INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

HORSES ARE REMARKABLE athletes capable of achieving mass specific O₂ uptake (O₂) levels which are two to three times those of elite human athletes. When compared with humans, the kinetics by which they achieve these high O₂ values are also extremely fast (6, 11, 15, 16). Although O₂ kinetics in the horse are quantitatively different from those present in humans, there is a qualitative similarity between the two regarding the exercise on-transient kinetics (11). Specifically, O₂ on-transient kinetics in the exercising horse demonstrate the same exercise-intensity dependency as in humans and can be modeled in a similar fashion. Whether this holds true for the off-transient in recovery has not been determined in the horse.

The off-transient response has been examined in the horse (16) and pony (15). However, it has not been modeled systematically across the spectrum of exercise-intensity domains from moderate [i.e., below the lactate threshold (<T_lac)] to heavy (>T_lac) and severe [i.e., O₂ projecting to maximal O₂ (O_{2 max})]. In humans, the O₂ off-transient response traditionally has been separated into two components: a fast component attributed to replenishment of creatine phosphate (CP) stores within the muscle and a slow component attributed to removal of lactic acid from the muscle and blood. This explanation is now recognized as too simplistic and largely erroneous. For example, the kinetics of O₂ recovery and either CP restoration or lactate disappearance have been dissociated in a variety of models (see Ref. 7 for review).

In humans, there is no consensus as to whether the O₂ off-transient is monoexponential (13) >T_lac or, alternatively, whether fast and slow components can be resolved (4). The purpose of the present investigation was to address this issue in the horse by characterizing the off-transient behavior by using a model validated previously for the on-transient and to determine whether O₂ off-transient behavior is exercise intensity domain dependent in this animal.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Animals. Six geldings (5 Thoroughbreds and 1 Quarter Horse; age, 4-14 yr; weight, 459-603 kg) were used for this study. The animals were housed in a dry lot (loafing shed with paddock) with free access to water and salt, and they were fed alfalfa, grass hay, and concentrate twice a day. They were dewormed and vaccinated at regular intervals and were acclimatized to exercise on a high-speed treadmill (SATO, Uppsala, Sweden). Their peak O₂ (O_{2 peak}) achieved during an incremental treadmill exercise test to fatigue was 70.5 ± 3.6 l/min (130.5 ± 6.6 ml · kg¹ · min¹). The protocol for this experiment was approved by the Institutional Animal Care and Use Committee of Kansas State University.

Animal preparation. Animal preparation, exercise protocol, and sampling techniques have been described previously (11). Briefly, all animals were instrumented with a thermistor catheter (Columbus Instruments, Columbus, OH) and a 7-Fr microtip pressure transducer (with lumen, SPC-471A, Millar Instruments, Houston, TX) that was advanced into the pulmonary artery to measure blood temperature and to sample mixed venous blood. Location of the pressure transducer was verified by the characteristic pressure wave. The Millar pressure transducer was calibrated before and immediately after each run by using a mercury manometer. No transducer drift was detected across any of the runs.

Respiratory gas measurements. O₂ and CO₂ output (CO₂) were measured by using an open-flow system described previously (11). Ambient air was drawn through a loosely fitting face mask past the horse's nose and mouth at a rate sufficient to prevent escape of expired gas (i.e., up to 7,000 l/min, depending on running speed). Expired O₂ and CO₂ concentrations were measured by using a gas analyzer (model 1100 Medical Gas Analyzer, Perkin-Elmer, Pomona, CA) and were recorded at a sample rate of 100 Hz by a computer-based data-acquisition system (DATAQ, Akron, OH). The pressure differential across the flow nozzle was determined by means of a differential pressure transducer (model MC1-3-871, Validyne, Northridge, CA) for determination of flow. Relative humidity and temperature in the open-flow system were also measured continuously (HS-ZCHDT-2R, Thunder Scientific, Albuquerque, NM). Ambient temperature and barometric pressure were measured before each run was started. These variables were used to correct O₂ and CO₂ to STPD conditions.

Exercise protocol. Data for recovery were taken following three different exercise protocols. Two exercise bouts (moderate and heavy domains) consisted of 6-min square-wave work (~50 and ~85% of the speed at which O_{2 peak} was attained) following a warm-up at 3 m/s. All runs were performed on the level treadmill (i.e., 0% grade). For recovery, the velocity was decreased to 3 m/s (in <5 s) for 800 m of trotting exercise (~4 min). The recovery from the severe domain followed the incremental test. The increments were selected so that each horse attained its speed at the point of fatigue in 8-10 min. Following a warm-up of 800 m at 3 m/s, the treadmill speed was increased at 1-min intervals in even increments of 1-1.5 m/s until the horse could no longer keep up with the treadmill. Maximal speed was 13-15 m/s, depending on the capability of each horse. The incremental exercise test was used to determine 1) T_lac measured directly from mixed venous blood samples and 2) O_{2 peak}. Blood samples for measurement of plasma lactate concentration ([lactate]) were taken during the last 5 s of each increment and at 2 and 4 min of recovery. Plasma [lactate] was measured with a lactate analyzer calibrated according to the manufacturer's specifications (model 23L, Yellow Springs Instruments, Yellow Springs, OH).

Data analysis. Values for O₂ and CO₂ were calculated from the measured variables and smoothed by using a 10-s moving average to reduce noise from the respiratory cycle, breath-to-breath tidal volume variation, and breath holding by the horse to give values for each second. The time course from the downward transition at the end of the exercise protocol to 3 m/s (~4 min) was analyzed. Values for the on-transient kinetics have been previously determined from the same exercise runs (11).

Modeling procedure. The time course for O₂ kinetics at exercise onset has been described in a previous paper (11). The off-transients were modeled in a similar manner by using the following equations (Fig. 1). For t less than time delay (TD) (phase 1) For t >TD (phase 2 + slow component) where t is the time in seconds from the downward transition; EEO₂ is the end-exercise value for O₂; A₀, A₁, and A₂ are the asymptotic values for the exponential terms; ₀, ₁, and ₂ are the time constants of the A₀, A₁, and A₂ responses; and TD is the common time delay. Phase 1 was terminated at the start of phase 2 (t = TD) and assigned the value of the exponential function at that time [A₀' = A₀(1  e^TD/₀)]. As previously described in human models (4), a single TD was used to describe the function, because it was assumed that the factors involved in both the fast and slow component would be in play at the end of exercise. Parameters were estimated by using the Levenberg-Marquardt algorithm to minimize the sum of squares. A two-phase model, with a single- or a double-exponential phase 2, was compared for all work rates. In cases where the more complicated (double-exponential phase 2) model produced a smaller sum of squares, the parameters were evaluated for significance and accepted when P < 0.05. If P > 0.05 and the dependencies exceeded 0.99, the second term was dropped. Plots of residuals were also examined to help determine the appropriate fit (Figs. 2 and 3).

View larger version (13K): [in this window] [in a new window]   Fig. 1.   Models for exercise; moderate (A), and heavy or severe (B). EEO2, end-exercise value for O2 uptake (O2); baseline, preexercise O2 at 3 m/s; TD, time delay; A0', value of A0 at TD; A1, asymptotic values for time constant 1; A1' is A0' + A1. O2 180  baseline, difference between O2 at 180 s of recovery and preexercise baseline value.

View larger version (15K): [in this window] [in a new window]   Fig. 2.   Representative recovery O2 curve for 1 horse in heavy-exercise domain. Top: delay and monoexponential fit. Note good fit to monoexponential phase 2. Plots of residuals for 2 models show no better fit with the more complex double-exponential phase 2 (bottom right) vs. monoexponential phase 2 (bottom left).

View larger version (14K): [in this window] [in a new window]   Fig. 3.   Profile from only horse in which data were significantly better fitted by the double-exponential phase 2 response. Plot of residuals shows improvement with more complicated model (double exponential; bottom right) over simpler model (monoexponential, bottom left).

O₂ deficit and excess postexercise O₂ consumption (EPOC). At moderate work rates, the O₂ deficit was calculated as the area above the O₂ curve and below the asymptotic value for that exercise bout. At heavy work rates, the deficit was divided into two portions, the first being derived from the asymptotic value of the fast component, and the second derived by using the EEO₂ value (Fig. 4).

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Calculation of O2 deficit and excess postexercise O2 consumption (EPOC) for moderate (A) and heavy (B) exercise. O2 deficit fast component, diagonal hatched area; slow component, light gray area. EPOC fast component, vertical hatched area. See text for details.

Statistical analysis. Differences for all variables between moderate- and heavy-intensity exercise were determined by using paired t-tests, as were differences between on- and off-transients. Comparisons among the three exercise protocols were made by ANOVA for repeated measures, with the Tukey test performed to detect specific differences. If the data failed the normality test, Friedman's repeated measures ANOVA on ranks was used with the Student-Newman-Keuls method to detect specific differences. Data are presented as means ± SE. Plots of accumulated lactate vs. excess recovery O₂ (as determined by subtracting the baseline value from the value of O₂ 180 s into recovery, during the apparent steady state following the completion of the fast phase of recovery) and temperature increase vs. excess recovery O₂ were analyzed by linear regression. Significance was accepted at P  0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

General trends. As with the on-transient responses, the recovery kinetics from all work intensities were well characterized by a three-phase exponential response with a monoexponential primary component (phase 2) starting after a TD (Fig. 5). In one Thoroughbred horse (during heavy exercise), phase 2 was best described by a double exponential (Fig. 3). There were no significant differences in the off-transient TDs or ₁ among the three exercise-intensity domains. As expected, preexercise baseline O₂ values were not different among the exercise protocols, but EEO₂ values were different, as were those at 180 s postexercise (Table 1).

View larger version (10K): [in this window] [in a new window]   Fig. 5.   Off-transient response for 1 horse from exercise of moderate (A), heavy (B), and severe (C) intensity. Straight solid line, baseline; curved solid line, three-phase (monoexponential phase 2) model; dashed line, response of 1 horse. Note excellent fit in each instance.

View this table: [in this window] [in a new window]   Table 1.   Baseline O2, on-transient and end-exercise variables, off-transient variables, and parameters and time constants

Moderate-intensity exercise. The off-transients were well fitted by a monoexponential phase 2 response (₁ = 16.4 ± 2.0 s) with a TD (18.0 ± 1.0 s). The sum of squares for some of the double-exponential fits were smaller, but because they did not significantly improve the fit, they were dropped (see METHODS). The recovery O₂ at 90 s (TD + 4 s) was not different from preexercise baseline values. The off-transient ₁ for moderate exercise was not different from the on-transient response (11.2 ± 1.4 s) for the same exercise bouts (Table 2).

View this table: [in this window] [in a new window]   Table 2.   Comparison between off and on responses for time delay and 1

Heavy-intensity exercise. The off-transients were well fitted by a monoexponential phase 2 response (₁ = 13.5 ± 1.0 s) with TD (17.6 ± 1.1 s) in five horses and a double-exponential phase 2 response (₁ = 10.5 s, ₂ = 180 s; A₂ = 4.1 l/min) with TD (15.6 s) in one horse. The values for O₂ had not returned to baseline (12.6 ± 0.7 l/min, 23.4 ± 1.3 ml · kg¹ · min¹) by 180 s postexercise (15.6 ± 1.2 l/min, 29.0 ± 2.3 ml · kg¹ · min¹). There was, however, no significant decrease in O₂ from 90 to 240 s postexercise. The magnitude of the difference between preexercise baseline and the end of the phase 2 response values (t = 180 s) was 3.0 ± 1.0 l/min, (5.6± 1.8 ml · kg¹ · min¹), which amounted to 6% of the EEO₂  baseline difference. The off-transient ₁ for heavy exercise was faster (13.5 ± 1.0 s) than the on-transient response (23.3 ± 3.8 s) for the same exercise bouts (Table 2).

Severe exercise. As in moderate and heavy exercise, the severe exercise off-transients were well fitted by a single monoexponential phase 2 response (₁ =14.6 ± 0.3 s) with a TD (17.8 ± 2.0 s) in all six horses. However, the O₂ for all horses had not returned to baseline (12.2 ± 0.6 l/min, 22.5 ± 1.0 ml · kg¹ · min¹) by 3 min postexercise (17.8 ± 1.4 l/min, 33.0 ± 2.7 ml · kg¹ · min¹). As was the case for heavy exercise, there was no significant decrease in O₂ from 90 s (TD + 4 s) postexercise to 240 s postexercise by pairwise analysis. The magnitude of the difference between baseline and end-phase 2 values (t = 180 s) was 5.7 ± 1.1 l/min (10.6 ± 2.1 ml · kg¹ · min¹), which amounted to 10% of the EEO₂  baseline difference.

O₂ deficit and EPOC. O₂ deficit and EPOC were not different for moderate work rates (Table 1). Also, the fast portion of O₂ deficit and EPOC were not different for heavy work. However, the combined O₂ deficit (i.e., fast and slow component) for heavy work was greater than the calculated EPOC (see METHODS for calculations).

Lactate and temperature. The magnitude of the differences between baseline and 180-s O₂ values (excess recovery O₂) were correlated with the EE increase in pulmonary arterial plasma [lactate] (r = 0.94, P < 0.001; Fig. 6). In contrast, there was no correlation between the excess recovery O₂ at 180 s for the heavy and severe work rate and pulmonary arterial blood temperature (r = 0.45, P > 0.05).

View larger version (13K): [in this window] [in a new window]   Fig. 6.   Plasma lactate concentration at end exercise vs. elevated postexercise O2 (O2 at time = 180 s)  baseline O2.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The principal original finding of the present investigation is that O₂ kinetics across the recovery transient from higher running speeds in the horse are independent of exercise intensity in regard to the TD and the predominant fast-phase response, i.e., there is no slowing of ₁ with work intensities >T_lac. Moreover, with respect to heavy and severe exercise, there is no discrete slow component of the recovery O₂ response that can be resolved. This is in marked contrast to the on-transients, where there is significant slowing of the fast component along with the addition of a slow component >T_lac (11). Off-transients <T_lac (i.e., moderate domain) are well fitted by a three-phase (monoexponential phase 2) model with a return to baseline levels within 90 s. Above T_lac, off-transients were well fitted by the same model in five of six horses, but O₂ did not return to that present before the exercise challenge, resulting in what appeared to be an elevated O₂ baseline (i.e., no detectable changes in O₂ from 90 to 240 s of recovery). The magnitude of this elevated baseline effect increases, as exercise intensity increases, in proportion to the absolute plasma [lactate]. The kinetics of this O₂ component could not be ascertained by these data, as O₂ appeared to be in a steady state for the period sampled. For heavy exercise, the magnitude of this elevated baseline was 3.0 ±1.0 l/min (5.6 ±1.8 ml · kg¹ · min¹) compared with the on-transient slow component of O₂ of 5.8 l/min (10.4 ± 4.9 ml · kg¹ · min¹). Following severe exercise, the baseline elevation was even larger (5.7 ±1.1 l/min, 10.6 ± 2.1 ml · kg¹ · min¹). This dissymmetry between on- and off-transients is closely comparable to that found by Paterson and Whipp in humans (13) where the off-transient kinetics for high work rates were not slowed as the on-transients were. However, these responses contrast markedly with other studies in humans, in which the recovery O₂ kinetics was also slowed at the higher work rates (4).

Physiological interpretation. For the on-transients, phase 1 is attributed largely to increased pulmonary blood flow, whereas phase 2, which elevates O₂ to the steady state (phase 3), is thought to reflect the arrival at the lung of venous blood, emanating from the exercising muscle (1, 18). The O₂ deficit is thought to comprise breakdown of CP, anaerobic glycolysis, and utilization of O₂ stores from muscle myoglobin and venous O₂ (see Refs. 8 and 18 for review). The off-transient has a similar phase 1, which is most likely associated with the rapid decrease in pulmonary blood flow and thus the delivery of deoxygenated venous blood to the lung (8). In support of this notion, pulmonary artery pressure drops dramatically within 10 s of decreasing treadmill speed, likely indicating a rapid decrease in cardiac output and thus pulmonary blood flow (Fig. 7).

View larger version (15K): [in this window] [in a new window]   Fig. 7.   Plots of pulmonary arterial pressure and O2 vs. time during recovery from heavy exercise. Solid line, pulmonary arterial pressure; dashed line, O2. Note close temporal correspondence between the 2 responses over initial 2-3 s postexercise (phase 1).

Symmetry of responses. At moderate exercise, there was symmetry between the on- and off-transients, consistent with traditional theories regarding O₂ deficit and EPOC, i.e., the postexercise O₂ matched quantitatively the O₂ deficit (Table 1).

O_{2 peak}. The kinetics following exercise in the severe domain (O_{2 peak}) were similar to those of heavy exercise, with the exception that the O₂ projected to a greater elevation from baseline (5.7 ± 1.1 vs. 3.0 ± 1.0 l/min; P < 0.05). The change in pulmonary arterial blood temperature for the two protocols was not different, suggesting again that the Q₁₀ effect was not responsible for the O₂ response, at least inasmuch as blood sampled at this remote site reflects the changes in muscle temperature. As mentioned earlier, the pulmonary artery plasma [lactate] was higher after severe than after heavy exercise, and the elevated baseline O₂ was significantly correlated with plasma [lactate] (Fig. 6).

O₂ deficit and EPOC. O₂ deficit and EPOC were not different for moderate exercise. This would be expected if the EPOC truly represented repayment of the deficit. In the heavy-exercise domain, the fast component of the O₂ deficit and EPOC were not different, as predicted (discussed above). However, the O₂ deficit calculated from EEO₂ values was substantially larger than the EPOC. Thus, calculation from EE values may over- estimate the O₂ deficit. With the development of a slow component, it is impossible to predict at any one time what the actual O₂ cost should be. It is likely that the O₂ cost is rising over time during the period of slow-component development and, therefore, the estimate of O₂ deficit by using EEO₂ may be inflated.

Relationship to previous studies. Rose et al. (16) found a large slow component in recovery from work at 120% O_{2 max}. There are four potential reasons for the difference between that study and the present study. 1) In the present study, the horses worked at or below O_{2 max}. In humans, there is a range >T_lac where the slow component is diminished or absent in recovery (13). 2) Training decreases or abolishes the recovery slow component (10). In many respects, horses respond in a fashion similar to that of very highly trained elite athletes. This may help explain the single-exponential response. 3) In the study by Rose et al., the horses were brought to a halt, whereas in the present study the horses were brought to a trot at 3 m/s. It is known that continued light exercise enhances the removal of lactate (and possibly other metabolites) from the plasma and muscle of the horse after exercise (12). This presumably occurs by utilization of lactate by the working muscles (2). It is possible that this lactate removal or some other temporally associated response affects the O₂ recovery kinetics. 3) Our study was not carried out for the same duration as that of Rose et al. Because there was a change in recovery baseline O₂ after heavy and severe exercise, it would be reasonable to expect that the O₂ would decrease over time, but so slowly as to not be significant or even detectable in our model.

Methodological considerations. To obtain signals suitable for modeling, we used a 10-breath moving average to smooth the raw data. Although this may slightly alter the  values (resulting in a modest slowing of the ), given the very fast kinetics in the horse, the effect should be minimal. This was confirmed by analyses of raw data; this yielded very similar values to those given in Table 2. We are confident that the single-exponential model was appropriate for <T_lac (moderate) runs, because the O₂ returned to preexercise levels before any slow component could be manifested. Thus, any contribution of this component must have been negligible. However, in the heavier workloads, O₂ did not return to baseline levels during the experiment; this suggests some type of slow component. Further analysis of the models revealed that the additional parameters were not significant (P > 0.05), because they had extremely large errors and dependencies >0.99 in all moderate and severe runs as well as in five of the six horses during the heavy run. Consequently, we feel justified in excluding the additional parameters and considering the extended response as an "elevated baseline," at least for the time period that was monitored.