our laboratory has previously shown that when normal subjects are exposed to hypergravity there is an impaired arterial oxygenation (19) and lung diffusing capacity (20), which is more severe in supine than in prone posture. The aim of this study was to further analyze the mechanisms for a better preserved lung function in the prone posture in healthy subjects during normal and hypergravity by using indirect and simultaneous assessments of ventilation and perfusion distributions. Such simultaneous measurements of ventilation and perfusion were considered to be essential for a comparison of their relative influence. We hypothesized that indexes of ventilation and perfusion heterogeneity would be worse in supine than in prone posture during hypergravity as a result of a more marked lung tissue compression of dependent parts in supine posture. We also hypothesized that the results in the prone posture at five times normal gravity (5 G) would be qualitatively similar to those obtained in sitting hypergravity because of the anatomic similarity with a stable position of the heart toward one wall of the thoracic cavity, in contrast to the supine posture, in which the heart rests on underlying lung tissue. Another similarity would be the position of the diaphragm; gravity results in a cephalad positional shift of the diaphragm in the supine posture and an associated mechanical distortion of the dorsocaudal parts of the lungs, in contrast to prone and sitting postures in which there is, if any, rather a caudal positional shift of the diaphragm.

MATERIALS AND METHODS

Subjects.

Eight men and two women were studied. Their ages, heights, and body masses ranged from 21 to 29 yr, 165 to 191 cm, and 55 to 92 kg, respectively. They had no history of cardiopulmonary disease and were not taking medications at the time. They were also instructed not to drink coffee or use nicotine-containing products on the day of the experiment.

Ethics.

The subjects received written information about the procedure, and informed consent was obtained. The experimental protocol used in the present study was approved by the Ethics Committee of Karolinska Institutet, Stockholm, Sweden.

Equipment and measurements.

The experiments were conducted in the human centrifuge at Karolinska Institutet, Stockholm, Sweden. The centrifuge has two arms: one with a gondola and one with a platform. A support structure was mounted on this platform. The subject was placed on a padded support surface that could be adjusted to be perpendicular to the resultant of the normal-G and the centrifugal-G vectors. The subject was secured on the surface by a five-point safety belt. The head and torso of the subject were covered with a cowling to reduce air draft, noise, and visual inputs. The rotational radius of the centrifuge was 7.2 m at the middle of the support surface. Slip rings at the center of rotation allowed for audiovisual monitoring, power supply, and transmission of physiological signals between the platform and a control room. The instrumentation for respiratory measurements included a quadrupole mass spectrometer (AMIS 2000, Innovision, Odense, Denmark) and a wide-bore three-way solenoid valve (type 323-F, Bürkert, Egelsbach, Germany) with a 4-liter rebreathing bag. The subject breathed through a mouthpiece and wore a nose clip. Between the mouthpiece and the solenoid valve, there was an inlet for gas sampling through a 10-m-long capillary tube to the mass spectrometer located at the center of the centrifuge. The subject breathed through a pneumotachometer (type 3719, Hans Rudolph, Kansas City, MO), coupled to a pressure transducer (model CD12, Validyne, Northridge, CA) with its membrane mounted parallel to the plane of rotation (i.e., the horizontal plane) to eliminate the influences of centrifugal gravity (G) on the transducer. The volume of the instrumental dead space was 150 ml. ECG was monitored from chest electrodes with a clinical monitoring system (type AS2, Datex, Helsinki, Finland). To allow inspirations and expirations at controlled flow rates, an audio feedback system was constructed. A voltage-frequency converter created an audio signal with a frequency proportional to the flow rate. That signal and a reference signal of constant frequency were fed into each of the two channels of a stereo amplifier with its two loudspeakers mounted on each side of the head of the subject. The voltage-frequency converter was set so that the pitch of the flow-generated tone and the reference tone were identical at the desired flow rate.

The subjects had an ear probe for pulse oximetry (type EarSat Sensor, Datex, Ohmeda Division, Helsinki, Finland). The probe was held in place by an elastic bandage, and the ear lobe was pretreated with capsaicin ointment for vasodilatation to ensure a satisfactory pulse oximetry signal. Our laboratory (19) has previously shown an excellent agreement between this technique and concomitant arterial samples. An accelerometer was positioned on the support surface, perpendicular to the rotational radius and the support surface. Continuous signals were recorded at 200 Hz per channel in a digital data-acquisition system (Biopac, Goleta, CA). Before each experiment, the mass spectrometer was calibrated against gases of known composition, and after the experiments a calibration check was done. The pneumotachometer was calibrated with a 3-liter syringe within the experimental flow range. The response latency of the mass spectrometer was determined from a sudden, simultaneous change of gas composition and flow direction at the inlet of the sampling capillary (2). The latency between the flowmeter and the mass spectrometer signals was ∼3 s. The 95% rise time for the response to a step change in gas concentration was 150 ms.

Experimental procedures.

The experiments were performed at 1 and 5 G, in prone and supine postures, with the order randomized between the conditions. Each subject participated in one experimental sequence at each G level and posture for a total of four runs. The subject rested on a support surface on the platform of the centrifuge and breathed air through the mouthpiece. Approximately 4.5 min after reaching the desired G level, the subject performed one combined rebreathing and single-breath washout (SBW) maneuver in which the subject expired to functional residual capacity (FRC) and then rebreathed the full bag volume eight times at a rate corresponding to 3 s/breath. After a final inhalation from the rebreathing bag, the subject switched the rotary valve and exhaled completely to residual volume (RV) at a constant flow rate of 0.5 l/s, aided by audio feedback as described above. The subject then inhaled one vital capacity (VC) of cabin air and made a second expiration to RV, both at a rate of 0.5 l/s. The gas mixture used for rebreathing contained 35% oxygen (O₂) and 5% argon (Ar). In addition, and for other purposes (20), the mixture contained 3% sulfur hexafluoride (SF₆), 5% helium (He), 0.63% acetylene (C₂H₂), 0.3% carbon monoxide (C¹⁸O), and balance of nitrogen (N₂). The rebreathing gas volume varied between 1 and 2 liters, depending on the stature and the preference of the subject. Repetitions of the combined maneuver were separated by at least 10 min to permit elimination of foreign gases. At least 8 of these 10 min were at normal gravity. Each rebreathing maneuver was ended with a slow (∼0.5 l/s) exhalation to RV, at which gas tracings were analyzed for phase IV phenomena (closing volume and phase IV amplitude), which, if present, would indicate poor intrapulmonary gas mixing at the end of the rebreathing.

Data analysis.

Offline data analysis was performed with an Acknowledge 3.2 Biopac digital data-handling system (Biopac). Offline computations included algorithms for total dry pressure correction (22) and computation of calibrated values for all dry-gas fractional concentrations. Also, concentration readings were corrected for the response latency of the mass spectrometer system (2), and gas volumes and flows were converted to btps (body temperature, ambient pressure, saturated with water vapor) when appropriate. The pulmonary capillary blood flow data obtained during the preceding rebreathing maneuver have been reported elsewhere (20) and were used to calculate stroke volume (SV) in the present study. Heart rate was determined from the ECG during 15 to 5 s preceding the rebreathing maneuver, and SV was obtained as the ratio of pulmonary capillary blood flow over heart rate, with the assumption that the pulmonary capillary blood flow during the SBW maneuver was the same as that during the immediately preceding rebreathing maneuver. Flow was integrated to obtain expired volume.

Analysis of expirograms.

Three kinds of expirograms were generated: 1) Expired Ar concentration was plotted as a function of expired volume (Fig. 1A). The Ar concentration at the end of the rebreathing and before the VC inspiration was defined as 100% and the Ar in cabin air was defined as 0%. The resulting expirogram is analog to a conventional SBW expirogram in which a homogenously distributed resident gas (classically N₂) is diluted with one VC of another gas (classically pure O₂). 2) Expired CO₂ was plotted in a similar way, also normalized to the preinspiratory level by setting it to 100% and the CO₂ level of cabin air to 0% (Fig. 1A). 3) Finally, the ratio of the expired CO₂ over expired Ar concentrations, as defined above, was plotted as a function of expired volume (Fig. 1B). Henceforth the terms Ar and CO₂ will refer to percentages of Ar and CO₂ as defined above (Analysis of expirograms), when not stated otherwise.

In our analysis, the alveolar part of the expirogram was divided into two parts, phase III and phase IV (11, 17). Several indexes were calculated: the size of VC, phase III slope, closing volume (CV), phase IV amplitude, and amplitude of cardiogenic oscillations (COS). VC was defined as the maximal expiratory volume. The start and end of the alveolar plateau were identified by visual inspection, and phase III slope (%/l) was determined as the least squares best-fit line. The onset of airway closure was defined as the end of phase III and onset of phase IV, although not by an iterative process as described by Guy et al. (11). CV was calculated from this point to the end of the expiration and was also expressed as a percentage of VC (CV/VC). Phase IV amplitude was calculated as the vertical distance between the extrapolated phase III slope and the maximal concentration deviation at the end of phase IV. We selected the two largest consecutive COS during phase III with Ar, CO₂, or CO₂/Ar concentrations plotted against volume. We used the R wave-R wave interval from the ECG to find the local maxima and minima, referenced to the line already fitted to phase III. Because COS is affected by the size of the SV (12), we normalized COS for changes in SV by dividing COS with concomitant SV values (COS/SV). For a more detailed description of the rationale for the present analysis, we refer to a previous publication from our laboratory (21).

Statistical techniques.

ANOVA (Statistica 6.0, Statsoft, Tulsa, OK) with repeated-measures design with two independent factors [gravity (in the anterior-posterior or the posterior-anterior direction) and posture] was used to test for differences between changing G levels and posture and for interaction between these parameters. Planned comparison was used as post hoc test. Results were considered statistically significant if P < 0.05, and all tests were two sided. Data are presented as means ± SE, unless otherwise stated.

RESULTS

All 10 subjects completed the experiments. No phase IV phenomena (closing volume or phase IV amplitude) could be identified in any of the analyzed gases during the slow exhalation to RV at the end of the rebreathing maneuver. Typical individual recordings of expirograms from a SBW maneuver are shown in Fig. 1. VC and parameters extracted from expirograms are shown in Tables 1 and 2 and in Figs. 2–4. In the presentation of our data in Figs. 2–4, previous results (21) for the sitting position are added for comparison.

VC.

There was no difference between postures for VC at normal gravity. VC was significantly more decreased during hypergravity in the supine (−38%) than in the prone posture (−29%) (Tables 1 and 2).

Phase III.

COS/SV for Ar and COS/SV for CO₂/Ar increased in hypergravity and showed no difference between postures at any of the two G levels (Table 1 and Fig. 2). There was a tendency for a steeper phase III slope for Ar in the supine than in the prone posture at normal gravity (P = 0.08 for planned comparison between prone and supine at this G level). The phase III slope for CO₂/Ar was steeper in the prone than in the supine posture at normal gravity (1 G; P = 0.01). The phase III slope changed markedly in the supine posture with increased G force; for Ar the increase was to 580% of control (Fig. 3A), and for CO₂/Ar the slightly positive slope at 1 G changed to a markedly negative slope at 5 G (Fig. 3B). Hypergravity-induced changes of phase III slope were more pronounced in the supine than in the prone posture for both Ar (P = 0.003) and CO₂/Ar (P = 0.009) (Table 1, Fig. 3). There was a striking similarity between the phase III slope for Ar in prone and in sitting humans (Fig. 3A).

Phase IV.

There was a higher phase IV amplitude for Ar in the supine than in the prone posture at normal gravity (P = 0.04) (Table 1 and Fig. 4A). All parameters extracted from Ar expirograms during phase IV (CV for Ar, CV/VC for Ar, and phase IV amplitude for Ar) showed significant interaction between posture and gravity so that these were more marked in the prone than in the supine posture during hypergravity (Tables 1 and 2). Phase IV amplitude for CO₂/Ar was lower in the prone than in the supine posture during hypergravity (Table 1 and Fig. 4B). For CV for CO₂/Ar and CV/VC for CO₂/Ar, there was no gravity-induced difference in either of the postures but a tendency for a difference between prone and supine independent of G level (P = 0.07 for CV for CO₂/Ar and P = 0.06 for CV/VC for CO₂/Ar).

DISCUSSION

One major finding of the study was that of more severe small-scale ventilation and perfusion heterogeneities in the supine vs. prone posture at 5 G as reflected by the phase III slope. In apparent contrast, we found at the same time fewer signs of large-scale heterogeneities in terms of phase IV amplitudes in Ar and CO₂/Ar expirograms in supine hypergravity. This was paired with a more reduced VC than in prone hypergravity. As will be discussed below, the findings with respect to large-scale heterogeneity can be explained in terms of a steeper and less continuous change of pleural pressure along the vertical axis in supine compared with prone posture. In supine, the weight of the heart and an upward displacement of the diaphragm act together to create a positive pleural pressure in dependent lung regions, promoting airway closure also at relatively large lung volumes. At the same time, nondependent lung regions are stretched by a downward displacement of the heart, creating more negative pleural pressure, which counteracts airway closure in these regions despite a decreasing lung volume as expiration proceeds.

Vital capacity.

The hypergravity-induced decrease in VC from 5.0 liters at 1 G to 3.1 liters at 5 G in supine posture is similar to previous results from Glaister (8). This difference does not seem to be due to variations of intrathoracic blood volumes because previous observations in recumbent humans have shown that neither gravity nor the shift between prone and supine posture altered the lung tissue volume (including the pulmonary capillary blood volume) (20). It is known that FRC is smaller in supine subjects because of a cephalad shift of the diaphragm. Because the weight of the abdomen increases in hypergravity, this, most likely, causes a more pronounced cranial shift of the diaphragm during hypergravity. Our laboratory showed earlier that FRC was 400 ml smaller during hypergravity in supine than in prone posture (20). The larger VC reduction at hypergravity when subjects were supine, as found in this study, was most likely caused by the changing size of the expiratory reserve volume because the RV is not significantly altered during hypergravity (4, 8).

The reduced VC and FRC in supine vs. prone hypergravity suggest that the effect of the headward shift of the diaphragm in supine is larger than any compression of the thorax in the prone posture. As an alternative explanation, it can be speculated that these two effects are similar and that the difference in VC between postures during hypergravity is caused by the weight of the heart. In prone subjects, the heart is supported by the anterior chest wall, but in the supine posture it compresses underlying lung tissue (1). The compressed lung tissue area is unlikely to participate fully, if at all, in the VC maneuver. Previous observations of severe breathing discomfort and substernal pain on inspiration in supine but not in prone hypergravity (6, 20) speak in favor of such a mechanism.

Cardiogenic oscillations.

A more detailed discussion of the origin of the cardiogenic oscillations and other expirogram parameters is to be found in a previous publication from our laboratory (21). The influence of gravity as such on ventilatory heterogeneity and COS amplitude has previously been demonstrated by a 44% decrease in amplitude during a SBW maneuver with an insoluble gas when comparing 1-G standing with microgravity (11) and a 25–55% increase in COS/SV in sitting subjects during hypergravity (2 and 3 G) compared with normal gravity (21). To assess lung perfusion distribution in microgravity, Prisk et al. (17) studied COS for CO₂ after hyperventilation and breath holding and found lower COS amplitudes than during sitting control at 1 G. In conformity with this, we found in the present study that COS/SV increased at 5 G in both supine and prone postures for both Ar and CO₂/Ar, without any significant difference between postures (Fig. 2). However, Rodriguez-Nieto et al. (18) studied healthy humans performing a SBW with an insoluble gas on a rotating stretcher and found smaller COS in prone compared with supine posture. A difference compared with our experimental setup is that their subjects were attached to the stretcher by a leather holding garment covering only the anterior rib cage and leather straps over the waist and the ankles. The garment exerted no pressure on the abdominal region and probably led to a displacement of the diaphragm in the abdominal direction and therefore also an increased FRC. This may be a reason why their result differs from the present data. A comparison with COS/SV results from sitting humans exposed to 1–3 G (21) suggests similar responses in sitting and recumbent subjects (Fig. 2).

Phase III slope.

Previous results in humans during SBW tests at normal gravity show that phase III slope for an insoluble marker gas is steeper in the supine compared with prone posture (5, 18). In the present study, there is a tendency for a steeper phase III slope for Ar (P = 0.08) in the supine compared with prone posture during normal gravity in keeping with previous results. It has been suggested that the lesser pleural pressure gradient in prone compared with supine dogs (24) may explain the more uniform ventilation distribution in the prone posture (23). Using magnetic resonance imaging in humans, Mayo et al. (14) estimated the vertical gradients of pleural pressure to be three times smaller in the prone compared with supine posture.

Gravity-dependent contributions to phase III slope of an insoluble marker gas in the supine posture during normal gravity have been suggested to be 18% (18) and 22% (11) compared with only 7% in the prone posture (18). The notion of a difference in gravity dependence between postures is supported by the results in the present study, in which the hypergravity-induced change in phase III slope in the Ar expirograms was more than five times larger in the supine posture whereas in the prone posture there was only a doubling of the same slope with hypergravity. To differentiate between diffusive and convective transport phenomena behind the slope of phase III, SBW maneuvers must be performed with the simultaneous inhalation of a pair of insoluble gases with widely differing diffusivities, such as He and SF₆ (10, 13). This was not done in the present study. We consider it, however, likely that the hypergravity-induced increase of the phase III slope was caused by alterations of convective gas transport secondary to hypergravity-induced increased gradients of the volume expansion rate along the direction of gravity. This conclusion is supported by findings from Gustafsson et al. (10), who showed, in sitting subjects, that the SF₆-He phase III slope did not change with gravity up to 3 G. At the same time, the phase III slopes for both gases increased by 40–60%, suggesting that convective-dependent heterogeneity (as reflected by the phase III slope of either of the insoluble gases), but not diffusion-convection interaction-dependent heterogeneity (as reflected by the SF₆-He phase III slope), is influenced by gravity.

Prisk et al. (17) used a hyperventilation-breath hold-SBW maneuver to study the distribution of pulmonary perfusion during sustained microgravity compared with standing and supine subjects in normal gravity. Phase III slope for CO₂ was the same in sustained microgravity compared with standing in normal gravity and it was steeper in 1 G supine. Prisk et al. suggested that the increase in pulmonary capillary blood flow in supine compared with standing contributed to the steeper CO₂ slope by providing a larger CO₂ delivery to the lungs during the maneuver.

In the present study, there was a twofold greater positive phase III slope for CO₂/Ar in the prone compared with supine posture at normal gravity (P = 0.01). The slightly higher pulmonary blood flow (approximately +5% at 1 G) (20) can only have accounted for a small part of that difference. The most likely explanation is that the effect of continuing gas exchange was partly offset by gradually decreasing circulatory delivery of CO₂ in lung units emptying toward the end of expiration in the supine posture.

In hypergravity, phase III slope for CO₂/Ar differed much more between prone and supine posture than at normal gravity. In prone posture, phase III slope for CO₂/Ar was slightly positive just as in 1 G, whereas in the supine posture the positive slope at 1 G changed to a markedly negative slope at 5 G. These differences between prone and supine phase III slopes for CO₂/Ar can only be accounted for to a minor extent by differences in continuing gas exchange; there was ∼20% less pulmonary blood flow in the supine compared with prone posture at 5 G (20). However, a corresponding reduction of the phase III slope for CO₂/Ar would, all other factors being equal, still be a positive slope and far from that found in supine hypergravity. We therefore conclude that in supine posture there is a much greater gradient of regional perfusions within the population of lung units that empty sequentially during phase III, compared with the prone posture. This conclusion is consistent with our laboratory’s previous suggestion that there is a more homogenous perfusion distribution in prone hypergravity (20). This difference between postures can most probably be explained by the significantly higher cardiac output in prone than supine hypergravity, 4,600 vs. 3,800 ml/min (20). The higher cardiac output in the prone posture is presumably accompanied by a larger number of pulmonary capillaries recruited and, therefore, a more homogenous perfusion distribution in that posture.

Phase IV.

Previous studies in anesthetized dogs (23) and humans (18) in normal gravity show a smaller phase IV amplitude during SBW tests for N₂ in the prone than in the supine posture, which is in agreement with the present Ar results. Also, studies in anesthetized dogs (23) at normal gravity have shown a smaller closing volume during SBW tests for N₂ in the prone than in the supine posture. Similarly, in the present study, there was a tendency for a smaller CV for Ar in the prone compared with the supine posture in normal gravity (P = 0.09). Thus large-scale ventilation seems to be more homogenously distributed in the prone than in the supine posture at normal gravity.

Our observation during hypergravity, that phase IV amplitude for Ar and CV/VC for Ar were larger in the prone than in supine posture, is contrary to what might be expected. We have previously shown that gas exchange in recumbent humans exposed to 5 G is more impaired supine compared with prone (19), and impaired gas exchange would intuitively be associated with airway closure as a sign of impaired distribution of ventilation. However, because phase IV represents the late emptying of lung units with poor ventilation per unit lung volume, a likely interpretation is that such units contribute to a lesser extent to the expirate when subjects are supine than when prone.

There is a striking similarity between the data obtained from sitting subjects (21), included in Figs. 2–4, and the present prone experiments. The similarities are greatest for phase IV amplitude of the Ar expirograms (Fig. 4A). We speculate that these similarities are caused by qualitatively similar heart-lung and diaphragm-lung interactions. In both body postures, the heart is located adjacent to one inner wall of the thoracic cavity: in sitting against the diaphragm and in prone against the anterior chest wall. Thus the lung tissues are likely neither extended nor compressed to any significant extent by any limited gravity-induced displacement of the heart, and consequently ventilation and perfusion distributions are chiefly determined by gravity-induced gradients within the lung tissue itself. Such gradients, although linear along the axis of gravity, include nonlinear threshold phenomena such as airway closure once the pleural pressure exceeds alveolar pressure down the lungs; this occurs to an increasing extent as expiration proceeds to lower lung volumes (9). In the supine posture during hypergravity, however, the downward displaced tissue includes the heart with its much larger density compared with the surrounding lung tissues. This is likely to contribute to more negative pleural pressures in lung tissues above the heart even at low lung volume when subjects are supine so that the conditions for airway closure are not likely to be met to the same extent as in prone and upright postures.

There is probably also a difference in lung-diaphragm interaction in supine compared with sitting and prone posture. In the supine posture, dorsocaudal lung parts are mechanically distorted because of a cephalad positional shift of the dorsal parts of the diaphragm (15). This is not the case in prone and sitting posture where there is rather a caudal positional shift of the diaphragm (8, 15). The compressed lung tissue in supine hypergravity is unlikely to participate fully, if at all, in the VC maneuver.

The expirogram for CO₂/Ar obtained in hypergravity showed a markedly more negative end-expiratory deviation in the prone compared with supine posture. From a purely mathematical point of view, it is clear that if the CO₂ tracing is relatively constant during the end of the VC expiration (Fig. 1) and the Ar tracing shows a marked upward deviation at the same time (Fig. 1), then the CO₂/Ar expirogram must show a corresponding negative deviation (Fig. 2). From a physiological point of view, however, our interpretation of the negative deviations of the CO₂/Ar expirogram is that lung units with gradually lower CO₂ delivery from perfusion contribute to an increasing extent to the expirate as expiration proceeds. At the same time, these units have been gradually less diluted by inspired air as shown by a gradual increase of Ar, reflecting a gradually lower degree of dilution of the resident CO₂ before expiration. With respect to CO₂ content in lung units toward the end of expiration, these two phenomena offset each other, creating a leveling off of the CO₂ tracing.

The markedly lower amplitude of the phase IV of the CO₂/Ar expirogram in the supine compared with prone posture in hypergravity can be explained by factors analogous to the corresponding Ar expirogram, namely that the units with the lowest perfusion per unit lung volume contributed less to the end-expiratory part of the expirogram in the supine compared with prone posture.

Comparison with previous data on alveolar-to-arterial gas transport.

When put together with our laboratory’s previous findings of more severe arterial hypoxemia (19) and more impaired lung diffusing capacity (20) in supine compared with prone hypergravity, the present data support the notion of a more severe ventilation-perfusion mismatch in the supine posture. We propose that the primary cause is reduced ventilation due to mechanical compression of some parts of the lung caused by the heart and the diaphragm in the supine posture. However, data on the topographical distributions of ventilation and perfusion in the lungs during hypergravity exposure are required to verify such a mechanism.

Conclusions.

The present data are consistent with more severe small-scale ventilation and perfusion heterogeneities in the supine vs. prone posture at 5 G as reflected by the phase III slope. The more pronounced VC reduction in supine hypergravity may be interpreted as an even more marked large-scale heterogeneity of ventilation distribution. There are probably greater vertical pleural pressure differences in supine subjects at 5 G, and therefore nondependent lung parts show only limited airway closure, whereas lung parts below the heart are compressed to the extent that they did not participate fully to the expirate. This explains the lower VC, lower CV, lower CV/VC, and reduced phase IV amplitude for Ar in supine hypergravity compared with prone. The more negative phase IV amplitude for CO₂/Ar prone is probably due to longer expiration in this posture when gas from poorly perfused nondependent lung regions continue to empty.

Clinical perspective.

Essentially we had the following line of thought when planning this study: patients with acute respiratory distress syndrome (ARDS) have an improved arterial oxygenation when turned from the supine to the prone posture (16). In those patients, several factors contribute to the marked impairment of oxygen transfer from the inspired gas to the arterial blood, e.g., 1) fluid accumulation in the alveolar space; 2) edema in interstitial lung tissue; 3) injuries to both the lung endothelium and epithelium (3); and 4) compression of pulmonary tissue, both by the heavy, fluid-filled lung itself and by the enlarged heart (1, 7). However, in critically ill patients it may be difficult to determine the relative contributions of these factors to the impairment of oxygenation in the lungs and to the improvement that occurs in the prone posture. We reasoned that, when exposed to short periods of hypergravity, healthy subjects would become partially analogous to ARDS patients in the sense that they would experience pronounced gravity-induced deformation of the lungs but without interstitial or alveolar edema or injuries to the lung endothelium and epithelium. Taken together, the present study and recent other work from our laboratory (19, 20) show that the gas-exchange impairment from lung tissue deformation is less severe in the prone than in the supine posture. These observations provide strong and new support for the notion that lung tissue compression is the most likely component to be involved in the improvement of the gas exchange seen in patients with ARDS when placed in prone posture.

GRANTS

This study was supported by the Swedish National Space Board, Fraenckel's Fund for Medical Research, The Laerdahl Foundation for Acute Medicine, and the Swedish Research Council (Project nos. 05020 and 10401).

FOOTNOTES

• 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. Section 1734 solely to indicate this fact.