lungs inflate and deflate in response to changes in transpulmonary pressure (Pl), the pressure difference between the airway opening and the pleural space (Pao − Ppl). The potential for damage to the lungs caused by mechanical ventilation depends on the magnitude of Pl (8). However, current recommendations for management, such as positive end-expiratory pressure, are usually made in terms of the pressure difference applied by the ventilator across the whole respiratory system (Pao − Pbs, where Pbs is pressure at the body surface). This approach is reasonable if Ppl lies within a narrow range, as it does in relaxed, healthy subjects, but could be seriously misleading if Ppl were to vary widely and unpredictably. Recently, our laboratory reported that esophageal pressure (Pes), which is considered to reflect the effective Ppl surrounding the lung in healthy, upright subjects (2), ranged from 0 to >30 cmH₂O at relaxed end exhalation in subjects with acute respiratory failure (22). This variation in Pes could reflect real changes in Ppl, or it could be due to artifactual differences between Pes and average Ppl caused by the supine position or local pathological abnormalities. To the extent that elevated Pes reflects high Ppl, it indicates a compressive influence of the chest wall (i.e., the thorax and abdomen) that might require compensatory adjustments in ventilator pressures.

Before interpreting Pes in patients with acute lung injury, we sought to fully characterize the magnitude and variability of postural effects on Pes in normal subjects and to determine the contribution of postural change in lung volume to the postural change in Pes at resting end exhalation. Changing from upright to supine caused only minor changes in the relationship between Pes and lung volume, and over one-half of the effect was due to a change in resting lung volume. It remains to be determined whether measurements of Pes can be helpful for managing mechanical ventilation in patients with acute lung injury.

METHODS

Subjects were healthy adults, nine men and one woman, recruited from the hospital community (Table 1). Each provided written, informed consent using a protocol approved by the Committee on Clinical Investigations at Beth Israel Deaconess Medical Center.

Change in lung volume was measured with an 8-liter dry seal spirometer (Collins Cybermedic, Braintree, MA). Pes was measured with an esophageal balloon catheter (Jaeger, Hoechberg, Germany), inflated with 0.5-ml air, passed by nose or mouth, and swallowed to position its tip 40 cm from the incisors or nares. Pao and Pes were measured differentially to obtain Pl (Pl = Pao − Pes). In all postures, proper balloon position was confirmed by absence of detectable Pl change during respiratory efforts against an occluded airway. Pressure and volume signals were digitized and recorded for analysis by using Windaq software (Dataq Instruments, Akron, OH).

The first of two protocols was designed to measure postural differences in total lung capacity (TLC). To do this, we used continuous spirometry during postural change instead of repeated helium dilution measurements, because continuous spirometry avoids the noise associated with repeated measurement of functional residual capacity (FRC). The spirometer was flushed with air before each breathing episode, typically lasting 60 s, and was used without CO₂ absorbent to reduce volume change due to metabolic gas exchange. Subjects were coached to relax completely during exhalation to FRC. Then, breathing on the spirometer, they performed two or three tidal breaths followed by relaxed exhalation to FRC to approximate the relaxation volume (V_rel). After this sequence of tidal breaths, subjects inhaled maximally to TLC and, while continuing to breathe from the spirometer, changed posture and repeated the entire sequence. This measured the volume difference between TLC in the two postures. After a recovery period off the spirometer, the sequence was repeated with postural change in the opposite direction. For each pair of postures (upright and supine, upright and prone, and supine and left lateral), the complete sequence with postural change was performed 10 times, 5 times in each direction, and the average postural difference in TLC was calculated. Postural difference between TLC values in the upright and left lateral postures was computed as the sum of TLC changes on going between upright to supine and supine to lateral.

The second protocol was designed to assess Pl-volume (PV) characteristics in the same four postures. With the esophageal balloon in place, subjects breathed from the spirometer. After two or three tidal breaths ending at V_rel, subjects exhaled to residual volume (RV) and then inhaled to TLC, measuring the volume differences between TLC, V_rel, and RV. At TLC, the breathing circuit was occluded for approximately 3–5 s for measurement of static Pl. The subject then made a slow exhalation interrupted by intermittent occlusions of 3–5 s to generate individual points of the static deflation PV curve of the lung (Fig. 1). This maneuver was repeated five times in each posture. Pressures used for analyses were taken midway between the extremes of the cardiogenic oscillation to minimize this source of variability.

Analysis.

Volumes in all postures were reported and plotted relative to upright TLC. For supine, prone, and lateral postures, volumes were adjusted by adding a postural correction, which is the difference in TLC between upright and the other posture. For example, if V_rel in the supine posture was 4.6 liters below supine TLC, and the supine TLC was 0.5 liters less than upright TLC, then V_rel in the supine posture was reported as 5.1 liters below upright TLC. Similarly, PV characteristics were compared at the same lung volume across different postures by plotting all pressures at volumes relative to upright TLC (Fig. 2). Average upright PV curves were then constructed for each subject using a forward, stepwise, least squares regression procedure to isolate the most linear segment. Beginning with PV data in the mid-vital capacity (VC) range, successive regressions of volume on pressure were made as the data set was augmented by addition of data points at consecutively higher and lower lung volumes. The most linear segment of the PV curve in the upright posture was defined by the set of points yielding the highest correlation coefficient. Points within the same volume range were then used to construct PV curves for each of the other postures in each subject (Fig. 3). The Pl at relaxation (Pl_rel) was determined for each posture from the PV curves by using the posture-specific V_rel measured in the first protocol.

Postural differences in Pl were assessed by comparing Pl_rel at both the upright V_rel and posture-specific V_rel. Next, PV curves were compared for systematic differences in slope [i.e., the static deflation lung compliance (Cl)]. Significance of differences in Pl_rel and slope of the PV curve among postures was assessed by using a mixed ANOVA model with posture as fixed and subject as random effects, and pairwise comparisons (within subject) of PV slopes and V_rel using SPSS version 12.0 (SPSS, Chicago, IL). Results are reported as means (SD).

RESULTS

Postural effects on TLC, V_rel, RV, and VC are shown in Table 2. SDs for 10 measurements of TLC difference between postural pairs in individual subjects were small; the average SDs for upright-to-supine, upright-to-prone, and prone-to-lateral differences were 97, 133, and 111 ml, respectively. TLC was slightly greater in the upright posture than in other postures, and V_rel decreased from 35% of VC upright to 13% of VC supine. VCs were closely similar in all postures.

The Pl_rel averaged 3.7 cmH₂O upright and −3.3 cmH₂O supine (Table 3). The average upright-to-supine decrease in Pl (Pl_rel) was −7.0 cmH₂O (Table 4). Based on the horizontal shift of the PV curve at V_rel from upright to supine, the isovolume decrease in Pl_rel with assumption of the supine posture accounted for 42% of the total observed decrease in Pl_rel, whereas 58% was attributable to the decrease in V_rel.

Each subject produced static deflation PV curves in the upright posture that were highly reproducible and linear [mean R² = 0.892 (SD 0.080)] over a large portion of the VC. On average, this volume range extended from 22 to 78% of the largest VC measured in the upright posture. Supine PV curves also were reproducible and linear over that volume range [R² = 0.946 (SD 0.035)]. The PV slope was lower supine than upright in 8 of 10 subjects (Table 5) and for the group (Table 6). Despite substantial interindividual variation, postural change in PV slope and relaxation pressure were significant by ANOVA (for slope, F = 3.95, P = 0.018; for Pl_rel, F = 34.7, P < 0.001).

DISCUSSION

Our subjects had an average Pl that was 7.0 cmH₂O lower in the supine than in the upright posture, similar to the7.3-cmH₂O difference observed in five subjects by Mead and Gaensler (16). Approximately 4.1 cmH₂O (58%) of this difference can be attributed to the decrease in lung volume when changing from the upright to supine position, and the remaining 2.9-cmH₂O change is due to horizontal displacement of the PV curve (6, 13, 16, 18). Assuming that Pl is Pao − Pes in our subjects when upright, we estimate Pl near V_rel in the supine posture to be, on average, Pao − Pes + 2.9 cmH₂O.

Pl at V_rel in the supine position was negative in 7 of 10 subjects, on average −3.3 cmH₂O, and still negative after correcting for the shift in the PV curve. Pressure in the pleural space varies regionally due to gravitational gradients (15) and changes in the shape of the lung caused by displacements of surrounding chest wall structures, such as the heart and diaphragm-abdomen (12). Thus the Pes can, at best, reflect the pressure only at one locus in the pleural space, overestimating it in the nondependent pleural space and underestimating it in dependent regions (12). At low lung volumes, especially near the bottom of the pleural space, Ppl may indeed be locally positive, and thus a negative value of Pl is not unreasonable. Negative Pl are consistent with the low V_rel in our supine subjects, which averaged only 13% of VC.

Reluctance to use esophageal manometry in supine patients stems, in part, from studies in normal subjects, demonstrating that Pes at a given lung volume is increased, and Cl is apparently reduced, in the supine position (6, 13, 16, 18). These upright-to-supine differences in pressure and compliance have been attributed to an artifact caused by direct compression of the esophagus by mediastinal contents, such as the heart (6, 13, 16, 18). Lesser changes in Pes have been observed with change from supine to prone or lateral positions (6, 18). Previous studies (6, 13, 16, 18), each performed in four to seven subjects, did not characterize the contribution of changes in lung volume to the increases in Pes at end exhalation.

We found the average slope of the PV curve to be 15% less in the supine position than upright. This difference in the static deflation Cl as measured with an esophageal balloon was attributed by Knowles et al. (13) to compression of the esophagus by overlying mediastinal content. According to this explanation, as the lung deflates toward RV, the heart and other structures settle onto the esophagus, compressing it and raising Pes. The increasing esophageal compression at progressively lower lung volumes results in both a lower slope of the transpulmonary PV relationship and higher Pes observed supine at the reference volume (upright V_rel). Our data for the upright, supine, and prone positions are consistent with this mechanism. In the supine position, both Pl_rel and compliance are lower than upright, whereas, in the prone position, in which the mediastinal content is below the esophagus, Pl_rel and compliance are similar to those upright. However, in the left lateral position, compliance was decreased relative to upright (Table 6) without a concomitant decrease in Pl_rel (Table 4). As shown in Figs. 2 and 3, the horizontal position of the left lateral PV curve is similar to those of the prone and upright curves, but its slope is similar to that of the supine curve. How do we explain Pl_rel and compliance in left lateral recumbency, if not by mediastinal compression of the esophagus?

Alternatively, we speculate that changes in posture could cause true changes in the PV characteristic of the lung, changes that are not simply an artifact of Pes measurement. Cl could be decreased by inhomogeneous inflation caused by increased interregional differences in Ppl, such as would be caused by changes in the shape of the chest wall. For example, cephalad displacement of the dorsal diaphragmatic surface by the abdominal contents supine could compress the dependent lung, increasing the gravitational Ppl gradient. Animal studies have shown greater gravitational gradients in Ppl supine than prone, an effect attributed to a change in shape of the lung and chest wall (12, 15, 23, 24). D'Angelo et al. (4) proposed that increasing the Ppl gradient in rabbits causes an increase in lung elastance. We suggest that, in the upright and prone positions, the lung's inherent shape is close to that of its container, the gravitational Ppl gradient is low, and Cl high. Neither prone nor upright positions cause the compression of the dependent lung by the heart that occurs in the supine position (3). By contrast, in the left lateral position, the mediastinum settles in the chest, compressing the dependent lung and increasing interregional differences in Ppl, thereby decreasing Cl. In the left lateral posture, in which the heart and mediastinal content is largely below the esophagus, Pes is not elevated at V_rel, and changes in compliance cannot be explained by direct compression of the esophagus by the mediastinum.

In critically ill, mechanically ventilated patients, clinical decisions are informed by assessments of transrespiratory pressures and not Pl. Most studies utilizing esophageal manometry report only the changes in Pl and Pes, discounting by subtraction the baseline values (10, 17, 21). This practice ignores information about the chest wall that could be potentially useful.

Studies in animals have shown that ventilator pressures that are too high at end inflation or inappropriately low at end exhalation can worsen lung damage or cause ventilator-induced lung injury (8, 19). Recent articles on mechanical ventilation in acute respiratory distress syndrome have recommended a ventilatory strategy that prevents alveolar collapse by maintaining relatively high levels of positive end-expiratory pressure (9, 14, 19, 20) while limiting tidal volume to avoid lung injury from overdistension (1). However, these recommendations specify airway pressures and do not account for the variability in Ppl observed in critical illness (22). We suggest that estimating Ppl with esophageal manometry could reduce ventilator-induced lung injury and improve patient care.

The utility of esophageal manometry to estimate Ppl in supine patients has been debated (5, 11) and is at present unknown. We have confirmed in healthy subjects that, when adjusted for decreased lung volume, the average decrease in Pl from the upright to supine is <3 cmH₂O. This difference is nearly an order of magnitude less than the range of Pes values observed in critically ill patients with acute respiratory failure (22). It remains to be determined whether esophageal manometry can be useful in managing mechanical ventilation in patients.

GRANTS

This study was supported by National Heart, Lung, and Blood Institute Grant HL-52586.

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.