A Prospective Randomized Study of Brain Tissue Oxygen Pressure-Guided Management in Moderate and Severe Traumatic Brain Injury Patients

The purpose of this study was to compare the effect of PbtO2-guided therapy with traditional intracranial pressure- (ICP-) guided treatment on the management of cerebral variables, therapeutic interventions, survival rates, and neurological outcomes of moderate and severe traumatic brain injury (TBI) patients. From 2009 to 2010, TBI patients with a Glasgow coma scale <12 were recruited from 6 collaborative hospitals in northern Taiwan, excluding patients with severe systemic injuries, fixed and dilated pupils, and other major diseases. In total, 23 patients were treated with PbtO2-guided management (PbtO2 > 20 mmHg), and 27 patients were treated with ICP-guided therapy (ICP < 20 mmHg and CPP > 60 mmHg) in the neurosurgical intensive care unit (NICU); demographic characteristics were similar across groups. The survival rate in the PbtO2-guided group was also significantly increased at 3 and 6 months after injury. Moreover, there was a significant correlation between the PbtO2 signal and Glasgow outcome scale-extended in patients from 1 to 6 months after injury. This finding demonstrates that therapy directed by PbtO2 monitoring is valuable for the treatment of patients with moderate and severe TBI and that increasing PaO2 to 150 mmHg may be efficacious for preventing cerebral hypoxic events after brain trauma.


Introduction
Survival rates and outcomes for patients suffering from severe traumatic brain injury (TBI) depend on the severity of secondary cerebral insults [1,2]. Previous studies have shown that cerebral ischemia, which may be caused by systemic hypotension, intracranial hypertension, impaired autoregulation, or hyperventilation, is a common and independent factor associated with neurological deterioration after injury [3][4][5][6][7]. Although no randomized trial has demonstrated an improved outcome for severe TBI patients provided with intracranial pressure (ICP) signal-guided treatment, current guidelines for severe TBI management recommend ICP monitoring to calculate and maintain cerebral perfusion pressure (CPP) and prevent cerebral ischemia and infarction [8]. However, certain studies have confirmed that CPP may not be correlated with cerebral blood flow and oxygen consumption in TBI patients [7,[9][10][11][12]. Cerebral ischemia and infarction may not be observed in patients with marginally low CPP, but these events may still occur, even with the maintenance of normal or supranormal CPP. In addition, certain CPPguided treatments, such as the administration of vasopressor agents or fluid expanders, have been demonstrated to cause adverse effects, including respiratory distress syndrome, and were shown to offset the original benefits of these treatments on the outcome of severe TBI patients [13,14].
A continuous supply of oxygen and glucose to brain tissue is important for maintaining a normal aerobic metabolism inside the brain cells. However, an excessive oxygen requirement and insufficient cerebral blood flow after trauma may cause cerebral hypoxia and transform a normal metabolism into an anaerobic condition [12,15,16]. The regional partial pressure of brain tissue oxygen (PbtO 2 ) has been described as an independent, sensitive, and direct predictor of cerebral ischemia and hypoxia [9,17,18]. It has been reported that the incidence, frequency, and duration of cerebral hypoxic events, indicated by PbtO 2 < 5, 10, or 15 mmHg, are significantly correlated with an unfavorable outcome in patients after trauma [7,[19][20][21][22][23]. The PbtO 2 can be manipulated by adjusting the ICP and CPP or the fraction of inspired oxygen (FiO 2 ). Decreasing ICP below 20 mmHg and increasing CPP above 60 mmHg, when guided by PbtO 2 monitoring, were reported to relieve cerebral hypoxia simultaneously through intact cerebral autoregulation [19,24]. However, the cardiopulmonary complications arising from ICP and CPP treatments should be managed carefully. Several studies have shown that elevating FiO 2 can increase PaO 2 , reduce hypoxia, and reinstate the aerobic metabolism inside the brain cells with fewer cardiopulmonary complications [16,25,26].
The effects of TBI treatment using additional PbtO 2 monitoring on cerebral hypoxia and patient outcome remain controversial, and evidence from randomized clinical studies is scant. Several investigators have reported that PbtO 2 -guided therapy significantly reduced cerebral hypoxic events and mortality rates and improved outcomes in severe TBI patients compared with historical controls [24,[27][28][29]. However, Martini and colleagues determined that severe TBI management guided by PbtO 2 monitoring was associated with a poor neurological outcome and was an inefficient use of hospital resources [30]. Therefore, the purpose of this prospective randomized trial was to compare the effect of PbtO 2 -guided therapy (maintaining PbtO 2 > 20 mmHg) with traditional ICP-guided treatment (maintaining ICP < 20 mmHg and CPP > 60 mmHg) on the management of cerebral variables, therapeutic interventions, survival rates, and neurological outcomes in moderate and severe TBI patients. The outcomes of TBI patients in this study were evaluated on both the GOS and the Glasgow Outcome Scale-Extended (GOSE).

Subjects.
This human study was approved by the Institutional Review Board (IRB) of 6 collaborative hospitals in northern Taiwan, including Taipei Veterans General Hospital, Tri-Service General Hospital, En Chu Kong Hospital, Taipei Medical University Hospital, Taipei Medical University-Wan Fang Hospital, and Taipei Medical University-Shuang Ho Hospital. The study was explained to patients by investigators, coinvestigators, or the nursing and research staff, and a statement of informed consent was signed by each patient. The medical records used in this study were also evaluated as part of an observational TBI registry with IRB approval.
Moderate TBI patients with Glasgow coma scale (GCS) scores of 9-12 and severe TBI patients with GCS scores <8 were considered for recruitment from January 2009 to December 2010. Eligible patients were aged 17-70 years. Patients were excluded if they presented with severe systemic injuries with hypotension or multiple trauma, a GCS = 3 with fixed and dilated pupils, or a supplied 100% FiO 2 with saturation of arterial O 2 below 93%. Patients whose families refused aggressive treatment, as well as those transferred from another institution more than 24 hours after injury, were also excluded, as were patients suffering from open gunshot or stab wounds, postcardiac pulmonary resuscitation or obvious hypoxia, or other major underlying diseases such as uremia, liver cirrhosis, congestive heart failure with pulmonary edema, or coagulopathy.

Patient
Monitoring. Enrolled patients were randomized into ICP-and PbtO 2 -guided groups. Patients in the ICP group received an ICP monitor for use only in traditional ICP and CPP therapy. In the PbtO 2 -guided group, patients were treated using both ICP and PbtO 2 monitors for combined FiO 2 modulation with ICP and CPP management. An intraparenchymal or subdural ICP monitor (Codman electrode MicroSensor, Johnson and Johnson Medical, Ltd., USA) was used for monitoring the ICP signal in both groups. The ICP monitor was connected to an HP monitor (model 66s-M116A) through a pressure transducer and module (Codman neuromonitor interface control unit, 82-6605). In the PbtO 2guided group, an intraparenchymal brain tissue oxygen probe (LICOX REFIT2, Integra NeuroSciences, Ltd., England) was used for monitoring the PbtO 2 signal. The PbtO 2 signal was transmitted to an HP monitor through a transducer box and monitor cable (LICOX REF POM.BOX and LICOX REF NL950-MC-01, Integra NeuroSciences, Ltd., England). Both ICP and PbtO 2 monitors were placed in the margin (2-3 cm) of necrotic brain tissue (hematoma) of TBI patients, located by the penumbra in preoperative brain computed tomography scans.

Patient Management.
All patients were monitored by nursing staff in the NICU and were positioned in bed with a 30 ∘ head-up position. General monitoring included continual or intermittent assessment of mean arterial pressure, central venous pressure, electrocardiogram, pulse oximetry values, end-tidal CO 2 , and body temperature. Routine physical and pharmacological methods were adopted to prevent patient body temperature elevations above 37.5 ∘ C. Mechanical ventilation was adjusted to keep end-tidal CO 2 between 30 and 35 mmHg. Osmotic modulation, sedation, and decompressive craniectomy were used to control intracranial hypertension.
The major differences between the ICP-and PbtO 2guided groups were the treatment goals. In the ICP-guided group, ICP was predominantly maintained at <20 mmHg, and CPP was maintained at >60 mmHg. However, in the PbtO 2 -guided group, PbtO 2 was maintained at >20 mmHg, accompanied by ICP monitoring. Patients in the PbtO 2guided group were treated in compliance with standard procedures under 3 sets of conditions. If patients had intracranial hypertension (ICP > 20 mmHg) but a PbtO 2 > 20 mmHg, then the primary strategy was the treatment of intracranial hypertension with mannitol, glycerol, colloid, sedatives, or decompressive craniectomy, which is similar to traditional ICP-guided management. Vasopressor agents and hyperventilation could also be used to raise CPP appropriately, while controlling PbtO 2 .
If patients had marginal cerebral hypoxia (PbtO 2 < 20 mmHg) but ICP at <20 mmHg, PbtO 2 was normalized with a 100% FiO 2 challenge for 5 minutes. If PbtO 2 was increased by the FiO 2 challenge, then FiO 2 was slowly tapered while maintaining PbtO 2 . The 100% FiO 2 challenge was limited to 5 hours or less to avoid oxygen intoxication. If 100% FiO 2 was needed for longer than 5 hours, CPP was elevated up to 80 mmHg, and arterial carbon dioxide pressure (PaCO 2 ) was elevated to 40 mmHg to replace long-term highpercentage FiO 2 administration. If PbtO 2 was not increased by the FiO 2 challenge, then CPP and PaCO 2 could also be raised to resolve continuous cerebral ischemia after confirmation by brain computed tomography that the sensor tip was in place, and there was no evidence of heart failure or lung problems (pulmonary edema or acute respiratory distress syndrome).
Each of the 6 collaborative hospitals offered training courses in the standard protocol for multisite principal investigators, coinvestigators, and research assistants to maintain optimal patient management and avoid cluster effects caused by differences in equipment, faculty, and patient sources.

Data Collection.
Collected patient data included age, the initial GCS score, recruiting year, body mass index, injury etiology, the pathological status diagnosed by brain computed tomography, and operative status. Cerebral variables, such as GCS score, ICP, CPP, PbtO 2 , PaO 2 , and PaCO 2 , were noted during the first 5 days in the NICU, as well as the therapeutic administration of drugs such as mannitol, glycerol, colloids, vasopressor agents, and sedatives. Intracranial hypertension was indicated as ICP > 20 mmHg, whereas cerebral ischemia was defined as CPP < 60 mmHg. The CPP was calculated as the mean arterial pressure minus the ICP. Cerebral hypoxia was defined as PbtO 2 < 15 mmHg and was measured only in the PbtO 2 group. In addition, the survival rates and outcomes at 1, 3, and 6 months following injury were evaluated using the GOS and GOSE questionnaires. The collaborative hospitals held regular meetings to verify that all eligible patients were enrolled and that the data recording was complete.

Statistical Analysis.
After the data were cleaned and checked for completeness, we used an 2 test and an independent Student's test to compare categorical and continuous data between the ICP-and PbtO 2 -guided groups. Categorical variables included age, initial GCS score, recruiting year, etiology, pathological diagnosis, and operative status, as well as the appearance of intracranial hypertension, cerebral ischemia, therapeutic interventions, survival, and favorable outcome. Continuous variables included the mean age, initial GCS score, and cerebral monitoring over 5 days in the NICU; the data were expressed as means ± SD. Linear regression was used to further evaluate the relationship between PbtO 2 and GOS and GOSE scores. All statistical calculations were performed using SPSS, version 17.0 (SPSS, Chicago, IL, USA). Differences were considered statistically significant at < 0.05.

Results
Twenty-seven TBI patients were treated with traditional ICP-guided therapy (mean age 53.3 ± 20.1 y), whereas 23 patients were treated with PbtO 2 -guided management (mean age 53.7 ± 19.4 y). Table 1 shows the demographic data for both ICP-and PbtO 2 -guided groups. The distribution of age, initial GCS score, recruiting year, body mass index, etiology, pathological diagnosis, and operative status was similar across the 2 groups.
Cerebral monitoring variables over 5 days in the NICU are shown in Table 2. In the ICP-guided group, mean ICP was significantly higher ( = 0.017), and the intracranial hypertensive events (ICP > 20 mmHg) were almost 5 times more frequent (22.2% versus 4.3%) than in the PbtO 2 -guided group. The average CPP was significantly higher ( = 0.013) in PbtO 2 -guided patients compared with ICP-guided patients, although these signals were >60 mmHg in both groups. The mean GCS score and PaCO 2 showed no difference between the ICP-and PbtO 2 -guided groups. The mean PaO 2 was significantly elevated in the PbtO 2 -guided group compared with the ICP-guided group ( = 0.033). Figure 1 shows that the mean PaO 2 should be adjusted to >150 mmHg to prevent cerebral hypoxic events (PbtO 2 < 15 mmHg) in  patients with moderate to severe TBI. Therapeutic interventions were comparable across both treatment groups, as shown in Table 3.
In Table 4, we provided the mortality rate of each group. Survival rates were significantly higher in patients guided by PbtO 2 monitoring at 3 and 6 months postinjury compared   with those guided by ICP monitoring (Figure 2(a)). However, a favorable outcome was observed in <30% of patients in either group at any point after injury (Figure 2(b)). Mean GOS scores were 2.2-2.3 and 2.6-2.7, and mean GOSE scores were 2.4-2.6 and 3.0-3.2 in the ICP-and PbtO 2 -guided groups, respectively, from 1 to 6 months postinjury. Although no differences were significant in favorable outcome rates across the 2 groups, patients in the PbtO 2 -guided group had a 1.8-2.9 times more favorable outcome from 1 to 6 months postinjury than patients in the ICP-guided group. Figure 3 shows the correlation between the PbtO 2 signal and the outcome scale of TBI patients at 1, 3, and 6 months after injury. The PbtO 2 signals were significantly correlated with GOS scores at 1 and 3 months postinjury (Figures  3(a) and 3(b)) and with GOSE scores from 1 to 6 months postinjury (Figures 3(d)-3(f)). However, at 6 months postinjury, the correlation between PbtO 2 signals and GOS scores (Figure 3(c)) was still marginally significant ( = 0.060).

Discussion
Several studies have suggested that PbtO 2 is an independent factor related to the neurological outcome of severe TBI patients and weakly correlated with ICP or CPP signals [7,31]. Other investigators have reported a positive correlation between PbtO 2 and CPP under a specific range of CPP [19,32]. Although our results show that both ICP and CPP were managed in the reference range in most patients, significantly lower ICP and higher CPP were observed in patients treated with PbtO 2 monitoring compared with those treated with ICP monitoring alone. Treatment interventions were comparable between the 2 groups. Intracranial hypertensive events (ICP > 20 mmHg) in the PbtO 2 -guided group were rare in our study, which was similar to the previous results by Meixensberger et al. [24].
Previous studies have demonstrated that a cerebral hematoma that impairs blood flow and oxygen delivery into the brain tissue and induces secondary ischemia can progress with time after an intracerebral hemorrhage [33][34][35]. Furthermore, ischemia is a cause of brain edema surrounding the hematoma region [34,35]. Irrespective of whether maintaining cerebral oxygen consumption could reduce ischemiainduced edema, further improving CPP and normalizing ICP, even when these signals have already been controlled, is an important issue that requires further investigation.
Fewer patients ( = 1; 4.3%) in the current study had pulmonary complications in the PbtO 2 -guided group compared with the ICP-guided group ( = 3; 11.1%). Stiefel et al. suggested that treating severe TBI patients with causespecific management might result in improved survival rates and improved outcomes compared with treating ICP or CPP alone [27]. Increasing FiO 2 in patients with severe TBI has been studied as a strategy for achieving elevated PaO 2 and adequate PbtO 2 to avoid cerebral hypoxia and further improve outcomes by returning the cerebral metabolism to aerobic conditions [16,25,26,32,[36][37][38]. As shown in this study, the mean PaO 2 increased in PbtO 2 -guided patients. Therefore, a combination of FiO 2 modulation with traditional ICP and CPP treatment may simultaneously prevent hypoxia and other secondary complications, especially in pulmonary events. We also suggest that the PaO 2 targets in moderate and severe TBI patients should be different from those in general neurological patients. In moderate and severe TBI patients, PaO 2 must be adjusted to a value >150 mmHg, in contrast with the PaO 2 standard (PaO 2 > 60 mmHg), to prevent cerebral hypoxic events after trauma.
Our results indicate that patients guided by PbtO 2 monitoring had increased survival rates at 3 and 6 months postinjury. Nevertheless, mean GOS scores were <3, and favorable outcome rates were <30% in both groups at all the time points postinjury. A trend emerged toward greater favorable outcomes (1.8-2.9 times greater) in PbtO 2 -guided patients versus ICP-guided patients from 1 to 6 months after injury. These results are similar to those of other studies that observed a positive trend toward more favorable outcomes in severe TBI patients treated with PbtO 2 -guided monitoring [24,27]. The small difference in favorable outcomes between groups in our study was possibly due to most patients (>70%) having a TBI that was too severe to permit recovery. Spiotta et al. reported a study with more patients ( = 123) and noted a significantly improved short-term outcome in PbtO 2 -guided patients [29]. Therefore, the small sample may be another reason that differences in favorable outcome were undetectable for patients in our study. Despite the differences in favorable outcomes between the 2 groups in the study, positive correlations were noted between PbtO 2 signals and outcome scales, especially in GOSE scores, at 1, 3, and 6 months postinjury.
This study has several strengths and limitations. This was the first prospective randomized clinical trial studying the application of PbtO 2 monitoring for patients suffering from moderate to severe TBI. Additional strengths were shown in the collaboration among the 6 hospitals in northern Taiwan, which managed unselected brain-trauma patients, and held regular meetings to standardize the collection and recording of data. One limitation was the small number of patients enrolled, which may have resulted in an inability to detect differences in patient outcomes between the ICPand PbtO 2 -guided groups. However, the positive correlation between PbtO 2 signal and GOSE score reported here reflects the importance of PbtO 2 monitoring in TBI patients. Another limitation was that PbtO 2 and FiO 2 signals could not be compared across groups. The PbtO 2 was not monitored in the ICP-guided group, and FiO 2 was adjusted only according to PbtO 2 signals in the PbtO 2 -guided group during the first 5 days in the NICU, rather than continually recorded. Hence, cerebral oxygen consumption in the PbtO 2 -monitored patients was increased in accordance with increased PaO 2 in this study and in previously published studies [26,32].
In conclusion, this study demonstrates that (1) there was an increase in PbtO 2 related to increase of PaO 2 . (2) The PbtO 2 signals demonstrate a close correlation with patient outcomes from 1 to 6 months postinjury. Other than increase of PaO 2 , hemoglobin transfusion, decreasing oxygen demand (increased sedation, paralysis, and barbiturates use), and increased CO 2 (if ICP is controlled) can improve cerebral hypoxia. Therefore, we propose that PbtO 2 control is correlated with increase of PaO 2 and that therapy directed by PbtO 2 monitoring may be valuable in treating patients with moderate or severe TBI. In addition, increasing PaO 2 above 150 mmHg seems to efficaciously prevent cerebral hypoxic events after trauma. The mechanism and effects of PbtO 2 manipulation on well-controlled ICP (<20 mmHg) and CPP (>60 mmHg) TBI patients require further investigation.
Due to the small sample size in this study, we concluded that PbtO 2 -monitoring therapy might be beneficial in clinical care for managing moderate to severe brain injured patients. Firm conclusion shall be drawn in a larger and adequately designed study in the future.

Conflict of Interests
There is no potential conflict of interests to be disclosed.