Despite being the subject of considerable research over the past 20 years, the mechanism of exercise-induced hypoxemia (EIH) remains uncertain. Although evidence exists in support of an end-capillary oxygen diffusion limitation (5,7,31), a ventilation-perfusion inequality (5,7,12,29,30), and an inadequate ventilatory response to exercise (2,3,8,19,22), there is no consensus. EIH was earlier defined as a phenomenon occurring in young, male endurance athletes (maximal oxygen uptake, V̇O2max > 68 mL·kg−1·min−1) (17) performing high-intensity exercise (>70% V̇O2max) (2). More recently, it has been reported in females (6,9,11,27) and older males (20), reflecting a much broader spectrum of aerobic power, and at exercise intensities as low as 40% of peak oxygen uptake (V̇O2peak) (22). The proportion of athletes with EIH in a given subject cohort is dependent, in part, on the definition applied. Reductions in the partial pressure of oxygen in arterial blood (PaO2) ranging through 8 (15,21), 10 (3,22,23,27), 12 (24), and 18 mm Hg (18,19) have all been used to define the phenomenon. Given that no change or an increase in PaO2 would seem the normal response to exercise of increasing intensity (24,31), any decrease in PaO2, beyond the typical error of measurement, could be deemed EIH. Of much greater import is that a number of previous studies have not corrected blood gas measurements for the thermogenic effect of exercise (3,5,6,8,21). Given that rectal temperature can increase by 1–2°C during high-intensity exercise, failure to correct the blood gas values for this change will overestimate any hypoxemia by ~5 mm Hg for each 1°C increase in body temperature above 37°C. In consequence, whichever definition is applied, the true proportion of athletes with EIH cannot be judged with any certainty. Furthermore, although rectal temperature has been a common core temperature reference site (3,18,22), it may not be the most appropriate in the context of defining oxygen status during exercise. A more complete picture may be provided by correcting each arterial blood sample for the temperature in the esophagus (reflecting oxygen loading temperature) and arterial blood (reflecting oxygen delivery temperature). Whereas some previous studies have applied radial arterial blood (16,23,30) and esophageal (10,11,13) temperature corrections, their effect on the frequency of EIH in humans remains unknown. One study of exercise in horses and steers (14) has demonstrated a reduced hypoxemic response when comparing blood gas samples corrected for muscle and pulmonary artery blood temperature with those corrected for rectal temperature. Therefore, the present study was undertaken to examine the frequency of EIH during incremental exercise in trained men, when the simultaneous changes in rectal, esophageal, and arterial blood temperatures were used to correct arterial blood gas measurements. For the purposes of data analysis, a decrease in PaO2 of ≥ 10 mm Hg was chosen as the arbitrary index of EIH.
Subjects and Experimental Conditions
Ten endurance-trained males with no history of lung disease participated in the study (Table 1). One week before the main experiment, all subjects were familiarized with treadmill running and the use of a mouthpiece for expired gas collection and performed a lung diffusion capacity for carbon monoxide (DLCO) test, measured on a SensorMedics VMAX system in accordance with the standards of the American Thoracic Society. All experiments were conducted in an air-conditioned laboratory with room temperature maintained at 24 ± 0.5°C, and a relative humidity of 38 ± 5.8%, where values are means ± SD. Subjects arrived at the laboratory for the main experiment having abstained from vigorous exercise for the previous 24 h, and food and caffeine products for the previous 4 h. The experiment took place at the same time of the morning for each subject. Subjects signed informed consent forms acknowledging and accepting the testing requirements and protocols. The Human Research Ethics Committee of the Royal Adelaide Hospital approved the protocol.
On arrival, subjects were asked to void and self-insert a rectal temperature probe. Chest electrodes were applied (Red DotTM, 3M Health Care, Borken, Germany) for recording the ECG and a chest transmitter attached to obtain a record of heart rate as consecutive 5-s averages (Polar Accurex heart rate monitor, Kempele, Finland). Subjects then lay supine on a couch while a Teflon catheter (20 gauge/32-mm, Critikon, Tampa, FL) was inserted under local anesthesia (2% lignocaine hydrochloride, Xylocaine) into the radial artery at the wrist of the nondominant arm, with the catheter tip directed toward the heart. One end of a 10-cm J-loop (Becton Dickinson Vascular Access, Utah) was attached to the radial artery catheter hub and the other end to a Y-adapter (Tuta Laboratories, Adelaide, Australia) with separate injection and sampling ports.
The disposable rectal temperature probe (12-French Gauge, series 400 Mon-a-thermTM, Mallinckrodt Inc., St. Louis, MO) was self-inserted such that the thermistor tip lay ~14 cm proximal to the anal sphincter.
Radial artery blood.
Before attachment of the J-loop to the catheter hub, a rapid response, Teflon-coated, thermocouple (IT-21, Physitemp Instruments, Clifton, NJ) was passed through the lumen of a 21-gauge needle inserted into the rubber cap covering the injection port of the Y-adapter. It was then maneuvered through the J-loop such that when connection was made with the radial artery catheter hub the thermocouple tip sat at the opening of the catheter in the radial artery.
The esophageal thermistor probe (9-French Gauge, Mon-a-thermTM) was inserted through an anterior naris and passed posteriorly to a depth of ~30 cm below the oropharynx. Before insertion of the probe, the nasal passage was anesthetized with local anesthetic gel (XylocaineTM Viscous, Astra Pharmaceuticals Pty. Ltd., New South Wales, Australia), and the final position of the thermistor tip was chosen as that which elicited the highest temperature reading, this being assumed to reflect pulmonary artery blood temperature most closely (26).
To obtain readings of radial artery blood temperature the probe was connected to a digital thermometer (TH-5 Physitemp) and the highest value observed during each blood sampling period was manually recorded. To obtain readings of rectal and esophageal temperatures, the probes were connected to a dual input digital thermometer (CIG Industries, Adelaide, SA). Readings from all three probes were taken simultaneously, at the time of arterial blood sampling.
Before each experiment, thermocouples were checked for accuracy, precision, and linearity in a stirred water bath, using a NATA certified spirit thermometer (Cameron Instrumentation, Adelaide, SA) as the reference. Water temperature spanned from 33 to 43°C, and measurements were taken at 0.5°C increments. Due to the need to maintain sterility, esophageal temperature probes were checked postexercise, while rectal probes were calibrated from samples of their batch code. Blood probes were also calibrated postexercise, sterilized in hydrogen peroxide vapor for ~1 h and then reused. Regression equations were developed for each probe and applied to correct the data collected throughout the exercise test protocol.
Exercise began with the subject standing astride the treadmill belt, a nose clip applied and an externally supported breathing valve (Hans Rudolph, HR2700) inserted into the mouth for collection of expired gases. The resting arterial blood sample was obtained at this time. The treadmill was then started with zero grade, and when the speed reached 4 km·h−1, the subject began walking on the treadmill belt. After 2 min, and every 2 min thereafter, the speed was increased by 2 km·h−1, until a speed of 12 km·h−1 was reached, after which further workloads were generated by ramping the treadmill grade 2% every 2 min until volitional exhaustion.
Blood sampling and analysis.
Whole blood samples (5 mL) for blood gas analysis were drawn in the final 15 s of each 2-min workload. The samples were drawn anaerobically into preheparinized ground-glass syringes, capped immediately, and stored vertically in melting ice. Before analysis (within 30 min of collection), the samples were remixed and then analyzed in duplicate at 37°C (ABL520, Radiometer, Copenhagen) for partial pressures of oxygen (PaO2) and carbon dioxide (PaCO2). Blood gases were corrected for rectal, arterial blood, and esophageal temperatures measured simultaneously at the time of blood sampling (25), both at rest and during exercise. Before obtaining each blood sample, the initial 4 mL of the contents of the catheter system (blood-saline mixture) were drawn into a separate sterile syringe. This volume was re-injected after each blood sampling and the catheter system then cleared with ~5 mL of heparinized saline.
Determination of V̇O2peak
Ventilation (V̇E) was calculated using a linear pneumotachograph (Hans Rudolph, Model No. 4813, Kansas City, MO) attached to the inspiratory port of the one-way breathing valve (Hans Rudolph, HR2700). The subject’s expirate was passed through a 5-L mixing chamber and subsampled for analysis of oxygen (O2) and carbon dioxide (CO2) percentages using a mass spectrometer (Amis 2000, Odense, Denmark). The spectrometer was calibrated before each exercise test with gas mixtures of known O2 and CO2 percentages covering the physiological range of measurement and the pneumotachograph was calibrated with a 3-L syringe at varying flow rates. Metabolic data were calculated every 30 s and recorded on a PC-based computer with corrections for temporal misalignment due to the dead space of the expiratory circuit. V̇O2was expressed in terms of total body mass (mL·kg−1·min−1), with V̇O2peak being the mean V̇O2 of the minute in which the highest 30-s value was recorded.
Values are expressed as mean ± SD. A one-way ANOVA was used to examine the main effects of temperature measurement site, and the effects of the several temperature corrections on Δ PaO2 and ΔPaCO2. Where overall significance was obtained, differences between means were identified with a Tukey’s honestly significant difference test. Strength of association between the dependent and independent variables was tested using Pearson’s product moment correlation. Alpha was set at P ≤ 0.05, and all analyses were conducted using Statistica software (Ver. 5.0, Statsoft, Tulsa, OK).
Peak cardiorespiratory responses to incremental exercise.
Ventilation and heart rate increased with workload in all subjects (Table 2) to reach mean peak values at volitional exhaustion of, respectively, 146.7 ± 22.4 L·min−1 and 190 ± 8 beats·min−1, with the latter being 100.7% of the mean age-predicted value [208 minus (0.7 × age in years)] (28). V̇O2peak ranged from 53.9 to 78.0 mL·kg−1·min−1, with a mean value of 65.4 ± 7.0 mL·kg−1·min−1 (Table 2). Mean lung diffusion capacity for carbon monoxide (DLCO) was 104.9% of the predicted value (36.2 mL·min−1·mm Hg−1).
Temperature changes during exercise.
The mean values for resting and end-exercise temperatures at all three measurement sites are shown in Table 3. Temperature at all sites increased in response to exercise with the mean changes at the arterial (2.3 ± 0.5°C) and esophageal (2.4 ± 0.5°C) measurement sites being nearly double that recorded with the rectal probe (1.4 ± 0.4°C) over the same time period (P < 0.001 in each case) (Table 2).
PaO2 changes from rest to end exercise.
The temperature corrected and uncorrected mean values of PaO2 at rest and end exercise are shown in Table 3. When values were not corrected for temperature change, PaO2 fell in all subjects (Table 4), and 8 of the 10 subjects met the chosen criterion for EIH. When the values were corrected for rectal temperature, PaO2 fell in 9 of the 10 subjects, but only 5 met the criterion for EIH. Using either arterial blood or esophageal temperatures, PaO2 fell in seven subjects with only two meeting the criterion for EIH, and in each case the mean changes were not significantly different from zero. Whereas the mean values for Δ PaO2 (end exercise minus rest) were not different when corrected for arterial and esophageal temperature changes, all other mean values were different from each other (F(3,27) = 115.2, P < 0.0001). There was no significant correlation between Δ PaO2 and either V̇Epeak or V̇O2peak, whether uncorrected for temperature (r = 0.20, P = 0.57; r = 0.31, P = 0.39, respectively), or corrected for rectal (r = 0.21, P = 0.56; r = 0.31, P = 0.39, respectively), arterial (r = 0.28, P = 0.43; r = 0.36, P = 0.31, respectively), or esophageal (r = 0.46, P = 0.17; r = 0.50, P = 0.14, respectively) temperature changes.
PaCO2 changes from rest to end-exercise.
The temperature corrected and uncorrected mean values of PaCO2 at rest and end exercise are shown in Table 3. PaCO2 fell in all subjects from rest to end exercise when uncorrected for temperature change (Table 5) and in 8 of the 10 when corrected for rectal, arterial blood, and esophageal temperatures. Whereas the mean values for Δ PaCO2 (end exercise minus rest) were not different when corrected for arterial and esophageal temperature changes, all other mean values were different from each other (F(3,27) = 110.8, P < 0.0001). The greatest fall in PaCO2was for uncorrected data, which was larger than the fall for rectal temperature corrected, and in turn was greater than those for the arterial blood and esophageal temperature corrected values. There was no correlation between Δ PaCO2 and either Δ PaO2, V̇Epeak or V̇O2peak, whether uncorrected for temperature (r = 0.39, P = 0.26; r = 0.38, P = 0.27; r = 0.24, P = 0.50, respectively) or corrected for rectal (r = 0.31, P = 0.38; r = 0.36, P = 0.30; r = 0.21, P = 0.55, respectively), arterial (r = 0.28, P = 0.43; r = 0.32, P = 0.37; r = 0.19, P = 0.59, respectively), or esophageal (r = 0.30, P = 0.39; r = 0.18, P = 0.61; r = 0.08, P = 0.83, respectively) temperature changes.
The clear implication from the results of the present study is that when investigating EIH, all PaO2 values should be corrected for any exercise-induced increase in body temperature. Failure to do so will overestimate the proportion of subjects with EIH, regardless of which definition is applied. Another critical issue to emerge from the present study is the effect of the temperature measurement site chosen for blood gas corrections. When the PaO2 values were corrected for rectal temperature, 50% of the subjects in the present study met the ≥10 mm Hg fall criterion for EIH, a result in agreement with previous reports in the literature where this site was used to monitor the thermogenic effect of exercise (3,22). However, when the values were corrected for either arterial blood or esophageal temperatures, only 2 of the 10 subjects met the criterion for EIH and the mean group change in PaO2 was not significantly different from zero. Although rectal temperature has been used widely as the index of core temperature changes with exercise, it is well known to be slow to respond to changes in blood temperature (4) with measurements biased by intrapelvic muscle temperature and a reduction in heat dissipation during leg exercise (1). The protocol chosen in the present study, where exercise intensity was incremented until volitional exhaustion and the exercise session was of moderate duration, contrasts with studies where steady-state, high-intensity exercise of short duration (5 min or less) was more usual (2,8,19,24). Quite clearly, the profile and magnitude of the temperature changes will vary at different “core” sites in such diverse protocols. Esophageal temperature, which provided the highest “core” temperature recorded during exercise in the present study (41°C), is regarded as that which most closely reflects changes in pulmonary artery blood temperature (26). Hence, correcting blood gas measurements in sampled arterial blood for the corresponding esophageal temperature should be the most appropriate procedure to identify any problem in oxygen loading during exercise.
EIH has been regarded as a phenomenon of endurance-trained athletes (2,3,17), yet, both in this (eight subjects) and a previous study (15 subjects) in our laboratory (Rice et al., 22), no significant correlation has been found between the hypoxemic effect and V̇O2peak. Despite relatively uniform changes in core temperature in all 10 subjects at each of the three measurement sites, the falls in the uncorrected values for PaO2 across the complete exercise spectrum ranged from 0.5 to 27.4 mm Hg. It is difficult to propose a physiological basis for this diversity, particularly given the lack of correlation of PaO2 with either ΔPaCO2, V̇O2peak, or V̇Epeak.
In conclusion, on the evidence presented in the present study, the frequency of EIH in a given population is overestimated when arterial blood gases are either not corrected for body temperature changes during exercise or are corrected for temperature measured at a rectal site.
The authors are grateful for the support provided by the Lung Function Unit in the Department of Thoracic Medicine at the Royal Adelaide Hospital. N. Shipp gratefully acknowledges his support from the Dr. Roger Angove Research Fellowship.
1. Aulick, L. H., S. Robinson, and S. P. Tzankoff. Arm and leg intravascular temperatures of men during submaximal exercise. J. Appl. Physiol.
2. Dempsey, J. A., P. G. Hanson, and K. S. Henderson. Exercise-induced arterial hypoxaemia in healthy human subjects at sea level. J. Physiol.
) 355:161–175, 1984.
3. Durand, F., P. Mucci, and C. Préfaut. Evidence for an inadequate hyperventilation inducing arterial hypoxemia at submaximal exercise in all highly trained endurance athletes. Med. Sci. Sports Exerc.
4. Edwards, R. J., A. J. Belyavin, and M. H. Harrison. Core temperature
measurement in man. Aviat. Space Environ. Med.
5. Gale, G. E., J. R. Torre-Bueno, R. E. Moon, H. A. Saltzman, and P. D. Wagner. Ventilation-perfusion inequality in normal humans during exercise at sea level and simulated altitude. J. Appl. Physiol.
6. Gore, C. J., S. C. Little, A. G. Hahn, et al. Reduced performance of male and female athletes at 580 m altitude. Eur. J. Appl. Physiol.
7. Hammond, M. D., G. E. Gale, K. S. Kapitan, A. Ries, and P. D. Wagner. Pulmonary gas exchange in humans during exercise at sea level. J. Appl. Physiol.
8. Harms, C. A., and J. M. Stager. Low chemoresponsiveness and inadequate hyperventilation contribute to exercise-induced hypoxemia. J. Appl. Physiol.
9. Harms, C. A., S. R. McClaran, G. A. Nickele, D. F. Pegelow, W. B. Nelson, and J. A. Dempsey. Exercise-induced arterial hypoxaemia in healthy young women. J. Physiol. (Lond.)
10. Harms, C. A., S. R. McClaran, G. A. Nickele, D. F. Pegelow, W. B. Nelson, and J. A. Dempsey. Effect of exercise-induced arterial O2
desaturation on VO2max
in women. Med. Sci. Sports Exerc.
11. Hopkins, S. R., R. C. Barker, T. D. Brutsaert, et al. Pulmonary gas exchange during exercise in women: effects of exercise type and work increment. J. Appl. Physiol.
12. Hopkins, S. R., D. C. McKenzie, R. B. Schoene, R. W. Glenny, and H. T Robertson. Pulmonary gas exchange during exercise in athletes. I. Ventilation-perfusion mismatch and diffusion limitation. J. Appl. Physiol.
13. Johnson, B. D., K. W. Saupe, and J. A. Dempsey. Mechanical constraints on exercise hyperpnea in endurance athletes. J. Appl. Physiol.
14. Jones, J. H., C. R. Taylor, A. Lindholm, R. Straub, K. E. Longworth, and R. H. Karas. Blood gas measurement during exercise: errors due to temperature correction. J. Appl. Physiol.
15. Mucci, P., F. Durand, B. Lebel, J. Bousquet, and C. Prefaut. Basophils and exercise-induced hypoxemia in extreme athletes. J. Appl. Physiol.
16. Nielsen, H. B., P. P. Bredmose, M. Stromstad, S. Volianitis, B. Quistorff, and N. H. Secher. Bicarbonate attenuates arterial desaturation during maximal exercise in humans. J. Appl. Physiol.
17. Powers, S. K., S. Dodd, J. Lawler, et al. Incidence of exercise induced hypoxemia in elite athletes at sea level. Eur. J. Appl. Physiol.
18. Powers, S. K., S. Dodd, D. Criswell, J. Lawler, D. Martin, and S. Grinton. Evidence for an alveolar-arterial PO2
gradient threshold during incremental exercise. Int. J. Sports Med.
19. Powers, S. K., D. Martin, M. Cicale, N. Collop, D. Huang, and D. Criswell. Exercise-induced hypoxemia in athletes: role of inadequate hyperventilation. Eur. J. Appl. Physiol.
20. Préfaut, C., F. Anselme, C. Caillaud, and J. Masse-Biron. Exercise-induced hypoxemia in older athletes. J. Appl. Physiol.
21. Préfaut, C., F. Anselme, and C. Caillaud. Inhibition of histamine release by nedocromil sodium reduces exercise-induced hypoxemia in master athletes. Med. Sci. Sports Exerc.
29:10 –16, 1997.
22. Rice, A. J., G. C. Scroop, C. J. Gore, et al. Exercise-induced hypoxaemia in highly trained cyclists at 40% peak oxygen uptake
. Eur. J. Appl. Physiol.
23. Rice, A. J., A. T. Thornton, C. J. Gore, et al. Pulmonary gas exchange during exercise in highly trained cyclists with arterial hypoxemia. J. Appl. Physiol.
24. Rowell, L. B., H. L. Taylor, Y. Wang, and W. S. Carlson. Saturation of arterial blood with oxygen during maximal exercise. J. Appl. Physiol.
25. Severinghaus, J. W. Simple, accurate equations for human blood O2
dissociation computations. J. Appl. Physiol.
26. Shiraki, K., N. Konda, and S. Sagawa. Esophageal and tympanic temperature responses to core blood temperature changes during hyperthermia. J. Appl. Physiol.
27. St Croix, C. M., C. A. Harms, S. R. McClaran, et al. Effects of prior exercise on exercise-induced arterial hypoxemia in young women. J. Appl. Physiol.
28. Tanaka, H., K. Monahan, and D. Seals. Age-predicted maximal heart rate revisited. J. Am. Coll. Cardiol.
29. Torre-Bueno, J. R., P. D. Wagner, H. A. Saltzman, G. E. Gale, and R. E. Moon. Diffusion limitation in normal humans during exercise at sea level and simulated altitude. J. Appl. Physiol.
30. Wagner, P. D., G. E. Gale, R. E. Moon, J. R. Torre-Bueno, B. W. Stolp, and H. A. Saltzman. Pulmonary gas exchange in humans exercising at sea level and simulated altitude. J. App. Physiol.
31. Wasserman, K., J. E. Hansen, D. Y. Sue, B. J. Whipp, and R. Casaburi (Eds.). Principles of Exercise Testing and Interpretation
, 2nd Ed. Philadelphia: Lea & Febiger, 1994, p. 69.