Journal of Applied Physiology Fuel your research with LabChart
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

J Appl Physiol 87: 1802-1812, 1999;
8750-7587/99 $5.00
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF) Free
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Rice, A. J.
Right arrow Articles by Hopkins, S. R.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Rice, A. J.
Right arrow Articles by Hopkins, S. R.
Vol. 87, Issue 5, 1802-1812, November 1999

Pulmonary gas exchange during exercise in highly trained cyclists with arterial hypoxemia

Anthony J. Rice1, Andrew T. Thornton1, Christopher J. Gore3, Garry C. Scroop4, Hugh W. Greville1, Harrieth Wagner2, Peter D. Wagner2, and Susan R. Hopkins2

1 Department of Thoracic Medicine, Royal Adelaide Hospital, Adelaide, South Australia 5000; 2 Department of Medicine, University of California San Diego, La Jolla, California 92093; 3 Australian Institute of Sport, Belconnen, Australian Capital Territory 2616; and 4 Department of Physiology, University of Adelaide, Adelaide, South Australia 5000, Australia


    ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES

The causes of exercise-induced hypoxemia (EIH) remain unclear. We studied the mechanisms of EIH in highly trained cyclists. Five subjects had no significant change from resting arterial PO2 (PaO2; 92.1 ± 2.6 Torr) during maximal exercise (C), and seven subjects (E) had a >10-Torr reduction in PaO2 (81.7 ± 4.5 Torr). Later, they were studied at rest and during various exercise intensities by using the multiple inert gas elimination technique in normoxia and hypoxia (13.2% O2). During normoxia at 90% peak O2 consumption, PaO2 was lower in E compared with C (87 ± 4 vs. 97 ± 6 Torr, P < 0.001) and alveolar-to-arterial O2 tension difference (A-aDO2) was greater (33 ± 4 vs. 23 ± 1 Torr, P < 0.001). Diffusion limitation accounted for 23 (E) and 13 Torr (C) of the A-aDO2 (P < 0.01). There were no significant differences between groups in arterial PCO2 (PaCO2) or ventilation-perfusion (VA/Q) inequality as measured by the log SD of the perfusion distribution (logSDQ). Stepwise multiple linear regression revealed that lung O2 diffusing capacity (DLO2), logSDQ, and PaCO2 each accounted for ~30% of the variance in PaO2 (r = 0.95, P < 0.001). These data suggest that EIH has a multifactorial etiology related to DLO2, VA/Q inequality, and ventilation.

ventilation-perfusion inequality; pulmonary diffusion limitation; exercise


    INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES

EXERCISE-INDUCED HYPOXEMIA (EIH), with a reduction in arterial PO2 (PaO2) and a widening of the alveolar-arterial O2 tension difference (A-aDO2), can be found in ~50% of trained athletes capable of sustaining metabolic rates in excess of 4.5 l/min (27). The possible causes of EIH include intra- and extrapulmonary shunt, ventilation-perfusion (VA/Q) inequality, and O2 diffusion limitation (5, 8, 19, 20). In addition, if alveolar ventilation does not rise sufficiently to match the increase in metabolic rate during heavy exercise, PaO2 will decrease because of the direct effect ventilation has on alveolar PO2 (PAO2) (5, 9, 16, 24, 28). Recently, the development of EIH has been linked to gender-related lung size (15, 18).

The multiple inert gas elimination technique (MIGET) has previously been used to measure VA/Q inequality and O2 diffusion limitation in healthy subjects while exercising at sea level, and in hypobaric and normobaric hypoxia (8, 11, 12, 20, 31, 34, 36, 39). However, only a handful of studies has attempted to quantify the relative contributions of VA/Q inequality and O2 diffusion limitation in athletes during heavy exercise (11, 20, 39) and none has used highly trained athletes with documented EIH.

Therefore, the purpose of this study was to evaluate the relative contributions of the above-mentioned mechanisms of EIH in highly trained cyclists with EIH and compare their results with those from a similar group of highly trained cyclists without EIH. This was achieved by employing MIGET to investigate VA/Q inequality and O2 diffusion limitation during and up to 45 min after heavy exercise. As MIGET is unable to distinguish between O2 diffusion limitation and extrapulmonary shunt (from bronchial and thebesian veins), and we felt it imperative to make such a distinction, we also measured VA/Q inequality and O2 diffusion limitation while breathing 13% O2. This has the effect of both increasing the amount of O2 diffusion limitation and decreasing the degree to which extrapulmonary shunt affects PaO2 due to the position on the oxyhemoglobin curve on which hypoxia places each individual, and thus qualifies the contribution of extrapulmonary shunt to the measured reduction in PaO2.


    METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES

Subject Selection and Preliminary Studies

Twelve healthy male cyclists were selected from a group of 20 who had previously been screened for the presence or absence of EIH (30). Their characteristics are shown in Table 1. All subjects had a peak O2 consumption (VO2 peak) > 65 ml · kg-1 · min-1, or 4.5 l/min, and reported no history of cardiovascular or respiratory disease. The subjects were divided into two groups before the commencement of the study: control (C; n = 5) and experimental (E; n = 7) on the basis of their PaO2 (temperature corrected) at the end of a progressive, incremental exercise test to exhaustion. Although the C group showed no significant reduction in PaO2 from rest to that at the end of exercise [92.1 ± 2.6 (SD) Torr; range 89.8-95.0], the E subjects each demonstrated a decline in PaO2 from rest of at least 10 Torr (81.7 ± 4.5 Torr; range 76.9-88.9, between-group P < 0.001).

                              
View this table:
[in this window]
[in a new window]
  		Table 1.   Anthropometric and preliminary data for control and experimental subjects

Experimental Design

The experimental protocol was approved by the Research Ethics Committee of the Royal Adelaide Hospital, and all subjects gave written informed consent. Experiments were conducted in an air-conditioned laboratory with a room temperature maintained in the range of 20-22°C. Subjects arrived at the laboratory having abstained from exercise for the previous 24 h and from food and caffeine products for the previous 4 h. Each subject was required to complete four separate visits to the laboratory. The initial two visits measured VO2 peak during an incremental exercise protocol while breathing either air (20.93% O2) or hypoxic gas (13.2 ± 0.1% O2). These two visits were used to establish the workloads for the subsequent components of this study. During the third visit, inert gas exchange was measured during submaximal exercise while breathing both normoxia and hypoxia. Cardiac output (Q) was measured during the final visit at times of the day and workloads identical to those employed during the third visit by using a quasi-steady-state acetylene (C2H2) uptake technique (1). It was necessary to measure Q on a separate day as the relatively high concentration of C2H2 used to measure Q can interfere with the measurement of MIGET inert gas concentrations. All four visits were completed within a 4-wk period, with the final two visits completed within a week of each other.

Preliminary Incremental Exercise Protocol to Establish VO2 peak

After entering the laboratory, each subject had his height and weight recorded, and chest electrodes were applied to measure both heart rate (Sport-Tester, Polar OY) and electrocardiogram (Hewlett-Packard, Wilmington, DE). The subject then assumed a racing position on a calibrated air-braked cycle ergometer (Peter Bundy Cycles, Sydney, NSW, Australia) with power output displayed on a monitor. Each subject commenced exercise at a workload of 150 or 100 W while breathing the normoxic or hypoxic gas mixtures, respectively. The workload was increased 25 W every 2 min thereafter until volitional exhaustion. All cardiorespiratory variables were monitored continuously throughout exercise. Normoxic and hypoxic trials were separated by at least 2 days, but both were completed within a 2-wk period.

Determination of VO2 peak

O2 consumption (VO2) was measured with an on-line, indirect calorimetry system that measured inspired ventilation and both inspired and mixed expired gas fractions. The order of the inspired gas was randomized and balanced for the 12 subjects. The dry inspired gas mixtures were contained in pressurized cylinders, verified by the distributor (BOC Gases, Chatswood, NSW, Australia) to be accurate within ±0.1% O2. The dry gases were passed through a heated water bath to humidify the gas to ~50% and were then stored in a 2,000-liter impermeable aluminium bag (Scholle Industries, Elizabeth, SA, Australia). Leading from the aluminium bag was a T-piece connector and 1.5 m of respiratory tubing (Vacu-Med, Ventura, CA). The T piece housed a rapid-response temperature and relative humidity probe (HMP230, Viasala OY) as well as a sample port connected to a paramagnetic O2 analyzer (Normocap 200, Datex Medical Instruments, Tewksbury, MA) for the measurement of inspired O2 fraction (FIO2). Downstream from the respiratory tube was a linearized pneumotachograph (model 3, Fleisch, Lausanne, Switzerland) and a ±2 cmH2O differential pressure transducer (Validyne DP45, Northridge, CA) that, combined with real-time measurements of FIO2, gas temperature, and relative humidity, allowed the calculation of inspired minute ventilation. The subjects inhaled through a nonrebreathing respiratory valve (R2700, Hans Rudolph, Kansas City, MO) with the expired volume directed to a 5-liter baffled mixing box. Expired O2 and CO2 fraction were subsampled from the mixing box at a rate of 550 ml/min, dried with CaCl2, and then analyzed on a rapid-response zirconia (PK Morgan Zirconia, Rainham, Kent, UK) and infrared (Beckman LB-2, Beckman, Anaheim, CA) O2 and CO2 analyzer, respectively. The gas analyzers were calibrated beforehand and checked for drift immediately after the test protocol with two precision-grade gases of known concentration (BOC Gases). An IBM-compatible computer (386-SX) was programmed to determine the 30-s averages of expired minute ventilation (VE; BTPS), VO2 (STPD), CO2 production (VCO2; STPD), respiratory exchange ratio (RER), heart rate (HR), and power output. The two highest consecutive 30-s values obtained for VO2 during the incremental protocol were averaged and designated as the VO2 peak for each FIO2.

Subject Preparation for Inert Gas-Exchange Study

Before catheterization each subject had his weight recorded and then underwent routine pulmonary function testing [forced vital capacity (FVC), forced expiratory volume in 1 s (FEV1.0), and lung diffusing capacity for CO (DLCO)]. Then, under local anesthesia (1% lignocaine hydrochloride), each subject had two catheters placed percutaneously: 1) a 20-G radial arterial catheter (Arrow, Erding, Germany) in the left wrist and 2) an 18-G peripheral intravenous catheter (Insyte, Becton-Dickinson, Franklin Lakes, NJ) in the right forearm. One end of a 10-cm J loop (Becton-Dickinson) was attached to the radial artery catheter hub and the other end to a Y adaptor (Tuta Laboratories, Adelaide, SA, Australia) with two injection ports: one to allow the withdrawal of arterial blood via a three-way stopcock and the other to measure blood temperature by using a rapid-response (<0.2 s) Teflon-coated thermocouple (IT21, Physitemp Instruments, Clifton, NJ) that was positioned in the hub of the radial artery catheter. This method for blood temperature measurement has previously been described (20). The catheter, J-loop, and Y-adaptor system was routinely flushed with heparinized saline (15 IU/ml) to prevent clotting. The peripheral intravenous line was connected to a 1,000-ml infusion bag containing six inert gases (sulfur hexafluoride, ethane, cyclopropane, enflurane, ether, and acetone) dissolved in a 5% dextrose solution (38). This solution was infused (Masterflex model 7521, Barnant, Barrington, IL) via a 0.22-µm high-pressure Millipore filter into a peripheral vein at a rate of ~0.25 ml solution/min per liters per minute of exercise ventilation.

Inert Gas-Exchange Study Protocol

The catheterized subjects were seated on the cycle ergometer and connected to the same respiratory circuit as used during the VO2 peak measurements, with slight modifications to the expired tubing and mixing boxes. To minimize the loss of soluble inert gases (acetone), the expired breathing circuit, which consisted of expired tubing, a 10-liter mixing box (used for resting measurements) or a 27-liter mixing box (used for exercise measurements) was heated to ~40-45°C. VE, mixed expired gas fractions, and all ancillary measurements (inspired gas temperature, relative humidity, HR, and power output) were monitored continuously throughout rest and during each workload. The exercise workloads were continuous, and measurements were made for 5 min at each of the following points: 1) rest; 2) ~30% VO2 peak (light exercise); 3) ~60% VO2 peak (moderate exercise); 4) ~90% VO2 peak (heavy exercise); 5) 5 min postexercise; 6) 15 min postexercise; 7) 30 min postexercise; and 8) 45 min postexercise, for a total of eight measurement times. The individual exercise workloads were based on the earlier incremental tests described above. The order of the FIO2 administration was balanced.

Mixed expired inert gas samples and arterial blood samples were collected simultaneously during the final 2 min of each stage outlined above. The gas analyzers, pneumotachograph, and ancillary instruments were calibrated before rest, and the 5-min postexercise measurements and were checked for drift immediately after the 45-min postexercise measurement. Postexercise pulmonary function tests (FEV1.0, FVC, DLCO) were completed within 20 min of the final MIGET measurement.

Inert Gas Analysis

At each workload, duplicate 6-ml arterial blood samples for inert gases and duplicate 4-ml samples for arterial blood gases were collected in preheparinized ground-glass syringes. Simultaneously, 30-ml duplicate mixed expired gas samples were collected from the mixing box in glass syringes. Drawing of blood and mixed expired samples was coordinated to allow for the transit time of expired gas through the mixing chamber. The mixed expired and arterial blood inert gas samples were analyzed by gas chromatography (37). Retention and excretion values for all six gases were used to estimate the VA/Q distributions by the method of enforced smoothing, as previously described (7, 37, 38). Predicted values for PaO2, PaCO2, and A-aDO2 were calculated from the recovered VA/Q distribution (11, 22) on the assumption of alveolar-capillary diffusion equilibration. The log SD of the perfusion distribution (logSDQ) and the log SD of the ventilation distribution (logSDV) were calculated from the recovered distribution and used as overall indexes of VA/Q inequality. Additional dispersion indexes (DispR*, DispE, DispR*-E) were derived directly from the retention and excretion data, respectively (8). These three indexes describe the extent of VA/Q inequality independently of the 50-compartment model. DispR* and logSDQ are comparable in that both are parameters of the perfusion distribution. Similarly, DispE and logSDV describe the ventilation distribution. DispR*-E is an overall index of dispersion that has no counterpart in the moment analysis using the 50-compartment model. Mixed venous inert gas levels were calculated from the arterial and expired samples by using the Fick relationship and the measured Q and VE.

The algorithm of Hammond and Hempleman (13) was used to estimate an effective pulmonary diffusing capacity for O2 (DLO2), assuming a uniform distribution of diffusing capacity to blood flow. It calculates the expected PaO2 associated with the measured degree of VA/Q inequality, excluding the possibility of alveolar-capillary diffusion disequilibrium. Agreement between expected and measured PaO2, therefore, supports the conclusion that diffusion equilibration is complete, but if expected PaO2 exceeds the measured value, diffusion limitation is inferred, although extrapulmonary shunt (originating from bronchial and thebesian veins) cannot be excluded.

Blood-Gas Analysis

The arterial blood-gas samples were kept in ice and analyzed in duplicate (ABL 520, Radiometer, Copenhagen, Denmark) within 20 min of collection for PaO2, PaCO2, pH, arterial hemoglobin O2 saturation, and hemoglobin concentration. The blood-gas analyzer was calibrated at hourly intervals throughout the day, and the results from this machine were routinely submitted to the external quality assurance program of the Royal College of Pathologists of Australia. Blood-gas values were measured at 37°C and corrected for the appropriate blood temperature measured at rest, exercise, and recovery (32). Blood temperature from the thermocouple was taken as the highest temperature recorded on a digital display (Thermalert 5, Physitemp Instruments) during withdrawal of each arterial blood-gas sample.

Measurement of Q

Q was measured in both normoxia and hypoxia by using a C2H2 quasi-steady-state technique. The details of this method have been previously reported and show excellent agreement with direct Fick methods for measurement of Q (1). The identical experimental system was used as that described for the VO2 peak testing. Each subject exercised for 5 min at 30, 60, and 90% VO2 peak during both normoxia and hypoxia with measurements of VE, VO2, and VCO2 recorded breath by breath and averaged every 30 s. End-tidal PCO2 (PETCO2) was measured by using an infrared CO2 analyzer (Beckman LB-2) and a strip-chart recorder (Neomedix Systems, Dee Why, NSW, Australia). After 4 min at each workload, subjects inspired a gas mixture (1% C2H2-20.9 or 13.1% O2-balance N2) for ~30-40 breaths. End-tidal concentrations of C2H2 were measured by using gas chromatography, and the blood-gas partition coefficient for C2H2 in blood was measured by using the method of Wagner et al. (38). The steady-state relationship between inspired and end-tidal C2H2 was determined and then extrapolated back to the first breath to account for C2H2 recirculation. Q was calculated according to the following equation
<A><AC>Q</AC><AC>˙</AC></A> = <FENCE><FR><NU><A><AC>V</AC><AC>˙</AC></A><SC>e</SC> ⋅ P<SC>e</SC><SUB>CO<SUB>2</SUB></SUB> ⋅ (P<SC>i</SC><SUB>C<SUB>2</SUB>H<SUB>2</SUB></SUB> − P<SC>et</SC><SUB>C<SUB>2</SUB>H<SUB>2</SUB></SUB>)</NU><DE>(&lgr; ⋅ P<SC>et</SC><SUB>CO<SUB>2</SUB></SUB> ⋅ P<SC>et</SC><SUB>C<SUB>2</SUB>H<SUB>2</SUB></SUB>)</DE></FR></FENCE>
where VE is expired minute ventilation (l/min, BTPS); PECO2 is mixed expired PCO2; PIC2H2 is inspired C2H2; PETC2H2 is end-tidal C2H2; and lambda  is the C2H2 blood-gas partition coefficient (BTPS). This equation reflects steady-state mass balance for C2H2 across the lung and specifically assumes negligible alveolar-arterial difference for C2H2 and no significant recirculation of C2H2 into mixed venous blood.

Data Analysis

Data are expressed as means ± SD. A three-factor repeated-measures ANOVA (group, FIO2, and workload) revealed that the relative exercise intensities (%VO2 peak) during moderate and heavy normoxic exercise and moderate and heavy hypoxic exercise were significantly different from each other, and thus no comparisons were made between the two FIO2 values. A two-factor repeated-measures ANOVA was used to determine significant differences by group (C vs. E) and workload during each FIO2. Where overall significance was obtained, differences between cell means were identified with Tukey's post hoc analysis for unequal numbers. Stepwise multiple linear regression was used to predict PaO2 on the basis of DLO2, logSDQ, and PaCO2 at 90% VO2 peak during normoxia. All the independent variables were introduced first for the analysis, and the more suitable variables were chosen so that the F-value of a final multiple regression model, and partial F-values of the independent variables, became maximum and significant for the F-distribution. Standardized regression coefficients (beta coefficients) of the independent variables, which were defined as regression coefficients standardized for the units of the variables, were considered to indicate the relative contribution of the independent variables to PaO2. The level of significance was set at P < 0.05. All analyses were conducted by using Statistica (version 5.0, Statsoft, Tulsa, OK).


    RESULTS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES

Analysis According to Experimental Grouping of Subjects

Observed A-aDO2. Exercise resulted in a significant widening of the measured A-aDO2 in both groups (Fig. 1, Table 2). During both moderate and heavy exercise, normoxia and hypoxia resulted in significantly larger A-aDO2 than that at rest for both the E and C groups. Additionally, light exercise caused a significant widening of the measured A-aDO2 in the hypoxia trial for both groups (Fig. 1, Table 3). The E group developed significantly larger A-aDO2 values than did the C group during moderate and heavy exercise while breathing air, but there was no difference between the groups during hypoxia.


View larger version (18K):
[in this window]
[in a new window]
  		Fig. 1.   Observed (solid symbols) and predicted (open symbols) alveolar-arterial oxygen pressure difference (A-aDO2) in control (circles, n = 5) and experimental (squares, n = 7) groups during normoxia (A) and hypoxia (B). Data are means ± SD. * Significantly different from rest, P < 0.05. dagger  Significantly different from control, P < 0.05.


                              
View this table:
[in this window]
[in a new window]
  		Table 2.   Metabolic and inert gas data at rest and during exercise in control and experimental subjects while breathing air


                              
View this table:
[in this window]
[in a new window]
  		Table 3.   Metabolic and inert gas data at rest and during exercise in control and experimental subjects while breathing hypoxic gas (FIO2 = 0.132)

VA/Q inequality. During both normoxic and hypoxic exercise and recovery (data not shown), there were no significant changes from rest in logSDQ or logSDV for either group, nor were there any differences between the two groups (Tables 2 and 3). The predicted values for A-aDO2 (Fig. 1) reflect these findings in both groups for both inspired gas mixtures. Although there were no significant changes in VA/Q inequality during exercise, the degree of inequality present accounted for 30% of the observed A-aDO2 in the E group and 35% in the C group during heavy exercise while breathing air.

The independently derived measures of VA/Q inequality during both normoxia and hypoxia are shown in Fig. 2. Compared with rest, there were no significant differences in any dispersion index during either FIO2 for both the C and E groups throughout all exercise workloads. Additionally, there were no differences in these indexes between groups at any exercise level. Intrapulmonary shunt was not detected in either group during rest, exercise, or recovery for both inspired gas mixtures (data not shown).


View larger version (20K):
[in this window]
[in a new window]
  		Fig. 2.   Mean ventilation-perfusion (VA/Q) inequality dispersion (Disp) retention (R) and excretion (E) indexes for control (n = 5) and experimental (n = 7) groups during normoxia (A) and hypoxia (B). , Control DispR; black-triangle, control DispE; , control DispR-E; , experimental DispR; triangle , experimental DispE; open circle , experimental DispR-E. There was no effect of exercise under any condition. There were no differences between subject groups under any condition. SD bars were omitted for clarity.

Observed-predicted A-aDO2 [A-aDO2 (o-p)]. During moderate and heavy exercise while breathing air, A-aDO2 (o-p) in both the C and E groups was significantly greater than that measured at rest (Table 2), and at these workloads was larger by 10 Torr in the E than the C group (P < 0.01). During hypoxia, the A-aDO2 (o-p) in both the E and C groups was significantly higher than at rest at all exercise workloads, but there were no significant differences between the two groups at any exercise workload (Table 3).

DLO2. The estimation of DLO2 is based on the degree to which measured PaO2 is less than that predicted from VA/Q inequality alone. At rest and during light exercise, and also for one control subject during moderate and heavy exercise, this requirement was not met; consequently, DLO2 could not be calculated at these times. On the basis of the data of the remaining 11 subjects for moderate and heavy exercise during normoxia and hypoxia, DLO2 was not different between the E and C groups despite the E group having consistently lower values at all time points (Tables 2 and 3).

Q. Q increased progressively with exercise intensity during normoxia and hypoxia in both the C and E groups (Tables 2 and 3). There were no significant differences between the groups at any workload during either FIO2.

DLO2-to-Q ratio (DLO2/Q). To investigate further the differences between the two groups with respect to O2 diffusion limitation, we used the ratio of DLO2 to pulmonary perfusion conductance (i.e., Q). Table 2 presents the DLO2/Q for the C and E group while breathing air. During both moderate and heavy exercise, the E group had significantly lower values for DLO2/Q than did the C group. Although during hypoxia the E group when compared with the C group continued to display lower DLO2/Q values at moderate and heavy exercise, the values between the groups were not significantly different from each other (Table 3).

PaO2. Because of the anticipation of the onset of exercise, two subjects from the C group hyperventilated during resting measurements and thus spuriously elevated PaO2 and reduced PaCO2. For this reason, no comparisons between the groups were made at rest with respect to PaO2 and PaCO2 while breathing air.

Compared with PaO2 during light exercise, moderate and heavy exercise invoked significant arterial hypoxemia in the E group, whereas the C group demonstrated no significant change (Table 2). As such, the E group had a significantly lower PaO2 (~10 Torr) than did the C group during both moderate and heavy exercise. During hypoxia, both groups had a significantly lower PaO2 during all exercise intensities compared with rest (Table 3), but there were no significant differences between the groups.

Ventilation and PaCO2. There were no significant differences between the E and C groups for VE (l/min, BTPS) during either normoxia or hypoxia (Tables 2 and 3). When VE was calculated relative to body weight, VE (l · min-1 · kg-1, BTPS) during normoxic heavy exercise was higher for the C group (2.12 ± 0.33 l · min-1 · kg-1) than for the E group (1.91 ± 0.27) l · min-1 · kg-1, but the means were not significantly different (P = 0.25). During hypoxic heavy exercise the corresponding values were 2.20 ± 0.37 and 1.89 ± 0.39 l · min-1 · kg-1 for the C group and the E group, respectively (P = 0.20). Normoxic exercise resulted in no significant change in PaCO2 for either group, and although PaCO2 for the C group was systematically lower at all exercise workloads, there were no significant differences between the groups (Table 2). However, hypoxia resulted in a significant decline in PaCO2 during heavy exercise in both the E and C groups. There were no significant differences between the two groups during any workload (Table 3).

Analysis of All Subjects by Linear Regression

Prediction of PaO2. When data for all 12 subjects were examined by linear regression analysis, PaCO2 was significantly associated with, and explained over 45% of, the variance in PaO2 (P < 0.05, Fig. 3A) during normoxic heavy exercise. During this workload, neither DLO2 nor VA/Q inequality (represented by logSDQ) were individually significantly associated with PaO2 (r2 = 0.12 and 0.26, respectively). However, by including PaCO2, DLO2, and logSDQ in a stepwise multiple linear regression model, 90% of the variance in PaO2 was explained by the equation -1.36 · PaCO2 + 0.33 · DLO2 - 73.98 · logSDQ + 142.05 (r = 0.95, P < 0.001, Fig. 3B). The corresponding beta coefficients were 0.64, 0.65, and 0.64 for PaCO2, DLO2, and logSDQ, respectively. On the basis of the results of the r2 and the beta coefficients, it was estimated that, on average, DLO2 accounted for 30.2% of the variance in PaO2 (0.90 · 0.65 · 100/0.64 + 0.65 + 0.64), logSDQ for 29.8%, and PaCO2 for 29.7%.



View larger version (29K):
[in this window]
[in a new window]
  		Fig. 3.   A: association of arterial PO2 (PaO2) with arterial PCO2 (PaCO2) with performance of heavy exercise under normoxic conditions. , Experimental subjects (n = 7); open circle , control subjects (n = 5). B: association between measured and predicted PaO2. , Experimental subjects (n = 7); open circle , control subjects (n = 4); dashed line, line of identity. Predicted PaO2 was calculated by equation -1.36 · PaCO2 + 0.33 · lung diffusing capacity for O2 (DLO2) - 73.98 · log SD for cardiac output (logSDQ) + 142.05 in 11 subjects while they performed heavy exercise under normoxic conditions.

Recovery. There were no significant changes from preexercise in logSDQ or logSDV for either group during recovery (data not shown). PaO2 and A-aDO2 (o-p) remained at or near resting levels from 5 to 45 min postexercise under normoxic conditions (Fig. 4). A-aDO2 (o-p) during recovery from normoxic exercise was not significantly greater than zero, indicating an absence of pulmonary diffusion limitation or shunt during this time. After exercise, DLCO decreased significantly within each group compared with preexercise values (Fig. 5), but there were no significant differences between groups either pre- or postexercise. When the data of both subject groups were pooled, there were no significant changes in pulmonary function from pre- to postexercise (FEV1.0: preexercise, 4.77 ± 0.68, postexercise, 4.85 ± 0.64 liters; FVC: preexercise, 6.01 ± 0.56, postexercise, 6.07 ± 0.63 liters).


View larger version (15K):
[in this window]
[in a new window]
  		Fig. 4.   Normoxic PaO2 and observed minus predicted A-aDO2 [A-aDO2 (o-p)] in control (n = 5) and experimental (n = 7) groups preexercise and up to 45 min postexercise. , PaO2 (experimental); , PaO2 (control); , A-aDO2 (o-p) (experimental); open circle , A-aDO2 (o-p) (control). There were no significant differences over time or between 2 groups.



View larger version (12K):
[in this window]
[in a new window]
  		Fig. 5.   Lung diffusing capacity for CO (DLCO) preexercise (open bars) and postexercise (solid bars). * Significantly different from preexercise, P < 0.05.


    DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES

The principal findings of this investigation are threefold: 1) during heavy exercise, the subjects with exercise-induced hypoxemia developed significantly more O2 diffusion limitation [measured by A-aDO2 (o-p)] than control subjects matched for age, lung function, and VO2 peak; 2) no subject developed significant increases from rest in VA/Q inequality during exercise while breathing either air or 13.2% O2; and 3) the majority (90%) of the variance in PaO2 during normoxic heavy exercise could be explained with a stepwise multiple linear regression model that combined the predictor diffusing capacity (DLO2), VA/Q inequality (logSDQ), and ventila-tion (PaCO2). Within this regression model the relative contribution of the predictor variables was 30.2% for DLO2, 29.8% for logSDQ, and 29.7% for PaCO2. The quantitative contribution of each of these three terms to PaO2 may confounded by a pseudocorrelation due to the calculation of DLO2 that includes both the PaO2 itself and inert gas data related to the logSDQ. However, the relatively narrow range of the shared variables most likely makes this a minor effect. In addition, DLO2 and logSDQ were not significantly correlated (r2 = 0.14), further suggesting that in the present study the risk of pseudocorrelation is small. This result suggests that in this group of athletes, arterial oxygenation may have a multifactorial rather than a single cause. However, because of the small number of subjects studied, the generalizability of the present study may be limited.

O2 Diffusion Limitation

Previously, it has been reported that substantial O2 diffusion limitation is evident during intense exercise, and this becomes increasingly important as VO2 increases (11). In the present study, the E and C groups were matched for VO2 during both normoxic and hypoxic exercise, which suggests that the significant differences between the two groups for A-aDO2 (o-p) during moderate and heavy exercise is likely to be an inherent difference in pulmonary gas exchange between the two groups of subjects.

Reduced red blood cell transit time in the pulmonary circulation is a possible explanation for the observed O2 diffusion limitation in the present study. Previously, red blood cell transit times have been shown to decrease with exercise (17, 40), and whole lung transit time has been significantly associated with A-aDO2 (o-p) (17). However, unequivocal evidence to show that transit times reach the minimum time for O2 equilibration of 0.35-0.40 s (6, 10) is not available. In the present study there were no significant differences between the C and E groups with respect to Q. Consequently, we can only speculate that, if transit time is the major cause of greater O2 diffusion limitation in the E subjects, then this group must have either or both of the following: 1) a smaller pulmonary capillary blood volume and/or 2) less recruitment of pulmonary capillaries at identical exercise intensities. The two groups of subjects in the present study were well matched for pulmonary function and lung size. As such, there is no obvious reason to suspect that there would be a significant difference in the pulmonary capillary structure of the two groups; however, this cannot be discounted. We cannot provide direct evidence of O2 diffusion limitation due to excessively low pulmonary capillary transit times, but there is indirect evidence to suggest a link between the two. First, when DLO2 was included as a variable in a multiple linear regression model, it accounted for ~30% of the variance in PaO2 during heavy exercise while breathing air. Second, the development of an alveolar-end capillary gradient in a single-compartment lung model can be shown to depend on the relative DLO2/Q (26). When the ratio is >3, VO2 is perfusion limited, as would be the case in normoxia at rest. As DLO2/Q falls below three, O2 diffusion limitation becomes evident. In the present study, during normoxic heavy exercise both the C and E groups had DLO2/Q values <3, suggesting O2 diffusion limitation in each. More importantly, however, the E group during normoxic moderate and heavy exercise had significantly lower DLO2/Q values than the matched C group, indicating significantly more O2 diffusion limitation in the former.

Another possible mechanism for arterial hypoxemia in the E subjects compared with the C subjects during heavy exercise is a significantly reduced mixed venous PO2 (35). Calculated mixed venous PO2 during heavy exercise while breathing air was not significantly different between groups [24 ± 4 (C) vs. 26 ± 6 Torr (E), P = 0.29] and therefore cannot explain the greater arterial hypoxemia in the E group.

Recently, St. Croix et al. (33) provided evidence to suggest that EIH was a result of a functionally based mechanism and not caused by a mechanism that persisted after exercise. These results support the contention of reduced red blood cell transit times as a plausible explanation for EIH, but additional direct evidence is required to confirm or refute the link between transit times and O2 diffusion limitation during high-intensity exercise.

Extrapulmonary Shunt

The MIGET technique is unable to distinguish the relative contributions that pulmonary capillary diffusion limitation for O2 and shunt from bronchial and thebesian veins (extrapulmonary) make toward the overall A-aDO2 (o-p). Therefore, the potential effect this shunt may have on EIH is worth exploring. During normoxic heavy exercise, the fall in PaO2 from rest measured in the C and E subjects can be explained entirely by a 1.5 ± 0.5 and 3.0 ± 1.4% extrapulmonary shunt, respectively. Shunts of this size are at the limits of expectation in healthy subjects, and previous studies have reported extrapulmonary shunt values in the range of 0.18-2% of total Q (11, 34). Extrapulmonary shunt can be measured by having the subject breathe 100% O2. We chose not to use this method in the present study because of the difficulty in measuring PaO2 accurately when breathing 100% O2. Indeed, the absolute O2 content differences between air and 100% O2 breathing are in the range of the experimental error (11). Instead, we used hypoxia (FIO2 = 0.13) to demonstrate that extrapulmonary shunt could not be the sole explanation for the hypoxemia measured during normoxic heavy exercise. Because of the steepness of the O2 dissociation curve at an FIO2 of 0.13, a 1-3% extrapulmonary shunt would decrease measured PaO2 by only 2-3 Torr. In contrast, the hypoxemia we measured during hypoxic heavy exercise in the present study would require an extrapulmonary shunt in the region of 17% for the C subjects and 24% for the E subjects, an enormous increase for these healthy normal subjects. Therefore, we conclude that pulmonary capillary diffusion limitation for O2 occurred in the C and E subjects during heavy exercise and that extrapulmonary shunt comprises an extremely small component of the overall A-aDO2 (o-p).

Ventilation

Inadequate hyperventilation during high-intensity exercise has been proposed as a significant contributor to EIH by many authors (5, 9, 16, 24). In the present study, when all subjects (n = 12) were considered by regression analysis, there was a significant association between PaO2 and PaCO2 during heavy exercise under normoxic conditions. This result suggests that those subjects with the most severe arterial hypoxemia were most likely to have the lowest hyperventilatory response to exercise. A number of possible mechanisms have been proposed for inadequate hyperventilation during heavy exercise: 1) a decreased peripheral chemoreceptor function (4, 16); 2) respiratory muscle fatigue (3); and 3) mechanical constraints imposed on inspiratory and expiratory flow (21). Regardless of the mechanism, the level of ventilation during heavy exercise while breathing air in the present study explained up to 47% of the variance in arterial oxygenation, and further investigation into the possible mechanisms is warranted.

VA/Q Inequality

Contrary to previous studies (11, 12, 20, 34) we found normal levels of VA/Q inequality at rest with no significant increase during exercise while breathing either normoxic or hypoxic gas mixtures. The technical quality of the inert gas data was excellent, as evidenced by a low residual sum of squares under all conditions (Table 2). Consequently, we believe this to be a characteristic of this particular population. A recent study demonstrated that lung size (expressed relative to body surface area) is an important determinant in the efficiency of VA/Q matching, in effect suggesting that those athletes with large lungs are less likely to develop significant VA/Q inequality during exercise (18). This hypothesis is supported by the present study as all subjects possessed FVCs and alveolar volumes greater than those predicted according to the subject's age, height, and race (107 ± 10 and 166 ± 16%, respectively). Furthermore, there were no significant differences between the C and E groups in any pulmonary function test, which may explain the lack of difference between the two groups in VA/Q inequality during exercise. The reasons why hypoxia did not induce significant VA/Q inequality in either group remains unclear but may reflect, additionally, that these subjects were generally resistant to the development of exercise-induced VA/Q inequality.

Hypoxia

As expected, hypoxia resulted in significantly lower VO2 peak values in both the C and E subjects (decrease from normoxia: C, 22.0 ± 6.6%; E, 26.6 ± 3.9%, P = 0.14, not significant). The larger, even if nonsignificant, reduction measured in the E group probably represents their position as a group on the oxyhemoglobin curve as this has been shown to directly affect the degree of decrease in VO2 peak (29). This result has implications for those athletes who wish to pursue competition in endurance events at altitude. It is most likely, with all other factors being equal, that subjects who develop EIH during sea-level exercise will have a greater reduction in performance on ascent to altitude than will those athletes who retain PaO2 near resting levels.

Recovery

DLCO postexercise was reduced significantly in all C and E subjects, but VA/Q relationships were unchanged from preexercise values. This has been reported a number of times previously (14, 23) although the reasons behind this reduction are not fully understood. The mechanisms proposed in the literature for a reduction in postexercise DLCO are 1) a transient change in the structure of the alveolocapillary membrane, thereby affecting the O2 diffusion (23); and 2) a decrease in the pulmonary capillary blood volume (14). In the present study we measured A-aDO2 (o-p) and PaO2 up to 45 min postexercise, and there were no significant changes in either variable from rest, suggesting no change in the alveolocapillary structure of sufficient magnitude to affect O2 diffusion. Second, previous studies have demonstrated that lung volume and function are temporarily impaired after exercise (2, 25), suggesting small-airway closure and possible subclinical edema, all contributing to a decreased diffusing capacity. In the present study, we found no significant change in pulmonary function postexercise; in fact, there was a slight trend toward an improvement in both lung volume and airflow rates postexercise. This does not support the contention that edema was present in either the C or E groups during either normoxic or hypoxic exercise, and thus edema is an unlikely explanation for the significant fall in DLCO in this subject population. On the basis of this evidence, we believe that the most likely cause for the significant postexercise decrease in DLCO in both the C and E groups is a reduction in the pulmonary capillary blood volume.

In summary, we have shown that highly trained cyclists, who have previously demonstrated significant exercise-induced hypoxemia during intense exercise, developed a significantly larger A-aDO2 (o-p) than did a control group matched for age, lung function, and VO2 peak. This result has been interpreted to primarily represent differences in O2 diffusion limitation between the two subject groups. We have also demonstrated that DLO2, ventilation, and VA/Q inequality each contribute to the level of arterial oxygenation in any subject and together explain the majority of the variance in PaO2. Therefore, the results from the present study suggest that, for highly trained cyclists, exercise-induced hypoxemia has a multifactorial etiology related to O2 diffusion limitation, inadequate hyperventilation, and VA/Q inequality.


    ACKNOWLEDGEMENTS

The authors thank the subjects for participation in the study.


    FOOTNOTES

This project was supported by a Special Purposes Grant from the Royal Adelaide Hospital. S. R. Hopkins and P. D. Wagner were supported by National Institutes of Health Grants MO1RR-0287 and HL-17731, respectively.

The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Address for reprint requests and other correspondence: A. Rice, C/-Chest Clinic, Royal Adelaide Hospital, 275 North Terrace, Adelaide, South Australia 5000, Australia (E-mail: arice{at}earthling.net).

Received 22 January 1999; accepted in final form 2 August 1999.


    REFERENCES
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES

1.   Barker, R., S. R. Hopkins, N. Kellogg, I. M. Olfert, T. Brutsaert, T. P. Gavin, P. Entin, A. J. Rice, and P. D. Wagner. Measurement of cardiac output during exercise by open-circuit acetylene uptake. J. Appl. Physiol. 87: 1506-1512, 1999[Abstract/Free Full Text].

2.   Buono, M. J., S. H. Constable, A. R. Morton, T. C. Rotkis, P. R. Stanforth, and J. H. Wilmore. The effect of an acute bout of exercise on selected pulmonary function measurements. Med. Sci. Sports Exerc. 13: 290-293, 1981[Medline].

3.   Bye, P. T., G. A. Farkas, and C. Roussos. Respiratory factors limiting exercise. Annu. Rev. Physiol. 45: 439-451, 1983[Medline].

4.   Byrne-Quinn, E., J. V. Weil, I. E. Sodal, G. F. Filley, and R. F. Grover. Ventilatory control and the athlete. J. Appl. Physiol. 30: 91-98, 1971[Free Full Text].

5.   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[Abstract/Free Full Text].

6.   Dempsey, J. A., D. Hanson, D. Pegelow, A. Claremont, and J. Rankin. Limitations to exercise capacity and endurance: pulmonary system. Can. J. Appl. Spt. Sci. 7: 4-13, 1982.

7.   Evans, J. W., and P. D. Wagner. Limits on VA/Q distributions from analysis of experimental inert gas elimination. J. Appl. Physiol. 42: 889-898, 1977[Abstract/Free Full Text].

8.   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. 58: 978-988, 1985[Abstract/Free Full Text].

9.   Gavin, T. P., A. Derchak, and J. M. Stager. Ventilation's role in the decline in VO2 max and SaO2 in acute hypoxic exercise. Med. Sci. Sports Exerc. 30: 195-199, 1998[Medline].

10.   Gledhill, N., A. B. Froese, and J. A. Dempsey. Ventilation to perfusion distribution during exercise in health. In: Muscular Exercise and the Lung, edited by J. A. Dempsey, and C. E. Reed. Madison, WI: Univ. of Wisconsin Press, 1977, p. 325-343.

11.   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. 60: 1590-1598, 1986[Abstract/Free Full Text].

12.   Hammond, M. D., G. E. Gale, K. S. Kapitan, A. Ries, and P. D. Wagner. Pulmonary gas exchange in humans during normobaric hypoxic exercise. J. Appl. Physiol. 61: 1749-1757, 1986[Abstract/Free Full Text].

13.   Hammond, M. D., and S. C. Hempleman. Oxygen diffusing capacity estimates derived from measured VA/Q distributions in man. Respir. Physiol. 69: 129-147, 1987[Medline].

14.   Hanel, B., I. Teunissen, A. Rabol, J. Warberg, and N. H. Secher. Restricted postexercise pulmonary diffusion capacity and central blood volume depletion. J. Appl. Physiol. 83: 11-17, 1997[Abstract/Free Full Text].

15.   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.) 507: 619-628, 1998[Abstract/Free Full Text].

16.   Harms, C. A., and J. M. Stager. Low chemoresponsiveness and inadequate hyperventilation contribute to exercise-induced hypoxemia. J. Appl. Physiol. 79: 575-580, 1995[Abstract/Free Full Text].

17.   Hopkins, S. R., A. S. Belzberg, B. R. Wiggs, and D. C. McKenzie. Pulmonary transit time and diffusion limitation during heavy exercise in athletes. Respir. Physiol. 103: 67-73, 1996[Medline].

18.   Hopkins, S. R., T. P. Gavin, N. M. Siafakas, L. J. Haseler, I. M. Olfert, H. Wagner, and P. D. Wagner. Effect of prolonged heavy exercise on pulmonary gas exchange in athletes. J. Appl. Physiol. 85: 1523-1533, 1998[Abstract/Free Full Text].

19.   Hopkins, S. R., and D. C. McKenzie. Hypoxic ventilatory response and arterial desaturation during heavy work. J. Appl. Physiol. 67: 1119-1124, 1989[Abstract/Free Full Text].

20.   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. 77: 912-917, 1994[Abstract/Free Full Text].

21.   Johnson, B. D., K. W. Saupe, and J. A. Dempsey. Mechanical constraints on exercise hyperpnea in endurance athletes. J. Appl. Physiol. 73: 874-886, 1992[Abstract/Free Full Text].

22.   Kelman, G. R. Digital computer subroutine for the conversion of oxygen tension into saturation. J. Appl. Physiol. 21: 1375-1376, 1966[Free Full Text].

23.   Manier, G., J. Moinard, P. Techoueyres, N. Varene, and H. Guenard. Pulmonary diffusion limitation after prolonged strenuous exercise. Respir. Physiol. 83: 143-154, 1991[Medline].

24.   Miyachi, M., and I. Tabata. Relationship between arterial oxygen desaturation and ventilation during maximal exercise. J. Appl. Physiol. 73: 2588-2591, 1992[Abstract/Free Full Text].

25.   O'Kroy, J. A., R. A. Loy, and J. R. Coast. Pulmonary function changes following exercise. Med. Sci. Sports Exerc. 24: 1359-1364, 1992[Medline].

26.   Piiper, J., and P. Scheid. Blood-gas equilibration in lungs. In: Pulmonary Gas Exchange, edited by J. B. West. New York: Academic, 1980, p. 132-173.

27.   Powers, S. K., S. Dodd, J. Lawler, G. Landry, M. Kirtley, T. McKnight, and S. Grinton. Incidence of exercise induced hypoxemia in elite endurance athletes at sea level. Eur. J. Appl. Physiol. 58: 298-302, 1988.

28.   Powers, S. K., S. Dodd, J. Woodyard, R. E. Beadle, and G. Church. Haemoglobin saturation during incremental arm and leg exercise. Br. J. Sports Med. 18: 212-216, 1984[Abstract].

29.   Powers, S. K., J. Lawler, J. A. Dempsey, S. Dodd, and G. Landry. Effects of incomplete pulmonary gas exchange on VO2 max. J. Appl. Physiol. 66: 2491-2495, 1989[Abstract/Free Full Text].

30.   Rice, A. J., G. C. Scroop, C. J. Gore, A. T. Thornton, M. J. Chapman, H. W. Greville, M. D. Holmes, and R. Scicchitano. Exercise-induced hypoxaemia in highly trained cyclists at 40% peak oxygen uptake. Eur. J. Appl. Physiol. 79: 353-359, 1999.

31.   Schaffartzik, W., D. C. Poole, T. Derion, K. Tsukimoto, M. C. Hogan, J. P. Arcos, D. E. Bebout, and P. D. Wagner. VA/Q distribution during heavy exercise and recovery in humans: implications for pulmonary edema. J. Appl. Physiol. 72: 1657-1667, 1992[Abstract/Free Full Text].

32.   Severinghaus, J. W. Simple, accurate equations for human blood O2 dissociation computations. J. Appl. Physiol. 46: 599-602, 1979[Abstract/Free Full Text].

33.   St. Croix, C. M., C. A. Harms, S. R. McClaran, G. A. Nickele, D. Pegelow, W. B. Nelson, and J. A. Dempsey. Effects of prior exercise on exercise-induced arterial hypoxemia in young women. J. Appl. Physiol. 85: 1556-1563, 1998[Abstract/Free Full Text].

34.   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. 58: 989-995, 1985[Abstract/Free Full Text].

35.   Wagner, P. D. Influence of mixed venous PO2 on diffusion of O2 across the pulmonary blood: gas barrier. Clin. Physiol. 2: 105-115, 1982[Medline].

36.   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. Appl. Physiol. 61: 260-270, 1986[Abstract/Free Full Text].

37.   Wagner, P. D., P. F. Naumann, and R. B. Laravuso. Simultaneous measurement of eight foreign gases in blood by gas chromatography. J. Appl. Physiol. 36: 600-605, 1974[Free Full Text].

38.   Wagner, P. D., H. A. Saltzman, and J. B. West. Measurement of continuous distributions of ventilation-perfusion ratios: theory. J. Appl. Physiol. 36: 588-599, 1974[Free Full Text].

39.   Wagner, P. D., J. R. Sutton, J. T. Reeves, A. Cymerman, B. M. Groves, and M. K. Malconian. Operation Everest II: pulmonary gas exchange during a simulated ascent of Mt. Everest. J. Appl. Physiol. 63: 2348-2359, 1987.[Abstract/Free Full Text]