Despite the well-described association between obesity and insulin resistance, the physiologic mechanisms that link these two states are poorly understood. The present study was performed to elucidate the role of visceral adipose tissue in whole-body glucose homeostasis. Dogs made abdominally obese with a moderately elevated fat diet had catheters placed into the superior mesenteric artery so that the visceral adipose bed could be insulinized discretely. Omental insulin infusion was extracted at ∼27%, such that systemic insulin levels were lower than in control (portal vein) insulin infusions. Omental infusion did not lower systemic free fatty acid levels further than control infusion, likely because of the resistance of the omental adipose tissue to insulin suppression and the confounding lower systemic insulin levels. The arteriovenous difference technique showed that local infusion of insulin did suppress omental lipolysis, but only at extremely high insulin concentrations. The median effective dose for suppression of lipolysis was almost fourfold higher in the visceral adipose bed than for whole-body suppression of lipolysis. Thus, the omental adipose bed represents a highly insulin-resistant depot that drains directly into the portal vein. Increased free fatty acid flux to the liver may account for hepatic insulin resistance in the moderately obese state.
Insulin resistance is an important risk factor for type 2 diabetes and plays a role in many other diseases, including polycystic ovarian syndrome, cancer, and atherosclerosis. Several studies have reported an association between abdominal obesity and insulin resistance (1,2), but the physiological causality underlying this linkage remains unclear. We and others have demonstrated a strong correlation between free fatty acid (FFA) levels and hepatic glucose output (3,4) and have hypothesized that omental fat contributes to liver insulin resistance by increasing portal delivery of FFAs (5–7). This could lead to inappropriately elevated hepatic glucose production, hyperinsulinemia, and eventual β-cell failure.
Omental adipose tissue has been shown in vitro to be relatively resistant to insulin suppression of lipolysis (8), and therefore would be expected to contribute disproportionately to hepatic FFA levels, especially in the hyperinsulinemic state. However, insulin sensitivity of visceral adipose tissue assessed in vivo is less well defined. In this regard, Meek et al. (9) recently measured splanchnic FFA release in vivo and reported that overall splanchnic lipolysis is relatively resistant to insulin suppression. However, the explicit in vivo sensitivity of visceral adipocytes remains to be directly measured.
The present study was performed to characterize the dynamic regulation of omental adipocyte lipolysis in vivo under hyperinsulinemic conditions. Male dogs were made abdominally obese with a modest daily supplement of 2 g/kg body wt of cooked bacon grease for 4 weeks. Euglycemic clamp experiments were performed in the conscious state, and insulin was infused either directly into the omentum via the superior mesenteric artery, causing relative omental hyperinsulinemia, or into the portal vein on a separate occasion (control). Portal vein sampling was used to directly measure the net omental lipolysis in response to selective increases in omental insulin.
RESEARCH DESIGN AND METHODS
Experiments were performed on eight male mongrel dogs (31.9 ± 2.2 kg [mean ± SE] at the time of experiments). For 4 weeks before experiments, the animals were fed a diet supplemented with fat, representing a normal weight-maintaining diet plus a supplement of ∼450 kcal from cooked bacon grease. The effects of fat supplementation on the development of insulin resistance and its compensation have been presented; five of these eight dogs were included in the previous study (10). One week before experimentation, with the dogs under general anesthesia, chronic catheters (Fig. 1) were implanted as previously described (11). Catheters were placed in the carotid artery for sampling of arterial blood and in the femoral vein for infusion of tritiated glucose. Two catheters were placed into the portal vein, one 4 cm upstream from the porta hepatis for portal replacement infusions of glucagon and insulin and one ∼1 cm upstream from the porta hepatis for sampling of portal vein blood. We have previously shown that infusions given >3 cm upstream from the liver are overall equally distributed among the liver lobes (12). An additional chronic catheter was placed in the superior mesenteric artery (SMA) ∼2–3 cm distal to its bifurcation from the abdominal aorta, such that the tip of the catheter was ∼7 or 8 cm from the bifurcation. This catheter was used to infuse insulin directly into the omentum, since the SMA perfuses >90% of the omental fat depot (13). On the morning of the experiment, an acute catheter was inserted into the saphenous vein for the variable infusion of labeled glucose.
Euglycemic clamp protocol.
Each dog underwent two types of euglycemic clamps in random order. All clamps were performed after a 16-h overnight fast. After a fasting sample was taken at −150 min, infusions of somatostatin (1 μg · kg–1 · min–1) (SRIF; Bachem California, Torrance, CA) and high-performance liquid chromatograpy–purified [3-3H]d-glucose (25 μCi + 0.25 μCi/min) (NEN Research Products, DuPont, Boston, MA) were started into the femoral vein. Portal replacement infusions of insulin (0.3 mU · kg–1 · min–1) (porcine insulin; Novo-Nordisk, Copenhagen, Denmark) and glucagon (1.0 ng · kg–1 · min–1) (porcine glucagon; Sigma Chemical, St. Louis, MO) were initiated at the same time. Glucose and glucose specific activity were clamped using a variable infusion of labeled glucose into the saphenous vein (“hot Ginf” method). After an equilibration period, basal samples were taken at −20, −10, and 0 min. At time 0, three successive 90-min insulin infusions were given at 0.3, 0.6, and 1.2 mU · kg–1 · min–1. These stepwise insulin infusions were given either directly into the superior mesenteric artery (in omental studies) or into the portal vein (in control studies). Samples were taken at 5, 10, 15, 20, 30, 45, 60, 75, and 90 min after the initiation of each insulin infusion.
Basal values were defined as the mean of all basal samples. Steady-state values were defined as the average of the last 30 min of each insulin infusion step (SS1, SS2, and SS3). Insulin area under the curve (AUCINSULIN) was calculated for each insulin infusion period using the trapezoidal rule and normalizing the basal AUC to a 90-min period. The average slope of the line for systemic AUCINSULIN versus insulin infusion rate reflects the efficiency with which a given insulin infusion raises systemic insulin concentration and was used to calculate extraction of insulin by the omentum before it reached the systemic circulation. Omental insulin extraction was estimated only for paired experiments as 1 minus the ratio of the average slopes of the systemic AUCINSULIN versus insulin infusion rate for each route (control and omental; Fig. 3). Omental insulin levels were estimated as equal to arterial levels during basal periods and throughout the control experiments; during omental insulin infusion, they were estimated using the equation
where omental plasma flow is estimated as 10 ml · kg–1 · min–1 (15). The KM parameters for insulin suppression of arterial FFA levels were calculated for each experiment by fitting a modification of the Michaelis-Menten equation:
where FFABASAL, Vmax, and KM are all fitted parameters.
Portal blood sampling was successful in four omental and four unpaired control experiments; therefore, all values of portal concentrations and omental lipolysis rates report only these eight experiments. Omental lipolysis was estimated by the equation
where FFA levels were presmoothed using an optimal segments algorithm (16) and PPF is portal plasma flow, estimated to be 16 ml · kg–1 · min–1 (17,18). This equation calculates net FFA output from the omentum, which is equal to total omental lipolysis minus omental FFA reesterification and oxidation. Whole-body lipolysis was estimated as a percent of its basal value by assuming it to be proportional to the systemic FFA levels (i.e., assuming no significant changes in FFA uptake). All calculations were made using Microsoft Excel 2000 on an IBM-compatible computer. All fits were performed with SlideWrite Plus for Windows (version 5.0; Advanced Graphics Software, Encinitas, CA).
All outcome variables were analyzed with general linear-model ANOVA, testing for effects of dog, insulin infusion route, insulin step, and interaction between insulin infusion route and step (Minitab, State College, PA). Individual comparisons between different insulin infusion rates for each infusion route were made by calculating Bonferroni 95% simultaneous confidence intervals, which adjust the P values for multiple comparisons.
Glucose and insulin.
Glucose concentration decreased slightly during insulin infusions (P < 0.001) but was not different between infusion routes (NS, ANOVA). Although glucose specific activity was not different between infusion routes, it tended to increase during the first insulin infusion period in both groups (P < 0.05 for basal vs. each SS period, ANOVA) but then remained stable for the remainder of the experiment (Table 1; Fig. 2). As expected, fasting insulin concentrations in these animals were elevated (163 ± 42 pmol/l) compared with the normal range previously observed in our laboratory (11,14,19), due to the high-fat diet and concomitant insulin resistance. Somatostatin and portal insulin replacement at 0.3 mU · kg–1 · min–1 therefore yielded basal insulin concentrations below the fasting levels in these animals, yet within the range expected for normal animals (59 ± 8 pmol/l) (Table 1). Although there was no significant effect of insulin infusion route on steady-state systemic insulin levels (P = 0.28, ANOVA), the AUCINSULIN was lower with omental infusion than control (P > 0.05, ANOVA). The slope of the relationship between AUCINSULIN and insulin infusion rate was decreased with omental infusion (Fig. 3). This was presumably due to first-pass omental extraction of insulin, which was calculated as 27 ± 7% with omental infusion.
FFAs and glycerol.
As expected, systemic FFA levels were suppressed with insulin infusion in a dose-response pattern, declining 80% at the highest dose (P < 0.01). The route of insulin infusion (SMA/omental versus systemic/control) had no significant effect on arterial FFA concentrations. At first glance, this might be interpreted to indicate that omental insulin was not more effective in overall suppression of lipolysis than intraportal insulin. However, this interpretation is confounded by the apparent degradation of insulin in the omental tissue bed. The effect of omental infusion on overall lipolysis is more correctly assessed from the dose response between systemic insulin concentration and FFA levels. With omental infusion, we observed a tendency for a leftward shift (KM = 35 ± 9 vs. 58 ± 12 pmol/l, omental versus control; P = 0.08, unpaired t test) (Fig. 4). That this tendency was slight suggested that the omental fat tissue was particularly resistant to direct SMA infusion of insulin. Therefore, net omental lipolysis was calculated from the four experiments in each group in which portal vein sampling was successful (Table 2; Fig. 5). Direct infusion of insulin into the omental adipose tissue bed tended to suppress omental lipolysis (omental, P = 0.12). Systemic insulin infusion had no effect on omental lipolysis (control, P = 0.85). Of course, direct infusion of insulin into the omentum via the SMA causes a marked increase in local insulin concentration far above the systemic increment. Therefore, we compared the relationship between local omental insulin concentration and net lipolysis of the omental tissue bed with suppression of lipolysis for the entire body (Table 3; Fig. 6). When the insulin and lipolysis data in Table 3 were fitted by linear regression, the slope of the line for suppression of omental lipolysis was less than one-third of that for suppression of whole-body lipolysis (control, −0.22 ± 0.04; omental, −0.07 ± 0.01; P = 0.011). This shift reflects a profound threefold increase in insulin resistance of the omental adipose bed in vivo relative to lipolysis of the body as a whole.
Glucose fluxes were not different between the two routes of insulin infusion (Fig. 7). However, omental insulin extraction yielded lower systemic insulin levels with omental infusions than with control infusions. Thus, omental insulin infusion suppressed endogenous glucose production (EGP) to the same extent as control infusion despite the lower systemic insulin levels. This implies that omental insulinization contributed to suppression of EGP independent of systemic insulin levels. The relationship between portal vein FFA levels and EGP was not different with route of insulin infusion (Fig. 8).
The omental adipose depot is unique in many respects. Unlike subcutaneous fat, effluent from omental fat empties directly into the portal vein. Omental fat has been reported to be relatively resistant to insulin suppression of lipolysis in vitro (8,20), and the tissue is sensitive to β3-adrenergic stimulation of lipolysis (21,22). Any or all of these characteristics may contribute to the observed strong correlation between abdominal obesity and whole-body insulin resistance. Although portal vein FFAs are strong candidate signals for hepatic insulin resistance, the possible contributions of central subcutaneous fat cannot be discounted (23). In a previous study, we found expansion of both subcutaneous and visceral fat in dogs fed fat for 4 weeks (10). In the present study, we implanted chronic catheters into the SMA of dogs to examine the physiologic role of the omental fat depot in vivo. Stepwise insulin infusions into the SMA during euglycemic clamps were used to generate supraphysiological omental insulin concentrations despite concentration of systemic insulin in the physiological range. Insulin was infused into the portal vein as a control for the elevations in portal and systemic insulin levels. In this manner, we were able to directly show severe insulin resistance of omental fat in vivo.
Our data contrast with studies examining insulin sensitivity of omental adipose tissue in vitro in human tissues. Studies have shown a substantial resistance of omental adipocytes to insulin compared with subcutaneous adipocytes in both obese (24) and nonobese (20,25) women. However, although the median effective dose (ED50) of visceral adipose tissue in the present study was approximately three times higher than that of subcutaneous adipose tissue, this parameter was only increased 2- to 2.5-fold in the human in vitro studies. Although this could be caused by sex and/or species differences, it could also reflect differences in the cellular milieu. Circulating hormones such as incretins and catecholamines may affect adipocyte sensitivity in vivo. Concentrations of paracrine and endothelial-derived factors would be altered in vitro. It is also possible that direct neuronal control of the omental fat depot by the central nervous system could affect insulin sensitivity, a finding that would remain undetectable in in vitro studies.
The present study confirms the findings of Meek et al. (9), who used hepatic vein sampling to calculate splanchnic lipolysis in humans. However, because hepatic vein plasma FFA levels reflect metabolism of both omental adipose tissue and liver, the Meek study relied on arteriovenous (A-V) difference of both labeled and unlabeled palmitate for the calculation of omental lipolysis. Because of the error inherent to the A-V difference technique, calculations relying on two A-V differences are prone to especially large errors. Also, because the splanchnic FFA uptake was much larger than release in the Meek study, small changes in lipolysis may have been difficult to detect. Finally, the Meek study used different patient groups for each insulin dose, which can increase error. Despite these problems, Meek et al. showed that splanchnic palmitate release was significantly more resistant to insulin suppression than was that from the leg, although ED50 values were not reported. Thus, the insulin resistance of lipolysis reported by Meek et al. may well have reflected omental adipose as opposed to lipolysis by the liver itself. An important finding in the present study is that omental lipolysis was not significantly suppressed when systemic insulin levels were raised to 365 pmol/l with portal insulin infusion. In our experience, insulin levels do not exceed 300 pmol/l after a mixed meal in normal dogs. Even in fat-fed dogs, systemic insulin levels only reach 430 pmol/l for 5 min after a mixed meal and decline to <360 pmol/l after that (data not shown). Therefore, postprandial insulin secretion would likely not suppress omental lipolysis. The nonsuppressible omental lipolysis in the postprandial state could help explain why abdominal obesity is strongly associated with hepatic insulin resistance. We and others have shown a strong correlation between FFA levels and hepatic glucose production (11); this could partially explain the inappropriately elevated glucose production in the face of hyperinsulinemia seen in obesity and insulin resistance. Also, increased hepatic delivery of FFAs should increase hepatic triglyceride levels, which also may play a role in insulin resistance.
Although insulin infusion rates were matched between omental and control experiments, systemic insulin levels were significantly lower with omental infusion. This was due to extraction of insulin by the omentum, which we calculated to be ∼27%. This is in contrast to the study of Pye et al. (26), which found no evidence for omental extraction of insulin in lean dogs. This may indicate that the abdominally obese dog exhibits a substantial increase in omental insulin extraction. Livingston et al. (27) reported no effect of obesity on insulin receptor affinity or number per cell in omental adipocytes. However, it is reasonable to assume that an increase in the total number of omental adipocytes would lead to a proportional increase in the number of insulin receptors, thereby increasing total omental insulin extraction.
Despite the suppression of omental lipolysis with omental infusion of insulin, we did not detect differences between the two protocols in systemic FFA and glycerol levels. There are several possible explanations for this apparent paradox. It may have been due to the substantial insulin extraction by the omentum, which confounds comparisons between omental and control infusion. Lower systemic insulin levels with omental infusion would be expected to yield less suppression of subcutaneous lipolysis, which could partially mask the effect of omental insulin to suppress omental lipolysis. In fact, there was an apparent leftward shift in the systemic FFAs versus systemic insulin relationship with omental infusion, such that at the same systemic insulin levels, omental infusion was able to suppress systemic FFAs further than control infusion.
Alternatively, it is possible that omental lipolysis did not contribute greatly to systemic FFA and glycerol levels in these experiments, and so its suppression by omental insulin did not yield observable effects in the systemic levels of these metabolites. However, Guo et al. (28) have shown the splanchnic bed to be a significant source (∼15–20%) of basal FFA release in obese women. It is possible that even the extremely high local insulin levels achieved with omental infusion of insulin were unable to appreciably suppress omental lipolysis. However, this latter possibility is disputed by the A-V difference results.
Omental lipolysis was difficult to quantify due to the inconsistency of portal sampling and the noise inherent in A-V difference measurements. However, portal sampling was successful with four dogs in each protocol. FFA release from the omentum was suppressed in a dose-dependent manner with omental insulin and not affected by portal insulin infusion. As a function of local insulin levels, there is a profound rightward shift of omental lipolysis compared with whole-body lipolysis. This demonstrates that omental adipose is very insulin resistant in vivo, even though it extracts one-fourth of the insulin that passes through it.
Additionally, the calculation of omental lipolysis depends on the assumption that portal blood flow remains constant throughout the euglycemic clamp procedure. Although portal blood flow was not measured in the present study, previous studies have shown no significant change in portal blood flow with portal insulin infusion (29).
Endogenous glucose production was similar with omental versus control insulin infusion. This was not surprising given the fact that neither arterial nor portal vein FFA levels were altered by route of insulin infusion. Because omental adipose is severely resistant to suppression of lipolysis by insulin, one would expect that it would tend to bolster hepatic glucose output during hyperinsulinemic states, such as after a meal. In fact, omental lipolysis was not suppressed at even the highest rate of portal insulin infusion (1.2 mU · min–1 · kg–1). Therefore, the omental release of FFAs is likely not suppressed by physiological insulin secretion. Over time, this might be expected to lead to increased hepatic triglyceride levels. When omental adiposity increases, this insulin resistant depot could exert even greater effects on the liver, perhaps representing an early step in the development of whole-body insulin resistance.
This work was supported by research grants from the National Institutes of Health (DK 27619 and DK 29867). S.D.M. performed this work while he was a predoctoral trainee supported by the National Institute of Aging (T32-AG-00093) and a postdoctoral fellow supported by a mentor-based grant from the American Diabetes Association. G.W.V.C. is supported by a supplement grant from the National Institutes of Health.
Address correspondence and reprint requests to Richard N. Bergman, University of Southern California School of Medicine, Department of Physiology and Biophysics, MMR 626, 1333 San Pablo St., Los Angeles, CA 90089-9142. E-mail:.
Received for publication 12 March 2001 and accepted in revised form 26 November 2001.
AUCINSULIN, insulin area under the curve; ED50, median effective dose; EGP, endogenous glucose production; FFA, free fatty acid; SMA, superior mesenteric artery.