Cerebrospinal fluid pressure dynamics across the intra-and postoperative setting: Retrospective study of a spine surgery cohort

Timely and sufficient decompression are critical objectives in degenerative cervical myelopathy (DCM) and spinal cord injury (SCI). We previously investigated intraoperative cerebrospinal fluid pressure (CSFP) for determining surgical outcomes. However, confounding factors during the intra-and postoperative setting need consideration. These are related to type of respiration (i.e., artificial vs. natural) and anesthesia, which affect CSFP dynamics through the interaction between the cardiorespiratory system and the CSF compartment. This retrospective cohort study (NCT02170155) aims to systematically investigate these factors to facilitate CSFP interpretation. CSFP was continuously measured through a lumbar catheter, intra-and postoperatively, in 21 patients with DCM undergoing decompression surgery. Mean CSFP and cardiac-driven CSFP peak-to-valley amplitude (CSFPp) were analyzed throughout the perioperative period, including the immediate extubation period in eight patients. Intraoperative mean CSFP had a median value and {interquartile range} of 10.8 {5.5} mmHg and increased 1.6-fold to 16.9 {7.1} mmHg postoperatively (p < 0.001). CSFPp increased 3-fold from 0.6 {0.7} to 1.8 {2.5} mmHg (p = 0.001). Increased CSFP persisted overnight. During extubation, there was a notable increase in CSFP and CSFPp of 14.0 {5.8} and 5.1 {3.1} mmHg, respectively. From case-based analysis, this was attributed to an arterial pCO 2 increase. There was no correlation between respirator settings and CSFP metrics. There were distinct and quantifiable changes in CSFP dynamics from the intra-to postoperative setting related to type of respiration, anesthesia, and level of consciousness. When monitoring CSFP dynamics in spine surgery across these settings, cardiorespiratory factors must be controlled for.


Introduction
Timely and sufficient surgical decompression are critical objectives with many pre-clinical and clinical studies focused on surgical timing.However, it remains difficult to quantify the degree of cord compression and provide intraoperative guidance in decompression surgery.For this purpose, several studies in patients with spinal cord injury (SCI) have investigated the utility of intraoperative monitoring of intraspinal pressure (ISP) [1] and cerebrospinal fluid pressure (CSFP) [2].Recently, monitoring of cardiac-driven CSFP peak-to-valley amplitude (CSFPp) was utilized in patients with degenerative cervical myelopathy (DCM) [3].These studies used ISP or CSFP dynamics to estimate effective cord decompression or as a proxy to calculate spinal cord perfusion pressure [4,5].Other studies investigated CSF drainage as a therapeutic option to lower CSFP and improve functional outcomes [6,7].During surgery, patients are sedated and artificially ventilated.At the end of surgery, anesthesia is stopped, spontaneous breathing starts prior to extubation, consciousness progressively returns, and vital signs settle to individual Abbreviations: ABP, Arterial blood pressure; ASA, American Society of Anesthesiologists scores; COPD, Chronic obstructive pulmonary disease; CSF, Cerebrospinal fluid; CSFP, Cerebrospinal fluid pressure; CSFPp, Cardiac-driven CSFP peak-to-valley amplitude; DCM, Degenerative cervical myelopathy; ICU, Intensive care unit; ICP, Intracranial pressure; ISP, Intra-spinal pressure; MAP, Mean arterial blood pressure; MRI, Magnetic resonance imaging; pCO 2 , Partial pressure of carbon dioxide; PEEP, Positive end-expiratory pressure; SCI, Spinal cord injury.baselines.However, there is insufficient published data on the impact of these factors on intraoperative CSFP in comparison with the postoperative setting.We hypothesize that CSFP metrics increase from the intraoperative (artificial ventilation) to the postoperative setting (natural ventilation), due to the switch from positive to negative pressure ventilation.
In healthy awake subjects, CSF flow studies using MRI have shown the greater impact of respirationin particular deep respirationon the overall CSF movement compared to arterial pulsation [8][9][10][11].While respiration induces a high global CSF flow at the smaller frequency of respiratory rate, cardiac pulsations elicit a lower local CSF flow at the higher frequency of heartrate.It has been shown that CSF generally flows cranially with inspiration and caudally with expiration.This is caused by changes in the intrathoracic pressure where with inspiration venous outflow increases and CSF fills the cranial cavities to maintain a constant intracranial volume and vice versa with expiration.Due to the significant contribution of respiration on CSF dynamics, any change in the type of respiration (i.e., artificial vs. natural) during CSFP monitoring must be accounted for.
The transition period from intraoperative to postoperative setting, i. e., extubation, is characterized by a rise of partial pressure of carbon dioxide (pCO 2 ) level in blood [12], which induces spontaneous breathing.In children, a direct relation between pCO 2 increase and CSFP has been demonstrated [13], and in animal models it has been shown that higher pCO 2 also leads to greater CSFPp [14].These findings were explained through CO 2 reactivity of intracranial vessels and, thus, related to arteriolar vasodilation [15].
In this sub-study of a prospective cohort study, we aimed at characterizing the physiological changes in CSFP from anesthetized and artificially ventilated to awake and spontaneously breathing in adult patients with DCM undergoing surgical decompression [3,16].Additionally, extubation was studied as the transition period between the intra-and postoperative setting.Such insight is critical for the interpretation of perioperative CSFP data.

Clinical characteristics and CSFP monitoring setup
Data were obtained within a prospective cohort study in patients with DCM undergoing surgical deCOMPression of the spinal CORD (COMP-CORD) (NCT02170155) [16].This study is a secondary analysis of CSFP dynamics related to type of respiration and anesthesia.The study conformed to the latest revision of the Declaration of Helsinki and was approved by the local Ethics Committee of the University Hospital of Zurich (KEK-ZH number PB-2016-00623).Before enrollment, patients were informed about risks of participation.This included postpuncture syndrome, CSF leakage, infection, injury to nerve roots, spinal hematoma and intracranial hemorrhage related to lumbar catheter.Additionally, patients were informed that they do not receive an immediate benefit from participation in the study, as CSFP monitoring in DCM is at an experimental stage.All patients (N=21) had lumbar intrathecal CSFP recording before, during and after the decompression surgery.All patients were in stable medical condition at the time of elective surgery.Postoperatively, all patients were routinely managed at the intensive care unit (ICU) (Fig. 1).Three patients suffered from asthma (ID9, 10,17).Two patients had chronic obstructive pulmonary disease (COPD), ID3 was graded Gold IV and ID16 Gold II (with Gold IV being the most severe).American Society of Anesthesiologists (ASA) scores are reported in Tabel 1 (range ASA: 1-3, M=2.1, SD=0.6).They present the overall physical health of the patients, with a higher score indicating worse health status [17].Intrathecal catheter placement was confirmed by respiratory modulation of the CSFP signal and backflow of CSF.Given that the median depth of lumbar catheter placement from skin was 28 cm (range = 20-35 cm), we consider the catheter to be located at the lumbar or low thoracic level of the spine.CSFP dynamics during extubation from artificial ventilation were only available in a subgroup of N=8 patients, as this step was amended to the initial protocol.Blood gas measurements were done in two of these patients.Mean arterial pressure (MAP) was obtained continuously from radial artery in all but two patients during surgery and extubation.The lumbar catheter (Neuromedex Lumbalkatheter 4.5F) was connected to an analog-todigital pressure converter (NeuromedexVentrEX), the digitized signal was fed to a Philips X2-Pat Interface + MX 700 Monitor, and recorded using ICM+recording software (University of Cambridge) with the sampling rate of 250 Hz.

Data analysis
Data were analyzed with MATLAB using discrete wavelet decomposition.This choice was due to the fact that biological signals are nonstationary (i.e., their properties change with time).Therefore, the application of Fourier analysis, which is the more conventional tool, is limited to short time windows where the stationarity assumption can be met [18].Wavelet transform offers information both in time and frequency, it can extract desired characteristic shapes (such as CSFPp and modulation related to ventilation) from the signal, and the decomposition of the signal into the frequency bands of interest is smooth (Fig. 2).Signals were decomposed into ten frequency bins.The sum of the signals in the first two frequency bins (0-0.5 Hz) represents mean CSFP and the respiratory modulation.Mean CSFP was calculated as the median value, and modulation related to ventilation was measured as the difference between each peak and its preceding valley.The sum of the contents of the next four frequency bins (0.5-8 Hz) was used to calculate CSFPp, which corresponds to the difference between the systolic peak and the associated diastolic valley.Fig. 2 displays a sham CSFP signal and the extracted features, i.e., mean CSFP, modulation related to ventilation and CSFPp.Statistical analysis was performed using paired t-test to evaluate the significance of the changes in CSFP metrics and MAP (intravs.postoperative and postoperative vs. post-24H) across all patients.For multiple comparison of the parameters across different settings ANOVA test was used.A p-value < 0.05 was considered statistically significant.For the investigation of interrelations between the CSFP metrics and potential confounding factors (i.e., surgery time, anesthesia time, number of decompressed levels, intraoperative bleeding, surgical approach, Pmax, Ppl, PEEP), we assessed Kendall correlation, as the data was not normally distributed.

Anesthesia procedures
Patients were premedicated with midazolam 7.5 mg an hour before the operation.In the induction room, electrocardiography, pulse oximetry, invasive arterial blood pressure (ABP) measurement, and Bispectral Index (BIS) monitoring were performed, and an IV catheter was placed.The initial propofol (Marsh model) blood target concentration was set at 4-6 µg/ml.After loss of consciousness (defined as a lack of eyelid reflex), the individual effect-site concentration was doubled, set as the new target concentration for anesthesia maintenance, and was adjusted based on BIS value for a target between 40 and 50.Induction of fentanyl 1-3 μg/kg led to analgesia, and its maintenance was managed with additional fentanyl and remifentanil (Minto model) for an effect site concentration of 1-6 ng/ml.Rocuronium 0.9 mg/kg IV was used to facilitate tracheal intubation.The ventilator mode was set to volume-controlled with a fixed positive end-expiratory pressure of 5-7 mbar and a tidal volume between 6-10 ml/kg body-weight/min, which depended on CO 2 from the blood gas analysis.The pCO 2 and MAP target were 4.5-5 kPa and 65-70 mmHg, respectively.MAP was kept constant with crystalloid solutions and/or continuous infusion of noradrenaline.A constant core temperature of 36-36.5 • C was achieved using a warm air system.Parameters such as pH, K + , Na + , lactate and base excess and glucose assessed in the blood gas analyses were maintained in normal ranges during the whole intraoperative measurement by substitution/ correction, hydration or modification of ventilation.

Surgical procedure
14 Patients underwent anterior cervical discectomy and fusion and the remaining had dorsal decompression with laminectomy performed, with additional laminotomy in some cases.Further specific details are provided in Table 1, including surgical and anesthesia time, number of decompressed levels, amount of intraoperative bleeding, and surgical approach.Three patients suffered from intraoperative CSF leakage (ID4, 6, 16) which was treated with closure, and was asymptomatic.One patient had post-operative C5 palsy (ID9).In terms of adverse events from catheter placement, two patients had self-limiting post-puncture headache (ID1, 19), one had post-operative CSF loss that needed skin suture (ID 19), and one had toxic skin lesion related to remnants of disinfection agent (ID 4)

Artificial ventilation vs. Natural ventilation
The evolution of CSFP metrics and MAP during transition from artificial to natural ventilation is summarized for individual patients in Table 2.The time interval between the intraoperative and the postoperative recordings was 1 h ± 30 min (mean ± standard deviation).The interval between the first and last measurement in the postoperative scenario was 17 h ± 4 h.Overall, there was a rise of mean CSFP from a median and {interquartile range} of 10.8 {5.5} to 16.9 {7.1} mmHg (1.6-fold, p < 0.001), CSFPp from 0.6 {0.7} to 1.8 {2.5} mmHg (3-fold, p = 0.001), and MAP from 77.4 {10.5} to 100.6 {15.0} mmHg (1.3-fold, p < 0.001), and a decline of the modulation related to ventilation from 1.8 {1.2} to 1.0 {1.0} mmHg (0.6-fold, p = 0.003), proceeding from one

Extubation
Monitoring during extubation was available in 8/21 patients.Fig. 4 illustrates the evolution of CSFP in four of the patients (ID13, 14, 15, 16) as the ventilator was turned off prior to extubation.After turning off the ventilator (indicated by solid vertical blue line), mean CSFP remained constant for a few seconds, while the modulation induced by the ventilator immediately disappeared (Fig. 4a).Consecutively, a constant rise of CSFP was observed until respiratory modulation reappeared while mean CSFP gradually settled at a higher baseline compared to the intraoperative level.Patients were not fully conscious when the ventilator was turned off.Measurements showed a mean CSFP rise with a magnitude of 11.1 {5.5} mmHg and a CSFPp rise of 5.9 {4.2} mmHg.MAP increased by 12.9 {9.6} mmHg.Patient ID14 had a pCO 2 value of 5.9 kPa before propofol was turned off, which increased to 6.1 kPa afterwards (Fig. 4c).Patient ID15 had a pCO 2 value of 5.9 kPa during artificial ventilation, which increased to 6.3 kPa after extubation (Fig. 4d).

Summary of main findings
CSFP dynamics were highly affected by physiological changes occurring between the intra-and the postoperative setting, namely type of respiration, level of consciousness and blood gases.Both mean CSFP and CSFPp strongly increased in the postoperative setting, as hypothesized.We mainly attributed this to the change from artificial ventilation to natural ventilation, and awake state.During extubation, mean CSFP, CSFPp and MAP increased substantially within seconds, mostly related to turning off the ventilator and a consecutive rise in pCO 2 .Overall, the interaction between CSFP and respiration is complex, involving the mode of respiration, anesthesia, and the vascular system.Of factors involved in short-term changes, pCO 2 seems to be of major relevance given its vasodilatory effect on intracranial vessels [15], and because it could be easily monitored during routine surgery.However, CSFP increments were preserved overnight, pointing towards the involvement of other persisting factors independent from pCO 2 , such as natural breathing and the awake state in contrast to artificial ventilation and narcosis intraoperatively.Due to cerebral autoregulation, variations in MAP in the ranges we measured may have little effect on CSFP metrics [19].Intraoperative back versus prone positioning was not identified as a systematic confounder for changes in CSFP dynamics between settings.Intra-and postoperatively (including post-24H), surgery and anesthesia parameters (as reported in Table 1) were not significantly correlated with CSFP modulation related to ventilation, and neither to MAP.Thus, we deem the respiration related CSFP metrics not to be biased by these factors.However, these findings may not be generalized to patients with acute conditions that involve the autonomous nervous system (i.e., SCI) and patients with anesthesia complications (e.g., blood pressure fluctuations, or respiratory dysfunction).To date, there is no evidence in support of a clinical use of intraoperative CSFP assessments.This work will aid in the interpretation of CSFP dynamics in spine surgery research across the intra-and postoperative settings.

Rapid changes in CSFP dynamics during extubation
With the ventilator off, a period of constant CSFP and CSFPp rise was observed until the patients started to breathe spontaneously, at which point a slight drop of CSFP occurred, although it did not reach its initial value within the observation period.We attributed the short delay between turning off the ventilator and CSFP rise to the fact that the patients did not start breathing immediately on their own.Regarding pharmacological effects, anesthetic drugs are stopped some minutes before extubating as a matter of routine.Given several hours of half-life for most of these drugs, a major change in their effect on CSFP at that point is unlikely.Lastly, an effect of pCO 2 must be considered.There is a rapid increase in pCO 2 levels in the time window from reducing anesthetics to reinstatement of natural ventilation and removal of the tube.This increase is essential to stimulate the respiratory drive.It leads to cerebral vasodilation, causing mean CSFP and CSFPp to rise [20].Given this physiological background, and exclusion of other factors, we suggest to mainly attribute the rapid changes in CSFP during extubation to pCO 2 level rise.

Relating intra-to postoperative CSFP dynamics
Patients presented with a persisting higher mean CSFP (1.6-fold) and CSFPp (3-fold) following extubation, while showing a drop in the modulation related to ventilation of the CSFP signal (0.6-fold).We attribute this observation to a volume shift along the craniospinal CSF compartment.In general, the sum of volumes of the brain, intracranial blood, and CSF is constant due to the rigidity of the skull.Any volume expansion will result in CSF volume shift and pressure change that translates to the lumbar site as well [21].There is need to further elaborate on how these volume shifts relate to characteristics of the intra-and postoperative settings.The main contributors include, but are not limited to, type of respiration, anesthesia, and vascular system. 1) Type of respiration: In theory, positive pressure ventilation leads to increased intrathoracic pressure, which causes a decrease in venous return and a rise of jugular venous pressure, and ultimately increased intracranial pressure (ICP) [20].However, previous studies across different patient populations showed variable effects of PEEP on ICP, depending on multiple factors [22].For instance, in artificially ventilated patients with normal ICP values, the effect of PEEP level was shown to be negligible [23].In our cohort, lumbar CSFP was measured in lateral recumbent position, which is not identical to ICP [24], but can be considered a surrogate of ICP in patients without spinal stenosis and brain injury [25].Intraoperatively, within the narrow range of PEEP (between 4 and 7 cmH 2 O), there was no correlation with mean CSFP, which is in line with the previous findings.However, with larger PEEP or varying mean CSFP, the effect of PEEP on CSFP might be more pronounced.
In the transition from intra-to postoperative modulation related to ventilation, the patients with COPD and asthma did not show specific differences on a single subject level compared to their lung-healthy peers.This does not exclude an effect of these lung conditions on CSFP in specific respiratory states, such as during hypercapnia or sleep.
In the awake and spontaneously breathing patients, respiration returns to its physiological state with negative intrathoracic pressure compared to the positive pressure applied during artificial ventilation.One preclinical study in rats demonstrated that natural ventilation leads to greater CSF displacement and a higher CSF velocity compared to artificial ventilation [9].This finding was mainly attributed to negative intrathoracic pressure in natural ventilation, leading to a higher pressure-gradient along the craniospinal axis.Our results support these findings.Apparently, the changes in intrathoracic pressure have a larger effect on mean CSFP than on the magnitude of modulation related to ventilation, which is higher during artificial ventilation compared to natural ventilation.
2) Anesthesia: As pointed out in the previous section, given the short interval between intra-and postoperative recordings, it is unlikely that the observed changes in CSFP are related to drug effects.However, as patients regain consciousness in the postoperative setting due to metabolization of these drugs, the awake state is associated with higher sympathetic tone, which in turn leads to an increase in ICP [26,27].3) Vascular system: Notably, there is an increase in MAP (1.3-fold) during the postoperative setting.The observed MAP change aligns with the different patients' status, since in the postoperative setting, they are conscious and spontaneously breathing.Both intra-and postoperative MAP lie within the autoregulation range, meaning that the brain is able to maintain an adequate cerebral perfusion pressure despite oscillations in MAP or ICP [19].Therefore, even though we cannot exclude that the observed increase in MAP may have provoked the detected CSFP rise, it is not the most likely factor based on the aforementioned reasoning.
Lastly, filling of epidural veins affects CSF displacement and thus CSFP.As mentioned earlier, with every respiratory cycle, variations in the venous return lead to a rise of CSFP and to CSF flow changing directions.Notably, the respiration mode also affects the venous filling of abdominal and thoracic vein plexus, which in turn relate to CSFP [28].With an overall higher MAP, we expect the influence of the venous side to be more pronounced.
The observed increase in CSFP persisted overnight despite likely (though not quantified) pCO 2 normalization as compared to the extubation period.All mentioned factors considered; natural ventilation may play a more important role than other factors in sustaining increased CSFP over an extended period.

Implications for clinical and preclinical trials of CSFP dynamics
Intraoperative CSFP monitoring in spine surgery was previously investigated to determine effects from decompression [2,3] and for measuring spinal cord perfusion pressure [4,29], aiming at optimization of surgical outcomes and hemodynamic management.In terms of application, detection of altered CSFP dynamics might relate to restricted CSF flow across a spinal stenosis or narrowed CSF space through cord swelling.Potential interventions aim at ensuring complete bony decompression, while expansion duroplasty is currently under investigation to maintain sufficient spinal perfusion and reduce detrimental effects from cord swelling (NCT04936620) [30,31].Furthermore, there is evidence for clinical utility of CSFP-guided hemodynamic management in SCI, which may help to overcome crude blood pressure targets [4,29,32].Additionally, CSF drainage was proposed to reduce intrathecal pressure and improve spinal perfusion [2,6].Currently, a prospective multicenter study is investigating the effects of CSF drainage on spinal perfusion and neurological outcomes (NCT03911492).These approaches are applicable in the intraoperative and postoperative setting.Our findings should be accounted for when comparing results from CSFP monitoring between different settings.This said, the interpretation of postoperative CSFP dynamics requires careful distinction between respiration-and surgery-related factors.With regards to decompression effects on postoperative CSFP dynamics, additional volume load in the lumbar CSF compartment may contribute to the findings shown here.Evaluating this hypothesis demands for infusiontesting or experimental assessments of the CSF pulsatility curve as recently developed by our group [33].However, a homogenous pattern of postoperative CSFP increase found in a heterogenous patient population indicates that setting changeswhich are present across all patientshave a significant impact on postoperative CSFP.In some cases, in which postoperative CSFP dynamics remained constant, there were explanations such as CSF leakage (ID4) and hypercapnia (ID16), but some remained unclear.To our knowledge, there were no critical clinical events related to lack of increase in the postoperative CSFP.Secondly, our findings should be considered when managing CSFP with lumbar drainage.Given higher postoperative mean CSFP in most patients, mean CSFP targets relative to the baseline might be better applicable throughout the intra-and postoperative setting than absolute mean CSFP values.Thirdly, having a controlled and steady intraoperative condition provides an opportunity to examine the CSF compartment.This includes the option of modulating CSFP dynamics for diagnostic purposes through adjustment of respirator settings.Consequently, besides its role as a potential confounder of CSFP dynamics, modulating pCO 2 may be a potential tool to provoke mean CSFP and CSFPp increase and aid in determining effective cord compression.Permissive hypercapnia [34] is achieved in artificially ventilated anesthetized patients through hypoventilation, where an autonomic nervous system response and consecutively vessel reactivity is expected, leading to mean CSFP and CSFPp rise.This might also be helpful to better characterize conditions with very low CSFPp through enhancing CSFP dynamics.

Study limitations
This study aimed at investigating physiological changes of CSFP dynamics across the intra-and postoperative setting.The collected data do not allow for statements on the potential clinical efficacy of CSFP monitoring in spine surgery.As this retrospective study is based on findings from [3], where CSFP metrics following spinal cord decompression were the primary objectives, several other parameters of interest remain unknown.These include physiological changes in pressure gradients, i.e., intrapulmonary, intrathoracic, and intraabdominal, as well as blood gas levels that were not continuously quantified.Thus, more detailed investigation of the underlying mechanisms was not possible.Similarly, anesthesia management was not controlled for.Although the interrelation of CSFP metrics and respiration was not influenced by the surgical approach, it cannot be excluded that with larger samples there would be differences.For instance, anterior decompression surgery can result in postoperative swelling around the trachea, affecting the postoperative respiratory status.We note that there were no respiratory complications in the postoperative setting in our cohort.Secondly, a more detailed analysis of the directionality of modulation related to ventilation was not possible, since the exact site of catheter placement, i.e., lumbar or thoracic spinal level, was neither quantified nor standardized.Based on MRI studies [35], respiration leads to major CSF movement, which is due to the pressure gradient along the craniospinal column, meaning that CSFP respiratory modulation depends on where the measurement is taken.Thirdly, the patient cohort examined here were not spine-healthy.To reduce the bias resulting from spinal stenosis, exclusively recordings from the postdecompression time were used.Importantly, as this study was primarily aiming at the investigation of CSFP dynamics related to decompression surgery in DCM, it is limited in applying correlations and reaching a definitive conclusion to what extent and for how long each of the cardiorespiratory factors account for the observed changes in CSFP from the intra-to the postoperative setting.Lastly, there was a delay between extubation and resuming of the recording.This loss of information could be potentially solved through extubating in the intensive care unit, where no further patient transport is required.

Conclusions
The interpretation of intra-and postoperative CSFP dynamics in spine surgery research should account for the type of respiration, anesthesia, level of consciousness, and related changes in the vascular system.These parameters can be partly controlled for, in particular during the intraoperative setting, through the maintenance of anesthesia.

Fig. 1 .
Fig. 1.Timeline and characteristics of the study conditions.

Fig. 2 .
Fig. 2. Schematic representation of cerebrospinal fluid pressure (CSFP) data.CSFP sham signal (a) and the extracted metrics, mean CSFP (b), respiratory modulation and the arrow representing modulation related to ventilation (c), cardiac-driven CSFP peak-to-valley amplitude (CSFPp) and the arrow representing the peak-to-valley difference (d).

Fig. 4 .
Fig. 4. Cerebrospinal fluid pressure (CSFP) recording during extubation in four patients (ID13, 14, 15, 16).The solid line represents the timepoint where the artificial ventilation is turned off and the dashed line represents the timepoint of extubation.

Table 1 Surgery and Anesthesia parameters of the patients. † in
millibar; † † in cmH 2 O; ASA: American Society of Anesthesiologists scores; PEEP: Positive End Expiratory Pressure; Pmax: Maximum Pressure of the ventilator; Ppl: Pleural Pressure.