The effect of increased positive end expiratory pressure on brain tissue oxygenation and intracranial pressure in acute brain injury patients

Cerebral hypoxia is an important cause of secondary brain injury. Improving systemic oxygenation may increase brain tissue oxygenation (PbtO2). The effects of increased positive end-expiratory pressure (PEEP) on PbtO2 and intracranial pressure (ICP) needs to be further elucidated. This is a single center retrospective cohort study (2016–2021) conducted in a 34-bed Department of Intensive Care unit. All patients with acute brain injury under mechanical ventilation who were monitored with intracranial pressure and brain tissue oxygenation (PbtO2) catheters and underwent at least one PEEP increment were included in the study. Primary outcome was the rate of PbtO2 responders (increase in PbtO2 > 20% of baseline) after PEEP increase. ΔPEEP was defined as the difference between PEEP at 1 h and PEEP at baseline; similarly ΔPbtO2 was defined as the difference between PbtO2 at 1 h after PEEP incrementation and PbtO2 at baseline. We included 112 patients who underwent 295 episodes of PEEP increase. Overall, the median PEEP increased form 6 (IQR 5–8) to 10 (IQR 8–12) cmH2O (p = 0.001), the median PbtO2 increased from 21 (IQR 16–29) mmHg to 23 (IQR 18–30) mmHg (p = 0.001), while ICP remained unchanged [from 12 (7–18) mmHg to 12 (7–17) mmHg; p = 0.42]. Of 163 episode of PEEP increments with concomitant PbtO2 monitoring, 34 (21%) were PbtO2 responders. A lower baseline PbtO2 (OR 0.83 [0.73–0.96)]) was associated with the probability of being responder. ICP increased in 142/295 episodes of PEEP increments (58%); no baseline variable was able to identify this response. In PbtO2 responders there was a moderate positive correlation between ΔPbtO2 and ΔPEEP (r = 0.459 [95% CI 0.133–0.696]. The response in PbtO2 and ICP to PEEP elevations in brain injury patients is highly variable. Lower PbtO2 values at baseline could predict a significant increase in brain oxygenation after PEEP increase.


Study population
We screened adult (> 18 years) patients admitted to the ICU due to an acute brain injury.Inclusion criteria were: (a) traumatic brain injury (TBI) or aneurysmal subarachnoid hemorrhage (aSAH) patients under controlled mechanical ventilation; (b) the presence of ICP and PbtO 2 monitoring within the first 48 h of admission and (c) patients underwent at least one increase in PEEP levels concomitantly with the use of neuromonitoring.Patients with imminent death were excluded.

Patients' management
We followed the current guidelines for the management of TBI 31 and SAH 32 ; invasive multimodal monitoring, including PbtO 2 , was implemented according to recent consensus 1 .PbtO 2 monitoring is considered as "standard of care" for ABI patients with a Glasgow Coma Score (GCS) < 9 and requiring intracranial pressure (ICP) monitoring; PbtO 2 probes were placed in the frontal region of the hemisphere at greatest risk for secondary brain injury.In TBI patients, probe was placed close < 5 cm to the most injured/contused area in TBI; in SAH patients probes were placed in the region at risk or with demonstrated delayed cerebral ischemia for aSAH).Probes were inserted through a frontal burr hole using a triple-lumen bolt.PbtO 2 was monitored continuously using a specific probe and the Integra Licox® Brain Tissue Oxygen Monitoring System (IM3.ST_EU, Integra LifeSciences Corporation, Plainsboro, NJ, USA).Probe location was confirmed by a cerebral CT-scan performed within 24 h from neuromonitoring placement.The adequate functioning of the probe was tested with a 100% oxygen fraction (FiO 2 ) test for 15 min (i.e. an increase of at least 5 mmHg of PbtO 2 indicated an adequate catheter function) .Intracranial pressure probes were placed either intraventricular or intraparenchymal.

PEEP trials
PEEP was increased according to the decision of the critical care team responsible for the treatment of the patient.There was no standardized protocol.However, PEEP was increased in increments of 2 mmHg for at least 1 h-aiming at improving systemic oxygenation (targeting an improvement in arterial partial pressure of oxygen and inspired fraction of oxygen ration) and respiratory mechanics while minimizing ventilator induced lung injury 33 by implementing a lung protective ventilation strategy and targeting Plateau pressure < 30 mmHg and driving pressure < 15 mmHg 34 .If after one 1 h of PEEP incrementation there was no benefit in oxygenation or if there were deleterious effects on respiratory mechanics or persistent hemodynamic instability, PEEP was reversed to baseline setting.The decision on the amount of PEEP to incrementation was done by the respiratory therapist in agreement with the treatment treating physician.No FiO 2 changes were implemented during the PEEP trial.

Data collection
Physiological variables, ICP and PbtO 2 were measured in real-time and collected prospectively on a patient data monitoring system.Cerebral perfusion pressure (CPP) was calculated as the difference between MAP and ICP; MAP was zeroed at the level of the left atrium.Intracranial hypertension was defined as ICP value above 20 mmHg for at least 5 min at any time.Brain tissue hypoxia was defined as a PbtO 2 < 20 mmHg for at least 5 min.
Baseline (T0) was defined as the hour immediately preceding a PEEP change and T1 was defined as the first 1 h with the new stable PEEP value.The 60-min mean of value of the following continuous variables was collected at T0 and T1: mean arterial pressure (MAP), heart rate (HR), ICP, CPP and PbtO 2 .ICP and PbtO 2 were recorded prospectively every minute.Data was extracted from the monitoring system and the mean value of ICP and PbtO 2 was calculated for every hour of monitoring.
ICP and PbtO 2 changes (ΔICP and ΔPbtO 2 ) were calculated as the difference between ICP or PbtO 2 at T1 and T0.A "decrease" in ICP or PbtO 2 was identified as a ΔICP/ΔPbtO 2 < 0; a "stable" value as a ΔICP/ΔPbtO 2 = 0 and an "increase" as a ΔICP/ΔPbtO 2 > 0. Patients with a PbtO 2 increase of more than 20% from baseline were considered as "responders"; a significant increase in ICP was defined as an increase of more than 20% from baseline or an increase that led to intracranial hypertension.The following ventilator parameters were recorded at T0 and T1: inspiratory pressure (Pins); tidal volume (V T ), PEEP; respiratory rate (RR) and inspired fraction of oxygen (FiO 2 ).Arterial oxygen saturation (SaO 2 ), lactate, pH, PaCO 2 , PaO 2 were also recorded.
We collected demographics, the presence of comorbidities, sequential organ failure assessment (SOFA) 35 and the Glasgow coma scale (GCS) 36 on admission.Hospital mortality and the Glasgow Outcome Scale (GOS) 37 at 3 months were collected, as previously reported 38 .Unfavorable neurological outcome (UO) was defined as GOS of 1-3.

Study outcomes
The primary outcome was the proportion of PbtO 2 responders.Secondary outcomes included: (a) the proportion of patients with an increase of ICP > 20% or an increase in ICP that resulted in intracranial hypertension after PEEP augmentation; (b) the correlation between the difference of PEEP at T1 and T0 (ΔPEEP) and ΔICP/ ΔPbtO 2 ; (c) baseline factors associated with significant increases in PbtO 2 and ICP after PEEP augmentation; (d) the proportion of patients with an decrease in CPP resulting on a CPP < 60 mmHg; (e) differences between the trends in PbtO 2 and ICP during PEP incrementation challenge in SAH and TBI patients.

Statistical analysis
Descriptive statistics were computed for all variables.Categorical variables were described as proportions (%) and compared using Chi square or Fisher's exact test.Normality was assessed using the Kolmogorov-Smirnov test.Normally distributed variables were expressed as mean (± SD) and compared using Student t test while non-gaussian continuous variables were described median [IQRs] and compared using Mann-Whitney test (independent variables) or Wilcoxon test (repeated measures of related variables).A Spearman correlation was computed between ΔPEEP , ΔPbtO2, ΔICP, ΔCPP and ΔPaO 2 .As a sensitivity analysis we also considered just the first PEEP increment of each patient to calculate the correlation between ΔPEEP , ΔPbtO 2 , ΔICP, ΔCPP and ΔPaO 2 .To account for multiple measures per patient a generalized mixed model with logit link was used to identify baseline variables which were independently associated with a PbtO 2 responder and a significant increase in ICP after PEEP increment; baseline variables with a p value < 0.1 in the univariate analysis were included in the multivariable analysis.Odds ratios (ORs) with 95% confidence intervals (CIs) were computed for all variables.The independence of errors, presence of multicollinearity and of influential outlier assumptions were checked; none were violated.A receiver operator curve was designed to assess the sensitivity and specificity of baseline PbtO 2 to identify responders.The area under the curve (AUROC) and CI 95% were computed.Youden's test was used to identify the cut-off with the best sensitivity and specificity.We used a similar model to identify variables associated with an absolute increase in PbtO 2 and ICP after PEEP increments.All statistical analyses were performed using SPSS 27.0 for MacIntosh.A p value < 0.05 was considered significant.

Ethics approval and consent to participate
The study protocol was approved by local ethics Committees (Erasme Hospital: P2022/449) and informed written consent was waived.All methods were carried out in accordance with relevant scientific and ethical guidelines and regulations.

Patients with significant ICP increase
A significant increase in ICP was observed in 109/295 (37%) episodes of PEEP incrementation.In 23/109 (21%) episodes of significant ICP increase there was an ICP increase > 20% resulting in an ICP > 20 mmHg at T1.No baseline factors were associated with a significant ICP increase (Table 4).In patients with a significant increase in ICP (Supplemental Fig. S2A), no correlation between ΔPEEP and ΔICP (r = − 0.063 [95% CI − 0.53 to 0.132) was observed, but there was a weak inverse correlation between ΔICP and ΔCPP (r = − 0.239 [95% CI − 0.413

Subarachnoid hemorrhage and TBI
In TBI patients (n = 48) PEEP was increased from 5 (5-8)   (18-32) mmHg (p = 0.001), while ICP did not vary significantly over Table 3. Univariable generalized mixed model for fixed effects logit link function to assess the impact of baseline variables on the significant increase (responders) in brain tissue oxygenation (PbtO 2 ) after positive end expiratory pressure (PEEP) increments.Data are expressed as odds ratio and 95% confidence intervals.Data from 112 patients with 163 episodes of PEEP incrementation were included in this analysis.SOFA sequential organ failure assessment, ICU intensive care unit, TBI traumatic brain injury, SAH subarachnoid hemorrhage, PEEP positive end expiratory pressure, ICP intracranial pressure, CPP cerebral perfusion pressure, PbtO 2 brain tissue partial pressure of oxygen, PaO 2 arterial partial pressure of oxygen, PaCO 2 arterial partial pressure of carbon dioxide.

Discussion
In this study, we observed a significant elevation in PbtO 2 in 35% of episodes following an increase in PEEP; a lower baseline PbtO 2 was found to be associated with a significant increase in brain oxygenation.We determined that a baseline PbtO 2 cut-off of 21 mmHg was optimal for identifying PbtO 2 responders to PEEP.Additionally, we observed a moderate correlation between changes in PEEP and changes in PbtO 2 , specifically within the group of PbtO 2 responders.Interestingly, we did not identify any baseline factors associated with a significant increase in ICP, which was observed in 37% cases following PEEP augmentation.
Mechanical ventilation is often required for patients with severe TBI and poor-grade SAH due to various factors, such as coma, compromised airway protection, risk of aspiration, seizures, elevated ICP, pulmonary dysfunction, and respiratory failure resulting from pre-existing conditions or new complications 37 .The use of PEEP is a crucial component of ventilation strategies in this context.Recent consensus guidelines 38 suggest that in acute brain injury patients without acute respiratory failure, PEEP levels should be similar to those used in www.nature.com/scientificreports/patients without brain injury, and lung protective ventilation strategies can be employed.In patients with both respiratory failure and acute brain injury, lung protective ventilation and higher levels of PEEP may be utilized, as long as clinically significant increases in ICP are not observed.However, there are currently no specific recommendations provided for patients with acute brain injury, respiratory failure and intracranial hypertension 38 .These consensus guidelines have also recommended targeting a PaO 2 range of 80-120 mmHg to avoid both hypoxemia and hyperoxia 38 .Positive end-expiratory pressure plays a crucial role in increasing lung functional residual capacity, preventing alveolar de-recruitment, and improving oxygenation 37 .Therefore, increasing PEEP can be a strategy to enhance oxygenation in these patients.Another important consideration in setting PEEP is PbtO 2 , as avoiding cerebral hypoxia, in addition to systemic hypoxemia, is crucial in preventing secondary brain injury in neurocritical care.Additionally, acute lung injury has been associated with brain tissue hypoxia 39 .However, there is limited research on the impact of PEEP on brain oxygenation.
One study involving 20 TBI patients with ARDS demonstrated that increasing PEEP from 5 to 10 to 15 cmH 2 O resulted in significant increases in PbtO 2 and oxygen saturation, without affecting ICP or CPP in patients without baseline intracranial hypertension 25 .Another study with 10 SAH patients showed that applying PEEP of 20 cmH 2 O resulted in a decrease in PbtO 2 due to a simultaneous decrease in CPP and cerebral blood flow, which was reversed upon restoration of MAP 27 .Furthermore, a study involving a mixed population of SAH and TBI patients with ARDS demonstrated that recruitment maneuvers and high levels of PEEP significantly improved both systemic and brain tissue oxygenation, leading to reduced oxygen requirements 26 .In our study, increasing PEEP improved PbtO 2 , particularly in patients with low baseline PbtO 2 .Similar responses have been observed with other strategies aimed at enhancing interstitial oxygen availability in acute brain injury patients, such as red blood cell transfusion 40 .
The application of PEEP can have an impact on cardiovascular physiology by increasing intrathoracic pressure; this can potentially decrease venous return, reduce cardiac output, and lead to hypotension, particularly in hypovolemic patients 41 .Hypotension is a significant cause of secondary brain injury, as it can reduce CPP and induce ischemia 42,43 .Previous studies that reported negative effects of PEEP on CPP are typically associated with decreased mean arterial pressure 27,28,[44][45][46] .In our study, both MAP and CPP remained unchanged.It is important to note that increased intrathoracic pressure can lead to decreased cerebral venous drainage, which may elevate ICP 28 .Additionally, higher levels of PEEP and low tidal volumes can result in increased arterial carbon dioxide levels, potentially causing cerebral vasodilation and a subsequent increase in ICP 47 .Consequently, numerous studies have investigated the effects of PEEP on ICP and cerebral hemodynamics.Wolf et al. demonstrated that employing an open lung strategy with PEEP appears to be safe in patients with acute brain injury and concomitant ARDS, as evidenced by no significant changes in ICP in a small case series (n = 11) 48 .Boone et al. showed minimal effects of PEEP on ICP in patients without severe lung injury, and even in severe lung injury patients, the increase in ICP due to PEEP was considered clinically irrelevant 44 .Other studies conducted in patients with ischemic stroke and TBI have also indicated that the response of ICP to PEEP, although variable, is moderate at best and lacks clinical impact, suggesting that PEEP can be safely utilized in acute brain injury patients when indicated 27,29,45,[49][50][51][52] .These findings are consistent with our study.However, an older study reported a significant increase in ICP after increments in PEEP among head-injured patients, which had neurological repercussions and necessitated immediate reduction in PEEP 28 .
The impact of PEEP on ICP appears to be influenced by the presence of baseline intracranial hypertension.One study that analyzed changes in ICP after incremental increases in PEEP found that in patients with baseline intracranial hypertension, higher PEEP levels did not affect ICP, while in patients with normal ICP, PEEP Table 4. Univariable generalized mixed model for fixed effects logit link function to assess the impact of baseline variables on the a significant increase in ICP (increase of more than 20% of baseline or increased that resulted in ICP > 20 mmHg) after positive end expiratory pressure (PEEP) increments.Data are expressed as odds ratio and 95% confidence intervals.Data from 112 patients with 295 episodes of PEEP incrementation were included in this analysis..Other studies have shown that in patients with ICP > 20 mmHg, ICP decreased or remained stable after PEEP increments, while CPP remained unchanged or even increased in some cases 45,52 .In our study, significant increases in ICP due to PEEP increments were observed in 37% (109/295) of episodes, with an ICP exceeding 20 mmHg in 34% (37/109) of episodes.However, baseline ICP was not found to be an independent factor associated with ICP increases following PEEP increments.We were unable to identify any baseline factors that could predict changes in ICP in response to PEEP.Similarly, a recent study also failed to identify baseline factors that could help identify patients at higher risk for ICP increases following PEEP changes 46 .
Another factor that appears to influence ICP responses to variations in PEEP is lung compliance.In patients with low compliance, as is often the case in ARDS patients, higher levels of PEEP had no impact on CPP, ICP or cerebral blood flow 53 .Conversely, in patients with normal compliance, there was a decrease in CPP and cerebral blood flow, but no changes in ICP 45 .A recent study has demonstrated that changes in ICP are inversely correlated with lung recruitability after PEEP application and recruitment maneuvers, suggesting that patients who would benefit from higher PEEP levels will have minimal side effects on intracranial pressure 46 .This can be explained by the fact that the detrimental effects of PEEP are related to alveolar hyperinflation, leading to a significant increase in arterial carbon dioxide and intrathoracic pressure.Conversely, when PEEP leads to alveolar recruitment, improved lung gas distribution, and optimization of ventilation/perfusion matching, ICP remains unchanged 46 .In our study, PaCO 2 remained stable and within the normal range, which may explain why overall ICP remained unchanged.
Our study has several limitations that should be acknowledged.Firstly, we did not perform a formal sample size calculation, which may have limited the power of our study and prevented us from drawing definitive conclusions.Additionally, this was a single-center study, and therefore, the generalizability of our findings to other centers may be limited.Due to the limited number of events, we were unable to conduct further subgroup analyses, such as comparing patients with baseline tissue hypoxia, baseline intracranial hypertension or differentiating those with ARDS from those without ARDS, which could provide valuable insights into specific patient populations.Another limitation is that we did not specifically assess the correlation between increases in PbtO 2 after PEEP increments and functional outcomes.Moreover, the retrospective design of our study prevents us from ruling out the potential influence of other concurrent interventions, sedation, or vasopressor therapy, which may have contributed to the observed changes in PbtO 2 following PEEP increments.Additionally, we lack information on the specific reasons for PEEP changes, as these were determined by the clinical care team.Importantly, we were unable to assess change in respiratory mechanics (such as lung compliance) and correlate them to PbtO 2 and ICP due to the retrospective nature of this study.Furthermore, the magnitude of PEEP changes in our study was generally small, which could have impacted our results.Moreover, due to how the data was collected (1 h mean) we were unable to perform a time series analysis of the impact of PEEP incrementation on physiological variables.It is also important to note that while PbtO 2 may assess interstitial oxygen availability, its influence on cellular oxygenation in the brain remains uncertain.Lastly, our findings are not conclusive, as the heterogeneity of our study cohort, which included both traumatic and non-traumatic brain injury patients, and the placement of the probe in hypo perfused cerebral areas (which are more sensitive to interventions aiming to improve oxygen delivery), as well as the use of regional rather than global oxygen monitoring, raise significant issues that should be addressed in future studies on this topic.

Conclusions
In this study, implementing PEEP to enhance PbtO 2 could be a viable approach in patients with acute brain injury, particularly those presenting with baseline tissue hypoxia.However, it is crucial to closely monitor ICP and systemic hemodynamics to ensure the safe and appropriate administration of PEEP. https://doi.org/10.1038/s41598-023-43703-9

Table 1 .
Characteristics of the study population.SOFA sequential organ failure assessment, WFNS world federation of neurological surgeons, NYHA new york heart association functional classification, COPD chronic obstructive pulmonary disease, RRT renal replacement therapy, ECMO extra corporeal membrane oxygenation, ICU intensive care unit, GOS Glasgow outcome scale, LOS length of stay.

Table 2 .
Ventilatory and physiological variables at T0 and T1.Data are expressed as median and interquartile range.Vt tidal volume, Pinsp inspiratory pressure, PEEP positive end expiratory pressure, RR respiratory rate, FiO2 inspired fraction of oxygen, MAP mean arterial pressure, ICP intracranial pressure, CPP cerebral perfusion pressure, PbtO 2 brain tissue partial pressure of oxygen, P/F arterial partial pressure of oxygen/ inspired fraction of oxygen, PaCO 2 arterial partial pressure of carbon dioxide.