Articles on Medical Diseases and Conditions

Category: Tests for Diabetes and Hypoglycemia

  • Neonatal and Childhood Hypoglycemia

    Neonatal and childhood hypoglycemia will be considered together, although some are of the fasting type and some are postprandial.

    Idiopathic hypoglycemia of infancy. It has been reported that as many as 10% of all neonates have at least one low blood glucose value. Neonatal reference values are lower than the adult reference range, and there is disagreement as to what level should be considered hypoglycemic. Values considered hypoglycemic in the literature range from 20-40 mg/100 ml (1.10-2.20 mmol/L). The most widely quoted reference value lower limits are 20 mg/100 ml for premature or low-birth-weight newborns and 30mg/100 ml for term normal-weight newborns during the first day of life. These values were originally derived from whole blood assay. Conversion to serum or plasma glucose assay values (15% higher than whole blood) would bring the premature lower limit on serum specimens to about 25 mg/100 ml (1.38 mmol/L) and the mature infant lower limit on serum to about 35 mg/100 ml (1.93 mmol/L). The other frequently used reference range consists of premature infant serum lower limit of 30 mg/100 ml (1.65 mmol/L) and mature infant serum lower limit of 40 mg/100 ml (2.20 mmol/L) during the first day of life. Use of these reference values results in at least 10% more diagnoses of hypoglycemia than use of the lower reference limits. Since a substantial number of these infants are asymptomatic, the lower values are more commonly used. In both systems, the serum glucose lower limit after the first day of life is 40 mg/100 ml. Idiopathic neonatal hypoglycemia is more common in male infants with low birth weight for age and in premature infants. In some cases the condition is aggravated by delay in beginning milk feeding. Clinical symptoms occur within 72 hours of birth and include some combination of tremors, twitching, cyanosis, respiratory difficulties, high-pitched or weak cry, refusal to eat, limpness, apnea, and convulsions. These symptoms are nonspecific and could be due to a wide variety of serious or life-threatening neonatal illnesses.

    Neonatal hypoglycemia is thought to be caused at least in part by low hepatic glycogen levels. The condition is usually transient and is treated by oral milk and glucose (if mild) or by parenteral glucose (if severe). If the infant does not develop a clinical problem until after milk feeding is begun and the problem continues, the possibility of leucine sensitivity should be considered.

    Leucine sensitivity. Symptoms of leucine sensitivity usually begin within the first 2 years of life and spontaneously disappear by age 5-6 years. Intake of foods that are rich in the amino acid leucine (e.g., cow’s milk) apparently stimulate the pancreatic beta cells to overproduce insulin, rather similar to the effect of tolbutamide. The patients typically have a low FBG level and demonstrate marked hypoglycemia after leucine administration. Diagnosis is made by a leucine tolerance test, similar to the OGTT but using oral leucine. Since leucine also stimulates 50%-70% of insulinomas, the possibility of insulin-producing tumor or nesidioblastosis must also be considered.

    Nesidioblastosis. Occasionally children with hypoglycemia have diffuse pancreatic islet cell hyperplasia known as nesidioblastosis. Insulinomas also occur but are rare. Diagnosis is the same as for insulinomas and is based on demonstrating inappropriate elevation of insulin when blood glucose levels are low.

    Galactosemia. Galactosemia is an inborn error of metabolism produced by deficiency of an enzyme necessary to metabolize galactose to glucose. Galactose is produced by metabolism of lactose, which is present in milk. Some of these patients develop hypoglycemic episodes, presumably due to abnormality of liver production of glucose from glycogen. A more complete summary of this condition, including laboratory diagnosis, is found in Chapter 34.

    Ketosis. Ketosis said to be the most common etiology of hypoglycemia in childhood. Onset is usually between age 1.5 and 5 years, and the condition usually disappears by age 10 years. It is more common in males of lower than normal birth weight. Episodes of hypoglycemia tend to be precipitated by prolonged lack of food intake or by a low-calorie, low-carbohydrate diet. The cause is thought to be at least partially due to depletion of hepatic glycogen. The child is normal between episodes. Hypoglycemic attacks are usually associated with readily detectable plasma ketone bodies and frequently with urine ketonuria. Blood glucose levels and plasma insulin values are both low.

    Other. Hypoglycemia may be associated with maternal toxemia of pregnancy, poorly controlled diabetes mellitus, neonatal severe hypoxia, or neonatal sepsis. Diagnosis would be based on recognition that these conditions are present.

  • Postprandial Hypoglycemia

    Some of the most common etiologies of postprandial hypoglycemia (which is also known as “reactive hypoglycemia”) include the following eiologies

    Alimentary. Postprandial Hypoglycemia of gastrointestinal tract origin (sometimes called the “dumping syndrome”) most often occurs after gastric surgery and results from unusually swift or complete gastric emptying of ingested carbohydrate into the duodenum, resulting in abnormally high blood glucose levels and temporary hypoglycemia after hastily produced insulin has overcome the initial hyperglycemia. Initial blood glucose elevation is definitely greater than that of a normal person.

    Diabetic. Some persons with subclinical or early diabetes mellitus of the NDDG type II (noninsulin-dependent) category may develop mild and transitory hypoglycemia 3-5 hours after eating. This seems to be an early manifestation of their disease, which often disappears as the disease progresses. The exact incidence in diabetics is unclear but is probably low. However, because of the large number of diabetic persons, it may be a relatively frequent cause of postprandial hypoglycemia. Initial elevation of blood glucose values may or may not be higher than in normal persons, but the 2-hour postprandial value is elevated. The rise in plasma insulin after eating tends to be delayed, and the insulin peak (when finally achieved) may be somewhat elevated, resulting in the hypoglycemic episode.

    Functional. Some patients develop symptoms of hypoglycemia after eating without known cause. This most often occurs 3-4 hours postprandially (range, 1-5 hours). In some cases symptoms can be correlated with acceptably low blood glucose values, in which case many physicians make a diagnosis of functional hypoglycemia (although there is controversy on this point, to be discussed later with the 5-hour OGTT). In other cases, probably the majority, symptoms cannot be adequately correlated with acceptably low blood glucose values. Either the OGTT serum glucose value is low but no symptoms occur, or (less commonly) symptoms occur at a relatively normal glucose level. In these cases some have used the diagnosis of “idiopathic postprandial syndrome.” Perhaps a better term would be “pseudohypoglycemia.” The peak blood glucose elevation after eating is not higher than in a normal person, and the 2-hour value is also normal.

    Other. A few patients with insulinoma or alcoholism may develop postprandial hypoglycemia, although fasting hypoglycemia is much more common.

    Laboratory tests

    The first consideration is to rule out a fasting hypoglycemia, especially when hypoglycemic symptoms occur several hours after food intake. Self-administered hypoglycemic agents are also possible, although for some reason this is not often mentioned when associated with postprandial symptoms. The best test would be a blood glucose measurement drawn at the same time that symptoms were present. Because symptoms in daily life usually occur at times when blood specimens cannot be obtained, the traditional laboratory procedure in postprandial hypoglycemia has been the 5-hour OGTT. In alimentary hypoglycemia the classic OGTT pattern is a peak value within 1 hour that is above OGTT reference limits, followed by a swift fall to hypoglycemic levels (usually between 1 and 3 hours after glucose). In diabetic hypoglycemia, there is an elevated 2-hour postprandial value, followed by hypoglycemia during the 3-5 hours postglucose time interval. In functional hypoglycemia, there is a normal OGTT peak and 2-hour level, followed by hypoglycemia during the 2 to 4-hour postglucose time interval (Fig.).

    Representative oral glucose tolerance curves.
    Representative oral glucose tolerance curves.

    Unfortunately, diagnosis has not been as simple as the classic OGTT findings just mentioned would imply. Many investigators have found disturbing variability in OGTT curves, which often change when testing is repeated over a period of time and may change when testing is repeated on a day-to-day basis. In some cases this is related to factors known to influence the OGTT, such as inadequate carbohydrate preparation, but in other cases there is no obvious cause for OGTT discrepancies. Another major problem is disagreement regarding what postprandial blood glucose value to accept as indicative of hypoglycemia. Most investigators use 50 mg/100 ml (2.76 mmol/L) of plasma glucose as the dividing line. However, there is considerable controversy in the literature on this point. In some cases it is not clear whether serum or whole blood was assayed (reference values for whole blood are 15% less than those for serum or plasma). An additional problem involves the concept of chemical hypoglycemia versus clinical hypoglycemia. A number of studies have shown that a certain percentage of clinically normal persons may have OGTT curves compatible with functional hypoglycemia without developing any symptoms. In some instances asymptomatic OGTT serum levels as low as 30 mg/100 ml (1.65 mmol/L) were observed. Moreover, in some studies continuous blood glucose monitoring systems disclosed hypoglycemic dips not evident in standard OGTT specimens. On the other hand, some persons with symptoms compatible with hypoglycemia do not develop OGTT values below reference limits. To make matters more confusing, some studies have found that when the 5-hour OGTT was repeated after it had initially displayed a hypoglycemic dip, a substantial number of the repeat OGTT results became normal (about 65% of cases in one report). Finally, several studies have indicated that hypoglycemic values found during an OGTT usually disappear when an ordinary meal is substituted for the carbohydrate dose. Since actual patients do not usually ingest pure carbohydrate meals, this casts doubt on the reliability and usefulness of the OGTT in the diagnosis of functional hypoglycemia. One study found that patients with symptoms of functional hypoglycemia had an increase in the plasma epinephrine level of at least 250 pg/ml over baseline at the time of the OGTT lowest (“nadir”) serum glucose level, regardless of the actual value, whereas those who did not develop symptoms had an epinephrine increase at the time of the glucose nadir that was less than 250 pg/ml.

    In summary, the diagnosis of postprandial hypoglycemia is clouded by controversy, especially the category of functional hypoglycemia. Probably the least debatable diagnosis would be one established by a serum glucose level less than 50 mg/100 ml obtained less than 5 hours after a regular meal while the patient is having hypoglycemic symptoms. A 5-hour OGTT could be done with a regular meal instead of the pure glucose dose. This would also help to exclude alimentary or diabetic etiologies.

  • Fasting Hypoglycemia

    Of the two clinical categories, fasting hypoglycemia is by far the more straightforward. The two chief mechanisms are insulin excess (either absolute or relative) and effects of carbohydrate deprivation. Conditions acting through these mechanisms are the following:

    1. The most well-known etiology of insulin excess is the beta-cell insulin-producing pancreatic islet cell tumor (insulinoma). About 80% of these tumors are single adenomas, about 10% are multiple adenomas, and about 10% are carcinomas. About 10% of patients with insulinoma also have the MEN type II syndrome (Chapter 33), and of these patients, about 80% have multiple insulinomas. Although insulinomas typically are associated with fasting hypoglycemia, one study reported that nearly 50% of their insulinoma patients developed symptoms less than 6 hours postprandially, which would clinically suggest a postprandial (reactive) etiology.
    2. Overdose of insulin in a diabetic person. Since the patient may be found in coma without any history available, insulin overdose must be a major consideration in any emergency room comatose patient.
    3. Hypoglycemia may be associated with deficiency of hormones that normally counteract the hypoglycemic effect of insulin. This group includes pituitary insufficiency (deficiency of growth hormone and cortisol) and adrenal insufficiency (cortisol). Glucagon and thyroxine also have a hyperglycemic effect, but hypoglycemia due to a deficiency of these hormones is rare.
    4. Prolonged carbohydrate deprivation, even when total calories are relatively adequate, has been reported to predispose to hypoglycemia. Occasionally patients with severe liver disease develop hypoglycemia, but this is uncommon.
    5. Certain nonpancreatic tumors have occasionally been reported to cause hypoglycemia, presumably either by glucose utilization or by production of an insulin-like substance. The great majority have been neoplasms of large size described as fibrosarcomas or spindle cell sarcomas, usually located in the abdomen (in one study, about two thirds occurred in the abdomen and one third occurred in the thorax). Hepatoma is next most frequent.
    6. Alcoholic hypoglycemia may occur in either chronic alcoholics or occasional drinkers. Malnutrition, chronic or temporary, is an important predisposing factor. In those persons who develop hypoglycemia, fasting for 12-24 hours precedes alcohol intake. Symptoms may occur immediately but most often follow 6-34 hours later. Therefore, alcohol-related hypoglycemia is most often a fasting type but occasionally may appear to be reactive.

    Laboratory tests

    The major consideration in these patients is to discover a pancreatic islet cell tumor or eliminate its possibility. Diagnosis of islet cell tumor is important since the condition is surgically correctable, but accurate diagnosis is a necessity to avoid unnecessary operation.

    Whipple’s triad. Whipple’s triad remains the basic screening procedure for insulinoma:

    1. Symptoms compatible with hypoglycemia while fasting.
    2. A FBG level of 10 mg/100 ml (0.55 mmol/L) or more below FBG normal lower limits at some time. This is most reliable when the specimen is obtained while the patient is having symptoms.
    3. Relief of symptoms by glucose.
    4. In addition, most endocrinologists would require an elevated serum insulin value during the hypoglycemia.

    Plasma (or serum) insulin/glucose ratio. In functioning insulinomas, one would expect serum insulin levels to be elevated. In some patients with serum glucose within glucose reference range, serum insulin may be elevated not only with insulinoma but also with Cushing’s syndrome, chronic renal failure, growth hormone overproduction, cortisol-type steroids, obesity, and estrogen therapy. On the other hand, about 10% (range, 0%-20%) of patients with insulinoma have been reported to show serum insulin levels in the upper 75% of insulin reference range during hypoglycemia. However, some of these patients with apparently normal insulin levels can be shown to have insulin values that are disproportionately high in relation to the glucose value. Therefore, some investigators consider the ratio of immunoreactive insulin to glucose (IRI/G ratio) to be more sensitive and reliable than blood levels or either glucose alone or insulin alone. Normally, the IRI/G ratio should be less than 0.3 (the immunoreactive insulin is measured in microunits per milliliter, the glucose in milligrams per 100 ml). The ratio is abnormal if it is greater than 0.3 in nonobese persons and greater than 0.3 in obese persons with serum glucose values less than 60mg/100 ml. Hyperinsulinism due to insulinoma results in serum insulin levels inappropriately high in relation to the low serum glucose values. In some institutions a G/IRI ratio is used instead of an IRI/G ratio.

    Amended insulin/glucose ratio for diagnosis of insulinoma. Some investigators have proposed variants of the IRI/G ratio to increase sensitivity or specificity for insulinoma. The most commonly used variant is the “amended ratio” of Turner, whose formula is:

    Serum insulin level x 100 / Serum glucose – 30 mg/100 ml

    when serum insulin is reported in µU/ml and serum glucose in mg/100 ml. A ratio over 50 suggests insulinoma, while a ratio less than 50 is evidence against insulinoma. The serum levels of glucose and insulin are obtained after fasting, which may have to be long-term. The amended IRI/G ratio is reported to be a little more sensitive than the standard IRI/G ratio in some reports, but in other reports there were 20%-35% false negative or nondiagnostic results in patients with insulinoma.

    Prolonged fasting. A considerable number of patients do not display symptoms or laboratory evidence of insulinoma after overnight fasting. When this occurs, the most useful procedure is prolonged fasting for a time period up to 72 hours long, with periodic insulin plus glucose measurements, or such measurements if symptoms develop. After an overnight fast, approximately 50% of insulinomas are revealed by the FBG level alone and about 66% by the IRI/G ratio. After a 48-hour fast, the blood glucose value uncovers about 66% of tumors and the IRI/G ratio about 85%. In 72 hours, the blood glucose level alone detects about 70% of patients and the IRI/G ratio is abnormal in more than 95%. In the Mayo Clinic series, Whipple’s triad appeared within 24 hours in 71% of insulinoma patients, within 36 hours in 79%, within 48 hours in 92%, and within 72 hours in 98%. Serum insulin assay alone reveals about the same percentage of patients with elevated values as the blood glucose level does with low values.

    Tolbutamide tolerance test. Tolbutamide (Orinase) is a sulfonylurea drug that has the ability to stimulate insulin production from pancreatic islet cells. The drug has been used to treat diabetics who produce insulin but not in sufficient quantity. In the tolbutamide test, a special water-soluble form of tolbutamide (not the therapeutic oral form) is given intravenously. In normal persons there is a prompt fall in blood glucose levels to a minimum at 30-45 minutes, followed by a return to normal values between 1.5 and 3 hours. In patients with insulinoma the fall in the blood glucose level is greater than that of most normal persons, declining to 40%-65% of baseline, whereas normal persons usually do not fall as low as 50% of baseline [Mayo Clinic recent criteria report sensitivity and specificity of 95% when the average of the 120-, 150-, and 180-minute plasma glucose specimens is less than 55 mg/100 ml (3.05 mmol/L) in lean persons and 62 mg/100 ml (3.44 mmol/L) in obese persons].

    Since there is occasional overlap between normal persons and those with insulinoma, of greater significance is the fact that hypoglycemia from insulinoma persists for more than 3 hours, whereas in most normal individuals the blood glucose level has returned to fasting values by 3 hours. In a few normal individuals and in those with functional hypoglycemia, values return to at least 80% of FBG levels by 3 hours. Adrenal insufficiency also returns to at least 80% of FBG levels by 3 hours, although the initial decrease may be as great as that for insulinoma. Some patients with severe liver disease have curves similar to those of insulin-producing tumor. However, this is not frequent and usually is not a real diagnostic problem. The tolbutamide test is definitely more sensitive than the OGTT for the diagnosis of islet cell tumor but has the disadvantage that the characteristic responses of diabetic or alimentary hypoglycemia to the OGTT cannot be demonstrated.

    The major drawback to the tolbutamide test is the necessity in some patients to stop the test prematurely because of severe hypoglycemic symptoms. Patients must be closely watched during the test. Sensitivity of the test for insulinoma seems to be approximately 80%-85% using the extended 3-hour time period (literature range, 75%-97%). A major advantage of the test is the short time required to obtain an answer. The tolbutamide test is usually not performed if the FBG value is already in the hypoglycemic range.

    Proinsulin. Insulin is derived from a precursor called proinsulin, which is synthesized in the pancreas, is metabolically inactive, and is larger in size (“big insulin”). Proinsulin consists of an alpha and a beta chain connected by an area called “connecting peptide” (C-peptide). Proinsulin is enzymatically cleaved within beta cells into equal quantities of insulin and C-peptide. Radioimmunoassay measurement of insulin includes both proinsulin and regular insulin. Normally, about 5%-15% of immunoreactive insulin (that substance measured by immunoassay) is proinsulin. In many (but not all) patients with insulinomas, the amount of circulating proinsulin is increased relative to total insulin. In diabetics with insulin deficiency who are being treated with insulin, the proinsulin fraction of the individual’s own immunoreactive insulin values may be increased. Measurement of proinsulin necessitates special procedures such as Sephadex column chromatography and is not widely available. Sensitivity of proinsulin assay for detection of insulinoma is reported to be about 80%.

    Serum connecting peptide (C-peptide) measurement. As noted previously, C-peptide is a by-product of insulin production. Although it is released in quantities equal to insulin, serum C-peptide levels do not exactly parallel those of insulin, due to differences in serum half-life and catabolic rate. Nevertheless, C-peptide values correlate well with insulin values in terms of the position of each in relation to its own normal range (i.e., if one is decreased, the other likewise is decreased). Therefore, C-peptide can be used as an indicator of insulin secretion.

    Some insulin is derived from animal pancreas (synthetic human insulin is available but not yet exclusively used). Use of this foreign substance may lead to antiinsulin antibody production. Pork insulin is less antigenic than beef insulin. Insulin antibodies falsely increase insulin assay values in most commercial kits (although in a few systems the values are decreased instead). C-peptide assay kits do not react with animal-origin insulin and therefore reflect only actual patient insulin production without being affected by the presence of insulin antibodies.

    C-peptide assay has been used in several ways: (1) most commonly, to detect or prove factitious (self-medication) insulin-induced hypoglycemia, (2) to detect insulinoma in diabetic patients requiring insulin therapy, (3) to evaluate pancreatectomy status, and (4) to evaluate insulin reserve or production in diabetics who have taken or are taking insulin.

    Occasionally patients become hypoglycemic by self-administration of insulin. Since insulin assay cannot differentiate between exogenous insulin and that produced by insulinoma, C-peptide assay should be performed on the same specimen that showed elevated insulin levels. In hyperinsulinism from islet cell tumor, C-peptide levels are elevated; in that due to exogenous (self-administered) insulin, C-peptide levels are low. Another cause of low C-peptide levels is a type I diabetic who cannot produce insulin; but the insulin assay value would be low. An elevation of the C-peptide level classically suggests insulinoma but may also be seen after taking oral hypoglycemic agents (since these agents stimulate the production of insulin).

    Connecting peptide assay has been used after pancreatectomy to evaluate the possibility of residual pancreatic islet tissue.

    Some investigators have used C-peptide assay in diabetics previously treated with insulin to see how much insulin production is possible. Those who have substantial capability for insulin production are treated differently from those who do not. This method can help diagnose the syndrome of peripheral insulin resistance. Also, if the diabetic patient has significant capability for insulin production, frequent and severe episodes of diabetic ketoacidosis may suggest some factor other than insulin deficiency. There is still controversy over criteria dividing type I and type II diabetics based on insulin or C-peptide assay and what therapeutic changes, if any, should be made based on the amount of insulin production using insulin or C-peptide information.

    Other tests. The 5-hour OGTT after overnight fasting was widely used in the past for detection of insulinoma. Patients characteristically had a low or low-normal FBG level and a normal sharp rise after glucose administration (the peak remaining within normal OGTT limits), then a slower fall to hypoglycemic levels that did not rapidly return to the reference range. However, the OGTT has been virtually abandoned for diagnosis of insulinoma because a considerable minority of insulinomas are reported to demonstrate a flat curve or sometimes even a diabetic-type curve, or the curve may be normal. The 5-hour OGTT, however, is sometimes used in patients with the postprandial type of hypoglycemia. Leucine and glucagon provocative tests for insulinoma have been reported. However, these are rarely used since they are somewhat less sensitive than tolbutamide and substantially less sensitive than the IRI/G ratio after prolonged fasting.

    Current status of tests for insulinoma

    At present, most investigators use Whipple’s triad and prolonged fasting (with insulin assay or the IRI/G ratio) as the primary screening tests for insulinoma, with the tolbutamide test available in equivocal or problem cases. The differential diagnosis of decreased glucose and elevated insulin includes insulinoma, factitious hypoglycemia, and antiinsulin antibodies. Insulinoma has increased C-peptide levels and the other two have normal or decreased C-peptide levels.

  • Hypoglycemia

    Hypoglycemia is a topic that has generated a great deal of confusion. Although the word means “low blood glucose,” the diagnosis of hypoglycemia is controversial, because it is sometimes defined strictly on the basis of an arbitrary blood glucose level (chemical hypoglycemia), sometimes in terms of symptoms (clinical hypoglycemia), and sometimes as a combination of glucose level and symptoms. The most readily accepted aspect of hypoglycemia is division into two clinical categories: one in which symptoms occur after fasting (fasting hypoglycemia) and one in which symptoms occur after eating (postprandial hypoglycemia). If the blood glucose level drops rapidly, symptoms tend to be similar to those associated with release of epinephrine (adrenergic) and include anxiety, sweating, palpitation, tremor, and hunger. If hypoglycemia persists, CNS glucose deprivation occurs (neuroglycopenia) and symptoms resemble those of cerebral hypoxia, such as lethargy, headache, confusion, bizarre behavior, visual disturbances, syncope, convulsions, and coma. Symptoms associated with fasting hypoglycemia tend to be one or more of those associated with CNS neuroglycopenia, and those associated with postprandial hypoglycemia tend to be adrenergic. However, there is some overlap in symptoms between the two groups. In general, symptoms due to fasting hypoglycemia have a much higher incidence of visual disturbances, bizarre behavior or personality changes, convulsions, and loss of consciousness, especially prolonged loss of consciousness. Postprandial hypoglycemia tends to be more abrupt in onset and is usually self-limited.

  • Glucosuria

    Besides measurement of blood glucose or carbohydrate tolerance, certain other procedures are widely used or proposed for the detection of diabetes mellitus. The appearance of glucose in the urine has long been used both for detection and as a parameter of treatment. As a clue to diagnosis, urine glucose depends on hyperglycemia that exceeds the renal tubular threshold for glucose. This threshold is most often considered to be a venous plasma true glucose value of 180 mg/100 ml (1.0 mmol/L); (however, there is a range in the literature of 150-200 mg/100 ml). Of some interest regarding the threshold concept in diabetics is evidence that some diabetics (especially the elderly) possess unusually high thresholds (up to 300 mg/100 ml; 16.6 mmol/L). It has also been shown that arterial blood glucose levels are much better correlated with glucosuria than venous ones. Nevertheless, routine urine testing provides one method for practical continuous outpatient monitoring of therapy and for the prevention of ketoacidosis. This aspect provides another argument for more routine use of the full GTT, since glucosuria can be correlated with degree of hyperglycemia. Incidentally, many diabetic patients and many of those involved in mass surveys have a urine glucose test before breakfast, which is the least likely time to produce glucosuria.

    The problem of causes of hyperglycemia not due to diabetes mellitus was discussed earlier. Renal threshold assumes importance in another way because of the condition known as “renal glucosuria.” This may be congenital or acquired; the acquired type may be idiopathic or secondary to certain diseases such as Fanconi’s syndrome, acute tubular necrosis, or renal rickets. In all these conditions there is glucosuria at lower blood glucose levels than normal renal threshold values. Some report that a significant number of patients with the nonfamilial idiopathic variety of renal glucosuria eventually develop overt diabetes mellitus, although others do not agree.

    Glucosuria of pregnancy occurs in the last trimester. Reported incidence depends on the sensitivity of the testing methods used, ranging from 5%-35% or even 70%. The etiology seems to be a combination of increased glomerular filtration rate and temporarily decreased renal threshold. Lactosuria is even more common. Glucosuria without hyperglycemia occurs in 20% of patients with lead poisoning. This is due to a direct toxic effect on the renal tubule cells. Glucosuria of a transient nature has been reported in 24% of normal newborn infants. A study utilizing paper chromatography found galactosuria, usually in amounts too small for detection by routine techniques, to be even more common.

    Mentioned here only for the sake of completeness are the two main types of urine glucose tests: the copper sulfate tests for reducing substances and the glucose oxidase enzyme papers. The merits, drawbacks, and technical aspects of these tests, as well as a general discussion of glucosuria.

    Diabetic proteinuria

    The earliest evidence of diabetic renal disease is glomerular basement membrane abnormality on renal biopsy using special stains or electron microscopy. This is present in nearly all type I patients by 2-5 years after onset. The structural changes initially produce increase in glomerular permeability that in turn results in increased urinary excretion of certain molecules (such as albumin and immunoglobulin-G) that are filtered by the glomeruli. Initially, the degree of abnormality is small enough that urinary albumin remains within reference range during at least the first 5 years after initial diagnosis of type I diabetes. Afterward, there is a variable number of years during which about 30% of patients (range, 12%-43%) increase urinary albumin above reference range but below threshold of detection (200-300 mg/L) by standard laboratory urinalysis dipstick protein tests. This “subclinical” state of selectively elevated albumin excretion rate (AER) is called “microalbuminuria”. This sequence also occurs in a substantial number of type II diabetics (about 30%, range 13%-59%). After a variable number of years, about 70% (range, 14%-100%) of type I patients gradually increase albumin excretion until it is “overt”; that is, detectable by routine laboratory protein dipstick screening methods. This eventually happens in at least 35%-40% (range, 2%-60%) of all type I diabetics, although about 40% never develop overt albuminuria. Progression from microalbuminuria to overt albuminuria in type II diabetics occurs in about 20%-25% of patients, with about 25% (range, 3%-40%) of all type II patients reaching this stage. Once overt albuminuria occurs, most type I diabetics eventually progress to renal failure (65%-75%, range, 50%-100%) unless death occurs from coronary heart disease or some other cause. Renal failure occurs in about 30% (range, 22%-40%) of all type I patients. Overall, diabetics comprised 30% of all patients in 1987 who had end-stage renal disease; of the diabetics, type I and II were represented in equal numbers (type I is more apt to progress to renal failure; but type II occurs nearly 10 times more frequently than type I). All these sometimes conflicting statistics are influenced by many factors, such as patient age at diagnosis, number of years followed, type of microalbumin test used, and patient racial group composition. African Americans, native Americans, and Hispanics have higher rates of progressive diabetic renal disease than Europeans. The rationale for detecting microalbuminuria is to find disease at a stage in which certain therapies might retard or even prevent further impairment. By the time overt albuminuria develops, there is no current way to prevent progression.

    Microalbuminuria has been defined in several ways; there is not unanimous opinion which is the best screening method or “gold standard” method. Based on a Consensus Conference held in 1989, the following definitions currently appear to be most widely accepted: excretion rate, 30-300 mg/24 hrs or 20-200 µg/min; excretion ratio, 20-200 mg albumin/gm creatinine (0.4-2.8 mg/mmol creatinine). There is also controversy whether to employ overnight specimens, 24-hour specimens, early morning specimens, or random specimens. 24-hour specimens are generally considered “gold standard”; however, since albumin excretion increases in the upright position and during exercise, plus the problems of incomplete 24-hour collections, many investigators prefer overnight collection as a baseline. In general, timed collections are thought to be more accurate than untimed ones, and some studies obtained more accurate and reproducible results using an albumin/creatinine ratio, which would partially correct for differences in urine volume and concentration. Finally, several investigators advocate an early morning untimed specimen for screening purposes (microalbuminuria range, 20-30 mg/L). Impacting on all these techniques is a rather high percentage of variability (30%-45%) in day-to-day albumin excretion in diabetics with or without microalbuminuria. There is also significant assay technical variation that can be as high as 20%-40%, depending on the analytical method, the quantity of albumin, and the laboratory. Therefore, it is strongly recommended that at least 2 of 3 specimens be abnormal during a 6 month time period before diagnosing microalbuminuria.

    Assay methods for microalbuminuria include quantitative immunoassay (ELISA or particle agglutination methods using nephelometry); qualitative “yes-no” agglutination slide immunoassay using anti-albumin antibodies (e.g., AlbuSure, 20mg/L detection level); and chemical methods (e.g., Microbumintest tablets). The quantitative assays are advantageous in establishing a baseline value and disclosing worsening of disease if it occurs. Of the qualitative screening tests, Microbumintest has been criticized by some for false positive results and some false negative results. AlbuSure is said to produce acceptable results. Other tests are available, but with insufficient evaluation data. With any method it is possible to obtain false positive results if the urine specimen is contaminated with blood. The specimen should be assayed fresh (i.e., within 12 hours); if not possible, it can be refrigerated (acceptable for 7 days) or a suitable preservative can be added. There are conflicting reports whether freezing lessens albumin content. Some albumin may adhere to the walls of glass collection bottles.

    Finally, it must be remembered that various conditions other than diabetes (e.g., atherosclerosis, hypertension, infection, collagen diseases, glomerulonephritis) may increase urinary albumin as a component of ordinary proteinuria induced by either focal or diffuse acute or chronic renal damage. Theoretically, these conditions would cause detectable proteinuria on standard dipstick protein screening tests.

    The American Diabetes Association (1989) recommends that urine microalbuminemia should be assayed yearly in all type II diabetics and yearly beginning 5 years after diagnosis in all type I diabetics (unless the patient has known diabetic progressive nephropathy).

    Diagnosis of diabetic coma

    Diabetic coma may occur without a history of diabetes or in circumstances where history is not available. Other major etiologies of coma must be considered; including insulin hypoglycemia, meningitis or cerebrovascular accident, shock, uremia and barbiturate overdose. A clear-cut, fast diagnosis of diabetic coma can be made with a test for plasma ketones (frequently called “acetone,” although acetone is not the only ketone substance). Anticoagulated blood is obtained, a portion is centrifuged for 2-3 minutes, and the plasma is tested for ketones. Diabetic acidosis severe enough to produce coma will be definitely positive (except for the rare cases of lactic acidosis or hyperosmolar coma). The other etiologies for coma will be negative, since they rarely produce the degree of acidosis found in diabetic coma. The presence of urinary glucose and ketones strongly suggests diabetes but may occur in other conditions. Such findings would not entirely rule out insulin overdose (always a consideration in a known diabetic), since the urine could have been produced before the overdose. An elevated blood glucose level also is strong evidence of diabetic coma, especially if the degree of elevation is marked. Other conditions that might combine coma with hyperglycemia (cerebrovascular accident, acute myocardial infarction) have only mild or moderate hyperglycemia in those instances where hyperglycemia is produced. Besides blood glucose determination, a simple empirical test to rule out hypoglycemia is to inject some glucose solution intravenously. Cerebral damage is investigated by cerebrospinal fluid examination or computerized tomography scan. Uremia is determined by means of the blood urea nitrogen (BUN) level, although other etiologies of coma besides primary renal disease may be associated with an elevated BUN level. Drug ingestion is established by careful history, analysis of stomach contents, and identification of the drug in blood samples (one anticoagulated and one clotted specimen are preferred) or urine samples. Shock is diagnosed on the basis of blood pressure; further laboratory investigation depends on the probable etiology.

    Hyperosmolar nonketotic coma. Hyperosmolar nonketotic coma is uncommon but is being reported with increased frequency. The criterion for diagnosis is very high blood glucose level (usually well above 500 mg/100 ml; 28.0 mmol/L) without ketones in either plasma or urine. The patients usually become severely dehydrated. Plasma osmolality is high due to dehydration and hyperglycemia. Most patients are maturity-onset mild diabetics, but nondiabetics may be affected. Associated precipitating factors include infections, severe burns, high-dose corticosteroid therapy, and renal dialysis. Occasional cases have been reported due to phenytoin and to glucose administration during hypothermia.

    Lactic acidosis syndrome. Lactic acidosis syndrome is rare and may have several etiologies. It used to be most frequently reported with phenformin therapy of diabetes but now is encountered as a nonketotic form of diabetic acidosis. The most common cause of elevated blood lactate levels is tissue hypoxia from shock. Arterial blood is said to be more reliable than venous for lactic acid determination. Tourniquet blood stagnation must be prevented, and the specimen must be kept in ice until analyzed.

  • Autoantibodies Associated with Diabetes

    About 60%-90% of type I (insulin-dependent) diabetics have antibody against islet cell cytoplasmic glycoprotein (“islet cell autoantibody”) at the time of diagnosis, and many of those initially without this antibody develop it later. This antibody disappears within 2 years after appearance in 85%-90% of type I diabetics. It has also been reported in about 20% of type II diabetics and about 10% of gestational diabetics at time of diagnosis. About 30%-50% of children have autoantibody against insulin (antiinsulin antibody) at time of diagnosis before beginning insulin therapy and some (much less than formerly) develop it after using therapeutic insulin. Some patients have autoantibodies against beta cell surface antigen (beta cell antibodies). Over 95% of type I patients possess the human lymphocyte antigen (HLA) DR3 or DR4. However, at present these autoantibodies and HLAs are not being widely used in clinical medicine or in diagnosis.

  • Serum Fructosamine Assay

    Besides Hb A, albumin and various globulins may undergo nonenzymatic glycosylation. In contrast to hemoglobin, which has a serum half-life of about 60 days, albumin has a half-life of about 17-20 days, and total protein (roughly one half albumin and one half globulins) has a half-life of about 30 days. Either glycosylated albumin or glycosylated total protein can be assayed, but most laboratories assay total protein using the fructosamine procedure. This does not involve the sugar fructose and is based on biochemical reaction with glucose bound to protein with a ketoamine linkage, most often using nitro blue tetrazolium as the reagent. Serum fructosamine assay results indicate average glycosylation within the preceding 2-week time period (range, 1-3 weeks). This time period is considerably shorter than that of glycoHb but substantially longer than that for labile hemoglobin glycosylation. Drawbacks of fructosamine assay include changes in serum level due to changes in albumin rather than blood glucose. According to one report, changes in albumin affect fructosamine levels significantly only if decreased albumin levels are due to increased catabolism (decreased half-life) or increased albumin loss, but not when there is decreased metabolism of protein. Reducing substances in serum may interfere with the assay in some methods.

  • Glycosylated Hemoglobin (GLYCOHB) Assay

    In adults, hemoglobin A (Hb A) constitutes about 97%-98% of normal hemoglobin; the remainder includes about 2.5% hemoglobin A2 and about 0.5% hemoglobin F (Hb F). About 6%-7% of Hb A consists of Hb A molecules that have been partially modified by attachment of a glucose molecule to the terminal valine amino acid of the globin beta chain. This process is called “glycosylation,” and this particular glycosylated hemoglobin is called “hemoglobin A1” (Hb A1). Although Hb A1 comprises the great majority of glycosylated hemoglobin under usual conditions, glycosylation to some degree may occur at other locations in the globin chain and in other hemoglobins besides Hb A. The sum of the various glycosylation activities occurring in all hemoglobins (normal or abnormal) in the patient is known as total glycosylated hemoglobin.

    Glycosylation of hemoglobin occurs during exposure of red blood cells (RBCs) to plasma glucose; hemoglobin and glucose can form a bond that initially is labile but then becomes stable. Once stable bonding occurs, it is very slowly and poorly reversible. In Hb A1, the labile bonding fraction normally constitutes about 10% of total glucose bonding. Formation of Hb A1 occurs very slowly during the entire 120-day life span of the RBC, and the number of Hb A molecules affected by glycosylation depends on the degree and duration of RBC exposure to glucose. Hemoglobin A1 is actually composed of three hemoglobins: A1a, A1b, and A1c. Of these, Hb A1c is about 70% glycosylated, whereas the other two are less than 20% glycosylated. In addition, Hb A1c constitutes about 60%-70% of total Hb A1. Since Hb A1 comprises the majority of the predominant glycosylated Hb A fraction, under usual conditions Hb A1c therefore represents the majority of glycosylated hemoglobin. Because of this relationship the term glycosylated hemoglobin (or glycoHb) has been used for both Hb A1 and its major component Hb A1c, which sometimes is confusing. The components of total glycosylated Hb, Hb A1, and Hb A1c are shown in the box. There is a strong correlation between all three parameters, and, in most circumstances, any of the three can provide clinically useful information. However, there are differences, and in some cases one or the other is more advantageous.

    Components of Hb A1, A1c, and Total GlycoHb

    Hb A1c
    Glycosylated Hb A1c
    Nonglycosylated Hb A1c
    Hb A1
    Glycosylated Hb A1c
    Nonglycosylated Hb A1c
    Hb A1a + Hb A1b
    Negatively charged non-A glycosylated hemoglobins*
    Total Glycosylated Hb (Affinity method)
    Glycosylated Hb A1c
    Nonglycosylated Hb A1c
    Hb A1a + Hb A1b
    Negatively charged non-A glycosylated hemoglobins*
    Positively charged non-A glycosylated hemoglobins†
    Hb A glycosylated elsewhere than Hb A1 sites
    __________________________________________________
    *Hb Bart’s, F, G, H, I, J (Baltimore), M, and N.
    †Hb A2, C, D, E, and S.

    An increase in glycoHb quantity can be produced by recent very high short-term increases in blood glucose (in which case labile bonding is primarily affected), but is most often caused either by relatively continual elevation of blood glucose or by intermittent elevations that are frequent enough to produce abnormally high average glucose levels (in both of these cases stable glycosylation is primarily affected). A measurable increase in glycosylated (stable) hemoglobin begins about 2-3 weeks (literature range, 1-4 weeks) after a sustained increase in the average blood glucose level and takes at least 4 weeks to begin decreasing after a sustained decrease in the average blood glucose level. GlycoHb assay represents the averaged blood glucose levels during the preceding 2-3 months (literature range, 1-4 months). In contrast, blood glucose increases or decreases of “moderate” (100 mg/100 ml; 5.55 mmol/L) degree that occur within the 3 days just before Hb A1 measurement add sufficient labile component so as to constitute as much as 15% (range, 12%-19%) of the glycoHb result. Spontaneous sudden decreases in blood glucose of this magnitude are not common, so that under most circumstances a normal glycoHb level is good evidence of relatively normal average blood glucose during at least the preceding 4 weeks. Most of the clinical problems with labile bonding component occur when it produces false increases in glycoHb levels. In summary, an elevated glycoHb level is most often due to long-term average blood glucose elevation over the preceding 2-3 months, but the possibility exists for elevation due to marked short-term blood glucose increase if an assay method is used that is not specific for stable bonding.

    GlycoHb measurement has been used to monitor effectiveness of (long-term) diabetic therapy, to monitor patient compliance with therapy, and to differentiate between short-term stress-related glucose tolerance abnormality (e.g., acute myocardial infarction) and diabetes. Of these, the most widely accepted indications are monitoring of diabetic therapy effectiveness and monitoring of patient compliance. GlycoHb assay has also been used to diagnose diabetes mellitus, but this is controversial.

    Laboratory methods

    As noted previously, glycoHb can be measured as either total glycoHb, Hb A1, or Hb A1c (since most of total glycoHb is Hb A1 and most of Hb A1 is Hb A1c). The majority of commercially available kits measure Hb A1 and report the result as a percentage of total hemoglobin. There are a variety of assay methods. Currently, most commercial kits assaying Hb A1 or A1c use some method involving ion exchange resin. Less than 20% use agar electrophoresis or high performance liquid chromatography. Total glycoHb is assayed by a special boronic acid resin that reacts only with the stable glycated fraction and does not need pretreatment. There are surprisingly few evaluations of different glycoHb kits. In some, it is difficult to tell what they are measuring. In several kits evaluated in my laboratory there was significant variation in reproducibility and accuracy.

    Sources of error. Most ion-exchange resin-based kits do not differentiate between labile and stable glucose bonding to hemoglobin. Certain techniques available can eliminate the labile fraction before testing the patient serum. Many hemoglobins can form glycoHb to some extent. However, with some ion-exchange resin methods for A1 or A1c, positively charged non-A hemoglobins do not elute from the resin with Hb A1 or A1c but instead remain on the resin with nonglycosylated Hb A (see the box). These hemoglobins (such as Hb S and HbC) may produce glycoHb assay values that are less than true levels because these abnormal hemoglobins have a glycosylated component that is not being measured along with Hb A1. On the other hand, negatively charged non-A hemoglobins such as Hb F and Hb H elute from the resin in the same fraction as Hb A1. Therefore, an increased Hb F value or the presence of Hb H could falsely increase Hb A1 or A1c values since they are included in Hb A1 assay. The hemoglobin F value may be increased in young infants, in up to 17% of pregnant women, and in patients with some of the hemoglobinopathies. Therefore, it may be advantageous to use a method such as total glycoHb by boronic affinity when there are significant numbers of patients who are not of northern European descent. Some of the resin methods are affected by temperature changes, and in some, chronic renal failure has been reported to produce falsely high results. A few reports have described false increase with aspirin, alcoholism, and lead poisoning. It is necessary to find out what will falsely increase or decrease any A1 or A1c method. Total glycoHb measure by boronic acid chromatography includes the results of abnormal hemoglobin glycation as well as Hb A glycation and is not affected by renal failure, aspirin, or temperature fluctuations. Hemolytic anemia may produce falsely low glycoHb values with any method because hemolysis results in a shortened RBC life span and RBCs therefore are not exposed to blood glucose as long as a normal RBC. This is accentuated by bone marrow reticulocyte response since the reticulocytes are young cells with no glucose exposure. Frequent episodes of hypoglycemia might decrease glycoHb levels somewhat. Finally, there is some difficulty in calibration of assay kits because primary standards (i.e., material with substance values that are known with absolute certainty) are not available.

    In summary, glycoHb assay provides information of great value in the treatment of diabetes and in certain cases may help in the diagnosis. However, the sensitivity and reliability of some commercial kits still need improvement.

  • Plasma (or Serum) Insulin Assay

    Insulin was the first hormone measured successfully by radioisotope immunoassay, and insulin assay is now available in most sizable reference laboratories. Insulin is excreted primarily through the kidneys. In general, juvenile diabetics have low fasting insulin levels, and an OGTT using insulin determinations usually produces a flat curve. Mild diabetics have normal fasting insulin levels and display an insulin GTT curve that has a delayed rise, either to normal height or to a point moderately above normal; in either case the curve thereafter falls in a normal fashion. Decreased tolerance due to many other causes produces similar curves; an insulin OGTT has not been more efficient in uncovering subclinical diabetes than blood glucose OGTT. Some maintain that the ratio of insulin values to glucose values obtained on the same specimen during the OGTT is more reliable than insulin values alone. At any rate, most investigators believe that, at present, plasma insulin levels should not be used for diagnosis of diabetes mellitus.

    Plasma anticoagulated with ethylenediamine tetraacetic acid (EDTA) is reported to produce plasma insulin values equal to serum, but heparin is said to be associated with plasma insulin values greater than serum.

    Patients being treated with insulin frequently develop antiinsulin antibodies after approximately 6 weeks. These antibodies interfere with insulin RIA measurement by competing with insulin antibodies used in the test. Whether values will be falsely increased or decreased depends on the method used. Endogenous antibodies do not interfere with tolerance tests, since the quantity of endogenous antibody remains unchanged throughout the test; only the baseline value is affected.

  • Intravenous Glucose Tolerance Test

    The intravenous glucose tolerance test (IVGTT) was devised to eliminate some of the objections to the OGTT. Standard procedure for the IVGTT is as follows: The patient has a 3-day high-carbohydrate preparatory diet. After the FBG level is measured, a standard solution of 50% glucose is injected intravenously over a 3- to 4-minute period 0.33 gm/kg ideal body wt. Blood is obtained at 0.5, 1, 2, and 3 hours, although it would seem more informative to omit the 30-minute specimen and substitute a 1.5-hour sample. The curve reaches a peak immediately after injection (300-400 mg/100 ml [16.7-22.2 mmol/L], accompanied by glucosuria), then falls steadily but not linearly toward fasting levels. Criteria for interpretation are not uniform. However, most believe that a normal response is indicated by return to fasting levels by 1-1.25 hours. The height of the curve has no significance. Most agree that the IVGTT response is adequately reproducible. In diabetes, fasting levels are not reached in 2 hours and often not even by 3 hours. The curve in liver disease most characteristically returns to normal in 1.25-2 hours; however, some patients with cirrhosis have a diabetic-type curve. Many of the same factors that produce a diabetogenic effect on the OGTT do likewise to the IVGTT; these include carbohydrate deprivation, inactivity, old age, fever, uremia, stress, neoplasms, and the various steroid-producing endocrine diseases. There are, however, several differences from the OGTT. Alimentary problems are eliminated. The IVGTT is said to be normal in pregnancy and also in hyperthyroidism, although one report found occasional abnormality in thyrotoxicosis. The IVGTT is conceded to be somewhat less sensitive than the OGTT, although, as just noted, a little more specific.