Introduction

In acute respiratory distress syndrome (ARDS) the analysis of the quasistatic pressure-volume (PV) curve may aid in ventilatory settings adjustment [1, 2, 3, 4, 5, 6, 7]. Inflation PV curve-based setting of positive end-expiratory pressure (PEEP) has been shown in lavage-injured sheep to be most effective in maintaining gas exchange and minimizing lung injury [8]. In ARDS the prone position results in decreased plateau transpulmonary pressure, increased static lung compliance, and improved gas exchange [9, 10, 11, 12, 13]. Previously demonstrated pronation benefits on lung mechanical behavior suggest that prone position improves the shape of lung inflation PV curves. However, to date relevant data are scarce and conflicting [14].

We theorized that in severe ARDS, prone position would result in (a) leftward displacement of the lung inflation PV curve and (b) decreased pressure at the lower inflection point (LIP) and increased volume at the upper inflection point (UIP) of the lung PV curve. We also sought to determine any relationship between the postural PV curve changes and any concomitant alterations in gas exchange.

Materials and methods

Patients

Institutional approval and informed, written next-of-kin consent were obtained. The study enrolled 13 consecutive, orotracheally intubated (cuffed endotracheal tube no. 7.5–8.5) and mechanically ventilated (Siemens 300C, Siemens, Berlin, Germany) patients with early-onset, early-stage (i.e., within 7 days of onset), and severe ARDS (Table 1) [9, 15, 16, 17]. Inclusion criteria were a PaO2/FIO2 ratio of100 mmHg or less, “white lungs” feature on chest radiography, and/or diffuse attenuations on thoracic computed tomography [16]. Exclusion criteria were age under 18 years, pregnancy, intracranial hypertension, burns over more than 30% of body surface area, spine fractures, smoking and/or chronic respiratory disease, hemodynamic instability (systolic blood pressure <95 mmHg), chest wall abnormalities and pneumothorax, chronic liver disease (Child-Pugh class C), neuromuscular disease impairing spontaneous breathing, sickle cell disease, and body mass index greater than 27.5.

Table 1 Patient characteristics and baseline ventilatory settings (volume assist-control mode). For males PBW was calculated as “50 + [height (cm)–152.4] ×0.91”; for females “50” was replaced by “45.5.” (BMI body mass index, SAPS simplified acute physiology score, FIO 2 inspired oxygen fraction, PEEP positive end-expiratory pressure, Vt tidal volume, PBW predicted body weight, P 2 ,aw plateau airway pressure, C,rs quasistatic compliance of the respiratory system, ARDS acute respiratory distress syndrome, S survival to intensive care unit discharge, D death in the intensive care unit)
Full size table

Electrocardiographic lead II, intra-arterial/pulmonary artery pressures, cardiac index (744HF75 CCOmbo Swan-Ganz catheter, Edwards Lifesciences, Irvine, Calif., USA), and peripheral oxygen saturation (SaO2) were monitored continuously. Esophageal and gastric balloons were inserted and their correct placement as previously verified [18, 19, 20]. Additional information is provided in the Electronic Supplemental Material.

Protocol

On enrollment the patients were anesthetized/paralyzed [9] and ventilated in a baseline preprone semirecumbent position (45° inclination). Baseline ventilation tidal volume (Table 1) was titrated to a plateau airway pressure less than 30 cmH2O or kept within 6–6.5 ml/kg if plateau pressure exceeded 30 cmH2O. Inspiratory flow ranged within 0.75–0.90 l/s. Respiratory rate (range 20–30/min) was adjusted to maintain pHa at or above 7.30. PEEP was set at 2 cmH2O above a mean LIP determined graphically [2, 7] by two independent clinicians within 24 h before enrollment. These physicians constructed respiratory system PV curves at zero PEEP with multiple inspiratory airway occlusion [21]; administered inflation volume ranged within 0.1–0.8 l, because the main objective was to determine the LIP.

Patients were sequentially studied in three body postures: (a) preprone semirecumbent, (b) prone, and (c) postrpone semirecumbent (45° inclination). For each posture the study duration was 135–140 min. Measurements were performed during PEEP application, zero PEEP, and after original PEEP restoration. Minimal allowable SaO2 was 89%. Following pronation a roll and a pillow were placed under the upper chest wall and pelvis, respectively [9, 10]. Figure 1 illustrates the time sequence of investigational interventions. The time points of recruitment with inflation volumes of 1.0 l or greater (see below) [9, 12, 22], and the periods of volume history standardization are also indicated. Four minutes prior to PEEP removal, inspired oxygen fraction (FIO2) was increased to 1.0. FIO2 was restored to its original level (Table 1) within 5–15 min after PEEP restoration (Fig. 1). Following PEEP removal volume history was standardized over only 5 min, because: (a) in ARDS, a single, prolonged expiration to zero PEEP leads to reversal of PEEP-induced recruitment [6], implying rapid respiratory system equilibration to zero PEEP conditions; (b) the expiration just before PV curve construction was prolonged [6] (see below); and (c) by protocol, the time of uninterrupted zero PEEP ventilation should be minimized. If hemodynamics/gas exchange determinations (required time=6 min) were performed immediately after the 5-min period of equilibration at zero PEEP, the time of uninterrupted zero PEEP ventilation would have risen to 11 min (Fig. 1).

Fig. 1
figure 1

Time sequence of investigational interventions for each body posture. Lung volumes included determination of (a) physiological dead space (2 min of collection of expired gases into a Douglas bag and determination of CO2 concentration with an infrared analyzer); (b) functional residual capacity (FRC) change (passive exhalation from end-expiratory lung volume to FRC lasting 10 s or less; the reaching of FRC level was confirmed by an airway pressure of 0 cmH2O during an expiratory occlusion of 2 s); and (c) FRC by helium-dilution (corresponding to approx. 2 min of ventilation with high inflation volumes of 1.0–1.3 l). In all postures during external positive end-expiratory pressure (PEEPe) application inspired oxygen fraction (FIO2) was identical to that reported in Table 1 (baseline FIO2); during the last 4 min of the first PEEPe period and during the first 15 min of the zero PEEPe period, ventilator FIO2 was increased to 1.0; during the last 2 min of the zero PEEPe period (and during all FRC measurements), FIO2 was 0.87, because FRC was measured with the use of 13% helium in oxygen [9, 10]; following PEEPe restoration, ventilator FIO2 was readjusted to baseline FIO2 within 5–15 min in all cases, with peripheral oxygen saturation being always maintained at or above 89%. Periods of volume history standardization (VHS) and time points of administration of recruitment maneuver(s) (RM) are highlighted (see also text)

Full size image

Complete sets of measurements were obtained from all patients and no protocol-related complications [11, 23] occurred. Eight patients were weaned from mechanical ventilation 15.8±1.2 days after its institution and discharged from intensive care unit after another 2.5±0.2 days. Five patients died of multiple system organ failure (Table 1).

PV curves

Flow, volume, and airway, esophageal, and gastric pressures were measured as previously described [9, 24]. Esophageal pressure measurements’ reliability was as previously tested [9, 24]. In each posture inflation PV curves of the respiratory system, lung, and chest wall were constructed at zero PEEP (Fig. 1). Functional residual capacity (FRC) was reached during a prolonged (5–10 s) exhalation [25]; this was confirmed by an expiratory occlusion pressure of 0 cmH2O [25]. A 2-l calibration syringe filled with 100% oxygen was then connected to the endotracheal tube. PV curves were obtained by sequentially adding 50–100 ml in steps of 2–3 s until the lungs were inflated with 1.2 l. Procedure duration was 40–45 s. Maximal allowable airway and transpulmonary pressures were 50 and 35 cmH2O, respectively.

PV curve data analysis

PV diagrams for the respiratory system, lung, and chest wall were constructed. Regression lines for three or four consecutive points were determined with the Statistical Package for Social Sciences version 10.0 (SPSS, Chicago, Ill., USA). The slopes of the regression lines were determined in Adobe Photoshop 5.0 (Adobe Systems, San Jose, Calif., USA) with the Measure Tool. On confirmation of a consistent change in regression line slope of 20% or more [19] the intersection point of the linear tracts identifying the zones of initial and maximal compliance was identified as LIP. Accordingly, the UIP was identified at the intersection of the linear tracts identifying the zones of final and maximal compliance. Linear tracts were drawn as regression line extensions with the Line Tool. Following LIP and UIP determination the identity of the slopes of the regression lines and their extensions was confirmed with the Measure Tool. Initial (starting) static compliance was determined as inflation volume at LIP divided by pressure at LIP. For the respiratory system and lung, maximal static compliance was determined as inflation volume difference between UIP and LIP divided by the corresponding pressure difference; final static compliance was determined as inflation volume difference between the PV curve’s maximal pressure point and UIP divided by the corresponding pressure difference. Chest wall PV curves did not exhibit UIPs. Therefore maximal static chest wall compliance was determined as inflation volume difference between the PV curve’s maximal pressure point and LIP divided by the corresponding pressure difference.

Each diagram was analyzed by four independent observers. Computer files containing diagrams were named (encoded) according to patient initials, body posture, and respiratory system component. Following completion of diagram evaluations, diagram codes were broken, and means of observer-determined variable values were analyzed statistically.

Hemodynamics, gas exchange, and lung volumes

Measurements were performed over 6-min periods (Fig. 1). Hemodynamic measurements included arterial, central venous, and pulmonary artery pressures, cardiac output/index, and mixed-venous oxygen saturation (SvO2). Arterial and mixed-venous blood gas analysis (ABL System 625, Radiometer, Copenhagen, Denmark) was concomitantly performed. Physiological dead space, FRC change, and FRC were measured as previously described [9, 10, 12, 24, 26].

Statistics

Data sets from different body postures or end-expiratory pressure were compared using SPSS by repeated-measures analysis of variance for two within-group factors, followed by Scheffé’s test as appropriate. Significance was set at p<0.05. Multiple linear regression analysis was performed by stepwise method; variable entry and removal criteria were p<0.05 and p<0.1, respectively. Data are reported as mean ±standard error.

Results

As previously [9], gastric and esophageal pressures at FRC, and mean maximal amplitude and mean rise rate in “cardiac oscillations” of esophageal pressure tracings [9, 24] did not differ significantly between study postures (Table 2). Consequently, (a) abdominal movement restriction was effectively minimized in the prone position [9, 10, 24], and (b) the initial, correct esophageal balloon positioning relative to the heart was maintained throughout the study period, and measurements of esophageal pressure changes were as accurate as possible in all postures [9, 24]. The esophageal balloon technique has previously been adopted by us and others and shown to be adequate also in the prone position [9, 10, 27, 28, 29].

Table 2 Esophageal and gastric pressures at functional residual capacity and cardiac oscillation-induced changes in esophageal pressure. For each study posture the mean maximal amplitude and mean rise rate in esophageal pressure (Pes) [9, 24] were averaged over six to ten consecutive cardiac oscillations recorded just prior to the pressure-volume curve construction (FRC functional residual capacity, Pga gastric pressure)
Full size table

Reproducibility of computerized methodology

Measurements of the LIP pressures of the respiratory system, lung, and chest wall exhibited average interobserver variabilities of 2.2–2.7%, 4.1–9.6%, and 3.7–4.3%, respectively. For the respiratory system and lung average interobserver variability values of UIP pressure measurements ranged within 1.1–1.3% and 1.3–2.2%, respectively. Thus LIP and UIP pressures were consistently determined within 0.37 and 0.40 cmH2O, respectively. (See also ESM.)

PV curves

Figure 2 displays average PV curves. Prone position resulted in leftward shifting of the respiratory system and lung PV curves. Table 3 summarizes results on determined variables of the respiratory system, lung, and chest wall PV curves. Main results were as follows: (a) Respiratory system PV curves: relative to preprone/postprone semirecumbent, the UIP corresponded to similar pressure but significantly higher volume in the prone position. Final respiratory system compliance was higher in prone vs. preprone semirecumbent. (b) Lung PV curves: the LIP corresponded to lower pressure/volume in prone vs. preprone/postprone semirecumbent (Fig. 3). The UIP corresponded to similar pressure but higher volume in prone than in preprone/postprone semirecumbent. Starting and maximal lung compliance were higher in prone than in preprone/postprone semirecumbent. Final lung compliance was higher in prone vs. preprone semirecumbent. (c) Chest wall PV curves: maximal chest wall compliance was lower in prone than in preprone semirecumbent. The relative contribution of the LIP of the chest wall PV curve to the respiratory system LIP (Fig. 3) was higher in prone than in preprone/postprone semirecumbent (69.3±2.0% vs. 59.6±1.4%/52.4±1.5%, respectively, p<0.001 for both).

Fig. 2
figure 2

Average pressure-volume curves of the respiratory system, chest wall, and lung. Squares Mean; horizontal bars standard error; SREC semirecumbent. *p<0.05 vs. preprone semirecumbent, †p<0.05 vs. postprone semirecumbent

Full size image
Table 3 Results on respiratory mechanics variables determined by analysis of inflation pressure-volume curves (LIP lower inflection point, UIP upper inflection point, SREC semirecumbent, NA not applicable)
Full size table
Fig. 3
figure 3

Individual values of pressure at the lower inflection point (LIP) of the pressure-volume (PV) curves of the respiratory system (A), chest wall (B), and lung (C); horizontal bars mean; SREC semirecumbent. *p<0.05 vs. preprone semirecumbent, †p<0.05 vs. postprone semirecumbent

Full size image

Hemodynamics, gas exchange, and lung volumes

Hemodynamic variables were unaffected by posture change (data not shown). PEEP caused significant drops in mean arterial pressure (by 8–12%) and rises in pulmonary artery pressure and pulmonary vascular resistance index (by 14–19% and 14–44%, respectively). Prior to PEEP removal PaO2/FIO2, PaCO2, shunt fraction, physiological dead space, and FRC were improved in prone and postprone semirecumbent relative to preprone semirecumbent. Following PEEP removal and after PEEP restoration, gas exchange, and lung volumes were improved in prone relative to preprone and postprone semirecumbent (Table 4). With respect to posture change, the results on end-expiratory lung volume (EELV) were similar to those on FRC (Table 4).

Table 4 Gas exchange, and lung volumes (Pa arterial partial pressure, FIO 2 inspired oxygen fraction, QS/QT shunt fraction, Vd/Vt physiological dead space, FRC functional residual capacity, ΔFRC change in FRC, EELV end-expiratory lung volume, i.e., the sum of FRC and ΔFRC, SREC semirecumbent, PEEP positive end-expiratory pressure of 10.2±0.6 cmΗ2O employed in each body posture, ZEEP zero PEEP, PEEP-R restored PEEP level of 10.2±0.6 cm Η2O employed in each body posture)
Full size table

The principal consequences of PEEP removal were (a) significant reductions in PaO2/FIO2 only in the semirecumbent postures, (b) significant increase in shunt fraction in all postures, and (c) significant FRC reduction only in postprone semirecumbent (Table 4). In postprone semirecumbent oxygenation and shunt fraction exhibited the maximal observed deterioration after 8–11 min of recruitment-interrupted, zero PEEP ventilation (Fig. 1), thus becoming comparable to oxygenation/shunt fraction in preprone semirecumbent (Table 4). Furthermore, after 14 min of recruitment-interrupted zero PEEP ventilation (Fig. 1), FRC became comparable to the FRC of preprone semirecumbent (Table 4).

PEEP restoration in preprone semirecumbent resulted mainly in full reversal of the changes induced by the preceding PEEP removal in gas exchange. In postprone semirecumbent (Fig. 1) PEEP restoration resulted in a slightly lower shunt fraction than in preprone semirecumbent (Table 4), with PaO2/FIO2 and FRC/EELV remaining substantially closer to preprone semirecumbent rather than prone. Consequently in postprone semirecumbent the employed ventilatory maneuvers and volume history standardization (Fig. 1) adequately changed the initial “prone-like” gas exchange and lung volumes to “preprone semirecumbent-like.”

Regression analysis

Increases in PaO2/FIO2 from preprone semirecumbent to prone either at PEEP or at zero PEEP were closely related to concomitant decreases in LIP pressure of the lung PV curve (R2=0.75–0.79, p<0.001 for all; Fig. 4); similar relationships were observed from prone to postprone semirecumbent (data not shown). Multiple linear regression analysis (dependent variable=postural change in PaO2/FIO2; independent variables=postural changes in PV curve-derived variables and lung volumes) revealed that the sole independent predictor of PaO2/FIO2 increases were the corresponding decreases in the LIP pressure of the lung PV curve. Significant relationships were also observed between increases in PaO2/FIO2 from preprone semirecumbent to prone and concomitant postural changes in PaCO2, FRC, and starting lung compliance (Fig. 5A–C). Postural changes in starting lung compliance were also significantly related with the concomitant increases in FRC (Fig. 5D).

Fig. 4
figure 4

Scatterplots of individual changes in arterial oxygenation (PaO2/FIO2) and changes in the pressure at the lower inflection point (LIP) of the lung pressure-volume curve; the presented PaO2/FIO2 differences are between preprone semirecumbent and prone position and were determined prior to the removal of the positive end-expiratory pressure (A), at zero end-expiratory pressure (B), and after the restoration of the positive end-expiratory pressure (C) (see also text); the curves at either side of the regression lines represent 95% confidence limits

Full size image
Fig. 5
figure 5

Relationships determined at zero end-expiratory pressure between changes from preprone semirecumbent to prone in PaCO2 and PaO2/FIO2 (A), PaO2/FIO2 and functional residual capacity (FRC) (B), PaO2/FIO2 and starting lung compliance (Cstart, L) (C), and Cstart, L and FRC (D)

Full size image

Discussion

The major findings of this study are that in early-stage, severe ARDS (characterized by diffuse lung damage with intra-alveolar and interstitial edema and hyaline membrane) (a) prone position vs. preprone semirecumbent causes leftward shift of the respiratory system and lung PV curves, and reduces LIP pressure and increases UIP volume of the lung PV curve; (b) prone position also increases starting, maximal, and final lung compliance and FRC/EELV, and improves gas exchange; and (c) body posture related changes in PaO2/FIO2 can be explained by concomitant changes in the LIP pressure of the lung PV curve.

Effects of prone position

Our findings on lung PV curve-derived variables suggest augmentation of the continuous [6], lung inflation-induced alveolar recruitment in the prone position, and attenuation of alveolar overdistension above the UIP. Prone ventilation renders lung topography more suitable to recruitment [12] by: (a) providing a more homogeneous distribution of pleural pressures and ventilation [10, 30, 31, 32], (b) relieving regional lung compression by the heart [33] and abdominal contents [34], and (c) improving aeration of dependent lung units [10].

A previous study by Pelosi et al. [10] found no pronation-induced increase in EELV or decrease in shunt fraction. However, they did not employ the criterion of “diffuse ARDS” [16] during patient selection. Also, they enrolled patients with both primary and secondary ARDS [10, 35]. However, patients with differing lung morphologies and/or ARDS may present different physiological responses to PEEP or prone positioning [36, 37].

In the present study oxygenation, PaCO2, shunt fraction, EELV/FRC, and lung compliance were early (within 80 min) and relatively uniformly improved in response to prone positioning (Table 4; Figs. 4, 5). Furthermore, these changes were partly explanatory of each other, thus implying the simultaneous presence of (a) improved ventilation homogeneity leading to both improved ventilation-perfusion matching and effective alveolar ventilation, as suggested by the improved gas exchange (Fig. 5A); (b) an increased FRC-induced improvement in oxygenation (Fig. 5B); and (c) an augmented alveolar recruitment (reflected by the increased starting lung compliance) that induced improvement in both oxygenation (Figs. 4, 5C) and FRC (Fig. 5D). Our results on EELV/FRC and oxygenation/shunt fraction are consistent with those of two recent studies [9, 12], and partly support the notion of “increased EELV-induced improvement in oxygenation” [38].

LIP, PEEP, and oxygenation

With respect to PEEP optimization the significance of the LIP of the respiratory system PV curve is questionable; the LIP may be partly due to the shape of the chest wall PV curve [39]. In such cases the setting of PEEP at “LIP + 2 cmH2O” may not improve oxygenation [39]. In contrast, the LIP of the lung PV curve may more accurately reflect changes in inflation-induced alveolar mechanical behavior, because it is not influenced by the chest wall mechanical abnormalities in ARDS [39]. These concepts are strongly supported by our results, because (a) the LIP pressures of the lung PV curves were different in all body postures, and the contribution of the chest wall PV curve to the respiratory system LIP was considerable in all postures (Table 3; Figs. 2, and 3); (b) the removal and subsequent restoration of PEEP had minor effects on gas exchange in the prone position (Table 4), where the chest wall contribution to the respiratory system LIP was maximized; and (c) the postural change in the LIP pressure of the lung PV curve was the sole independent predictor of the postural change in oxygenation.

Gas exchange, body posture, and PEEP

In prone position PaO2/FIO2 was affected only arithmetically by PEEP. However, the use of PEEP resulted in augmentation of the pronation-induced reduction in shunt fraction, thus being consistent with a recently reported additive effect of PEEP and prone position on true pulmonary shunt [40]. Lastly, the postprone semirecumbent results on gas exchange and lung volumes clearly show that (a) PEEP is necessary for the maintenance of prone position benefits [9], (b) pronation benefits are lost within 14 min of recruitment-interrupted, zero PEEP ventilation, and (c) once lost, these favorable changes cannot be regained by restoring PEEP.

Methodological considerations

The measurements performed in the present study exhibited a limited variability (see T.S1–T.S3 of the Electronic Supplementary Material). A likely source was the manual determination of the pressure-volume coordinates of LIPs and UIPs. The variability in these measurements (also performed with the Measure Tool) most likely resulted in the observed, “roughly stable,” levels of consistency (e.g., 0.24–0.40 cmH2O for pressure measurements; see Electronic Supplementary Material). In our opinion, these consistency limits are acceptable for clinical measurements, thus supporting the combined use of widely available computer programs (SPSS and Photoshop) for PV curve analysis. PV curves were constructed only at zero PEEP, mainly because our objective was to determine the effects of body posture alone (i.e., at zero PEEP conditions) on PV curve variables (e.g., LIP and UIP), according to which ventilator settings could be adjusted [1, 2, 3, 4, 6, 7, 8, 9].

Ethical considerations

In ARDS the prolonged use of zero PEEP is potentially deleterious. In each posture in the present study, the continuous use of zero PEEP was minimized over two consecutive periods of 5 and 9 min, each of which was followed by recruitment maneuver(s) (Fig. 1). Furthermore, despite setting PEEP at 0 cmH2O total PEEP of the respiratory system (determined just prior to PV curve construction) actually ranged within 2.0–5.7 cmH2O. Total PEEP was due to patient intrinsic PEEP at the employed ventilatory settings.

Conclusions

In early, severe, and diffuse ARDS, prone position substantially attenuates the LIP and displaces toward higher inflation volume the UIP of the lung PV curve. The pronation-induced change in the LIP of the lung PV curve explains the postural improvement in arterial oxygenation. The latter is also associated with an increased FRC.