sometimes may include fasting hyperglycemia.
Malignancies of varying types are reported to produce decreased carbohydrate tolerance in varying numbers of patients, but the true incidence and the mechanism involved are difficult to ascertain due to the presence of other diabetogenic factors such as fever, cachexia, liver dysfunction, and inactivity.
Liver disease often affects the OGTT response. This is not surprising in view of the importance of the liver in carbohydrate homeostasis. Abnormality is most often seen in cirrhosis; the degree of abnormality has a general (although not exact) correlation with degree of liver damage. In well-established cirrhosis, the 2-hour postprandial blood glucose level is usually abnormal. The FBG level is variable but is most often normal. Fatty liver may produce GTT abnormality similar to that of cirrhosis. In hepatitis virus infection there is less abnormality than in cirrhosis; results become normal during convalescence and may be normal at all times in mild disease.
Acute myocardial infarction has been shown to precipitate temporary hyperglycemia, glucosuria, or decreased carbohydrate tolerance. In one representative study, 75% of patients had abnormal GTT responses during the acute phase of infarction, with 50% of these being frankly diabetic curves; follow-up showed that about one third of the abnormal curves persisted. Besides the well-known increased incidence of atherosclerosis (predisposing to infarction) in overt or latent diabetics, emotional factors in a stress situation and hypotension or hepatic passive congestion with liver damage may be contributory.
Emotional hyperglycemia is considered a well-established entity presumably related to epinephrine effect.
OGTT responses in pregnancy were discussed earlier. Some investigators believe that the OGTT curve in gravid women does not differ from that in nulliparas, and thus any abnormalities in the curve are indicative of latent diabetes. Others believe that pregnancy itself, especially in the last trimester, tends to exert a definite diabetogenic influence (controversy here concerns how much change is considered “normal.” This view is reinforced by observations that the original synthetic estrogen-progesterone combinations used for contraception (with higher estrogen content than most current types) often mimicked the diabetogenic effects of pregnancy. This occurred in 18%-46% of patients, and, as in pregnancy, the FBG level was most often normal. The exact mechanism is not clear; some suggest altered intestinal absorption.
Salicylate overdose in children frequently produces a clinical situation that closely resembles diabetic acidosis. Salicylate in large quantities has a toxic effect on the liver, leading to decreased glycogen formation and to increased breakdown of glycogen to glucose. Therefore, there may develop a mild to moderate elevation in blood glucose level, accompanied by ketonuria. Plasma ketone test results may even be positive, although usually only to mild degree. Salicylic acid metabolites give positive results on tests for reducing substances such as Clinitest, so that results of such tests falsely suggest glucosuria. In addition, salicylate stimulates the central nervous system (CNS) respiratory center in the early phases of overdose, so that increased respiration may suggest the Kussmaul breathing of diabetic acidosis. The carbon dioxide (CO2) content (or partial pressure of CO2, (PCO2) is decreased. Later on, a metabolic acidosis develops.
Differentiation from diabetic acidosis can be accomplished by simple tests for salicylate in plasma or urine. A dipstick test called Phenistix is very useful for screening purposes. A positive plasma Phenistix reaction for salicylate is good evidence of salicylate poisoning. A positive result in urine is not conclusive, since in urine the procedure will detect nontoxic levels of salicylate, but a negative urine result is strong evidence against the diagnosis. Definitive chemical quantitative or semiquantitative tests for blood salicylate levels are available. It is important to ask about a history of medication given to the patient or the possibility of accidental ingestion in children with suspicious clinical symptoms.
Salicylate intoxication is not frequent in adults. When it occurs, there is much less tendency toward development of a pseudodiabetic acidosis syndrome. In fact, in adults, salicylate in nontoxic doses occasionally produces hypoglycemia, which tends to occur 2-4 hours postprandially.
Phenytoin (Dilantin) is reported to decrease glucose tolerance, and overdose occasionally produces a type of nonketotic hyperglycemic coma.
Posthypoglycemic hyperglycemia (Somogyi phenomenon) refers to fasting hyperglycemia produced by hormonal (epinephrine, catecholamines, growth hormone) response to previous hypoglycemia induced by too much administered insulin. This can produce the false appearance of treatment failure due to insufficient insulin.
Dawn phenomenon refers to circadian increase in plasma insulin levels between 3 A.M. and 7 A.M. This occurs in response to increased plasma glucose level produced by circadian increase in pituitary growth hormone secretion. In nondiabetic persons the plasma glucose level remains normal in response to the insulin. In diabetics, including those who are insulin dependent and some who are noninsulin dependent, impaired glucose tolerance permits plasma glucose levels to become elevated to some degree over baseline values during this time. In some patients the 6 A.M. or 7 A.M. fasting glucose level becomes sufficiently elevated to simulate necessity for higher doses of daily insulin, whereas more insulin is needed only during this limited time period.
Controversy on clinical relevance of the oral glucose tolerance test
Complete discussion of the OGTT must include reference to various studies that attack the clinical usefulness of the procedure. These consist of reports of large series of normal persons showing up to 20% flat GTT results, studies that showed different curves in repeat determinations after time lapse, and others in which various types of curves were obtained on repeated tests in the same individual. Based on the criteria in use before the NDDG recommendations, some investigators believed there is inadequate evidence that GTT abnormality actually indicates true diabetes mellitus, since many persons with abnormal GTT responses failed to progress to clinical diabetes, and the population incidence of diabetes was far short of that predicted by GTT screening. Since the NDDG criteria require somewhat more abnormality than previous criteria to make a diagnosis of diabetes mellitus, these problems have been partially corrected. Even so, not all drawbacks of the OGTT have been solved regarding problems of sensitivity, specificity, reproducibility, and clinical relevance even when the test is performed under optimal conditions. Nevertheless, at present the OGTT is still the standard test of carbohydrate tolerance and the laboratory basis for the diagnosis of diabetes mellitus.