REVIEW ARTICLE
Sports Medicine II (2): 102-124. 1991 0112-1642/ 91 / 0002-0 I02/ $1 1.50/ 0 © Adis Intt' rnational Limited. All rights reserved. SP01353
Plasma Glucose Metabolism During Exercise in Humans Andrew R. Coggan Exercise Physiology Laboratory, School of Health, Physical Education, and Recreation, The Ohio State University, Columbus. Ohio, USA
Contents 102 104 104 104 105
105 105 /10
I/O I/O 1/1 1/2 1/3 1/3 114 115 116 116 117 117 118
1/9 120
Summary
Summary I. Methods for Studying Glucose Metabolism During Exercise 1.1 Arteriovenous Balance Method 1.2 Isotopic Tracer Method 1.3 Comparison of the Arteriovenous Balance and Isotopic Tracer Methods 2. Glucose Utilisation During Exercise 2.1 Effect of Exercise Intensity and Duration 2.2 Effect of Exercise Mode/Active Muscle Mass 2.3 Effect of Other Substrates 2.3.1 Muscle Glycogen 2.3.2 Plasma Free Fatty Acids 2.4 Effect of Diet 2.5 Effect of Exercise Training 2.5.1 During Exercise at the Same Absolute Intensity 2.5.2 During Exercise at the Same Relative Intensity 3. Glucose Production During Exercise 3.1 Hepatic Glycogenolysis and Gluconeogenesis 3. Ll Effect of Exercise Intensity and Duration 3. L2 Effect of Exercise Mode/ Acti ve Muscle Mass 3. 1.3 Effect of Diet 3. 1.4 Effect of Exercise Training 3.2 Hormonal Regulation of Hepatic Glucose Production 4. Conclusions
Plasma glucose is an important energy source in exercising humans, supplying between 20 and 50% of the total oxidative energy production and between 25 and 100% of the total carbohydrate oxidised during submaximal exercise. Plasma glucose utilisation increases with the intensity of exercise, due to an increase in glucose utilisation by each active muscle fibre, an increase in the number of active muscle fibres, or both. Plasma glucose utilisation also increases with the duration of exercise, thereby partially compensating for the progressive decrease in muscle glycogen concentration. When compared at the same absolute exercise intensity (i.e. the same V02), reliance on plasma glucose is also greater during exercise performed with a small muscle mass, i.e. with the arms or just I leg. This may be due to differences in the relative exercise
Glucose Metabolism During Exercise
103
intensity (i.e. the %V02peak), or due to differences between the arms and legs in their fitness for aerobic activity. The rate of plasma glucose utilisation is decreased when plasma free fatty acid or muscle glycogen concentrations are very high, effects which are probably mediated by increases in muscle glucose-6-phosphate concentration. However, glucose utilisation is also reduced during exercise following a low carbohydrate diet, despite the fact that muscle glycogen is also often lower. When exercise is performed at the same absolute intensity before and after endurance training, plasma glucose utilisation is lower in the trained state. During exercise performed at the same relative intensity, however, glucose utilisation may be lower, the same, or actually higher in trained than in untrained subjects, because ofthe greater absolute V02 and demand for substrate in trained subjects during exercise at a given relative exercise intensity. Although both hyperglycaemia and hypoglycaemia may occur during exercise, plasma glucose concentration usually remains relatively constant. Factors which increase or decrease the reliance of peripheral tissues on plasma glucose during exercise are therefore generally accompanied by quantitatively similar increases or decreases in glucose production. These changes in total glucose production are mediated by changes in both hepatic glycogenolysis and hepatic gluconeogenesis. Glycogenolysis dominates under most conditions, and is greatest early in exercise, during high intensity exercise, or when dietary carbohydrate intake is high. The rate of gluconeogenesis is increased when exercise is prolonged, preceded by a restricted carbohydrate intake, or performed with the arms. Both glycogenolysis and gluconeogenesis appear to be decreased by endurance exercise training. These effects are due to changes in both the hormonal milieu and in the availability of hepatic glycogen and gluconeogenic precursors. Hepatic glucose production during exercise is stimulated by glucagon and the catecholamines and suppressed by insulin or an increase in plasma glucose concentration. In contrast to earlier suggestions, it appears that a decrease in insulin and an increase in glucagon are both required for hepatic glucose production to increase normally during moderate intensity, moderate duration (40 to 60 minutes) exercise. Changes in the catecholamines. however, may still prove to be important, especially during more intense or more prolonged exercise.
Both fats and carbohydrates are oxidised by skeletal muscle to provide energy during exercise (Krogh & Lindhard 1920). Although the fat stores of the body are quite large, even in lean people, the body's total carbohydrate stores are limited, amounting to some ".,2000 to 2500mmol glucosyl units (Christensen & Hansen 1939; Hedman 1957). Of this total, liver glycogen in the postabsorptive state may contribute ".,400 to 550mmol (Hultman 1977), with circulating glucose in body fluids constituting another ".,90mmol (DeFonzo et a1. 1979). Thus, on the basis of mass alone, it is apparent that glucose transported through the plasma represents about one-fourth of the carbohydrate available to provide fuel for skeletal muscle during exercise. When one considers that only glycogen within the active muscles is readily accessible as a fuel during exercise (since muscle lacks g1ucose-6-phosphatase), and that the liver can also synthesise glucose from 3-carbon intermediates such as lactate, the
potential importance of plasma-borne glucose to carbohydrate energy production becomes even greater. Indeed, at times plasma glucose supplies 70 to 100% of the carbohydrate oxidised during exercise (Ahlborg et a1. 1974; Broberg & Sahlin 1989; Wahren et a1. 1971), even at intensities as high as ".,75% Y02peak (Coggan & Coyle 1987) [Y02max, maximal oxygen uptake measured during treadmill running; Y02peak, peak oxygen uptake measured during cycle ergometer exercise]. The rate of plasma glucose utilisation during exercise is dependent upon both the intensity and duration of the exercise bout, as well as upon the mode of exercise itself. In addition, plasma glucose utilisation during exercise may be influenced by the availability of other substrates and by the pre-exercise diet and training status of the individual. The present review considers these factors, as well as the role of the liver in supplying glucose during exercise. The emphasis is placed upon studies of
104
whole-body exercise in normal individuals. For information regarding the effects of exercise on skeletal muscle glucose transport, or on the effects of exercise on plasma glucose metabolism in disease state such as diabetes, readers are referred to other recent reviews (Bjorkman 1986; Ivy 1987; Wasserman & Vranic 1986).
1. Methods for Studying Glucose Metabolism During Exercise 1.1 Arteriovenous Balance Method
Glucose metabolism during exercise has traditionally been studied by catheterising both the arterial blood supply and the venous drainage of a given tissue bed, thereby allowing measurement of the arteriovenous glucose difference and the rate of blood flow. While the product of these 2 parameters (i.e. the net arteriovenous glucose balance) does not indicate the ultimate fate (or source) of the glucose taken up (or released), it does indicate the maximum possible contribution of glucose to (or from) the metabolic processes in the tissue bed under study. The arteriovenous balance method provides the ability to examine glucose metabolism in specific tissue beds during exercise. For example, this technique has been used to compare glucose uptake by active vs nonactive limbs (Ahlborg et al. 1975), or to quantify the contributions of renal and hepatic tissues to overall glucose release (Wahren et al. 1971). However, because it is usually not possible to catheterise the portal vein in humans, it is actually splanchnic (rather than hepatic) glucose release that is normally measured, resulting in a 5 to 10% underestimation of the true rate of hepatic glucose release (Wahren et al. 1971; Wasserman et al. 1987). Net splanchnic release is usually the parameter of interest, however, as it indicates the rate at which glucose is being made available for metabolism in peripheral tissues. The arteriovenous balance technique has several disadvantages, perhaps the greatest of which is its invasive nature. As indicated above, simple measurement of the arteriovenous balance also does not provide any indication of the metabolic fate
Sports Medicine II (2) 1991
(i.e. oxidation, glycogen synthesis, lactate production, etc.) or source (i.e. glycogenolysis, gluconeogenesis) of the glucose that is exchanged during exercise. In addition, the arteriovenous glucose difference across exercising muscle is extremely small (0.1 to 0.4 mmol/L) [Ahlborg et al. 1974; Ahlborg & Felig 1982; Wahren et al. 1971], and therefore can be difficult to measure accurately. Variability in the determination of blood flow also adds to the variability of the calculated glucose arteriovenous balance. Despite these limitations, however, determination of the arteriovenous balance of glucose and other metabolites has added immeasurably to the understanding of substrate utilisation during exercise.
1.2 Isotopic Tracer Method Radioactive or stable isotope tracers are being used more and more frequently to measure substrate kinetics during exercise. The rate of wholebody glucose appearance (Ra), for example, can be calculated from the dilution of infused (or injected) glucose molecules labelled with 2H or 3H or with I3C or 14c. The rate of whole-body glucose disappearance (Rd) is then inferred from changes (or lack thereof) in plasma glucose concentration. When the concentrations of unlabelled and labelled glucose are not changing, the system is in steady-state, and Rd must equal Ra. However, when the concentrations of unlabelled or labelled glucose are changing (as is usually the case during exercise), the equations used to calculate Ra and Rd must be modified to reflect these non-steady-state conditions (Steele 1959). The use of isotopic tracers to study glucose metabolism has the advantage of being much less invasive than arteriovenous balance measurements. In addition, when using a glucose tracer labelled with 14C or I3C, it is possible to follow the flux of carbons from glucose to C02, making it possible to quantify the rate of glucose oxidation (Rox). This calculation, however, requires steady-state labelling of glucose and expired C02; conditions which are difficult to achieve during exercise unless the
105
Glucose Metabolism During Exercise
exercise bout is relatively long (i.e. ~ 1.5 hours) [Coggan et al. 1990; Young et al. 1967]. Isotopic tracer methods cannot provide information about glucose metabolism in specific tissue beds during exercise, unless also combined with catheterisation. In addition, in non-steady-state the tracer method may underestimate actual glucose production and utilisation (see below). Although the use of radioactive tracers such as 3H or 14C carries with it some inherent risk, this can be avoided through the use of stable isotope tracers such as 2H or I3c. 1.3 Comparison of the Arteriovenous Balance and Isotopic Tracer Methods
Although the arteriovenous balance and isotopic tracer methods have not been directly compared during exercise in humans, examination of the literature suggests that Ra may underestimate splanchnic glucose release during exercise (table I). Because Rd is derived from the measurement of R a, the rate of whole-body glucose utilisation may also be underestimated. A similar underestimation is often observed during euglycaemic glucose clamps, where Ra is often less than the rate of exogenous glucose infusion (Argoud et al. 1987; Cobelli et al. 1987; Finegood et al. 1987, 1988). Because the isotopic tracer method has proven to be very accurate in estimating glucose production under steady-state conditions (Allsop et al. 1978), this underestimation appears to be due primarily to the inability of the pool-fraction mod~1 of Steele (1959) to adequately model glucose kinetics during non-steady-state. Isotopic discrimination (Argoud et al. 1987; Finegood et al. 1988), and/or radioactive contaminants present in some commercial tracers (McMahon et al. 1989) have also been suggested as contributing factors. Another potential difficulty, especially in studies in which the tracer is infused for a prolonged period before exercise, is incorporation of the label into hepatic glycogen at rest and subsequent re-release during the exercise itself (Sonne & Galbo 1985), which would also result in an underestimation of glucose production.
One way to minimise errors resulting from the pool-fraction model of Steele (1959) is to increase the rate of tracer infusion with the onset of exercise, similar to the approach that has been used during euglycaemic clamps (Finegood et al. 1988). This procedure minimises changes in the concentration of glucose tracer in plasma (fig. 1), thereby also minimising the influence of non-steady-state assumptions. Consequently, the tracer-determined Ra more closely approximates the expected pattern of splanchnic glucose release (fig. I). However, this approach requires accurately anticipating Ra a priori, which can be difficult, especially when studying an intervention that may alter Ra. Therefore, other means of improving non-steady-state estimates of Ra , such as multiple compartment models (Cobelli et al. 1987), may prove to be useful during exercise. The possible underestimation of glucose production and utilisation during exercise by the isotopic tracer method does not invalidate its use, especially for comparisons within the same subject (e.g. pre- and post-training, etc.), where absolute rates are less important than relative changes. It does suggest, however, that the contribution of plasma glucose to substrate metabolism during exercise may be greater than indicated by studies which have used the isotopic tracer approach. Despite the theoretical as well as actual differences between the arteriovenous balance and isotopic dilution techniques, most authors have not differentiated between the measurement of glucose release/uptake and RalRct. In the present manuscript, however, these terms will be used exclusively to refer to measurements made using the arteriovenous balance method and the isotopic tracer method, respectively. Glucose production and utilisation will be used as more general terms to refer to the underlying physiological processes, irrespective of the method of measurement.
2. Plasma Glucose Utilisation During Exercise 2.1 Effect of Exercise Intensity and Duration At rest, skeletal muscle demonstrates a small but positive net glucose uptake (Andres et al. 1956; 10rfeldt & Wahren 1970; Katz et a1. 1986, Klassen
Sports Medicine 11 (2) 1991
106
Table I. Splanchnic glucose release and plasma glucose appearance in untrained men after 40 minutes of cycle ergometer exercise at 50 to 60 %V02peak Reference
n
Intensity (%V02peak)
Splanchnic glucose release or plasma glucose appearance (mmol/min)a
Splanchnic gluco8e relea8e Wahren et a!. (1971) Wahren et a!. (1975) Bjorkman et a!. (1981) Ahlborg & Felig (1982) Bjorkman et a!. (1983)
7 8 10 10 5
60 56 52 58 55
2.23 2.44 2.18 2.43 2.53
56
2.35
60 60 55 60 55 53
1.85 2.15 1.58 1.56 2.12 2.00
57
1.85
Total number of subjects Weighted means Pla8ma glucose appearance Chisholm et a!. (1982) Jenkins et al. (1985) KOivisto et a!. (1985) Hoelzer et a!. (1986a) Jenkins et a!. (1986) Jenkins et a!. (1988) Total number of subjects Weighted means a
40
5 5 8 8 8 5 39
Splanchnic glucose release was determined from the arterial-hepatic venous glucose difference and the estimated splanchnic blood flow. Plasma glucose appearance was determined via primed, continuous infusion of [3-3 Hlglucose, except for the study of Jenkins et a!. (1986). in which [2-3 Hlglucose was used. The percentage of V02peak for the subjects studied by Wahren et a!. (1971) was estimated from their age and bodyweight as presented in the companion paper on amino acid metabolism during exercise (Felig & Wahren 1971). The subjects studied by Hoelzer et a!. (1986a) were assumed to weigh 70kg.
et al. 1970}. This glucose is apparently directed to nonoxidative pathways, as the respiratory quotient (RQ) of resting muscle is close to 0.7, commensurate with predominantly fat oxidation (Andres et al. 1956; Jorfeldt & Wahren 1970; Klassen et al. 1970). With the onset of exercise, the arteriovenous glucose difference initially decreases and may become negative, indicating a net release of glucose from muscle (Jorfeldt & Wahren 1970; Klassen et al. 1970; Wahren 1970). Because skeletal muscle lacks glucose-6-phosphatase, this glucose is thought to be derived from the small percentage of free glucose that is formed from degradation of muscle glycogen. Within a few minutes, however, this glucose release soon reverts to glucose uptake (Jorfeldt & Wahren 1970; Wahren 1970). The rate of skeletal muscle glucose uptake after the first few minutes of exercise is positively re-
lated to the exercise intensity (Katz et al. 1986; Keul et al. 1967; Wahren et al. 1971). During cycle ergometer exercise, glucose uptake by the legs increases curvilinearly from < 0.3 mmol/min at rest to almost 8 mmol/min during maximal exercise (fig. 2). This curvilinear increase in glucose uptake is due to a curvilinear increase in the arteriovenous glucose difference, because blood flow increases in a linear manner (fig. 2). The gradual widening of the arteriovenous glucose difference with increasing exercise intensity may be due to a greater demand for glucose by each active muscle fibre (Nesher et al. 1985), an increase in the number of active muscle fibres (Gollnick et al. 1974; Vollestad & Blom 1985), or (most likely) a combination of these 2 factors. Because glucose uptake by muscle is concentration dependent (Grubb & Snarr 1977; Lewis et al. 1977), the increase in arterial glucose
107
Glucose Metabolism During Exercise
concentration usually observed during intense exercise (Boje 1936; Calles et al. 1983; Kjaer et al. 1986; Wahren et al. 1971) probably also contributes to this widening of the arteriovenous glucose difference. Because glucose uptake by the brain (Ahlborg & Wahren 1972), splanchnic bed (Wasserman et al. 1987), and resting muscle (Ahlborg et al. 1975) remain relatively constant during exercise, Rd also increases curvilinearly with increasing intensity. Thus, from a resting rate of ""0.8 mmoljmin, during short term exercise Rd increases by 1.4-fold at 40 %,,"02peak (Stanley et al. 1988; Wolfe et al. 1986a), 2.5-fold at 85 %,,"02peak (Calles et al. 1983), and 3-fold at 110 %,,"02max (Kjaer et al. 1986). In contrast, Cooper et al. (1989) have recently concluded that Rd does not increase during exercise below the so-called 'anaerobic threshold'. These authors suggested that, during low intensity exer-
15 "2
·E 10
d: ~
.g
~
5
ID
:J"
~
0
g
0.8
E
B c: !!! (I)
== i5
0.6
(I)
8'":::I
0.4
'"0c:
0.2
c;, :::I (I)
.~ 1: < 0.0
8
150
~ 125 ••••••• ~~100 .. .g ~ +-~e4I,.~>Oo-o---.=.t..~........• ~~ ~ 0
6 % E
2-# 50 ~~25
a:>
0
.\.!.Q
0
000 0
0
0000
"2
g ~
Sports Medicine II (2): 102-124. 1991 0112-1642/ 91 / 0002-0 I02/ $1 1.50/ 0 © Adis Intt' rnational Limited. All rights reserved. SP01353
Plasma Glucose Metabolism During Exercise in Humans Andrew R. Coggan Exercise Physiology Laboratory, School of Health, Physical Education, and Recreation, The Ohio State University, Columbus. Ohio, USA
Contents 102 104 104 104 105
105 105 /10
I/O I/O 1/1 1/2 1/3 1/3 114 115 116 116 117 117 118
1/9 120
Summary
Summary I. Methods for Studying Glucose Metabolism During Exercise 1.1 Arteriovenous Balance Method 1.2 Isotopic Tracer Method 1.3 Comparison of the Arteriovenous Balance and Isotopic Tracer Methods 2. Glucose Utilisation During Exercise 2.1 Effect of Exercise Intensity and Duration 2.2 Effect of Exercise Mode/Active Muscle Mass 2.3 Effect of Other Substrates 2.3.1 Muscle Glycogen 2.3.2 Plasma Free Fatty Acids 2.4 Effect of Diet 2.5 Effect of Exercise Training 2.5.1 During Exercise at the Same Absolute Intensity 2.5.2 During Exercise at the Same Relative Intensity 3. Glucose Production During Exercise 3.1 Hepatic Glycogenolysis and Gluconeogenesis 3. Ll Effect of Exercise Intensity and Duration 3. L2 Effect of Exercise Mode/ Acti ve Muscle Mass 3. 1.3 Effect of Diet 3. 1.4 Effect of Exercise Training 3.2 Hormonal Regulation of Hepatic Glucose Production 4. Conclusions
Plasma glucose is an important energy source in exercising humans, supplying between 20 and 50% of the total oxidative energy production and between 25 and 100% of the total carbohydrate oxidised during submaximal exercise. Plasma glucose utilisation increases with the intensity of exercise, due to an increase in glucose utilisation by each active muscle fibre, an increase in the number of active muscle fibres, or both. Plasma glucose utilisation also increases with the duration of exercise, thereby partially compensating for the progressive decrease in muscle glycogen concentration. When compared at the same absolute exercise intensity (i.e. the same V02), reliance on plasma glucose is also greater during exercise performed with a small muscle mass, i.e. with the arms or just I leg. This may be due to differences in the relative exercise
Glucose Metabolism During Exercise
103
intensity (i.e. the %V02peak), or due to differences between the arms and legs in their fitness for aerobic activity. The rate of plasma glucose utilisation is decreased when plasma free fatty acid or muscle glycogen concentrations are very high, effects which are probably mediated by increases in muscle glucose-6-phosphate concentration. However, glucose utilisation is also reduced during exercise following a low carbohydrate diet, despite the fact that muscle glycogen is also often lower. When exercise is performed at the same absolute intensity before and after endurance training, plasma glucose utilisation is lower in the trained state. During exercise performed at the same relative intensity, however, glucose utilisation may be lower, the same, or actually higher in trained than in untrained subjects, because ofthe greater absolute V02 and demand for substrate in trained subjects during exercise at a given relative exercise intensity. Although both hyperglycaemia and hypoglycaemia may occur during exercise, plasma glucose concentration usually remains relatively constant. Factors which increase or decrease the reliance of peripheral tissues on plasma glucose during exercise are therefore generally accompanied by quantitatively similar increases or decreases in glucose production. These changes in total glucose production are mediated by changes in both hepatic glycogenolysis and hepatic gluconeogenesis. Glycogenolysis dominates under most conditions, and is greatest early in exercise, during high intensity exercise, or when dietary carbohydrate intake is high. The rate of gluconeogenesis is increased when exercise is prolonged, preceded by a restricted carbohydrate intake, or performed with the arms. Both glycogenolysis and gluconeogenesis appear to be decreased by endurance exercise training. These effects are due to changes in both the hormonal milieu and in the availability of hepatic glycogen and gluconeogenic precursors. Hepatic glucose production during exercise is stimulated by glucagon and the catecholamines and suppressed by insulin or an increase in plasma glucose concentration. In contrast to earlier suggestions, it appears that a decrease in insulin and an increase in glucagon are both required for hepatic glucose production to increase normally during moderate intensity, moderate duration (40 to 60 minutes) exercise. Changes in the catecholamines. however, may still prove to be important, especially during more intense or more prolonged exercise.
Both fats and carbohydrates are oxidised by skeletal muscle to provide energy during exercise (Krogh & Lindhard 1920). Although the fat stores of the body are quite large, even in lean people, the body's total carbohydrate stores are limited, amounting to some ".,2000 to 2500mmol glucosyl units (Christensen & Hansen 1939; Hedman 1957). Of this total, liver glycogen in the postabsorptive state may contribute ".,400 to 550mmol (Hultman 1977), with circulating glucose in body fluids constituting another ".,90mmol (DeFonzo et a1. 1979). Thus, on the basis of mass alone, it is apparent that glucose transported through the plasma represents about one-fourth of the carbohydrate available to provide fuel for skeletal muscle during exercise. When one considers that only glycogen within the active muscles is readily accessible as a fuel during exercise (since muscle lacks g1ucose-6-phosphatase), and that the liver can also synthesise glucose from 3-carbon intermediates such as lactate, the
potential importance of plasma-borne glucose to carbohydrate energy production becomes even greater. Indeed, at times plasma glucose supplies 70 to 100% of the carbohydrate oxidised during exercise (Ahlborg et a1. 1974; Broberg & Sahlin 1989; Wahren et a1. 1971), even at intensities as high as ".,75% Y02peak (Coggan & Coyle 1987) [Y02max, maximal oxygen uptake measured during treadmill running; Y02peak, peak oxygen uptake measured during cycle ergometer exercise]. The rate of plasma glucose utilisation during exercise is dependent upon both the intensity and duration of the exercise bout, as well as upon the mode of exercise itself. In addition, plasma glucose utilisation during exercise may be influenced by the availability of other substrates and by the pre-exercise diet and training status of the individual. The present review considers these factors, as well as the role of the liver in supplying glucose during exercise. The emphasis is placed upon studies of
104
whole-body exercise in normal individuals. For information regarding the effects of exercise on skeletal muscle glucose transport, or on the effects of exercise on plasma glucose metabolism in disease state such as diabetes, readers are referred to other recent reviews (Bjorkman 1986; Ivy 1987; Wasserman & Vranic 1986).
1. Methods for Studying Glucose Metabolism During Exercise 1.1 Arteriovenous Balance Method
Glucose metabolism during exercise has traditionally been studied by catheterising both the arterial blood supply and the venous drainage of a given tissue bed, thereby allowing measurement of the arteriovenous glucose difference and the rate of blood flow. While the product of these 2 parameters (i.e. the net arteriovenous glucose balance) does not indicate the ultimate fate (or source) of the glucose taken up (or released), it does indicate the maximum possible contribution of glucose to (or from) the metabolic processes in the tissue bed under study. The arteriovenous balance method provides the ability to examine glucose metabolism in specific tissue beds during exercise. For example, this technique has been used to compare glucose uptake by active vs nonactive limbs (Ahlborg et al. 1975), or to quantify the contributions of renal and hepatic tissues to overall glucose release (Wahren et al. 1971). However, because it is usually not possible to catheterise the portal vein in humans, it is actually splanchnic (rather than hepatic) glucose release that is normally measured, resulting in a 5 to 10% underestimation of the true rate of hepatic glucose release (Wahren et al. 1971; Wasserman et al. 1987). Net splanchnic release is usually the parameter of interest, however, as it indicates the rate at which glucose is being made available for metabolism in peripheral tissues. The arteriovenous balance technique has several disadvantages, perhaps the greatest of which is its invasive nature. As indicated above, simple measurement of the arteriovenous balance also does not provide any indication of the metabolic fate
Sports Medicine II (2) 1991
(i.e. oxidation, glycogen synthesis, lactate production, etc.) or source (i.e. glycogenolysis, gluconeogenesis) of the glucose that is exchanged during exercise. In addition, the arteriovenous glucose difference across exercising muscle is extremely small (0.1 to 0.4 mmol/L) [Ahlborg et al. 1974; Ahlborg & Felig 1982; Wahren et al. 1971], and therefore can be difficult to measure accurately. Variability in the determination of blood flow also adds to the variability of the calculated glucose arteriovenous balance. Despite these limitations, however, determination of the arteriovenous balance of glucose and other metabolites has added immeasurably to the understanding of substrate utilisation during exercise.
1.2 Isotopic Tracer Method Radioactive or stable isotope tracers are being used more and more frequently to measure substrate kinetics during exercise. The rate of wholebody glucose appearance (Ra), for example, can be calculated from the dilution of infused (or injected) glucose molecules labelled with 2H or 3H or with I3C or 14c. The rate of whole-body glucose disappearance (Rd) is then inferred from changes (or lack thereof) in plasma glucose concentration. When the concentrations of unlabelled and labelled glucose are not changing, the system is in steady-state, and Rd must equal Ra. However, when the concentrations of unlabelled or labelled glucose are changing (as is usually the case during exercise), the equations used to calculate Ra and Rd must be modified to reflect these non-steady-state conditions (Steele 1959). The use of isotopic tracers to study glucose metabolism has the advantage of being much less invasive than arteriovenous balance measurements. In addition, when using a glucose tracer labelled with 14C or I3C, it is possible to follow the flux of carbons from glucose to C02, making it possible to quantify the rate of glucose oxidation (Rox). This calculation, however, requires steady-state labelling of glucose and expired C02; conditions which are difficult to achieve during exercise unless the
105
Glucose Metabolism During Exercise
exercise bout is relatively long (i.e. ~ 1.5 hours) [Coggan et al. 1990; Young et al. 1967]. Isotopic tracer methods cannot provide information about glucose metabolism in specific tissue beds during exercise, unless also combined with catheterisation. In addition, in non-steady-state the tracer method may underestimate actual glucose production and utilisation (see below). Although the use of radioactive tracers such as 3H or 14C carries with it some inherent risk, this can be avoided through the use of stable isotope tracers such as 2H or I3c. 1.3 Comparison of the Arteriovenous Balance and Isotopic Tracer Methods
Although the arteriovenous balance and isotopic tracer methods have not been directly compared during exercise in humans, examination of the literature suggests that Ra may underestimate splanchnic glucose release during exercise (table I). Because Rd is derived from the measurement of R a, the rate of whole-body glucose utilisation may also be underestimated. A similar underestimation is often observed during euglycaemic glucose clamps, where Ra is often less than the rate of exogenous glucose infusion (Argoud et al. 1987; Cobelli et al. 1987; Finegood et al. 1987, 1988). Because the isotopic tracer method has proven to be very accurate in estimating glucose production under steady-state conditions (Allsop et al. 1978), this underestimation appears to be due primarily to the inability of the pool-fraction mod~1 of Steele (1959) to adequately model glucose kinetics during non-steady-state. Isotopic discrimination (Argoud et al. 1987; Finegood et al. 1988), and/or radioactive contaminants present in some commercial tracers (McMahon et al. 1989) have also been suggested as contributing factors. Another potential difficulty, especially in studies in which the tracer is infused for a prolonged period before exercise, is incorporation of the label into hepatic glycogen at rest and subsequent re-release during the exercise itself (Sonne & Galbo 1985), which would also result in an underestimation of glucose production.
One way to minimise errors resulting from the pool-fraction model of Steele (1959) is to increase the rate of tracer infusion with the onset of exercise, similar to the approach that has been used during euglycaemic clamps (Finegood et al. 1988). This procedure minimises changes in the concentration of glucose tracer in plasma (fig. 1), thereby also minimising the influence of non-steady-state assumptions. Consequently, the tracer-determined Ra more closely approximates the expected pattern of splanchnic glucose release (fig. I). However, this approach requires accurately anticipating Ra a priori, which can be difficult, especially when studying an intervention that may alter Ra. Therefore, other means of improving non-steady-state estimates of Ra , such as multiple compartment models (Cobelli et al. 1987), may prove to be useful during exercise. The possible underestimation of glucose production and utilisation during exercise by the isotopic tracer method does not invalidate its use, especially for comparisons within the same subject (e.g. pre- and post-training, etc.), where absolute rates are less important than relative changes. It does suggest, however, that the contribution of plasma glucose to substrate metabolism during exercise may be greater than indicated by studies which have used the isotopic tracer approach. Despite the theoretical as well as actual differences between the arteriovenous balance and isotopic dilution techniques, most authors have not differentiated between the measurement of glucose release/uptake and RalRct. In the present manuscript, however, these terms will be used exclusively to refer to measurements made using the arteriovenous balance method and the isotopic tracer method, respectively. Glucose production and utilisation will be used as more general terms to refer to the underlying physiological processes, irrespective of the method of measurement.
2. Plasma Glucose Utilisation During Exercise 2.1 Effect of Exercise Intensity and Duration At rest, skeletal muscle demonstrates a small but positive net glucose uptake (Andres et al. 1956; 10rfeldt & Wahren 1970; Katz et a1. 1986, Klassen
Sports Medicine 11 (2) 1991
106
Table I. Splanchnic glucose release and plasma glucose appearance in untrained men after 40 minutes of cycle ergometer exercise at 50 to 60 %V02peak Reference
n
Intensity (%V02peak)
Splanchnic glucose release or plasma glucose appearance (mmol/min)a
Splanchnic gluco8e relea8e Wahren et a!. (1971) Wahren et a!. (1975) Bjorkman et a!. (1981) Ahlborg & Felig (1982) Bjorkman et a!. (1983)
7 8 10 10 5
60 56 52 58 55
2.23 2.44 2.18 2.43 2.53
56
2.35
60 60 55 60 55 53
1.85 2.15 1.58 1.56 2.12 2.00
57
1.85
Total number of subjects Weighted means Pla8ma glucose appearance Chisholm et a!. (1982) Jenkins et al. (1985) KOivisto et a!. (1985) Hoelzer et a!. (1986a) Jenkins et a!. (1986) Jenkins et a!. (1988) Total number of subjects Weighted means a
40
5 5 8 8 8 5 39
Splanchnic glucose release was determined from the arterial-hepatic venous glucose difference and the estimated splanchnic blood flow. Plasma glucose appearance was determined via primed, continuous infusion of [3-3 Hlglucose, except for the study of Jenkins et a!. (1986). in which [2-3 Hlglucose was used. The percentage of V02peak for the subjects studied by Wahren et a!. (1971) was estimated from their age and bodyweight as presented in the companion paper on amino acid metabolism during exercise (Felig & Wahren 1971). The subjects studied by Hoelzer et a!. (1986a) were assumed to weigh 70kg.
et al. 1970}. This glucose is apparently directed to nonoxidative pathways, as the respiratory quotient (RQ) of resting muscle is close to 0.7, commensurate with predominantly fat oxidation (Andres et al. 1956; Jorfeldt & Wahren 1970; Klassen et al. 1970). With the onset of exercise, the arteriovenous glucose difference initially decreases and may become negative, indicating a net release of glucose from muscle (Jorfeldt & Wahren 1970; Klassen et al. 1970; Wahren 1970). Because skeletal muscle lacks glucose-6-phosphatase, this glucose is thought to be derived from the small percentage of free glucose that is formed from degradation of muscle glycogen. Within a few minutes, however, this glucose release soon reverts to glucose uptake (Jorfeldt & Wahren 1970; Wahren 1970). The rate of skeletal muscle glucose uptake after the first few minutes of exercise is positively re-
lated to the exercise intensity (Katz et al. 1986; Keul et al. 1967; Wahren et al. 1971). During cycle ergometer exercise, glucose uptake by the legs increases curvilinearly from < 0.3 mmol/min at rest to almost 8 mmol/min during maximal exercise (fig. 2). This curvilinear increase in glucose uptake is due to a curvilinear increase in the arteriovenous glucose difference, because blood flow increases in a linear manner (fig. 2). The gradual widening of the arteriovenous glucose difference with increasing exercise intensity may be due to a greater demand for glucose by each active muscle fibre (Nesher et al. 1985), an increase in the number of active muscle fibres (Gollnick et al. 1974; Vollestad & Blom 1985), or (most likely) a combination of these 2 factors. Because glucose uptake by muscle is concentration dependent (Grubb & Snarr 1977; Lewis et al. 1977), the increase in arterial glucose
107
Glucose Metabolism During Exercise
concentration usually observed during intense exercise (Boje 1936; Calles et al. 1983; Kjaer et al. 1986; Wahren et al. 1971) probably also contributes to this widening of the arteriovenous glucose difference. Because glucose uptake by the brain (Ahlborg & Wahren 1972), splanchnic bed (Wasserman et al. 1987), and resting muscle (Ahlborg et al. 1975) remain relatively constant during exercise, Rd also increases curvilinearly with increasing intensity. Thus, from a resting rate of ""0.8 mmoljmin, during short term exercise Rd increases by 1.4-fold at 40 %,,"02peak (Stanley et al. 1988; Wolfe et al. 1986a), 2.5-fold at 85 %,,"02peak (Calles et al. 1983), and 3-fold at 110 %,,"02max (Kjaer et al. 1986). In contrast, Cooper et al. (1989) have recently concluded that Rd does not increase during exercise below the so-called 'anaerobic threshold'. These authors suggested that, during low intensity exer-
15 "2
·E 10
d: ~
.g
~
5
ID
:J"
~
0
g
0.8
E
B c: !!! (I)
== i5
0.6
(I)
8'":::I
0.4
'"0c:
0.2
c;, :::I (I)
.~ 1: < 0.0
8
150
~ 125 ••••••• ~~100 .. .g ~ +-~e4I,.~>Oo-o---.=.t..~........• ~~ ~ 0
6 % E
2-# 50 ~~25
a:>
0
.\.!.Q
0
000 0
0
0000
"2
g ~
