Detection of ‘best’ positive end-expiratory pressure derived from electrical impedance tomography parameters during a decremental positive end-expiratory pressure trial

Mechanical ventilation acts as a stress raiser to lung tissue adjacent to collapsed tissue in the dependent lung and the risk of alveolar hyperinflation in the non-dependent lung [1]. An alveolar recruitment maneuver and the use of positive end-expiratory pressure (PEEP) are applied to open up and to keep open the atelectatic lung tissue in an attempt to minimize this stress and over-distention.

The original definition of the best PEEP as proposed by Suter et al. is the PEEP level with the best compromise between lung aeration and circulatory depression [2]; circulatory depression is caused by compression of capillaries due to hyperinflation. Tusman et al. performed an incremental and decremental PEEP trial in eight volume-controlled ventilated surfactant-depleted pigs; they showed that calculation of dead space detected early signs of lung collapse, which correlated well with findings on computed tomography (CT) [3]. In a second study, these authors reported that continuous compliance monitoring could identify the onset of alveolar collapse as confirmed by CT, as well as changes in partial arterial oxygen pressure (PaO₂) values [4]. Another experimental study, using an oleic acid-induced acute lung injury (ALI) model, showed that minimizing elastance of the respiratory system could be used to titrate PEEP settings [5]. Maisch et al. performed an incremental and decremental PEEP trial in 20 anesthetized patients undergoing elective surgery and confirmed that dynamic compliance and dead-space calculations were able to detect alveolar collapse and/or over-distention, whereas functional residual capacity (FRC) and PaO₂ changes were less sensitive [6]. From these studies it has been concluded that dynamic compliance and dead-space calculations are the most reliable global parameters to define the best PEEP at the bedside.

CT is regarded as the gold standard to assess the effect of a recruitment maneuver and the applied PEEP on aeration of the lung [7, 8]. However, the obvious drawbacks of repeated CT scans (that is, transfer of the mechanically ventilated patient and excessive radiation exposure) reduce the application of CT as a tool for assessment of PEEP settings. On the other hand, electrical impedance tomography (EIT) is a real-time bedside monitoring tool, which has proven to correlate well with CT for assessment of changes in gas volume and tidal volume [9‐11].

Several EIT parameters have been developed to collect more data on ventilation distribution in order to optimize ventilator settings [12‐15]. The present study examines whether one specific EIT parameter is able to describe the optimal PEEP level at the bedside; for this, we defined the best PEEP as the PEEP level with minimal lung collapse and minimal over-distention.

Details of patient characteristics are presented in Table 1. During the entire PEEP trial, patients were ventilated with an inspiratory pressure above PEEP of 10 ± 2 cm H₂O. A PaO₂/FiO₂ ratio ≥350 mmHg was defined as an open lung; in two patients we were unable to open up the lung despite the recruitment maneuver and use of a PEEP level of 15 cm H₂O. Figure 1 shows the distribution of TIV for one representative patient during the decremental PEEP trial. The effects of decremental PEEP on TIV, regional compliance, VSA, COV, RVD and GI index are presented in Figure 2A-F.

Figure 3 shows the effects of the decremental PEEP trial on the PaO₂/FiO₂ ratio and dynamic compliance for the entire study population. The PaO₂/FiO₂ ratio had the highest value at 15 cm H₂O PEEP but showed a significant decrease after lowering PEEP from 10 to 5 to 0 cm H₂O compared with 15 cm H₂O PEEP (Figure 3A). Dynamic compliance had the highest value at 10 cm H₂O PEEP but showed a significant decrease at a PEEP level of 0 compared with 15 cm H₂O (Figure 3B).

In the non-dependent lung regions, TIV (Figure 2A), regional compliance (Figure 2B) and VSA (Figure 2C) reached the maximum value at 5 cm H₂O PEEP. In contrast, in the dependent region TIV, VSA and regional compliance reached a maximum at the highest PEEP level applied and then decreased during the entire PEEP trial. During the decremental PEEP trial, COV increased steadily towards the anterior part of the thorax cavity, indicating loss of TIV in the dependent region (Figure 2D). During the decremental PEEP trial the RVD index increased in both lung regions and the RVD values of the non-dependent region remained significantly lower than those in the dependent region, except at zero end-expiratory pressure (Figure 2E). The GI index had the lowest values at 15 and 10 cm H₂O PEEP and then increased steadily at lower PEEP levels (Figure 2F), indicating more homogeneous ventilation at higher PEEP levels.

Figure 4 presents the results of intratidal gas distribution. At the highest levels of PEEP, the intratidal gas distribution to the dependent region was higher than that to the non-dependent region. Decreasing the PEEP level resulted in a higher overall gas distribution to the non-dependent region compared with the dependent region. At a PEEP level of 10 cm H₂O, the intratidal gas distribution curves of both regions crossed each other during a breath (Figure 4). Figure 5 shows the calculated ITV index as percentage of 1 for each PEEP level in each individual patient. There was a correlation between the PaO₂/FiO₂ ratio and the ITV index (−0.762; P <0.001).

This study demonstrates that intratidal gas distribution visualizes the best PEEP as compared with dynamic compliance in post cardiac surgery patients. In addition, the ITV index is able to determine a specific PEEP level in each individual patient, resulting in an even distribution of tidal volume to the non-dependent and dependent lung regions. Below this specific PEEP level, the intratidal gas distribution is predominantly distributed to the non-dependent region. This indicates that, at these PEEP levels, there is less ventilation in the dependent region due to lung collapse. In contrast, at PEEP levels above this specific level, there is less ventilation in the non-dependent region indicating overdistention.

In an experimental study, Protti et al. showed that ventilation with high tidal volumes, resulting in an expiratory volume of 1.5 times FRC (equal to a strain of 1.5), caused severe lung edema; all their study animals died within the observation period of 54 h [22]. In a second study, the authors ventilated all animals with a strain of 2.5 and showed that high tidal volumes with a low level of PEEP damaged the lungs and increased mortality, whereas high PEEP levels together with low tidal volume but with the same strain of 2.5, did not result in edema and all animals in this group survived [23]; it was suggested that application of high PEEP levels might lead to more homogeneous lung ventilation. In 1970, Mead et al. estimated that forces acting on lung tissue increase with a factor 4.5 when lungs are inhomogeneously ventilated [24]. This was recently confirmed by Rausch et al. who performed x-ray tomographic microscopy (generating detailed three-dimensional alveolar geometry) in rat lungs and found local strain values four times the global strain [25]. Therefore, a parameter that describes the ventilation distribution could be of importance in finding the best PEEP in patients with acute respiratory distress syndrome (ARDS).

Intratidal gas distribution was first described by Löwhagen et al. who used this technique in 16 volume-controlled ALI/ARDS patients to describe ventilation distribution to different lung regions within an inspiration [14]; they found that the intratidal gas distribution of the dorsal and mid-dorsal regions increased at higher PEEP levels, indicating redistribution of ventilation to the dependent region [14]. We modified their analysis by combining the ventral and mid-ventral regions into a non-dependent region and their mid-dorsal and dorsal regions into a dependent region [26]. Previously, we used the intratidal gas distribution technique to assess the effect of different assist levels during pressure support ventilation (PSV) and neurally adjusted ventilatory assist (NAVA) on ventilation distribution. We demonstrated that NAVA improved ventilation of the dependent lung region compared with PSV, leading to a more homogeneous ventilation of the lung [26]. In that study using the intratidal gas distribution technique, we demonstrated for the first time, less over-assistance during NAVA whereas there was marked over-assistance at higher pressure support levels [26]. This latter finding indicates that PSV with higher support levels mimics control ventilation with predominantly ventilation of the non-dependent lung. We also used the intratidal gas distribution in an experimental study comparing global and regional parameters to detect best PEEP in healthy and in ARDS lungs [27]. In these animals we found the same trend as in the present study, that is, that the intratidal gas distribution curves of the dependent and non-dependent regions reached each other at a specific PEEP level. Below this specific PEEP level, ventilation is mainly distributed to the non-dependent lung and above this level ventilation is mainly distributed to the dependent lung region. Above this specific PEEP level ventilation distribution to the non-dependent lung region diminished at higher PEEP levels, indicating that this region is overdistended.

In the present study we used a decremental PEEP trial and, as long as the PEEP level is adequate to keep the dependent lung region open, the used pressure amplitude of around 10 cm H₂O was sufficient to ventilate this region; however, after collapse the inspiratory pressures are too low to open up this dependent region. This indicates that ventilation distribution to the dependent lung will be improved if higher inspiratory pressures are used.To make the intratidal gas distribution an easy-to-use parameter at the bedside, we introduced the ITV index. An ITV index of 1 indicates homogeneous distribution of tidal volume to the non-dependent and dependent lung regions. The intratidal gas distribution shows the behavior of the distribution of tidal volume during one breath. Figure 5 illustrates the ITV index, in which 100% indicates an ITV index of 1, that is, an equal distribution of ventilation to the dependent and non-dependent lung regions.

To test whether the intratidal gas distribution reliably defines the optimal PEEP, we compared the optimal PEEP values defined by the intratidal gas distribution, PaO₂/FiO₂ ratio and dynamic compliance. We found a negative correlation between the PaO₂/FiO₂ ratio and the ITV index. A lower ITV index means that more tidal ventilation is distributed to the better perfused dependent lung region, leading to less shunt and improved PaO₂/FiO₂ ratio. In contrast, we found no correlation between ITV and dynamic compliance. Alveoli that open up during inspiration but collapse during expiration give higher compliance values, whereas this cyclic collapse is harmful to the lung. In addition, recruitment of a lung region, but collapse in another lung region at the same time, will not increase dynamic compliance of the entire lung. This latter effect can be visualized by regional or pixel compliance from EIT. Costa et al. introduced the regional or pixel compliance based on EIT measurements during a decremental PEEP trial in two patients with pneumonia [12]. These patients were ventilated with PCV using a constant pressure amplitude and the change in impedance represents the change in volume; thus, for each pixel the compliance could be calculated. The highest compliance at a specific PEEP level was indicated as the best PEEP. Above this PEEP value, compliance decreased due to over-distention and below this value compliance decreased due to collapse. The authors showed that the optimal PEEP level was different for the dependent and non-dependent region [12]. Also, in their experimental study, Dargaville et al. demonstrated that ventral, medial and dorsal lung regions have different optimal PEEP levels based on regional compliance values in both normal and surfactant-depleted lungs [28]. This was confirmed by our results, in which the best PEEP level for the non-dependent and dependent regions were 5 and 15 cm H₂O, respectively (Figure 2B).

In accordance with our previous studies [16, 29], we found more ventilation of the dependent region at higher PEEP levels, as described by the parameter COV (Figure 2D). COV describes the ventilation distribution in the ventral-to-dorsal direction and a value of 50% reflects an even distribution [13]. Thus, COV provides information about the optimal distribution of tidal volume to the non-dependent and dependent regions at a certain PEEP level, but it does not give information about collapse or overdistention during a breath, as is seen with the intratidal gas distribution.

The RVD was developed to assess the homogeneity of aeration of the lung regions [15, 19] during a slow inflation maneuver. It has been demonstrated in pigs that ALI lungs are more inhomogeneous compared with healthy lungs [21]; however, application of higher PEEP levels improved the homogeneity of lung ventilation. In addition, the authors showed that the dependent region was slower inflated as compared with the non-dependent region, indicating inhomogeneous ventilation of the dorsal lung parts; however, they used the SDs of the RVD index to create a ventilation homogeneity map [21]. In the present study we calculated the time needed to reach a threshold of 40% of the regional impedance change compared with the total inspiratory time and without the use of a slow flow inflation. We found that RVD values in both lung regions increased at each decremental PEEP step. Therefore, we were unable to detect the best PEEP level by means of RVD, with the used PEEP levels. However, Wrigge et al. described that RVD could only be used during a slow flow inflation maneuver with a tidal volume of 12 mL/kg to describe tidal recruitment [15]. Therefore, in the present study the results of the RVD are incorrect. Another index to describe homogeneous ventilation is the GI index, which quantifies the variation in tidal ventilation distribution [20, 21] and shows the lowest value in healthy patients and the highest value in patients with ARDS [21]. However, because this index describes homogeneity based on differences in measured impedance, the index value does not take into account the presence of atelectasis or over-distention and, therefore, we believe that this is not an appropriate index. As within our range of PEEP levels we were unable to detect an optimal PEEP level based on the GI index and RVD index, it is possible that an optimum may have been found if higher PEEP levels had been applied than used in the present study.

This study has some limitations. First, EIT measures ventilation distribution in a lung slice of approximately 5 to 10 cm [30, 31]; therefore, information gathered by EIT has a limited external validity for the remaining lung tissue. However, placing the EIT belt at a higher position is known to reduce the probability to detect inhomogeneity of the lungs at decreasing PEEP levels [16] and, thus, the probability to detect lung areas susceptible to the development of VILI. Therefore, we placed the EIT belt just above the diaphragm to increase the probability of detecting lung collapse.

Second, based on the protocol of our study described previously, EIT measurements were not recorded continuously [16]. However, by treating the EIT data as percentages (instead of absolute values) the baseline shifts are corrected. In addition, the EIT belt was not disconnected from the patient to measure the same lung slice during each measurement. Third, this study was performed with post cardiac surgery patients who respond well to a recruitment maneuver. However, Reis Miranda et al. demonstrated that a PEEP level of 15 cm H₂O was necessary to keep the lung open (PaO₂/FiO₂ > 350 mmHg) in cardiac surgery patients [32]; the authors also showed that if patients are ventilated according to the open lung concept, fewer cytokines are released, FRC recovers faster, less oxygen is required in the normal ward, and that afterload of the right ventricle is lower in these patients [32‐35]. Despite the fact that adequate PEEP settings are also important in cardiac surgery patients, most patients in need of a recruitment maneuver in combination with the best PEEP are ARDS patients. However, ARDS patients show less response to a recruitment maneuver compared with cardiac surgery patients, due to fewer gravitational-dependent infiltrates. Particularly in ARDS patients it is important to achieve homogeneous ventilation to reduce the stress acting on lung tissue. Our earlier experimental study showed the same trend of tidal ventilation distribution during a decremental PEEP trial in both ARDS-induced and healthy lungs [27]. However, additional studies are required to test whether application of this specific PEEP level leads to better outcome and also to confirm the present findings in patients with ARDS.

The ITV index was comparable with dynamic compliance to indicate the best PEEP level in post cardiac surgery patients. The intratidal gas distribution is able to identify the onset of over-distention in the non-dependent part and recruitment in the dependent part. We believe that the ITV index may be the ideal bedside tool to detect the best PEEP; however, its preventive effect on ventilator-induced lung injury (VILI) and thereby on outcome still needs to be examined in patients with ALI/ARDS.

● Several EIT parameters have recently been developed to optimize ventilator settings.

● Tidal impedance variation, regional compliance and ventilation surface area showed different optimal PEEP levels for the dependent and non-dependent lung regions.

● The intratidal gas distribution is able to detect the onset of alveolar overdistention and collapse, within one inspiration.

● Best PEEP as defined by the intratidal gas distribution shows good agreement with best PEEP as defined by the global parameter dynamic compliance.

● Equal distribution of ventilation to the dependent and non-dependent lung regions, as defined by the intratidal gas distribution, might lower stress and strain in the lung.