Individualized flow-controlled ventilation compared to best clinical practice pressure-controlled ventilation: a prospective randomized porcine study

Ventilator-induced lung injury (VILI) is a serious complication in mechanically ventilated patients, significantly contributing to patient morbidity and mortality [1, 2]. Despite numerous efforts during the last decade to optimize standard ventilation methods in order to minimize the probability of VILI, e.g., by decreasing tidal volume (V_T) and adjusting positive end-expiratory pressure (PEEP), the incidence of VILI is still high leading to pulmonary complications postoperatively as well as in ICU patients [1‐3]. One inherent problem of artificial ventilation is the difficulty entailed in comprehensively determining individual lung mechanics from ventilation parameters and displayed measurements. In addition, setting the ventilator based on (predicted) body weight does not adequately address variations in lung mechanics [4].

Flow-controlled ventilation (FCV) is a ventilation mode in which flow is kept constant during both inspiration and expiration [5, 6]. This is a novelty in artificial ventilation. The resulting constant flow coupled with direct intratracheal pressure measurements allows much more precise analysis of individual lung mechanics than is possible in conventional ventilation modes, where flow varies over a wide range and intratracheal pressure is not directly accessible. In particular, FCV allows accurate determination of dynamic compliance and—based on this—precise adjustment of PEEP and peak pressure (P_peak), thereby ensuring that ventilation occurs in the range between the (so-called) lower and upper inflection points of each patient’s pressure-volume curve.

In addition, it has been shown that the energy applied to the lung tissue is a significant factor contributing to the development of VILI [7]. Energy dissipation in lung tissue is related to flow. Therefore, ventilation at constant flow (which avoids the high flow peaks as in conventional ventilation) leads to minimization of applied and—more importantly—dissipated energy [5]. There is growing awareness that energy dissipation during the ventilation cycle is related to VILI and thus its reduction seems to be a key factor in developing more protective ventilation strategies [6].

The aim of this study was to evaluate the applicability of individualized FCV and to examine—for the first time—the effects of individualized FCV on respiratory and metabolic parameters as well as on lung aeration in comparison to evidence-based, best clinical practice pressure-controlled ventilation (PCV) during long-term ventilation in pigs.

The experiment protocol was completed in 12 pigs.

The main finding of this study is the applicability of compliance-guided individualization of FCV settings. Tailored ventilator settings meet today’s ambitions of precision medicine and individual patient needs much better than do fixed numbers or thresholds for every patient.. We showed that compliance-guided pressure adjustment with FCV did not cause more regional overinflation of lung tissue when compared to evidence-based, low tidal volume PCV even though higher V_Ts were applied with FCV. Several studies report a significant improvement in ventilation efficiency and aeration of lung tissue when FCV is applied [12‐16]; however, these studies did not use an individualized ventilation approach. Our study shows that normocapnia was maintained in individualized FCV despite a remarkable reduction of respiratory minute volume by 50% compared to PCV. Second, individualized FCV resulted in significantly improved oxygenation and a significant reduction of non-aerated lung tissue indicating an improvement of ventilation efficiency and lung aeration.

Several factors contributing to improved ventilation efficiency were found for FCV. First, in animals ventilated with FCV dead space ventilation was significantly reduced by increasing V_T. The finding of increased V_T may be counterintuitive in terms of lung-protective ventilation at first sight since studies have clearly shown that higher tidal volumes may be associated with a negative outcome in ARDS [17]. However, the primary idea of FCV is to physically reduce dissipated energy applied to lung tissue as much as possible. The only way to minimize dissipated energy is to keep gas flow constant during the entire ventilation cycle and at the lowest possible level [5, 6]. With minimized gas flow, ventilation must be as efficient as possible; therefore, tidal volume was increased within lung mechanic limits to reduce dead space ventilation but simultaneously decrease the risk of atelectasis and/or overdistension (Fig. 1). This process is aided by the viscoelastic properties of the lung tissue—it has more time to “creep” and relieve stress in FCV. In fact, the combination of direct intratracheal pressure measurement and a constant flow allows for the first time to measure dynamic compliance during ventilation and pressure settings were adjusted accordingly. The novelty in individualized FCV is that tidal volume is naturally strongly related to individual lung compliance as a result of the individualization process, thereby representing the ventilation of the available aerated lung tissue. This can lead to a higher V_T in lung healthy individuals (as shown in this study) but would also result in decreased V_T if the compliance of an injured lung is reduced.

In addition, improved lung aeration with less atelectasis was observed in FCV animals, even though a lower PEEP level was established than with PCV. These findings agree with those of previous studies, where a recruiting effect due to a linearized expiratory airway pressure decline was supposed for FCV [12, 13]. In fact, controlling expiratory flow mimics physiological effects provided by the glottis (acting as a dynamic resistor to the egress of gas) and the diaphragm (slowing down expiratory flow by controlled muscle relaxation). As already mentioned above, the observed higher V_T when administering FCV did not increase the proportion of overdistended lung tissue. This fact underlines previous findings in non-individualized FCV showing a more even ventilation pattern [14‐16]. The observed better lung aeration and more homogeneous gas distribution improve ventilation efficiency, not only by increasing gas exchange surface and diffusion capacity, but also by reducing intrapulmonary shunt fraction and improving the ratio between ventilation and perfusion. However, further studies are needed to investigate the exact shunt fraction and ventilation/perfusion ratio in FCV.

The results of a trial in healthy piglets suggest that the energy delivered by the ventilator to the lung tissue contributes to VILI [7]. However, established calculations for mechanical power applied to the lung tissue consider only the inspiratory effort since expiration is presumed to be a passively occurring process. In earlier work on the concept of FCV, we already hypothesized that energy dissipated in the lung tissue during both inspiration and expiration is an important contributor to VILI [5, 6]. In mechanical ventilation, energy is necessarily dissipated because of resistive work that needs to be performed in order to overcome airway and tissue resistance during inspiration as well as expiration. Part of the applied energy is stored during inspiration (by expanding the (visco)elastic lung tissue and chest), can be partly recovered during expiration (by elastic recoil), and leaves the lung tissue again (during the egress of respiratory gas). By contrast, the dissipated energy (which represents the unavoidable “loss” of energy of any working mechanical system) remains in the lung tissue. In order to minimize dissipated energy, a constant flow of respiratory gas at the lowest level is most favorable [5]. Decelerating flow in PCV during inspiration and expiration results in two spikes where energy is dissipated. Contrarily, constant flow in FCV lacks any energy spikes delivered to and dissipated in the lung tissue [6]. Therefore, individualized FCV is performed not only with the greatest efficiency with respect to lung mechanics at the lowest flow guaranteeing normoventilation, but also with the lowest possible energy input to lung tissue.

Moreover, dynamic stress and strain, both undisputed contributors to VILI [17‐20], are reduced in FCV due to steady, slow changes in pressure and smooth increase or decrease in lung volume during the whole ventilation cycle at an optimally low RR (see digital content: Additional file 1).

Our study has several limitations. First, the measured technical dead space of the Evone® ventilator is 25 ml lower than that of the EvitaXL® (50 ml vs. 75 ml), contributing to a higher MV in the PCV group. Moreover, the RR required to achieve normocapnia was significantly higher with PCV. However, correction of MV with consideration of technical and RR-related dead space ventilation effects still yields a significant MV reduction in FCV animals (5.0 vs 9.6 l/min). Taking into account the significantly higher baseline MV of 2.4 l/min in the PCV group due to a slightly higher body weight (Table 1), a considerably reduced MV still results in the FCV animals (approx. 5.0 vs. 7.1 l/min). This aligns our findings of improved ventilation efficiency with previous findings [12].

Second, PEEP in the FCV group was adjusted by means of measurement of dynamic lung mechanics. In the PCV group, however, PEEP was set at 5 cm H₂O according to the recommendations of the ARDS Network [17] for low tidal volume ventilation in healthy lungs. We acknowledge, this may not represent a truly “fair fight” between FCV and PCV; however, a PEEP level of 5 cm H₂O was used in a previous preclinical study investigating the effects of FCV versus VCV making our results very roughly comparable to this trial [12]. It may be argued that oxygenation could be improved by applying a slightly higher PEEP level in the PCV group; on the other hand, we observed improved oxygenation in the FCV group despite a significantly lower PEEP level. Finally, the aim of the current study was to compare ventilation based on evidence-based fixed numbers with settings based on an individualized approach.

Although more recent studies suggest using personalized ventilator settings based on electrical impedance tomography measurements and repeated CT scans [4], this approach was not feasible in our study due to a lack of resources and does not reflect clinical routine nor was it possible to perform PEEP titration guided by transpulmonary pressure due to anatomical issues related to the measurement of esophageal pressure (P_es) as a surrogate parameter for pleural pressure (Table 1). The location of the esophagus is far more dorsal in pigs than in humans and therefore P_es reflects only pleural pressure at the very dependent parts of the lungs and would have led to unacceptably high PEEP titration in order to keep transpulmonary pressure positive. Besides the fixed PEEP level in PCV, no recruitment maneuvers were performed either with PCV or FCV. Recruited lung tissue might have led to redistribution of the fixed tidal volume in PCV and thus a more homogeneous distribution of transpulmonary pressure through interdependence properties and, after an increase of lung compliance, a reduction of driving pressure. In the absence of recruitment maneuvers, it has to be stated that ventilator settings based on inspiratory pressure-volume curves may not provide enough information to guarantee safe mechanical settings.

Third, it is not possible to directly quantify atelectasis in the Hounsfield unit (HU) distribution, because atelectatic tissue lies in the same HU range as blood vessels and soft and fat tissue (− 100 to 100). Therefore, technically it is not possible to distinguish between atelectasis and non-alveolar lung tissue. However, since the amount of non-alveolar lung tissue inside the thoracic cavity should be roughly equal in healthy pigs, we hypothesize that the significant differences seen in the CT scan presentation are attributable to atelectasis. Additionally, CT images were taken during a hold maneuvers representing a static CT acquisition of a dynamic process; this fact is even more important for FCV than for PCV as in FCV normally any pause during ventilation is absent.

Finally, this study investigated lung healthy pigs with a comparable low compliance compared to humans. Thus, accepted VILI indicators such as tidal volume, driving pressure, and mechanical power are not fully transferable to humans, especially not under different pathophysiological conditions such as ARDS.

This porcine long-term ventilation study demonstrates the applicability of compliance-guided individualization of FCV settings. Additionally, compared to low tidal volume PCV, more efficient gas exchange with improved oxygenation, more efficient CO₂ elimination, and more homogeneous gas distribution without signs of overinflation in FCV were confirmed. Our results for individualized FCV and the underlying concept for reducing dissipated energy (according to accepted thermodynamic considerations in the technical field) and minimizing stress and strain are suggestive of a more lung-protective ventilation strategy than is the current best clinical practice PCV. Further studies and clinical trials are consequently needed to determine and investigate long-term effects of individualized FCV under different pathophysiological conditions.