INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

DURING HEAVY EXERTION, THE horse (3, 4, 29, 43) and humans (2, 10, 20, 21) develop exercise-induced arterial hypoxemia associated with increased alveolar (PA_O2)-arterial oxygen tension (Pa_O2) difference. Possible explanations include relative hypoventilation, diffusion limitation, and shunt and ventilation perfusion inequality (4, 35). Wagner et al. (43) found that alveolar-capillary diffusion limitation of oxygen was the major cause of hypoxemia at intense exercise in horses, similar to that described in humans. Erickson and colleagues (16) observed a reduction in PA_O2-Pa_O2 difference in exercising horses breathing air, in which helium was substituted for nitrogen, and speculated that improvement in either ventilation-perfusion relationships, gas-phase diffusion limitation, or alveolar-capillary diffusion equilibration contributed to the improved oxygen transport. Subsequently, this group attributed the improved PA_O2-Pa_O2 difference to an increase in PA_O2 that increased the driving pressure and permitted hemoglobin to operate on the steeper portion of the slope of its dissociation curve (17, 18).

Alteration of the diffusion barrier itself by the accumulation of interstitial water could contribute to both diffusion limitation and increased ventilation-perfusion mismatch. Under conditions of maximal exertion, vascular pressures in the pulmonary capillary bed of the horse increase greatly, thereby increasing the Starling forces that transfer water into the interstitium (6, 7, 33, 36). Abnormal ventilation-perfusion ratios after exercise in humans may accompany interstitial edema, thereby contributing to altered gas exchange (34). Pulmonary perivascular edema has been described in pigs after bouts of exercise (33). Exercise-induced pulmonary hemorrhage occurs in a large majority of horses (45), strongly suggesting that the alveolar-capillary interface is damaged during exertion in this species. Double-dilution indicator studies performed in humans suggest that the amount of extravascular lung water increased after exercise (26, 34). Because our initial observations on the horse appeared to confirm an increase in lung water (13, 46), we explored this possibility systematically during and after exertion. We postulated that the exercise-induced pulmonary vascular hypertension that is observed consistently in horses would cause pulmonary edema to an extent that could be detected by a dual-indicator dilution technique that can be applied during exercise. Specifically, we hypothesized that extravascular lung water would increase during exertion in the horse.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Horses. Seven standardbred horses, two mares and five geldings, ranging in age from 2 to 5 yr of age, 353-540 kg body wt, were studied. The horses, housed in stalls bedded with wood shavings, were fed twice daily on a diet of grain concentrate and grass hay, totaling ~2% of body weight. Water was freely available at all times. Horses were clinically normal at the time of study, and the Institutional Animal Care and Use Committee approved all interventions.

Instrumentation. For each experiment, catheter sites were aseptically prepared and infiltrated with mepivicaine hydrochloride. Two 8.5-Fr vascular catheter introducers with hemostasis valves (CR Bard, Billerica, MA) were placed in the left jugular vein. A 7-Fr vascular catheter introducer with a hemostasis valve (CR Bard) was placed in a portion of the right carotid artery that previously had been surgically relocated to a subcutaneous position. A cardiotachometer (Hippocard PEH 200, Kentucky Equine Research, Versailles, KY) was placed on the chest under a harness.

View larger version (8K): [in this window] [in a new window]   Fig. 1.   Schematic diagram of the tip of a pulmonary arterial catheter. Diagram is not to scale; diameter is ~2.5 mm.

Experimental protocol. After instrumentation, horses were exercised on the horizontal treadmill, and measurements were made during seven sequential exercise periods. Initial measurements were obtained with the horses standing quietly on the treadmill. Horses were then walked at 1.8 m/s. Treadmill speed was increased for 2 min to a speed previously calculated for each horse to reach 75% of HR_max. Treadmill speed was decreased to 1.8 m/s for a short (<5 min) walking period and then increased for ~3 min to a speed previously estimated for each horse to reach 100% HR_max. Treadmill speed was then decreased to 1.8 m/s, and measurements were obtained at three distinct walking recovery periods, 2, 10, and 20 min, after cessation of 100% HR_max.

Control experiment. Four of the horses were subjected to a similar exercise protocol without instrumentation or saline injections. Serum sodium concentration was measured in arterial blood that was collected at the end of each exercise level as in the protocol described above.

Data acquisition and analysis. Data were collected continuously at 64 Hz using an analog-to-digital board (DT2800, Data Translation Board, Marlboro, MA) and analyzed under the control of ASYST 4.01 (Keithley Instruments, Cleveland OH). The program permitted manual correction of baseline drift for any curve where necessary. Cardiac output (CO; l/min) was calculated by application of Stewart-Hamilton principles to the impedance dilution transient at the carotid artery using an algorithm based on observations of Trautman and Newbower (41, 42)
where Z_b () is the blood impedance, Z_o () is the impedance with the probe in 0.9% NaCl, S (/min) is the area of the distribution of transit times, V (liters) is the volume of NaCl injected, and c is concentration (%, wt/vol). The carotid artery impedance signal was used to calculate CO because its signal-to-noise ratio was marginally better than that at the pulmonary artery. CO was normalized to body weight. The extravascular thermal volume of the lung was determined using a dual-indicator technique, with electrical impedance as the nondiffusible intravascular indicator and heat as the diffusible total lung water indicator. Transit times for the impedance transients in the blood were timed from the centroid of the injection flow transient to the centroid of the impedance transient in the pulmonary artery (Z_tt pa) and the centroid of the impedance transient in the carotid artery (Z_tt ca) (Fig. 2). Active circulation time was measured using time between the peaks in the first and second impedance transients at the carotid artery. Temperature transients in the blood were timed from the centroid of the thermal input profile because this corrected for warming of the cold injectate in the syringe and catheter before it was injected (14). Transit times of the temperature transients at their centroids in the pulmonary (T_tt pa) and carotid arteries (T_tt ca) were corrected for response time delay of the respective thermistors.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Injection flow (A), electrical impedance (B and D), and temperature (C and E) in the pulmonary arteries (B and C) and carotid arteries (D and E) of a standing horse. The transients in impedance and temperature are associated with the cold hypertonic saline injections but are superimposed on slow oscillation of the baseline. Baseline was reset manually at X. Arrow indicates the onset of walking. Time is compressed ×64, and the signals are unfiltered.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

No significant changes in lung water were detected with exercise (Table 1). Because these techniques have not been applied before in the exercising horse, we must comment on the quality of the signals observed and state some of the compromises that were made. In the standing horse, the long transients (Fig. 2) generated in the electrical impedance make recirculation impossible to differentiate. Thus recirculation times and actively circulating blood volumes cannot be determined in the standing animal. The temperature signal (Fig. 2) from the standing horse is quite noisy because it hunts erratically by ~0.05-0.10°C at ~0.03 Hz. These values are not dissimilar to the 0.1°C fall in temperature with a transient of ~30-s duration produced by the 55-ml cold injection (Fig. 2). Generally, the temperature oscillations are more damped in the carotid artery than in the pulmonary artery. Once the animal is moving, both impedance and temperature signals become steadier (Fig. 3). Hunting of the temperature disappears shortly after walking begins. The second pass of the saline bolus produces a transient change in electrical impedance that can be clearly separated from that of the first pass, with a return to baseline between the two (Fig. 3). This suggests that recirculation can make only a negligible contribution to the first pass area. Sometimes at HR_max it is possible to discern a third pass of the bolus. Exertion is associated with steadily rising baseline in blood impedance and temperature that was compensated for during analysis (Fig. 3). In our opinion, the data become much more reliable once the horse first starts walking and remain so during and after heavy exercise.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Injection flow (A), electrical impedance (B and D), and temperature (C and E) in the pulmonary arteries (B and C) and carotid arteries (D and E) of a horse galloping at 100% maximal heart rate. Note the transients in impedance associated with 1st, 2nd, and 3rd pass of the indicator, the steady increase in baseline temperature, and the absence of slow oscillations. Time is compressed ×4, and the signals are unfiltered.

Within each horse, data derived from multiple injections during each exercise period were surprisingly consistent with one exception. In the standing horse, the thermal delay at the carotid artery increased consistently (P < 0.03) between the first and second injections. Because CO did not change under these circumstances, the lung water calculated from the second injection transient was consistently larger than that calculated from the first injection transient. An increase in lung water secondary to a hypertonic bolus passing through the lung has been reported with injection of radiopaque material (37, 38).

There were significant differences among horses for extravascular lung water, CO, stroke volume, total actively circulating blood volume, central blood volume, systemic actively circulating volume, pulmonary arterial pressure, Z_tt pa, Z_tt pa  Z_tt ca, Pa_O2, Pa_CO2, pH_a, hemoglobin, plasma protein, serum sodium, baseline impedance in the carotid artery, and temperature in the carotid artery.

Transit time from the jugular vein to the carotid artery decreased ~60% with the onset of walking and decreased a further ~50% with the increase in speed from walking to HR_max (Table 2). These values returned to preexertion walking values within 2 min of cessation of maximal exertion. Transit between the pulmonary and carotid arteries was ~60% of jugular-to-carotid transit time at all levels of exertion (Table 2). In the standing horse, the thermal transient in the carotid artery was delayed by ~2.4 s with respect to the impedance transient. This delay was more than halved by the onset of walking and at maximal exertion had been reduced to ~0.4 s. Walking after exertion was associated with a return of thermal delay to preexercise walking values (Table 2).

Heart rate, CO, and pulmonary arterial pressure increased with exertion and returned to preexercise walking levels while the horses were walking after maximal exertion (Table 1). Although none exhibited obvious epistaxis, all horses had fresh blood visible in the trachea at tracheoscopy. Treadmill speed at 75 and 100% HR_max was 6.6 ± 0.9 (SD) and 12.1 ± 0.6 m/s, respectively. At 100% HR_max, central blood volume increased by ~8 ml/kg, and systemic actively circulating volume increased by ~17 ml/kg with respect to the corresponding walking values before exertion (Table 1). The rise in stroke volume and total actively circulating blood volume did not reach significance (Table 1). With increasing CO, impedance transit time from the pulmonary artery to the carotid artery decreased first and then formed a plateau (Fig. 4).

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Impedance transit time (s) from the pulmonary artery to the carotid artery plotted against cardiac output (ml · kg1 · min1) in 7 horses. HRmax, maximal heart rate.

Blood temperature increased by ~3°C with exercise and remained increased throughout the walking recovery period, even after 20 min (Table 3, Fig. 3). Pa_O2 decreased significantly with maximal exercise but was greater than preexercise standing values at 2 and 10 min of walking recovery. Pa_CO2 was greater at maximal exercise than at 75% maximal exertion and was less than standing values at all periods of walking recovery. pH_a decreased with exertion but had returned to preexercise values after 10 min of walking. Serum sodium increased with exertion and remained increased during all walking recovery periods. At no time was serum sodium concentration different between exertion with saline injections and exertion without saline injections. Baseline blood impedance and hemoglobin concentration were increased over standing values at all exercise and recovery levels. The changes in baseline impedance were significantly related to the changes in Hb and the plasma protein concentration (P). After correcting for the temperature and expressing the impedance changes as a percentage of the walking levels (Z%), regression analysis gave the relation Z% = 53.5 + 7.7 × Hb  7.3 × P; r² = 0.83. 

One horse (horse 7) exhibited inspiratory stridor on exertion and, on subsequent examination, was found to have partial left laryngeal paresis. This horse had the lowest Pa_O2 during maximal exertion (37 Torr) but otherwise did not stand out from the other data.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

The most important result is the absence of significant change in the lung water with exercise. Based on the least significant difference on ANOVA, the technique we used should have detected an increase of ~2 ml/kg (40). A change of this magnitude was quite absent during exercise and was approached only between 2 and 20 min of walking recovery from exercise. Although we cannot rule out a smaller increase in lung water as an impediment to oxygen diffusion from the alveoli to the blood during exertion, these data suggest that substantial pulmonary edema is an unlikely factor in the hypoxemia seen in exercising horses. Nevertheless, an increase in lung water below that which we could detect could impair diffusion, especially were it to be preferentially distributed to the pericapillary interstitium rather than perivascular spaces. The increase in water flow into the interstitium adduced from the increase in pulmonary vascular pressures is likely to be offset by a corresponding increase in lymphatic drainage (28).

Lung water measured in these experiments was 2.6-4.3 ml/kg body wt. Using the tritiated water and ¹³¹I-labeled albumin dual-indicator technique, lung water in humans has been reported to increase from 126 ml at rest to 229 ml at exertion (26); assuming a body weight of 60 kg, this gives an increase from 2.1 mg/kg at rest to 3.8 ml/kg with exercise, values that are quite similar to those reported here. Using dual impedance/thermal methods in normal anesthetized dogs, lung water has been reported to range from 6.0 to 8.3 ml/kg body wt (25, 30) and in anesthetized horses to be ~7.8 ml/kg body wt (24). Lung water values obtained in our experiments were substantially less than those cited above. Direct comparison with these data from horses is difficult because CO and transit times were not reported for the dual-indicator dilution technique (24). Theoretically, such a discrepancy in the values for lung water could be associated with large volumes of lung that are unperfused and, therefore, unavailable for heat diffusion in our horses (15). Studies with microspheres have failed to detect large, unperfused regions of lung, either in healthy, standing horses (12, 23) or in exercising horses (5). However, capillary recruitment beyond the site of microsphere wedging may not be complete, particularly in the standing animal, preventing microsphere-based methods from detecting substantial unperfused tissue (22).

Studies in dogs (25) using dual indicators to measure lung water gave no dependence on CO for these techniques, suggesting that, even at the increased CO values in these horses, the technique could give useful data. Although exercise-associated accumulation of lung water in humans has been suggested by qualitative imaging studies, with the use of radiography and computerized tomography (1, 8), quantitative studies measuring lung water accumulation in large numbers of exercising people have yet to be reported (11).

Exercise produces modest increases in pulmonary arterial pressure in most exercising mammals, and exercise-induced hypoxemia only occurs in ~50% of very fit male human athletes during exertion (31, 44). The much greater increase in pulmonary arterial pressure and consistent exercise-related hypoxemia led us to expect any increase in lung water to be more easily detected in exercising horses than in other species, such as humans. Our early experiments suggested that extravascular lung water rose with increasing workload in horses and concurred with electron microscopic findings in exercising horses and light microscopic findings in exercising pigs (13, 33, 45, 46). Once we recognized two systematic errors, both tending to delay the transients in temperature but not in impedance, it became apparent that lung water had been overestimated in our early experiments. The first source of error was the lag of the thermal input transient behind the injection flow transient caused by unavoidable warming of the injectate in the dead space of the catheter and syringe. If uncorrected, this warming caused the heat removed by the injection to be overestimated and the centroid of the heat transfer to lag behind the centroid of the injection flow profile by 0.10 ± 0.03 s. The error in lung water that this caused was exaggerated during exertion, because the timing error is multiplied by the CO.

The principle behind the temperature measurement of the injectate and its use to correct the systematic error described above deserve a brief explanation. The transit time flowmeter attached to the injection flow probe uses the timing difference between upstream and downstream passage of ultrasound signals to measure flow. The sum of these two signals measures the time, independent of the flow, for the sound to pass between the two crystal sensors and hence gives a sensitive measure of the sound velocity between the crystals. Because this velocity is determined solely by the temperature of an injectate of constant composition, the output voltage can be calibrated to yield a dynamic temperature with negligible signal delay. The instantaneous product of the injection flow and the difference of injectate temperature from body temperature give the instantaneous heat load administered to the horse and, hence, the actual thermal input transient. By using the centroid of this transient, any warming of injectate in the dead space before injection was automatically compensated for in the temperature transit times.

The second source of systematic error in our laboratory's earlier experiments (13, 46) arose from the rate of equilibration of the thermistor. To make catheters that were sufficiently robust for insertion, the thermistors (Fig. 1) were partially imbedded in plastic, which, because of its specific heat, increased the half-equilibration time of temperature from a nominal ~0.05 s to 0.08 ± 0.02 s. The transit times of the temperature transients reported here were corrected for the actual response time of each thermistor measured after each experiment.

The injections of hypertonic saline constituted a sodium load but did not increase the serum sodium concentration compared with a control. The increase in sodium concentration with exertion that we observed was similar to that in the horses that received no injections and was also similar to that described by others in normal exercising horses (9). It is possible that the injections contributed to the changes in blood volume or CO, but we cannot control for this.

Actively circulating blood volume has been measured from the exponential decay of radioactivity observed immediately after bolus injections of isotope-labeled albumin and erythrocytes in anesthetized dogs (32). Actively circulating blood volume measured directly from the recirculation of a bolus, as reported here, can be shown to be the same volume (V. V. Kislukhin, personal communication). The situations differ in the degree to which the signal from the bolus is damped on passing through the vascular system; damping is much reduced in the exercising horse. Total blood volume, measured after allowing time for thorough mixing of a label throughout the whole blood volume (9 , 27), would be much larger than actively circulating volume.

The effect of exercise on actively circulating blood volume has not been described previously. In exertion, the increase in total actively circulating blood volume proved too variable to reach significance (Table 2), but the central blood volume, the volume of blood between the jugular vein and carotid artery, increased by ~8 ml/kg (47%) during the transition from walking to maximal exertion. This accompanied the rise in pulmonary arterial pressure; hence the lung vasculature increased its capacity with a compliance of ~0.2 ml · mmHg¹ · kg¹. The systemic part of actively circulating blood volume increased by ~17 ml/kg (66%) during the transition from walking to maximal exertion. This increase at least partially accommodates the increased red cell volume associated with splenic contraction in the horse and may also represent a shift from more slowly perfusing regions, such as the splanchnic bed, to regions with vasodilatation; presumably it is largely accommodated in muscle. The magnitude of the increase in these volumes is reduced by exercise-induced reduction in plasma volume (27). In both pulmonary and systemic vascular beds, part of the increase in blood volume depends on passive vessel dilatation and part on capillary recruitment. Both mechanisms increase capillary capacitance and capillary diffusive surface area and, in turn, offset the reduction in transit times that the eightfold increase in CO would otherwise impose on a rigid system. This is illustrated in Fig. 4, where the majority of the reduction in transit time across the pulmonary bed and left heart occurs during the transition from standing to walking; further increases in CO produce only minor reductions in transit time. Nevertheless, at maximal exertion, the transit time from the pulmonary artery to the carotid artery was reduced to ~2.7 s, and recirculation time was reduced to ~7.1 s, suggesting that capillary transit time and the time for exchange of oxygen in both pulmonary and systemic vascular beds are substantially reduced. The plateau in transit time (Fig. 4) at high CO values suggests that the decrement in oxygen tension between 75% and 100% HR_max is unlikely to be due to a further decrease in the time for equilibration of capillary blood with alveolar gas. The 17% increase in central blood volume (Table 1) that occurs between 75 and 100% HR_max suggests that the corresponding decrement in Pa_O2 may be due to increased distance for oxygen diffusion within the capillary.

Most hemodynamic variables recovered quickly from exertion. After a 2-min walking period, blood transit times, CO, central blood volume, and systemic circulating blood volume were indistinguishable from preexertion walking values, but hemoglobin concentration decreased more slowly with a half-time of ~10 min. Pa_O2 increased and Pa_CO2 decreased after maximal exertion. This hyperventilation was possibly driven by the accumulated heat load and oxygen debt resulting from exertion.

It is likely that lung water must increase ~30% before it can be reliably detected by methods such dual-indicator dilution (39). Despite identification of and correction of important sources of error in the thermal-impedance, dual-indicator technique, an increase in lung water with exertion was not detected in these horses. Limitations in the ability of the technique to detect measurable increases in lung water with exercise in horses does not preclude a smaller accumulation of lung water under these conditions. The technique did allow us to measure substantial decreases in transit time for blood and large increases in the actively circulating components of the blood volume.