Effect of Mechanical Ventilation on Inflammatory Mediators in Patients With Acute Respiratory Distress Syndrome: Title and subTitle BreakA Randomized Controlled Trial FREE

Acute respiratory distress syndrome (ARDS) is a clinical syndrome characterized by severe hypoxemia, stiff lungs, and decreased respiratory system compliance. Early ARDS is characterized by acute and diffuse endothelial and epithelial injury termed diffuse alveolar damage,¹ which leads to increased vascular permeability with protein-rich exudative edema. Although originally thought to be relatively homogeneous, a number of recent studies have highlighted the marked heterogeneity of the pathological process with consolidation in the dependent regions of the lung and relatively normal aeration of the nondependent regions.^{2 - 3}

Despite apparent improvement in management and outcome of ARDS, the mortality rate of ARDS remains high, ranging from 35% to 65%.^{4 - 5} Mechanical ventilation delays mortality in many patients with acute respiratory failure and is used to maintain adequate systemic oxygenation and to rest the respiratory muscles. However, over the last 2 decades, it has become evident that mechanical ventilation itself can augment or cause acute lung injury. This has been demonstrated in animal studies^{6 - 9} and has been highlighted in a recent randomized trial, which found that the mortality rate from ARDS could be reduced by 40% when a lung-protective strategy was used compared with a conventional strategy of mechanical ventilation.¹⁰ The lung-protective strategy in this trial involved limiting peak airway pressures or tidal volumes and using positive end-expiratory pressure (PEEP) levels that were individualized to each patient based on respiratory system mechanics, as assessed by the volume-pressure curve.^{10 - 11} One reason for the efficacy of this approach may have been the decreased stress on the lung since Mead et al¹² have predicted that in a nonuniformly inflated lung, shear forces due to expansion of collapsed alveolar regions surrounded by expanded regions could exceed 100 cm H₂O even though the transpulmonary pressure may only be 30 cm H₂O.

ARDS is an inflammatory disease and clinical studies have suggested increased mortality in patients who continue to manifest elevated cytokine levels during their clinical course.^{13 - 16} Recent experimental studies in various animal models have provided 3 lines of evidence suggesting that mechanical ventilation can initiate or exacerbate an inflammatory response: (1) pathologic evidence of neutrophil infiltration,^{17 - 18} (2) increased cytokine levels in lung lavage,⁸ and (3) increased cytokine levels in the systemic circulation.^{9 ,19} In addition, ventilatory strategies that were designed to minimize ventilator-induced lung injury (low end-inspiratory lung volumes or high end-expiratory lung volumes) in these animal studies were associated with strikingly lower cytokine levels.⁸ To examine the influence of mechanical ventilation on lung and systemic cytokine levels in patients with ARDS, we compared a ventilatory strategy designed to minimize ventilator-induced lung injury (high PEEP, low end-inspiratory stretch) with a conventional ventilatory strategy.

Following approval of institutional review boards and after obtaining informed consent from the patient or next of kin, patients were recruited from the intensive care units of the university hospitals of Bari, Italy, and Geneva, Switzerland. Inclusion criteria were (1) age of 18 years or older and (2) diagnosis of ARDS based on American-European Consensus Conference criteria.²⁰ Exclusion criteria were (1) anticipated to require mechanical ventilation less than 48 hours (determined by the attending physician), (2) more than 8 hours of mechanical ventilation prior to admission to the study, (3) cardiogenic pulmonary edema (clinically suspected, or pulmonary artery occlusion pressure >18 mm Hg), (4) history of ventricular fibrillation or tachyarrhythmia, unstable angina, or myocardial infarction within preceding month, (5) preexisting chronic obstructive pulmonary disease, (6) major chest wall abnormalities (kyphoscoliosis, open or flail chest), chest tube with persistent air leak, or abdominal distention, (7) pregnancy, (8) known intracranial abnormality, and (9) enrollment in another interventional study.

Patients were sedated (0.01 mg/kg of fentanyl citrate and 5-20 mg of diazepam), paralyzed (4-8 mg of pancuronium bromide), and ventilated for 1 to 2 hours with a fixed ventilatory management protocol consisting of volume-targeted control mechanical ventilation: PEEP of 10 cm H₂O, fraction of inspired oxygen (FIO₂) of 100%, tidal volume of 5 to 8 mL/kg (ideal body weight), and inspiratory to expiratory ratio of 1:2. The PEEP was then removed and a volume-pressure curve was measured as described below. The volume-pressure curve of the respiratory system of patients with early ARDS often has a characteristic sigmoidal shape with a lower inflection point (P_flex) thought to approximate the pressure required to reopen collapsed lung regions and an upper inflection point (UIP) thought to correspond to the pressure at which overdistention of some lung units occurs.³ The PEEP was then restored and bronchoalveolar lavage fluid and blood samples were taken 20 to 30 minutes later. A full set of laboratory, hemodynamic, and respiratory variables was obtained. The Acute Physiology and Chronic Health Evaluation II²¹ and lung injury scores²² (for both scores, higher values indicate greater illness severity) and the number of organ failures as defined by Knaus et al²³ were also calculated. At the end of this fixed ventilatory period, patients were randomly assigned to a control group or a lung-protective strategy group. The patients were assigned by a concealed allocation approach using opaque sealed envelopes containing the randomization schedule. Baseline blood gases were then measured 2 to 3 hours after the assigned ventilatory strategy was initiated.

Mechanical ventilation was performed using control mechanical ventilation with a respiratory rate of 10 to 15/min, an inspiratory to expiratory ratio of 1:2, and a tidal volume targeted to maintain the PaCO₂ between 35 and 40 mm Hg. For safety reasons, when a plateau airway pressure of 35 cm H₂O was reached, the tidal volume was not increased further, irrespective of PaCO₂.²⁴ A PEEP trial on 100% FIO₂ was performed using incremental (3-5 cm H₂O) levels from 3 to 15 cm H₂O to determine the PEEP level that produced the greatest improvement in arterial oxygen percent saturation (SaO₂) without worsening hemodynamics (>10% drop in mean blood pressure). The FIO₂ was then decreased until SaO₂ was decreased by 1% to 2% or below 90%.

The following aspects of the ventilatory strategy were used as in the control group: control mechanical ventilation, respiratory rate of 10 to 15/min, inspiratory to expiratory ratio of 1:2, and FIO₂ of 100%. The tidal volume and PEEP values were set to minimize stress on the lung. Tidal volume was set to obtain a value of plateau pressure (P_plat) less than pressure at the UIP regardless of PaCO₂. The PEEP was set at 2 to 3 cm H₂O higher than the pressure at P_flex. If UIP and P_flex could not be determined on the volume-pressure curve, a tidal volume of 5 to 8 mL/kg (ideal body weight) and a PEEP_total level of 15 cm H₂O were applied, respectively.¹⁰ If arterial pH was less than 7.15, tidal volume was increased until a P_plat of 35 cm H₂O was reached or until arterial pH was more than 7.15 mm Hg. The FIO₂ was then decreased until the SaO₂ fell by 1% to 2% or below 90%.

After admission, protocol withdrawal could occur if 1 of the following a priori conditions occurred within the first 24 hours: (1) a 20% increase or decrease in Acute Physiology and Chronic Health Evaluation II²¹ and lung injury scores,²² (2) an increase or decrease in the initial number of organ failures,²³ (3) need for high levels of epinephrine (>0.15 mg/kg per minute) or norepinephrine (>0.1 mg/kg per minute), (4) the patient was ready to be weaned from mechanical ventilation, as determined by the attending intensivist, and (5) high risk of death defined as hemodynamic instability, arrhythmia, or hypoxemia, not responsive to 20 to 30 minutes of standard treatment. Additional ARDS cointerventions (inhaled nitric oxide or prostacycline, administration of almitrine or steroids, prone position) were not allowed in either group during the study. Patients were cared for by attending physicians not involved in the protocol and other management decisions were made at their discretion. All measurements were repeated 24 to 30 hours and 36 to 40 hours after randomization. Sedation was maintained throughout the study.

The volume-pressure curve was constructed by plotting the different inflation volumes against the corresponding values of P_plat as assessed by end-inspiratory occlusion at different tidal volumes.³ Each occlusion was maintained until a stable P_plat was observed. Total PEEP (applied PEEP plus auto-PEEP) was measured by performing an end-expiratory occlusion³ (Servo 300, Siemens-Elema, Stockholm, Sweden). Flow (pneumotachograph, Fleisch, Lausanne, Switzerland), pressure (Validyne pressure transducer, Northridge, Calif), and volume (digital integration of flow signal, Anadat software package, Montreal, Quebec) were measured using standard techniques described previously.³ The P_flex and UIP on the volume-pressure curve were quantified using a computer step-by-step regression analysis.³

Blind bronchoalveolar lavage was performed using a telescoping catheter (Ballard, Draper, Utah) with 2 aliquots of 40 to 50 mL of sterile isotonic saline. Lavage with a third aliquot was performed if there was less than 30 to 40 mL of recovered fluid from the first 100 mL. When a diffuse infiltrate was seen on a chest x-ray, bronchoalveolar lavage was performed in the right lower or middle lobe. When an area of localized pulmonary infiltration was present, bronchoalveolar lavage was blindly performed in the lower lobe of the opposite lung.²⁵ The first aliquot was discarded^{13 - 15} and the remaining bronchoalveolar lavage fluid was rapidly filtered through sterile gauze and then spun at 4°C at 400g for 15 minutes. A microscopic cell count was performed on the cell pellet using standard techniques.²⁵ The supernatant was centrifuged at 80,000g for 30 minutes at 4°C to remove the surfactant-rich fraction and then concentrated 10-fold on a 5000 molecular weight cut-off filter (Amicon, Beverly, Mass) under nitrogen. The concentrated supernatant was then frozen at −70°C. Blood samples (20 mL) obtained from a central venous line were placed in a specimen tube containing heparin, centrifuged at 1500g for 10 minutes, and then the plasma was aspirated and stored at −70°C. All cytokine determinations on the bronchoalveolar lavage fluid and plasma were carried out in duplicate in Geneva, Switzerland (with the technician blinded to ventilation strategy) using a solid-phase enzyme-linked immunosorbent assay method based on the quantitative immunometric sandwich enzyme immunoassay technique.²⁵ Reagents for the various cytokines were obtained from several sources (tumor necrosis factor [TNF] α, soluble TNF-α receptors [TNF-αsR55 and TNF-αsR75], interleukin [IL] 6, and IL-8 from Medgenix, Fleures, Belgium); IL-1β and IL-1 receptor antagonist [IL-1Ra] from Immunotech, Marseille, France).

In a post hoc fashion, we calculated the number of ventilator-free days (days without mechanical ventilation after extubation) during the 28 days immediately after study entry.²⁶ Results were scored from 0 (worst outcome) to 28. The number of patients who were alive at 28 days was also recorded.

The values for the cytokine concentrations were not normally distributed so we performed log₁₀ transformations to normalize the data to permit the application of parametric statistics. To evaluate differences over time of cytokine values within each group, repeated measures analysis of variance by Bonferroni method was used. To evaluate differences between the 2 groups, the Fisher exact test for categorical variables, the t test with unequal variance for continuous variables, and the Mann-Whitney rank sum test for ordinal variables were used. All tests of significance were 2-tailed, and P<.05 was accepted as significant.

From November 1995 to February 1998, 44 patients were enrolled (34 in Bari and 10 in Geneva). Seven protocol patients withdrew because of an increase in the initial number of organ failures (1 in each group), need for high doses of norepinephrine (1 in each group), high risk of death (1 in the lung-protective strategy group), and initiation of weaning from mechanical ventilation (1 in each group). Baseline characteristics, underlying condition responsible for ARDS, and respiratory mechanics on entry are presented in Table 1. Most patients had sepsis and multiple trauma as the underlying condition responsible for ARDS. The P_flex was measurable in all patients, while the UIP was not measurable in 10 patients (4 in the control group and 6 in the lung-protective strategy group). There was no significant difference in P_flex or UIP between groups (Table 1).

Values of ventilator settings and of arterial blood gases 2 to 3 hours after randomization are presented in Table 2. In the control group, PEEP levels were lower than P_flex in every patient and values of P_plat were higher than UIP in all but 4 patients. By design, PEEP levels were greater than P_flex and P_plat levels were lower than UIP in every patient in the lung-protective strategy group. The PaO₂ did not differ between groups, although FIO₂ was slightly higher in the control group. A significant increase in PaCO₂ (permissive hypercapnia) was observed in the lung-protective strategy group.

With the exception of bronchoalveolar lavage concentrations of TNF-αsR75 (lower in the control group, P<.05; Table 3) and plasma values of IL-6 (lower in the control group, P<.01), values of inflammatory mediators prior to randomization did not differ between groups (Figure 1 and Figure 2, Table 3).

There was a trend for polymorphonuclear cell levels in bronchoalveolar lavage to increase over time in the control group (Figure 1). In the control group, bronchoalveolar lavage fluid concentration of IL-1β (P<.001), TNF-α (P<.05), and IL-6 (P<.01) increased over time, as did plasma levels of TNF-α (P<.01) and IL-6 (P<.001) (Figure 1 and Figure 2). The concentrations of other inflammatory mediators did not change significantly. A significant (P<.01) increase over time of both TNF-α receptors in the bronchoalveolar lavage fluid and plasma concentrations and in the bronchoalveolar lavage IL-1Ra (P<.01) was observed in the control group (Table 3).

A significant reduction over time in bronchoalveolar lavage concentrations of polymorphonuclear cells (P<.001), IL-1β (P<.05), TNF-α (P<.001), IL-8 (P<.001), and IL-6 (P<.005), and in plasma concentrations of IL-6 (P<.002) was observed in the lung-protective strategy group (Figure 1 and Figure 2). Plasma concentrations of TNF-α and IL-8 did not change significantly over time in the lung-protective strategy group (Figure 2). A significant reduction over time of bronchoalveolar lavage concentrations of TNF-αsR55 (P<.001) and bronchoalveolar lavage and plasma concentrations of TNF-αsR75 (P<.001) and IL-1Ra (P<.05 and P<.001, respectively) were observed in the lung-protective strategy group (Table 3).

With the exception of IL-1Ra in the bronchoalveolar lavage, values of inflammatory mediators 36 to 40 hours after randomization were significantly (P<.05 to P<.001) lower in the lung-protective strategy group.

The post hoc analysis showed that the mean (SD) number of ventilator-free days²⁶ in the lung-protective strategy group was higher (P<.01) than in the control group (12 [11] vs 4 [8] days, respectively). Mortality rates at 28 days from admission were 38% and 58% in the lung-protective strategy and control groups, respectively (P=.19).

The major finding of this study is that mechanical ventilation—the therapeutic modality that is invariably used in the treatment of ARDS—can itself lead to an increase in cytokine levels in the lung, as well as in the systemic circulation. These results may partially explain the development of multiple organ failure in many patients with ARDS,²⁷ the high mortality rate of this syndrome (35%-65%),^{4 - 5} and perhaps the decrease in mortality observed in a recent study that used a lung-protective strategy.¹⁰

Over the past decade, numerous studies have suggested that mechanical ventilation can cause or exacerbate acute lung injury. This is particularly true in patients with ARDS because of the widespread, heterogeneous distribution of consolidated/atelectatic regions, which produces a small lung volume available for ventilation.² Using computed tomography, Gattinoni et al² showed that the lungs of patients with ARDS are highly asymmetrical along the vertical axis with a small nondependent lung region continuously open to ventilation, and a dependent consolidated, atelectatic region. In between, there is a region that can be recruited or derecruited depending on the particular ventilatory strategy used.² In such patients, mechanical ventilation could lead to injury due to overdistention as more of the tidal volume is distributed to the small, relatively normal alveolar regions^{2 ,7 ,28} and/or repeated recruitment or derecruitment of alveolar units that may be exacerbated with ventilation at low PEEP levels.^{2 ,29 - 30}

Recent in vitro studies using human alveolar macrophage cultures³¹ and ex vivo preparations of rat and mice lungs^{8 ,19} "stretched" with injurious ventilatory strategies have demonstrated an up-regulation of the cytokine response, which may also lead to a systemic cytokine response in vivo.^{9 ,19} The present study is the first in humans to demonstrate that mechanical ventilation per se may be an important factor in determining the pulmonary and systemic cytokine levels in patients with ARDS. Of particular importance, the lack of improvement in the cytokine response in the control group occurred using tidal volumes and PEEP levels commonly used in the treatment of patients with ARDS. Furthermore, our data indicate that bronchoalveolar lavage concentrations of neutrophils and various cytokines decreased significantly over time with a ventilatory strategy aimed at minimizing injury by maintaining the lung recruited throughout the respiratory cycle, and by avoiding overdistention.

Cytokines are a diverse group of inflammatory mediators (low-molecular-weight proteins) that are produced by numerous cell types, which initiate and orchestrate the host's response to different stresses such as bacteremia, shock, and thermal injury.³² Cytokines act on many target cells and can affect all organs to elicit physiological and biochemical responses to critical illness.³³ Cytokines interact with highly specific cell-surface receptors, causing a series of intracellular signaling events that typically result in de novo protein synthesis and the elaboration of other cytokines within the target cell. If these processes are not regulated, they may result in excessive amplification of the inflammatory cascade and the overproduction of proinflammatory mediators with the consequent uncontrolled activation of the immune system.^{32 - 33} This can lead to a number of clinical sequelae, including disseminated intravascular coagulation, renal insufficiency, acute pancreatitis, multiple system dysfunction syndrome (MODS), and ARDS.^{1 ,34} Indeed, there is evidence that in patients with MODS, the highest levels of cytokines in the blood are found downstream from the most affected organ.³⁵ The relevance of cytokines in ARDS is evident from human studies showing an increase in bronchoalveolar lavage IL-8 concentrations that precede clinical evidence of disease,³⁶ and the observation that bronchoalveolar lavage concentrations of TNF-α, IL-1β, IL-6, and IL-8 were higher at presentation in patients who died of ARDS, compared with those who survived.^{13 - 15 ,25} Concentrations of bronchoalveolar lavage cytokines remained elevated over time in nonsurvivors, suggesting that persistent elevation of polymorphonuclear cells and inflammatory cytokines may delay or preclude resolution of the pulmonary^{14 - 15} and systemic^{13 ,16} inflammatory processes. Recent data suggest that modulation of the inflammatory response by administration of methylprednisolone in patients with unresolving ARDS was associated with improvement in lung injury and MODS scores and reduced mortality.³⁷

In patients developing ARDS, only a small percentage die of hypoxia and/or hypercarbia.²⁷ Rather, lung injury appears to predispose patients to the development of a systemic inflammatory response that culminates in MODS and death.^{4 ,27} The classic explanation for the development of MODS in ARDS is the development of superinfection.²⁷ However, a number of recent studies have suggested another mechanism. In ARDS, the alveolar epithelial-endothelial barrier is disrupted,¹ and cytokines produced in the lungs may enter the systemic circulation.³⁸ This represents a potential mechanism leading to the subsequent development of MODS.^{35 ,38} It has been suggested that mechanical ventilation may play a major role in this regard³⁸ by causing an increase in production of pulmonary cytokines as well as by increasing alveolar capillary permeability that would increase the transfer of intrapulmonary cytokines from the alveolar or airway compartment to the systemic circulation. In addition, both high-peak pressures as well as the absence of PEEP have been shown to increase bacterial translocation from the lung into the bloodstream in animal models of intratracheal instillation of bacteria,³⁹ providing another mechanism by which mechanical ventilation can produce systemic manifestations. Our data suggest that, at least in some patients, the persistence of elevated lung lavage and systemic cytokine concentrations over 36 hours may be due in part to the ventilatory strategy used to treat the underlying respiratory failure, although it is possible that the somewhat higher FIO₂ levels in the control group may play a role in this regard.⁴⁰

Recommendations from a consensus conference on mechanical ventilation²⁴ suggested that in patients with ARDS, inflation pressure should ideally be maintained at less than 35 cm H₂O by decreasing tidal volume to as low as 5 mL/kg. The consensus conference also suggested that PEEP is useful in supporting oxygenation and may prevent lung damage. To verify the clinical impact of these recommendations, several randomized clinical trials have been conducted.^{10 ,41 - 42} Stewart et al⁴¹ found no difference in mortality in their comparison of low (6-7 mL/kg) and high (10-12 mL/kg) tidal volume. In that study,⁴¹ PEEP was set as the minimal value that achieved acceptable arterial oxygen saturation (89%-93%) with nontoxic FIO₂ values (≤0.5). Similar results were obtained by Brochard et al⁴² in a study in which they compared a strategy aimed at limiting P_plat to 25 cm H₂O vs a more conventional ventilatory approach using a mean tidal volume of 10.3 mL/kg; both arms used a similar level of PEEP. In another study, Amato et al¹⁰ tested the hypothesis that the limitation of end-inspiratory volume and the use of PEEP levels that would prevent end-expiratory collapse would decrease mortality in patients with ARDS. Ventilation settings in the protective ventilation group were defined according to the inspiratory volume-pressure curves and resulted in a tidal volume and PEEP levels of 4 to 6 mL/kg and 15 to 20 cm H₂O compared with 10 to 12 mL/kg and 6 to 8 cm H₂O in the control group, respectively. This strategy was associated with significantly improved survival.¹⁰ Given the correlation between the persistence of the pulmonary and systemic inflammatory response and outcome in patients with ARDS,^{13 - 15} our data may help explain these findings.¹⁰

In our study, the post hoc analysis of ventilator-free days²⁶ showed a significant reduction in the lung-protective strategy group, and mortality at 28 days after randomization tended to be lower in the lung-protective strategy group. These results were somewhat surprising to us since the protocol was designed to be a 36-hour study to assess cytokine levels. However, on termination of the study, there was a strong tendency for patients to be maintained on the ventilatory strategy that they had been assigned to in the experimental protocol until they were ready to be weaned. We should emphasize, however, that the study design did not include a strict weaning protocol for both groups and was not designed to assess these clinical end points; hence, these comparisons should be viewed as post hoc analyses, which must clearly be examined in appropriate randomized clinical trials.

In conclusion, we found that conventional mechanical ventilation of patients with ARDS is associated with a local and systemic cytokine response that is sustained over 36 hours; furthermore, this response may be attenuated by a ventilatory strategy designed to minimize ventilator-induced lung injury. Additional studies, assessing optimum ventilatory strategies, as well as other novel approaches specifically targeting the inflammatory response, are required to evaluate their effect on physiological as well as clinical end points in patients with ARDS requiring mechanical ventilation.