INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

DURING EXERCISE in the postabsorptive state, the release of glucose to the blood is almost exclusively caused by an increase in hepatic glucose production (1, 20). Two different methods have been widely used to estimate the hepatic glucose production. Measurement of arterial and hepatic venous concentrations of glucose in combination with determination of splanchnic blood flow (SBF)the so-called arteriohepatic venous (a-hv) balance techniqueallows for determination of net splanchnic glucose production (1, 20). However, because glucose is metabolized in the gut, the glucose concentration in portal blood may be different from arterial glucose concentrations. Therefore, determination of net hepatic glucose release would require sampling of portal venous blood, and this is not possible in humans. The requirement of hepatic vein catherization may limit the use of the method in many laboratories.

A less invasive alternative is the glucose tracer-dilution method. Whole body appearance of glucose [or rate of appearance (R_a)] can be calculated from the dilution measured in arterial blood of glucose that is labeled (by radioactive or stable isotopes) infused into a peripheral vein. Given that the tracer mimics the behavior of endogenous glucose and that the label does not recycle, if a steady-state condition is achieved, this method allows for accurate determination of glucose R_a in the resting state (9). However, gross glucose release from all tissues is measured by this means, and, during some conditions, the kidneys may play a quantitative role in glucose production (19). Therefore the method may overestimate net hepatic glucose production. In contrast, underestimation may occur if recycling of label takes place (17). Equations have been constructed to calculate glucose R_a under non-steady-state conditions, and these equations have been found to be valid during various experimental conditions (13, 17, 18). However, the accuracy of such calculations is still debated (3) and must be particularly doubtful in conditions, such as physical exercise, during which the size and physiological characteristics of the various body compartments change (17).

The a-hv balance technique and the tracer-dilution method have been applied simultaneously in dogs that were fasted overnight and were running at an intensity corresponding to 50% maximal O₂ uptake (O_{2 max}) (21). While the animals were in the resting state, overall R_a that was determined with the latter method tended to be slightly higher than net splanchnic glucose production that was determined by the former technique. During prolonged exercise, results obtained with the two methods were comparable. However, differences in metabolism and circulation between dogs and humans make it impossible to predict the outcome of a similar evaluation in humans. Because the tracer-dilution method has become widely used in human exercise studies, a comparison, performed in exercising humans, between the two methods available for estimation of hepatic glucose production is much warranted (4). In the present study, such a comparison was carried out at both moderate and heavy exercise.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects. Eight healthy young men [mean age, 25 yr (range, 23-30 yr); height, 182 cm (range, 174-190 cm); weight, 79 kg (range, 65-97 kg); and O_{2 max}, 50 ml · kg¹ · min¹ (range, 46-60 ml · kg¹ · min¹)] gave informed written consent to participate in a study approved by the Ethics Committee of Copenhagen. The O_{2 max} for each individual was determined 4-8 days before the experimental study by using a protocol of incremental workloads (2 min at each workload) to exhaustion on a modified electromagnetically braked Krogh cycle ergometer. While the subjects cycled, the upper part of the body formed an angle of 45° with the horizontal plane. O_{2 max} was the highest O₂ consumption (O₂) attained during the latter stages of the test and was accompanied by a respiratory exchange ratio of >1.1, a heart rate close to the age-predicted maximum (220 beats/min  age in yr), and a leveling-off phenomenon of O₂. None of the subjects was taking medications or had a history of endocrine disease. All subjects refrained from training, smoking, and drinking alcohol the day before the experiment.

Procedures. Subjects arrived at the laboratory at 800, after an overnight fast. For the purpose of taking blood samples, a cannula was inserted in the radial artery of the nondominant arm of each subject. Two separate venous lines were inserted in two forearm veins for infusion of radiolabeled glucose and indocyanine green (ICG) dye, respectively. In all subjects, a catheter was introduced into a femoral vein and was advanced, under fluoroscopy, into a right-sided hepatic vein to ~3-4 cm from the wedge position. Its location was verified, both during and after exercise, by using ultrasonography and fluoroscopy, respectively. Patency of the catheter was maintained by flushing with heparinized isotonic saline solution (10 U/ml).

SBF and net glucose output. The splanchnic plasma flow (SPF) was estimated by the ICG dye-excretion method (14). This technique involves a primed (1,000 µg), constant (200 µg/min) infusion of ICG (prepared in a 5% solution of human serum albumin in isotonic saline), with an equilibration period of 45 min before blood sampling begins. Arterial and hepatic venous blood was sampled every 10 min, starting 30 min before exercise. Plasma concentrations of ICG were determined spectrophotometrically (805 nm) in duplicate, with correction for plasma turbidity measured at 900 nm, and hematocrit (Hct) was measured with the microhematocrit method. Plasma glucose concentrations were immediately determined with an automatic glucose analyzer (YSI 23AM; Yellow Springs Instruments) and were converted to whole blood concentrations on the basis of a-hv Hct values determined throughout the experiment. SPF was estimated according to a modification for non-steady-state conditions where I is the infusion rate of ICG, t is the time interval between time 1 and time 2, V_d is the volume of distribution of ICG approximated to 5% of body weight (14), and C_a and C_v are the arterial and hepatic venous plasma concentrations of ICG, respectively. Thus SBF = SPF/(1  Hct). Net splanchnic glucose output was calculated as the product of SBF and a-hv glucose-concentration differences.

Glucose turnover. Glucose R_a was determined by a primed (24 µCi) continuous (0.24 µCi/min) infusion of [3-³H]glucose [Amersham; pharmaceutically prepared and dissolved in isotonic saline (2.0 µCi/ml) by Isotopapoteket, Copenhagen, Denmark] begun at 120 min and continued during the exercise period. During the 2-h equilibration period, subjects rested in the supine or in the semisupine position (last 30 min). From measurements of plasma glucose concentration and specific activity of the radiolabeled glucose in the last 30 min of the resting period, individual distribution volume of glucose (glucose pool/plasma glucose concentration) was calculated according to Hetenyi and Norwich (7). R_a was calculated by using the formulas for non-steady-state conditions (13, 18). These calculations are based on a modified single-compartment analysis that assumes that the pool fraction in which rapid changes in concentration and specific activity of glucose take place is 0.65 (18). The calculated mean glucose distribution space of our subjects was 23.3 ± 1.1% of body weight. For the assay of [3-³H]glucose radioactivity, 2-ml plasma samples were deproteinized with 1 ml perchloric acid and centrifuged. Triplicate aliquots (500 µl) of the supernatant were evaporated overnight under a stream of air to remove tritiated water. The dry residue was redissolved in 1 ml of water and counted in 9 ml of Lumagel (Lumac) in a liquid scintillation spectrophotometer. Correction for counting efficiency was always carried out by means of dilutions of the infusate, with plasma run in parallel with plasma samples.

Statistical analysis. Statistical analysis was performed by using nonparametric tests. Overall differences between curves obtained in the two types of experiments were evaluated with the Kruskal-Wallis test. Friedman's test was used to evaluate whether changes became manifest with time, and such changes were located with Wilcoxon's rank-sum test for paired data (15). A 95% level of confidence (two-tailed testing) was accepted for all comparisons. All data are reported as means ± SE.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects completed 40 min of cycling at 50.4 ± 1.5% O_{2 max}, followed by 30 min of cycling at 69.0 ± 2.2% O_{2 max}, and heart rate increased from 59 ± 3 beats/min at rest to 138 ± 3 beats/min (50% O_{2 max}) and 175 ± 3 beats/min (~70% O_{2 max}) during exercise. The glucose production was similar at rest, whether determined by the a-hv balance technique or by the tracer-dilution method (Fig. 1; P > 0.05). During moderate exercise (50% O_{2 max}), values almost doubled, yet no difference was found between methods (Fig. 1). Similarly, during intense exercise (~70% O_{2 max}), during which glucose production rose to 250-300% of initial resting values, no statistically significant difference was obtained between the two methods (Fig. 1; P > 0.05). Arterial glucose concentration remained stable throughout the experiment (5.08 ± 0.03 mmol/l) except at the very end of exercise, when it decreased slightly but significantly compared with rest [4.79 ± 0.11 mmol/l (70 min), P < 0.05; Fig. 2]. The a-hv glucose difference increased from 0.77 ± 0.06 mmol/l at rest to 1.53 ± 0.11 and 4.29 ± 0.57 mmol/l during moderate and intense exercise, respectively. The glucose specific activity was constant for the last 30 min preceding the exercise session and decreased by 12 ± 3% during low-intensity exercise in arterial blood. During intense exercise by subjects, the fall in glucose specific activity in arterial blood was more pronounced, and values were 70 ± 6% of the initial values at the end of that exercise period (Fig. 3). Glucose specific activity in arterial blood became statistically lower than the resting values were after 50 min of exercise and in hepatic venous blood after 30 min of exercise (Fig. 3). Although concentration of radioactivity (Beq/ml blood) tended to be ~10% higher in arterial compared with hepatic venous blood, neither at rest nor during exercise was this difference statistically significant (data not shown). SBF was calculated to be 1.56 ± 0.11 l/min at rest and decreased during exercise to 1.39 ± 0.12 l/min (50% O_{2 max}, P > 0.05 vs. rest) and 0.87 ± 0.19 l/min (69% O_{2 max}, P < 0.05 vs. rest; Fig. 4). Hct values rose from 45.0 ± 1.4% at rest to 45.8 ± 1.3 and 47.7 ± 1.5% during moderate and intense exercise, respectively.

View larger version (19K): [in this window] [in a new window]   Fig. 1.   Effect of exercise at 50% maximal O2 uptake (O2 max) (0-40 min) and at 70% O2 max (40-70 min) on splanchnic glucose output [, arteriohepatic venous (a-hv) difference] and tracer-determined glucose production (, radioisotopic). Values are means and SE; n = 8 healthy young men. * Time points when values are significantly elevated vs. rest; P < 0.05.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Plasma glucose concentrations in arterial (, solid line) and hepatic venous (, dotted line) blood at rest and during exercise at 50% O2 max (0-40 min) and at 70% O2 max (40-70 min) in 8 males. Values are means and SE; n = 8 men. * Value below basal resting value, P < 0.05; § values significantly above basal resting value, P < 0.05.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   Specific activity of radiolabeled [3-3H]glucose in both arterial ()and hepatic venous () plasma at rest (30 to 0 min) and during semisupine cycling at 50% O2 max (0-40 min) and at 70% O2 max (40-70 min). Values are means and SE; n = 8 men. * Time points that are significantly decreased compared with resting values, P < 0.05.

View larger version (12K): [in this window] [in a new window]   Fig. 4.   Estimated splanchnic blood flow (ESBF) determined at rest and during exercise at 50 and 70% O2 max by continuous infusion of indocyanine green. Values are means and SE; n = 8 men. * Values significantly different from rest, P < 0.05.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

An important finding in the present study of humans is that, whether determined by the use of an isotope-dilution method or by an a-hv balance technique, the same glucose production is found during exercise. Radiolabeled [³H]glucose was administered by primed and constant infusion to determine whole body glucose R_a, and, simultaneously, in the same individuals, splanchnic glucose production was determined from a-hv sampling and measurement of SBF by ICG dye infusion. To our knowledge, a comparison between these two methods has not been performed previously in humans while they exercised.

Previously, the a-hv balance and isotopic-tracer methods have been compared only indirectly by examination of the literature (4). From this comparison, it has been suggested that R_a may underestimate splanchnic glucose release during exercise. This could be due to inability of the pool fraction to model adequately the model glucose turnover during non-steady-state conditions (3, 18). Other possibilities are an increase in the total V_d for glucose with the transition from rest to exercise (21), isotope discrimination between labeled and unlabeled glucose that influences tissue uptake (2, 6), and potential incorporation of labeled glucose into hepatic glycogen at rest and subsequent re-release during exercise (17). In the present study, a primed, constant infusion of tracer was chosen. Other techniques have been used that employed variable tracer-infusion rates to avoid concerns in regard to lack of tracer equilibration between body pools. However, it has been debated how large an advantage one gets by use of that approach. Furthermore, to estimate how tracer infusion should be changed with the transition from rest to exercise, additional studies would be required for each individual.

A good agreement between the arterio-venous balance and the isotopic tracer methods has been found in the resting, postabsorptive state in humans (5). Only one previous study, performed in dogs, has directly compared the two methods during exercise (21). As in the present study, neither at rest nor during prolonged exercise was it possible to demonstrate that tracer-determined R_a was significantly different from the net splanchnic glucose output as determined by the a-hv method (21). In the dog study, sampling of portal venous blood revealed that glucose utilization by gut and liver was constant during rest and exercise. When splanchnic glucose uptake was added to net splanchnic glucose output, the derived absolute splanchnic glucose output turned out to be higher than tracer-determined R_a during the first part of the exercise period. Ideally, R_a should equal absolute splanchnic glucose output, and the underestimation by the tracer method was ascribed to a possible increase in effective V_d for glucose. In the present human study, portal blood was not obtained, but from concentration or glucose tracer in both arterial and hepatic venous blood, no statistically significant difference was observed between these parameters, either at rest or during exercise. On the basis of the present findings, splanchnic glucose uptake seems to be minimal. The finding, both in the dog study (21) and in the present study of humans, that overall tracer-determined R_a seems to agree with net splanchnic glucose production during exercise is satisfactory from a practical point of view, but it also shows that assumptions about the tracer technique cannot be completely fulfilled.

The ICG method for determination of SBF is subject to experimental problems. During an ideal steady state, SPF can be determined, according to Fick's principle, with constant infusion of ICG and measurement of arterial and hepatic venous concentrations. During the non-steady state, a correction for the change in the amount of ICG in the body is necessary (16). The correction used in the present experiments is based on the assumption that ICG distributes only in plasma. This assumption has been questioned. On the basis of findings in pigs (11, 12), ICG also seems to be distributed in the extravascular space. Using different kinetic models for the redistribution of ICG, Ott (10) calculated that the SPF may be underestimated by up to 33% or overestimated by up to 45%, depending on the model and the length of the equilibrium period. However, during a reasonable steady state, i.e., when the arterial ICG concentration changes <5%/h, the main source of variation is the variation in ICG concentrations between different hepatic veins, which may be on the order of 8-11% (22). Arteriovenous difference techniques always face a problem when used in a non-steady-state situation, and, in the present case, transit time across the liver might interfere with the results if the non-steady state is pronounced. However, although no absolute steady state was found during exercise in the present study, at least during moderate workloads, metabolic changes occurred so slowly that one can assume a steady state over the period when blood samples were taken.

In conclusion, during exercise in humans, determination of glucose production can be performed equally well, whether one uses a primed, constant infusion of radiolabeled [³H]glucose, together with the pool-fraction model for non-steady-state conditions to determine glucose R_a, or one uses the a-hv balance technique, including blood flow determinations with infusion of ICG, to quantitate splanchnic glucose release.