INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

AT THE ONSET of light- to moderate-intensity submaximal exercise (i.e., below the ventilatory threshold), O₂ uptake (O₂) measured at the mouth increases as a two-phase response. Phase 1, lasting ~15 s, is a function of increased return of venous blood, most of which was pooled in the periphery before exercise (11). Because phase 1 is mainly a consequence of increased venous return, it is often called the cardiodynamic phase. Phase 2, the primary phase, of the adaptive process reflects the change in muscle oxidative metabolism as venous return continues to increase and more O₂ is extracted with exercise. Phase 2 of the submaximal exercise response has been modeled with an exponential function (8, 25, 40).

As the intensity of exercise increases above the ventilatory threshold but remains below peak O₂ (O_{2 peak}), O₂ adapts after a delay with an additional phase (i.e., phase 3) during constant-load exercise (8, 9, 14). Phase 3 has been described as an additional phase in which the O₂ adds on top of the metabolic requirement of that work rate (31). The results of some experiments indicated no difference in phase 2 kinetics of O₂ at these heavier work rates between ventilatory threshold and O_{2 peak} compared with the responses during light-intensity exercise (7, 9, 31). In contrast, other data, some obtained by the same research groups who found no difference, were consistent with slower O₂ kinetics during phase 2 for the heavier work rates below O_{2 peak} (10, 14, 15, 27, 28, 41). These different observations raise questions about the mechanisms responsible for establishing the adaptation of oxidative metabolism at the onset of heavy submaximal exercise.

For work rates near or above O_{2 peak}, there has been little research of the kinetics of O₂ (9, 20, 29). Margaria et al. (29) suggested that the rate of increase in O₂ should be proportional to the difference between required O₂ and actual O₂ for work rates above O_{2 peak}. They concluded that there was no difference between the time constant () across a range of work rates that caused exhaustion in 30-120 s. Furthermore, they suggested that this was not different from the  measured during submaximal exercise. However, a technical limitation is apparent in their study, inasmuch as they show measured O₂ values during heavy exercise that exceeded the reported values of O_{2 peak} in their subjects. Hebestreit et al. (20) observed faster kinetics for phase 2 at 100-130% O_{2 peak} in boys and men but noted that the curve-fitting procedure might have caused an apparent speeding of O₂ kinetics at the higher work rate. Therefore, new data are required to resolve this question.

The precise regulation of the cardiovascular system to match O₂ transport to O₂ utilization is achieved by a control system that integrates various feedforward and feedback signals to achieve a coordinated distribution of blood flow between exercising and nonexercising tissues. With a transition between two levels of energy expenditure, error signals are established in proportion to the difference between the present and the required values of O₂ transport and muscle O₂ (21, 23, 41). The physiological responses to these error signals differ between individuals with varying levels of physical fitness (19) as well as between exercise intensities, as considered above. It is desirable to have a rapid adaptation to increased metabolic demand, inasmuch as this will minimize accumulation of an O₂ deficit and cause less production of lactic acid. An understanding of the responses during transition states can provide important information about the mechanisms that limit the adaptive processes.

The ability of the cardiovascular system to achieve adequate O₂ transport throughout the early phase of very heavy submaximal work rates has been questioned (15, 27, 28). This would lead one to suspect that, during the high demand of work rates above that required to achieve O_{2 peak}, O₂ transport might also be limiting. Therefore, in contrast to the conclusion of Margaria et al. (29), we hypothesized that O₂ kinetics would be slowed for work rates near or above O_{2 peak} compared with subventilatory threshold work rates. The purpose of this study was to examine the O₂ during exercise at ~96 and 125% O_{2 peak} and to compare the results with those from exercise at 57% O_{2 peak}. To achieve the best mathematical description of the kinetics response for O₂, it was necessary to incorporate some modifications of the exponential fitting based on estimates of the predicted metabolic requirements (O_{2 pred}) of the exercise load. Thus we present the justification for this approach and provide an indication of potential error in estimation of kinetic parameters. These results are considered with respect to a model of dynamic cardiovascular control.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Eight healthy young men (20.8 ± 0.7 yr) participated in the study. After receiving complete written and verbal details of the experimental protocol, the subjects signed a consent form approved by the Office of Human Research of the University of Waterloo. Ventilatory threshold, measured as the point of increase in ventilatory equivalent for O₂ (E/O₂) with no change in ventilatory equivalent for CO₂ output (E/CO₂) (22), occurred at 2,480 ± 135 ml/min or ~62% O_{2 peak}.

Initial testing. After the subjects were familiarized with the laboratory setting, each completed two incremental exercise tests to exhaustion on an electrically braked cycle ergometer (Lode Excalibur, Groningen, The Netherlands). Subjects were allowed to select a comfortable pedal frequency near 80 cycles/min. In the first, after a 5-min period at 30 W, the work rate increased at a rate of 30 W every 3 min (Fig. 1). The purpose of this test was to establish the O₂ at each submaximal work rate. The second test also began with 5 min at 30 W. After this, the work rate increased by 30 W every 1 min. The O_{2 peak} was defined as the highest O₂ over a 15-s period during the first or second test. The rationale for two separate tests was that the total duration of the first test was sufficiently long that most subjects achieved lower O_{2 peak} in these tests because of cumulative fatigue. To determine the O₂-work rate relationship, O₂ values below ventilatory threshold in the slow-incrementing test were examined for a linear relationship to work rate. Additional values were considered only if the correlation coefficient of the linear regression did not decrease. This provided a conservative estimate of the lowest value of the slope of O₂ and work rate (Fig. 1). On the basis of this linear regression, the work rate and the metabolic cost were predicted for exercise approximating 57, 96, and 125% of the individual subject's O_{2 peak}.

View larger version (22K): [in this window] [in a new window]   Fig. 1.   O2 uptake (O2) for a single subject as a function of time for incremental exercise tests to exhaustion. Data points are mean values in final 30 s of each 3-min stage at a work rate. Only work rates less than that at ventilatory threshold (horizontal dotted line at O2 = 2,600 ml/min) are included in regression used to predict O2-work rate relationship for work rates near and above peak O2 (O2 peak).

Constant work rate tests. In subsequent testing, each subject completed four repetitions of the work rate at 57% O_{2 peak} and two repetitions of each of the 96 and 125% work rates. On a given test day, the subjects completed the 57% test first and then rested for 10 min before the 96 or 125% test. There is no evidence that the light work rate tests might alter O₂ kinetics in the two heavier-exercise tests (15). Exercise began with 4-5 min of cycling at a baseline work rate of 30 W. The work rate then increased as a step function to the required level for an additional 5 min for the 57% tests and as long as the subjects could maintain it for the 96 and 125% tests. No warning was given at the time the work rate was increased. Subjects were required to maintain a pedal frequency of ~80 cycles/min. Breath-by-breath data for O₂ and the mean heart rate over each breath were collected continuously throughout the tests.

Measurement of O₂. Breath-by-breath O₂ was measured on a computerized system (First Breath, St. Agatha, ON, Canada) that sampled inspired and expired gas volumes with an ultrasonic flowmeter (Kou Consulting, Redmond, WA) and fractional concentrations of O₂, CO₂, and N₂ by mass spectrometry (model MGA-1100A, Marquette, Milwaukee, WI). The flowmeter was placed in direct line with the mouthpiece, and air was continuously sampled near the mouth. The total dead space of this configuration was ~100 ml. O₂ was calculated with an algorithm that allowed for breath-by-breath changes in lung gas stores and with computation of the effective lung volume (24). Calibration of the flow/volume sensor was achieved immediately before each test by manually pumping a 3-liter syringe through the flowmeter at a rate similar to that achieved during the exercise test. The mass spectrometer was calibrated with two precision gas tanks that spanned the ranges of gas concentrations encountered during the tests. Corrections were made for inspired and expired water vapor and temperature to yield breath-by-breath values expressed in milliliters per minute STPD. Heart rate was measured with an electrocardiograph (model 7803A, Hewlett-Packard) with a standard bipolar electrode placement.

Data analysis. The individual test data were linearly interpolated between points to give values at 1-s intervals (Fig. 2). The multiple repetitions at each work rate for an individual subject were averaged and then processed by two different means. The first was a nonlinear curve-fitting procedure described previously for application to submaximal exercise intensities (25). The second incorporated a semilogarithmic transformation of the data and is similar to that used previously (21, 29, 41). The outcome of this analysis would be almost identical to that obtained by nonlinear curve fitting, with the amplitude of the second term fixed at the predicted value.

View larger version (18K): [in this window] [in a new window]   Fig. 2.   O2 as a function of time during constant-load exercise for same subject as in Fig. 1. Actual or predicted O2 (O2 pred) values for this subject were 62% (), 99% (), and 129% () O2 peak. Individual data points are 1-s average values obtained from multiple repetitions. Time constants for 2nd component (2) from nonlinear exponential curve fitting were 30.5, 20.6, and 25.7 s for 62, 99, and 129% O2 peak, respectively.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Logarithms of difference between O2 pred and measured O2 at any time [O2(t)] for same data sets presented on linear axes in Fig. 2. Top: 62 and 99% O2 peak tests; bottom: 129% O2 peak test. Small dots, ±5% error range for each test (i.e., 94-104% O2 peak and 124-134% O2 peak). A linear regression was fit to transformed data over approximately same range as that used in nonlinear curve fitting to estimate time constant of 2nd component (). For this subject, estimate of time constant for 62% O2 peak test was 30.6 s; estimates of time constants (and range determined from ±5%) were 39.5 s (32.7-46.2 s) and 43.4 s (39.5-47.7 s) for 99 and 129% O2 peak tests, respectively.

Statistics. Because we used different methods to analyze the data for the 57% vs. the 96 and 125% tests, we had to divide the statistical analysis. First, we focused on the ₂ value from the nonlinear curve fitting by completing a one-way repeated-measures ANOVA on the main effect of work rate. Next, we compared the ₂ value from nonlinear curve fitting with the log(₂) value from the semilogarithmic transformation for the 96 and 125% tests only by two-way repeated-measures ANOVA with main effects of fitting procedure and work rate. Significant differences were accepted for P < 0.05. Values are means ± SE.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Incremental exercise tests. The peak work rate attained in the slowly incrementing exercise tests was 296 ± 14 W. At this work rate, the peak O₂ was 3,857 ± 154 ml/min and the peak heart rate was 189 ± 4 beats/min. During the more rapidly incrementing exercise tests, peak values were work rate 353 ± 14 W, O₂ 3,963 ± 167 ml/min, and heart rate 187 ± 4 beats/min.

57% O_{2 peak} tests. O₂ increased rapidly at the onset of exercise at 57% O_{2 peak} (Fig. 2, Table 1). The first component of the exponential model accounted for ~25% of the total response and lasted ~18 s. The ₂ value for the O₂ kinetics response was 29.4 ± 4.1 s.

View this table: [in this window] [in a new window]   Table 1.   Parameter estimates for exponential curve fitting of O2 response at three different exercise intensities

96 and 125% O_{2 peak} tests. The amplitude of phase 1 of the kinetic response at the two higher intensities of exercise was slightly greater than that observed at 57% O_{2 peak} exercise, but it represented a smaller percentage of the total response (Table 1). Phase 2 started ~12-14 s after the onset of exercise at these two higher intensities. The ₂ values for the O₂ kinetics responses were 22.1 ± 2.1 and 16.3 ± 3.1 s for the 96 and 125% tests, respectively. Phase 3 started after ~40-73 s (Table 1), with the shorter duration for the 125% tests. The ₃ was considerably slower than for the phase 2 response.

View this table: [in this window] [in a new window]   Table 2.   Estimates of 2 of O2 in phase 2 component at high work rates with exponential and semilogarithmic models showing ±5% error estimates

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this study, we examined the rate of change in O₂ during the transition from baseline to light or very heavy exercise. In particular, we focused on phase 2 of the adaptive response, because this phase has been taken to reflect oxidative metabolism in the exercising muscles (6, 17) and to provide indications of the rate-limiting steps (23, 37, 39). It should be noted that phase 2 during the two higher intensities had an average duration of only 28 s for the 125% test to 59 s for the 96% test. The magnitude of O₂ increase during phase 2 (see G₂ in Table 1) was ~1,700 ml/min, which amounted to 60% of the increase above baseline cycling. Thus our analysis concentrated on a specific region of the adaptive response.

When we fit the data with the semilogarithmic transformation, we found significantly slower kinetics of O₂ at the work rates that corresponded to ~96 and 125% O_{2 peak} compared with those measured at 57% O_{2 peak}. These data supported our hypothesis that kinetics would be slower at these high exercise intensities because of a limitation in O₂ transport. The results further showed the limitation of nonlinear curve fitting to attempt to describe the kinetics of O₂ at high work rates. With the curve-fitting approach the "required" O₂ determined from the amplitude of the asymptote of the second component achieved only ~80% of O_{2 peak} in contrast to the 96-125% O_{2 peak} that was actually required. Our findings contrast with those of Margaria et al. (29), who reported no change in time course of O₂ as work rate increased above O_{2 peak}. As noted in the introduction, methodological problems might have limited their study. Our results also provide an explanation for the apparent acceleration of O₂ kinetics when nonlinear curve fitting was applied to data from 100 to 130% O_{2 peak} (20).

Investigation of O₂ kinetics at maximal and supramaximal exercise intensities requires a number of assumptions, and there are inherent limitations. We had to assume that we could predict the energy demand of high-intensity exercise on the basis of a linear extrapolation from submaximal work rates. Furthermore, it was necessary to assume that the energy demands of these high-intensity work rates remained constant throughout the period over which we evaluated the kinetics. Given the potential for error in the linear extrapolation, we examined the influence of error on the calculated time constant.

Linear extrapolation to establish energy demand. Several researchers have presented the assumptions and limitations inherent in attempting to use the energy cost at submaximal exercise intensities to predict those at higher intensities (3, 4, 30, 38). Across a range of submaximal work rates up to approximately the ventilatory threshold, O₂ increases as a linear function of work rate (38). The O₂ at these intensities stays relatively constant with continued, constant-load exercise (33). For work rates above this level, the O₂ varies as a function of time while the work rate remains constant so that O₂ could equal the predicted value but is more likely to be higher than predicted as the exercise time is prolonged. For intensities slightly above ventilatory threshold, O₂ will normally increase slightly and then remain at this higher O₂ (10, 33, 38, 41). However, as a given constant-load exercise intensity approaches O_{2 peak}, the O₂ will continue to increase and often reach O_{2 peak} at these supposedly submaximal work rates (2, 10, 38). The mechanisms responsible for the elevated O₂ during high-intensity exercise will be considered in Constant energy demand of high-intensity exercise.

Constant energy demand of high-intensity exercise. We assumed that energy demand would be unchanged from the start to the end of exercise. This is probably true for the 57% O_{2 peak} tests, inasmuch as little change in O₂ is observed after 3 min of exercise. It is unlikely that the total energy demand would remain constant with time for the 96 and 125% O_{2 peak} tests, inasmuch as O₂ often increases with time during heavy constant-load, but sub-O_{2 peak}, exercise (2, 9, 10).

Derivation of the model. The exponential model that we used is similar to that employed in most other research of O₂ kinetics. The inclusion of one, two, or three exponential components is based on goodness of fit derived from various nonlinear curve-fitting functions (25, 27, 40). However, each of these models assumes that the asymptote for each component represents the "true" end point. We believe this to be the case for the 57% O_{2 peak} tests. The difference between the measured O₂ at any time after the onset of exercise and the required O₂ (i.e., steady-state O₂) during this 57% test represented the error signal. The magnitude of the error signal was reduced in an exponential manner, as typical of a linear first-order system with a  (21). By forcing the curve fitting of the phase 1 component to be complete before the start of the phase 2 component, we avoided any interference with the estimate of the dynamic response of phase 2. In this sense, the model was identical to that employed by Barstow et al. (8). Furthermore, by allowing the phase 3 component to begin at some time after the origin of the phase 2 component (TD₃ = 40-73 s; Table 1), we did not bias the estimate of ₂.

O₂ and heart rate in heavy exercise. Previous research has not reached a firm conclusion about the effect of work rate on the time course of O₂ across a range of work rates. Several early studies suggested that O₂ kinetics were slower as the work rate increased, with a clear slowing at work rates above ventilatory threshold (10, 27, 31, 41). Next, there were studies that suggested that O₂ kinetics were not different when ₂ were compared for work rates above and below ventilatory threshold (7, 9). More recently, several investigations have indicated that kinetics of phase 2 are indeed slower with heavy exercise (14, 15, 28). Part of the reason for the different outcomes of this research stems from the different mathematical modeling approaches used. Early investigations that used a model in which phases 1, 2, and 3 were mixed showed slowing of the responses (10, 27, 31, 41). As models were developed that split the phases and treated each separately, the outcome of research indicated no difference (7, 9) or slower kinetics (14, 15, 28). Of the few studies that have actually looked at kinetics during exercise at 100% O_{2 peak}, ₂ was not different from that at lower work rates (9, 29). Indeed, when we applied a model similar to that of Barstow and Molé (9), we found slightly faster ₂ for the 96% tests and significantly faster ₂ at 125% O_{2 peak}. This latter observation was consistent with previous studies of very-high-intensity exercise compared with exercise intensity below the ventilatory threshold (20) and emphasizes the importance of the choice of model.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Heart rate responses for same subject shown in Figs. 1-3 in transition from baseline work rate to exercise requiring ~62% () 99% (), and 129% () O2 peak. Note similar pattern for heart rate and O2 shown in Fig. 2. Subject's peak heart rate during incremental exercise was 180 beats/min.

Cardiovascular control model. Feedback and feedforward components have been identified in regulation of the cardiovascular adaptation to exercise. The feedforward component is proportional to the activation of the motor cortex and appears to play a role, at least for static exercise, in the regulation of heart rate by vagal withdrawal (34). A feedback regulatory model requires that the controlled variable (muscle blood flow or O₂ transport) be altered in proportion to an error signal. Two competing factors, the need for increased muscle blood flow and the necessity to maintain arterial blood pressure, interact to determine the appropriate signals to activate an increase in cardiac output and to evoke a coordinated cardiac and vasoconstrictor response. Neither the error signal nor the specific mechanism has been defined for the regulation of the cardiovascular response to support aerobic metabolism. Neural pathways have been identified that are responsive to mechanical or chemical signals (1). In very heavy exercise, a chemical error signal might be generated by the accumulation of metabolic by-products in proportion to the anaerobic contribution to energy supply in addition to factors, such as extracellular potassium, that are associated with muscle contraction itself. With rapid depletion of phosphocreatine and lactate accumulation, there will be increases in the concentrations of inorganic phosphate and hydrogen ion. These factors, as well as adenosine or other vasoactive metabolites, will act directly on the local vasculature to cause vasodilation, which could cause a drop in arterial blood pressure at the onset of exercise (36) and activate the arterial baroreflex mechanisms in support of increased blood flow.

Conclusions. The present results are consistent with the hypothesis that O₂ kinetics at the onset of exercise near or above O_{2 peak} are slower than observed for submaximal exercise. These data support the notion that adaptation to exercise at high intensities is limited by O₂ transport (15, 28, 38). The observation that O₂ at any time point after the onset of exercise is higher for higher work rates should not be confused with faster kinetics. Previous investigations that referenced the O₂ response to the estimated plateau at the end of phase 2 gave an incorrect estimate of the kinetics of O₂ because the asymptote was not determined by the metabolic error signal but by the limitation of O₂ transport. That is, it must be seen that the cardiovascular and metabolic systems attempted to adapt on the basis of the larger error signal that occurs with this very heavy exercise. When O₂ was referenced with respect to the magnitude of O_{2 pred}, the slower adaptation [log(₂) in this study] indicates that the muscle O₂ cannot adapt at the same proportional rate for very heavy exercise as observed at lower work rates because of functional limitations of O₂ transport. Estimation of O₂ during very heavy exercise required certain assumptions. As indicated previously, our assumptions probably tended to underestimate O_{2 pred}. If the predicted value had been greater because of changes in muscle recruitment or some other factor, then the kinetics of O₂ would have been even slower.