Measuring Brain Tissue Oxygenation Compared with Jugular Venous Oxygen Saturation for Monitoring Cerebral Oxygenation After Traumatic Brain Injury

Abstract

Jugular bulb oximetry is the most widely used method of monitoring cerebral oxygenation.More recently, measurement of brain tissue oxygenation has been reported in head-injured patients. We compared the changes in brain tissue oxygen partial pressure (PbO₂) with changes in jugular venous oxygen saturation (SjVO₂) in response to hyperventilation in areas of brain with and without focal pathology. Thirteen patients with severe head injuries were studied. A multiparameter sensor was inserted into areas of brain with focal pathology in five patients and outside areas of focal pathology in eight patients. A fiberoptic catheter was inserted into the right jugular bulb. Patients were hyperventilated in a stepwise manner from a PaCO₂ of approximately 35 mm Hg to a PaCO₂ of 22 mm Hg. There was no significant change in cerebral perfusion pressure or arterial partial pressure of oxygen with hyperventilation. In areas without focal pathology, there was a good correlation between changes in SjVO₂ and PbO₂ (Delta SjVO₂ and Delta PbO₂; r² = 0.69, P < 0.0001). In areas with focal pathology, there was no correlation between Delta SjVO₂ and Delta PbO₂ (r² = 0.07, P = 0.23). In this study, we demonstrated that measurement of local tissue oxygenation can highlight focal differences in regional cerebral oxygenation that are disguised when measuring SjVO₂. Thus, monitoring of PbO₂ is a useful addition to multimodal monitoring of patients with traumatic head injury. Implications: Brain oxygenation is currently monitored by using jugular bulb oximetry, which attracts a number of potential artifacts and may not reflect regional changes in oxygenation. We compared this method with measurement of brain tissue oxygenation using a multiparameter sensor inserted into brain tissue. The brain tissue monitor seemed to reflect regional brain oxygenation better than jugular bulb oximetry.

(Anesth Analg 1999;88:549-53)

Section Editor: David S. Warner.

After severe head injury, patients are at risk of developing secondary cerebral hypoxia and ischemia, which causes additional brain damage ^[1,2]. These insults may occur more frequently than previously assumed and often go undetected ^[3]. Both global and regional cerebral blood flow (CBF) and cerebral metabolism are disturbed during the first few hours and days after injury. This change in CBF is rarely uniform throughout the brain with evidence of profound reductions of flow in pericontusional areas ^[4]. The regional heterogeneity of CBF and metabolism in the brain is now well recognized ^[5-7].

The primary therapeutic aim in severe head injury is directed at maintaining cerebral perfusion and avoiding tissue hypoxia. Continuous monitoring of jugular venous oxygen saturation (SjVO₂) has become an accepted method of monitoring global cerebral oxygenation and metabolism, but it may be difficult to use for continuous monitoring in clinical practice. Catheters require constant care and frequent recalibration if artifactual data are to be excluded. More recently, sensors measuring brain tissue partial pressure of oxygen (PbO₂), carbon dioxide (PbCO₂), and brain pH have been used to measure regional changes in tissue oxygenation and may provide an alternative, less artifactprone method of assessing cerebral oxygenation ^[8-11].

In the context of head injury, the use of tissue oxygen sensors is significantly different from SjVO₂ monitoring with respect to one other important attribute. Whereas SjVO₂ monitoring provides an assessment of global cerebral oxygenation and metabolism, tissue oxygen sensors reflect metabolic conditions in a much smaller area. Regional changes measured by these multiparameter sensors have the disadvantage that they may not reflect physiology across a metabolically heterogeneous brain. However, the limited field of view may be used to advantage. Careful placement of tissue sensors in structurally normal and abnormal areas may allow us to monitor metabolic conditions in targeted regions, thus providing spatially (and, hence, pathologically) resolved, on-line data concerning brain oxygenation and biochemistry.

The aim of this study was to measure PbO₂ during graded hyperventilation in head-injured patients and to compare changes in PbO₂ (Delta PbO₂) with changes in SjVO₂ (Delta SjVO₂) in areas with and without focal pathology.

Methods

After local research ethics committee approval and with written consent from the patients' next of kin, we studied 13 patients with severe head injuries. All patients were sedated with IV infusions of propofol (2-5 mg [center dot] kg^-1 [center dot] h^-1), fentanyl (2-3 [micro sign]g [center dot] kg^-1 [center dot] h^-1) and paralyzed with atracurium (0.5-1.0 mg [center dot] kg^-1 [center dot] h^-1). After tracheal intubation, the lungs were mechanically ventilated.

Routine invasive monitoring of these patients included mean arterial pressure (MAP), central venous pressure (CVP), intracranial pressure (ICP), and SjVO (₂). Pulmonary artery catheters were inserted into those patients requiring inotropic support, those with preexisting cardiac disease, or those with significant myocardial contusion.

A multiparameter sensor (Paratrend 7[trade mark sign], Diametrics Medical, High Wycombe, UK) was used to measure PbO₂, PbCO₂, brain pH, and temperature. This sensor, which is 0.5 mm in diameter, contains a miniature Clarke electrode for measuring PbO₂, two fiberoptic elements to measure PbCO₂ and pH, and a thermocouple to measure temperature. Although this sensor was originally designed for continuous arterial blood gases measurement, its use as a device for brain tissue measurement has been evaluated in animal models ^[8].

Before insertion, the sensor was calibrated with three precision gases. The first calibration gas contained 2% CO₂ and 15% O₂, the second contained 5% CO₂ and 15% O₂, and the third contained 10% CO₂ and 15% O₂. Two sensors were inserted into each patient. One was inserted through an 18-gauge cannula into a femoral artery for continuous blood gas analysis. The second was inserted into brain tissue through a modified Camino bolt, which was placed adjacent to the ICP monitoring bolt. Subsequent computed tomographic (CT) scans of the head identified whether the sensor was placed in an area of focal pathology (e.g., contusion) or in an area with no focal lesion.

Baseline readings were taken after a 30-min equilibration period. The arterial Paratrend 7[trade mark sign] sensor readings were verified with intermittent arterial blood gas estimations, and SjVO₂ readings were calibrated with an aspirated sample before the protocol and again if there was a poor signal quality on the monitor or if artifact was suspected during the course of the protocol. A calibration sample was also taken at the end of the protocol. Patients were then hyperventilated in a stepwise manner by decreasing PaCO₂ concentrations by 4-mm Hg increments, down to a level of 22 mm Hg. Readings were taken at each PaCO₂ level after stabilization for 15 min. Hyperventilation was abandoned if SjVO₂ decreased below 50%.

MAP, ICP, SjVO₂, arterial blood gases, PbO₂, PbCO₂, brain pH, and temperature were measured. Cerebral perfusion pressure (CPP) was calculated by the relationship CPP = MAP - ICP. Signals were time averaged over 4 s, digitized, and collected on a laptop computer using specialized multimodality software ^[12]. Data were analyzed using repeated-measures analysis of variance, and results were considered significant when P < 0.05. Correlation plots were obtained using a commercially available statistics package.

Results

We studied eight men and five women with a mean age of 50 yr (range 26-78 yr). All patients had Glasgow Coma Scale scores of <8 (median score 5). CT scans after insertion of monitors revealed that the Paratrend 7[trade mark sign] sensors were inserted into areas of focal pathology in five patients and into areas without focal pathology in eight patients. No complications were associated with insertion of the cerebral Paratrend 7[trade mark sign]. Intermittent blood gas analysis demonstrated accurate readings and long-term stability from the femoral artery Paratrend 7[trade mark sign] sensors.

(Table 1) shows the variation of CPP and PaO₂ at different levels of PaCO₂. CPP was maintained at a mean of 81 mm Hg, and there were no significant or systematic changes in CPP or PaO₂ with changes in PaCO₂. Figure 1 shows representative CT scans of the sensor in "normal" brain (Figure 1a) and in an area of focal pathology (Figure 1b). There was no significant difference among PbO₂ values at different PaCO₂ levels (P = 0.4). The response of PbO₂ to PaCO₂ changes in the two groups revealed no significant change in PbO₂ with hyperventilation in the areas without pathology (P > 0.49), or in regions with focal pathology (P = 0.12) (Figure 2).

Table 1: Cerebral Perfusion Pressure (CPP) and PaO₂ at Different Levels of PaCO₂

Figure 1: a, Computed tomographic scan of tissue sensor in area of no focal injury. b, Computed tomographic scan of tissue sensor in area of focal pathology (arrow denotes tip of sensor).

Figure 2: Effect of decreasing PaCO₂ on the change in tissue oxygen partial pressure (Delta PbO₂) in the groups with focal pathology and with no focal pathology.

Although there was a trend for SjVO₂ values to decrease with hyperventilation, these changes did not reach significance within the range of PaCO₂ values studied (P = 0.19).

We compared changes in PbO₂ and SjVO₂ produced by hyperventilation. For all patients there was a poor correlation between Delta SjVO₂ and Delta PbO₂ (r² = 0.274, P < 0.0001). However, there were clear differences in the Delta SjVO_2/Delta PbO₂ relationship in the two patient groups (Figure 3 and Figure 4). There was little correlation between Delta SjVO₂ and Delta PbO₂ in the focal group (r² = 0.07, P = 0.23), whereas there was a strong correlation between these variables in the nonfocal group (r² = 0.69, P < 0.0001).

Figure 3: Change in jugular venous oxygen saturation (Delta SjO₂) plotted against change in tissue oxygen partial pressure (Delta PbO₂) with the sensor in an area of focal pathology (r² = 0.07, P = 0.23).

Figure 4: Change in jugular venous oxygen saturation (Delta SjO₂) plotted against change in tissue oxygen partial pressure (Delta PbO₂) with the sensor not in an area of focal pathology (r² = 0.69, P < 0.0001).

Discussion

The findings of this study show that as PaCO₂ decreases PbO₂ does not change significantly. However, when the patients were separated by sensor site pathology, there was a gradual decrease in Delta PbO₂ in the areas with no focal pathology, whereas there was an increase in Delta PbO₂ in areas of focal pathology as PaCO₂ decreased. The decrease in Delta PbO₂ in areas without focal pathology concurs with findings of other investigators who used brain tissue probes during hyperventilation therapy ^[8,13-15]. The increase in PbO₂ in areas of focal pathology may reflect disturbed vascular reactivity in these areas. Indeed, Marion and Bouma ^[16] suggested that regional heterogeneity of CO₂ vasoresponsivity is much more common than global estimates have suggested.

The limitations of SjVO₂ monitoring are well recognized. Continuous SjVO₂ monitoring has a number of technical difficulties, whereas intermittent sampling only gives a "snapshot" of the state of cerebral oxygenation. Aspiration of blood from the jugular bulb should ideally be representative of mixed cerebral blood. Stocchetti et al. ^[17]. attempted to identify the most appropriate side for catheterization of the jugular bulb by simultaneously sampling both jugular veins. The variation between the two sides was insignificant; thus, neither side was shown to be more appropriate.

Our results suggest a further drawback of SjVO₂ monitoring. The lack of correlation between changes in SjVO₂ and PbO₂ in regions of focal pathology implies that SjVO₂ monitoring is poor at assessing the adequacy of oxygenation in areas that are most at risk from secondary physiological insults. These findings add weight to suggestions from Andrews et al. ^[18] and Chieregato et al. ^[19] that SjVO₂ alone should not be relied on when assessing cerebral oxygenation in the brain injured patient. Indeed, Keining et al. ^[20] suggested that measurement of brain tissue PO₂ is more suitable than SjVO₂ in head-injured patients. Adverse changes in cerebral oxygenation in a head-injured patient detected by a brain tissue sensor but not identified by jugular venous bulb oximetry indicate that a normal SjVO₂ does not guarantee against regional ischemia ^[21].

These data confirm regional heterogeneity in physiological responses between injured and noninjured brain. They also underline the value of targeted tissue PO₂ monitoring for following such physiology. When the well-being of the noninjured brain is in question, the placement of a tissue PO₂ sensor in noninjured areas can provide a method of monitoring cerebral oxygenation that may be more invasive, but that is more reliable and technically easier to maintain. Conversely, placement of probes in injured regions may allow the selective monitoring of tissue most at risk. These considerations may prompt the placement of multiple PO₂ sensors in selected patients. Furthermore, the strong correlation between the recordings of changes in SjVO₂ and PbO₂ indicate that the tissue sensor reflects changes in tissue oxygenation in a wider compartment of brain tissue, rather than a smaller area resulting from local trauma due to probe insertion and disruption of the blood-brain barrier. The in vivo validation of the tissue sensor against a "gold standard" is presently underway.

In summary, in this study, we demonstrated a difference in the response of brain tissue oxygen with changes in PaCO₂ in areas of brain with focal pathology compared with areas without focal pathology. Furthermore, changes in SjVO₂ did not correlate with changes in PbO₂ when patients' data were pooled, which implies that measurement of brain tissue oxygenation may be a more sensitive measure of regional oxygenation than jugular bulb oxygen saturation.