Introduction
Mechanical ventilation is a life-saving strategy in patients with acute respiratory failure. Nevertheless, unequivocal evidence from both experimental and clinical studies indicates that mechanical ventilation has the potential to aggravate lung injury [
1‐
3]. Data from three randomized controlled clinical trials confirmed the existence of ventilator-associated lung injury in patients with acute lung injury (ALI) or its more-severe form, acute respiratory distress syndrome (ARDS), by showing reduced morbidity and mortality in the lower tidal-volume arm [
4‐
6]. As a result of these studies, current guidelines now clearly support the use of lower tidal volume in patients with ALI/ARDS [
7]. In contrast, little evidence supports the use of lower tidal volume in critically ill patients without ALI/ARDS, partly because of a lack of randomized controlled trial evidence on the best ventilator strategies in these patients [
8].
Pneumonia, aspiration, sepsis, trauma, shock, and multiple blood transfusions are well-described risk factors for ALI/ARDS [
9]. Animal studies demonstrate that mechanical ventilation with conventional tidal volume not only may aggravate, but also may initiate lung injury [
1,
2]. The role of tidal-volume size as a contributor to the development of lung injury in humans is, however, less clear. One study on reduced tidal volume and pressure settings did not show a reduction in mortality but suggested more side effects of lower tidal-volume ventilation in patients at high risk for ALI/ARDS [
10]. Conversely, pulmonary procoagulant changes and increased systemic cytokine production were observed in patients without preexisting lung injury receiving conventional-tidal-volume mechanical ventilation during surgery [
11,
12].
Other studies have challenged these findings [
13,
14]. Results from cohort studies suggest that mechanical ventilation with conventional tidal volumes may cause or contribute to development of lung injury in critically ill patients who did not have ALI/ARDS at the onset of mechanical ventilation [
15,
16]. The inconclusive results from the studies in surgical patients may arise from the fact that these patients were ventilated for only a short period, whereas the patients of the larger cohort studies were critically ill patients who had been ventilated for a longer period.
As ALI/ARDS is characterized by a profound production of inflammatory mediators, it might be expected that if conventional tidal volumes contribute to development of lung injury, the injury also may be associated with increased production of cytokines. We therefore conducted a trial to determine whether mechanical ventilation with conventional or lower tidal volume would be associated with different cytokine patterns in the lungs and the plasma of critically ill patients without ALI at onset of mechanical ventilation. Secondary end points were development of lung injury, duration of mechanical ventilation, and mortality.
Materials and methods
Participants
From January 2005 until December 2007 patients were recruited in the intensive care departments of one academic and one regional teaching hospital in the Netherlands. The academic ICU is a 28-bed "closed format" department where medical/surgical patients (including neurosurgery/neurology, cardiothoracic surgery, and cardiology patients) were under the direct care of the ICU team. The ICU team comprised 10 full-time ICU physicians, eight subspecialty fellows, 12 residents, and occasionally one intern. The regional teaching ICU is an eight-bed "open format" department with medical/surgical patients (not including neurosurgery and cardiothoracic surgery patients). The ICU team comprised three full-time ICU physicians, five physicians who participate in evening and night shifts, and one resident. The two ICUs had similar standards of practice in terms of mechanical-ventilation and sedation protocols.
Patients were eligible for the study if they did not meet the consensus criteria for ALI/ARDS [
17] and needed mechanical ventilation for an anticipated duration of more than 72 hours. Patients had to be randomized less than 36 hours after the onset of mechanical ventilation. Exclusion criteria were age younger than 18 years, participation in other clinical trials, pregnancy, increased uncontrollable intracranial pressure, chronic obstructive pulmonary disease (defined as a forced expiratory volume in 1 second to a forced vital capacity ratio less than 0.64 and daily medication), restrictive pulmonary disease (evidence of chronic interstitial infiltration on chest radiograph), use of immunosuppressive agents (100 mg hydrocortisone per day was allowed), pulmonary thromboembolism, previous pneumectomy or lobectomy, and previous randomization in this study. Randomization was performed by using sealed opaque envelopes in blocks of 50 patients. Each study center had its own randomization block. The protocol was approved by the medical ethics committees of both hospitals, and written informed consent was obtained from the patient or closest relatives before entry in the study. All procedures were done in compliance with the Helsinki declaration.
Interventions
The volume-controlled mode was used for mechanical ventilation. To calculate tidal volume, predicted body weight was used, as described [
4]. The target tidal volume in the conventional group was 10 ml/kg of predicted body weight, which was routine practice at the time of the conduct of the study. Patients from the intervention group were ventilated at tidal volumes of 6 ml/kg of predicted body weight. In case patients were randomized to 6 ml/kg, the attending physician was allowed to increase tidal-volume size to 7 to 8 ml/kg if patients had severe dyspnea, as identified by increased respiratory rate (more than 35 to 40 breaths per minute) accompanied by increasing levels of discomfort (with or without need for more sedation). Levels of PEEP were set, together with the level of inspired oxygen (Fi
O2) depending of the Pa
O2 according to a local protocol.
The ventilator was routinely (3 times/day) switched to the pressure support mode. If the pressure support mode was tolerated, this mode was used for further mechanical ventilation. Toleration of pressure support mode was assessed at the discretion of the attending physician. The pressure support was adjusted to reach the target tidal volumes. In case the attending physician preferred pressure-support ventilation in a patient randomized to the lower-tidal-volume group, and the applied tidal volume exceeded the target tidal volume because of high levels of pressure support, then this was accepted. Such patients were kept in their original randomization group in the statistical analyses.
As soon as patients were ready to be weaned from the ventilator, the pressure-support level had to be lowered stepwise to 5 cm H2O within 24 hours. If this was not possible because of severe dyspnea, then the pressure support had to be increased to maintain tidal-volume size based on randomization group. Attending physicians decided to extubate the patient, based on general extubation criteria (that is, responsive and cooperative, adequate oxygenation with FiO2 of 40% or less, hemodynamically stable, no uncontrolled arrhythmia, and having a rectal temperature greater than 36.0). If a patient had been weaned from the ventilator but was reintubated for additional mechanical ventilation within 28 days, the same tidal-volume protocol was resumed.
Lung injury was diagnosed if a patient met the consensus criteria [
17]. If it was diagnosed by the attending physician, the local protocol mandated mechanical ventilation with a tidal volume of 6 ml/kg in a pressure-controlled mode for the remaining ventilation period.
Objective and outcomes
The primary outcome was cytokine levels in blindly obtained bronchoalveolar lavage fluid and plasma. Development of lung injury (according to consensus criteria for ALI/ARDS) [
17], duration of mechanical ventilation, and mortality were secondary outcomes.
Data collection
Demographic data, ventilation parameters, and clinical and radiologic data were recorded immediately after the ventilator settings were changed on day 0. Each second day, ventilator settings, blood-gas parameters, radiographic data, and medication use were recorded until the patient was weaned from the ventilator. The oxygenation index was calculated as described earlier [
18]. Mean airway pressure was measured with the ventilator. The lung-injury score (LIS) was calculated. On the day of enrollment and each second day until the patient was weaned from the ventilator, a bronchoalveolar minilavage was performed for the measurement of levels of tumor necrosis factor α (TNF-α), interleukin-1β (IL-1β), and interleukin-6 (IL-6). Simultaneously, blood samples were drawn from an indwelling arterial catheter for IL-6 measurements. Minilavage was performed as described previously [
19]. The recovered fluid was centrifuged at 1,500
g for 10 minutes at 4°C. The supernatant was collected and stored at -80°C until measurements were performed. All markers were measured with an enzyme-linked immunoassay (Sanquin, Amsterdam, The Netherlands).
Definitions
Sepsis was defined by the Bone criteria [
20]. Septic shock was present in cases of persisted hypotension (mean, less than 60 mm Hg) despite fluid resuscitation or vasopressor use [
20]. Pneumonia was diagnosed from new infiltrates on chest radiograph together with clinical signs of infection and positive sputum culture with no other explanation for the symptoms [
21]. Chronic alcohol abuse was defined as a previously established diagnosis of chronic alcoholism, a prior admission for alcohol detoxification, or alcohol withdrawal [
22].
Sample size
The power calculation was based on a previous study on ventilator-associated lung injury [
23]. In this study, bronchoalveolar lavage fluid levels of IL-6 increased by ± 20% in ALI/ARDS patients ventilated with a conventional regimen and decreased by ± 20% in patients ventilated with a protective regimen. Based on these differences and expected baseline IL-6 levels of 250 pg/ml [
19], we calculated that to detect a difference in changes from baseline between groups of 100 pg/ml, with a two-sided significance level of 0.05 and a power of 80%, 49 patients had to be included in each group. As we studied patients without ALI/ARDS, we chose to study twice as many patients, resulting in a total of 200 patients.
Lung injury diagnosis for interim analysis
For reasons of safety, interim analyses on the development of lung injury were conducted after the inclusion of 100 and 150 patients. For this, all chest radiographs were reviewed by two independent physicians who were blinded to all clinical parameters and randomization groups. Any new or worsening abnormality was scored. Chest radiographs showing new or worsening abnormalities were selected for further review. During the review process, they had access to Pa
O2/Fi
O2 (P/F), echocardiography, and fluid-balance data, pulmonary capillary wedge pressures (if measured), and the admission diagnosis. Both physicians were familiar with the consensus criteria for ALI/ARDS [
17]. In case of disagreement, consensus had to be obtained while reviewing the patient together.
Statistical analysis
Data are presented as mean with standard deviation for parametric data or as medians with interquartile range (IQR) for nonparametric data. Baseline comparisons between groups were made with the Student t test, Mann-Whitney U test, χ2 test, or Fisher Exact test where appropriate. The Mann-Whitney U test was used to compare baseline levels of cytokines between groups. To study the primary outcome, a linear mixed model was constructed on cytokine levels, adding time and randomization group as factors in the model. In this model, the interaction between time and randomization group was used to study differences over time between groups. If the residuals were not normally distributed in linear mixed-model analyses, the data were transformed to the natural logarithm of the original data. The relation between cytokine levels and development of ALI/ARDS was studied with a multivariate logistic regression analysis. For the secondary outcome, development of ALI/ARDS was studied with the χ2 test. To show the incidence of ALI/ARDS over time, a Kaplan-Meier curve was constructed, and the log-rank test was used to calculate differences between groups.
To study the effect of tidal volumes while correcting for risk factors for ALI, a multivariate logistic regression analysis was performed. Variables with a P value < 0.10 in univariate analysis were considered for a multivariate model. If collinearity between variables was found, then the weaker variables were removed from the multivariate model. A backward elimination method was used for the final model.
A two-tailed P value < 0.05 was considered to be statistically significant. Data were analyzed by using SPSS, version 14.02 (SPSS Inc., Chicago, IL).
Discussion
ALI/ARDS is rarely present at the time of hospital admission but develops over a period of hours to days in patients with predisposing conditions, such as trauma, shock or sepsis, and associated interventions, including mechanical ventilation [
24]. Therefore, ALI/ARDS may be viewed as a potentially preventable complication. Implementation of prevention strategies, such as lung-protective mechanical ventilation with lower tidal volumes lead to a significant decrease in ALI/ARDS and the mortality of mechanically ventilated patients [
25]. Although both groups in the present study had a comparable prevalence of risk factors for ALI/ARDS, mechanical ventilation with conventional tidal volumes was associated with a delayed decrease in plasma IL-6 levels and an increased frequency of lung injury after the initiation of mechanical ventilation. The benefit of the use of lower tidal volumes occurred without the need for additional sedation or vasopressor use and was not associated with altered requirements for higher PEEP or additional Fi
O2.
Conventional mechanical ventilation was accompanied by an altered plasma cytokine profile but not an altered pulmonary cytokine profile. We found plasma IL-6 levels to decrease over time in both groups. The decrease was, however, more pronounced in patients ventilated with lower tidal volumes. This was not accompanied by different cytokine profiles in the lavage fluids. This is in contrast with findings of earlier studies in patients undergoing elective surgery. In a recent study, procoagulant changes in lavage fluid of patients with healthy lungs were observed after 5 h of mechanical ventilation with large tidal volumes [
11]. Another recent trial showed increased blood cytokine levels in surgical patients ventilated with conventional tidal volumes compared with those in patients ventilated with lower tidal volumes [
12]. The increase in lung-injury score in the conventional-tidal-volume group may not have been reflected by increases in cytokine levels because of a different timing of lavage, deteriorations in P/F, and changes on chest radiographs. A lavage was performed each second day, but the chest radiographs could be made daily, and blood-gas analyses were routinely performed at least 4 times per day. Moreover, in seven patients, the attending physician reduced the tidal volume size to 6 ml/kg after the development of lung injury. The study protocol allowed only lung lavage every second day. Lavage procedures were not always performed on the moment of ALI/ARDS diagnosis and before tidal volumes were reduced.
Although we did not observe a general increase in cytokine levels, the mechanism by which mechanical ventilation with conventional tidal volumes leads to full-blown ALI/ARDS in critically ill patients may be as follows. A second-hit model theory can be suggested as a mechanism by which mechanical ventilation may lead to ALI/ARDS. The patients in whom lung injury developed in our study had increased IL-6 levels in their lavage fluid, a higher level of PEEP, and a worse oxygenation index. Although the IL-6 level and oxygenation index were not independent predictors in multivariate analysis, it does show that these patients had some pulmonary inflammation at baseline. It may be speculated that patients with a certain level of inflammation are the patients at risk for ventilator-induced lung injury. Moreover, the baseline level of PEEP was significantly associated with development of lung injury. Larger studies also showed that next to tidal-volume size, risk factors for ALI/ARDS, level of PEEP, and P/F are significant predictors of ALI/ARDS, which is in line with the second-hit model theory [
15,
16].
The study was stopped after the second interim analysis. Interim analyses were not planned at first and were not taken into account on the calculation of sample size, which is a limitation of our study. Although development of lung injury was a secondary end point in our study, the attending physicians of the Academic Medical Center insisted on interim analyses, as they had concerns about the safety of the study. They assumed that the development of lung injury was more frequent with the use of conventional tidal volumes. Therefore, interim analyses were planned halfway and after 150 patients. A stopping boundary was not determined beforehand. As the P value was as low as 0.01 on the second interim analysis, the investigators had no other choice than to stop the trial.
The multivariate analysis showed that tidal volume was an independent predictor of ALI/ARDS development, together with the level of PEEP. Our study was not powered to investigate various risk factors for ALI/ARDS separately in a multivariate model. Therefore, the results of this analysis should be taken with caution. As the level of PEEP may simply have been a marker of disease severity, tidal-volume size may be the only risk factor that can be influenced by the attending physician. Risk factor, oxygenation index, number of transfused blood products, and the baseline lavage-fluid IL-6 level were all associated with lung injury in the univariate analysis and tended to have a significant influence in the multivariate model. As lung injury was a secondary outcome, this study was not powered to investigate all these variables, but larger studies showed that these are significant predictors of ALI/ARDS [
15,
16]. This is in line with the suggested second-hit model described earlier.
Our findings are in line with earlier reports in patients without ALI/ARDS at the onset of mechanical ventilation [
15,
16,
23,
24]. The odds ratio of 5.1 for a tidal volume of 10 ml/kg
versus 6 ml/kg in the present study is in line with earlier findings of an odds ratio of 1.3 for each milliliter above 6 ml/kg [
15]. Although the incidence of development of lung injury in the conventional-tidal-volume group of our study seems quite high (13.5%), it is still low compared with a comparable cohort of patients without preexisting lung injury in another setting [
24]. Of interest, the frequency of ALI/ARDS decreased approximately 65% after implementation of lung-protective measures, resulting in an incidence of ALI/ARDS comparable to that in our study [
25].
The reported mortality rates in ALI/ARDS patients are, however, relatively high, even with lower-tidal-volume mechanical ventilation [
4]. We did not find differences in either mortality or the number of ventilator-free days, however. First, our study was not powered to these end points. Second, the protocol required that if lung injury developed, tidal volumes were reduced to 6 ml/kg. This may have underestimated the effect of higher tidal volumes.
Although diagnosing ALI/ARDS is susceptible to subjective interpretation [
26], the final decision on the development of lung injury was made by two physicians who were independent of the study. Their judgment was supported by differences in LIS, compliance, and pulmonary and plasma IL-6 levels between patients with and without lung-injury development. Nevertheless, although the physicians were experienced, it may still be possible that some patients who were identified as having ALI/ARDS had hydrostatic pulmonary edema or pulmonary infection and that patients identified as not having ALI/ARDS actually had ALI/ARDS. This may have happened in cardiac-arrest patients, as they are prone to have elevated left atrial pressures and cardiogenic pulmonary edema. Furthermore, as this patient group has a poor prognosis, ALI/ARDS may have been overlooked by the attending physicians. However, all patients with respiratory failure in this group were reviewed by the independent intensivists. As the clinical assessment of ALI/ARDS is hampered by the lack of a gold standard [
26], the sensitivity and specificity of any scoring system is moderate at best. The same holds true for the clinical assessment of infection and the identification of patients with ventilator-associated pneumonia [
27]. Despite these limitations, we observed an incidence of 8% in the whole group, which is in line with an earlier report of 6% in a large international cohort study [
16].
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
The study was designed by RMD, EKW, LC, and MJS. Acquisition of the data was performed by RMD, AR, EKW, APV, GC, FP, JJH, and MJG. Analysis and interpretation of data was done by RMD, JK, and MJS. The manuscript was drafted by RMD, JK, and MJS. Critical revision of the manuscript for important intellectual content was done by RMD, AR, EKW, AV, GC, FP, JJH, MJG, JK, and MJS. The statistical analysis was done by RMD, JK, and MJS. Final approval of the manuscript was done by RMD, AR, EKW, APV, GC, FP, JJH, MJG, JK, and MJS.