Predicting diffusive alveolar oxygen transfer from carbon monoxide-diffusing capacity in exercising foxhounds

Abstract

Although lung diffusing capacity for carbon monoxide (Dl_CO) is a widely used test of diffusive O₂ transfer, few studies have directly related Dl_CO to O₂-diffusing capacity (Dl_O2); none has used the components of Dl_CO, i.e., conductance of alveolar membrane and capillary blood, to predict Dl_O2 from rest to exercise. To understand the relationship between Dl_CO and Dl_O2 at matched levels of cardiac output, we analyzed cumulative data from rest to heavy exercise in 43 adult dogs, with normal lungs or reduced lung capacity following lung resection, that were studied by two techniques. 1) A rebreathing (RB) technique was used to measure Dl_CO and pulmonary blood flow at two O₂ tensions, independent of O₂ exchange. Dl_CO was partitioned into CO-diffusing capacity of alveolar membrane and pulmonary capillary blood volume using the Roughton-Forster equation and converted into an equivalent Dl_O2, [Dl_O2(RB)]. 2) A multiple inert-gas elimination technique (MIGET) was used to measure ventilation-perfusion distributions, O₂ and CO₂ exchange under hypoxia, to derive Dl_O2 [Dl_O2(MIGET)] by the Lilienthal-Riley technique and Bohr integration. For direct comparisons, Dl_O2(RB) was interpolated to the cardiac output measured by the Fick principle corresponding to Dl_O2(MIGET). The Dl_O2-to-Dl_CO ratio averaged 1.61. Correlation between Dl_O2(RB) and Dl_O2(MIGET) was similar in normal and post-resection groups. Overall, Dl_O2(MIGET) = 0.975 Dl_O2(RB); mean difference between the two techniques was under 5% for both animal groups. We conclude that, despite various uncertainties inherent in these two disparate methods, the Roughton-Forster equation adequately predicts diffusive O₂ transfer from rest to heavy exercise in canines with normal, as well as reduced, lung capacities.

the impetus for measuring lung diffusing capacity arose because two famous investigators (Christian Bohr and John Scott Haldane) reported in the 1890's using independent methods (2, 8) that O₂ was actively secreted into pulmonary capillary blood, so that O₂ tension of blood leaving the lung was higher than that in alveolar air. August and Marie Krogh (26, 27) were skeptical of the secretion theory, owing to uncertainties in the techniques. Krogh and Bohr recognized a need to measure diffusing capacity using a gas other than O₂ to determine whether passive diffusion could support the levels of O₂ uptake during exercise. Haldane was the first to measure carbon monoxide (CO) uptake at rest on himself (7). Based on Haldane's measurements, Bohr (1) published the first values of CO-diffusing capacity (Dl_CO) and translated Haldane's Dl_CO into O₂-diffusing capacity (Dl_O2), taking into account the relative molecular weight (MW) and solubility (α, in ml·ml⁻¹·atm⁻¹ at 37.5°C) of the two gases. The generally used conversion factor is Dl_O2/Dl_CO = /·α_O2/α_CO = /·0.0239/0.0182 = 1.23 (range 1.1–1.4), where MW_CO and MW_O2 are the molecular weight of CO and O₂, and α_O2 and α_CO are the solubility of O₂ and CO, respectively (24, 30, 42). Bohr also derived the famous Bohr integral to interpret the data but erroneously concluded that O₂ secretion was required to raise O₂ uptake above 1.5 l/min. Marie Krogh, using a breath-holding method, measured Dl_CO at rest and exercise (27) and showed that Dl_O2 rose to two to four times that measured at rest by Bohr, sufficient to support O₂ uptake during exercise, even at high altitude, without invoking O₂ secretion. Thus diffusion was established as the process of O₂ transfer within pulmonary alveoli.

Marie Krogh (27) in 1915 first proposed the use of Dl_CO to infer physiological limits of diffusive O₂ transfer, assuming Dl_O2 ≈ 1.23·Dl_CO, which took into account the relative tissue diffusivities for O₂ and CO, but not the relative conductance of capillary blood (Θ_O2 and Θ_CO, respectively, in ml gas·min⁻¹·ml blood⁻¹·mmHg⁻¹). In the 1950's, Roughton and Forster (34) partitioned Dl_CO into separate resistances imposed by the alveolar membrane (Dm_CO) and capillary blood (Θ_CO·Vc), i.e.,

(1)

where Θ_CO (ml CO·min⁻¹·ml blood⁻¹·mmHg⁻¹) is the empirical CO uptake by whole blood measured in vitro, and Vc is pulmonary capillary blood volume (ml). Half a century later, the true values of Θ_O2 and Θ_CO remain controversial, and no study has formally established whether Dm_CO and Vc accurately predict diffusive O₂ transfer. Resistance of the blood for CO uptake is thought to be greater than that for O₂ uptake. There have been concerns that true Dm_CO and Vc are sensitive to inspired O₂ tension (18), thereby introducing errors when inferring O₂ transfer from CO uptake measured at different alveolar O₂ tensions. As stated by Hughes and Bates in a recent review (24): “…there is still uncertainty over the relative importance of the alveolar capillary membrane versus the red cells as rate-limiting steps in the overall transfer of carbon monoxide from gas to blood.”

In view of the above uncertainties, we tested the assumption that Dm_CO and Vc in the Roughton-Forster equation accurately predict Dl_O2 from rest to heavy exercise. In separate cohorts of dogs with normal or reduced lung capacities, we compared the relationships between Dl_O2 and cardiac output derived using two independent methods. 1) A rebreathing (RB) technique was used to simultaneously measure Dl_CO and pulmonary blood flow from the uptake of inhaled tracer gases, CO and acetylene, respectively, independent of O₂ and CO₂ exchange. Measurements were made at two different O₂ tensions to obtain Dm_CO and Vc (Eq. 1), which were converted into an equivalent Dl_O2 [Dl_O2(RB)] using the tissue diffusivity factor (1.23) and published values of Θ_O2 and Θ_CO for dog blood. 2) A multiple inert-gas elimination technique (MIGET) was used to estimate Dl_O2 [Dl_O2(MIGET)] from invasive measurement of ventilation-perfusion (V̇a/Q̇) distributions, and O₂ and CO₂ exchange during hypoxic exercise, utilizing the Lilienthal-Riley technique and Bohr integration. Whole lung Dl_O2(MIGET) represents the value that would be necessary to explain the part of the measured alveolar-arterial O₂ tension gradient (A-aDo₂) unexplained by V̇a/Q̇ mismatch (10). For a direct comparison of the two Dl_O2 estimates, Dl_O2(RB) was interpolated to the same cardiac output as that measured using the Fick principle, in conjunction with Dl_O2(MIGET) in the same animal. We reasoned that correspondence between two independent estimates of Dl_O2 at the same cardiac output supports the use of Dm_CO and Vc to predict the limits of diffusive O₂ transfer during exercise.

METHODS

Animals.

The Institutional Animal Care and Use Committee approved all protocols and procedures. A total of 43 adult male foxhounds (2.5–5 yr of age) were studied in separate projects. Data from some of these animals have been published (6, 14, 15, 20–22, 29, 43). Twenty-six animals (body weight 26.4 ± 3.8 kg) had intact lungs. Another 17 animals (26.2 ± 5.1 kg) were studied 4–24 mo post-lung resection. Of the latter, 11 underwent right pneumonectomy (removal of 58% of lung units). Of these, nine animals had a custom-made silicone space-filling prosthesis implanted into the empty hemithorax at the time of pneumonectomy. The prosthesis was kept inflated via a subcutaneous injection port in four animals and remained uninflated in five animals; the remaining two animals had undergone right pneumonectomy without implantation of prosthesis. Six other animals underwent bilateral surgical resection of 70% of the lung units in two stages. In all animals, bilateral carotid arteries were exteriorized into dermal loops to permit acute catheterization without risking the complications of indwelling vascular catheters. Exercise measurements were obtained at least 4 mo after recovery from surgical procedures when adaptation to lung resection has stabilized. The RB and MIGET studies could not be performed together and were separated by an average of 4 mo. In most animals, RB measurements were performed before and after studies by MIGET.

Exercise training and apparatus.

Detailed description of these methods has been published elsewhere (19, 38). Briefly, animals were trained to run on a motorized treadmill wearing a sealed respiratory mask for measuring ventilation and gas exchange. Regular physical training continued throughout the studies. A built-in plastic cylindrical breathing orifice connects to an automatic pneumatic balloon three-way valve attached to a one-way, low-resistance valve and to a RB bag. Inspired and expired ventilation were measured using separate screen pneumotachographs. Expired gas concentrations are sampled continuously by a mass spectrometer distal to a mixing chamber. Minute ventilation, O₂ uptake, CO₂ output, and respiratory rate were followed breath by breath. Heart rate and rectal temperature were continuously recorded. All signals are digitized at 50 or 100 Hz and averaged every 10 breaths.

RB measurements.

The technique has been described extensively (6, 20, 29). Based on individual performance, measurements were obtained at three to five workloads of differing intensities. After warm-up at 6 mph and 0% grade for 3–5 min, the treadmill workload was increased to a preselected level. The RB bag was automatically prefilled with a volume of gas equal to the animal's average tidal volume plus 200 ml (atpd). The RB gas mixture contained 0.6% acetylene, 0.3% C¹⁸O, 9% He, and 30% O₂ in a balance of either N₂ or O₂. After 3 min of exercise at each workload, the pneumatic valve was automatically switched at end expiration to allow the animal to inspire one bolus of the test gas mixture and then rebreathe this mixture in and out of an anesthetic bag for 6–12 s, while gas concentrations at the mouth were monitored. Complete (>90%) mixing was accomplished within three breaths in all animals. For RB maneuvers using the mixture containing 90% O₂, the animal inspired 100% O₂ from a large reservoir for 1 min before the maneuver to increase alveolar O₂ tension to that in the test gas mixture.

System volume was measured from helium dilution, pulmonary blood flow from the slope of the exponential acetylene disappearance, and Dl_CO from the slope of the exponential C¹⁸O disappearance (3, 4). From Dl_CO measured at two different O₂ tensions, Dm_CO and Vc were calculated using Eq. 1 and the value of Θ_CO for dog blood given by Holland (12), corrected by the Arrhenius equation to a rectal temperature of 39°C, i.e.,

(2)

where Pa_O2 is alveolar Po₂. The equivalent Dm_O2 and Dl_O2 were calculated from Dm_CO and Vc using the value of θ_O2 = 3.9 ml O₂·min⁻¹·mmHg⁻¹·ml blood⁻¹ given by Yamaguchi et al. (44), and the tissue diffusivity between O₂ (Dm_O2) and CO based on relative MWs and solubility, i.e., Dm_O2 = 1.23 Dm_CO, giving:

(3)

One can reasonably assume this θ_O2 value when alveolar oxyhemoglobin saturation falls to ∼85%, a level reached in dogs at heavy exercise while breathing room air (25).

MIGET.

This technique has also been described extensively (14, 15, 17, 21, 22). Briefly, both external jugular veins and one exteriorized carotid artery were cannulated on the day of study. A Swan-Ganz catheter was advanced via one jugular catheter into the pulmonary artery for blood sampling and pressure monitoring. Six inert gases (SF₆, ethane, cyclopropane, enflurane, acetone, and ether) were dissolved in saline and infused via the contralateral jugular vein at a constant rate, adjusted to ensure an adequate signal-to-noise ratio. Animals inspired 14% O₂ from a large reservoir. At rest, the infusion was started 20 min before sampling to ensure equilibrium of inert-gas exchange at the blood-gas interface. During exercise, the infusion began at the onset of exercise. After running 3 min at a given workload when O₂ uptake and ventilation neared a plateau, duplicate arterial and mixed venous blood samples and timed quadruplicate mixed expired gas samples were collected in glass syringes for measuring inert-gas concentrations by gas chromatography. Blood-gas partition coefficients of the inert gases were measured for each animal and corrected for the difference between the animal's body temperature during exercise and the water bath temperature at which the samples were equilibrated. Cardiac output was calculated by the direct Fick method.

From the multicompartmental distribution of V̇a/Q̇ ratios, the log scale second moments of the ventilation (log SDV̇) and perfusion (log SDQ̇) distributions were calculated. From the inert-gas data, barometric pressure, cardiac output, ventilation, hemoglobin, body temperature, base excess, inspired and mixed venous Po₂ and Pco₂, arterial Pco₂, and A-aDo₂ attributable to the observed V̇a/Q̇ mismatch were calculated, assuming complete alveolar-capillary diffusion equilibrium (40). The actual A-aDo₂ was measured from arterial Po₂ (Pa_O2) and the ideal Pa_O2. The difference between A-aDo₂ predicted from V̇a/Q̇ mismatch and the measured A-aDo₂ were assumed to reflect contribution from a combination of alveolar-end capillary diffusion limitation and postpulmonary shunting. Assuming negligible diffusion-perfusion (Dl/Q̇) inhomogeneity, a trial value of total lung Dl_O2 was selected and divided among the V̇a/Q̇ compartments, according to their blood flow. Bohr integration was performed on each V̇a/Q̇ compartment to account for nonlinearity of the O₂ dissociation curve (10). The compartmental Dl_O2 are iterated until the predicted mixed Pa_O2 for the whole lung matches the measured value, at which point the whole lung Dl_O2 estimate represents that which accounts for the difference between measured and predicted A-aDo₂ during hypoxic exercise. This estimate of Dl_O2 represents the minimal value that can account for the portion of A-aDo₂ not explained by V̇a/Q̇ mismatch. Compared with breathing room air, breathing 14% O₂ accentuates A-aDo₂ caused by diffusion impairment and extrapulmonary shunt, while minimizing the A-aDo₂ caused by V̇a/Q̇ mismatch and intrapulmonary shunt (28); differences arise because both O₂ and inert gases are perturbed by V̇a/Q̇ mismatch and intrapulmonary shunt, but only O₂ is diffusion limited and sensitive to extrapulmonary shunt.

Comparison of Dl_O2 estimated by two methods.

To directly compare Dl_O2 estimated by the RB method to that estimated by MIGET, it was necessary to match the cardiac output. For each animal, Dl_CO [Dl_CO(RB)] and Dl_O2(RB) were plotted against their respective simultaneously measured cardiac output, and the slope and intercept of individual regression lines were used to interpolate Dl_CO and Dl_O2 to the same cardiac output reached during hypoxic exercise using MIGET.

RESULTS

Measurements obtained at rest and during heavy exrecise in hypoxia are summarized in Table 1. The relationships between cardiac output, measured from acetylene disappearance during RB, and O₂ uptake were similar to that measured by the direct Fick principle in conjunction with MIGET (Fig. 1), indicating no systematic difference in estimation by the two methods. No significant shunt flow developed in these animals. Using the RB technique, there is a linear relationship between Dl_CO and pulmonary blood flow in animals with intact lungs and in those post-resection (pooled data shown in Fig. 2). Using MIGET and Fick principle, there is a similar linear relationship between Dl_O2 and cardiac output in both groups (pooled data shown in Fig. 3). The slope of the correlation between Dl_O2(MIGET) and Dl_O2(RB) was not significantly different between normal animals with intact lungs [Dl_O2(MIGET) = 1.568 Dl_CO(RB), R² = 0.63] and postresection animals [Dl_O2(MIGET) = 1.656 Dl_CO(RB), R² = 0.49, Fig. 4]. The overall regression line through all data points was Dl_O2(MIGET) = 1.608 Dl_CO(RB), R² = 0.59. Applying Eq. 3, the corresponding RB estimates of Dm_CO and Vc were converted to their equivalent Dl_O2 value [Dl_O2(RB)] and compared with Dl_O2(MIGET) measured at the same cardiac output. The correlations between Dl_O2(RB) and Dl_O2(MIGET) were similar for normal [Dl_O2(MIGET) = 0.944 Dl_O2(RB)] and post-resection [Dl_O2(MIGET) = 0.997 Dl_O2(RB)] groups. The overall correlation for all data points has a slope close to 1.0 [Dl_O2(MIGET) = 0.975 Dl_O2(RB)] (Fig. 5A). The residual vs. predicted values of the correlation are shown in Fig. 5B. The difference between estimates by the two methods averaged 4.5 ± 2.6% and −0.9 ± 2.6% (means ± SE) for normal and post-resection groups, respectively (Fig. 5C).

Table 1.

Multiple inert-gas elimination technique measurements breathing 14% O₂

	Normal		Resection
	Rest	80% Peak Workload	Rest	80% Peak Workload
Ventilation, l·min−1·kg−1	0.588±0.242	4.084±1.587	0.794±0.427*	3.493±0.784
Tidal volume, ml/kg	29.6±7.8	39.5±7.0	21.1±8.1*	33.5±5.2
Respiratory rate, beats/min	22±15	104±26	48±42*	107±28
Oxygen uptake, ml·min−1·kg−1	9±3	61±32	10±3	47±11
CO2 output, ml·min−1·kg−1	8±2	58±18	9±2*	46±12
Heart rate, beats/min	158±24	247±49	159±32	219±73
Mean systemic arterial pressure, mmHg	129±13	148±17	129±6	149±11
Mean pulmonary arterial pressure, mmHg	21±9	37±12	35±7*	61±11
Cardiac output, ml·min−1·kg−1	211±76	529±129	200±73	473±109
Hemoglobin, g/dl	13.9±0.8	15.9±1.6	14.1±0.8	16.0±1.7
Arterial pH	7.39±0.02	7.24±1.02	7.38±0.02*	7.34±0.07*
Arterial Pco2, mmHg	36.0±2.1	33.6±3.7	37.6±3.0*	39.7±6.2
Arterial Po2, mmHg	50.3±4.9	47.4±4.3	54.3±4.1*	40.8±5.0
Arterial O2 saturation, %	80.4±4.3	70.5±10.1	82.4±3.7*	59.2±10.7
Arterial O2 content, ml/dl	15.3±1.2	15.2±2.2	16.4±1.5*	12.9±1.6
O2 delivery, ml·min−1·kg−1	32±11	78±16	32±12	60±13
Mixed venous pH	7.37±0.02	7.31±0.08	7.36±0.02*	7.27±0.08*
Mixed venous Pco2, mmHg	40.1±3.4	47.6±7.6	42.7±3.4*	54.6±10.8
Mixed venous Po2, mmHg	36.0±4.7	21.6±5.6	37.1±3.4	17.6±4.8
Mixed venous O2 saturation, %	58±6	22±11	57±6	15±9
Mixed venous O2 content, ml/dl	11±1	5±3	15±4	16±3
O2 extraction, %	28±7	70±13	30±5	71±11
Log SDV̇	1.0±0.7	1.2±0.3	1.0±0.6	1.2±0.2*
Log SDQ̇	0.4±0.1	0.6±0.1	0.4±0.1	0.6±0.1*
A-aDo2 predicted, mmHg	3.7±2.7	6.5±1.6	4.3±3.7	9.7±5.0
A-aDo2 measured, mmHg	4.0±5.1	12.2±3.8	2.3±4.6	17.1±6.3
A-aDo2 (measured-predicted), mmHg	0.4±4.0	5.7±3.5	−2.0±6.6	7.4±9.0
DlO2, ml·min−1·mmHg−1·kg−1	0.6±0.2	2.1±0.6	0.7±0.3	1.6±0.5

Values are means ± SD. Log SDV̇ and log SDQ̇, log scale second moments of ventilation and perfusion distributions, respectively; A-aDo₂, alveolar-arterial O₂ tension gradient; Dl_O2, O₂ diffusing capacity.

^*P < 0.05 vs. corresponding value in normal group by ANOVA.

Fig. 1.

The relationships of cardiac output (Q̇_c) to O₂ uptake (V̇o₂) from rest to exercise measured by rebreathing (RB) and direct Fick methods were not significantly different. Regression line through pooled data: Q̇_c(RB) = 3.6 V̇o₂ + 260, R² = 0.41; Q̇_c(Fick) = 4.7 V̇o₂ + 225, R² = 0.70.

Fig. 2.

Relationships between carbon monoxide-diffusing capacity (Dl_CO) expressed under standard conditions (A), and the corresponding O₂-diffusing capacity (Dl_O2) derived from Dl_CO of alveolar membrane (Dm_CO) and pulmonary capillary blood volume (Vc) (B) with respect to Q̇_c were measured by RB technique in normal and post-resection groups. Regression lines are through pooled data. A: normal, Dl_CO = 0.0020 Q̇_c + 0.351, R² = 0.84; resection Dl_CO = 0.0020 Q̇_c + 0.153, R² = 0.74. B: normal Dl_O2 = 0.0036 Q̇_c + 0.470, R² = 0.77; resection Dl_O2 = 0.0028 Q̇_c + 0.431, R² = 0.56.

Fig. 3.

Relationship between Dl_O2 and Q̇_c was measured by multiple inert-gas elimination technique (MIGET) in normal and post-resection animals. Pooled data: normal, Dl_O2 = 0.0038 Q̇_c + 0.26, R² = 0.66; resection, Dl_O2 = 0.0035 Q̇_c + 0.14, R² = 0.44.

Fig. 4.

Relationship between Dl_O2 measured by MIGET and Dl_CO measured by RB. Normal: Dl_O2 = 1.57 Dl_CO, R² = 0.62; resection: Dl_O2 = 1.66 Dl_CO, R² = 0.49; combined: Dl_O2 = 1.61 Dl_CO, R² = 0.58.

Fig. 5.

A: relationship between Dl_O2 estimated from MIGET [Dl_O2(MIGET)] and from RB technique [Dl_O2(RB)] at corresponding levels of Q̇_c. Slope of the correlation through all data, labeled Linear (All), is not significantly different from identity. Normal: Dl_O2(MIGET) = 0.94 Dl_O2(RB), R² = 0.43; post-resection: Dl_O2(MIGET) = 1.00 Dl_O2(RB), R² = 0.27; combined: Dl_O2(MIGET) = 0.97 Dl_O2(RB), R² = 0.31. B: residuals and predicted values for the correlation between Dl_O2(MIGET) and Dl_O2(RB) for normal and post-resection groups. C: differences in Dl_O2 estimated by RB and MIGET were not different between normal and post-resection groups (4.5 ± 2.6 and −0.9 ± 2.6%, respectively; means ± SE).

DISCUSSION

Summary of findings.

In 43 normal and post-lung resection dogs, we measured the relationship between Dl_CO and pulmonary blood flow at incremental treadmill workloads using a RB technique at two O₂ tensions to estimate Dm_CO and Vc. From these estimates, the equivalent Dl_O2 value was calculated based on the relative diffusivity of the two gases, as well as the empiric in vitro uptake of CO and O₂ by whole blood. The same animals were studied using the MIGET at rest and exercise while breathing 14% O₂ to estimate V̇a/Q̇ distributions. From measurements of blood gases and cardiac output, we predicted the A-aDo₂ that would result from the observed degree of V̇a/Q̇ mismatch. If measured A-aDo₂ exceeded that which was predicted, the difference was attributed to diffusion limitation and was used in conjunction with blood and expired gas parameters to derive an estimate of Dl_O2.

Both methods confirm a linear relationship between Dl_O2 and cardiac output, which reflects functional recruitment of the alveolar microvasculature. The Dl_O2-to-Dl_CO ratio averaged 1.61 for pooled data. There is reasonable correspondence between Dl_O2 estimated by these two methods with a mean difference of under 5% in animals with normal, as well as reduced, lung capacities. Results support the robustness of the Roughton-Forster model that uses serial CO conductance of the membrane and blood for assessing the limits of alveolar O₂ uptake.

Critique of the methods.

The two methods of estimating Dl_O2 involve different theoretical principles, assumptions as well as experimental conditions, and invoke different potential sources of error. RB measurements were performed under ambient normoxia and hyperoxia. Measurements by MIGET were performed under ambient hypoxia. The two techniques were performed at least 4 mo following any surgical intervention, i.e., in a physiologically stable period. The two types of studies could not be performed together and were separated by an average of 4 mo. In most but not all animals, RB measurements were performed before and after studies by MIGET. Where available, the longitudinal RB data show no interval change in the relationship between Dl_CO(RB) and pulmonary blood flow. Although average body weight was similar between sham and post-resection animals, we normalized the data by body weight to minimize any discrepancy that might be caused by interval weight change. We expected the highly derived parameters (Dl_O2 calculated from Dm_CO and Vc) to exhibit greater variability than the primary measurements [Dl_O2(MIGET) and Dl_CO(RB)], and this was the case.

In the RB method, Dm_CO and Vc are solved by the Roughton-Forster equation (Eq. 1), using two simultaneous equations and Dl_CO measured at two alveolar O₂ tensions. The Roughton-Forster equation implicitly assumes that Dm_CO and Vc are independent of inspired O₂ tension. We have shown this assumption to be untrue. In finite-element model simulation of alveolar-capillary CO flux by diffusion, Dm_CO was found to decrease with increasing alveolar O₂ tension (18), implying an overestimation of Dm_CO and, consequently, an obligatory underestimation of Vc by the RB method. The magnitude of the errors is modest (under 5%) compared with true alveolar CO flux in model simulation within a physiological range of capillary hematocrit (18). Although our group and others have used nitric oxide diffusing capacity (Dl_NO) as presumably a more direct measure of Dm_CO in human subjects at rest and exercise and in anesthetized dogs (16, 32), the interpretation and limits of Dl_NO remain to be understood. We have not been able to measure Dl_NO in exercising dogs due to their high respiratory rate, which renders the response time of the NO analyzer inadequate.

Using MIGET, Dl_O2 is estimated as the value that explains the portion of measured A-aDo₂ that cannot be attributed to V̇a/Q̇ mismatch, assuming complete Dl/Q̇^.homogeneity. Therefore, Dl_O2(MIGET) represents a minimal value that explains the blood and inert-gas data. If significant Dl/Q̇ mismatch developed in these animals upon exercise or after lung resection, the estimated Dl_O2(MIGET) would be correspondingly higher, i.e., MIGET could potentially overestimate true Dl_O2 in the presence of Dl/Q̇ heterogeneity. The magnitude of Dl/Q̇ heterogeneity and hence overestimation of Dl_O2 by MIGET is unknown.

Red cell uptake of CO and O₂.

Variability in the assumed values of Θ_CO and Θ_O2, which were empirically measured in vitro, could alter the absolute values of Dm_CO and Vc, thereby altering the magnitude of derived Dl_O2(RB), but not its correlation with respect to Dl_O2(MIGET). A range of Θ_CO values has been reported (5, 12, 33, 34, 39), including two values for dog blood given by Holland, representing different values of red cell membrane diffusivity (12). We reported our results using one of these values expressed at a realistic body temperature during exercise (Eq. 2). Repeating the analysis using the second Θ_CO value for dog blood given by Holland (12) yields negligible difference in the resulting regression comparison between the Dl_O2 derived from Dl_CO and that measured as part of MIGET (data not shown). Reeves and Park (33) had shown, in a whole blood thin film preparation with minimal unstirred layer, that red cell CO uptake is entirely reaction rate limited at Po₂ >100 Torr, and that capillary red cell CO diffusion equilibrium is rapidly achieved during uptake. In our normal and post-resection animals, the mean Pa_O2 at peak exercise workload while rebreathing a 30% O₂ mixture averaged 117 ± 14 and 96 ± 15 Torr, respectively (means ± SD). Despite the lower mean Pa_O2 in the resection group, there was no difference in the relationship between Dl_O2 estimated by the two methods between normal and post-resection animals, suggesting that any variability in Θ_CO due to intra-erythrocytic diffusion limitation had minor effects on the Dl_O2 comparison. A range of Θ_O2 values also exists (31, 37, 44); we used the most recent value obtained by Yamaguchi et al. (44) in human blood using a stopped-flow technique after minimizing the effect of an unstirred layer around erythrocytes. It is uncertain whether a significant unstirred layer around capillary erythrocytes exists in vivo.

Previous comparisons of Dl_CO to Dl_O2.

The Dl_O2-to-Dl_CO ratios we observed (1.57–1.66) are consistent with that from earlier comparisons (1.36–1.71) and summarized by Hughes and Bates (24). The range of Dl_O2-to-Dl_CO ratios partly results from the fact that the ratio varies with the Pa_O2 at which Dl_CO is measured. The in vivo magnitude and distribution of Vc may also differ in normoxia (RB) and hypoxia (MIGET). Hempleman and Hughes (11) attempted to estimate Dl_O2 from Dl_CO using noninvasive data obtained from patients with interstitial fibrosis who show a diffusion limitation on exercise. Turino et al. (41) measured Dl_CO by a steady-state method compared with simultaneously measured Dl_O2 using the Lilienthal-Riley method under hypoxic conditions in normal subjects at rest and during supine and upright exercise. They showed an overall Dl_O2-to-Dl_CO ratio of 1.45, which is not very different from our results. Meyer et al. (30) used a RB method and ¹⁸O isotope to directly measure Dl¹⁸O₂ and Dl_C¹⁸_O in normal subjects at rest and during exercise. In another study from the same laboratory, Dl_C¹⁸_O measured by RB technique was compared with Dl_O2 estimated by the Lilienthal-Riley technique in tracheostomized dogs during low-intensity hypoxic exercise (36). In both studies, the average Dl_O2/Dl_CO was 1.2, close to the theoretical ratio based on Krogh's tissue diffusion constants (K_O2/K_CO = 1.23). In contrast, Savoy et al. (35) used a steady-state technique in anesthetized, paralyzed dogs ventilated with a hypoxic gas mixture and reported a paradoxical Dl_O2/Dl_CO of 0.40–0.62; the lower ratio was likely due to the high sensitivity of the steady-state method to V̇a/Q̇ heterogeneity, that develops in anesthetized preparations. Such heterogeneity is minimized in our data obtained during exercise using the RB technique. None of the earlier studies compared Dl_CO to Dl_O2 at the same cardiac output, which is a major determinant of alveolar microvascular recruitment and contributes significantly to the variability in the measured values within and between techniques.

Using a similar approach as that described here, we previously reported an indirect comparison of Dl_O2 estimated using the Roughton-Forster equation and using MIGET in separate studies involving normal human subjects. Dm_CO and Vc were obtained by a RB technique in healthy nonsmokers at rest and during exercise using published Θ_CO value for human blood (23), and compared Dl_O2 estimated by MIGET during hypoxic exercise in healthy subjects studied by Hammond and Hempleman (9) and in athletes studied by Hopkins et al. (13). The RB results were converted to equivalent Dl_O2 values using Eq. 3, expressed with respect to the simultaneously measured pulmonary blood flow, and compared with the relationships of Dl_O2 to cardiac output derived from V̇a/Q̇ distributions measured by MIGET. This indirect comparison (25) also demonstrates consistent agreement between Dl_O2 estimated by the two techniques.

We conclude that, despite the unavoidable variability involved in this analysis and the respective inherent sources of experimental error, the significant correlation, as well as the correspondence of Dl_O2 estimates obtained by two independent techniques support the use of Dl_CO, Dm_CO, and Vc in the Roughton-Forster model to derive an estimate of Dl_O2, to directly interpret alveolar O₂ transfer from rest to heavy exercise in subjects with normal, as well as reduced, lung capacities.

GRANTS

This research was supported by National Heart, Lung, and Blood Institute Grants R01 HL045716, HL040070, HL054060, and HL062873.

DISCLOSURES

The contents of this article are solely the responsibility of the authors and do not necessarily represent the official views of the National Heart, Lung, and Blood Institute or of the National Institutes of Health.