ABSTRACT
A new design of flowmeter is described and used in a comprehensive study of the respiratory and cardiovascular adjustments that occur during a standardised exercise test in Thoroughbred horses. The flowmeter system and associated lightweight, fibreglass mask (total mass, 0.7kg) have a maximum dead space of 500ml and negligible resistance to airflow. They have no systematic effect on blood gases and, together with a rapidly responding mass spectrometer, enable an accurate computation of gas exchange to be performed together with breath-by-breath determination of other respiratory variables.
At the highest level of exercise (12 ms−1 on a 3° incline), the rate of oxygen uptake and carbon dioxide production increased to 29.4 times and 36.8 times their resting values, respectively. Respiratory minute volume increased to 27.0 times its resting value, with respiratory frequency (fR) making the major contribution at the walk and trot. However, with increasing cantering speeds, fR changed little as it was locked in a 1:1 fashion to stride frequency, and tidal volume (VT) then made the major contribution to the increase in . The ratio of ventilatory dead space (VD) to VT in resting horses was lower than that previously reported in the literature and this could be the result of the different respiratory recording systems that were used. There was a close relationship between VT and stride length at increasing cantering speeds.
Despite the fact that alveolar ventilation was well matched to , there was a significant reduction in arterial when the horses cantered at 8 ms −1 and this eventually fell to 34% below the resting value. The present data tend to support the idea that (where is cardiac output) inequalities are important in causing this hypoxaemia. However, the reduction in was more than compensated for by an increase in haemoglobin concentration, [Hb], so the concentration of oxygen in the arterial blood was significantly above the resting value at all levels of exercise. Both lactate concentration and increased during exercise, causing substantial reductions in pH of both arterial and mixed venous blood. This would have inevitably shifted the oxygen equilibrium curve of the Hb to the right, desaturating the arterial blood and thus exacerbating the effect of the hypoxaemia, as would the almost 4°C rise in blood temperature. The tight respiratory/locomotor linkage might prevent the acidosis and hyperthermia having the stimulatory effects on that they have in humans at high work loads.
Cardiac output and heart rate were 5 and 6 times their resting values, respectively, at the highest level of exercise, while cardiac stroke volume did not change significantly. There was both systemic and pulmonary hypertension, with mean pulmonary blood pressure reaching 2.6 times its resting value at the highest level of exercise. It is concluded that the pulmonary hypertension does not result from the hypoxaemia and that it could cause pulmonary oedema, which could contribute to the postulated diffusion limitation of the lungs during exercise.
Although this report confirms some results obtained by other workers, some discrepancies have been found. It is probable that the flowmeter and mask used here have a significantly reduced effect on the measured variables compared with methods used in many of the previous studies.
INTRODUCTION
Thoroughbred horses, together with dogs, foxes and pronghorned antelope, are considered to be elite athletes amongst terrestrial mammals (Taylor et al. 1981, 1987; Longworth et al. 1989; Lindstedt et al. 1991) in as much as their maximum oxygen uptake is several times greater than that predicted for similar-sized terrestrial mammals. There is, therefore, much interest in the physiological mechanisms involved in the demand for and delivery of such high rates of oxygen. Clearly, the important variables in the process of oxygen delivery (and CO2 removal) are those associated with ventilation and gas exchange itself. It is not easy to determine these variables in any animal during exercise, let alone in large animals such as horses. Masks incorporating one-way valves have been used and these enable expired gas to be collected, thus allowing the determination of tidal volume and respiratory minute volume, as well as gas exchange (Bayly et al. 1987b; Evans and Rose, 1987; 1988a). However, such a system has long lengths (2–3m) of wide-diameter (10cm) tubing attached to it and a relatively large (1.5l) dead space. It would be surprising if such a system did not affect the ventilatory characteristics of the horse in some way; it certainly causes a significant reduction in and a significant increase in at all levels of exercise (Bayly et al. 1987c). The magnitude of these changes can be up to approximately 3kPa. A flow-through system using flow rates in excess of 6000 l min−1, as used by Jones et al. (1989a), has no effect on blood gases (Bayly et al. 1987c), but does not allow the determination of tidal volume or respiratory minute volume. Neither of these systems allows the determination of respiratory flow rates or breath-by-breath analyses of respiratory variables.
The traditional method for obtaining such data from air-breathing vertebrates is to use a Fleisch pneumotachograph, which has to be connected to a differential pressure transducer. Such a system has been used with exercising ponies (Art and Lekeux, 1988), but two different pneumotachographs had to be used, one for the animals at rest and the other when they were exercising. With the latter, the total dead space was 700ml and the mask, plus the resistance to flow of the transducers, had a significant depressive effect on respiratory frequency during exercise. It is likely, therefore, that tidal volume was elevated to compensate for this or that respiratory minute volume was suppressed. In their studies, Art and Lekeux (1988) did not exercise the ponies above a fast trot (3–5 ms−1). Presumably, when cantering and galloping, the adverse effects of the mask on ventilation would have been greater as respiratory frequency and possibly tidal volume are linked to stride frequency and stride length, respectively (Bramble and Carrier, 1983). A flow-through system incorporating a Fleisch pneumotachograph has been used by Landgren et al. (1991). Flows of up to 10000 l min−1 had to be generated and, again, long lengths of tubing were used. Although there appeared, from the data presented, to be little influence of this system on blood CO2 levels, the authors do concede that the substantial loads imposed by the system ‘probably influence breathing’. It must also be remembered that any deposition on the resistive element of the pneumotachograph will affect its performance (Art and Lekeux, 1988).
A different system has been developed in which airflow is determined by the force it exerts on a sensing flag that projects into the flow (Hörnicke et al. 1974; Kimmich and Spaan, 1980). The system is of complex mechanical design, requiring secondary sensors to compensate for the effects of gravity and acceleration and it will not cover the whole range of flows from rest to heavy exercise without a change in the internal diameter of the respiratory tube.
The problems and limitations associated with these various methods have been discussed by Rose and Evans (1987) who wrote: ‘It is apparent that some effort needs to be put into the design and manufacture of a mask that is capable of handling the high respiratory flows that are generated at maximal exercise, without disturbing the respiratory pattern by increasing flow resistance. It is also important to minimize extra respiratory dead space and wasted ventilation due to the mask. This problem is particularly relevant in the horse during maximal exercise at the canter and gallop, when respiratory frequency is relatively high and tidal volume is relatively low’.
A new design of flowmeter has been developed which overcomes many of the problems mentioned above. This report details its use in a comprehensive study of the respiratory and cardiovascular adjustments that occur during a standardised incremental exercise test in Thoroughbred horses. This is the first time that such a range of variables has been measured directly in horses exercising at close to their with a system that has been demonstrated to have a minimal effect on levels of blood gases and no obvious effect on ventilation. The data obtained will be compared with those collected previously by the various methods described above. A brief report of a preliminary study has already been published elsewhere (Woakes et al. 1987), as have some of the present data (Anderson et al. 1989).
MATERIALS AND METHODS
Seven Thoroughbred horses (5 geldings, 2 fillies) were used in the present study; they had an average mass of 440±12kg and mean age of 5.3±1.3 years. They were all stabled under the same conditions and fed hunter cubes (Dodson and Horrell Ltd) and hay, with water provided ad libitum. A carotid artery was raised to a subcutaneous position in each horse in order to aid its subsequent catheterization. The operation was performed under general anaesthesia in the Clinical Unit of the Animal Health Trust. Following this procedure, the animals were allowed a recovery period of at least 2 months before being exercised. They were then trained for 4–6 months to bring them to a level of racing fitness prior to the studies.
The studies were performed on a treadmill (Sato, Sweden) housed in an air-conditioned building maintained at a temperature of 20°C and a humidity of 60%. While on the treadmill, the horse wore a safety harness clipped onto a strap attached to a support frame above the treadmill. One air fan, placed in front of the horse, was used during the exercise tests to assist heat loss.
An incremental exercise test was used. This was preceded by a warm-up period consisting of a 6min walk at 1.6 ms−1 followed by a 4min trot at 4 ms−1 and a further 5 min walk at 1.6 ms−1, all at 0° incline. The horses were then exercised for 2min at speeds of 6, 8, 10 and 12 ms−1 (0° incline) and 12 ms−1 (3° incline) with each speed separated by a 5min walk (1.6 ms−1). This was followed by a recovery period consisting of 15min of walking (1.6 ms−1, 0° incline) and 15min of standing. Exercising the horses for 2min enabled the present data to be compared with those obtained from previous studies (Bayly et al. 1987c; Evans and Rose, 1988b; Jones et al. 1989a;,Wagner et al. 1989). Although there was a 5min walk between each exercise level, the data are presented graphically as exercise intensity increasing continuously.
The horses were fitted with a lightweight, fibreglass mask which housed two flow tubes (total mass 0.7kg), one for each nostril, leaving the mouth free (Woakes et al. 1987). The velocity of the airflow in the tubes was detected by phase shifts in beams of ultrasound transmitted alternately in one and then the other direction, passing diagonally across the airflow (Woakes and Butler, 1980). The dead space of the mask and flow tubes was, at most, 500ml. The resistance to airflow of the mask and tubes was determined at flow rates up to 20–30 ls−1 and was found to be negligible. This is also the case for flow rates up to 60 ls−1 (D. J. Marlin, in preparation). Instantaneous respiratory gas concentrations were monitored by a mass spectrometer (Airspec Ltd, MGA 2000) via a flexible capillary positioned in one of the flow tubes.
The efficacy of the face mask in monitoring all of the ventilatory gas flow and its effect on blood gases were determined. The mask was fitted to a horse, as usual, and an additional undertray, of the same material, was added to enclose the mouth and jaw. These two units were then completely sealed to the head by a flexible plastic film. A third flow tube was connected to the undertray so that any gas entering or leaving the horse’s respiratory system, but bypassing the two tubes attached to the mask, could be monitored. This additional system was fitted to three horses, and flows were measured from rest to canter at 8 ms−1. Peak airflows measured from the mouth chamber represented approximately 3.6% of the peak flows measured by the respiratory flowmeters. The flow profiles were dissimilar, and appeared to be related to the movement of the head within the additional mask structure. Leakage of respiratory airflow from around the mask seal was, therefore, considered not to represent an important source of error. The variation of flowmeter accuracy with gas concentration was determined through the use of a motor-driven piston pump (Boutellier et al. 1981) pushing a constant volume of gas of variable composition through the flowmeter into a distendable but non-elastic bag. The change in gain of the flowmeters was as predicted by the theory of operation of such ultrasonic devices (inversely varying with the square of the velocity of sound), and appropriate correction factors were used to adjust the flow signal when determining exhalation variables. These corrections were less than 3% of the measured flow.
Two different assessments of the effect of the mask on levels of blood gases were made. In the first assessment, five horses completed the incremental exercise test described earlier both with and, at a later date, without the mask. In the second assessment, three horses performed another exercise test, in which the speed was raised in steps of 1ms−1 after 1min at each speed up to 11ms−1 (Butler et al. 1991), in the morning and afternoon of the same day, one with and the other without the face mask. Two-way analysis of variance on both sets of data indicated that the mask had a significant effect on partial pressure of CO2 in arterial blood during the first assessment, but not during the second. Subsequent statistical analysis (Tukey multi-comparison test) of individual pairs of data points from the first assessments (Fig. 1) indicates that there was no systematic effect of the mask on during this test (i.e. there were significant effects of the mask at speeds of 6, 8 and 12ms−1, but not at 10ms−1).
Instantaneous respiratory airflow through both tubes and respiratory O2 and CO2 concentrations (Fig. 2) were recorded on a six-channel recorder with rectilinear coordinates (Gould, USA). The following variables were determined as described by Woakes et al. (1987) and Butler et al. (1991): peak inspiratory and expiratory flows, end tidal (ET) O2 and CO2 partial pressures, respiratory frequency (fR), tidal volume (VT), respiratory minute volume , rate of oxygen consumption , rate of CO2 production . Values for VT and V were corrected to BTPS and those for and are at STPD. Alveolar ventilation was calculated from the formula:
The accuracy of the basic measurement system and associated computer analyses for determining was tested in humans (N=6) against an open-flow system in which air was drawn through a T-piece, mounted on the end of an ultrasonic flowmeter held in the mouth, at a known rate which always exceeded the peak inspired and expired flow rates. This flowmeter was a more sensitive version of the system used on horses. The oxygen contents of air entering and leaving the T-piece were determined by a paramagnetic oxygen analyser (Servomex Ltd). The subjects exercised on a bicycle ergometer, and measurements were made over a range of exercise levels, from rest to the maximum of which each subject was capable. When the values of obtained simultaneously by the two methods were plotted against one another, the intercept (-0.046±0.078) was not significantly different from zero and the slope (0.954±0.047, N=29) was not significantly different from unity.
Heart rate (fH) in exercising horses was recorded telemetrically (PEH 100 Hippocard, Switzerland) from electrodes on the mid and lateral thorax and synchronised with the other recording systems. The accuracy of this system has already been tested (Physick-Sheard et al. 1987). A veterinary surgeon monitored the performance of the horse and, by pressing a button on a recording watch, marked the recording so that specific times during the exercise could be recognised.
Four catheters were inserted into the left jugular vein and the raised carotid artery while the horse was standing in stocks. The areas of insertion were clipped and scrubbed with povidone–iodine (Pevidine, BK Veterinary Products Ltd, UK) and surgical spirit. After the area had been shaved, local anaesthetic (lignocaine hydrochloride, 2% aqueous solution without adrenaline; Astra Pharmaceuticals, UK) was administered before incisions were made in the skin. A polyethylene catheter (i.d. 1.67mm, o.d. 2.42 mm, Portex, UK) for sampling mixed venous blood and a transducer-tipped catheter (6F, Millar Mikro-tip, Millar Instruments, USA) were inserted into the left jugular vein. The former was inserted until its tip lay in the right atrium and the latter was advanced until it recorded pulmonary artery pressure. A catheter for temperature measurement and blood sampling (7F, Criticath, Spectramed Inc., USA) and a transducer-tipped catheter (6F Millar Mikro-tip) were inserted into the raised carotid artery. The catheters were flushed regularly with heparinized saline (10–20 i.u.ml−1) to keep them patent. Preliminary tests on two horses at different levels of exercise indicated that blood taken from the right atrium was representative of mixed venous blood as there was no significant difference in the measured at this site and in samples taken simultaneously from the right pulmonary artery. Pulsatile and mean blood pressures from the pulmonary and carotid arteries were displayed on a six-channel recorder (Gould, USA).
Arterial (a) and mixed venous blood samples were taken, in plastic syringes, during the last 10s of each work level for the analysis of pH, partial pressures and contents of oxygen and carbon dioxide , and lactate and haemoglobin, [Hb], concentrations. After withdrawal of the blood sample, the syringe was placed on crushed ice and water for storage (<1h) before analysis. Before a sample was analysed for blood gases and pH, it was mixed thoroughly and the blood in the tip of the syringe discarded. Storage on ice had no significant effect on pH (+0.003±0.004) or (+0.04±0.02kPa) for periods up to 2.5h. There was, however, a significant increase in (+0.46±0.11kPa) (see Mahoney et al. 1991).
Blood gas measurements and pH were performed on a blood gas analyser (ABL 330, Radiometer, Denmark) with an automatic temperature correction facility based on the characteristics of human blood. These temperature corrections are also accurate for horse blood (Butler et al. 1991). Since the blood gas and pH measurements were corrected to the temperature recorded in the arterial blood of the animal at the time they were taken, they are correct (see Jones et al. 1989b). Oxygen content was measured by the method of Tucker (1967). Cardiac output was calculated from and using the Fick equation. Packed cell volume (PCV) was measured by a microhaematocrit (Hawksley, UK), and [Hb] was determined by a haemoglobinometer (Coulter Electronics Ltd, UK). Blood lactate concentration was determined from venous blood by an enzymatic assay (Hohorst, 1963).
All variables are calculated as means ± S.E.M. A repeated-measures analysis of variance (ANOVA) test was used to determine the effect of exercise on each variable. If significant (P<0.05), pairwise comparisons were made between the resting values and those at different levels of exercise using Fisher’s PLSD test.
RESULTS
Exercise
The mean resting values for the measured variables are given in Table 1. Both and increased significantly above the resting value during the walk. There was a linear relationship between the different levels of exercise and exchange of the two respiratory gases (Fig. 3) and, although the horses were exercising strenuously at 12 ms−1+3°, they did not appear to reach their maximum rate of oxygen uptake , despite attaining 29.4 times the resting value at 60.6±3.6 l min−1 (137.1±10.3mlmin−1 kg−1), with at 36.8 times the resting value (75.1±4.3 l min−1). There was a significant increase in the respiratory exchange ratio (RE) at the walk; it then decreased to levels similar to that at rest until the horse was galloping at 12 m s−1, when it was again significantly above the resting value. also increased significantly at the walk and at the highest levels of gas exchange it was 27.0 times the resting level, although it can be seen from Fig. 4 that this variable is essentially the same at the two highest levels of . When the horses were running at 12 ms−1 on the flat, was 25.7 times the resting value, was 28.9 times the resting value and was 26.6 times the resting value.
The relative contributions of VT and fR to the changes in are related to the gait of the horse. fR increased significantly at the walk, whereas there was no significant increase in VT above the resting value until the trot (Fig. 4). From rest to canter at 6 ms−1, increased to 13.3 times the resting value, with fR increasing to 8.2 times the resting value and VT to only 1.6 times the resting value. Between walking and cantering at 6m s−1, there was no significant change in VT. As exercise intensity increased, there was only a slight (8%) further change in fR, as it was then locked in a 1:1 fashion with stride frequency (fS), which also changed little. The further (twofold) increase in up to 12 ms−1 was largely the result of a similar (1.8-fold) increase in VT. Thus, the relationship between VT and was biphasic, with an abrupt change in slope once fR stopped increasing, i.e. from 6 ms−1 to 8 ms−1 (Fig. 4). increased significantly at the trot and then rose linearly with to 1121±91 l min−1 at 12 ms−1+3°, an increase of almost 30-fold. VD/VT increased significantly at the walk and then decreased to values similar to that at rest.
Another interesting feature shown in Fig. 4 is the close relationship during cantering between VT and stride length (SL). Both variables increased by a similar proportion (approximately 2 times) between 6 and 12ms−1, but neither changed at 12ms−1+3 °. The constant VT at 12ms−1 and 12ms−1+3 ° is compensated by a decrease in the partial pressure of end-tidal oxygen . Although there is no significant difference between the values of at 10ms−1 and 12ms−1 (14.1±0.2 kPa and 13.9±0.2kPa), the value at 12 ms−1+3 ° (13.5±0.2kPa) is significantly lower than that at 10ms−1 (see Fig. 6). Peak airflow was significantly greater during expiration than during inspiration at the trot and the difference between them became progressively greater at higher levels of exercise (Fig. 5).
Despite the fact that was well matched to , there was a progressive reduction in below the resting value as soon as the horses began to canter (Fig. 6). was significantly below the resting value at 8 ms−1 and eventually reached a value some 34% below the resting value. , after an initial significant increase, was not significantly below the resting value even at the highest level of exercise, so the difference between and became progressively greater (Fig. 6). and followed each other closely and it was not until the highest levels of exercise (12 ms−1 and 12 ms−1+3°) that these two variables were significantly above the resting levels (at 12 ms−1+3°, they were both approximately 26% above the resting value). At the lower levels of exercise, mirrored in that it was significantly below the resting value at walk, trot and 6 ms−1. From walking onwards, showed an exponential increase. It was significantly above the resting value at 8 ms−1 and at the highest level of exercise was 2.4 times the resting value.
Lactate concentration in the venous blood increased exponentially and was significantly above the resting value at 10 ms−1. At 12 ms−1+3° it was 28 times the resting value (Fig. 7). Mixed venous blood became significantly acidotic at 8 ms−1, whereas arterial blood became acidotic at 10 ms−1. The pH of both then declined at approximately the same rate, being, on average, 0.38 and 0.30 of a pH unit, respectively, below their resting values at 12 ms−1+3°. Fig. 7 also illustrates the large (almost 4°C) increase in arterial temperature that occurred during the exercise test. It had increased significantly above the resting value at the trot.
As soon as the horses began to exercise, i.e. to walk, there were significant increases in PCV and [Hb] (Fig. 8). Approximately half of the eventual increases in these two variables occurred while the horses were trotting. Thus, although there was no significant decrease in until the horses were galloping at 8 ms−1 (Fig. 6), there was a significant increase in while they were walking (Fig. 8). peaked at 8–10 ms−1 and then began to decline. At 12 ms−1+3° it was significantly below the value at 8 ms−1. However, because the rise in [Hb] more than compensated for the fall in was 24% above the resting value at 12 ms−1+3°. The decline in was matched by an equally dramatic fall in , eventually to 13% of its resting value was significantly below the resting value at 6 ms−1. The difference between and increased from 4.3±0.27mldl−1 at rest to 23.7±0.6mldl−1 at 12 ms−1+3°, the latter value being 14% greater than in the horses at rest.
Heart rate and cardiac output had both increased significantly at the walk and at 12 ms−1+3° were 6 and 5 times their resting values, respectively (Fig. 9), while cardiac stroke volume did not change significantly (1.1l at rest, 1.0l at 12 ms−1+3°). increased to 4.33±0.14 at 12 ms−1+3°. Associated with the increase in were increases in mean pressures in both carotid and pulmonary arteries (Fig. 9). Although both peaked at 10–12 ms−1, with the former being 1.6 times the resting value and the latter 2.6 times the resting value, pulmonary artery pressure was significantly above its resting value at the trot whereas carotid artery pressure was not significantly elevated above the resting value until 8 ms−1. Resistances to flow in the two major circuits were 0.32±0.05kPa min l−1 in the systemic and 0.10±0.02kPaminl−1 in the pulmonary at rest. They fell to 0.10±0.01kPamin l−1 and 0.05±0.01kPamin l−1, respectively, at 12 ms−1.
Recovery
Even after 30min of recovery (15min of walking and 15min of standing), was still more than twice the pre-exercise resting value (Fig. 3). However, and fR were, at the same time, approximately 5.5 times their resting values, while VT had returned to its resting value (Fig. 4). VD/VT was significantly above the resting value for the whole of the recovery period. and were, respectively, above and below their resting values 5 min after the end of exercise (Fig. 6). The former had returned to its resting value but the latter was still significantly below the resting value 30min after the cessation of running. Lactate concentration and arterial temperature remained significantly above their resting values, and arterial pH significantly below, even after 30min of recovery (Fig. 7).
Mean systemic arterial blood pressure had returned to its resting value by 15min of recovery, although mean pulmonary arterial pressure was still significantly above the resting value. After 30min of recovery, the latter had returned to its resting value (Fig. 9). fH and were approximately twice their resting values 30min after cessation of running. However, after the 15min of recovery walking, both of these variables were at the same levels as those recorded during the initial period of walking before the exercise test began.
DISCUSSION
The mask and airflow system used in the current experiments copes with the full range of respiratory airflows and gas exchange of Thoroughbred horses from rest to high levels of exercise and it has no systematic effect on the partial pressures of blood gases. The airflow profiles were indicative of those from healthy horses without chronic obstructive pulmonary diseases (Pollman and Hörnicke, 1987). It is in this context that the present data must be compared with those obtained using different mask and airflow systems.
Although it appears that was not reached in the present study, the highest mean value of is very similar to some reported values of for Thoroughbreds (Rose et al. 1988; Evans and Rose, 1988c), but somewhat less than those reported for fully fit Thoroughbred horses (Butler et al. 1991; Knight et al. 1991). As previously described, respiratory minute ventilation increased linearly with the increase in , although it was higher in the present study than in that of Evans and Rose (1987) at similar levels of (1807 versus 1250 l min−1 at a of 140mlkg−1 min−1). It is, perhaps, strange that minute ventilation does not show a disproportionate increase at high work loads, when metabolic acidosis and hyperthermia are apparent, as is the case in man (Wasserman et al. 1967). This may be because the mechanical coupling that exists between limb movements and ventilation in cantering and galloping horses (Bramble and Carrier, 1983; Young et al. 1992) actually restrains lung ventilation and prevents increases in both frequency and volume beyond those related to the mechanical factors. At lower speeds (up to and including 6 ms−1) the large increase in fR is the major contributor to the increase in pulmonary ventilation, whereas above that speed fR changes very little and VT increases substantially. As predicted by Bramble (1989), tidal volume is proportional to stride length, at least at speeds at and above 6ms−1, even at 12 ms−1 and 12 ms−1+3°, despite the increased work load associated with the latter compared with the former. This relationship is not apparent from the data of Bayly et al. (1987b), who found that VT was constant between speeds of 5 and 10 ms−1 (on a 10° incline). Since increased significantly (see Fig. 3 of Bayly et al. 1987b), there must have been an equally significant increase in fR, which is strange in view of the fact that, as already mentioned, fR normally changes little above a speed of 6 ms−1. Maximum peak inspiratory and expiratory airflows from the present study were similar to those reported by Hörnicke et al. (1987).
Few determinations of VD/VT have been made in horses. Three studies report a value at rest of around 0.6 (Rose and Evans, 1987; Bayly et al. 1987b; Pelletier et al. 1987). The value from the present study is lower than this, but similar to that measured in man (Jones et al. 1966). The latter authors also show that this ratio declines to around 0.15 at high work loads. Bayly et al. (1987b) report a drop to a similar level in Thoroughbred horses running at 10 ms−1 on a 10° incline, whereas in the present study there was no such decline below the resting value. The reasons for these differences in the values of VD/VT and in the response of this ratio to exercise in the Thoroughbred horse may be related to the different respiratory equipment used. Bayly et al. (1987b) and Pelletier et al. (1987) obtained their values of VT, and VD/VT using a valved mask system. This has a relatively large dead space (1.2–1.5l) and is known to cause large increases in , particularly at high levels of exercise (Bayly et al. 1987c), which would influence the calculation of VD/VT (see Materials and methods). If these authors did not take into account the (large) dead space of the mask system they used when calculating VD/VT, and there is no indication that they did, this would explain why they obtained such high resting values. Also, the value of minute ventilation that Bayly et al. (1987b) obtained at an oxygen uptake of approximately 140mlkg−1 min−1 was, at 1200 l min−1 BTPS, substantially lower than the 1800 l min−1 obtained in the present study and in that of Hörnicke et al. (1983: approximately 16001 min−1 STPD) at a similar oxygen uptake.
Thus, unlike the situation in man (Jones et al. 1966; Wasserman et al. 1967), there may not be a reduction in VD/VT below the resting value in exercising Thoroughbred horses and this may, at least partly, explain why these animals become hypercapnic and hypoxaemic at higher levels of exercise. It has been demonstrated, using the multiple inert gas elimination technique, that the major cause of the arterial hypoxaemia in exercising horses is diffusion limitation, but that relative hypoventilation and mild mismatch also contribute (Wagner et al. 1989). A morphometric and physiological study (Constantinopol et al. 1989) however, indicates the opposite, with relative hypoventilation and inhomogeneities being the most likely cause of the hypoxaemia. The relative hypoventilation during exercise may not result entirely from an inadequate increase in minute ventilation, but may be aggravated by an unusually high VD/VT ratio. It seems that in the present study the physiological dead space increases approximately in proportion to the increase in VT. The increase in physiological dead space is thought to be due to an increase in the anatomical, or airway, component (Jones et al. 1966), and this could be exaggerated in horses if areas of the lung are ventilated but remain relatively underperfused during exercise (Wasserman et al. 1967). A contributory factor to this condition could well be the anomalous blood supply of large portions of the lung (up to 22% of total lung volume) by bronchial arteries in horses with exercise-induced pulmonary haemorrhage (O’Callaghan et al. 1987). Also, the hypoxaemia that develops during higher levels of exercise could exacerbate any local tendencies towards hypoxia in the lung and this could lead to reduced alveolar perfusion (Kuriyama et al. 1984).
Mean pulmonary blood pressure increased to a similar level to that reported by Erickson et al. (1990). As there was a significant increase in this variable (at the trot) before there was a significant fall in (at 8 ms−1), the latter cannot be the cause of the former, at least at relatively low levels of exercise. The study of Pelletier and Leith (1991) suggests that pulmonary hypoxia is not an important mechanism in pulmonary hypertension, even at high levels of exercise. As both cardiac output and packed cell volume increase significantly, even when the horses are walking, these are the most likely causes of pulmonary hypertension (Erickson et al. 1990). It is interesting to speculate that the high pulmonary pressure could cause oedema (cf. Younes et al. 1987), which would contribute to the diffusion limitation of the lungs during exercise (but see West et al. 1991).
The significance of the large increases in heart rate and in , despite the arterial hypoxaemia in the cardiovascular response to exercise in the horse, has been noted previously (Evans and Rose 1988c; Jones et al. 1989a). The values of cardiac output, cardiac stroke volume and in the present study are similar to those determined by the above groups. is similar to that reported by Wagner et al. (1989).
Changes in arterial and mixed venous pH values were similar to those reported by Bayly et al. (1989) and the increase in blood lactate concentration was similar to that reported by Seeherman and Morris (1990) for adult Thoroughbred horses. However, high-level exercise to exhaustion in the laboratory (Rose et al. 1988) and when racing in the field (Bayly et al. 1987a) causes lactate concentrations that are twice as great as those determined in the present study. There is evidence that, as might be expected, blood pH is lower at the higher values of lactate concentration (Bayly et al. 1983). Presumably, such low pH values tend to shift the oxygen equilibrium curve (OEC) to the right and desaturate the arterial blood, thus exacerbating the effect of the hypoxaemia.
Another factor that is almost bound to influence the OEC is the large rise in arterial temperature that occurs during intense exercise. It is clear from the study of Jones et al. (1989b) that deep rectal temperature gives a false impression of good thermoregulation during heavy exercise in horses since it is approximately 2.5°C lower than arterial blood temperature after 6min of exercise at . An increase in blood temperature of 4°C would be expected to be a powerful stimulus to ventilation and, as discussed earlier, the mechanical linkage of ventilation with limb movements may prevent thermal and metabolic stimuli from exerting their influence on ventilation during exercise. Certainly, during the post-exercise recovery period, respiratory minute volume remains significantly elevated above the resting value, mainly as a result of the high respiratory frequency, despite a dramatic reduction in gas exchange. Arterial and mixed venous pH return almost to the resting values, but blood temperature is still over 1°C above the resting value. It is deduced, therefore, that immediately upon cessation of intense exercise, minute ventilation is high because of thermal and metabolic stimuli. As the recovery period continues, the latter exerts proportionately less influence. A strong thermal stimulus may also explain the high value of VD/VT during recovery.
Does the temperature in the arterial blood reflect that in the brain? Certainly during cooling in the cat it does (Kiley et al. 1984). However, in a number of mammals it has been demonstrated that, during heat load, brain temperature may be lower than the temperature of the arterial blood (e.g. Baker and Hayward, 1968; Baker, 1972; Johnson et al. 1987) and this is also true during exercise-induced hyperthermia (Taylor and Lyman, 1972; Caputa et al. 1983; Chesy et al. 1985). Whether this exists in the horse remains to be seen, but it could be that an elevated brain temperature has an inhibiting effect on exercise and, therefore, contributes to the phenomenon of fatigue (see Caputa et al. 1986). It is also evident from the data presented here that limitations of the respiratory system may well restrict the maximum performance of the Thoroughbred horse.
This report confirms some results obtained by other workers, but discrepancies have been found between previous work and the current study. It is probable that the new technique of measuring respiratory variables used here (the ultrasonic flowmeter) significantly reduces the effect of the measuring system on the measured variables, an effect very evident in many of the previously reported studies.
Acknowledgements
The authors wish to thank Drs L. Anderson and S. Young for their assistance with some of the mask validation experiments and Professor A. Guz for the use of his ‘metabolic pump’. The work was supported by the Horserace Betting Levy Board and Sheikh Mohammed al Maktoum. K.S. was in receipt of an SERC studentship.