Marilyn Ader

Why visceral fat is bad : Mechanisms of the metabolic syndrome

A consensus has emerged that fat stored in the central segment of the body is particularly damaging in that it portends greater risk for diabetes, cardiovascular disease, hypertension, and certain cancers (1–3). It is also accepted that insulin resistance is a related characteristic that may be an essential link between central fat and disease risk. Additionally, it is possible that the hyperinsulinemia that accompanies insulin resistance in non-diabetic but at-risk individuals may magnify, or even mediate, some of the detrimental effects of visceral adiposity (4–6). However, there is less information regarding the mechanisms that may link visceral fat with risk for disease. For example, there is controversy regarding the specific mechanisms by which fat in the visceral compartment confers greater risk than subcutaneous fat. Many investigators have suggested that one or more moieties secreted by the visceral adipocyte might mediate insulin resistance. Among the socalled “bad actors” are free fatty acids (FFAs) themselves (“portal theory”) (7–9) or the adipose tissue–released cytokines (adipokines) such as interleukin-1, interleukin-6, tumor necrosis factor, resistin, or a reduction in adiponectin, which has been repeatedly shown to be associated with reduced insulin resistance (10–13). Of course, insulin itself could be involved, as other adipose-secreted protein compounds not yet identified. But why visceral fat? Is it because of the unique anatomical position of the visceral fat depot, with effluent entering the liver, or is it because of molecular characteristics of visceral fat itself, which may favor release of damaging molecules into the systemic circulation? These questions remain unanswered. However, in our laboratory, we have developed the obese dog model, which has led to some understanding of the pathogenesis of the metabolic syndrome. The dog model has not been widely used for the study of the metabolic syndrome, but we have found it to have several important characteristics that we have been able to exploit: the ability to make longitudinal measurements and the ability to access the portal vein. In that sense the dog is a unique model, in that these latter measurements are daunting in rodents, and carrying out repetitive, invasive clinical measurements in non-human primates is challenging. Also, the dog with visceral obesity has turned out to be a reasonable model for a similar syndrome in humans (Figure 1). In fact, the dog is genetically more similar to humans than is the rodent. Here we summarize a significant amount of evidence in which we examined what we considered to be the simplest hypothesis composed of two postulates: 1) that FFAs per se are among the most important products of the visceral adipocyte to cause insulin resistance (and hence the metabolic syndrome) and 2) that the anatomical position of the visceral adipose depot (i.e., portal drainage into the liver) plays an important role in the pathogenesis of the metabolic syndrome. While we cannot say that these postulates are proven, there are data that support them, and Occam’s razor instructs us to accept them until proven untrue. Whether true or not, it appears that examining them has led us to a deeper understanding of the physiological basis for the metabolic syndrome itself. One similarity between dogs and humans is the wide variance in fat deposition in a “wild” or “natural” population. We measure distribution of fat about the truncal region using magnetic resonance imaging [Figure 2; 11 axial slices: 1-cm landmark slice at the umbilicus (left renal artery) 5 cm]. Similar to human subjects (14,15), there is surprising variability in distribution. Some animals are strikingly lean, with total fat varying over a factor of 5, from 10 to 50 cm/cm non-fat tissue. Interestingly, there is a tendency for visceral adiposity to increase rapidly as one examines animals with increasing body fat; the visceral fat depot tends to plateau, and subcutaneous fat increases more rapidly with overall obesity. This tendency for visceral fat to Department of Physiology and Biophysics, Keck School of Medicine, University of Southern California, Los Angeles, California. Address correspondence to Richard N. Bergman, Department of Physiology and Biophysics, MMR 630, 1333 San Pablo Street, Los Angeles CA 90033. E-mail: rbergman@usc.edu Copyright

Insulin transport across capillaries is rate limiting for insulin action in dogs

This study examined the relationship between transcapillary insulin transport and insulin action in vivo. During euglycemic clamps (n = 7) in normal conscious dogs we simultaneously measured plasma and thoracic duct lymph insulin and glucose utilization (Rd). Clamps consisted of an activation phase with constant insulin infusion (0.6 mU/kg per min) and a deactivation phase. [14C]Inulin was infused as a passively transported control substance. While [14C]inulin reached an equilibrium between plasma and lymph, steady-state (ss) plasma insulin was higher than lymph (P less than 0.05) and the ratio of 3:2 was maintained during basal, activation, and deactivation phases: 18 +/- 2 vs. 12 +/- 1, 51 +/- 2 vs. 32 +/- 1, and 18 +/- 3 vs. 13 +/- 1 microU/ml. In addition, it took longer for lymph insulin to reach ss than plasma insulin during activation and deactivation: 11 +/- 2 vs. 31 +/- 5 and 8 +/- 2 vs. 32 +/- 6 min (P less than 0.02). Rd increased from 2.6 +/- 0.1 to a ss of 6.6 +/- 0.4 mg/kg per min within 50 +/- 8 min. There was a remarkable similarity in the dynamics of insulin in lymph and Rd: the time to reach ss for Rd was not different from lymph insulin (P greater than 0.1), and the relative increases of the two measurements were similar, 164 +/- 45% and 189 +/- 29% (P greater than 0.05). While there was only a modest correlation (r = 0.78, P less than 0.01) between Rd and plasma insulin, the dynamic changes of lymph insulin and Rd showed a strong correlation (r = 0.95, P less than 0.01). The intimate relationship between lymph insulin and Rd suggests that the transcapillary insulin transport is primarily responsible for the delay in Rd. Thus, transcapillary transport may be rate limiting for insulin action, and if altered, it could be an important component of insulin resistance in obesity and diabetes mellitus.

Importance of Glucose Per Se to Intravenous Glucose Tolerance: Comparison of the Minimal-Model Prediction with Direct Measurements

Glucose disappearance after an oral or intravenous challenge is a function of the effects of both endogenously secreted insulin and of glucose itself. We previously introduced the term “glucose effectiveness,” or SG, defined as the ability of glucose per se to enhance its own disappearance independent of an increment in plasma insulin. The present investigation, performed in conscious dogs, was undertaken to quantify this glucose effect by minimal-model-based analysis of insulin and glucose dynamics after a frequently sampled intravenous glucose tolerance test (FSIGT). The values from the standard FSIGT were then compared with direct measurementsobtained from experiments in which the dynamic insulin response to glucose was suppressed with somatostatin (SRIF). In addition, we examined SG values from the modified FSIGT protocol, which involves both glucose and tolbutamide injections. Protocol I (N = 9): FSIGTs were performed and the glucose and insulin data were analyzed by computer. KG was 2.65 ± 0.28 min−1, S1 was 4.09 ± 0.34 × 10−4 min−1/(μU/ml), and SG was 0.033 ± 0.004 min−1. Protocol II (N = 6): FSIGTs were performed on animals in which SRIF was infused (0.8 μg/min−1kg) to obliterate the dynamic insulin response to glucose injection. Before the FSIGT, insulin and glucagon were infused intraportally to reattain basal glycemia. Without dynamic insulin, KG was reduced to 0.96 ± 0.18 min−1 (P < 0.0001). However, SG, estimated from the exponential rate of fall of plasma glucose in the absence of dynamic insulin, was similar to the standard FSIGTs: 0.025 ± 0.004 (P > 0.25). Protocol III (N = 6): modified FSIGTs wereperformed using glucose and tolbutamide injections for a better estimate of model parameters. Model parameters S1 and SG, and the KGwere not different from standard FSIGTs (P > 0.3). In fact, the value of SG (0.028 ± 0.003 min−1) was nearly identical to the direct measure from protocol II. Therefore, the effect of glucose per se on glucose decline, estimated by modeling the standard and modified FSIGTs, was confirmed by a direct measurement with the endogenous insulin response suppressed with SRIF. Also, the time course of the insulin effect to enhance net glucose disappearance from plasma [leff(t)] was calculated from the data of protocol II, and was the same as the time coursepredicted by the model. These studies demonstrate the ability of the computer modeling approach to separate insulin-dependent and glucose-dependent glucose disappearance, and represent a direct confirmation of the minimal model. More important, they confirm that glucose per se, independent of the dynamic insulin response, is a very significant factor in the determination of glucose tolerance.

Diurnal Variation in Glucose Tolerance: Cyclic Suppression of Insulin Action and Insulin Secretion in Normal-Weight, But Not Obese, Subjects

The relative roles of insulin sensitivity, insulin secretion, and glucose effectiveness to the diurnal rhythm of glucose tolerance were examined in normal-weight (n = 12) and obese (n = 11) subjects. Two frequently sampled intravenous glucose tolerance tests were performed in each subject at 0800 on one occasion and 1800 on a separate day. Tests were preceded by identical fasts of 10–12 h. In nonobese subjects, glucose tolerance, expressed as the 10- to 16-min KG value (KGS), was much reduced in the evening (AM 2.98 ± 0.45, PM 1.86 ± 0.33 min−1 P < 0.002). In the obese subjects, tolerance was lower in the morning than normal-weight subjects (2.19 ± 0.31 min−1), but unlike in nonobese subjects, tolerance was not significantly reduced during the day (1.90 ± 0.18 min−1 P > 0.40). The reduction in glucose tolerance in the normal-weight subjects was caused by diminished insulin sensitivity (parameter S1, AM 15.4 ± 2.9, PM 10.2 ± 1.9 × 10−5 min−1/pM, P < 0.01) and reduced β-cell responsivity to glucose. The evening decrease in the latter was reflected both in first-phase plasma insulin (AM 2466 ± 441, PM 1825 ± 381 pM/10 min, P < 0.05) and the potentiation slope (AM 462 ± 68, PM 267 ± 35 pM/mM, P < 0.01). In contrast, consistent with no diurnal variation in glucose tolerance, obese subjects exhibited no decline in insulin sensitivity in the evening (AM 3.6 ± 0.7, PM 4.9 ± 1.0 × 10−5 min−1/pM). A marginally significant decline in (β-cell responsiveness was observed in the obese group (potentiation slope AM 388 ± 78, PM 233 ± 58 pM/mM, P < 0.05). Glucose tolerance is maintained when the pancreas adequately compensates for prevailing insulin resistance, a relationship described by the disposition index (SI, × Φ1). This product decreased 46% in nonobese subjects from morning to evening (P < 0.05), reflecting the inability of the pancreas to overcome prevailing insulin resistance, and resulting in evening glucose intolerance. No diurnal decline in the disposition index was observed in obese subjects. The diurnal rhythms in insulin sensitivity and secretion in nonobese subjects are not consistent with known rhythms in growth hormone or cortisol, but could be due to the effects of a putative diurnal rhythm of sympathetic activity, which would suppress both insulin sensitivity and secretion in the evening. The absence of a rhythm in the obese could reflect an autonomic or hormonal dysfunction that may be related to body-weight regulation.

Transendothelial insulin transport is not saturable in vivo. No evidence for a receptor-mediated process.

In vitro, insulin transport across endothelial cells has been reported to be saturable, suggesting that the transport process is receptor mediated. In the present study, the transport of insulin across capillary endothelial cells was investigated in vivo. Euglycemic glucose clamps were performed in anesthetized dogs (n = 16) in which insulin was infused to achieve concentrations in the physiological range (1.0 mU/kg per min + 5 mU/kg priming bolus; n = 8) or pharmacologic range (18 mU/kg per min + 325 mU/kg priming bolus; n = 8). Insulin concentrations were measured in plasma and hindlimb lymph derived from interstitial fluid (ISF) surrounding muscle. Basal plasma insulin concentrations were twice the basal ISF insulin concentrations and were not different between the physiologic and pharmacologic infusion groups (plasma/ISF ratio 2.05 +/- 0.22 vs 2.05 +/- 0.23; p = 0.0003). The plasma/ISF gradient was, however, significantly reduced at steady-state pharmacologic insulin concentrations (1.37 +/- 0.25 vs 1.98 +/- 0.21; P = 0.0003). The reduced gradient is opposite to that expected if transendothelial insulin transport were saturable. Insulin transport into muscle ISF tended to increase with pharmacologic compared with physiologic changes in insulin concentration (41% increase; 1.37 +/- 0.18 10(-2) to 1.93 +/- 0.24 10(-2) min-1; P = 0.088), while at the same time insulin clearance out of the muscle ISF compartment was unaltered (2.53 +/- 0.26 10(-2) vs 2.34 +/- 0.28 10(-2) min-1; P = 0.62). Thus, the reduced plasma/ISF gradient at pharmacologic insulin was due to enhanced transendothelial insulin transport rather than changes in ISF insulin clearance. We conclude that insulin transport is not saturable in vivo and thus not receptor mediated. The increase in transport efficiency with saturating insulin is likely due to an increase in diffusionary capacity resulting from capillary dilation or recruitment.

Angiotensin II induces insulin resistance independent of changes in interstitial insulin

We set out to examine whether angiotensin-driven hypertension can alter insulin action and whether these changes are reflected as changes in interstitial insulin (the signal to which insulin-sensitive cells respond to increase glucose uptake). To this end, we measured hemodynamic parameters, glucose turnover, and insulin dynamics in both plasma and interstitial fluid (lymph) during hyperinsulinemic euglycemic clamps in anesthetized dogs, with or without simultaneous infusions of angiotensin II (ANG II). Hyperinsulinemia per se failed to alter mean arterial pressure, heart rate, or femoral blood flow. ANG II infusion resulted in increased mean arterial pressure (68 ± 16 to 94 ± 14 mmHg, P < 0.001) with a compensatory decrease in heart rate (110 ± 7 vs. 86 ± 4 mmHg, P < 0.05). Peripheral resistance was significantly increased by ANG II from 0.434 to 0.507 mmHg ⋅ ml-1 ⋅ min ( P < 0.05). ANG II infusion increased femoral artery blood flow (176 ± 4 to 187 ± 5 ml/min, P < 0.05) and resulted in additional increases in both plasma and lymph insulin (93 ± 20 to 122 ± 13 μU/ml and 30 ± 4 to 45 ± 8 μU/ml, P < 0.05). However, glucose uptake was not significantly altered and actually had a tendency to be lower (5.9 ± 1.2 vs. 5.4 ± 0.7 mg ⋅ kg-1 ⋅ min-1, P > 0.10). Mimicking of the ANG II-induced hyperinsulinemia resulted in an additional increase in glucose uptake. These data imply that ANG II induces insulin resistance by an effect independent of a reduction in interstitial insulin.We set out to examine whether angiotensin-driven hypertension can alter insulin action and whether these changes are reflected as changes in interstitial insulin (the signal to which insulin-sensitive cells respond to increase glucose uptake). To this end, we measured hemodynamic parameters, glucose turnover, and insulin dynamics in both plasma and interstitial fluid (lymph) during hyperinsulinemic euglycemic clamps in anesthetized dogs, with or without simultaneous infusions of angiotensin II (ANG II). Hyperinsulinemia per se failed to alter mean arterial pressure, heart rate, or femoral blood flow. ANG II infusion resulted in increased mean arterial pressure (68 +/- 16 to 94 +/- 14 mmHg, P < 0. 001) with a compensatory decrease in heart rate (110 +/- 7 vs. 86 +/- 4 mmHg, P < 0.05). Peripheral resistance was significantly increased by ANG II from 0.434 to 0.507 mmHg. ml(-1). min (P < 0.05). ANG II infusion increased femoral artery blood flow (176 +/- 4 to 187 +/- 5 ml/min, P < 0.05) and resulted in additional increases in both plasma and lymph insulin (93 +/- 20 to 122 +/- 13 microU/ml and 30 +/- 4 to 45 +/- 8 microU/ml, P < 0.05). However, glucose uptake was not significantly altered and actually had a tendency to be lower (5.9 +/- 1.2 vs. 5.4 +/- 0.7 mg. kg(-1). min(-1), P > 0.10). Mimicking of the ANG II-induced hyperinsulinemia resulted in an additional increase in glucose uptake. These data imply that ANG II induces insulin resistance by an effect independent of a reduction in interstitial insulin.

Mechanism of protracted metabolic effects of fatty acid acylated insulin, NN304, in dogs: retention of NN304 by albumin

Aims/hypothesis. The provision of stable, reproducible basal insulin is crucial to diabetes management. This study in dogs examined the metabolic effects and interstitial fluid (ISF) profiles of fatty acid acylated insulin, LysB29-tetradecanoyl, des-(B30) human insulin (NN304). Methods. Euglycaemic clamps were carried out under inhalant anaesthesia during equimolar intravenous infusions (3.6 pmol · min–1· kg–1 for 480 min) of human insulin or NN304 (n = 8 per group). Results. Steady-state total NN304 (albumin-bound and unbound) was considerably higher in plasma compared with human insulin (1895 ± 127 vs 181 ± 10 pmol/l, p < 0.001) and increased in interstitial fluid (163 ± 14 vs 106 ± 9 pmol/l, p < 0.01). The halftime for appearance of NN304 in interstitial fluid was slower than human insulin (92 vs 29 min, p < 0.001). Yet, equivalency of action was shown for glucose turnover; steady-state glucose uptake (Rd) of 7.28 ± 0.55 and 6.76 ± 0.24 mg · min–1· kg–1 and endogenous glucose production of 0.11 ± 0.12 and 0.22 ± 0.03 mg · min–1· kg–1 (p > 0.40; NN304 and human insulin, respectively). Similar to interstitial fluid, half times for Rd and endogenous glucose production were delayed during NN304 infusion (162 vs 46 min and 80 vs 31 min, respectively; p < 0.01 vs human insulin). Conclusion/interpretation. Firstly equivalency of steady-state action is found at equimolar physiologic infusions of human insulin and NN304. Secondly NN304 binding to plasma albumin results in slower NN304 appearance in the interstitial compartment compared with human insulin. Thirdly the delay in appearance of NN304 in interstitial fluid may not in itself be a source of the protracted action of this insulin analogue. The protracted effect is due primarily to albumin binding of the insulin analogue NN304. [Diabetologia (1999) 42: 1254–1263]

Dynamics of Glucose Production and Uptake Are More Closely Related to Insulin in Hindlimb Lymph Than in Thoracic Duct Lymph

We previously reported a striking similarity between the dynamics of both glucose turnover and thoracic duct lymph insulin during euglycemic clamps (J Clin Invest 84:1620, 1989), which suggested that transendothelial insulin transport (TET) is rate-limiting for insulin action in vivo. Thoracic duct lymph, however, is primarily derived from insulin-insensitive tissues, which raises questions as to the physiological significance of this relationship. The relationship between glucose turnover and TET was thus examined in insulin-sensitive tissues by the simultaneous measurement of insulin in plasma, thoracic duct lymph, and hindlimb lymph during euglycemic clamps in normal anesthetized dogs (n = 8). Clamps consisted of two 3-h phases: a 0.6 mU · min−1 · kg−1 insulin infusion (activation phase) followed by termination of the insulin infusion (deactivation phase). Lymph insulin was < plasma insulin during both phases (P < 0.01) with steady-state hindlimb (120 ± 12 pM) and thoracic duct lymph insulin (138 ± 12 pM) 38 and 45%, respectively, lower than steady-state plasma insulin (222 ± 24 pM) at the end of the activation phase (P < 0.05). Also, the rate of increase of lymph insulin was slower than plasma insulin during hormone infusion; half-time to steady-state was 8.8 ± 2.0 min for plasma insulin, but longer for thoracic (25.8 ± 3.5) and hindlimb lymph insulin (40.7 ± 5.7 min). A very close relationship was observed during activation between the rate of increase of glucose uptake (Rd) and the increase in hindlimb lymph insulin (r2 = 0.92); this relationship was weaker for thoracic lymph (r2 = 0.74) and much weaker between glucose uptake and plasma insulin (r2 = 0.35). These data support the concept that interstitial insulin (represented by hindlimb lymph) is the signal that determines glucose uptake by insulin-sensitive tissues and that the rate of increase of glucose uptake is determined by transendothelial insulin transport into insulin-sensitive tissue. Also, during activation, hindlimb lymph insulin was a very strong predictor of the rate of suppression of hepatic glucose output (HGO) (r2 = 0.96), and the correlation with HGO was stronger than that for thoracic lymph (r2 = 0.85). The evidence that the rate of increase of Rd and the rate of suppression of HGO during insulin infusion are very strongly predicted by the time course of insulin in hindlimb lymph is consistent with the single-gateway hypothesis: the insulin transport rate across endothelium in insulin-sensitive tissue (skeletal muscle) determines the rate of glucose utilization and the suppression of hepatic glucose output. It is suggested that there is a yet-undefined signal that controls HGO generated at the level of insulin-sensitive peripheral tissues.

Dose-response relationship between lymph insulin and glucose uptake reveals enhanced insulin sensitivity of peripheral tissues

To examine the role of transcapillary insulin transport to peripheral insulin sensitivity in vivo, we performed dose-response experiments in which both plasma and thoracic duct lymph insulin and glucose utilization (Rd) were measured in conscious dogs. Euglycemic clamps (n = 22) consisted of a 3-h activation period in which insulin was infused (rates: “physiological” 3.6, 5.4, 7.2 pmol · min−1 · kg−1; “pharmacological” 108 pmol · min−1 · kg−1), followed by a 3-h deactivation period. [14C]inulin was also infused as a diffusionary marker. Insulin sensitivity was estimated as the ED50. When based on plasma insulin, ED50 was 480 pM. However, when calculated from lymph (i.e., interstitial) insulin measurements, ED50 was 240 pM. Thus, interstitial insulin measurements reveal that insulin sensitivity of peripheral tissues is approximately twice that estimated from plasma insulin and is similar to sensitivity reported for suppression of hepatic glucose production. Furthermore, although [14C]inulin achieved equilibrium between plasma and lymph within 180 min, within the physiological range, steady state plasma insulin was higher than insulin in lymph (306 ± 18, 474 ± 42, and 780 ± 60 pM vs. 180 ± 18, 318 ± 12, and 504 ± 36 pM; P < 0.0001); plasma insulin achieved steady state faster than lymph insulin (6 ± 1, 6 ± 2, and 11 ± 3 min vs. 29 ± 4, 16 ± 6, and 44 ± 8 min; P < 0.01) and disappeared faster (5 ± 2, 7 ± 2, and 15 ± 6 min vs. 37 ± 8, 32 ± 4, and 43 ± 9 min; P < 0.01). The time course of lymph insulin at each dose was similar to that of Rd, and at each dose, unlike plasma insulin, lymph insulin was strongly correlated with Rd (r = 0.93 or better). At pharmacological hyperinsulinemia (plasma 35232 ± 5250 pM, lymph 27366 ± 4380 pM), Rd rose faster than lymph insulin and disappeared more slowly than insulin. Thus, lymph insulin data indicate that the periphery is more sensitive to insulin than previously realized from estimates based solely on plasma hormone. Furthermore, lymph insulin is proportional to Rd within the physiological but not pharmacological range of insulin, indicating that transcapillary insulin transport is rate limiting for insulin action in this range. Finally, based on in vivo lymph (i.e., interstitial) insulin measurements, peripheral tissue is almost twice as sensitive to insulin than previously realized.

6 Insulin sensitivity in the intact organism

Summary Insulin resistance is a classic characteristic of type II diabetes. Precise quantification of insulin sensitivity in vivo is essential for elucidation of the pathogenesis of observed glucose intolerance. The euglycaemic glucose clamp of Andres and colleagues assesses insulin action by disrupting the negative feedback relationship of glucose and insulin. Insulin is infused, but euglycaemia is maintained by exogenous glucose infusion (GINF). At steady state, GINF ( M ) represents total insulin action, i.e. suppression of hepatic glucose output (HGO) plus acceleration of peripheral uptake ( R d ). Coupled with the isotope dilution techniques of Steele et al, these hepatic and peripheral effects can be partitioned, although this approach may underestimate HGO and R d . Modifications of the Steele method may be required for more reliable quantification of glucose turnover. The dose-response relationship between insulin and R d also provides a measure of sensitivity from the glucose clamp: the ED 50 (and R dmax ), analogous to the K m and V max of Michaelis—Menten analysis. However, extensive labour requirements, cost and difficulties in ED 50 estimation limit its widespread use. Glucose clearance ( R d / G ) has also been used to assess insulin sensitivity in subjects of differing glycaemia, but because it fails to rise in proportion to glucose it is an inappropriate measure. Hence, we have introduced the insulin sensitivity index S IP(clamp) , defined as the action of insulin to increase glucose clearance (Δ R d /( G Δ I )), which is independent of prevailing glycaemia and insulinaemia. Lastly, we proposed the minimal model method, which determines insulin sensitivity ( S I ) from analysis of the simple intravenous glucose tolerance test (IVGTT). By adding tolbutamide injection after glucose, the resultant model-based S I is equivalent to S IP(clamp) . Furthermore, minimal model analysis will also yield S G , the parameter of insulin-independent glucose disappearance. We conclude that assessment of insulin sensitivity, whilst important, must be considered in the context of its relation to pancreatic function and insulin-dependent glucose disappearance. Only a clear understanding of the complex interrelation of these factors will lead to elucidation of the mechanisms underlying glucose intolerance.