We tested the hypothesis that the asymptote of the hyperbolic relationship between work rate and time to exhaustion during muscular exercise, the “critical power” (CP), represents the highest constant work rate that can be sustained without a progressive loss of homeostasis [as assessed using 31P magnetic resonance spectroscopy (MRS) measurements of muscle metabolites]. Six healthy male subjects initially completed single-leg knee-extension exercise at three to four different constant work rates to the limit of tolerance (range 3–18 min) for estimation of the CP (mean ± SD, 20 ± 2 W). Subsequently, the subjects exercised at work rates 10% below CP (<CP) for 20 min and 10% above CP (>CP) for as long as possible, while the metabolic responses in the contracting quadriceps muscle, i.e., phosphorylcreatine concentration ([PCr]), Pi concentration ([Pi]), and pH, were estimated using 31P-MRS. All subjects completed 20 min of <CP exercise without duress, whereas the limit of tolerance during >CP exercise was 14.7 ± 7.1 min. During <CP exercise, stable values for [PCr], [Pi], and pH were attained within 3 min after the onset of exercise, and there were no further significant changes in these variables (end-exercise values = 68 ± 11% of baseline [PCr], 314 ± 216% of baseline [Pi], and pH 7.01 ± 0.03). During >CP exercise, however, [PCr] continued to fall to the point of exhaustion and [Pi] and pH changed precipitously to values that are typically observed at the termination of high-intensity exhaustive exercise (end-exercise values = 26 ± 16% of baseline [PCr], 564 ± 167% of baseline [Pi], and pH 6.87 ± 0.10, all P < 0.05 vs. <CP exercise). These data support the hypothesis that the CP represents the highest constant work rate that can be sustained without a progressive depletion of muscle high-energy phosphates and a rapid accumulation of metabolites (i.e., H+ concentration and [Pi]), which have been associated with the fatigue process.
- fatigue threshold
- maximal steady state
- exercise tolerance
- oxygen kinetics
the hyperbolic relationship between work rate and time to exhaustion (defined here as an inability to maintain the external work rate) is a fundamental property of exercise performance in humans (37, 38, 44) and other species (6, 15, 16, 34). Monod and Scherrer (37) first reported this hyperbolic relationship in a single muscle group, and this relationship was subsequently demonstrated during whole body exercise, such as cycling (38, 63), treadmill running (24), swimming (60), and rowing (22). The work-rate asymptote of this hyperbolic relationship has been termed the “critical power” (CP) or “fatigue threshold,” whereas the curvature constant, which represents the total amount of work that can be performed above the CP, has been termed the W′ (37, 38, 44, 63). The parameters CP and W′ can also be derived through linear regression analysis after transformation of the hyperbolic relationship to a linear formulation by plotting total work done during the series of exercise tests vs. time to exhaustion (37) or by plotting power output vs. the inverse of time to exhaustion (13, 44).
Moritani et al. (38) proposed that the CP represented the highest rate at which aerobic energy could be supplied during sustained exercise and that exercise below the CP could be sustained indefinitely (in reality, until the subject becomes limited by factors such as substrate depletion and hyperthermia). In contrast, sustained exercise above the CP requires utilization of the finite W′ at a predictable rate until it is expended, at which point exercise cannot be continued at the same intensity. The W′ therefore represents a constant amount of work that can be performed above the CP and is notionally equivalent to an energy store consisting of O2 reserves (myoglobin and venous blood), high-energy phosphates, and a source related to “anaerobic” glycolysis (37, 38). The higher the sustained power output above the CP, the more rapidly the W′ will be expended, and the greater will be the rate at which metabolites that have been associated with the fatigue process (e.g., Pi, ADP, H+, and extracellular K+) accumulate (11, 48, 61, 66).
In support of the theoretical constructs of CP and W′, studies have shown that the CP is increased by endurance training (25, 45) and reduced by inspiration of a hypoxic gas (38) and that the W′ is increased by creatine supplementation (35, 55) and high-intensity strength and sprint training (26) but reduced after glycogen depletion (36) and prior high-intensity exercise with limited recovery (19). Moreover, the CP appears to broadly correspond to the exercise intensity at the maximal lactate steady state (i.e., the highest power output that can be sustained without a continued increase in blood lactate concentration with time) (47, 54). Indeed, Poole et al. (44) demonstrated that arterialized venous blood lactate and bicarbonate concentrations ([lactate] and [bicarbonate]) and pH ultimately attained steady values when subjects exercised for 24 min at the CP but that blood [lactate] increased and pH and [bicarbonate] fell progressively during exercise performed at a work rate that was 5% above CP, such that exhaustion occurred before the 24-min target was attained. Moreover, a steady state in O2 uptake (V̇o2) was attained at the CP (although it was delayed and elevated above the expected value by the V̇o2 slow component), whereas during exercise above the CP, V̇o2 increased inexorably throughout exercise until V̇o2 that was not different from the independently determined maximal V̇o2 (V̇o2 max) was reached (44).
On the basis of these distinct blood acid-base and pulmonary gas exchange profiles, the CP has been accepted as an important demarcator of exercise intensity, with corresponding implications for the predominant mechanism(s) of muscular fatigue and the tolerable duration of exercise (28, 64). However, to our knowledge, no previous studies have investigated the temporal profile of the muscle metabolic responses to prolonged <CP and >CP exercise. If the CP is a valid theoretical and practical construct, one would predict that <CP exercise would result, after an initial rapid change, in a stabilization of phosphorylcreatine (PCr) and Pi concentrations ([PCr] and [Pi]) and pH in the contracting muscles. During >CP exercise, where the W′ will be utilized at a predictable rate until fatigue prevents the continuation of exercise, however, a continued fall in muscle [PCr] and pH and a continued rise in muscle [Pi] would be expected. The concentration of muscle high-energy phosphate metabolites and pH can be assessed noninvasively, in real time, and with high temporal resolution via 31P magnetic resonance spectroscopy (31P-MRS).
In the present study, we therefore sought to investigate the mechanistic bases for the CP concept by using 31P-MRS to resolve the muscle metabolic responses to constant-work-rate single-leg knee-extension exercise at ∼10% below and 10% above the predetermined CP. We hypothesized that <CP exercise would result, after an initial rapid change, in the early attainment of stable values of [PCr], [Pi], and pH and the completion of 20 min of exercise without appreciable duress. We further hypothesized that >CP exercise would be characterized by a continued increase in [Pi] and a continued reduction in [PCr] and pH until the subjects reached their limit of tolerance.
Six male subjects (mean ± SD: 30 ± 8 yr, 1.82 ± 0.04 m, 82.5 ± 7.6 kg) volunteered and gave written informed consent to participate in the present study, which had been approved by the local research ethics committee. All the subjects were recreationally active in a variety of sport and exercise activities, were nonsmokers, and were familiar with the exercise mode and experimental procedures used in the present study. The subjects were instructed to report to the laboratory in a rested state, having completed no strenuous exercise within the previous 24 h and having consumed a light carbohydrate-based meal 2–3 h before the scheduled appointment. The subjects were also instructed to abstain from alcohol and caffeine for the preceding 12 and 6 h, respectively. Adherence to these instructions was verified on the subjects' arrival at the laboratory.
The study was conducted in two parts. The work rate vs. time to exhaustion relationship was first established during single-leg knee-extension exercise for each subject from three to four separate exercise bouts. From this relationship, the CP (and W′) was estimated. Then subjects performed single-leg knee-extension exercise at work rates that were ∼10% below and 10% above their CP while positioned inside the bore of a 1.5-T superconducting MR scanner for 31P-MRS assessment of muscle metabolic responses. In all cases, the right leg was used for the experiments, and subjects exercised using the same ergometer in both parts of the study. All exercise tests were conducted on separate days and at a similar time of day (within ±3 h).
Part I: derivation of the work rate vs. time to exhaustion relationship and estimation of CP.
The subjects initially completed a minimum of three exercise bouts, each at a constant, but different, work rate for determination of the hyperbolic relationship between work rate and time to exhaustion. The work rates were selected, on the basis of previous tests performed by these same subjects during other experiments, to result in exhaustion in 2–15 min. Specifically, the highest and lowest work rates in the series were selected to elicit exhaustion in 2–3 min and 12–15 min, respectively, with the other work rate(s) selected to cause exhaustion in ∼4–8 min. In four subjects, the first three work rates produced times to exhaustion within the desired range; in the remaining two subjects, an additional work rate was required to ensure an appropriate range of times to exhaustion. The exercise bouts were performed on separate days and presented in random order. During all exercise tests, the subjects were verbally encouraged to continue exercising for as long as possible. The time to exhaustion, defined as the time at which the subject could no longer keep pace with the required rate of muscle contraction (see below), was recorded to the nearest second by a hand-held stopwatch.
The CP and W′ were estimated using three separate models: 1) the nonlinear power-time model [time = W′/(P − CP)], in which the asymptote of the hyperbolic relationship is the CP and the curvature constant is the W′; 2) the linear work-time model, in which work done is plotted against time to exhaustion (W = CP·time + W′) and linear regression analysis is used to estimate the CP (slope of the line) and the W′ (y-intercept); and 3) the linear 1/time model, in which work rate is plotted against the inverse of the time to exhaustion and linear regression analysis is used to estimate the CP (y-intercept) and the W′ (slope of the line) (13, 17, 23, 40, 44, 47). The 95% confidence intervals for the estimation of CP from model 3 were used to calculate work rates that were just below (∼10%) and just above (∼10%) the CP. For example, if the CP was estimated as 25 W with a 95% confidence interval of ±2 W, then the subject exercised at a work rate of 22 W during the <CP condition and 28 W during the >CP condition. We reasoned that this approach would provide reasonable assurance that the work rates were just below and above the CP for all subjects.
Part II: 31P-MRS assessment of muscle metabolic responses to <CP and >CP exercise.
Within 7 days of completion of the predictive trials for estimation of the CP, the subjects reported to the MRS laboratory on two separate occasions to perform <CP and >CP exercise with simultaneous measurement of muscle metabolic responses by 31P-MRS. The order in which the <CP and >CP trials was presented was randomized. Subjects were required to complete 20 min of <CP exercise and to exercise for as long as possible at >CP. With the exception of the measurement of the muscle metabolic response, the equipment and general procedures were identical to those described for part I. The time to exhaustion for >CP exercise was predicted [i.e., time = W′/(P − CP)] and compared with the actual time to exhaustion.
Equipment and MRS Measurements
The single-leg knee-extension exercise tests were conducted in the prone position, with subjects secured to the ergometer bed with Velcro straps at the thigh, buttocks, and lower back to minimize extraneous movement during the protocol. The ergometer was constructed in-house and consists of a self-supporting nylon frame, which fits onto the bed and is placed close to the subject's feet, and a base unit placed at the end of the bed. The subject's right foot was connected to a rope that runs over the top of the frame to the base unit on which a mounted pulley system permitted nonmagnetic weights to be lifted and lowered. Within the pulley system, a shaft encoder (type BDK-06, Baumer Electronics, Swindon, UK) was fitted to record distances through which the load was moved, alongside a nonmagnetic load cell (type F250, Novatech Measurements, St. Leonards-On-Sea, UK) to record applied forces. Exercise was performed at a rate of 40 rpm, with the subjects lifting and lowering the weight over a distance of ∼0.22 m in accordance with a visual cue. The equipment and procedures were the same in parts I and II, except, in part II, subjects were positioned inside a whole body MRI system. A 6-cm 31P transmit/receive surface coil was placed within the subject bed, and the subject was asked to lie on it, such that the coil was centered over the quadriceps muscle of the leg to be exercised.
MRS was performed in the Peninsula Magnetic Resonance Research Centre (Exeter, UK) with a 1.5-T superconducting MR scanner (Intera, Philips). Initially, fast field echo images were acquired to determine that the muscle was positioned correctly relative to the coil. Placement of cod liver oil capsules, which yield high-intensity signal points within the image, adjacent to the coil, allowed its orientation relative to the muscle volume under examination to be assessed. A number of preacquisition steps were carried out to optimize the signal from the muscle under investigation. An automatic shimming protocol was undertaken within a volume that defined the quadriceps muscle to optimize the homogeneity of the local magnetic field, thereby leading to maximum signal collection. Tuning and matching of the coil were then performed to maximize energy transfer between the coil and the muscle. To ensure that the muscle was consistently scanned with the same level of contraction, thereby ensuring a consistent distance from the coil at the time of data sampling, the subject was visually cued via a display consisting of two vertical bars, one that moved at a constant rate with a frequency of 0.67 Hz and one that monitored foot movement via a sensor within the pulley to which they were connected. Thus the subject endeavored to match the movements of these two bars. The contraction phase of the knee extensors and 31P-MRS examination of the quadriceps were synchronized. At the time of data acquisition, a trigger pulse was sent from the phosphorous MR amplifier to the visual display to ensure that the two maintain synchronization. Thus, provided that subjects correctly match their movements to the display, the leg will always be in the same position when the signal is collected. The work done by the subject was recorded via a nonmagnetic strain gauge within the pulley mechanism, allowing the actual work rate to be calculated.
Before exercise, during exercise, and during recovery, data were acquired every 1.5 s, with a spectral width of 1,500 Hz and 1,000 data points. Phase cycling with four phase cycles was employed, leading to a spectrum being acquired every 6 s. The subsequent spectra were quantified via peak fitting, with the assumption of prior knowledge, using the jMRUI (version 2) software package and the AMARES fitting algorithm (58). Spectra were fitted with the assumption that Pi, PCr, α-ATP (2 peaks, amplitude ratio 1:1), γ-ATP (2 peaks, amplitude ratio 1:1), β-ATP (3 peaks, amplitude ratio 1:2:1), and phosphodiester peaks were present. In all cases, relative amplitudes were corrected for partial saturation due to the repetition time relative to T1. Intracellular pH was calculated using the chemical shift of the Pi spectral peak relative to the PCr peak (56).
[PCr] and [Pi] data were expressed as percent change relative to resting baseline, which was assumed to represent 100%. The change in [PCr] from rest to end exercise for the <CP and >CP conditions was divided by the respective work rate to provide information on the response “gain,” that is, the change (Δ) in [PCr] per unit increase in work rate. Also, to investigate the existence of a [PCr] “slow component” during >CP exercise, the difference in [PCr] between 3 and 6 min of exercise was computed.
31P-MRS data ([PCr], [Pi], and pH) for the <CP and >CP conditions were expressed at specific absolute time points (baseline, 1 min, 3 min, 6 min, 10 min, and end exercise) and as a fraction (0%, 25%, 50%, 75%, and 100%) of the time to exhaustion (for >CP) or time to end exercise (20 min, for >CP). Statistical analysis of measurements was performed using ANOVA. Two two-way ANOVAs were used, one with repeated measures across intensity (<CP and >CP) and absolute time (0 min, 1 min, 3 min, 6 min, 10 min, and end exercise) and one with repeated measures across intensity (<CP and >CP) and relative time (0%, 25%, 50%, 75%, and 100%). When intensity-by-time interactions were significant, post hoc one-way ANOVAs were undertaken on the relevant data and Bonferroni-adjusted paired t-tests were used as appropriate to identify differences between responses at specific time points. A paired t-test and Pearson's product-moment correlation coefficient were used to investigate the relationship between the predicted and actual times to exhaustion during >CP exercise. A paired t-test was used to assess differences in the change in [PCr] between 3 and 6 min of <CP and >CP exercise. Significance was accepted when P < 0.05. Values are means ± SD unless stated otherwise.
Part I: Derivation of the Work Rate vs. Time to Exhaustion Relationship and Estimation of CP
All subjects successfully completed a minimum of three constant-work-rate exercise trials for the estimation of CP. There was no significant difference in the parameter estimates derived from the three models for CP (19.6 ± 2.2, 19.8 ± 1.9, and 19.9 ± 1.8 W for models 1, 2, and 3, respectively; F2,5 = 1.75, P = 0.22) or W′ (2.09 ± 0.76, 2.02 ± 0.65, and 1.98 ± 0.60 kJ for models 1, 2, and 3, respectively; F2,5 = 1.27, P = 0.32). The work done vs. time and work rate vs. 1/time relationships were highly linear for all subjects (R2 = 0.99 ± 0.01). The hyperbolic relationship between work rate and time to fatigue, along with the linear transformation of the relationship of work rate to 1/time to exhaustion, is shown for a representative subject in Fig. 1. The subjects' individual CP and W′ values (derived from model 3), along with the work rates applied in part II, are shown in Table 1.
Part II: 31P-MRS Assessment of Muscle Metabolic Responses to <CP and >CP Exercise
During <CP exercise, all subjects were able to complete the prescribed 20-min exercise period without difficulty. However, during >CP exercise, time to exhaustion was 14.7 ± 7.1 min. This value was not significantly different from the predicted time to exhaustion for the work rate (15.1 ± 3.3 min), and the actual and predicted times to exhaustion were significantly correlated (r = 0.87, P < 0.05).
Two-way ANOVA with repeated measures across intensity and absolute time revealed a significant intensity-by-time interaction (F5,25 = 31.29, P < 0.01) and significant main effects for intensity (F1,5 = 42.26, P < 0.01) and time (F5,25 = 44.07, P < 0.01). Subsequent one-way ANOVAs with repeated measures across time also indicated the existence of significant differences (F5,25 = 25.5, P < 0.01 for >CP; F5,25 = 47.37, P < 0.01 for >CP). The ANOVA results were essentially the same when data were expressed as a fraction of the time at the end of exercise and, thus, are not reported. The pertinent statistical comparisons are presented in Table 2 (for absolute time) and Table 3 (for relative time). For the >CP exercise bout, after an initial fall, muscle [PCr] stabilized at ∼75% of the baseline value after ∼3 min. There was no significant change in muscle [PCr] between 3 min and the end of exercise at this intensity. In contrast, during >CP exercise, muscle [PCr] did not stabilize but, rather, continued to fall with time until exercise was terminated (Table 2). The muscle [PCr] at exhaustion at this intensity was 26 ± 16% of the baseline value. Muscle [PCr] was significantly lower during >CP than <CP exercise at 1, 3, 6, and 10 min and at the end of exercise (P < 0.05). The change in [PCr] between rest and end exercise per unit work rate was almost twice as great for >CP exercise (1.86 ± 0.55 vs. 3.40 ± 0.82%Δ[PCr]/W, P < 0.01). The difference in [PCr] between 3 and 6 min of exercise (used here as an index of a possible slow component in the response) was significantly different between the two intensities (4 ± 3% and 14 ± 8% for <CP and >CP, respectively, P < 0.05). The muscle [PCr] responses for <CP and >CP exercise are shown in Fig. 2 for a representative subject and for the entire group when the data are plotted as a function of time to exhaustion.
Two-way ANOVA with repeated measures across intensity and absolute time revealed a significant intensity-by-time interaction (F5,25 = 4.99, P < 0.01) and significant main effects for intensity (F1,5 = 62.58, P < 0.01) and time (F5,25 = 19.75, P < 0.01). Subsequent one-way ANOVAs with repeated measures across time also indicated significant differences (F5,25 = 6.91, P < 0.01 for >CP; F5,25 = 21.34, P < 0.01 for >CP). ANOVA results were essentially the same when data were expressed as a fraction of the time at the end of exercise. The pertinent statistical comparisons are therefore presented in Table 2 (for absolute time) and Table 3 (for relative time). For the <CP exercise bout, muscle [Pi] rose immediately after the onset of exercise but then reached a stable level after ∼1 min. There was no significant further increase in [Pi] between 1 min (270 ± 56% of baseline) and the end of exercise (314 ± 216% of baseline). [Pi] increased more rapidly for >CP than <CP exercise; however, the increase in [Pi] between 3 min (449 ± 141% of baseline) and the end of exercise (564 ± 167% of baseline) was not significant. [Pi] was significantly greater at the end of exercise in >CP than <CP exercise (P < 0.05). Also the mean change in [Pi] per unit work rate was almost twice as great for >CP exercise, but interindividual variability in the responses precluded the attainment of statistical significance (12.1 ± 11.9 and 21.8 ± 9.3%Δ[Pi]/W for <CP and >CP, respectively, P > 0.05). The muscle [Pi] responses for <CP and >CP exercise are shown in Fig. 3 for a representative subject and for the entire group.
Two-way ANOVA with repeated measures across intensity and absolute time revealed a significant intensity-by-time interaction (F5,25 = 6.73, P < 0.01) and a significant main effects for time (F5,25 = 29.40, P < 0.01) but not intensity (F1,5 = 1.69, P = 0.26). Subsequent one-way ANOVAs with repeated measures across time indicated significant differences (F5,25 = 9.06, P < 0.01 for <CP; F5,25 = 25.07, P < 0.01 for >CP). ANOVA results were essentially the same when data were expressed as a fraction of the time at the end of exercise. The pertinent statistical comparisons are presented in Table 2 (for absolute time) and Table 3 (for relative time). During <CP exercise, muscle pH increased transiently during the 1st min before falling to reach a nadir at ∼3 min. There was no significant difference in pH from 3 min (6.99 ± 0.05) until the end of exercise (7.01 ± 0.03). Indeed, the slight recovery of pH after ∼6–8 min of exercise meant that pH at the end of exercise was not significantly different from resting pH. During >CP exercise, however, muscle pH fell precipitously to reach 6.88 ± 0.07 at 6 min, which was not significantly different from the value reached at the termination of exercise (6.87 ± 0.10). Muscle pH was significantly lower for >CP than <CP exercise at 1, 3, 6, and 10 min and at the end of exercise (P < 0.05). The change in pH from rest to end exercise per unit work rate was three times greater for >CP exercise (0.003 ± 0.003 vs. 0.009 ± 0.004ΔpH/W, P < 0.05). The muscle pH responses for <CP and >CP exercise are shown in Fig. 4 for a representative subject and for the entire group.
The principal novel finding of the present investigation was that the muscle metabolic response to exercise (as assessed by 31P-MRS) differs profoundly according to whether the work rate is incrementally <CP or >CP. The results of the study were, in part, consistent with our hypothesis and support the theory that constant-work-rate exercise above (but not below) the CP results in the utilization of the finite capacity for ATP resynthesis through nonoxidative pathways until the W′ is exhausted and exhaustion ensues. During exercise that was just below CP, a steady state in muscle [PCr] and pH was attained within 3 min and a steady state in [Pi] was attained within 1 min after the onset of exercise; moreover, the metabolic perturbation at this intensity was relatively slight: after 20 min of exercise, [PCr] had fallen to approximately three-quarters of the baseline value and pH was 7.01 (not significantly different from the value measured at rest). In stark contrast, during exercise that was just above CP, there was a precipitous change in pH and [Pi] over the first 3–6 min of exercise and [PCr] fell progressively with time until subjects were unable to sustain the work rate. The results therefore indicate that the CP demarcates an exercise intensity domain within which these muscle metabolic variables can be rapidly stabilized and sustained close to resting values from one in which they change precipitously (pH and [Pi]) or progressively ([PCr]) to values that may be limiting to performance (10, 11, 33, 52, 61).
Part I: Work Rate vs. Time to Exhaustion Relationship and Estimation of CP
That the tolerable duration of high-intensity exercise is hyperbolically related to the work rate is well established (20, 24, 37, 38). This relationship is very well described by a hyperbolic function in humans (37, 38, 44) and in other species, including the crab (16), the lungless salamander (15), the horse (34), and the mouse (6). It has been posited that the asymptote of the hyperbolic relationship, the CP, reflects the highest oxidative metabolic rate that can be sustained for a long time without fatigue, whereas the curvature constant of the relationship (the W′) represents a constant amount of work that can be performed above the CP irrespective of the rate at which the work is done (37, 38, 63). The tolerable duration of >CP exercise is therefore related to the difference between the work rate being sustained and the CP: the higher the work rate above the CP, the more rapidly the stored energy sources (O2 reserves, high-energy phosphates, and a source related to anaerobic glycolysis) should be depleted, and the more rapidly the by-products of these reactions (ADP, Pi, H+, and lactate) should accumulate in the contracting muscles and/or blood (37, 38, 44). Although there are numerous assumptions surrounding the CP concept and its predictions, some of which are problematic (21, 39), numerous studies have demonstrated that physiological responses to >CP exercise generally conform to the predictions of the CP concept (14, 19, 23, 29, 35, 36, 44, 45, 47, 54, 55). In particular, it appears that the CP broadly corresponds to the work rate at the maximal (lactate) steady state (44, 47, 54). The CP has therefore been suggested to demarcate the “heavy”-intensity exercise domain, within which steady states in V̇o2 and blood [lactate] are attainable, from the “severe”-intensity exercise domain, within which V̇o2 and blood [lactate] do not stabilize but continue to rise with time until the V̇o2 max is reached (27, 44, 65).
In the present study, the work rate vs. time to exhaustion relationship was established from three to four constant-work-rate trials performed on different days, with the tolerable duration of exercise ranging from 3 to 18 min, i.e., consistent with recommendations for accurate assessment of CP (7, 8, 21). Consistent with previous studies (23, 40; cf. 17), the nonlinear power-time model, the linear work-time model, and the linear 1/time model produced similar values for CP and W′. The linear work rate vs. 1/time model (13, 44, 47), in which the y-intercept from the linear regression model gives the CP while the gradient of the relationship gives the W′ (Fig. 1), produced R2 > 0.97 for all subjects (Table 1) and was used to calculate work rates that were ∼10% below and ∼10% above the CP.
Part II: 31P-MRS Assessment of Muscle Metabolic Responses to <CP and >CP Exercise
To our knowledge, the present study is the first to investigate the muscle metabolic responses to constant-work-rate exercise performed below and above the CP. During exercise performed at a work rate that was ∼10% below the CP (∼2 W below CP on average), it is clear that a steady state in muscle metabolites is achieved relatively rapidly (i.e., within ∼3 min) and that the muscle metabolic perturbation is relatively slight (Figs. 2–4). For example, at the end of exercise, muscle [PCr] had fallen, on average, to 68% of the baseline value, and pH had fallen by 0.05 unit from the resting value. In all cases, the subjects were able to complete 20 min of <CP exercise without significant duress. These data support the suggestion that <CP exercise can be sustained for an appreciable duration without significant fatigue development (37, 38).
The muscle metabolic response to constant-work-rate exercise performed ∼10% (∼2 W) above the CP was strikingly different from the response observed below the CP (Figs. 2–4). During >CP exercise, [PCr] fell progressively, reaching ∼26% of the baseline value at the end of exercise (Fig. 2); this continued fall in [PCr] is consistent with the predictions of the CP concept (37, 38, 44). Moreover, pH fell abruptly (reaching 6.87 at exhaustion) and [Pi] increased rapidly after the onset of exercise (Figs. 3 and 4). The nonlinearity of the muscle metabolic responses to <CP and >CP exercise is underlined when the change in the metabolic variables from rest to end exercise is expressed relative to the change in work rate, that is, for the same unit change in work rate, the change in pH was three times as great, and the change in [PCr] and [Pi] was almost twice as great during >CP exercise.
The time to exhaustion during the >CP exercise bout averaged 14.7 min, from a minimum of 6.9 min to a maximum of 23.8 min. Clearly, exercise tolerance is limited during >CP exercise. Clues to the mechanism(s) responsible for the more rapid development of muscular fatigue during >CP exercise can be obtained from the present study. Classically, fatigue during >CP exercise has been attributed to a depletion of the W′, which is considered to be a finite energy store that is predominantly (i.e., energy contained in the high-energy phosphates and a source linked to anaerobic glycolysis), although not exclusively (i.e., energy equivalent contained in stored O2 reserves in myoglobin and venous blood), anaerobic (37, 38). Data from the present study certainly indicate a significant depletion of the high-energy phosphates: [PCr] fell by an average of 74% at exhaustion during >CP exercise. It should be remembered, however, that this value represents the “mean” [PCr] in the region of muscle interrogated by MRS. Several microdissection studies have demonstrated that the depletion of muscle high-energy phosphates during exercise is heterogeneous, with the heterogeneity appearing to be related to muscle fiber type (32, 52, 53). These studies indicate that a reduction in the ability to generate muscle power during exercise might ultimately be attributed to fatigue and reduced force production in a relatively small population of muscle fibers (32, 53). Therefore, it is possible that [PCr] (and, perhaps, ATP concentration) fell to levels sufficiently low in at least some of the recruited muscle fibers during >CP exercise that the required muscle force could not be maintained. It should also be considered that, during >CP exercise, “higher-order” muscle fibers are progressively recruited to maintain muscle power output as the initially recruited fibers become fatigued until, eventually, the maximal voluntary contraction force falls below the force required for the work rate (5, 59).
The limited exercise tolerance above the CP might also be attributed to the extent, or the rate, of accumulation of metabolites that have been associated with the fatigue process, such as Pi and H+. Historically, a reduction in muscle pH has been linked to the fatigue process through H+ competition with Ca2+ for binding to troponin, interference with Ca2+ release from the sarcoplasmic reticulum, and inhibition of phosphofructokinase (11, 12, 57). More recently, the role of acidosis in muscle fatigue has been questioned, while the possible role of Pi accumulation (specifically, in its diprotonated form) has received increasing attention (1, 10, 41, 48, 61, 62, 66). In the present study, muscle [Pi] and pH changed precipitously after the onset of >CP exercise to values that might potentially have limited muscle contractile function. However, [Pi] did not increase significantly between 3 min and the end of exercise, and pH did not fall significantly between 6 min and the end of exercise. The depletion of muscle [PCr] would therefore appear to have a stronger temporal relationship with muscle fatigue development during >CP exercise. It cannot be ruled out, however, that [Pi] and H+ concentration continued to increase with time to the point of exhaustion in individual muscle fibers and that this was “masked” by interfiber, intersubject, and measurement variability, and it remains possible that metabolite accumulation was responsible for, or at least contributed to, the limited exercise tolerance above the CP. In this respect, it is noteworthy that the change in [PCr] and [Pi], which would be expected to be stoichiometric, demonstrated somewhat different temporal profiles during >CP exercise; that is, the continued fall of [PCr] with time during >CP exercise was not mirrored by a continued rise in [Pi] but, rather, [Pi] rose rapidly after the onset of exercise to reach a high but stable value. It is possible that this was caused by difficulty in precisely quantifying the changes in [Pi] during high-intensity exercise, in which a low pH can result in a broadening or splitting of the [Pi] peak or by changes in the longitudinal relaxation time (T1) for Pi (49, 50).
Muscle fatigue during high-intensity exercise is highly complex and is likely explained by a combination of interacting mechanisms (2, 10, 33, 41, 43, 48, 61). However, on the basis of the muscle metabolic responses to <CP and >CP exercise reported in the present study, it seems reasonable to suggest that the CP might differentiate exercise intensities that are principally limited by high-energy phosphate depletion and/or the effects of metabolite accumulation on cross-bridge function (>CP) from those that are chiefly limited by the availability of oxidative substrate (chiefly, muscle glycogen) and/or hyperthermia or, perhaps, central fatigue (<CP), as suggested in some of the original CP investigations (38, 44).
The kinetics of the muscle [PCr] response to exercise have been reported to be similar to those of pulmonary V̇o2, and this has been suggested to reflect feedback control of oxidative phosphorylation through one or more of the reactants and/or products of high-energy phosphate hydrolysis (3, 49, 50). It is therefore of interest to note that the response profiles of muscle [PCr] in the present study had characteristics similar to those that would be expected for pulmonary V̇o2 during <CP and >CP exercise (42, 65). The present data also confirm the existence of a [PCr] slow component during high-intensity constant-work-rate exercise (18, 30, 50, 51): the change in [PCr] between 3 and 6 min of exercise was significantly greater for >CP than <CP exercise. As recognized previously, the existence of a [PCr] slow component implies an increasing phosphate cost of force production as exercise proceeds, an effect that might be attributed to the effects of fatigue (e.g., on the ATP cost of ion pumping) and/or an increasing contribution of type II muscle fibers to force production (18, 46, 50). Conley et al. (9) argue that the highest sustainable oxidative metabolic rate (i.e., ostensibly the CP) is determined by the interaction between glycolytic flux and oxidative phosphorylation and can be understood with reference to the creatine kinase reaction. Specifically, they propose that the increased H+ concentration resulting from a greater activation of glycolysis during high-intensity exercise limits the extent to which ADP concentration ([ADP], believed to be a key signal activating mitochondrial respiration) can rise, such that the highest steady-state V̇o2 (and the highest sustainable work rate) will occur at ∼80% of the respective maxima achieved during an incremental (V̇o2 max) test (4). However, ADP can be made to increase, allowing exercise to be continued for a short period (∼5–10 min) and V̇o2 to reach its maximal rate, if [PCr] is reduced to offset the fall in pH (9, 31). The continued fall in muscle [PCr] during >CP exercise might therefore be necessary to enable the appropriate [ADP] stimulus to oxidative phosphorylation but is also characteristic of a fatiguing muscle with a limited capacity for continued force generation.
Perspectives and Significance
We have demonstrated that the dynamics of the muscle metabolic response to exercise (as assessed using 31P-MRS) can be differentiated according to whether the exercise is performed just below or just above the CP, therefore providing empirical verification of the CP construct. Specifically, the CP appears to demarcate a range of work rates within which muscle [PCr], [Pi], and pH can be rapidly stabilized and sustained close to resting values from those within which they change precipitously (pH and [Pi]) or progressively ([PCr]) to values that might predispose the muscle to the development of fatigue. The CP concept therefore has theoretical and practical utility in exercise physiology as a model for exploring the mechanistic bases of muscular fatigue and the determinants of human exercise tolerance.
The authors thank Anni Vanhatalo for assistance with nonlinear modeling of the power-time relationship.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society