Bedside selection of positive end-expiratory pressure by electrical impedance tomography in hypoxemic patients: a feasibility study

Acute respiratory distress syndrome (ARDS) is a relatively common and severe form of respiratory failure, characterized by massive non-cardiogenic pulmonary edema, with consequent loss of aeration in the alveolar spaces [1]. ARDS patients require intubation and mechanical ventilation as lifesaving procedures, and positive end-expiratory pressure (PEEP) is a key element of mechanical ventilation settings. Since both lower and higher PEEP levels may be associated with significant adverse consequences [2], personalized PEEP setting might be of cornerstone importance. Ideally, PEEP should optimize recruitment to improve oxygenation and reduce lung strain, without causing excessive overdistension, but controversy exists on the best bedside method to select PEEP. A common clinical approach is based on the severity of hypoxemia, relying on the use of PaO₂/FiO₂ tables [3]. Other approaches to select “personalized PEEP” are based on its effect on respiratory mechanics, focusing on plateau pressure [4], on stress index [5] or on transpulmonary pressure [6, 7]. These methods, however, do not provide consistent finding [8] and share the limitation of “lumping” into one measurement heterogeneous processes within the lung (i.e., recruitment, tidal opening–closing and overdistension [9, 10]) and of using surrogate rather than direct measures for lung recruitment induced by PEEP. In summary, bedside personalization of PEEP is still quite far from clinical practice. Randomized clinical trials have not shown clear benefit by indiscriminate application of high PEEP levels: Although a meta-analysis suggested that “higher” PEEP might be beneficial in moderate–severe ARDS [11], in everyday clinical practice clinicians still tend to apply relatively low PEEP levels even in severe ARDS patients [12].

In the present study, we hypothesized that the optimal PEEP level for each patient may be selected by assessing its efficacy in maintaining alveolar recruitment induced by a recruitment maneuver (RM). RMs are transient and voluntary increases in transpulmonary pressure that could reopen previously collapsed alveoli; they typically consist of application of continuous positive airway pressure of 30–50 cmH₂O for 20–40 s, or transient increases in PEEP and/or inspiratory pressure, with a consequent increase in end-expiratory lung volume (EELV), decrease in lung strain and improvement in patient’s oxygenation [13, 14].

However, if RM is not followed by the application of adequate PEEP, EELV will progressively decrease over time (alveolar de-recruitment). At the opposite, RM plus adequate PEEP level will minimize de-recruitment and maintain sustained recruitment.

Electrical impedance tomography (EIT) is a noninvasive, radiation-free, bedside lung monitoring technique [15], that tracks real-time changes in regional lung ventilation and end-expiratory lung impedance (EELI), which closely correlate with EELV changes [16]. While several studies demonstrated the ability of EIT in assessing alveolar recruitment [17, 18], only a few used this method to guide therapy in humans [19]. The aim of the present study was to assess the feasibility of personalized PEEP selection based on its efficacy in stabilizing the EELV increase induced by a RM, using EIT as tool to monitor EELV changes. Moreover, we compared the effects of the selected PEEP on gas exchange, respiratory mechanics, hemodynamics and tidal recruitment/de-recruitment and overdistension with those induced by the application of PEEP levels selected according to PEEP/FiO₂ tables. This comparator was chosen because of its large acceptance in the clinical practice and the frequent use of this approach as a control in other studies aimed at testing physiological-based methods to titrate PEEP [4, 6].

The study was conducted in the general intensive care units of the university-affiliated hospitals San Gerardo, Monza and Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Milan, both in Italy. Institutional ethical committees of each institution approved the study, and informed consent was obtained according to local recommendations. Inclusion criteria were as follows: patient with acute hypoxemic respiratory failure (PaO₂/FiO₂ ratio ≤300 mmHg) of non-cardiogenic origin undergoing controlled mechanical ventilation with PEEP ≥ 5 cmH₂O. Exclusion criteria were as follows: age <18 years, pregnancy, hemodynamic instability (requiring vasoactive drugs), presence of pneumothorax or lung emphysema, previous history of severe chronic obstructive pulmonary disease (GOLD III or IV), contraindications to the use of EIT (e.g., presence of pacemaker or automatic implantable cardioverter–defibrillator) and impossibility to place the EIT belt in the right position (e.g., presence of surgical wounds dressing). At enrollment, the following variables were collected: sex, age, predicted body weight (PBW), body mass index, Simplified Acute Physiology Score II (SAPS II) value at ICU admission, etiology of acute respiratory failure, diagnosis of ARDS and duration of intubation before enrollment. Vital status was recorded for all patients at ICU discharge.

The main findings of this study can be summarized as follows: Bedside PEEP setting based on sustained recruitment following a RM as visualized by EIT was feasible in the majority of patients with acute hypoxemic respiratory failure (most of whom fulfilled ARDS criteria). This method invariably led to the application of higher PEEP levels in comparison with the commonly used ARDSnet table and was associated with improved oxygenation. Furthermore, EIT allowed to disclose and quantitate the presence of regional overdistension associated with PEEP increase. PEEP setting in hypoxemic respiratory failure and ARDS remains controversial. A meta-analysis on three large randomized trials [3] showed that higher PEEP could be beneficial in more severe ARDS patients, but no indication was provided on how to titrate this higher PEEP: Indeed, while in two studies an oxygenation-based criterion was used, in the third trial a respiratory mechanics-based method was used. Later, a secondary analysis by Goligher et al. [26] showed that the benefit of higher PEEP on mortality was limited to the patients who had an oxygenation improvement, likely indicating the presence of recruitment. Even if the aforementioned paper does not necessarily proof a cause relationship effect (i.e., that higher PEEP prevented death in recruiters), we reasoned that it would have been desirable to have a method to set PEEP directly targeting alveolar recruitment combined with prevention of de-recruitment, and we focused on EIT as a bedside noninvasive tool. While the ability of EIT in assessing lung recruitment has been previously shown, most studies were conducted on animal models and/or explored the effects of relatively high PEEP changes (in the order of 10 cmH₂O) [27, 28]. On the contrary, we aimed to test the feasibility of a specific protocol, with a “fine tuning” of PEEP in steps of 2 cmH₂O. Since we expected that in most patients the EIT-based approach would have led to increase in PEEP from the NIH table, we added PEEP_EIT+2 phase, in order to establish the effects on oxygenation and respiratory mechanics of a further increase in PEEP above the level set by EIT. The method proposed appeared feasible in most patients: This is encouraging in prospect of future evaluation of the protocol in a clinical setting. A PEEP increase will always be associated with an increment of EELV, even if no recruitment occurs, simply because of the expansion of ventilated alveoli. Hence, to dissect these two phenomena, we took advantage of RM. Although available literature shows that RM does not impact outcome, it also shows that it is safe and devoid of major complications [29]. Grasso et al. [30] showed how RMs are unlikely to benefit patients with more than 5 days of ARDS; plus, Borges et al. [31, 32] obtained best results in terms of recruitability with RMs involving stepwise increases in PEEP compared with sustained inflation methods; however, we included in the algorithm RMs more as “diagnostic” tools to exploit potential for lung recruitment, rather than therapeutic measures; for this reason, we did not focus on a specific category of patients and we chose the most simple and immediate method of recruitment, usually used in our clinical practice.

As a reference method, we used the ARDSnet table, one of the most frequently applied methods. Not surprisingly, EIT led to an increase in PEEP in all subjects: This finding is not surprising, since the ARDSnet table has been used to set PEEP in the control group of all studies testing “higher” PEEP strategies.

In the majority of patients, we found that the protocol was feasible: It led to the univocal identification of a PEEP level associated with sustained recruitment after a RM, without exceeding the upper safety limit set to 18 cmH₂O. In the majority of patients, the PEEP changes were within a relatively narrow range (mean 4 cmH₂O) and hence safe to apply. Despite relatively small, however, these changes were clinically relevant, leading to improvement in oxygenation. It was likely due to lung recruitment and increased lung size (as indicated also by the positive correlation between these two variables), and this might possibly lead to a decreased injury from mechanical ventilation. Interestingly, these improvements were not due to the increase in PEEP per se, since a further increase in PEEP above PEEP_EIT was not associated with further improvement in gas exchange and respiratory mechanics.

The global change in compliance between PEEP_ARDSnet and PEEP_EIT phases was the net result of two opposed changes in regional compliance: increasing in the dorsal layers (likely due to recruitment) and decreasing in the ventral layers (likely due to overdistension). We cannot exclude that part of the improvement in the compliance of dorsal layers might be due to the presence of intratidal recruitment; however, this effect was unlikely since previus studies showed that increasing PEEP leads to a decrease in intratidal recruitment [2]. FiO₂ was kept stable in all study phases (0.5 ± 0.1); in this way, we avoided erroneous estimation of alveolar collapse and recruitment due to low alveolar oxygen concentration [31, 33]. We believe that this result further underlines the need for a regional real-time monitoring of the distribution of ventilation, which could prompt a decrease in tidal volume until overdistension of ventral regions drops back to baseline.

This study has some limitations that need to be acknowledged. EIT measurement encompasses only a cross-sectional slice of 5–10 cm of the thorax, and we assumed that other lung regions behave similarly. However, previous studies on similar patient populations showed that ∆EELV measured by EIT well represents the entire lung [34]. The study population was relatively small, but large enough to test the feasibility, safety and efficacy on selected physiological endpoints. Our results do not provide any evidence that a strategy aimed at obtaining stable recruitment leads to a decreased lung injury, but show that such aim can be achieved and that the overdistension also induced by PEEP can be simultaneously monitored and could be used to further adjust ventilation (e.g., reducing tidal volume). Therefore, these results pave the way to a larger study, aimed at assessing whether this approach to PEEP setting leads to benefit in outcome.

Finally, we have not limited our population to ARDS patients: This choice was on one the hand pragmatic, facilitating the enrollment of patients, but on the other hand it also acknowledges the fact that alveolar de-recruitment is not unique of ARDS and that mechanical ventilation can be a challenge also in acutely hypoxemic patients. As this study aimed to test the feasibility of the method, we exclude that the inclusion of non-ARDS patient introduced a significant bias, while we are uncertain of what would be the best approach in an outcome study.

This study shows that, in a cohort of patients with acute hypoxemic respiratory failure undergoing lung protective ventilation, a PEEP selection strategy aimed at maximizing alveolar recruitment and preventing de-recruitment, guided by EIT at the bedside, is feasible, simple and safe, leading to systematically higher PEEP values than the ARDSnet table, with positive effects on gas exchange and respiratory system mechanical properties. This strategy also allows estimating the relative weight of overdistension and recruitment following a PEEP change. These results do not necessarily imply that benefits of recruitment achieved out weight the negative effects induced by overdistension and larger study are required to elucidate if this strategy could also lead to improved clinical outcome.