Intracranial pressure responsiveness to positive end-expiratory pressure is influenced by chest wall elastance: a physiological study in patients with aneurysmal subarachnoid hemorrhage

Acute lung injury is prevalent in patients with acute brain injury [1‐4]. Mechanical ventilation is needed in this population, and positive end-expiratory pressure (PEEP) is often used to improve oxygenation and prevent or recruit alveolar collapse [5‐7]. However, there have long been concerns that the use of PEEP in brain-injured patients could reduce cerebral perfusion pressure due to both increased intracranial pressure (ICP) and decreased mean arterial pressure [8, 9]. Previous studies yielded inconsistent effects of PEEP on ICP [10] and diverse individual responsiveness [11‐13], suggesting that the mechanism of action might be multifactorial. Several possible determinants for the influence of PEEP on ICP have been proposed, including baseline ICP [14], intracranial compliance [15, 16], respiratory mechanics [16, 17], dead space change and alveolar recruitment by PEEP [18].

Theoretically, PEEP may increase ICP by reducing venous return via elevating intrathoracic pressure [9]. Thus, ICP responsiveness to PEEP might largely depend on pressure transmission from the lung to the pleural cavity, which is determined by the distribution of lung elastance (E_L) and chest wall elastance (E_CW) in respiratory system elastance (E_RS). Although clinical studies have suggested that an increased E_RS might attenuate the effect of PEEP on ICP, E_L and E_CW were not differentiated in these studies [16, 17]. A given increased E_RS might mainly contribute to the increase in E_L due to pulmonary causes, such as acute respiratory distress syndrome (ARDS), or the increase in E_CW due to the chest wall impairment, such as intra-abdominal hypertension or massive pleural effusion [19]. In mechanical ventilated patients, the reported E_CW to E_RS ratio (E_CW/E_RS ratio) varied from 0.2 to 0.8 [19]. To date, no study has been performed to determine the different effects of E_L and E_CW on ICP responsiveness to PEEP, which is worthy of investigation.

We hypothesized that the patients with high E_CW or a high E_CW/E_RS ratio might exhibit more significant ICP responsiveness to PEEP. In the present study, patients with aneurysmal subarachnoid hemorrhage after clipping surgeries were enrolled, and two different levels of PEEP were applied. Changes in ICP after PEEP adjustment were monitored. Esophageal pressure was measured as a surrogate for pleural pressure, and the distribution of E_L and E_CW in E_RS was determined. The aim was to investigate the possible influencing factors in ICP responsiveness to PEEP, especially for E_CW and the E_CW/E_RS ratio.

From February to November 2016, 30 patients were studied. No emergent termination occurred during the procedures. Table 1 presents the main characteristics of the enrolled patients at the low (5 cm H₂O) and high (15 cm H₂O) PEEP levels. After the PEEP level was increased, ICP significantly increased (p <  0.001). The median (IQR) change in ICP was 2.5 (1.0–4.0) mm Hg. According to the predefined criterion, each of the 15 patients was allocated to the “responder” group or the “non-responder” group, which had respective ICP changes of 4.0 (3.0–5.0) and 1.0 (1.0–2.0) mm Hg. Baseline characteristics were comparable between the two groups except for responders with an older age and a lower body mass index (Table 2).

Detailed data of respiratory mechanics, ICP and hemodynamic parameters and blood gas analysis are presented in the Additional file 2 (Table E1 to E6).

Esophageal pressure at end-expiratory occlusion increased significantly after PEEP adjustment (5.4 [4.1–8.2] versus 9.5 [6.8–12.2] cm H₂O, p <  0.001), and the magnitude of the increase was significantly higher in the responder group than in the non-responder group (4.1 [2.7–4.1] versus 2.7 [0.0–2.7] cm H₂O, p = 0.033, Fig. 1). At the low PEEP level, there were no significant differences in E_RS (p = 0.136) and E_L (p = 0.863) between two groups, but E_CW and the E_CW/E_RS ratio were significantly higher in responders than in non-responders (p = 0.021 and 0.017) (Fig. 2a). As the PEEP level increased, no significant difference existed in the magnitude of change in E_RS, E_L, and E_CW between the two groups (p = 0.468 to 0.787, Fig. 2b). Thus, significant differences in E_CW and the E_CW/E_RS ratio between the two groups remained at the high PEEP level (p = 0.011 and 0.025, Fig. 2c).

No significant difference was found in baseline ICP (3.0 [2.0–6.0] versus 4.0 [4.0–10.0] mm Hg, p = 0.163) and intracranial ventricular compliance (1.0 [0.7–1.0] versus 1.0 [0.7–1.0] mL/mm Hg, p = 0.723) between the ICP responder and non-responder groups. Although central venous pressure increased significantly after the PEEP level was increased in all enrolled patients (Table 1, p <  0.001), no significant difference was observed in the change of central venous pressure between ICP responders and non-responders (5.0 [4.0–5.0] versus 3.0 [1.0–6.0] mm Hg, p = 0.077, Fig. 3). Similarly, significant decreases in mean blood pressure were observed after PEEP increase in all enrolled patients (81.0 [75.6, 92.1] versus 76.2 [67.4, 90.0], p = 0.002), but no difference in the change of mean blood pressure between groups (Table E3 and E4). Cerebral perfusion also decreased after PEEP was increased, the responder had a significantly greater decrease in cerebral perfusion pressure than the non-responder (− 13.0 [− 18.3, − 6.7] versus − 6.7 [− 11.3, 1.0], p = 0.011; Table E3 and E4).

There was no significant difference in PaO₂, PaO₂/FiO₂ and PaCO₂ after PEEP adjustment (p = 0.052 to 0.061, Table 1). P_ETCO₂ significantly decreased (29.5 [27.8–31.2] versus 29.0 [25.8–30.3] mm Hg, p = 0.003); thus, V_Dalv/V_T significantly increased (13.2 [8.5–19.9] versus 18.8 [11.9–27.7] %, p <  0.001). No significant differences were observed in P_ETCO₂ and V_Dalv/V_T between ICP responders and non-responders at either the low or high PEEP levels.

The main finding of our study was that patients with a greater ICP responsiveness to increased PEEP exhibit a higher E_CW and E_CW/E_RS ratio. This finding was consistent with our hypothesis.

Regarding the influence of PEEP on ICP, clinical studies have reported conflicting results, with ICP increasing [14‐18], not markedly changing [25‐27] or even decreasing [28] after the application of PEEP. In patients with increased ICP, the effect of PEEP on ICP becomes evident whenever the elevation of the intrathoracic pressure induced by applied PEEP exceeds the ICP [14]. This mechanism was referred to as the Starling resistor model, which represented the role of damaged venous outflow in the response of ICP to PEEP [29]. In a study conducted by McGuire et al., 15 mmHg was used as the upper limit of the normal baseline value of ICP at zero PEEP [14]. ICP increased at PEEP levels of 10 and 15 cm H₂O in patients with normal baseline ICP but did not significantly change at PEEP levels of 5 to 15 cm H₂O in patients with elevated baseline ICP. In our group of patients, all ICP values were less than 15 mmHg at 5 cm H₂O PEEP (4.0 [2.0–10.0] mm Hg) and increased after the PEEP level was adjusted to 15 cm H₂O (7.0 [5.8–11.0] mm Hg). The difference in ICP at the two PEEP levels (2.5 [1.0–4.0] mm Hg) in our patients was comparable to those previously reported [13, 14].

The PEEP-induced increase of pleural pressure may reduce cerebral venous drainage and eventually increase ICP [9]. The pleural pressure serves as an intermediate link from the lung to the cranium. In the present study, we measured esophageal pressure as a surrogate of pleural pressure and found that esophageal pressure increased more significantly in ICP responders than in non-responders (Fig. 1). This finding suggested a role of pressure transmitted to the pleural cavity in ICP responsiveness. In addition, central venous pressure increased significantly at the high PEEP level (Table 1), indicating the impairment of venous return by PEEP, which eventually resulted in reduced mean blood pressure and cerebral perfusion pressure. ICP responders had a higher tendency to have elevated central venous pressure (Fig. 3); however, statistical significance was not achieved, potentially due to the limited sample size.

The elastic properties of the respiratory system may be a key factor in determining the effect of PEEP on pleural pressure and subsequently ICP [9]. Pressure transmission from the lung to the pleural cavity depends on the distribution of E_L and E_CW in E_RS. The higher the proportion of E_CW is, the greater the impact of PEEP on pleural pressure [19]. Therefore, E_CW and the E_CW/E_RS ratio may be a more important determinant of ICP responsiveness to PEEP. In previous clinical studies, only E_RS was used to explore the influence of respiratory mechanics in the PEEP and ICP relationship [16, 17]. After distinguishing E_L and E_CW from E_RS, we found that the effect of PEEP on ICP was more profound in patients with higher E_CW and a higher E_CW/E_RS ratio (Fig. 2). Although patients with brain injury exhibited higher E_CW (without detailed values) in previous studies [30], E_CW and the E_CW/E_RS ratio measured in our group of patients were relatively “normal” (Table 1). Our results were comparable to those reported in patients under general anesthesia [31] but higher than those reported in ARDS patients [32]. However, even a slightly higher E_CW and E_CW/E_RS ratio might contribute to ICP responsiveness (Fig. 2). Our data suggested the importance of separate E_L and E_CW monitoring during the clinical application of high PEEP in brain-injured patients.

There are two consequences after the application of PEEP: recruitment of collapsed alveoli and/or overdistention of normal alveoli. In patients with severe brain injury and acute lung injury or ARDS, Mascia et al. defined two groups based on recruitment volume: ≥ 110 mL as recruiters and < 110 mL as non-recruiters [18]. As the PEEP level increased, E_RS and PaCO₂ significantly increased in non-recruiters, indicating hyperinflation. On the contrary, E_RS significantly decreased, and PaCO₂ remained unchanged in recruiters. ICP increased only in non-recruiters. In their patient population, the PaO₂/FiO₂ ratio was 95–141 mmHg, and E_RS was 24–25 cm H₂O/L at 5 cm H₂O PEEP [18]. In our group of patients, only six ARDS cases (20%) were enrolled, and the overall severity of lung injury was mild, as represented by the median PaO₂/FiO₂ ratio of 264 and an E_RS of 15.3 cm H₂O/L at the low PEEP level. Alveolar collapse might also be minor; therefore, increases in the PEEP level produced increases in E_RS (15.3 to 20.1 cm H₂O/L) and V_Dalv/V_T (13.2 to 18.8%). Despite lacking statistical significance, PaCO₂ and V_Dalv/V_T were high in ICP responders. Therefore, PEEP-induced hyperinflation might also contribute to ICP responsiveness in our patients.

The effect of PEEP on ICP might be affected by ventricular compliance [15, 16]. Apuzzo et al. reported that a significant increase in ICP was only observed in patients who manifested increased cerebral elastance when PEEP was applied [15]. Similarly, Burchiel et al. reported that PEEP increased ICP in patients with decreased intracranial compliance and that decreased lung compliance may buffer this effect [16]. In the present study, given that patients with normal ICP were enrolled and that cerebrospinal fluid was continuously drained before the start of the study, the ventricular compliance was “normal” in all patients. Additionally, for ethical reasons, we cannot measure the ventricular compliance by injecting normal saline into the ventricle instead of draining the cerebrospinal fluid. Because intracranial hypertension and ventricular compliance impairment are prevalent in patients with severe brain injury, further animal experiments should be performed to validate the role of ventricular compliance in the ICP and PEEP relationship.

Increase in PEEP decreases systematic venous return and potentially lowers the cardiac output, this will lower blood pressure and cerebral perfusion pressure. If the cerebral perfusion pressure break through the lower limit of cerebral autoregulation, cerebral perfusion will not be maintained and thus decrease cerebral blood flow, cerebral blood volume and ICP. In this case, the decrease of cerebral perfusion (and thereby decrease in ICP) will somehow contaminate the effect of PEEP on the venous side. In our study, blood pressure and cerebral perfusion pressure were maintained in a normal range, even in a high PEEP level. Therefore, although without cerebral blood flow being measured, we assumed that the cerebral perfusion was comparable between the two groups.

The responders were significant older in this study. It has been reported that the elders have higher E_CW due to structural changes to the thoracic cage [33, 34], which is the result of age-related osteoporosis and calcification of the rib cage that reduce the ability of the thoracic cage to expand. Our data also showed higher E_CW in the elders, probably due to the same reason. The responders also had significantly lower BMI. Although obesity people have higher E_RS, it can be the result of either increased E_L or increased E_CW, or even both [35]. This makes the E_CW unpredictable simply base on BMI. Moreover, our patients had relatively normal BMI, the small difference of BMI between the two group (although statistically significant) is unlikely the determinant of the difference of E_CW.

The main strength of the present study was that the contribution of the chest wall and the lung in ICP responsiveness to PEEP was distinguished. The adverse effect of PEEP on ICP may be amplified in patients with higher E_CW and a higher E_CW/E_RS ratio. Our data suggest the potential necessity of respiratory mechanics monitoring when applying PEEP in brain-injured patients, especially in patients with a risk of chest wall impairment.

There were several limitations of the study. First, we used the median of the increased value rather than other physiological parameters to define the ICP responders and non-responders. Given that no widely accepted standard is available to discriminate whether the patient’s ICP is responsive to increased PEEP, the division of patients into two groups is reasonable and enables us to compare potential determinants between the two groups. Second, we only included postoperative subarachnoid hemorrhage patients with relatively normal ICP and respiratory mechanics. Therefore, our results might not be directly applied to other populations. Third, although esophageal pressure has been employed as a surrogate of pleural pressure, the use of the absolute value of esophageal pressure remains questionable [19]. However, in the calculation of E_CW, we used the change in esophageal pressure between end-inspiration and end-expiration occlusion, which has been shown to reliably reflect the change in pleural pressure [23, 36, 37].

Patients with greater ICP responsiveness to increased PEEP had higher E_CW and a higher E_CW/E_RS ratio. Potential factors contributing to the increase in E_CW and the E_CW/E_RS ratio should be assessed and eliminated if possible to avoid the adverse effects of high PEEP levels on the brain.