Dorsal recruitment with flow-controlled expiration (FLEX): an experimental study in mechanically ventilated lung-healthy and lung-injured pigs

A special focus of current research in intensive care medicine and anesthesiology is on strategies for lung-protective mechanical ventilation. Current concepts include variation of end-inspiratory volume or peak or driving pressure [ 1– 3]. However, except for positive end-expiratory pressure (PEEP), which certainly effects inspiration as well, little attention has been drawn so far on modifying the expiratory phase of the ventilation cycle. Consequently, in routine mechanical ventilation the expiration is guided solely by the passive recoil forces of the patients’ respiratory system [ 4]. Thus, for a patient suffering from acute respiratory distress syndrome (ARDS), the injured respiratory system determines the processing of the expiration phase, leading to high expiratory peak flow rates and rapid lung deflation as a consequence of the short time-constant of passive expiration. In addition to the high flow rates, airway pressure drops quickly during passive expiration and the lungs stay at PEEP level until the next inspiration cycle starts. By contrast, flow-controlled expiration (FLEX) decelerates the initial expiratory flow in favor of a moderate flow persisting throughout nearly the whole expiration phase [ 5]. It has been found that FLEX results in lower PEEP requirement, improved dynamic lung compliance and could better attenuate experimental ARDS compared to conventional volume-controlled ventilation (VCV) in a porcine lung injury model [ 5]. In addition, decelerating the expiratory flow improves the homogeneous distribution of ventilation in lung-healthy patients [ 6] and pigs [ 7]. In the light of these findings we hypothesized that FLEX affects the distribution of regional ventilation and provides additional recruiting effects. Therefore, we assessed the regional ventilation using electrical impedance tomography (EIT) during mechanical ventilation in the volume-controlled mode with and without FLEX in a pig model before and after induction of lung injury.

The study was approved by the Animal Welfare Committee of the University of Freiburg and carried out in compliance with the German animal protection law and the European Union Directive on the protection of animals used for scientific purposes (2010/63).

The main findings of our study are that ventilation with FLEX resulted in an elevated mean airway pressure and a lower expiratory peak flow compared to VCV with otherwise identical respiratory parameters. In terms of regional ventilation, FLEX was associated with an elevated baseline in the global and in the regional impedance curves, a shortened end-expiratory zero-flow period, a redistribution of tidal ventilation from ventral to dorsal regions and a shorter and more homogeneously distributed delay in regional ventilation. The effects of FLEX were present in the healthy lungs but even more pronounced in injured lungs. These findings support our hypothesis that FLEX is associated with lung recruiting effects.

General recruitment was most intuitively reflected in the elevated impedance curve baselines when FLEX was switched on and their immediate drops when FLEX was switched off. This occurred regardless of the health condition of the lung and was evident in all regions of interest. Similar behavior of impedance curves is seen when PEEP is increased and decreased [ 16, 19– 21]. The observed baseline shifts can therefore be attributed to an increased functional residual capacity generated by the decelerated expiration during ventilation with FLEX.

In this context, an important observation was on the lung behavior during the end-expiratory zero-flow period. During this period, convective gas movement comes to a halt and only intra-pulmonary redistribution processes occur. These however, resulted in a further decrease in the lung impedance indicating end-expiratory derecruitment, probably based on pendelluft effects and redistribution of volume out of the observation plane of the EIT measurement. With conventional VCV such derecruitment during the end-expiratory zero-flow period could only be reduced by increasing PEEP or by reducing expiration time. However, increasing PEEP would undesirably increase peak pressure and therefore increase the risk of overdistension in the non-dependent lung areas. Adjusting expiratory time might increase the risk of auto-PEEP due to incomplete expiration. By contrast, with FLEX, the end-expiratory zero-flow period per se is shorter and accordingly, the end-expiratory derecruitment was found to be less pronounced during ventilation with FLEX compared to conventional VCV. It has to be noted that the risk of generating auto-PEEP during ventilation with FLEX was thereby minimized since FLEX was set in a fashion that at the end of expiration, flow subsided completely. A second effect might be explained with respect to temporal effects of alveolar collapse, as proposed by a mathematical alveolar model [ 22], which postulates a population of alveoli that close “in a matter of seconds or less” after alveolar pressure is below closing pressure. The FLEX-induced slow pressure decrease results in longer maintenance of alveolar pressure above closing pressure during expiration. As a consequence the percentage of breath cycle time at which the alveolar pressure is below closing pressure is significantly lower compared to conventional passive expiration and consequently there is less time for the alveoli to collapse. In total, with FLEX the dynamics of expiration were fundamentally changed compared to passive expiration: the lungs stayed in tidal motion for a longer period of expiratory time, which we consider to be more physiological because it is closer to the spontaneous breathing pattern.

The behavior of the impedance baseline after switching from FLEX to conventional VCV is of special interest. In the healthy lungs, the global baseline levels during conventional VCV sequences before and after ventilation with FLEX were comparable. However, the baselines of the dorsal impedance curves were significantly lower in the second VCV sequence. This effect is not fully understood yet - one could speculate that due to recruitment a three-dimensional redistribution of ventilation occurred, which could not be visualized owing to the limitation of EIT to a single observation plane. Further studies would be needed to investigate this in greater depth. Contrary to this, both ventral and dorsal baselines were significantly elevated in the injured lungs during the period of conventional VCV following FLEX ventilation (VCV2) compared to the conventional VCV period before (VCV1). This shift provides evidence that the recruitment prevailed at least in the short term after FLEX was switched off and some recruitment could be sustained in the injured lung. However, since only 1 min of the second volume-controlled sequence was recorded, no prediction on long-term effects after FLEX ventilation can be given here and further studies would be required to investigate those.

If we assume that FRC is higher during FLEX ventilation than during conventional VCV, it is of special interest how this surplus is distributed within the lung. The fractional impedance distribution showed a redistribution of the ventilation to dorsal areas. Even though the majority of ventilated lung area remained in the ventral region, a significant shift towards increased ventilation in dependent parts of the lung occurred with FLEX. Ventral areas were less aerated during FLEX ventilation than during VCV ventilation whereas dorsal areas showed an increase in ventilation. This effect again occurred in the healthy and in the injured lungs and again it was more pronounced when the lung was injured. Probably, the healthy lungs had a lower potential for recruiting effects compared to the injured ones and hence the shift to dependent areas was smaller for this case.

Homogenizing effects of FLEX were impressively illustrated by the RVD. With FLEX, larger areas participated earlier in ventilation than during conventional VCV. In the injured lungs the effect was more pronounced. The delay in inspiration during conventional VCV is likely related to the location of the injured lung areas. The RVD maps drafted by Muders and colleagues displayed increased delay times in the dorsal lung regions due to end-tidal lung collapse when ventilating lung-injured pigs [ 18]. With FLEX ventilation, the delay is reduced in the dorsal areas, indicating faster and more homogeneous dispersal of gas in the initial inspiration phase and consequently a better recruitment state, compared to VCV.

In this study, we recorded EIT sequences over a rather short period of time. Thus, long-term effects of FLEX ventilation cannot be deduced from this setting. However, the effects of FLEX ventilation on lung recruitment set in immediately and were well observable within our time window. We did not measure blood gases, since we did not expect noticeable changes in these within our short observation window. Nevertheless, the effects of FLEX ventilation on pH and partial pressure of arterial carbon dioxide (paCO ₂) would be important in the view of long-term ventilation. A prior study suggests that paO ₂ improves and paCO ₂ is lower during 6 h of FLEX ventilation compared to VCV [ 5]. In addition, our study was restricted to using only one tidal volume, one frequency and the supine position. It would be of interest to investigate the recruiting effects of FLEX with respect to systematic variation of these parameters.

Our protocol did not contain an auto-PEEP measurement. The gold standard of auto-PEEP measurement is to apply an end-expiratory pause of appropriate length (5–10 s). Such a hold maneuver would provide prolonged static conditions. As we assume that the effects of FLEX rely on a continued dynamic situation, this would immediately diminish the effects of FLEX. Therefore, such measurement was not part of our protocol. Consequently, we cannot fully exclude the generation of a minor auto-PEEP. The calculated P ₀ values were always lower than the respective PEEP. The underestimation of PEEP (in the absence of auto-PEEP) is in accordance with the findings of Eberhard and colleagues [ 11], who attributed this to inhomogeneity and tissue stress. Both are more effective during fast expiration since the lung tissue needs more time to reach equilibrium, than during slow expiration. Correspondingly, the values for P ₀ converged with PEEP for the longer active expiration time during FLEX ventilation; they were however always below PEEP and did thus not provide evidence for a relevant auto-PEEP. Further, the mean dynamic compliance of the healthy animals was 51 ml/cmH ₂O and of animals with lung injury 16 ml/cmH ₂O. If auto-PEEP was the reason for the observed FLEX-related volume increases of 57 ml and 115 ml, an auto-PEEP of about 2 cmH ₂O or 7 cmH ₂O would be required, respectively. We would expect that an auto-PEEP of such an amount would have caused a measurable end-expiratory flow.

The technique of slowing down the expiratory phase is not entirely new. Earlier approaches have also aimed at decelerating expiration by adding expiratory resistance [ 23, 24]. Although the ideas behind these approaches are similar to that of FLEX, the constant additional expiratory resistances increased expiratory time by increasing the time constant. Incomplete expiration might therefore have prevented clinically relevant benefits [ 24, 25]. By contrast, FLEX decouples the expiration from the time constant. It causes flow deceleration only in the early expiration phase and an improved flow rate in the late expiration phase. As a consequence, the lung volume empties in a linearized fashion and an expiratory gas flow is achieved nearly throughout expiration. Thus expiration time is not necessarily prolonged.

FLEX ventilation elevated the baseline of impedance curves, shortened the no-flow phase at the end of the expiration, redistributed ventilation from ventral towards dorsal areas and homogenized the regional ventilation. The effects of FLEX were more pronounced in injured lungs. These findings support our hypothesis that FLEX ventilation has recruiting effects and indicate potential lung-protective effects of FLEX during mechanical ventilation.