In the postabsorptive state (10-12 hour overnight fast), the majority of total body glucose disposal takes place in insulin dependent tissues (1). Under basal conditions approximately 50% of all glucose utilization occurs in the brain, which is insulin independent and becomes saturated at a plasma glucose concentration of about 40 mg/dl (2). Another 25% of glucose uptake occurs in the splanchnic area (liver plus gastrointestinal tissues) and also is insulin independent (3). The remaining 25% of glucose metabolism in the postabsorptive state takes place in insulin-dependent tissues, primarily muscle (4,5). Basal glucose utilization averages ~2.0 mg/kgomin and is precisely matched by the rate of endogenous glucose production (1,3-7) . Approximately 85% of endogenous glucose production is derived from the liver, and the remaining amount is produced by the kidney (1,8,9). Approximately half of basal hepatic glucose production is derived from glycogenolysis and half from gluconeogenesis (9,10).

Following glucose ingestion, the balance between endogenous glucose production and tissue glucose uptake is disrupted. The increase in plasma glucose concentration stimulates insulin release from the pancreatic beta cells, and the resultant hyperinsulinemia and hyperglycemia serve (i) to stimulate glucose uptake by splanchnic (liver and gut) and peripheral (primarily muscle) tissues (Table 1) and (ii) to suppress endogenous glucose production (1,3-7,11-14). Hyperglycemia, in the absence of hyperinsulinemia, exerts its own independent effect to stimulate muscle glucose uptake and to suppress endogenous glucose production in a dose dependent fashion (14-16). The majority (~80-85%) of glucose that is taken up by peripheral tissues is disposed of in muscle (1,3-7,11-14), with only a small amount (~4-5%) being metabolized by adipocytes (17). Although fat tissue is responsible for only a small amount of total body glucose disposal, it plays a very important role in the maintenance of total body glucose homeostasis (see below). Insulin is a potent inhibitor of lipolysis and even small increments in the plasma insulin concentration exert a potent antilipolytic effect, leading to a marked reduction in the plasma free fatty acid level (18). The decline in plasma FFA concentration results in increased glucose uptake in muscle (19) and contributes to the inhibition of hepatic glucose production (16,20). Thus, changes in the plasma FFA concentration in response to increased plasma levels of insulin and glucose play an important role in the maintenance of normal glucose homeostasis (21,22).

Glucagon also plays a central role in the maintenance of normal glucose homeostasis (16,23). Under postabsorptive conditions approximately half of total hepatic glucose output is dependent upon the maintenance of normal basal glucagon levels and inhibition of basal glucagon secretion with somatostatin causes a profound reduction in endogenous glucose production and decline in plasma glucose concentration (16,23). Following a glucose-containing meal, alpha cell secretion of glucagon is inhibited by hyperinsulinemia, and the resultant hypoglucagonemia plays an important role in the suppression of hepatic glucose production and maintenance of normal postprandial glucose tolerance (23).

The route of glucose administration also plays an important role in the distribution of the administered glucose load and overall glucose homeostasis (1,4,5,7,11-13,16,24,25). Intravenous administration of insulin does not augment splanchnic (liver plus gut) glucose uptake, while intravenous glucose enhances splanchnic glucose uptake in direct proportion to the increase in plasma glucose concentration (3,24,25). In contrast, oral glucose administration markedly enhances splanchnic glucose uptake (1,4,16,24,27). The portal signal that stimulates hepatic glucose uptake following oral glucose is directly proportional to the negative hepatic artery-portal vein glucose concentration gradient (26). Widening of this gradient stimulates the splanchnic nerves, activating a neural reflex, that enhances vagal activity and inhibits sympathetic nerves which innervate the liver (16,26,27). These neural changes augment glucose uptake by the hepatocyte and stimulate hepatic glycogen synthase, while simultaneously inhibiting glycogen phosphorylase. In contrast to the intravenous administration of glucose/insulin, where muscle accounts for the majority (~80-85%) of glucose disposal, the splanchnic tissues are responsible for the removal of approximately 30-40% of an ingested glucose load (6,7,11-13).

Glucose administration via the gastrointestinal tract also has a potentiating effect on insulin secretion (28). Thus, the plasma insulin response following glucose ingestion is approximately twice as great as that following intravenous glucose despite equivalent increases in the plasma glucose concentration. This potentiating effect of oral glucose administration is related to the release of glucagon-like peptide-1 (GLP-1) and glucose-dependent insulinotropic polypeptide (previously called gastric inhibitory polypeptide) (GIP) from the gastrointestinal tissues (29,30).

GLUCOSE HOMEOSTASIS IN TYPE 2 DIABETES MELLITUS

Type 2 diabetic individuals are characterized by: (i) defects in insulin secretion; (ii) insulin resistance involving muscle, liver, and the adipcoyte; and (iii) abnormalities in splanchnic glucose uptake.

Insulin Secretion in Type 2 Diabetes Mellitus

Impaired insulin secretion is a uniform finding in type 2 diabetic patients and the evolution of beta cell dysfunction has been well characterized in diverse ethnic populations (1,4,31-40). Early in the natural history of type 2 diabetes, insulin resistance is well established (1,4,35,37-39,41-45) but glucose tolerance remains normal because of a compensatory increase in insulin secretion (1,4,33-35,37-39,42-44,46-48). This dynamic interaction between insulin secretion and insulin resistance has been well documented (49). Within the normal glucose tolerant population, approximately 20-25% of individuals are severely resistant to the action of insulin (49) (measured with the euglycemic insulin clamp) and subjects in the lowest quartile of insulin sensitivity are as insulin resistant as type 2 diabetic patients . However, insulin secretion (measured with the hyperglycemic clamp technique) in these insulin resistant, non-diabetic individuals, is increased in proportion to the severity of the insulin resistance and glucose tolerance remains normal (49). Thus, the pancreas in people with a normal-functioning beta cell is able to "read" the severity of insulin resistance and adjust its secretion of insulin to maintain normal glucose tolerance.

In type 2 diabetics, the fasting plasma insulin concentration invariably has been found to be normal or increased (1,4,31,33,34,39,42,48) and basal insulin secretion, measured from C-peptide kinetics, is elevated (1,3,4,52,53). DeFronzo et al (54) measured the fasting plasma insulin concentration and performed oral glucose tolerance tests in 77 normal-weight type 2 diabetic patients and over 100 lean subjects with normal or impaired glucose tolerance. The relationship between the fasting plasma glucose concentration and the fasting plasma insulin concentration resembles an inverted U or horseshoe. Because this curve closely resembles Starling's curve of the heart, it has been referred to as Starling's curve of the pancreas (4). As the fasting plasma glucose concentration rises from 80 to 140 mg/dl, the fasting plasma insulin concentration increases progressively, peaking at a value that is 2-2.5 fold greater than in normal weight, non-diabetic, age-matched controls. The progressive rise in fasting plasma insulin level can be viewed as an adaptive response of the pancreas to offset the progressive deterioration in glucose homeostasis. However, when the fasting plasma glucose concentration exceeds 140 mg/dl, the beta cell is unable to maintain its elevated rate of insulin secretion and the fasting insulin concentration declines precipitously. This decrease in fasting insulin level has important physiologic implications, since it is at this point that hepatic glucose production (the primary determinant of the fasting plasma glucose concentration) begins to rise (54).

The relationship between the plasma insulin response during an oral glucose tolerance test and the fasting plasma insulin concentration also resembles an inverted U-shaped curve (1,4) . However, the inflection point, ~120 mg/dl, at which the beta cell no longer can maintain its accelerated rate of insulin secretion is shifted to the left compared to the basal insulin secretory rate. A type 2 diabetic patient with a fasting plasma glucose concentration of 150-160 mg/dl secretes an amount of insulin that is similar to that in a healthy non-diabetic individual. However, a "normal" insulin response in the presence of this degree of hyperglycemia and underlying insulin resistance is markedly abnormal. When the fasting plasma glucose concentration exceeds 150-160 mg/dl, the plasma insulin response, when viewed in absolute terms, becomes insulinopenic. Finally, when the fasting glucose exceeds 200-220 mg/dl, the plasma insulin response to a glucose challenge is markedly blunted. Nonetheless, the postabsorptive rate of insulin secretion remains elevated and fasting hyperinsulinemia persists, even with fasting plasma glucose concentrations as high as 250-300 mg/dl (1,4,31,33-37,42,44,45,48,51,54,55). Despite severe fasting hyperglycemia, 24 hour integrated plasma insulin and C-peptide profiles in lean type 2 diabetic patients remain normal (56,57). These normal day-long values result from the combination of elevated fasting and decreased postprandial insulin and C-peptide secretory rates.

The natural history of type 2 diabetes, starting with normal glucose tolerance, insulin resistance, and compensatory hyperinsulinemia with progression to impaired glucose tolerance (IGT) and overt diabetes mellitus has been observed in a variety of populations including Caucasians, Native-Americans, Mexican Americans, and Pacific Islanders (4,33-39,44-47,58), and in the rhesus monkey (59), an animal model that closely resembles type 2 diabetes in man. In all of these population studies, as well as in the rhesus monkey, the development of type 2 diabetes is strongly associated with the presence of obesity and the close association of these two common metabolic disorders has been referred to as diabesity. In these populations at high risk to develop type 2 diabetes mellitus, the progression from normal to impaired glucose tolerance is associated with a marked increase in both fasting and glucose-stimulated plasma insulin levels (4,34,58,59) and a decrease in tissue sensitivity to insulin (1,4,35) . Even though the plasma insulin response is increased two- to threefold above that in normal-glucose-tolerant subjects, overt diabetes does not develop unless a concomitant defect in insulin secretion is present. The defect in insulin secretion can be appreciated when ß-cell function is viewed relative to the prevailing severity of insulin resistance. The progression from IGT to type 2 diabetes with mild fasting hyperglycemia (120-140 mg/dl, 6.7-7.8 mmol/l) is heralded by an inability of the ß-cell to maintain its previously high rate of insulin secretion in response to a glucose challenge (4,31,34,35,44,59) without any further or minimal deterioration in tissue sensitivity to insulin. It should be noted, however, that increased basal insulin secretion and fasting hyperinsulinemia are maintained until fasting plasma glucose levels exceed 140 mg/dl. A similar picture of insulin secretion is observed in the rhesus monkey (48). As monkeys age, they become obese and a high percentage develop typical type 2 diabetes. In monkeys, as in humans, the earliest detectable abnormality that precedes the onset of diabetes mellitus is an increase in the fasting and glucose-stimulated plasma insulin concentration and a decrease in tissue sensitivity to insulin (48). With time, this high rate of insulin secretion cannot be maintained and the downward slope of Starling's curve commences . At this point, marked fasting hyperglycemia and glucose intolerance ensue. In summary, these studies have demonstrated that hyperinsulinemia precedes the development of type 2 diabetes (4,34-37,58-60), and hyperinsulinemia is a strong predictor of the development of IGT and type 2 diabetes (33,34,37-39,41,44,58,61-64). However, these studies also demonstrate that overt type 2 diabetes (fasting glucose > 126 mg/dl) does not develop in the absence of a significant defect in ß-cell function. The nature of this ß-cell defect is considered in subsequent sections.

Type 2 Diabetes with Hypoinsulinemia

Despite an impressive body of evidence that implicates hyperinsulinemia and insulin resistance as the antecedents of type 2 diabetes, a number of studies have documented that absolute insulin deficiency, with or without impaired tissue sensitivity to insulin, can lead to the development of type 2 diabetes. This is best exemplified by patients with maturity onset diabetes of youth (MODY), which represents a familial subtype of type 2 diabetes characterized by an early age of onset, autosomal dominant inheritance with high penetrance, mild-to-moderate fasting hyperglycemia, and impaired insulin secretion (65-68).

The initial description of MODY was made by Fajans and Bell et al subsequently demonstrated that MODY1 was related to a nonsence mutation in exon 7 of the HNF4? gene (69). Froguel and colleagues (65,70) later demonstrated that MODY in French families was associated with mutations in the glucokinase gene on chromosome 7p (MODY2). To date, six specific mutations in different genes have been implicated in the MODY profile including glucokinase and five transcription factors: MODY1 = hepatic nuclear factor 4a; MODY2 = glucokinase; MODY3 = HNF-1a; MODY4 = insulin promoter factor 1; MODY5 = HNF-1ß; MODY6 = neurogenic differentiation 1/ß-cell E-box transactivator 2 (67-73). MODY individuals are characterized by impaired insulin secretion in response to glucose and other secretagogues (31,66,74). Although diminished insulin secretion represents the primary defect in MODY, peripheral tissue resistance to insulin and disturbances in hepatic glucose metabolism also have been shown to play some role in the development of impaired glucose homeostasis (75,76). Although glucokinase mutations, leading to impaired insulin secretion, are well-established features of MODY2, intensive investigations in typical older-onset type 2 diabetic individuals in a variety of ethnic populations have demonstrated that glucokinase mutations account for less than 1% of the common form of type 2 diabetes (77).

The view that insulin deficiency represents the primary defect responsible for glucose intolerance in individuals who present with typical type 2 diabetes and who do not have a glucokinase or other MODY mutations has been championed by Cerasi, Luft, Hales, and coworkers (32,55,77-79). According to these investigators, a defect in early insulin secretion leads to an excessive rise in plasma glucose concentration and the resultant hyperglycemia is responsible for late hyperinsulinemia. Hales and colleagues (55,79) have demonstrated that many lean Caucasian individuals with mild fasting hyperglycemia (<140 mg/dl, 7.8 mmol/l) are characterized by insulin deficiency at all time points during an OGTT. An impaired early insulin response also has been a characteristic finding in Japanese Americans who progress to type 2 diabetes (80). Unfortunately, none of these studies provided information about insulin sensitivity. Arner et al (81) and the EGIR study group (82) in Caucasians have demonstrated normal insulin sensitivity in a minority of type 2 diabetic individuals and Banerji and colleagues (83) have suggested that up to 50% of African American type 2 diabetic patients who reside in New York City are characterized by severely impaired insulin secretion and normal insulin sensitivity. A similar defect in insulin secretion has been described in black African type 2 diabetics living in Cameroon (84).

In summary, it is clear that impaired insulin secretion-in the absence of insulin resistance-can lead to the development of full-blown type 2 diabetes. However, it is unclear how frequently a pure ß-cell defect results in typical type 2 diabetes.

First-Phase Insulin Secretion

In humans (85) and animals (86), insulin secretion in response to intravenous glucose is biphasic, with an early burst of insulin release within the first 10 minutes followed by a gradually increasing phase of insulin secretion that persists as long as the hyperglycemic stimulus is present. This biphasic insulin response is not easily identifiable after oral glucose, because of the more gradual rise in plasma glucose concentration. The loss of first phase insulin secretion is an early abnormality in patients who are destined to develop type 2 diabetes (31,32,66,78). The early phase of insulin secretion during the OGTT (0-30 min) and during the IVGTT (0-10 min) tends to be reduced in most type 2 diabetic subjects with fasting plasma glucose levels >110-120 mg/dl (6.1-6.7 mmol/l) (1,4). During the OGTT, the defect in early insulin secretion is most readily detected if one calculates incremental plasma insulin response at 30 min divided by the incremental plasma glucose response at 30 min (?I30/?G30) (31,32,37-39,55,61,66,79,87). Although the first-phase insulin secretory response is characteristically lost in type 2 diabetes, this defect is not consistently observed until the fasting plasma glucose concentration rises to approximately 115-120 mg/dl (6.4-6.7 mmol/l) (88). Tight metabolic control partially restores the defect in first-phase insulin response (89,90), indicating that at least part of the defect is acquired, most likely secondary to glucotoxicity (91) or liptoxicity (92-94) (see subsequent discussion). The loss of the first phase of insulin secretion has important pathogenetic consequences since this early burst of insulin release primes insulin target tissues, especially the liver, that are responsible for the maintenance of normal glucose homeostasis (95,96).

Etiology of Impaired Insulin Secretion in T2DM

The progression from normal to impaired glucose tolerance (IGT) to type 2 diabetes with mild fasting hyperglycemia (<120-140 mg/dl, 6.7-7.8 mmol/l) is characterized by hyperinsulinemia . However, when the fasting glucose concentration exceeds ~120 mg/dl (6.7 mmol/l) and ~140 mg/dl (7.8 mmol/l), respectively, there is a progressive decline in fasting and glucose-stimulated plasma insulin levels. These observations suggest that the decline in ß-cell function is acquired, and this conclusion is supported by the studies of Polonsky (31,66) using more sophisticated techniques to evaluate insulin secretion. A number of pathogenic factors, both genetic and acquired, have been implicated in the progressive impairment in insulin secretion. A growing body of evidence indicates that pancreatic beta cells are in a constant state of dynamic change, with continued regeneration and apoptosis (97). Islet neogenesis occurs from ductal endothelial cells of the exocrine pancreas. In adults there is continued renewal and loss of beta cells and a number of factors have shown to disturb the delicate balance between beta cell replication and apotosis (97) (see subsequent discussion).

Investigations in first degree relatives of type 2 diabetic patients (40,98-100) and in twins (101,102) have provided strong evidence for the genetic basis of beta cell dysfunction. Impaired insulin secretion also has been shown to be an inherited trait in Finnish families with type 2 diabetes mellitus with evidence for a susceptibility locus on chromosome 12 (103,104).

Of the acquired defects leading to impaired insulin secretion, "glucotoxicity" (91) and "lipotoxicity" (22,92-94) have received the most attention . Evidence in support of the glucotoxicity hypothesis in man derives from the observation that improved glycemia, however achieved (diet, insulin therapy, sulfonylureas, metformin), leads to enhanced insulin secretion (89,90). A more rigorous test of the glucotoxicity hypothesis requires that the plasma glucose concentration be reduced without altering other circulating substrate levels and without administering an agent (i.e., such as insulin or metformin) that has direct effects on cellular glucose metabolism. This has been accomplished by utilizing phlorizin, a potent inhibitor of renal tubular glucose transport. When administered to partially pancreatectomized, chronically hyperglycemic diabetic rats, phlorizin restores normoglycemia in association with a dramatic improvement in both the first and second phases of insulin secretion (105). Weir and colleagues (106,107) have used both partially pancreatectomized and neonatal streptozotocin-induced diabetic rat models to produce a state of mild chronic hyperglycemia. When the pancreatic remnant from these diabetic rats was perfused in vitro, the insulin response to hyperglycemia was markedly impaired in diabetic compared with control animals (108). A rise in the plasma glucose concentration of as little as 15 mg/dl caused a 75% inhibition of insulin secretion by the pancreas perfused in vitro. These results provide support for the concept that a minimal elevation in mean plasma glucose concentration, in the presence of a reduced ß-cell mass, can lead to a major impairment in insulin secretion by the remaining pancreatic tissue (91,105-108). Prolonged ß-cell exposure to high glucose concentrations also has been demonstrated to impair insulin gene transcription, leading to decreased insulin synthesis and secretion (109).

Liptotoxicity also has been implicated as an acquired cause of the progressive decline in beta cell function as individuals progress from impaired glucose tolerance to overt type 2 diabetes mellitus. Short term beta cell exposure to physiologic increases in free fatty acids stimulates insulin secretion (22). Within the beta cell, long chain fatty acids are converted to their fatty acyl-CoA derivates (92-94,110), which lead to increased formation of phosphatidic acid and diacylglycerol. These lipid intermediates activate specific protein kinase C isoforms, which stimulate the exocytosis of insulin (92,110,111). Long chain fatty acyl-CoAs also directly enhance exocytosis (92), cause closure of the K+-ATPase channel (92), and stimulate Ca++-ATPase and increase intracellular calcium, thus augmenting insulin secretion (92,110,111). However, chronic exposure of beta cells to elevated free fatty acyl CoA levels inhibits insulin secretion via operation of the Randle cycle (112), in which increased beta-oxidation increases acetyl-CoA, leading to an inhibition of pyruvate dehydrogenase and elevated citrate levels, which inhibit phosphofructokinase and subsequently glycolysis (92-94,110,112). Since glucose metabolism via the glycolytic pathway with subsequent oxidation of pyruvate in the Krebs cycle to generate ATP is an essential requisite for glucose-stimulated insulin secretion, chronic exposure to elevated plasma FFA levels leads to an inhibition of insulin secretion. Increased fatty acyl CoA levels within the beta cell also stimulate ceramide synthesis, which augments inducible nitric oxide synthase (94,113). The resultant increase in nitric oxide increases the expression of inflammatory cytokines, including interleukin-1 and tumor necrosis factor alpha, which impair beta cell function and cause apoptosis of beta cells.

"Incretin" deficiency and/or "incretin" resistance also has been implicated in the pathogenesis of beta cell dysfunction in type 2 diabetic patients (29,30,114,115). Glucose administered via the gastrointestinal route causes a much greater stimulation of insulin secretion than a comparable glucose challenge given intravenously (28). This observation prompted a search for the responsible "incretins" or gut-derived hormones that enhance glucose-stimulated insulin secretion. Two gastrointestinal hormones, glucose-dependent insulinotropic polypeptide (GIP) and glucagon-like peptide-1 (GLP-1), are responsible for more than 90% of the incretin effect observed following glucose or mixed meal ingestion (29,30,114-116). Both GIP and GLP-1 are released from endocrine cells of the duodenum and jejunum in response to intraluminal carbohydrate but not in response to circulating glucose, and their release is proportional to the oral glucose load and dependent upon absorption of glucose across the intestinal mucosa (30,114,115). The effectiveness of both GIP and GLP-1 to stimulate insulin secretion is proportional to the increase in plasma glucose concentration (29,30,114,115). Antibodies that neutralize GIP and GLP-1 (117) and GLP-1 antagonists (118) impair glucose tolerance in a variety of animal species, including primates. Although the amount of GLP-1 released is considerably less than those of GIP, GLP-1 is such a potent glucose-sensitive potentiator of insulin secretion that it is thought to be the major incretin (30,114,115,119). In type 2 diabetic humans the GIP response to glucose ingestion consistently has been shown to be normal, indicating the presence of beta cell resistance to the incretin-effect of GIP (30,119). In contrast, the GLP-1 response to oral glucose has been shown to be reduced (30,114,115,120). Intravenous administration of GLP-1 in type 2 diabetic patients markedly enhances the postprandial insulin secretory response in type 2 diabetic patients (119,121) and chronic continuous GLP-1 administration restores near-normal glycemia in type 2 diabetic patients (122). GLP-1 also has been shown to augment islet regeneration in rodents (123).

Amylin (islet amyloid polypeptide, IAPP) has been implicated in the progressive beta cell failure in type 2 diabetes mellitus (124). IAPP is produced by the ß-cell, packaged with insulin in secretory granules, and cosecreted into the sinusoidal space (125). This peptide has been shown to be the precursor for the amyloid deposits frequently observed in patients with type 2 diabetes (125). At very high doses, amylin inhibits insulin secretion by the perfused rat pancreas in vitro (126), and pancreatic amylin depots have been shown to precede the appearance of glucose intolerance in spontaneously diabetic monkeys (127). Elevated plasma islet amyloid polypeptide levels have been demonstrated in type 2 diabetic subjects (125), in obese glucose-intolerant subjects (128), in glucose-intolerant first-degree relatives of type 2 diabetic patients (124), and in animal models of diabetes (125). Using immunohistochemical and in situ hybridization techniques, islet amyloid polypeptide mRNA has been demonstrated within the islets of Langerhans (130). Following secretion, amylin accumulates extracellularly in close contact with the ß-cell, and it has been suggested that these amylin deposits cause ß-cell dysfunction and eventually death by impairing the transport of nutrients from the plasma to the beta cell or by interfering with the glucose-sensing and/or insulin-secreting apparatus of the ß-cell (125-127). Although attractive, this theory has been seriously challenged by Bloom and coworkers, who failed to find any inhibitory effect of amylin on insulin secretion when the peptide was infused in pharmacological doses in rats, rabbits, and humans (131,132). Studies in transgenic mice, with the gene encoding either human or rat islet amyloid polypeptide under control of an insulin promoter, also mitigate against an important role of IAPP in the development of beta cell dysfunction (133). Although pancreatic and plasma amyloid polypeptide levels were significantly elevated in these transgenic mice, hyperglycemia and hyperinsulinemia did not develop. A recent provocative review suggests that islet amyloid polypeptide in the islets of Langerhans may serve a protective role under conditions of increased insulin secretion (134). In summary, definitive evidence that amylin is responsible for the defect in ß-cell dysfunction in type 2 diabetes in humans remains elusive, although recent evidence suggests that the combination of elevated plasma FFA levels and hypersecretion of amylin may interact synergistically to impair insulin secretion and cause injury to the pancreatic beta cells (124).

The number of ß-cells is a critical determinant of the amount of insulin that is secreted by the pancreas. Most (135-139), but not all (140,141) studies have demonstrated a modest reduction (20-40%) in ß-cell mass in patients with long-standing type 2 diabetes. By routine histology, the islets of Langehans appear normal morphologically, with the exception of beta cell degranulation. Insulitis is not observed. The factors responsible for the decrease in ß-cell mass in type 2 diabetics remain to be identified. A recent study suggests that new islet formation from exocrine ducts is markedly reduced in type 2 diabetic individuals (142). Obesity, another insulin-resistant state, is characterized by a significant increase in ß-cell mass (137). Thus, the finding of even a modest reduction (by 20-40%) in ß-cell mass in any insulin-resistant state is most impressive. Although recent studies with well matched (for age, gender and obesity) controls suggest that beta mass is reduced even during the early stages of the development of type 2 diabetes (139), it seems likely that factors in addition to ß-cell loss must be responsible for the impairment in insulin secretion.

An association between low birth weight and the development of IGT and type 2 diabetes later has been demonstrated in a number of populations (143,144) Developmental studies in animals and humans have demonstrated that poor nutrition and impaired growth are associated with reduced insulin secretion and/or reduced ß-cell mass (145). Fetal malnutrition also can lead to the development of insulin resistance later in life (146). One could postulate that an environmental influence, e.g., impaired fetal nutrition leading to an acquired defect in insulin secretion or reduced ß-cell mass, when superimposed on an inherited defect in insulin action, could eventuate in type 2 diabetes later in life. Thus, with the normal aging process, the onset of obesity, or a worsening of the genetic component of the insulin resistance, the ß-cell would be called on to augment its secretion of insulin to offset the defect in insulin action. If ß-cell mass or function were reduced by an environmental insult during fetal life, the result would be the emergence of IGT or overt type 2 diabetes. Note that, although such a defect may place a limit on the maximum amount of insulin that is secreted, it would not explain the progressive decline in the amount of insulin that is secreted in response to physiological stimuli as individuals progress from IGT to mild-moderately severe type 2 diabetes .

Insulin Resistance and Type 2 Diabetes

Both longitudinal and cross-sectional studies have documented conclusively that hyperinsulinemia precedes the development of type 2 diabetes in most ethnic populations with a high incidence of type 2 diabetes (33-39,41,43,44,58-63,98,147,148). Studies employing the euglycemic insulin clamp, minimal model, and insulin suppression techniques have provided direct quantitative evidence that the progression from normal to impaired glucose tolerance is associated with the development of severe insulin resistance (1,4,34,37-39,41,43-45,58-60,149,150), whereas plasma insulin concentrations, both in the fasting state and in response to a glucose load are increased (see above discussion about insulin secretion).

Himsworth and Kerr (151), using a combined oral glucose and intravenous insulin tolerance test, were the first to demonstrate that tissue sensitivity to insulin was diminished in diabetic patients. In 1975 Reaven and colleagues (152), using the insulin suppression test, found that the ability of insulin to promote tissue glucose uptake in type 2 diabetes was severely reduced. A defect in insulin action in type 2 diabetes also has been demonstrated with limb infusion of insulin using the forearm and leg catheterization techniques (3-5,153), as well as with radioisotope turnover studies (154), the modified intravenous glucose tolerance test (155), and the minimal model technique (156).

DeFronzo et al, using the more physiologic euglycemic insulin clamp technique (85), have provided the most conclusive documentation that insulin resistance is a characteristic feature of lean type 2 diabetic individuals (1,3,4,18,158,159) . Because diabetic patients with severe fasting hyperglycemia (>180-200 mg/dl, 10.0-11.1 mmol/l) are known to be insulinopenic , and insulin deficiency is associated with the emergence of a number of intracellular defects in insulin action (160,161), only diabetics with mild to modest elevations in the fasting plasma glucose concentration (mean = 150±8 mg/dl, 8.3±0.4 mmol/L) were included in these studies in order to focus on the early stages of the evolution of type 2 diabetes in normal-weight subjects. Insulin-mediated whole-body glucose disposal was reduced by ~40-50% in type 2 diabetic subjects, conclusively demonstrating the presence of moderate to severe insulin resistance. Three additional points are noteworthy: (i) in lean type 2 diabetics with more severe fasting hyperglycemia (198±10 mg/dl), the severity of insulin resistance is only slightly (10-20%) greater than that in diabetic patients with mild fasting hyperglycemia (1,4,162,163); (ii) the defect in insulin action is observed at all plasma insulin concentrations, spanning the physiological and pharmacological range (1,4,18,164,165) ; (iii) in diabetic patients with overt fasting hyperglycemia, even maximally stimulating plasma insulin concentrations are not capable of eliciting a normal glucose metabolic response under euglycemic conditions (1,4,18,164,165). With a few exceptions (81,83), the great majority of investigators have demonstrated that lean type 2 diabetic subjects are resistant to the action of insulin (1,3,4,18,37-39,41,42,47,51,59,60,98,150,152,154,156-159,164-166). The ability of glucose, i.e. hyperglycemia, to stimulate its own uptake also is impaired in type 2 diabetics (15,157,168).