Introduction

Direct invasive monitoring of brain tissue oxygenation (PbtO2) has been routinely utilized to predict cerebral ischemia and to prevent secondary injury in patients with traumatic brain injury (TBI) and vasospasm secondary to subarachnoid hemorrhage (SAH). Long-term follow-up has focused largely on the relationship between PbtO2 and neurologic outcome in adult patients following TBI. These studies have generally demonstrated that decreased brain tissue oxygenation is associated with poor outcome and interventions to increase brain tissue oxygenation improve outcome [1–12].

Smaller studies have focused on the impact of brain tissue oxygenation in the setting of severe TBI in children. Optimizing PbtO2 improved outcomes, even in the setting of optimal intracranial pressure (ICP), cerebral perfusion pressure (CPP), hemoglobin (Hb), and partial pressure of oxygen in arterial blood (PaO2) [13–16]. Despite widespread use in trauma and adult neuroscience intensive care units, there are no published studies on its use in patients with stroke and in particular pediatric stroke. In this manuscript, we present our early experience utilizing PbtO2 monitoring in two children with stroke. We postulate that brain tissue oxygen monitoring is a powerful tool that can be effectively used in optimizing pediatric neurocritical care after stroke.

Case Report

Patient 1

The first patient is a 3-year and 6-month old female struck by a motor vehicle. The patient did not lose consciousness and was brought to the emergency room (ER) within 30 min of the accident. On arrival, the patient had a Glascow Coma Score (GCS) of 15. Emergent head computed tomography (HCT) revealed no intraparanchymal, subdural, or epidural hemorrhage (Fig. 1a), a closed, displaced fracture of the left occipital bone (Fig. 1b), a non-displaced right temporal bone fracture with underlying SAH, and a left temporal bone fracture through the foramen magnum and the posterior margin of the left jugular foramen extending into the base of the left sphenoid bone and pterygoid plate (Fig. 1c). Luminal compression of the left internal jugular vein (IJ) versus intraluminal clot was noted, and a magnetic resonance venogram (MRV) was performed to better characterize this finding. Magnetic resonance imaging (MRI) and MRV revealed thrombosis of the left transverse and sigmoid sinuses extending to the left jugular vein (Fig. 1d). Incidentally noted on the MRI, the left internal carotid artery (ICA) was occluded secondary to the skull base fracture with distal reconstitution of flow to the left anterior cerebral artery (ACA) and middle cerebral artery (MCA) (Fig. 1e). Based on the presence of a venous sinus thrombosis and carotid dissection, the patient was started on anti-platelet therapy with aspirin despite the presence of intracranial hemorrhage. She remained at her neurologic baseline until approximately 40 h after admission, when she became combative and aphasic with a right hemiparesis. HCT suggested extensive cerebrovascular compromise within the left MCA territory and hyperdensity within the proximal sylvian fissure suggesting an embolic occlusion of the left MCA (Fig. 1f). The decision was made to perform an emergent hemicraniectomy (Fig. 1g) with insertion of an ICP monitor in the right frontal cortex and a PbtO2 probe at the penumbra of the ischemic area in the left frontal lobe (Fig. 1h). Immediate post-operative angiogram confirmed dissection of the post-bifurcation ICA (Fig. 1i) and revealed the MCA occlusion and the route through which the patient had adequately perfused her left hemisphere before the embolic event, through collateral flow from her posterior circulation (Fig. 1j).

Fig. 1
figure 1

Imaging for patient 1 throughout hospital course. a HCT, no signs of intraparenchymal, subdural, or epidural hemorrhage. b HCT, closed and displaced fracture of the left occipital bone. c HCT, non-displaced skull base fracture through the foramen magnum and left jugular foramen. d MRV, thrombosis of the left transverse and sigmoid sinuses extending to the left jugular vein. e MRI, occlusion of left ICA with distal reconstitution of flow to the left ACA and left MCA. f HCT, extensive cortical ischemia in the territory of the left MCA and hyperdensity within proximal Sylvain fissure. g HCT, post-hemicraniectomy. h HCT, placement of PbtO2 probe within penumbra. i Angiogram, dissection of the left post-bifurcation ICA. j Angiogram, collateral perfusion of left hemisphere through the posterior circulation. k, l Post-intervention MRI, no new infarcts and expected evolution of previous ischemic area in the left MCA territory

Full size image

Post-operatively, the patient was sedated, paralyzed, cooled to 32° and placed on a ventilator to allow optimization of ICP, CPP, mean arterial pressure (MAP), and PbtO2. Our goals were standard: maintenance of ICP < 20 mmHg, CPP > 40 mmHg, MAP > 60 mmHg, but added the criteria of PbtO2 > 20 torr which is based on adult literature. The patient received several directed care decisions over the next 2 days, primarily for optimization of PbtO2 (Table 1). During 4 days of monitoring, PbtO2 dropped below threshold six times and ICP was elevated above threshold three times. These increases in ICP correlated with a decrease in PbtO2 at only one time point. All other measurements remained well above established thresholds for the duration of monitoring. Imaging was not performed during this period of intensive care; however, imaging performed after removal of the PbtO2 probe demonstrated no new infarcts and expected evolution of the previous ischemic area in the left MCA territory (Fig. 1k, l).

Table 1 Physiological response to interventions in brain hypoxic episodes for patient 1
Full size table

On post-operative day (POD) #3, the patient’s PbtO2 stabilized above 20 torr, and the PbtO2 probe was removed on POD #5. The patient was slowly weaned from sedation and paralytics and was successfully extubated on POD #11. On extubation, the patient was following commands on the left, with minimal spontaneous movement on the right. Her exam steadily improved over the next 2 months, and she was discharged to a rehabilitation facility with normal arousal and interaction, mild impairment in speech production, full strength on the left, and mild decrement in strength on the right. Full psychomotor and neurologic evaluation 6 months post-stroke, found normal verbal IQ, full visual fields, and a mildly spastic hemiparesis. She is independently ambulatory with an ankle foot orthosis and resumed full day nursery school 1 year after her accident.

Patient 2

The second patient is a 4-year and 6-month old female with stage 4 neuroblastoma who presented unresponsive with dilated, minimally reactive pupils, and decorticate posturing. HCT revealed a large left frontal intraparenchymal hemorrhage (IPH) with a presumed underlying metastasis (Fig. 2a). She was taken emergently to the operating room for surgical evacuation of the hematoma and metastatic lesion. Post-operatively, the patient was kept intubated and sedated to maximize control of ICP. MRI of the brain on POD #1 revealed bilateral caudate (Fig. 2b), left putaminal, and left medial frontal lobe infarcts (Fig. 2c), consistent with transtentorial herniation. A small focus of enhancement along the left M1 segment was also noted, indicating possible leptomeningeal disease (Fig. 2d). The patient was successfully extubated on POD #1 but remained somnolent. She moved her left upper extremity (LUE) spontaneously and purposefully with minimal spontaneous movement of the right upper extremity (RUE). Bilateral lower extremities demonstrated spontaneous movement. The patient experienced new onset seizure activity on POD #4, and HCT demonstrated progression of prior ischemia with new areas of susceptibility in the distribution of the left anterior, middle, and posterior cerebral arteries (Fig. 2e). The patient was again sedated, intubated, and taken emergently for cerebral angiography, which demonstrated no evidence of vasospasm or vascular occlusion (Fig. 2f, g). Subsequent MRI showed bilateral hemispheric ischemia, worse on the left than the right (Fig. 2h). Based on the fact that the etiology of the vascular compromise was unknown, and the patient’s clinical exam was difficult to monitor, the decision was made to place a right-sided external ventricular drain (EVD) and bilateral PbtO2 probes in an attempt to protect the ischemic regions from infarction. Placement of the left PbtO2 probe was complicated by an ICP spike to 50 torr with arterial bleeding requiring gel foam and thrombin to achieve hemostasis; placement of the left PbtO2 probe was therefore aborted. Following this intervention, the patient was managed with escalating sedation, mannitol, and initiation of paralytics and pressors. HCT post-procedure demonstrated no evidence of intracranial bleeding and good placement of the EVD and PbtO2 probe (Fig. 2i).

Fig. 2
figure 2

Imaging for patient 2 throughout hospital course. a HCT, left frontal IPH with underlying metastatic lesion. b MRI, bilateral caudate and putaminal infarcts. c MRI, left medial frontal lobe infarct. d MRI, small focus of enhancement along left M1 segment. e HCT, progression of previous ischemia and new areas of susceptibility in distribution of left ACA, MCA, and MCA. f, g Angiogram, no evidence of vasospasm or vascular occlusion. h MRI, bilateral hemispheric ischemia, L > R. i HCT, placement of EVD and PbtO2 probe with no evidence of intracranial hemorrhage. j HCT, stable infarcts with no new ischemia. k, l MRI, evolution of infarction in bilateral cerebral hemispheres and basal ganglia

Full size image

After placement of the monitoring devices, the patient’s ICP remained elevated above threshold despite increased sedation, cerebrospinal fluid (CSF) drainage, increased pentobarbital, mild hyperventilation, and hyperosmolar therapy; PbtO2 remained normal. A bifrontal decompressive craniectomy was performed and a left Licox probe was placed. The patient required multiple interventions over the next 2 days to maintain PbtO2 > 20 torr (Table 2). Over the period of PbtO2 monitoring, subsequent to placement of bilateral PbtO2 probes, there were six points during which PbtO2 was below threshold; none of these events correlated with abnormal values of ICP, CPP, or MAP, with ICP, CPP, and MAP remaining above threshold. However, two of these declines in PbtO2 occurred in conjunction with decreases in systemic oxygenation.

Table 2 Physiological response to interventions in brain hypoxic episodes for patient 2
Full size table

Imaging performed 3 days after placement of PbtO2 probes showed no evidence of new infarcts. On the following day, the probes were removed after more than 24 h with stable readings. The patient was subsequently extubated and imaging on the day of extubation again demonstrated stable infarcts with no new areas of ischemia (Fig. 2j). At discharge, the patient was awake and minimally responsive with choreoathetoid movements of her left hand. MRI before discharge demonstrated expected evolution of multifocal areas of restricted diffusion within the bilateral cerebral hemispheres (Fig. 2k) and bilateral basal ganglia (Fig. 2l), likely secondary to the patient’s herniation on presentation. There were no new areas of ischemia.

Discussion

Over the past decade, brain tissue oxygenation (PbtO2) has been studied in the setting of traumatic brain injury, subarachnoid hemorrhage, and intra-operatively during aneurysm clipping [3–5]. The placement of the PbtO2 probes has been associated with low morbidity, and reliability of data has been confirmed after the first hour of insertion [2, 4, 5]. As such, the application of this technology has been sporadically utilized in the neuro-intensive care unit, primarily in the setting of traumatic brain injury. There are, however, unknown variables in the application of PbtO2 monitoring to stroke patients. PbtO2 monitors have traditionally been placed in the right frontal lobe. The safety of proposed utilization of these probes in the penumbra may theoretically carry a higher risk of hemorrhage. Further studies determining safety are needed for this particular patient population.

Multiple studies demonstrate that episodes of cerebral hypoxia (PbtO2 < 20 torr) may occur following trauma, even in the setting of optimal ICP and CPP. Longer duration and severity of low PbtO2 is significantly correlated with outcome [8, 9, 11]. Stiefel et al. [6, 8] have shown that PbtO2-based treatment in the setting of TBI significantly improved mortality and outcome, although Meixensberger et al. report no change in the patients with optimized brain oxygenation versus those without. It appears based on this data, in addition to ICP, CPP, TCD, and microdialysis, that PbtO2 monitoring may be important to prevent secondary injury following acute TBI.

The safety and utility of PbtO2 monitoring in pediatric TBI has also been studied [7, 13–16]. Figaji et al. [13, 15, 16] have reported that cerebral hypoxia is common in the setting of normal ICP, CPP, systemic oxygenation, and normal hemoglobin following trauma in children, and that longer duration of hypoxia is associated with poor outcome and higher mortality. Narotam et al. [7] found that institution of eubaric hyperoxemia via increased FiO2, pressors, or transfusion to treat low PbtO2 resulted in lower mortality and improved outcome in pediatric patients.

Given the use of PbtO2 monitoring to prevent secondary ischemia in trauma and SAH, application of this technology to prevent brain infarction in stroke patients is a logical extension. Hesitancy to extend hyperoxemia to stroke patients has traditionally stemmed from concern over reperfusion injury, free-radical effect on tissue, the effect of neuronal hyperexcitability on NMDA and glutamate receptors, and the deleterious systemic effects of hyper-oxygenation through increased systemic vascular resistance, decreased heart rate, and decreased cardiac output [17]. Animal models of stroke, however, have demonstrated that hyperoxia can reduce the volume of ischemic territory on MRI. This effect correlated with the time to treatment in relation to the time of injury, and there was no increase in the risk of hemorrhage during reperfusion [17–19]. Singhal et al. [20] studied the effects of normobaric oxygen therapy in humans at the onset of stroke, and found that therapy initiated within 12 h of vascular insult improved clinical outcome and ischemia on MRI. In a follow-up study, they additionally found hyperoxia to improve brain metabolism in the ischemic penumbra [21]. Dani et al. [22] again showed improved metabolism in hyperacute ischemic penumbra following hyperoxia. Hoffman et al. [23] have shown the ability of PbtO2 monitors to detect cerebral ischemia during intra-operative arterial occlusion and aneurysm clipping, confirming the appropriateness of this monitor in detection of early brain ischemia. Implementation of a monitoring mechanism to detect early ischemia in stroke patients, and PbtO2 directed therapy to prevent further infarction, appears to be feasible.

We therefore report on our experience using PbtO2 monitors in two children with stroke with the aim of initiating investigation into whether current standard systemic and intracranial measurement parameters adequately predict and reflect cerebral hypoxia, and whether alteration of PbtO2 affects outcome in this population. In the two cases presented here, initial imaging indicated a primary ischemic event, and PbtO2 probes were placed for the purpose of protecting the penumbra and minimizing the territory of infarction. While there were instances where ICP and PbtO2 were simultaneously compromised, brain oxygenation was also compromised in the setting of normal ICP, MAP, and peripheral oxygenation. For every decrease in PbtO2 below threshold, targeted-therapy was performed which resulted in return of PbtO2 to target within 1 h. Temporary normobaric hyperoxia was frequently used to address acute drops in PbtO2, with weaning occurring as PbtO2 stabilized. While ischemic brain may respond favorably to hyperoxia, prolonged increases in FiO2 have been shown to cause minor decreases to global cerebral blood flow [24–26]. Respiratory rate and PEEP were also used to increase the systemic delivery of oxygen, especially in the patient who was unable to maintain systemic oxygenation at lower values. Pressors were used to increased mean arterial blood pressure and, by effect, global cerebral blood flow. Infrequently, pentobarbital was given to increase PbtO2, which acts theoretically by decreasing the demand and consumption of oxygen in the brain [27–29].

For both patients, imaging following the period of PbtO2 monitoring demonstrated stability in infarct size. It is likely that protection from progression of ischemic injury was influenced to a large extent by control of ICP, as well as by maintenance of normal brain oxygenation. Due to the fact that we have reviewed our experience in only two patients, it is impossible to separate the effect of PbtO2 from the other commonly monitored parameters.

Conclusions

There is currently inadequate data to support the application of PbtO2 monitoring in children with stroke to prevent progressive ischemia and to improve outcome. In young patients with cerebral ischemia, optimization of brain oxygenation is assumed in the presence of ICP < 20 mmHg, age-appropriate CPP, arterial O2 > 60 torr, normotension, and Hb >8 g/dl [13, 30]. Based on extensive study by Figaji et al., however, it is apparent that these measures do not accurately predict brain tissue oxygen tension. Clinical outcome in the patients with primary stroke is highly dependent upon the territory and duration of ischemia. Logic dictates that maintaining adequate brain tissue oxygen in the aftermath of pediatric stroke may be a means to minimize the ischemic territory and improve clinical outcome.