The main findings of our study were: (1) ICP was elevated by increased PEEP in both normal ICP and intracranial hypertension conditions in animals with normal blood volume (Series I and II); (2) Increasing PEEP resulted in decreased ICP if the animals were volume-depleted (Series III); (3) Increasing PEEP resulted in decrease in PtiO2 in either normovolemic or hypovolemic conditions (Series I–III); (4) The impact of PEEP on hemodynamical parameters, ICP and PtiO2 became more evident when ECW was increased (Series I–III); (5) Compare to normovolemic condition, common carotid arterial blood flow was further lowered when PEEP was raised in the condition of hypovolemia, which was probably the cause of lowered ICP and PtiO2 (Series IV).
As shown by Chapin et al., the transmission of airway pressure to pleural pressure depends on the relative compliance of the lung and the chest wall. The transmission is less “efficient” with increased EL and decreased ECW [19]. This means that the applied PEEP in ARDS (a common condition of increased EL) is less likely to be transmitted into the pleural space, and thus has less impact on central venous pressure and cerebral venous return. In other words, although a high PEEP is usually applied in the treatment of ARDS, the pulmonary disease per se might be a “protective” factor against the deleterious effects of PEEP to the brain. However, EL is not routinely measured in clinical settings; instead, ERS was often considered synonymous of EL. One should keep in mind that ERS and EL are not interchangeable because ECW is also a key determinant of ERS. Our data from Series I to III clearly demonstrated that an increased ECW resulted in a more considerable influence of PEEP on ICP. Our finding suggests that the possible underlining conditions which can increase ECW (e.g., intra-abdominal hypertension due to gastrointestinal retention, or the need for rib fixation due to traumatic rib fractures, which are not rare in clinical settings) must be ruled out before a high PEEP is applied in brain-injured patients.
In contrast to the normovolemic conditions (Series I and II), raising PEEP resulted in decreased ICP if the animals were volume-depleted (Series III). This can be explained by the influence of PEEP on CO and thus cerebral perfusion. As proposed in a conceptual framework that the two important determinants of the impact of elevating PEEP on ICP are the intracranial compliance and the “net” change of cerebral blood volume (CBV); the latter is determined by the balance between the arterial inflow (regulated by CO, cerebral autoregulation, PaCO2, etc.) and venous outflow (regulated by PEEP, Starling resistor, etc.). Increased PEEP and subsequent increased pleural pressure impede the global venous return to the heart, lead to a decrease in CO and cerebral perfusion [22, 23]. The effect is more profound in the volume-depleted condition [24], as was observed in Series III. In this case, the influence of decreased CO (which reduces arterial inflow and decreases CBV) exceeded the influence of decreased venous return (which reduces venous outflow and increases CBV), which resulted in a negative “net” change of CBV and eventually a decreased ICP.
Mean arterial pressure (MAP) and CPP were also lower when high PEEP levels were applied. In animals with intact cerebral autoregulation, vessels in the brain can maintain a constant cerebral blood flow via regulating the vascular tone throughout a wide range of MAP or CPP [25, 26]. In the present study, CPP dropped beyond the lower limit of autoregulation when high PEEP levels were selected in animals received exsanguination (Series III). Therefore, the most possible cause of lowered ICP and PtiO2 was decreased cerebral perfusion. More data was obtained to support the concept in Series IV. We measured the blood flow in the common carotid artery and a significantly greater decrease of blood flow was observed after exsanguination when the same PEEP was applied. Take all data together, we can conclude that PEEP can affect ICP predominantly via its hemodynamical effects, which always functions on both the arterial (perfusion) and venous (returning) sides simultaneously. And here, increased ECW amplifies the effects of PEEP.
PtiO2 was also measured in the present study, and we found that increasing PEEP resulted in decreased PtiO2 in either normovolemic or hypovolemic conditions. The impact of PEEP on cerebral oxygenation was not well understood yet. Results from previous studies are variable [3, 9, 27]. In traumatic brain injury patients with ARDS, a progressively raising of PEEP (from 5 to 15 cmH2O) improved PtiO2 [9]. In an experimental study, a raising-PEEP strategy was carried out in healthy pigs where no impact on PtiO2 was observed. The conflicts of our data to previous studies can be explained as follows: assuming a relatively constant O2 consumption, the change of PtiO2 should be the consequence of changed O2 delivery. Thus, the two key determinants of O2 delivery to the brain were CO and arterial O2 saturation. Unlike Nemer’s study where increasing PEEP resulted in a considerable improvement in P/F ratio (from 108.5 to 203.6 mmHg), the improvement in P/F ratio was less obvious (although statistically significant, Additional file 1: Fig. S1–S3) in the present study; moreover, O2 saturation was 100% already and thus further increase was not possible since the FiO2 was 1.0 in the first place. On the other hand, CO was decreased following PEEP increments. Therefore, PtiO2 decreased due to a decreased CO.
It is also interesting to know the influence of increased PEEP on ICP and other parameters in animals with normal ICP (like in Series I) but with depleted blood volume (like in Series III). In fact, we did perform some pilot experiments to exam this. Briefly, a crossover rather than a randomized-control experiment was completed in four pilot animals with normal ICP and depleted blood volume (two from chest wall strapping to control condition and the other two in the inverse order). ICP decreased with the increment of PEEP in all animals, but the magnitude was small (from 8.5 to 6.8 mmHg in the control condition and from 7.1 to 4.8 mmHg in the chest wall strapping condition with PEEP increment, Additional file 2: Fig. S4). The reason we did not further perform a randomized-control study with a larger sample size were: first, the trend of ICP change in the pilot study was the same as we observed in Series III, and the underlying pathophysiology rationale could be the same; second, considering the small difference in the change of ICP between the two conditions, large sample size is required to obtain a statistical significance, which may deviate from the ethical requirements of animal experiments. Therefore, we finished only one randomized-control study (Series III) to exam the impact of hypovolemia and speculated that the conclusion might also be true in animals with normal ICP, although without additional experiments.
Our study has some limitations: (1) The data were obtained with experiments of short duration; therefore, the long term effects of PEEP were still unclear; (2) Blood flow was measured at the common carotid artery rather than the internal carotid artery (which perfuse the brain). In the present study, we used a Vevo 3100 ultrasonic system with an ultra-high-frequency probe designed for small animal research. The ultra-high-frequency probe provided high resolution while the maximal scan depth was limited. The internal carotid artery was too deep to display by using the current device, so we chose common carotid artery instead; (3) Although a decreased cerebral venous return was proposed as the main reason for increased ICP, the outflow was not measured in this study. We measured arterial blood flow via Doppler ultrasound; however, using the same methodology to assess cerebral venous return was suggested to be inaccurate [28]; (4) Indeed, ICP has four components: besides arterial blood inflow and venous blood outflow discussed above, cerebrospinal fluid (CSF) circulation and volumetric changes of brain tissue or contusion volume also play an important role. We did not directly measure the change of CSF circulation and brain tissue volume in the present study. Considering the short experimental period, brain tissue volume can be assumed unchanged, while the change of CSF circulation was not investigated in a quantitative way but a qualitative way instead (Series I vs. Series II). In other words, the four components of ICP were not investigated in isolation and the quantitative way in the present study and thus may limit our interpretation of the effects of a specific factor. Moreover, several other factors that may have an impact on these four components, such as pulse amplitude, respiratory wave amplitude and pressure reactive index (as a surrogate of cerebral autoregulation) were not considered in the present study. (5) The present study investigates only two different volume statuses (normovolemia vs. massive blood loss) while intermediate states, for example, mild hypovolemia due to diuretics or hyperosmolar agents, or small amount of blood loss, are more likely to occur in the clinical setting. Further studies are necessary to confirm the effects of PEEP in these situations.