State-of-the-ART Trial: Do Recruitment Maneuvers & Higher PEEP Raise Mortality?
Jon-Emile S. Kenny MD [@heart_lung]
“To believe in medicine would be the height of folly, if not to believe in it were not a greater folly still.”
A 32 year old man with no past medical history save for a BMI of 51 is admitted with severe acute pancreatitis following a large intake of ethanol at a recent bachelor party. 72 hours into his hospitalization he becomes severely short of breath; he has gained 3 kg since admission but has no peripheral edema. Plain film of his chest reveals bilateral infiltrates; he is placed on a Venturi mask and his calculated PaO2:FiO2 ratio is 140. As he is unable to tolerate both high flow nasal cannula and BiPAP, an awake intubation is performed and he is ventilated with a tidal volume of 6 mL/kg of predicted body weight and a PEEP of 12 cm H2O. He is sedated and paralyzed, but his oxygenation progressively worsens.
At the recent European Society of Intensive Care Medicine in Vienna, the Alveolar Recruitment Trial was presented; the results immediately launched Twitter into a digital chorus of shocked songbirds. The open lung approach is dead, they sang – and this isn’t fake news.
The premise behind the ‘open lung approach’ is to open collapsed lung – often using a recruitment maneuver – and keep it open with positive end-expiratory pressure [PEEP]. The putative benefit from this exercise being mitigation of atelectrauma – the injury sustained by lung tissue from repetitive opening and closing – particularly about areas of lung inhomogeneity, or stress raisers.
Prior to the ART Trial, three large randomized controlled trials evaluated the effects of high PEEP versus low PEEP in the context of lung protective tidal volumes. In 2004, the ARDSNet ALVEOLI trial [549 patients] was the first, followed by the LOVS [985 patients] and EXPRESS [767 patients] trials in 2008. Of these 3, only the LOVS trial routinely used recruitment maneuvers as a part of the experimental paradigm. In each of the three trials there was a trend towards improvement in mortality for the patients receiving high PEEP, however, none were statistically significant.
Following these three trials, a meta-analysis was performed which showed, overall, no mortality benefit to high PEEP. However, in sub-group analysis, those with a PaO2:FiO2 ratio of less than 200 had improved mortality with augmented end-expiratory pressure; thus the ART Trial only enrolled those with moderate-to-severe ARDS.
1013 patients were randomized in the ART Trial to either low PEEP which was guided by the same PEEP-FiO2 table used in the low PEEP group in both the ALVEOLI and LOVS trials [see table 1] or to recruitment maneuver and titrated PEEP. Overall, 62% of the patients had ARDS of pulmonary origin while 38% were of extra-pulmonary origin and these proportions are similar to previous trials.
Table 1: Rendition of FiO2-PEEP table used in low PEEP arm of ALVEOLI and LOVS. PEEP is positive end-expiratory pressure and FiO2 is fraction of inspired oxygen.
When enrolled in the experimental group, paralytics were given followed by IV fluid if there were signs of ‘fluid responsiveness.’ The patients then received a recruitment maneuver using incremental PEEP to open the lung. Importantly, after about 50% of the patients in the trial were enrolled, the recruitment maneuver protocol was changed. There were 3 cases of resuscitated cardiac arrest associated with their initial recruitment protocol which required a successive PEEP titration [25 cm H2O for 1 minute, 35 cm H2O for 1 minute and 45 cm H2O for 2 minutes] in pressure control – with a driving pressure of 15 cm H2O! Following recruitment, PEEP was decreased slowly with calculation of static respiratory system compliance at each step-wise drop in PEEP. The level of PEEP at which the best compliance was achieved was considered optimal; to this value, 2 cm H2O was added and this was the PEEP value used for ventilation following a second recruitment maneuver. For the second 50% of enrollment, PEEP was titrated upwards more quickly [1 minute at each step] with a maximum value of 35 cm H2O for recruitment.
Death at 28 days was higher in the experimental group with an absolute risk increase of 6% [number needed to harm of about 17]; additionally, there was increased pneumothorax & barotrauma in the intervention group. Hypotension and vasopressors required within the first hour was higher in the experimental group and there was a clinically meaningful, but not statistically significant increase in cardiac arrest within the first day of randomization in the experimental group as well.
Recruitment & Cor Pulmonale
Firstly, the intervention group was more than simply increased PEEP – it was also a fluids, paralytics and recruitment maneuver [RM] group, so teasing out the reason for the increased mortality becomes slightly challenging. Nevertheless, paralytics – based on current evidence – are not thought harmful in moderate-to-severe ARDS and the difference in fluid administration between the two groups was only 300 mL at the end of day 1. Thus, the difference between the two arms, potentially, relates to the recruitment maneuver and/or the higher level of applied PEEP.
Interestingly, only the LOVS trial [highlighted in green] used routine recruitment for open lung in the previous trials [see table 2]. In that trial, a much ‘gentler’ method of recruitment was used – CPAP at 40 cm H2O for 40 seconds. Even with this, the authors of LOVS noted a 22% immediate complication rate with recruitment which was mostly hypotension and hypoxemia; further, 3 patients were noted to have signs of barotrauma immediately following the RM.
Table 2: comparison table between trials discussed within text. This represents data at the end of day 1 following randomization. Values in parentheses indicate change from baseline value if reported. *the ALVEOLI trial did not report death at 28 days, but instead in house mortality. Figure 1 from ALVEOLI depicts survival curves for the keen reader to scrutinize.
Given the hypotension and need for vasopressors observed following RM in both LOVS and ART, concern for acute cor pulmonale is real. At the levels of PEEP applied within this patient population, RV afterload is particularly sensitive. Without echocardiographic or invasive monitoring, it is hard to detect and treat this common complication of ARDS. Notably, the patients in ART had the highest PaCO2 with comparable respiratory rates and tidal volumes. This may reflect a high dead space secondary to cor pulmonale, but small tidal volumes themselves do also raise the dead space fraction.
Respiratory Mechanics & Mechanical Power
As can be seen in the summary table 2 – the amount of PEEP in the ‘low PEEP’ group in ART was substantially higher than in the other trials. Given that the same PEEP-FiO2 table was used as in the ALVEOLI and LOVS studies, this speaks to the sicker population in ART – which only enrolled patients with a PaO2:FiO2 ratio of less than 200. This may partly explain why the mortality in ART was much higher than the other trials and further complicates comparisons. Interestingly, in the meta-analysis combining the ALVEOLI, LOVS and EXPRESS trials, the sub-group of patients with a PaO2:FiO2 ratio of less than 200 had hospital death rates of 34.1% and 39.1% for high versus low PEEP, respectively. It is not entirely clear why mortality is considerably higher in ART.
It may be tempting to relate indices of respiratory mechanics, such as the driving pressure, between these trials, but it is important to recognize that driving pressure will vary as a function of both tidal volume delivered and respiratory system compliance. Thus, that the experimental group in ART had the lowest driving pressure may reflect better respiratory mechanics, but also that these patients had the smallest tidal volume, on average, as well.
Nevertheless, the two arms of the ART trial had very similar tidal volumes and in the experimental group, the driving pressure fell. This finding echoes improvement in respiratory system compliance, which was also demonstrated. In other words, the recruitment maneuver and higher PEEP did what was desired. Despite this, outcomes were worse. One potential unifying concept is that of mechanical power which takes into consideration both the static and dynamic energies required to ventilate the lung over time. The application of PEEP too, raises the energy applied to the lung skeleton. Thus, while RM and high PEEP improve the static contribution of energy applied to the lung, to the extent that higher respiratory rate and higher peak pressures result, power will increase and so too will VILI.
Recently, sub-phenotypes of ARDS have been identified. Importantly, the hyper-inflammatory sub-phenotype 2 has been associated with improved outcome in patients who receive less fluid and higher PEEP. To my eye, there is overlap between sub-phenotype 2 and extra-pulmonary ARDS, first coined by Gattinoni in the late 1990s. To the credit of the authors’ of the ART trial, they identified pulmonary versus extra-pulmonary ARDS patients a priori for sub-group analysis. There was a wide confidence interval in the hazard ratio for death at 28 days in those with extra-pulmonary ARDS, while those with pulmonary ARDS clearly benefited from the lower PEEP strategy. Accordingly, it’s possible that RM and higher PEEP may be less dangerous in those with extra-pulmonary ARDS [e.g. pancreatitis], so long as mechanical power doesn’t rise as a consequence.
Secondly, assessment of driving pressure and respiratory system compliance often assumes a normal compliance of the chest wall. In patients with morbid obesity, the high driving pressure and low respiratory system compliance may be a function not of the lung, but rather the elastic load imparted by body habitus. In such patients, recruitment maneuvers and higher PEEP will affect the trans-pulmonary pressure [and therefore the mechanical power applied to the lung itself] much less so. There was no mention in the ART trial of body habitus or BMI of the patients randomized to each intervention.
Return to the Case
The overnight critical care fellow was concerned about loss of FRC given paralysis and the patient’s body habitus. Esophageal pressure monitoring was not available, so she utilized the stress index as a surrogate for detecting terminal alveolar recruitment or distention. At a PEEP of 12, the patient’s stress index was less than 1, so PEEP was slowly increased until the terminal morphology of the pressure-time curve was straight. At 20 cm H2O of PEEP, the patient’s oxygenation improved as did PaCO2. His mean arterial pressure increased following the rise in PEEP.