The following is the abstract of the article discussed in the subsequent letter:
Friedlander, Anne L., Gretchen A. Casazza, Michael A. Horning, Melvin J. Huie, and George A. Brooks. Training-induced alterations of glucose flux in men. J. Appl. Physiol. 82(4): 1360–1369, 1997.—We examined the hypothesis that glucose flux was directly related to relative exercise intensity both before and after a 10-wk cycle ergometer training program in 19 healthy male subjects. Two pretraining trials [45 and 65% of peak O2 consumption (V˙o 2 peak)] and two post- training trials (same absolute and relative intensities as 65% pretraining) were performed for 90 min of rest and 1 h of cycling exercise. After training, subjects increasedV˙o 2 peak by 9.4 ± 1.4%. Pretraining, the intensity effect on glucose kinetics was evident with rates of appearance (Ra; 5.84 ± 0.23 vs. 4.73 ± 0.19 mg ⋅ kg−1 ⋅ min−1), disappearance (Rd; 5.78 ± 0.19 vs. 4.73 ± 0.19 mg ⋅ kg−1 ⋅ min−1), oxidation (Rox; 5.36 ± 0.15 vs. 3.41 ± 0.23 mg ⋅ kg−1 ⋅ min−1), and metabolic clearance (7.03 ± 0.56 vs. 5.20 ± 0.28 ml ⋅ kg−1 ⋅ min−1) of glucose being significantly greater (P ≤ 0.05) in the 65% than the 45%V˙o 2 peak trial. When Rd was expressed as a percentage of total energy expended per minute (Rd E), there was no difference between the 45 and 65% intensities. Training did reduce Ra(4.63 ± 0.25), Rd(4.65 ± 0.24), Rox(3.77 ± 0.43), and Rd E(15.30 ± 0.40 to 12.85 ± 0.81) when subjects were tested at the same absolute workload (P ≤ 0.05). However, when they were tested at the same relative workload, Ra, Rd, and Rd E were not different, although Roxwas lower posttraining (5.36 ± 0.15 vs. 4.41 ± 0.42,P ≤ 0.05). These results show 1) glucose use is directly related to exercise intensity; 2) training decreases glucose flux for a given power output; 3) when expressed as relative exercise intensity, training does not affect the magnitude of blood glucose use during exercise; 4) training alters the pathways of glucose disposal.
Training-Induced Alterations in Glucose Metabolism During Exercise
To the Editor: Friedlander et al. (8) have confirmed our findings that endurance training reduces glucose appearance, disappearance, and oxidation rates (Ra, Rd, and Rox, respectively) during moderate exercise performed at the same absolute power output before and after training (4). Their data also corroborate our report that the training-induced decrease in Ra is due, in part, to a decrease in the rate of gluconeogenesis (6). Finally, their Rox data agree with our conclusion, based on cross-sectional observations, that training reduces glucose utilization even when the exercise is performed at the same relative intensity [i.e., at the same percentage of peak O2 uptake (V˙o 2 peak)], as in the untrained state (5). Paradoxically, however, Friedlander et al. (8) did not find any training-induced changes in Ra/Rdunder the latter conditions, leading them to conclude that “… when expressed as relative exercise intensity, training does not affect blood glucose flux… .” Instead, they argue that training affects the pathways of glucose disposal, possibly by enhancing muscle glycogen synthesis during exercise.
I believe that there is a simpler explanation for their findings. As we (5, 6) and others have done, Friedlander et al. (8) increased the rate of tracer infusion (F) at the onset of exercise to minimize changes in glucose enrichment. In contrast to previous studies, however, they did not use the same F under all conditions. Instead, they increased F fourfold during exercise at 65% ofV˙o 2 peak after training, compared with only threefold in the other three trials. As shown previously (2, 7), the magnitude of the step increase in F has a direct impact on the calculated Ra during exercise. Importantly, this is true, even though an apparent plateau in plasma enrichment may be obtained (2). The reason for this is that, even though exercise increases the rate of glucose exchange between various pools in the body (3), considerable time is still required before complete reequilibration of the tracer is achieved. For example, a computer simulation (MLAB, Civilized Software, Bethesda, MD) using the latter compartmental modeling data indicates that ∼90 min would be required to reach a new steady state in plasma enrichment at 65% ofV˙o 2 peak [i.e., the intensity used by Friedlander et al. (8)]. Because the increase inF and its ultimate effect on plasma enrichment are dissociated with respect to time, yet the increase in F enters immediately into the calculations, a larger step increase in F results in a higher estimated Ra, even when it is calculated by using the non-steady-state Steele equation. Because Friedlander et al. increased F fourfold after training vs. only threefold before training, it is likely that they overestimated Ra after training compared with before training and, consequently, underestimated the change due to training. Indeed, both the computer simulation and my unpublished observations suggest that, because of the difference in tracer-infusion protocols, Ra was biased upward by ∼15% after training compared with before training, such that Friedlander et al. missed a similar percent decrease in Ra (and thus Rd) due to training. An ∼15% lower Ra and Rd during exercise at 65% ofV˙o 2 peak after training would agree quite nicely with their finding that training reduced Rox by ∼18% under these conditions (the calculated Rox would be much less affected by the difference inF, because of the intervening effect of the multiple-pool bicarbonate system). This interpretation also agrees with the fact that when the subjects exercised at the same absolute intensity as before training (i.e., at 58% of their new, higherV˙o 2 peak), their Ra and Rd were the same as when they had exercised at 45% ofV˙o 2 peak before training. In other words, Friedlander et al.’s (8) own data show that, whenF is the same, a higher relative exercise intensity is required after training to elicit the same response in Ra and Rd.
Despite this limitation in experimental design, the study of Friedlander et al. (8) seemingly refutes the most important new tenet of Brooks’ and Mercier’s “crossover” concept, which is the prediction that “… after endurance training blood glucose appearance rate and the rate of hepatic gluconeogenesis are increased… .” during intense exercise at the same relative intensity (1). Perhaps the authors would like to comment on this issue.
- Copyright © 1998 the American Physiological Society
To the Editor: We appreciate the opportunity to respond to Coggan’s letter, but we are confused by several aspects of it. At the outset of his letter, Coggan claims that our results (1-6) are like his (1-4), but then he goes on to criticize our methodology with reference to his own work. An inspection of both Coggan’s letter and paper (1-4) reveals flaws in both.
Coggan criticizes our work by asserting that the magnitude of the step increase in tracer infusion rate at the onset of exercise affects the computed Ra during exercise, even after an “apparent plateau” in plasma isotopic enrichment (IE). He hypothesizes a slow equilibration of tracer distribution after the onset of exercise and predicts that a steady state in plasma enrichment in our experiment would require ∼90 min, based on a computer simulation with unspecified structure or parameters.
In assessing Coggan’s assertion, it is important to consider our design and examine the data. We used a steady-state design so that we could calculate and report glucose flux rates after 30 min of exercise when blood IE values were constant, or nearly so. Our Fig. 1 shows IE as a function of time. In none of the records does IE continue to increase after 30 min of exercise, as Coggan predicts. In three of four experimental conditions, IE is constant during the last 30 min of exercise, the period during which we computed Ra. Examination of the data shows that this is a statistical fact, not just an “apparent plateau.” In steady state, the magnitude of the infusion rate F does not affect the computed Ra: a larger F causes a proportionally larger enrichment. In the fourth experimental condition [65% ofV˙o 2 peakpretraining], the IE was not constant but fell in the last 30 min of exercise. We used the Steele equation (1-10) to compute the non-steady-state Ra in this case. It is important to recognize that the variable-volume model cited in Coggan’s letter (1-5) accounts for the effect of the declining Ra in exactly the same way: the declining IE leads to a larger estimate of Ra. It should be noted that in this experimental condition the step in F was insufficient to maintain a constant IE. This decline in IE after a peak, and a corresponding increase in the estimate of Ra, are just the opposite of the effect hypothesized by Coggan.
Without an adequate description of the computer simulation mentioned by Coggan in his letter, it is difficult to comment on any errors he might have made. That mistakes were made is clear when the data (Fig. 1, showing IE stabilized in 30 min) are compared with the model prediction (90 min). Furthermore, a direct experimental test also denies Coggan’s hypothesis: in a parallel experiment, performed on nine different subjects, but with a step to four times the resting F during exercise trials eliciting 65% ofV˙o 2 peak, pre- and posttraining, we (B. C. Bergman, G. E. Butterfield, G. A. Casazza, M. A. Horning, E. E. Wolfel, and G. A. Brooks, unpublished observations) reproduced our previously published glucose flux rates (1-6).
In contrast to our results (1-6), in which F was held constant in subjects during steady-rate exercise, are those of Coggan et al. (1-4). By varying the isotope infusion rate continuously during their experiments, Coggan and associates created the worst possible scenario. Exercise power output and metabolic rate (V˙o 2) were constant as was blood glucose concentration. However, by grading the isotope infusion rate over time, Coggan et al. ensured a changing blood IE, slow and continuously changing tracer mixing in various compartments, and constantly changing (calculated) glucose Ra and Rd. In other words, Coggan et al. turned a steady-state metabolic condition into a nonsteady isotope tracer condition, an unfortunate situation, which complicated and compromised calculation of glucose fluxes.
In our study, tracer-measured blood glucose Ra rose rapidly at the start of exercise and then plateaued, perhaps rising slowly over time. Splanchnic (hepatic) glucose release is, unfortunately, infrequently measured, but the data available on humans (1-11) and dogs (1-12) show the same pattern as we observed. Moreover, reports on human limb glucose net uptake during exercise (1-7-1-9) show a similar pattern of rapid adjustment at the outset of contractions, followed by slow adjustment over time. In contrast, the experimental paradigm employed by Coggan et al. (1-4) of ramping the F continuously produced the result of glucose Ra and Rd rising linearly over time. In our view, this result was unphysiological and an artifact of the methodology employed. Thus the relatively small differences between athletes and nonathletes observed by Coggan et al. need to be viewed with a degree of uncertainty.
Irrespective of methodological differences, from the physiological standpoint, an important point addressed by Coggan et al. (1-4) and us (1-6) is that of the effect of chronic physical activity on glucose kinetics. We utilized a longitudinal design and studied healthy young men before and after training; in contrast, Coggan et al. (1-4) compared responses in athletes and nonathletes. We note that it is generally recognized that physiological and metabolic responses in athletes are distinct from those of nonathletes (1-1). Consequently, it is seldom asserted that responses of athletes to any condition uniquely demonstrate the effects of environment (i.e., training), as opposed to the interactive effects of environment and genetics.
From the scientific standpoint, we would like to be able to state that the effects of training on glucose flux are sufficiently robust that they are demonstrable when a variety of designs and experimental approaches is used. It is our position that the more recent data obtained on humans by using stable-isotope tracers are similar to those obtained previously by us on laboratory animals (1-2). Thus we are confident about the generality of the findings. We do acknowledge that it is Coggan’s right to object to our position and we view his objections as useful, since they focus debate on key issues, even if we find his objections groundless.
Finally, with regard to Coggan’s objections to the “crossover” concept (1-3), we remain confident that data obtained from measurements of respiratory gas exchange, isotopic tracers, muscle biopsies, and arteriovenous concentration differences in humans and other mammals (1-2,1-4, 1-6-1-9) show crossover to carbohydrate dependency during exercises eliciting 50% maximalV˙o 2, regardless of training status.
- Copyright © 1998 the American Physiological Society