Low tidal volume pressure support versus controlled ventilation in early experimental sepsis in pigs

In the course of non-pulmonary sepsis respiratory failure is a common cause and occurs in about 50% of the patients with severe sepsis [1]. Patients suffering from sepsis often require mechanical ventilation, even if they do not fulfill the criteria of an acute respiratory distress syndrome (ARDS). On the other hand, mechanical ventilation itself can represent the second hit leading to the development of ARDS. Independent from the underlying illness, lung protective strategies that aim to minimize ongoing pulmonary damage by targeting low tidal volumes and limitation of the inspiratory pressure are regarded as the key interventions when the criteria of ARDS are met [2],[3]. In severe ARDS there is evidence that short-term neuromuscular blockade enables the consequent realization of low tidal volume ventilation and increases survival rates [4]. In mild to moderate or post-acute ARDS, however, the admittance of spontaneous breathing is reported to improve gas exchange, reduce diaphragmatic dysfunction and enable a faster weaning [5],[6]. Several experimental models report beneficial effects of spontaneous breathing in various patterns on gas exchange, edema formation, and lung injury [7]-[9]. These findings, though, have not been verified in primary sepsis-related lung injury. Furthermore, some clinical and experimental data also suggest the value of preventive initiation of lung protective ventilation [10]. But currently the appropriate guidelines do not state on the preemptive application of lung protective ventilation or spontaneous breathing in early sepsis-induced lung injury [11].

We hypothesized that in early sepsis immediate application of pressure support ventilation (PSV) targeted to a tidal volume (V_t) of 6 ml kg^-1 will improve the pulmonary function in comparison to conventional volume controlled ventilation (VCV, V_t 6 ml kg^-1). Hence, we compared the early effects of PSV and VCV on gas exchange, ventilation/perfusion distribution and histopathological lung injury in a porcine model of systemic, lipopolysaccharide (LPS)-induced sepsis subsequently leading to lung injury.

The study was approved by the State and Institutional Animal Care Committee (Landesuntersuchungsamt Rheinland-Pfalz, Koblenz, Germany; approval number: G10-1-004). 18 juvenile pigs (Sus scrofa domestica, weight 27 ± 2 kg) were examined in a prospective-randomized setting.

The study protocol was completed in all 18 animals. Table 1 summarizes the ventilatory and hemodynamic data and shows no intergroup differences at baseline.

The present study features the following main findings: in a porcine model of exclusively sepsis-related lung injury the immediate initiation of low V_t-PSV was feasible, but not superior to low V_t-VCV in terms of gas exchange, respiratory pattern, hemodynamic stability, and did not improve histopathological parameters over six hours.

Sepsis is one of the most frequent risk constellations of ARDS. Systemic LPS exposition triggers inflammation by releasing pro-inflammatory cytokines and leucocyte accumulation. The early response in experimental models is characterized by acute leucopenia and immense cytokine levels. Hemodynamic findings include decreased systemic blood pressure and pulmonary arterial hypertension [17],[18]. Our model adequately reproduces these common early findings. Following central venous LPS infusion the lungs are the first microcirculatory bed to pass. Nevertheless, short-term LPS exposition hardly leads to immediate or persisting ARDS [19],[20]. On the other hand, occurrence of severe hemodynamic failure or septic shock conditions limits the systemic LPS application in experimental models. We therefore chose a two-staged infusion regime that caused significant gas exchange deterioration. The reduced maintenance dosage of LPS may also account for a clinically relevant reduction of bacteremia due to therapy. In contrast, primary pulmonary models like bronchoalveolar lavage or acid aspiration rapidly generate atelectasis, reduced lung compliance and gas exchange impairment directly from the beginning on.

The admittance of assisted spontaneous breathing activity can improve the pulmonary function in comparison to conventional lung protective ventilation [6]. Beneficial effects of PSV from several experimental studies amongst others include improved pulmonary blood flow redistribution and overall gas exchange, as well as attenuation of lung injury and IL-6 levels [8],[21]. Sophisticated variable pressure support ventilation has shown the potential to further increase these effects in several experimental studies [7],[8],[22] and is currently tested for clinical application [23],[24]. The early use of PSV can decrease sedation requirements, improve the cardiopulmonary function and V_A/Q matching. The number of days under mechanical ventilation on can also be reduced [25]-[28]. However, more severely lung injured patients tend to respond poorly to PSV [29]. Additionally, it is reported that early PSV increases patient-ventilator asynchrony [6],[30]. This effect may lead to high and harmful tidal volumes even in the early phase of ventilation [31].

Our present findings do not reproduce a significant improvement of gas exchange or lung injury. Ventilation was more homogenously distributed between the central and non-dependent compartment during PSV, but merely a non-significant amount of high V_A/Q compartments indicating hyperinflation developed in the VCV group. This is a considerable contrast to the upper mentioned results. However, it is worth to take model dependent characteristics into account: for most experimental studies focusing on various assisted spontaneous breathing modes the bronchoalveolar lavage/surfactant-depletion model was used. Furthermore, the assumed mechanisms that mitigate lung injury through spontaneous ventilation vary between several studies and are not fully elucidated [6]. The lavage model immediately induces atelectasis due to surfactant depletion, which are relative easy to recruit in the early phase [29], whereas LPS-injured lungs tend to respond poorly to recruitment strategies [32]. In a rabbit model of lavage-induced lung injury PSV was beneficial only in mild ARDS, though aggravated a pre-existing severe lung injury [33]. The present data, in this context, assume that not only the severity but also etiology and pathophysiologic considerations may considerably influence the response to early PSV. Furthermore, PSV was started before outright fulfillment of ARDS criteria in our model.

Endotoxemia alone causes a relatively moderate V_A/Q impairment in the short run [34]. This is reflected in the moderate V_A/Q changes in our data. Interestingly, previous data from a porcine sepsis model combined with non-protective ventilation reported a shunt fraction of 17.3 ± 7.5 [34] without occurrence of low V_A/Q units, whereas low V_A/Q units represent the predominant pattern of impairment during our low V_t modes (Figure 5). Furthermore, LPS but not bronchoalveolar lavage compromises the hypoxic pulmonary vasoconstriction [35], which should partially compensate the gas exchange deficit. Supporting our results, LPS administration did not significantly affect the ventilation distribution, but only influenced the perfusion pattern [36].

The present study has some limitations: the group sizes were adapted to previous publications that showed beneficial effects of PSV in early ARDS without a prior power analysis. Due absence of a clear-cut trend, however, it appears unlikely that the lack of effect is essentially influenced by the group sizes. We applied standard PSV and not promising but sophisticated spontaneous ventilation approaches like variable, proportional or neutrally adjusted ventilation [6],[37]. However, the initiation of early PSV in beginning sepsis should be feasible almost anywhere, not just in specialized intensive care units. Several studies showed that spontaneous ventilation attenuates histopathological lung injury in mild to moderate ARDS models [7],[8],[33]. But inflammatory response was only slightly altered in comparison to conventional lung protective ventilation [8], while gene expression analysis yielded no significant differences in pulmonary mRNA expression of inflammatory marker genes [7]. With regard to the reported model characteristics and ongoing LPS exposition, which is documented in the high plasma cytokine levels (Figure 3), significant variances in tissue contents of inflammatory markers are highly improbable without the presence of an anti-inflammatory agent over six hours. The present study was designed to focus the early phase of sepsis with a developing lung injury. If possible effects over six hours proceed towards improved long-term outcome, is merely speculative. Nevertheless, adequate identification and selection of patients may considerably influence the effectiveness of early PSV.

In a porcine model of early LPS-induced lung injury direct initiation of low V_t-PSV did not improve pulmonary function or affect lung injury in comparison to low V_t-VCV within six hours. This is a contrast to several studies that report beneficial effects of assisted spontaneous breathing modes in non-septic experimental models of mild to moderate ARDS. Early response to PSV in ARDS seems to be determined not exclusively by severity but also by etiology of the developing lung injury.