The Impact of Spontaneous Ventilation on Distribution of Lung Aeration in Patients with Acute Respiratory Distress Syndrome: Airway Pressure Release Ventilation Versus Pressure Support Ventilation

Abstract

BACKGROUND: 

In this study, we sought to determine which mode, airway pressure release ventilation (APRV) or pressure support ventilation (PSV), decreases atelectasis more in patients with acute lung injury/acute respiratory distress syndrome (ARDS).

METHODS: 

This was a retrospective study in the intensive care unit. Between 2006 and 2007, we identified 18 patients with acute lung injury/ARDS who received either APRV or PSV and had a helical computed tomography scan twice in 3 days.

RESULTS: 

Computed tomography data from the APRV and PSV groups (n = 9 each) were analyzed for 3-dimensional reconstruction and volumetry. Aerated lung regions (normally aerated, poorly aerated, nonaerated, and hyperinflated) were identified by their densities in Hounsfield units. The Pao₂/Fio₂ ratio and alveolar-arteriolar oxygen gradient after ventilation were improved in both groups (P = 0.008); however, the improvements in the APRV group exceeded those in the PSV group when delivered with equal mean airway pressure (P = 0.018 and 0.015, respectively). Atelectasis decreased significantly from 41% (range, 17%–68%) to 19% (range, 6%–40%) (P = 0.008) and normally aerated volume increased significantly from 29% (range, 13%–41%) to 43% (range, 25%–56%) (P = 0.008) in the APRV group, whereas lung volume did not change in the PSV group.

CONCLUSIONS: 

Spontaneous ventilation during APRV improves lung aeration by decreasing atelectasis. PSV for gas exchange is effective but not sufficient to improve lung aeration. These results indicate that APRV is more efficient than PSV as a mode of primary ventilatory support to decrease atelectasis in patients with ARDS.

Acute lung injury (ALI)/acute respiratory distress syndrome (ARDS) causes alveolar collapse primarily in dependent lung regions leading to a mismatch of ventilation-perfusion (V/Q) and intrapulmonary shunting.¹ The preservation of spontaneous ventilation has been recognized as critical to improving V/Q matching and intrapulmonary shunting.^2,3 During spontaneous breathing, posterior muscular sections of the diaphragm are able to move more than the anterior tendon plate when there is sufficient diaphragmatic contraction.^4,5 Compared with controlled mechanical ventilation, this diaphragmatic contraction will mainly lead to redistribution of gas to dependent, well-perfused lung regions,^6,7 resulting in atelectasis improvement.^8–10 Therefore, partial ventilatory support that preserves spontaneous ventilation is increasingly used as a primary ventilatory mode in patients with ALI/ARDS.^2,11–14

There are 2 alternatives often available in the intensive care unit to maintain sufficient spontaneous ventilation in mechanically ventilated patients: airway pressure release ventilation (APRV) and pressure support ventilation (PSV). In an oleic acid lung injury model^15–18 and in patients with ARDS,^2,11,12 the effects of spontaneous ventilation on gas exchange during APRV have been well investigated and have included comparisons with controlled mechanical ventilation, APRV without spontaneous ventilation, PSV, and synchronized intermittent mandatory ventilation. However, computed tomography (CT) has not yet been used to evaluate differences in spontaneous ventilation and changes in lung aeration when APRV and PSV are used in a clinical setting in patients with ALI/ARDS.

The aim of our retrospective study was to evaluate the impact of spontaneous ventilation during APRV and PSV on distribution of lung aeration in patients with ALI/ARDS by analyzing volumetric quantification of lung aeration and collapse.

METHODS

All mechanically ventilated patients with ALI/ARDS admitted to the 6-bed intensive care unit of Osaka City General Hospital, Osaka, Japan (1063 beds) from December 2006 to November 2007 were identified retrospectively. ALI/ARDS was defined according to the criteria of the American-European Consensus Conference on ARDS.¹⁹ We included the following patients from this group: 1) those who had met the ALI/ARDS criteria within the past 3 days, 2) those who had thoracic helical CT twice within 3 days (baseline and follow-up), and 3) those who received APRV or PSV continuously during the study period. The exclusion criteria were a history of severe chronic lung, heart, hepatic, and/or renal diseases, malignancies, immunosuppressive drugs, age <18 yr, and “Do Not Resuscitate.”

The ethics committee approved the retrospective analysis of CT data, and informed consent was obtained from the patients or their next of kin. Organ failure was defined with a scoring system described by Knaus et al.²⁰ Severity of illness was evaluated by the Sequential Organ Failure Assessment (SOFA) score.²¹ Organ failure and the SOFA score were calculated from the medical charts at the time the first CT scans were taken.

Study Design

Eighteen patients with ALI/ARDS were included in this study (APRV: n = 9, PSV: n = 9). None of the patients had received neuromuscular blocking drugs. APRV was provided with a demand valve continuous positive airway pressure circuit of a standard ventilator (Evita, Dräger Medical AG & Co., Lübeck, Germany), which made it possible to maintain spontaneous ventilation throughout the ventilatory cycle at 2 airway pressure levels. In addition to fraction of inspired oxygen (Fio₂) and slope of the pressure curve, we controlled the following 4 variables: P_high, P_low, T_high, and T_low.¹⁴P_high was set at no more than 30 cm H₂O; P_low was fixed at 0 cm H₂O; T_high was arbitrarily set at 4 s; and T_low was adjusted to maintain peak expiratory flow rate termination at 50%–75% to generate end-release pressure and minimize derecruitment. Fio₂ was adjusted to maintain Pao₂ above 60 mm Hg. The imposition of PSV above P_high was not used, and we focused on adequate tidal volume (<6 mL/kg) as a target to titrate P_high. Patients’ lungs in the PSV group were ventilated with a Servo-I ventilator (Maquet, Bridgewater, NJ). Ventilator settings in the PSV group were adjusted so that tidal volume was >4 mL/kg and <6 mL/kg and respiratory rate (RR) was <35/min by titrating PSV level and sedatives, and positive end-expiratory pressure (PEEP) and Fio₂ were set according to the PEEP/Fio₂ titration table.²² Inspiratory effort triggered an insufflation at −2 cm H₂O of trigger sensitivity. The preset ventilatory mode was not changed until a follow-up CT was read.

Helical CT Scans

All patients enrolled in this study received multidetector row helical CT scans with an Aquilion 8 (Toshiba Medical Systems, Tokyo, Japan) at baseline and again within 3 days after ventilation (follow-up). Scanning variables of the unenhanced CT scans were a voltage of 120 kV, a tube current of 100–400 mA adjusted by automatic exposure control, 0.5 s per rotation resulting in an acquisition time of 12–15 s, a detector row beam collimation of 2 mm, and a beam pitch of 0.875. Images reconstructed using the FC 14 algorithm were transferred to and analyzed with an AZE Virtual Place (AZE, Tokyo, Japan) for segmentation, 3-dimensional reconstruction, and volumetry of CT data. The segmentation with a region-growing algorithm was performed semiautomatically. The entire lung was identified by manually drawing the external boundaries of the lungs at the inside of the ribs and the internal boundaries along the mediastinal organs. Differently aerated lung regions were identified by their densities in Hounsfield units (HU). The lung regions were classified into 4 categories by their densities, as described previously²³: hyperinflated lung regions, −1000 to −900 HU; normally aerated lung regions, −900 to −500 HU; poorly aerated lung regions, −500 to −100 HU; and nonaerated lung regions, −100 to +100 HU. To certify reproducibility, an intensivist and a respiratory physician twice created 3-dimensional images with manual editing for volumetry.

Measurement Variables

We assessed the following variables from medical charts. Mean arterial blood pressure (MAP), heart rate, central venous pressure, number of pressors used, and lactate acid were used to quantify hemodynamics; peak airway pressure (P_{aw, peak}), mean airway pressure (P_{aw, mean}), PEEP, PSV level, RR, minute ventilation (MV), alveolar-arteriolar oxygen gradient (AaDO₂), Paco₂, and Pao₂/Fio₂ (P/F) ratio were used as measures of pulmonary function; and the number of extrapulmonary organ failures and SOFA score considered measures of severity of illness. Fio₂ was set at 1.0 at least 15 min before arterial blood gases were measured using the ABL 725 (Radiometer, Copenhagen, Denmark).

Statistical Analysis

Statistical analyses were performed using a statistical software package (ystat2004.xls, Igakutosho Shuppan, Tokyo, Japan). Results are expressed as median and range. In the figures, the boxes extend from the 25th to the 75th percentiles, the horizontal line indicates the median, and the upper and lower lines indicate extreme values. Differences between groups were evaluated with the Mann-Whitney U-test, and changes within groups were evaluated with the Wilcoxon’s signed rank test, because normal distribution could not be assured. The χ² or Fisher’s exact test was used to compare proportions. All tests were 2-tailed and differences were considered to be statistically significant at P < 0.05.

RESULTS

Patient Characteristics

Eighteen patients with ALI/ARDS were analyzed: the APRV group (n = 9) and the PSV group (n = 9). The baseline demographics, etiology, and severity of lung injury are summarized in Table 1. Severity of lung injury and severity of illness were similar in both groups. The high proportion of P/F ratio <100 (APRV 55.6%, PSV 66.7%) indicated severe lung injury. The time span between baseline and follow-up CT scans was similar (APRV 2 days [range, 1–3], PSV 2 days [range, 1–3]; P = 0.794).

Table 1: Demographic Data and Clinical Characteristics (n = 18)

Ventilatory and Hemodynamic Variables

The ventilatory variables, other than the P/F ratio and AaDO₂, are shown in Table 2. The P_{aw, peak} and P_{aw, mean} in both groups were moderate, and there were no significant differences during the study periods between the ventilatory modes. There were no significant differences in Paco₂ and in MV at baseline and follow-up between the groups, whereas during PSV, RR increased significantly to maintain alveolar ventilation at follow-up (P = 0.006). PEEP and tidal volume (calculated using MV and RR) in the PSV group were adequate and in accordance with the Acute Respiratory Distress Syndrome Network ventilator strategy.²³ The changes in the P/F ratio and AaDO₂ are shown in Figures 1 and 2. Although arterial oxygenation increased significantly during the study period in both groups (APRV 79 [range, 40–253] mm Hg to 398 [range, 191–502] mm Hg, P = 0.008; PSV 96 [range, 66–123] mm Hg to 249 [range, 112–436] mm Hg, P = 0.008), the improvement in the P/F ratio in the APRV group exceeded that in the PSV group when delivered with equal airway pressure (APRV 398 [range, 191–502] mm Hg, PSV 249 [range, 112–436] mm Hg; P = 0.018). Similarly, AaDO₂ improved significantly during the study period in both groups (APRV 546 [range, 390–607] mm Hg to 262 [range, 139–459] mm Hg, P = 0.008; PSV 553 [range, 535–599] mm Hg to 409 [range, 231–549] mm Hg, P = 0.008). However, the improvement in AaDO₂ in the APRV group exceeded that in the PSV group when delivered with equal airway pressure (APRV 262 [range, 139–459] mm Hg, PSV 409 [range, 231–549] mm Hg, P = 0.015).

Table 2: Ventilatory Variables Except Pao₂/Fio₂ Ratio and AaDO₂

Figure 1.: Changes in the Pao₂/Fio₂ (P/F) ratio in the airway pressure release ventilation (APRV) group and the pressure support ventilation (PSV) group. The P/F ratio in the APRV group is represented by white boxes and the ratio in the PSV group by gray boxes (§P < 0.01 compared with baseline; *P < 0.05 APRV versus PSV).

Figure 2.: Changes of alveolar-arteriolar oxygen gradient (AaDO₂) in the airway pressure release ventilation (APRV) group and the pressure support ventilation (PSV) group. AaDO₂ in the APRV group is represented by white boxes and AaDO₂ in the PSV group is represented by gray boxes (§P < 0.01 compared with baseline; *P < 0.05 APRV versus PSV).

The hemodynamic variables are summarized in Table 3. Although the number of vasopressors used was not different between the groups at baseline or follow-up, APRV increased MAP significantly at follow-up (P = 0.018). No differences were observed in central venous pressure, heart rate, or lactate acid between the groups.

Table 3: Hemodynamic Variables

Volumetric Analysis of Lung Aeration

The volumetric analyses of differently aerated lung regions in both groups are presented in Figure 3. In the APRV group, improvement of lung aeration by decreasing the collapsed tissue was clearly observed. A significant gain in normally aerated lung volumes from 29% (range, 13%–41%) to 43% (range, 25%–56%) was observed (P = 0.008), whereas nonaerated lung volumes significantly decreased from 41% (range, 17%–68%) to 19% (range, 6%–40%) (P = 0.008). Neither poorly aerated lung volumes (23% [range, 12%–45%] to 27% [13%–43%], P = 0.520) nor hyperinflated lung volumes (2.4% [range, 0.2%–18%] to 2.2% [range, 0.2%–36%], P = 0.490) changed during the study period in the APRV group. Likewise, there were no significant changes in any differently aerated lung volumes in the PSV group: hyperinflated lung volumes 2.6% (range, 0.4%–41%) to 2.2% (range, 0.3%–23%) (P = 0.725), normally aerated lung volumes 39% (range, 27%–70%) to 44% (range, 13%–63%) (P = 0.445), poorly aerated lung volumes 17% (range, 12%–23%) to 16% (range, 12%–30%) (P = 0.953), and nonaerated lung volumes 39% (range, 9%–49%) to 29% (range, 17%–68%) (P = 0.379). An example of a CT scan image after manual drawing and reconstruction using an AZE Virtual Place is presented in Figure 4.

Figure 3.: A, Distribution of hyperinflated, normally aerated, poorly aerated, and nonaerated lung volume, defined by helical computed tomography scans of total lung in the airway pressure release ventilation (APRV) group. B, Distribution in the pressure support ventilation (PSV) group. Baseline and follow-up values are compared for each aerated lung volume. Baseline is shown by white boxes and follow-up by gray boxes (*P < 0.01 baseline versus follow-up).

Figure 4.: A representative 3-dimensional reconstructed computed tomography image using an AZE Virtual Place. Nonaerated lung regions are colored dark gray. These images clearly show that atelectasis occurred predominantly in dependent lung regions.

DISCUSSION

In a clinical setting in patients with ALI/ARDS, this study was designed to evaluate the impact of 2 different partial ventilatory modes on distribution of lung aeration and gas exchange using analysis for volumetric quantification of lung aeration and collapse. The study revealed the following results: 1) pulmonary oxygenation during spontaneous ventilation with APRV was better than during PSV when delivered with equal mean airway pressure, 2) spontaneous ventilation with APRV improved lung aeration by decreasing the amount of collapsed tissue in a very short time, and 3) pulmonary oxygenation improved but PSV had no impact on lung aeration. These results suggest that spontaneous ventilation with APRV decreased atelectasis more than ventilation with PSV and led to better V/Q matching and pulmonary oxygenation.

In this study, it seems likely that the improvement in gas exchange in the APRV group was mainly caused by the ability of lung recruitment to reduce the amount of collapsed tissue. Previous animal studies support our observation.^17,18 Wrigge et al.¹⁷ have shown that 4 h of APRV with spontaneous ventilation resulted in improved oxygenation, higher end-expiratory lung volume, and less nonaerated tissue in diaphragmatic slices compared with APRV without spontaneous ventilation.

The conceptual advantage of APRV is thought to come from substantial mean airway pressure and preservation of spontaneous ventilation,²⁴ whereas numerous studies have shown that APRV without spontaneous ventilation did not affect gas exchange^2,15–18 and was not different from conventional pressure-controlled MV.^2,14,17 In this study, airway pressures in both groups were similar and moderate, which means that the advantage of APRV seems to come from preservation of spontaneous ventilation rather than substantial mean airway pressure. During APRV, patients can breathe spontaneously without a trigger and support in any phase, and can control frequency and duration of spontaneous inspiration and expiration,¹⁴ which is a unique mechanism that makes it possible to maintain a sinusoidal flow pattern similar to normal spontaneous breathing and to generate full diaphragmatic contraction. Therefore, the results showing that APRV improved lung aeration by decreasing atelectasis are mainly attributable to the above-mentioned spontaneous ventilation benefits.

We have been searching for the optimal method to provide spontaneous ventilation activity in ALI/ARDS while maintaining lung-protective ventilation; therefore, we compared APRV and PSV as primary ventilatory strategies in this study. PSV has been used to wean patients from MV and is now increasingly used even for patients with acute respiratory failure.^13,25 Despite the fact that PSV has proven to have various benefits,^26–30 its effectiveness in patients with ALI/ARDS remains controversial.^{2,8,13,31–33} In our study, in agreement with some previous studies,^25,33 improvements in the P/F ratio and AaDO₂ in PSV were clearly observed; however, improvements did not exceed those in APRV with spontaneous ventilation, and no impact on lung aeration was observed even when equal mean airway pressure was delivered. Considering these results, although PSV used with the Acute Respiratory Distress Syndrome Network ventilator strategy might possibly be beneficial for patients with ALI/ARDS, apparently it is not the best use of the spontaneous ventilation benefits to recruit collapsed alveoli. These effects may be explained by a different mechanism.

First, PSV produces a decelerating gas flow pattern, not sinusoidal gas flow, because the gas flow and patient effort do not follow a similar time course.¹⁴ Gas distribution during PSV is similar to an assisted breath rather than a spontaneous breath. As a result, PSV reduces the spontaneous ventilation benefits. Second, Uchiyama et al.³⁴ have shown that the lower the PSV level, the higher the diaphragmatic activity in an animal model. Because the PSV levels applied in previous studies^{2,13,16,25,33} were set from 10 to 20 cm H₂O, they were too high to cause diaphragmatic contractions sufficient to provide spontaneous ventilation benefits, which result in patient-triggered pressure-controlled MV rather than PSV. In addition to moderate PEEP, the moderate PSV level in this study compared with these previous studies might lead to a beneficial effect on gas exchange.

Accordingly, we believe, when comparing unsupported spontaneous ventilation (APRV) with assisted spontaneous ventilation (PSV) under equal mean airway pressure, that improving lung aeration by decreasing atelectasis depends above all on the amount of diaphragmatic activity, which is observed to be highest in unsupported spontaneous ventilation. This effect, however, is observed only when sufficient positive pressure is maintained.

In this study, APRV with spontaneous ventilation was associated with an increase in MAP compared with PSV. Previous studies also showed that spontaneous ventilation with APRV increased cardiac index, right ventricular end-systolic volume index, and right ventricular ejection fraction compared with PSV and APRV without spontaneous ventilation.^2,35 Habashi¹⁴ reported that during spontaneous ventilation with APRV, sufficient diaphragmatic contraction resulted in compressing abdominal viscera propelling blood into the inferior vena cava, improving hemodynamic effects. During PSV, weakened diaphragmatic activity caused by assisted support may play an important role in limiting venous return and MAP.

This study had several limitations. First, the study was limited to patients with early ALI/ARDS who had repeat CT scans within 3 days and who received APRV or PSV continuously during the study period. This explains why the number of patients was small. When evaluating the impact of different ventilatory strategies on lung structure, it is important to select patients manifesting pathologically identical phases of ARDS. Because of the minimizing factors of lung recovery, except for ventilatory strategies such as antibiotic effect and spontaneous recovery, to the extent that it was possible, we selected only patients with a short time span between CT scans for this study. Another limitation was related to the CT-based analyses of lung aeration. The partial volume effect may have caused overestimation of poorly aerated lung volumes and underestimation of hyperinflated lung volumes if a voxel contained elements of highly different densities. The final limitation was that there was no assessment of intrinsic PEEP (PEEPi) and transpulmonary pressure (P_L), by measuring esophageal pressure (P_es). We used a fixed P_low of 0 cm H₂O because it was simple to control end-release pressure/volume (PEEPi) by 1 parameter (T_low). A P_low of 0 cm H₂O accelerated peak expiratory flow rate, concluding the release phase earlier and enabling the P_high phase to be resumed earlier in the cycle, which leads to the production of less spontaneous ventilation variability and higher mean airway pressure. Because this approach was based on a previous review,¹⁴ we assume that our ventilator settings for APRV were adequate; however, we could not evaluate the possible effects of PEEPi on our results. The measurement of P_es provided us not only with the amount of diaphragmatic activity but also with the estimation of P_L. It seems important to emphasize that intrathoracic inflation pressure generated by diaphragmatic activity could add alveolar-stretching pressure to P_high. For example, when P_high is set at 30 cm H₂O according to a lung-protective strategy and the patient generates an additional 10 cm H₂O through spontaneous effort, P_L is 40 cm H₂O, which could be injurious. In future studies, therefore, P_L should be analyzed using P_es and in APRV settings; not only P_high but also P_L should be set at no more than 30 cm H₂O.

In conclusion, our results demonstrate that spontaneous ventilation during APRV improves lung aeration by decreasing atelectasis in a short time, which contributes to improved pulmonary oxygenation. PSV has a good effect on gas exchange; however, it is apparently not sufficient to improve lung aeration. These results indicate that APRV with spontaneous ventilation is more reasonable than PSV as a mode of primary ventilatory support to decrease atelectasis in patients with ALI/ARDS.