We tested the impact of long-term near normoglycemia (HbA1c <7% for >1 year) on glycogen metabolism in seven type 1 diabetic and seven matched nondiabetic subjects after a mixed meal. Glycemic profiles (6.2 ± 0.10 vs. 5.9 ± 0.07 mmol/l; P < 0.05) of diabetic patients were approximated to that of nondiabetic subjects by variable insulin infusion. Rates of hepatic glycogen synthesis and breakdown were calculated from the glycogen concentration time curves between 7:30 p.m. and 8:00 a.m. using in vivo 13C nuclear magnetic resonance spectroscopy. Glucose production was determined with d-[6,6-2H2]glucose, and the hepatic uridine-diphosphate glucose pool was sampled with acetaminophen. Glycogen synthesis and breakdown as well as glucose production were identical in diabetic and healthy subjects: 7.3 ± 0.9 vs. 7.1 ± 0.7, 4.2 ± 0.5 vs. 3.8 ± 0.3, and 8.7 ± 0.5 vs. 8.4 ± 0.7 μmol · kg−1 · min−1, respectively. Although portal vein insulin concentrations were doubled, the flux through the indirect pathway of glycogen synthesis remained higher in type 1 diabetic subjects: ∼70 vs. ∼50%; P < 0.05. In conclusion, combined long- and short-term intensified insulin substitution normalizes rates of hepatic glycogen synthesis but not the contribution of gluconeogenesis to glycogen synthesis in type 1 diabetes.
Early studies on glycogen metabolism in type 1 diabetic patients using liver biopsies revealed controversial results, reporting either increased or decreased liver glycogen concentrations (1,2,3,4). When liver glycogen was continuously measured with 13C nuclear magnetic resonance (NMR) spectroscopy, it became clear that under physiologic conditions of mixed meal ingestion, poorly controlled type 1 diabetic patients indeed exhibit a defect of net liver glycogen synthesis that accumulates throughout the day and is most pronounced after dinner (5). Furthermore, these authors reported higher contribution of the indirect (carbon 3 compounds → glucose → glucose-6-phosphate → glucose-1-phosphate → UDP-glucose → glycogen) compared with the direct pathway of glycogen synthesis (glucose → glucose-6-phosphate → glucose-1-phosphate → UDP-glucose → glycogen) (5,6). We recently found that type 1 diabetic subjects with poor metabolic control, as evidenced by an HbA1c of ∼9%, also present with reduced glycogen breakdown during the night after mixed meal ingestion (7). Short-term intensified insulin treatment for 24 h, resulting in near-normal plasma glucose concentrations, improved both defects in glycogen synthesis and breakdown that, however, were still ∼52 and ∼26% lower, respectively, compared with nondiabetic subjects.
At present, even advanced insulin substitution regimens do not resemble the physiologic insulin secretion pattern, since peripheral administration of insulin distorts the portal-to-peripheral insulin gradient and thereby affects hepatic glycogen turnover (8). Furthermore, the physiologic inhibition of glucagon secretion by insulin is impaired (9), giving rise to the portal vein glucagon-to-insulin ratio. Even a small increase in this ratio would dramatically reduce hepatic glycogen accumulation (10,11,12). Finally, a defect in the “portal signal,” which stimulates hepatic uptake of glucose on its enteral or portal vein delivery (13,14,15), could be present in diabetic patients.
Such defects likely contribute to the excessive fasting endogenous glucose production (EGP) and plasma glucose concentrations that are found in poorly controlled type 1 diabetic patients (16,17). Long- and short-term near normoglycemia exerts an additive effect to decrease EGP and fasting hyperglycemia in those patients (18,19). It is conceivable that long- and short-term near normoglycemia could further improve hepatic glycogen metabolism and glucose production.
Therefore, this study was designed to test the hypothesis that long-term intensive insulin therapy normalizes hepatic glycogen metabolism in type 1 diabetes. To this end, rates of hepatic glycogen synthesis and breakdown, pathways of glycogen synthesis, and rates of EGP were simultaneously examined in type 1 diabetic patients with excellent metabolic control (HbA1c <6.5%) and matched nondiabetic subjects under “normal life” conditions, i.e., ingestion of mixed meals.
RESEARCH DESIGN AND METHODS
Type 1 diabetic patients (six men and one woman, aged 30 ± 1 years, BMI 23.1 ± 1 kg/m2, mean HbA1c 6.2 ± 0.2% [range 6.0–6.5%], diabetes duration 12 ± 1 years, mean daily insulin dosage 73 ± 5 units) and nondiabetic subjects without family history of diabetes, matched for age and body weight (six men and one woman, aged 29 ± 3 years, BMI 23.3 ± 1 kg/m2, mean HbA1c 5.4 ± 0.1%) were studied. After 1 week of intensive training, all type 1 diabetic patients were under excellent long-term metabolic control (Fig. 1A) as a result of multiple insulin injections using the basal-bolus principle and algorithms derived from physiological knowledge (20). They were recruited from our diabetes outpatient service and seen by a trained physician at least four times per year since the onset of diabetes. The type 1 diabetic patients had no history of hypoglycemic episodes for at least 2 weeks before the study and did not present with any diabetes-related complications. Written consent was obtained from all subjects after explanation of the purpose, nature, and potential risks of the studies. The protocol was reviewed and approved by the Human Ethics Comitee of the University of Vienna Medical School.
All subjects were on a carbohydrate rich, weight-maintaining diet and refrained from strenous physical exercise for at least 3 days before the study. Patients were instructed to omit NPH- or Zn-insulin and to correct their plasma glucose concentration with regular insulin only (Actrapid; Novo Nordisk, Bagsvaerd, Denmark) during 24 h before admission.
On day 1, the type 1 diabetic subjects arrived at the metabolic ward between 11:00 p.m. and 12:00 a.m. After insertion of Teflon catheters into antecubital veins of the left and right arm for blood sampling and insulin infusion, respectively, insulin was administered as a variable intravenous infusion of 0.2–1.5 mU · kg−1 · min−1 to slowly lower plasma glucose and maintain its concentration between 5.56 and 6.67 mmol/l (7), which was monitored in hourly intervals.
On day 2, three standard mixed meals (60% carbohydrate, 20% protein, and 20% fat) were served at 8:00 a.m. (720 kcal), 12:30 p.m. (710 kcal), and 5:00 p.m. (800 kcal). Healthy subjects were admitted on day 2, and catheters were inserted as described above. Blood was drawn from all subjects to measure plasma glucose every 15 min from 8:00 a.m. to 8:00 p.m. and, when appropriate, during the night. Hormones and metabolites were sampled before dinner and at timed intervals thereafter. At 7:00 p.m., participants were transferred to the magnetic resonance unit, where in vivo 13C NMR spectroscopy was performed from 7:30 p.m. until the plateau of hepatic glycogen concentration was reached at ∼10:30 p.m. Thereafter, 13C NMR spectroscopy was performed from 2:00 a.m. to 3:00 a.m. and again between 6:45 a.m. and 7:45 a.m. Rates of EGP were determined using a bolus [5 (mg/kg) × body weight (kg) × fasting blood glucose (mg/dl)/90 (mg/dl)]-continuous [0.05 (mg/kg) × body weight (kg)/(min)] infusion of [6,6-2H2]glucose (Cambridge Isotope Laboratories, Andover, MA) for 3 h from 3:00 a.m. to 6:00 a.m. Blood samples to measure 2H mol percent enrichments (MPEs) in glucose were taken every 15 min during the last 45 min of the infusion (21).
At 8:00 a.m. on day 3, all subjects were given a breakfast that was identical to that of day 2 but with 10 g [1-13C]glucose added, and they ingested 500 mg acetaminophen. Blood was drawn for measurement of 13C MPEs in glucose and acetaminophen-glucuronide at −15, 90, 120, 150, 180, 210, and 240 min to assess the contribution of the direct and indirect pathways of glycogen synthesis.
Hormones and metabolites.
Plasma glucose concentrations were measured using glucose oxidase (Glucose Analyzer II; Beckman, Fullerton, CA). HbA1c (intra- and interassay coefficients of variation [CVs] <3%) was measured after high-performance liquid chromatography (HPLC) separation by cation exchange columns as previously described (22). Plasma concentrations of free fatty acids (FFAs) (CVs <6%) were determined enzymatically (Wako, Neuss, Germany). Plasma insulin (CVs <8%), C-peptide (CVs <9%), glucagon (CVs <8%), and growth hormone (CVs <7%) were measured by radioimmunoassay (RIA) (insulin: Pharmacia-Upjohn, Uppsala, Sweden; C-peptide: CIS, Gif-Sur-Yvette, France; glucagon: Linco Research, St. Charles, MI; growth hormone: Sorin Biomedica, Saluggia, Italy). Plasma cortisol (CVs <6%) was determined after extraction and charcoal-dextran separation by RIA (23). Plasma epinephrine and norepinephrine (CVs < 5%) were analyzed by reverse-phase HPLC using extraction tubes (ESA, Chelmsford, MA) for the isolation procedure (24).
Gas chromatography-mass spectrometry.
After deproteinization with Ba(OH)2/ZnSO4, plasma glucose was derivatized as its penta-acetate and analyzed on a Hewlett-Packard 5890 gas chromatograph equipped with a CP-Sil5 25 m × 0.25 mm × 0.12 μm capillary column (Chrompack, Middelburg, the Netherlands) interfaced to a Hewlett-Packard 5971A mass-selective detector operating in the electron impact ionization (25). Grass chromatography conditions were 80°C for 30 s, followed by an increase of 30°C/min to 140°C, an increase of 10°C/min to 230°C, and an increase of 5°C/min to 255°C. Ions with m/z 187 and 189 were selectively monitored for 2H enrichment in C3-C6.
Liquid chromatography tandem mass spectrometry.
After deproteinization with methanol/H2O, plasma acetaminophen-glucuronide 13C MPE was determined by liquid chromatography, and tandem mass spectrometry (26) was determined on a Perkin-Elmer Sciex API 3000 Electronspray Interface Tandem Mass Spectrometer (Perkin-Elmer, Foster City, CA) interfaced with a Hewlett-Packard 1090 liquid chromatograph (Hewlett-Packard, Palo Alto, CA). Multiple-reaction monitoring was adapted in such a way that acetaminophen glucuronide and its 13C isotopic form were monitored through the transition from precursor to product ion (m/z 326–113 and m/z 327–114, respectively) in the negative ion mode. Liquid chromatography conditions were 50:50 isocratic of 5% methanol/10 mmol/l ammonia acetate and 95% methanol/10 mmol/l ammonia acetate, with a flow rate of 200 μl/min. A sample with natural abundance 13C was included with each batch. Peak areas were taken for M = 326/113 and (M+1) = 327/114, and the 13C MPE in the sample was calculated as follows: 1/[1 + (M+1)/Msample + (M = 1)/Mnatural].
In vivo 13C NMR spectroscopy.
Liver glycogen concentrations were measured in all subjects with natural abundance 13C NMR spectroscopy, i.e., without administration of 13C-enriched glucose. The peak intensity of the C1 resonance of the glucosyl units of hepatic glycogen was quantified with a double-tuned 1H (125.6 MHz) and 13C (31.5 MHz) 10-cm circular coil on a 3 T Medspec 30/80-DBX system (Bruker Medical, Ettlingen, Germany) (7). Spectra were acquired using a modified 1D-ISIS sequence (27) without proton decoupling. Two blocks of measurements (5,000 scans each) were combined for determination of hepatic glycogen concentrations at a single time point. Hepatic glycogen was quantified by integration of the C1 glycogen doublet at 100.5 ppm using the same frequency bandwidth for all spectra (±300 Hz). Individual rates of net glycogen synthesis and net glycogen breakdown were calculated from linear regression of the glycogen concentration time curves between 7:30 p.m. and 10:30 p.m. and from 10:30 p.m. to 8:00 a.m., respectively.
Liver volumes were measured in a 1.5 T Vision imager (Siemens, Erlangen Germany) using a body array coil and in-phase and postphase multislice FLASH imaging sequences (7).
Calculations and statistics
Rates of EGP were calculated as the tracer infusion rate divided by the mean 2H MPEs of plasma glucose at steady-state conditions (28) and are given as μmol · kg body wt−1 · min−1.
Net hepatic glycogen synthesis, glycogenolysis, and gluconeogenesis.
Rates of net hepatic glycogen synthesis and glycogenolysis (glycogen breakdown) were calculated from the best linear fit of the liver glycogen time curve by the method of least squares (7,27). Rates of gluconeogenesis are given as the difference between rates of EGP and net glycogenolysis. All data were normalized to liver volume as well as body weight and are given as μmol · kg body wt−1 · min−1.
Pathways of glycogen synthesis.
The percent contribution of the direct pathway to hepatic glycogen synthesis (% direct pathway) is represented by the fraction of UDP-glucose formed by the direct pathway, as determined from the 13C MPEs in plasma glucose and acetaminophen-glucuronide using the formula (12,29): % direct pathway = [(C1 MPE-C6 MPE) glucuronide]/[(C1 MPE-C6 MPE) glucose].
Data are presented as means ± SE. One-way analysis of variance with Bartlett’s test for equal variances and post hoc testing by the Newman-Keuls test was used for statistical comparisons within the different groups. In addition, data of type 1 diabetic and nondiabetic subjects were compared using unpaired Student’s t test. Statistical significance was considered at P < 0.05. All calculations were done using the Sigma Stat software package (Jandel, San Rafael, CA).
Overnight intravenous insulin infusion normalized fasting plasma glucose concentration in type 1 diabetic patients at 8:00 a.m. (Fig. 1B). Throughout day 2, the variable insulin infusion resulted in close approximation of plasma glucose concentrations to those of the nondiabetic subjects, which were only slightly higher in diabetic patients during ingestion of mixed meals (control subjects 5.9 ± 0.1 mmol/l vs. type 1 diabetic patients 6.2 ± 0.1 mmol/l; P < 0.05). Plasma FFA concentrations before (control subjects 233 ± 68 μmol/l vs. type 1 diabetic patients 217 ± 49 μmol/l; P = NS) and after (control subjects 73 ± 11 μmol/l vs. type 1 diabetic patients 48 ± 14 μmol/l; P = NS) dinner were not different between diabetic and healthy subjects. During the night, plasma FFAs increased similarly in both groups (7:00 a.m. on day 3: control subjects 294 ± 56 μmol/l vs. type 1 diabetic patients 399 ± 112 μmol/l; P = NS).
The time course of plasma hormone concentrations, as monitored immediately before and after dinner on day 2, is depicted in Fig. 2. The variable insulin infusion increased plasma insulin concentrations in type 1 diabetic patients (Fig. 2A). Fasting plasma C-peptide concentrations were below the detection limit of 0.25 ng/ml in six type 1 diabetic patients and 1.1 ng/ml in one type 1 diabetic patient, respectively. After ingestion of the mixed meal, plasma C-peptide increased in control subjects but did not change in type 1 diabetic patients (P < 0.0001). Postprandial plasma glucagon concentrations were slightly lower in type 1 diabetic patients but exhibited a similar time course, with peak concentrations at 2 h (P < 0.05 vs. basal) after dinner in both groups (Fig. 2B). Plasma growth hormone and cortisol concentrations declined within ∼2 and ∼8 h, respectively (P < 0.05 vs. basal), but were not different between the groups (Figs. 2C and D). Plasma concentrations of epinephrine (5:00 p.m. control subjects 116 ± 28 pmol/l vs. type 1 diabetic patients 79 ± 22 pmol/l; 1:00 a.m. 82 ± 19 vs. 46 ± 4 pmol/l; 7:00 a.m. 67 ± 24 vs. 84 ± 23 pmol/l; P = NS) and norepinephrine (5:00 pm control subjects 1.32 ± 0.33 nmol/l vs. type 1 diabetic patients 1.78 ± 0.47 nmol/l; 1:00 a.m.: 1.15 ± 0.17 vs. 1.43 ± 0.56 nmol/l; 7:00 a.m. 0.77 ± 0.15 vs. 1.63 ± 0.44 nmol/l; P = NS) remained unchanged throughout the study periods and again did not differ between the groups.
After dinner, hepatic glycogen concentrations increased linearly from 8:00 p.m. to ∼10:30 p.m. (control subjects 8:00 p.m. 286 ± 9 mmol/l vs. 10:30 p.m. 322 ± 7 mmol/l; P < 0.001; type 1 diabetic patients 8:00 p.m. 267 ± 15 mmol/l vs. 10:30 p.m. 308 ± 16 mmol/l; P < 0.001), without any difference between the groups. Rates of glycogen synthesis were comparable between control subjects and type 1 diabetic patients (Fig. 3A). During the subsequent overnight fast, hepatic glycogen concentrations similarly declined in both groups (control subjects: 3:00 a.m. 271 ± 5 mmol/l and 7:00 a.m. 233 ± 7 mmol/l; P < 0.001 vs. 10:30 p.m.; type 1 diabetic patients 3:00 a.m. 264 ± 14 mmol/l and 7:00 a.m. 211 ± 10 mmol/l; P < 0.001 vs. 10:30 p.m.). Hepatic glycogen concentrations as well as rates of glycogen breakdown did not differ between control subjects and type 1 diabetic patients (Fig. 3B). Mean liver volume was almost identical in both groups (control subjects 1,608 ± 70 cm3 vs. type 1 diabetic patients 1,585 ± 45 cm3; P = NS). Rates of EGP were also similar in both groups (control subjects 8.84 ± 0.18 μmol · kg−1 · min−1 vs. type 1 diabetic patients 8.68 ± 0.28 μmol · kg−1 · min−1) (Fig. 3B), resulting in comparable rates of gluconeogenesis (control subjects 4.56 ± 0.59 μmol · kg−1 · min−1 vs. type 1 diabetic patients 4.50 ± 0.25 μmol · kg−1 · min−1).
The percent contribution of the flux through the direct pathway for hepatic glycogen synthesis, as determined from 13C enrichments in acetoaminophen-glucuronide after breakfast, was markedly reduced in type 1 diabetic patients (control subjects 49 ± 7% vs. type 1 diabetic patients 30 ± 8%; P < 0.05) (Fig. 4). Accordingly, the contribution of the indirect pathway from glucose precursors to glycogen synthesis was ∼51 and ∼70%, respectively.
This study shows that both long- and short-term near normoglycemia, resulting from tight metabolic control, are required to normalize hepatic glycogen metabolism in type 1 diabetic patients under physiologic conditions of mixed meal ingestion. Whereas short-term near normoglycemia for 24 h only partially ameliorates the derangement of glycogen metabolism in moderately controlled type 1 diabetic subjects (7), combined long- and short-term near normoglycemia resulted in peak hepatic glycogen concentrations as well as net rates of hepatic glycogen synthesis and breakdown in type 1 diabetic patients that were identical to those of nondiabetic humans (5,30,31,32). Consequently, EGP was also normal in those well-controlled type 1 diabetic patients, consistent with results of previous studies (18,19).
Until now, it was not resolved whether the abnormalities of hepatic glycogen metabolism in type 1 diabetes result from intrinsic factors or simply from relative portal vein hypoinsulinemia and/or hyperglucagonemia present under “normal life” conditions. Previously, net glycogen synthesis could only be normalized in type 1 diabetic subjects using hyperinsulinemic-hyperglycemic clamp tests along with suppression of glucagon secretion by somatostatin (6). Similarly, intraduodenal glucose infusion combined with somatostatin-insulin-glucagon clamp tests for 4 h also gave comparable UDP-glucose flux rates, as measured with labeled galactose in healthy and type 1 diabetic subjects (13). Although this method does not quantify hepatic glycogen synthetic rates (33), it is consistent with previous findings that hepatic glycogen synthesis can be corrected under such artificial conditions, which maximally stimulate glycogen synthesis and simultaneously inhibit glycogen breakdown.
Intrinsic factors such as protein phosphorylase-1 (PP-1) are important modulators of glycogen metabolism. PP-1 phosphorylates and thereby activates glycogen synthase and is a key factor in regulation of this enzyme (34,35). Livers of insulin-deficient rats only insufficiently express PP-1, and 96 rather than 24 h of insulin substitution are necessary to restore the deficient production of the PP-1 glycogen-binding domain (36). It is tempting to assume that due to a similar mechanism, only long-term near normoglycemia can normalize glycogen synthesis in type 1 diabetic patients.
Portal vein concentrations of insulin and glucagon are key factors regulating hepatic glycogen metabolism (8,12,37). Assuming a portal venous-to-systemic insulin gradient of 2.4 (38), postprandial portal insulin concentrations were ∼twofold higher in the type 1 diabetic patients than in the nondiabetic subjects (723 ± 96 vs. 1,634 ± 548 pmol/l; P < 0.001). On the other hand, postprandial plasma glucagon concentrations were lower, possibly resulting from suppressed glucagon secretion induced by the high degree of insulinemia present in the type 1 diabetic subjects (39). Thus, the normalization of hepatic glycogen metabolism in type 1 diabetic patients was only achieved by an increased portal vein insulin-to-glucagon ratio that favors glycogen synthesis and reduces glycogenolysis (8,12). This implies that even combined long- and short-term near normoglycemia still does not completely correct the impaired insulin sensitivity in the liver of type 1 diabetic patients.
Likewise, despite strenuous efforts to normalize glycemia, the contribution of the direct pathway to glycogen synthesis remained reduced in our type 1 diabetic patients. Such a defect was not only detected in moderately controlled type 1 diabetic subjects (5) during mixed meal ingestion but also under insulin clamp conditions when portal vein concentrations of insulin and glucagon were matched to those of nondiabetic subjects (6). It is not yet clear why this defect persists even in well-controlled type 1 diabetic patients, but animal models hint at impaired glucokinase acitivity (31,40) or increased contribution of the indirect gluconeogenic pathway due to increased phosphoenolpyruvate carboxykinase (PEPCK) activity (41,42). On the other hand, type 1 diabetic patients may present with peripheral insulin resistance (16,18), which could augment the availability of gluconeogenic precursors, thus increasing gluconeogenesis and stimulating the flux through the indirect pathway of glycogen synthesis. It is noteworthy that 2 weeks of improved metabolic control were sufficient to normalize postprandial supression of EGP in type 1 diabetic patients (43).
With regard to clinical practice, our results demonstrate that if the patient has excellent long-term metabolic control, portal vein insulin-to-glucagon ratios, which ensure sufficient postprandial net glycogen accumulation despite the unphysiologic peripheral route of insulin delivery, can be achieved. This is important because even the intensive insulin regimens that are currently being used for type 1 diabetes rely on such peripheral insulin administration. Restoring postprandial hepatic glycogen synthesis in type 1 diabetes will ameliorate the glucose peaks and the prolonged hyperglycemia after ingestion of mixed meals. Furthermore, normalization of hepatic glycogen stores could help to prevent severe hypoglycemic episodes because glucagon-mediated glycogenolysis, which is the first-line defense against hypoglycemia (44), depends on the availability of sufficient liver glycogen stores.
In addition, the present study found that combined long- and short-term tight metabolic control normalized hepatic glycogenolysis and EGP. This can be accounted for by similar or only slightly different plasma concentrations of major glycogenolytic and gluconeogenetic hormones (44,45,46) and by comparable peak postprandial glycogen concentrations that strongly correlate with net hepatic glycogenolysis (21). Because natural abundance 13C NMR spectroscopy determines net hepatic glycogen fluxes, an alteration in glycogen cycling, resulting from simultaneous changes in the activities of glycogen synthase and phosphorylase, cannot be completely ruled out but is unlikely given nearly identical net glycogen synthesis and breakdown rates in type 1 diabetic and nondiabetic subjects.
In conclusion, tight long- and short-term metabolic control using peripheral insulin substitution normalized 1) hepatic glycogen synthesis after ingestion of a mixed meal dinner, 2) hepatic glycogenolysis, and 3) subsequent nocturnal EGP fasting in type 1 diabetic subjects. However, the contribution of the indirect (gluconeogenic) pathway of glycogen synthesis remained increased, indicating augmented gluconeogenesis in type 1 diabetic patients, even after combined long- and short-term near normoglycemia.
This study was supported by grants from the Austrian Science Foundation (Fonds zur Förderung der Wissenschaftlichen Forschung: P13213-MOB and P13722-MED) (to M.R.) and the National Institutes of Health (R01DK49230 and P30DK34576) (to G.I.S.) and by an institutional grant from Novo Nordisk to (W.W.).
We gratefully acknowledge the excellent assistance of A. Hofer, H. Lentner, the laboratory staff of the Division of Endocrinology and Metabolism, and the radiological technical assistants of the Department of Radiology. Thanks also to Prof. E. Moser, PhD, Institute of Medical Physics, University of Vienna, for cooperation and support.
Address correspondence and reprint requests to Michael Roden, MD, Division of Endocrinology and Metabolism, Department of Internal Medicine III, University of Vienna Medical School, Waehringer Guertel 18-20, A-1090 Vienna, Austria. E-mail: firstname.lastname@example.org.
Received for publication 24 July 2001 and accepted in revised form 25 October 2001.
M.G.B. and E.B. contributed equally to this study.
CV, coefficient of variation; EGP, endogenous glucose production; FFA, free fatty acid; HPLC, high-performance liquid chromatography; MPE, mol percent enrichment; NMR, nuclear magnetic resonance; PP-1, protein phosphorylase-1; RIA, radioimmunoassay.