| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Internal Medicine, Divisions of Endocrinology and Metabolism and Cardiology, Iowa City Veterans Affairs Medical Center and University of Iowa, and University of Iowa General Clinical Research Center, Iowa City, Iowa 52246
Address all correspondence and requests for reprints to: Dr. William Sivitz, Department of Internal Medicine, University of Iowa Health Care, 3E-17 VA, Iowa City, Iowa 52246. E-mail: william-sivitz{at}uiowa.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Hence, prior reports variably suggest that circulating leptin is unchanged (11, 12, 13, 14, 15), reduced (16, 17), or less responsive to glucocorticoid stimulation (18) as a consequence of type 2 diabetes. However, studies of humans with untreated type 1 diabetes and animals with insulin deficiency consistently demonstrate that plasma leptin concentrations and adipose leptin mRNA are reduced (19, 20, 21, 22). The variable results in type 2 diabetes are not surprising given that subjects differed with respect to extent of obesity, treatment regimens, age, and gender. Moreover, leptin was reported either as plasma concentration per se or normalized to various measures of adiposity, such as total fat, percent fat, or body mass index (BMI). Thus, there is still controversy concerning circulating leptin concentrations in type 2 diabetes.
We recently examined plasma leptin and insulin levels and indexes of adiposity in groups of age-matched subjects receiving no antihyperglycemic medications classified as 1) type 2 diabetes; 2) impaired fasting glucose or mild diabetes mellitus (IFG/mild DM); 3) nondiabetic, matched for BMI; and 4) nonobese, nondiabetic. Diabetic subjects were also studied prospectively before and after treatment with glyburide or metformin. These studies were carried out as part of the work of our Veterans Administration/Juvenile Diabetes Research Foundation-sponsored diabetes center primarily directed at vascular disease in diabetes. In the context of this project, all participants underwent metabolic studies and measures of in vivo (forearm) endothelial function. The endothelial studies are the subject of a separate report.
In recruiting study participants, nondiabetic control subjects were grouped as adiposity-matched or nonobese on the basis of the commonly used BMI. However, as dual energy x-ray absorptiometry (DEXA) studies of fat distribution were carried out, it became clear that percent and total body fat varied considerably at any given range of BMI, raising some questions as to the most appropriate measure of adiposity. In particular, questions arose about how to evaluate metabolic data affected by degree of obesity. In this regard we considered different measures of adiposity and the consequences of using these for interpretation of plasma leptin concentrations in obesity and diabetes.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The recruiting and assignment of subjects are depicted in Fig. 1
. Three groups of subjects were recruited: 1) type 2 diabetes, defined for the purposes of this study as fasting plasma glucose (FPG) greater than 6.9 mmol/liter (125 mg/100 ml); 2) BMI-matched, nondiabetic control subjects with FPG less than 6.1 mmol/liter (110 mg/100 ml) matched to the diabetic subjects for age and gender; and 3) nonobese, nondiabetic subjects (FPG, <6.1 mmol/liter) matched for age and gender. Based on observations of the mean and range of BMI in the diabetic subjects, the two control groups were separated and assigned (for the purposes of this study) as BMI of 28 or more (BMI-matched controls) and BMI below 28 (nonobese controls). Otherwise, the inclusion and exclusion criteria (see below) did not differ between these groups.
|
Inclusion criteria
For inclusion in the group with diabetes, subjects were required to have 1) FPG greater than 6.9 mmol/liter, 2) age between 25–70 yr, 3) no history of ketoacidosis, 4) ability to perform self-blood glucose determinations, 5) stable blood pressure not anticipated to require new treatment with antihypertensive medication or adjustment of existing antihypertensive medication during the study period, and 6) no antilipid medication or stable with current dose of antilipid medication (not anticipated to require adjustment of existing antilipid medication during the study period). For inclusion as BMI-matched or nonobese controls, subjects were required to meet inclusion criteria 2, 3, 5, and 6 and have FPG less than 6.1 mmol/liter. Inclusion as a nonobese, nondiabetic subject required a BMI less than 28. Inclusion as a BMI-matched, nondiabetic control required a BMI of 28 or more.
Exclusion criteria
Participation as a subject in any group was precluded by 1) contraindication to glyburide or metformin; 2) FPG greater than 13.9 mmol/liter (250 mg/100 ml) or hemoglobin A1c (HbA1c) greater than 11.6%; 3) clinical evidence of coronary artery disease, or symptomatic cerebral or peripheral vascular disease; 4) low density lipoprotein cholesterol greater than 4.14 mmol/liter (160 mg/100 ml), triglycerides greater than 4.52 mmol/liter (400 mg/100 ml); 5) any acute or chronic medical or psychiatric condition likely to preclude compliance with diet regimen; 6) history of surgical treatment of obesity including liposuction; 7) BMI greater than 45; 8) Cushing’s syndrome, acromegaly, or any other disorder likely to alter glycemia; 9) any active medical or surgical disorder judged likely to interfere with safe participation of the subject or with the scientific validity of the studies to be performed; 10) alcoholism or drug abuse during the past 5 yr; and 11) smoking in past 3 months.
Experimental design
A randomized, open label, cross-over design was followed. Subjects followed the protocol graphically depicted in Fig. 1
.
Potential participants in the diabetes group underwent an initial screening visit for fasting glucose and HbA1c and were interviewed by a research nurse coordinator. After this, if they appeared to meet inclusion and exclusion criteria they returned for a second screen that included a history and physical examination, fasting glucose measurement, and electrocardiogram. As fasting glucose was measured twice in this protocol, the inclusion and exclusion criteria (see above) were based on an average value.
If study criteria were met, subjects returned for visit 1 (V1). Diabetic subjects were advised on a weight maintenance meal plan and given supplies for self-glucose monitoring. Oral glycemic medication and/or insulin were discontinued. Subjects returned in 10 wk for V2 if discontinued from antihyperglycemic medication or in 4 wk if not originally taking antihyperglycemic medication. Depending on the judgment of the investigators, some diabetic subjects discontinued from medication returned in the interim (between V1 and V2) for a review of their glucose-monitoring data to assure they were not ill or excessively hyperglycemic off medication.
At V2, fasting serum glucose was determined and used in the classification and assignment of participants as depicted in Fig. 1. Two weeks after V2, subjects were admitted to the GCRC for a baseline evaluation visit. Subjects were admitted in the evening or the morning before 0800 h. Overnight fasting (10 h) blood samples (
120 cc) were obtained for measurement of leptin, FFA, lipid profile, and storage for potential future biochemical analysis not currently specified. After blood sampling, subjects were taken to our cardiovascular laboratory for evaluation of forearm vascular reactivity. These studies, which will be reported in a separate manuscript, required the subjects to lie still on an examination table for 3 h while plethsmographic data were obtained. Short-acting, locally vasoactive substances were injected into a brachial artery as previously described (23). These compounds were injected in amounts insufficient for systemic effects. In the afternoon (after lunch), subjects underwent measurement of body fat distribution by DEXA. The next morning, after a 10-h overnight fast, insulin sensitivity and release were assessed by the insulin-modified, frequently sampled, iv glucose tolerance test (FSIVGTT).
Nondiabetic subjects and subjects with IFG/mild DM completed the protocol at this point. After baseline studies, participants with diabetes (Fig. 1) were randomly assigned to receive metformin or glyburide for 3 months, followed by cross-over to the opposite drug. Subjects returned to the GCRC for repeat baseline studies after 3 months of treatment with each drug (Fig. 1).
Pharmacological treatment
Standard clinical guidelines were followed for pharmacological treatment. Patients were seen at the GCRC 4 wk after initial randomization and cross-over. Doses were adjusted until a goal of FPG less than 6.9 mmol/liter (125 mg/100 ml) was achieved or until a maximum dosage (20 mg/d glyburide or 2550 mg/d metformin) of oral agent was reached, after which no additional change was made for the 3-month period and no additional drugs were added. Dose increments were implemented at 2 wk (phone contact) and 4 wk (study visit) after randomization or cross-over based upon an average self-reported fasting glucose over the prior week of greater than 6.9 mmol/liter (125 mg/100 ml). If at least five of seven values (prior week) were not reported, no change was made, the subject was asked to record values again for 1 additional week, and subsequent adjustment was made based upon an additional phone contact. Increments in dosing were as follows. Glyburide was administered in an initial dose of 5 mg prebreakfast and incrementally raised to 5 mg prebreakfast and presupper and to 10 mg prebreakfast and presupper. Metformin was administered in an initial dose of 500 mg prebreakfast and presupper and was incrementally raised to 500 mg prebreakfast, prelunch, and presupper and to 850 mg prebreakfast, prelunch, and presupper. For glyburide, safety measures were included to prevent hypoglycemia. If the average of self-monitored fasting glucose values in the 2 wk before initial glyburide treatment was less than 8.3 mmol/liter (150 mg/100 ml), then all glyburide doses were reduced by 50%, with the maximal dose being 5 mg prebreakfast and presupper rather than 10 mg. In addition, in some subjects receiving once daily glyburide, the drug was given at bedtime rather than in the morning, if, in the judgment of the investigator, the glucose monitoring pattern revealed higher morning glucose readings. Study medication was provided in bottles free of cost to the subjects, and pill counts were made to assure compliance.
Specific procedures
Insulin-modified FSIVGTT. Subjects were fasted overnight (10 h) before study. Baseline blood samples (3 ml) were obtained at -15, -10, -5, and 0 min through a heparin-locked indwelling catheter fitted to a three-way stopcock. At zero time, glucose (0.3 g/kg) was injected iv over 1 min through a catheter in the opposite forearm. Additional blood was drawn at 2, 5, 10, 15, and 20 min postinjection. At 20 min, regular human insulin (0.05 U/kg) was injected as an iv bolus through the line previously used for glucose injection. Additional blood samples were obtained at 23, 24, 25, 27, 30, 35, 40, 50, 60, 70, 80, 90, 100, 120, 140, 160, and 180 min. All samples were analyzed for glucose and insulin. The data were fit to the Bergman minimal model of glucose kinetics (24) using the program MINMOD. Insulin sensitivity was expressed as the sensitivity index (SI; minutes per microunits per milliliter), calculated from the computer data fit. SI measures the effect of insulin in an active compartment remote from the plasma to restore glucose toward basal levels. We also calculated the area under the insulin curve in the first 20 min after glucose injection as an index of the insulin secretory capacity. The insulin modification of the FSIVGTT (25) was used to induce a substantial change in circulating insulin during the period of changing glucose, a circumstance needed for accurate curve fitting in patients with type 2 diabetes whose endogenous insulin response to the glucose injection was expected to be impaired.
The FSIVGTT was completed for only 11 of the 27 nonobese subjects (BMI <28 as defined for recruiting purposes). This procedure was discontinued after the first 11 of these subjects because of an unacceptable incidence of hypoglycemia (glucose <2.8 mmol/liter or symptoms). The FSIVGTT was completed in only 22 of the 26 BMI-matched controls because of technical problems (hemolysis or venous access) or the subject declining the procedure.
Body fat distribution. Regional adipose mass was measured by DEXA, performed using a QDR 4500 Elite Scanner (Hologic, Inc., Bedford, MA).
Plasma determinations. The human leptin concentration was determined by RIA using kits purchased from Linco Research, Inc. (St. Charles, MO). Free insulin concentrations were determined by Microparticle Enzyme Immunoassay (Abbott Laboratories, North Chicago, IL) automated on the Abbott IMx after precipitation by polyethylene glycol.
Data analysis Differences between groups were determined by ANOVA, as specified in the text and figure legends. These analyses as well as the linear correlations and curve fitting were performed using commercial software (PRISM, GraphPad Software, Inc., San Diego, CA).
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Table 1 lists the characteristics of the study participants by group. Twenty-six diabetic subjects participated. At baseline, DEXA studies of body fat were performed in 23 subjects, as 3 subjects weighed more than 113.6 kg (250 lb), the maximum allowable for the procedure. All studies of parameters normalized to measures of body fat included those 23 diabetic subjects compared with those in the other groups listed in Table 1. In comparison with the BMI-matched control group, nonobese controls had significantly lower weight, percent body fat, and absolute fat mass as well as higher high density lipoprotein cholesterol. Diabetic subjects had been advised on a weight maintenance diet, and body weights were stable at the time of GCRC admission (92.3 ± 3.1 kg at the initial GCRC admission and 92.4 ± 3.1 when determined at V2, 2–3 wk before admission). The diabetic group had lower high density lipoprotein cholesterol and higher HbA1c compared with the BMI-matched subjects.
|
lists the characteristics of the diabetic subjects at baseline (before drug treatment) and after 3 months of each drug. Two subjects with diabetes who were close to 113.6 kg before glyburide treatment increased to above 113.6 kg during treatment. Therefore, 21 subjects, rather than 23, were included in Table 2, so that all subjects had DEXA measurements at baseline and after both drug treatment periods. These 21 subjects were included in the analyses comparing the diabetic subjects at baseline to those after treatment with glyburide or metformin.
|
and 2 did not significantly differ if all 26 subjects, as opposed to 23 or 21, were included. Plasma leptin is strongly related to percent body fat and is independent of gender
Among the combined nonobese and BMI-matched nondiabetic control groups, a strong curvilinear relationship was observed when leptin was plotted against percent body fat that fit well to an exponential curve. Subsequently, we noted strong correlations between log leptin and various measures of adiposity (Fig. 2). The strongest occurred when leptin was expressed relative to percent fat (Fig. 2A) as opposed to total body fat mass (Fig. 2B) or BMI (Fig. 2C). The relationship between log leptin and percent body fat appeared independent of gender. The linear curve of leptin vs. percent fat was used to calculate normalized leptin values expressed as percent predicted. This normalized value was calculated as 100 x the unmodified measured leptin concentration divided by the antilog of the predicted log leptin based on the linear relationship shown in Fig. 2A. Likewise, leptin values were normalized to total fat and BMI. As shown in Fig. 2D, leptin differed by gender when normalized to total fat or BMI. As Rosenbaum et al. (26) reported a strong linear correlation between unmodified plasma leptin concentration and fat mass, we also examined our data in this way and observed a similar strong relationship (Fig. 2E) and a difference in leptin/fat mass by gender (Fig. 2F).
|
). Leptin normalized to nontruncal fat (expressed as the percent predicted, calculated as described above) did not significantly differ by gender (102 ± 4 in females compared with 96 ± 7 in males), but differed when normalized to truncal fat (112 ± 4 in females compared with 80 ± 4 in males). Females had a higher percent truncal fat than males (17.0 ± 0.8 compared with 11.9 ± 1.0; P < 0.001, by unpaired t test). Plasma leptin is decreased in diabetic subjects
Plasma leptin was plotted against percent body fat in the control (combined nonobese and BMI-matched groups) and diabetic subjects at baseline, and exponential curves were fit to both groups (Fig. 3A). When examined as an exponential function of percent body fat, plasma leptin is reduced at any given percent fat above 25%, the range in which our diabetic subjects were largely distributed (range, 22.1–56.8%, with only three subjects <26%). The difference between the control and diabetic subjects is evident in linear plots of log leptin vs. percent fat (Fig. 3B), wherein there is no overlap between the 95% confidence intervals for the slopes of these lines (slope for control subjects, 0.042; confidence interval, 0.036–0.047; r2 = 0.826; P < 0.0001 compared with a slope in the diabetic group of 0.027; confidence interval, 0.020–0.035; P < 0.0001). Subjects with IFG/mild DM were intermediate in normalized leptin per degree of percent fat (slope of leptin vs. percent fat, 0.035 ± 0.008; r2 = 0.688; P = 0.016).
|
to determine the percent predicted leptin for both the control and diabetic subjects. The resulting values for normalized leptin were compared in the nonobese, BMI-matched, IFG/mild DM, and diabetic groups (Fig. 3C). This revealed a decrease in normalized leptin in the diabetic compared with the BMI-matched control group and an intermediate result for the IFG/mild DM group.
We further examined the effect of glucose intolerance and type 2 diabetes on plasma leptin in separate groups of male and female subjects. In these analyses, plasma leptin was expressed in alternative ways based on indexes of body fat. As indicated in Table 3, plasma leptin was significantly lower in female diabetic subjects compared with BMI-matched controls when normalized to any measure of adiposity. Table 3 also shows that plasma leptin was lower in the male diabetic subjects compared with BMI-matched controls; however, those data did not achieve statistical significance. In this regard, we noted divergent relationships for log leptin or unmodified leptin concentration vs. indexes of adiposity (Fig. 3, B, D, and E). Thus, the greater adiposity among the female subjects may explain why the decrease in leptin secondary to diabetes (Table 3) achieved significance only for the female subjects.
|
Insulin release and sensitivity were assessed using the insulin-modified FSIVGTT. Leptin normalized to percent fat correlated significantly with insulin release measured as the integrated area under the curve of glucose vs. time over 0–20 min of the FSIVGTT (insulin given at 20 min), but not to SI (Fig. 4). Leptin normalized to percent fat was poorly related to fasting plasma insulin (Fig. 4).
|
Table 4 shows that plasma leptin, leptin normalized in different ways to indexes of adiposity, and leptin concentration/fat mass were all significantly increased after glyburide treatment compared with no treatment or metformin treatment. In contrast to glyburide treatment, metformin had no effect on the various measures of leptin despite an equivalent improvement in glycemia (Table 1). As these analyses were carried out in paired fashion (repeated measures) using each subject as his/her own control, gender was not at issue.
|
Insulin sensitivity, measured as SI, was progressively impaired in subjects classified as BMI-matched, IFG/mild DM, and type 2 diabetes compared with nonobese controls (Fig. 5A). Likewise, insulin secretion, measured as insulin area under the curve for the first 20 min after iv glucose (Ins AUC0–20 min), was progressively reduced in the IFG/mild DM group and type 2 diabetic subjects compared with BMI-matched controls, although it was considerably greater in the BMI-matched compared with the nonobese group (Fig. 5B). Insulin sensitivity improved after treatment of type 2 diabetes with either metformin or glyburide (Fig. 5C), but the difference from baseline (pretreatment) was significant only for metformin. Glucose-induced insulin release was significantly improved by both drugs (Fig. 5D).
|
, A and B) or HbA1c (Fig. 6, C and D). The data reveal a rapid decline in insulin release beginning with a minimal degree of glucose intolerance. In contrast, plasma insulin varies somewhat differently, approaching the pattern of the well described inverse U-shaped curve (27, 28).
|
FFA in control and diabetic subjects
Figure 7 demonstrates that, among the combined nonobese and BMI-matched control groups, FFA correlated weakly, but significantly, with truncal fat, (Fig. 7B), but not with percent body fat (Fig. 7A). FFA also did not correlate significantly to either total fat mass or BMI (not shown). Among the four groups of subjects in Table 1, FFAs were significantly lower in both the nonobese and BMI-matched control groups compared with the group with type 2 diabetes (Fig. 7C). Treatment with neither metformin nor glyburide altered circulating FFA concentrations compared with the baseline levels (Fig. 7D).
|
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Plasma leptin was also examined in alternative ways, although this necessitated separation by gender to avoid confounding effects of this factor. At first, the data in Table 3 may suggest that the difference in leptin between the diabetic and BMI-matched subjects manifests only in female subjects. However, it is also quite possible that the apparent female specificity is explainable based on the diverging relationships between curves of leptin vs. adiposity in diabetes vs. controls. As this higher range of adiposity encompassed mostly female subjects, increased fat per se, rather than gender, may well explain the differences in the data between males and females. Unfortunately, our data cannot sort this out, as we did not have enough male diabetic subjects of sufficiently high body fat.
Although leptin appears to be reduced in untreated type 2 diabetes by any of the measures in Table 3, there is strong rationale supporting the normalization of leptin to percent fat rather than BMI or total fat mass. The close correlation of log leptin to percent fat was first appreciated shortly after the hormone was discovered in a series extending to over 500 subjects (4, 29), and, as in our data, this relationship was independent of gender (29). The physiological rationale may be that plasma leptin is by definition a measure of concentration and not secretion rate. Thus, the plasma concentration would depend on the distribution volume. Therefore, a higher percent fat at a given total fat mass would imply a smaller distribution volume, and the concentration, for a given amount secreted, would be greater at higher percent fat. The observation that plasma leptin clearance does not appear to be altered by adiposity (30) is consistent with this contention. The use of log leptin as opposed to unmodified plasma leptin values is also based on physiological reasoning, as dose-response characteristics for hormones and hormone-receptor binding curves are nonlinear and usually represented by response vs. log dose.
Our results are important, as there is still controversy concerning circulating leptin concentrations as affected by type 2 diabetes. Surveys of Mexican-American (11) and German (10) subjects showed that leptin did not differ between subjects with type 2 diabetes and controls with matched BMI. Percent or total adipose mass was not determined in these studies, and pharmacological treatment effects were not considered. In another report, baseline plasma leptin did not significantly differ between subjects with newly diagnosed or long-standing type 2 diabetes compared with nondiabetic controls matched for BMI; however, plasma leptin responsiveness to dexamethasone was impaired in the diabetic groups (18). Other reports comparing plasma leptin between controls and weight-matched subjects with type 2 diabetes have led to discrepant conclusions, showing no effect (12, 13, 14, 15) or a decrease in leptin (16, 17).
As opposed to type 2 diabetes, plasma leptin is consistently reduced in human type 1 diabetic subjects when they are insulin deficient (19, 20) and in rodents (21, 22) with hyperglycemia and hypoinsulinemia. In rodents, insulin replacement rapidly restores plasma leptin (21). Furthermore, insulin is a leptin secretagogue in isolated fat cells (31). In addition, insulin-treated type 1 diabetes is reportedly associated with increased plasma leptin, a state possibly explained by chronic hyperinsulinemia (32). Increased plasma leptin was also associated with insulin therapy, independent of BMI, in type 2 diabetic patients participating in the United Kingdom Prospective Diabetes Study (33).
Given these considerations, it seems reasonable that variations in circulating insulin explain much of the controversy concerning leptin in type 2 diabetes. Thus, differences in treatment with drugs or diet that alter insulin secretion may account for the differences between studies. This might also affect insulin sensitivity, although, as shown in Fig. 4, our data suggest that leptin correlates better with insulin release than sensitivity.
In like fashion, the effect of glyburide to increase plasma leptin, even when normalized in different ways to adiposity, may be explained through enhanced insulin secretion. Glyburide treatment resulted in weight gain even though plasma leptin concentrations increased, suggesting that the glyburide-induced increase in insulin was potent enough to restore fat despite leptin. Our results for glyburide are consistent with those reported by Haffner et al. (34), who found that sulfonylurea treatment increased leptin adjusted for BMI, but differ from the findings of Ozata et al. (15), who reported a decrease in circulating leptin concentration after sulfonylurea therapy. In the former study, subjects gained weight, whereas in the latter, there was actually a slight, but significant, weight loss despite sulfonylurea treatment and glycemic improvement. This suggests that sulfonylurea treatment in the study by Ozata et al. (15) was accompanied by caloric restriction, as the expected effect of sulfonylurea-induced glycemic improvement per se would be to slightly increase weight. In contrast to glyburide, it is interesting that metformin also increased glucose-induced insulin secretion and improved insulin action (Fig. 5), yet had no effect on plasma leptin (Table 4). The reason for this is unclear. The results suggest that metformin has some other effect to limit fat mass that does not involve insulin or leptin. In this regard, biguanides have been used worldwide for decades despite the poor understanding of their effects at the cellular and molecular level.
Although speculative, the reduction in leptin per adipose mass in diabetes may have metabolic consequences. First, leptin may enhance whole body insulin sensitivity (6) and increase glucose turnover (7). Moreover, leptin may improve glucose uptake by muscle (7) and decrease hepatic glucose production (35). Finally, there is evidence that leptin may protect against the adverse effects of fat accumulation within nonadipose cells (36), although, with respect to this latter point, this evidence derives from studies in rodents with a genetic reduction in leptin that is considerably greater than that seen in our human subjects with type 2 diabetes.
Notwithstanding the above, the potential consequences of the reduction in leptin (per adipose mass) in diabetes could be confounded by the putative effects of leptin to induce leptin resistance (37). Given this consideration, less circulating leptin may lead to improved sensitivity to leptin, possibly offsetting the consequences of the reduction. However, presumably, any such increase in leptin sensitivity would be compensatory and thus not able to completely prevent such consequences.
Moreover, although metformin treatment did not increase circulating leptin, we do not believe the considerations discussed above should discourage the use of this drug. Metformin improved glycemia and both insulin sensitivity and acute glucose-induced insulin release. Furthermore, metformin may have intracellular effects in hepatocytes or other cells to suppress lipogenic transcription factors (38). Finally, if leptin does induce leptin resistance, then the lack of an increase in leptin as a result of metformin would obviate any such leptin resistance.
Defects in first phase or acute insulin release upon glucose stimulation are demonstrable even with very mild impairments in glucose tolerance (39). Our results for acute glucose-induced insulin release are consistent. Further, our overall data seem to support the importance of this early reduction in insulin secretory capacity in the pathogenesis of type 2 diabetes. Plasma leptin correlated more with glucose-induced insulin secretion than sensitivity. As described above, an early defect in insulin secretion would lower circulating leptin and exacerbate insulin resistance as well as further compromise islet function. If this occurred in individuals with preexisting insulin resistance of obesity or genetic insulin resistance, this vicious cycle would be all the more active.
We also examined FFA concentrations in control and diabetic subjects and in response to glyburide and metformin. FFA were significantly elevated in type 2 diabetes compared with controls and were intermediate in the IFG/mild DM group. This was expected, consistent with the well established effect of insulin to inhibit FFA release. Our data also confirm the relation of FFA to intraabdominal fat (40), at least as estimated here by truncal fat. However, FFA, as opposed to leptin, correlated only weakly with percent truncal fat and insignificantly with percent body fat, with far more scatter, suggesting that additional factors beyond fat mass are of major importance in regulating FFA concentrations. Given that insulin and catecholamines regulate leptin as well as FFAs, it seems somewhat remarkable that leptin, compared with FFAs, remains so closely dependent on fat mass per se.
Finally, our data provide a head to head, nonindustry-sponsored comparison of the metabolic effects of glyburide and metformin. As expected, the two drugs induced similar glycemic improvement. Less predictably, both drugs improved both insulin secretion and sensitivity, although the effect of glyburide on SI missed significance. Possibly, the effects of the insulin sensitizer, metformin, on insulin release and those of the insulin secretagogue, glyburide, on insulin sensitivity were secondary to glucose reduction and consequent improvement in glucotoxicity (41).
In summary, our data show that the plasma leptin concentration, normalized to adiposity, is reduced in untreated diabetes, probably as a consequence of insulin deficiency. Moreover, the fasting circulating leptin concentration is differentially affected by mono-drug therapy with glyburide or metformin. Glyburide increased plasma leptin as well as insulin release, probably accounting for the weight gain noted with the use of this drug. Metformin increased insulin secretion and improved insulin sensitivity, but did not alter weight or leptin concentrations, suggesting that metformin has another off-setting effect to limit adiposity.
|
Footnotes |
|---|
Abbreviations: BMI, Body mass index; DEXA, dual energy x-ray absorptiometry; DM, diabetes mellitus; FFA, free fatty acids; FPG, fasting plasma glucose; FSIVGTT, frequently sampled, iv glucose tolerance test; GCRC, General Clinical Research Center; HbA1c, hemoglobin A1c; IFG, impaired fasting glucose; Ins AUC0–20 min, insulin area under the curve for the first 20 min after iv glucose; SI, sensitivity index; V1, V2, visits 1 and 2.
Received July 30, 2002.
Accepted January 7, 2003.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. Klein, S. Westphal, D. Kraus, B. Meier, N. Perwitz, V. Ott, M. Fasshauer, and H H. Klein Metformin inhibits leptin secretion via a mitogen-activated protein kinase signalling pathway in brown adipocytes J. Endocrinol., November 1, 2004; 183(2): 299 - 307. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Piemonti, G. Calori, A. Mercalli, G. Lattuada, P. Monti, M. P. Garancini, F. Costantino, G. Ruotolo, L. Luzi, and G. Perseghin Fasting Plasma Leptin, Tumor Necrosis Factor-{alpha} Receptor 2, and Monocyte Chemoattracting Protein 1 Concentration in a Population of Glucose-Tolerant and Glucose-Intolerant Women: Impact on cardiovascular mortality Diabetes Care, October 1, 2003; 26(10): 2883 - 2889. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
) subjects and indexes used to assess body fat. The logarithm of plasma leptin is plotted against percent body fat (A), total body fat mass (B), and BMI (C). Normalized leptin (expressed as a percentage of the predicted value based upon percent body fat, total fat mass, and BMI) was compared by gender (D). The unmodified plasma leptin concentration is also plotted against fat mass (E) and leptin/fat mass compared by gender (F). ns, Not significant.
