New Approaches to Bedside Monitoring in Stroke
Patients with intracranial hemorrhage have may have seizures or abnormal EEG, which can be associated with worsened neurological outcomes and are not related to underlying structural changes seen on noncontrast head computed tomography (CT), such as periodic lateralized epileptiform discharges. Finally, 19% of medical and neurologically critically ill patients who underwent diagnostic CEEG had abnormal EEG (Figure 4). Of these patients, 95% had NCSs. Therefore, EEG may help guide anticonvulsant prophylaxis or treatment as shown in SAH patients, in whom routine prophylaxis was associated with worse cognitive outcomes.
Table 1
(Enlarge Image)
Figure 4.
Continuous EEG and quantitative EEG example. (A) The regular EEG with suppression-burst pattern in a patient with cardiac arrest and (B) the corresponding image of the quantitative EEG over a longer period of time with discrete 'spikes' on Rosetta seizure probability index (red upper part of figure) and corresponding discrete dark blue areas of Rhythmic Run Detection Display for each hemisphere (middle panels below Rosetta), which correspond to high electrical bursts as seen in regular EEG. The asymmetry index next to the bottom is another quantitative EEG panel that can be useful in stroke if there is hemispheric asymmetry to trend in terms of electrical power. The bottom Q-EEG panel 'aEEG 25–75%' is useful if there is a normal a pattern to trend for differences between hemispheres. In this patient's case the amplitude-integrated EEG is confounded by large electrical bursts shown in the upper panel followed by near isoelectric EEG pattern. Copyright Mayo Foundation for Medical Education and Research.
Transcranial Doppler (TCD) uses sound waves typically around the range of 2–5 MHz through the skull to find intracranial flow of the main circle of Willis arteries such as the first segments of the middle cerebral artery (i.e., MCA-M1 segment), the anterior cerebral artery first segment (ACA-A1) and posterior cerebral artery first segment (PCA-P1). Evaluation of these proximal arterial segments has been well studied in the stroke subtype of SAH in obtaining baseline mean flow velocities (MFVs) and following trends. This is based on the physics of 'flow', which is perhaps best derived conceptually from Ohm's law, V = IR, where V = voltage or pressure difference, I = current or flow, and R is resistance. TCD only measures flow (i.e., the velocity of MCA-M1 blood flow in cm/s). However, as flow changes over time, based on baseline MFV trends, one can infer that resistance might be changing and in the case of VSP (or narrowing of the MCA-M1 due to luminal stenosis) the MFV increases. One must also understand that flow can artifically increase with measures that increase cardiac output to treat VSP, such as use of vasopressors and dopamine. Such increases in systemic cardiac output will increase MCA-M1 MFV as well. Therefore, the Lindegaard ratio (also known as hemispheric ratio) is used to account for hemodynamic augmentation. The Lindegaard ratio is the MCA-M1 MFV divided by the ipsilateral internal carotid artery MFV (i.e., MCA/ICA). Thus, true VSP or luminal narrowing will typically have a Lindegaard ratio of three or greater, whereas hyperdynamic states with increased flow are typically less than three.
TCD VSP sensitivity varies between 50 and 100%, and is vessel-dependent due to location and size, but has a specificity of >90% as compared to the 'gold standard' of digital subtraction angiography (Table 1). TCD is also heavily operator-dependent in terms of education and experience. The ACA and PCA arteries are also more difficult to obtain compared with MCA, no matter which machine is utilized to get a continuous wave (Figure 5B) versus transcranial color Doppler imaging (TCDI) or pulse wave (PW) (Figure 5A), hence sensitivity is reduced for these vessels (Table 2).
(Enlarge Image)
Figure 5.
Sensitivity and specificity of tests for vasospasm. (A) Example of transcranial color Doppler imaging (TCDI) using Sonosite with labeling of the anterior cerebral artery and middle cerebral artery (MCA)-M1 vessels. The hypoechoic structure off to the right of the red and blue color is the midbrain in cross-section and is one of the intracranial landmarks that can be seen with TCDI that is not possible to visualize with routine 'blind' transcranial color Doppler. (B) The 'blind' or non-imaging-based transcranial color Doppler via transtemporal window. The upper part of the image is the depth-based screen with red colors indicating intracranial flow headed toward the probe, whereas blue goes away from the probe and is the anterior cerebral artery. Therefore, the upper flow pattern with the horizontal (yellow) line is the MCA. This can only be known by an experienced examiner who knows the depth of the MCA and other intracranial artery segments, which is also seen on the left around 5 cm from the temporal bone probe location. The advantage of TCDI is you can see the anatomic landmarks and vessels as shown in (A), which is essentially the Circle of Willis. Copyright Mayo Foundation for Medical Education and Research.
There is some concern whether TCDI should not include angle correction, given conflicting studies between TCD continuous wave and pulse wave or TCDI-obtained MFV values, which are perhaps best studied in sickle cell disease. The summary of data in patients with sickle cell disease comparing TCDI and continuous PW TCD has shown that TCDI velocities are lower by approximately 10–20% compared to TCD velocities. A Mayo Clinic Florida institutional review of select patients in which angle correction was carried out and compared against a nonangle correction showed a 10–20% difference, comparable with larger studies. This 'corrected MFV' of approximately 20–30% typically affected the ACAs and PCAs since the MCA had little need for angle correction similar to Krejza. However, TCDI might be able to identify the major cerebral arteries more effectively than TCD. It has also been our experience that TCDI may have difficulty with those with thick temporal bone (Figure 6), approximately >8 mm or a combination of bone thickness and high Hounsfield units (HU; bone density).
(Enlarge Image)
Figure 6.
In patients where thickness of the skull bone is an obstacle to obtain a window for transcranial Doppler pulse waves, near-infrared spectroscopy may be an alternative noninvasive method for continuous screening for early signs of vasospasm. HU: Hounsfield units; ICA: Internal carotid artery; MCA: Middle cerebral artery; NIRS: Near-infrared spectroscopy; TCD: Transcranial Doppler. Copyright Mayo Foundation for Medical Education and Research.
Overall, for SAH VSP stroke detection, the American Academy of Neurology has published guidelines suggesting TCD monitoring of the basal cerebral arteries (MCA-M1, PCA-P1 and ACA-A1). This obviously depends on the skill and resources of each individual institution. For example, some institutions perform this monitoring 7 days a week, 365 days a year, like ours, whereas others perform this monitoring only from Monday to Friday, which can lead to decreased detection of VSP.
An established method to measure hemispheric brain mixed arteriovenous oxygenation is via the jugular venous oxygen saturation (SjvO2). Cannulation of the jugular venous bulb allows continuous monitoring of the delivery and consumption of oxygen in the brain. The catheter is equipped with an oximeter at the tip, and accuracy depends on correct positioning. This method requires systemic oxygen saturation, hematocrit and calculations of arteriovenous O2 saturation differences (AVDO2). As 70% of the cerebral venous blood drains via the ipsilateral jugular veins, some clinicians advocate cannulating at the side of injury. In the case of diffuse cerebral injury, most clinicians would monitor the right side, as it is commonly dominant, whereas some clinicians would advocate monitoring the side of dominant flow in all situations. The interpretation of SjvO2 values remains nevertheless complex, with a constant cerebral oxygen consumption, and with interpretation in the context of the hematocrit and SjvO2 variations correlates with CBF variations. The interpretation is complex, based on 'supply' and 'demand' physiology such as cardiac output, hemoglobin, and factors such as cerebral O2 demand, and the impact of oxygen extraction (e.g., fever, diseased or normal brain O2 uptake) (Table 3). Also, interpretation should correct for reduced systemic oxygen saturation and reduced oxygen demand due to hypothermia, sedatives, pathologic arterial–venous communications and brain death, which may result in increased SjvO2. After brain trauma, SjvO2 <50 or >75% is associated with a poor prognosis. To maintain SjvO2 >50% constitutes a reasonable therapeutic goal, but the benefit associated with such a strategy has not yet been validated. The jugular venous SjvO2 method is invasive and there are risks of complications including pneumothorax, arterial puncture, thrombosis and infection.
Near-infrared spectroscopy (NIRS) is a relatively new bedside application, which facilitates clinicians with a noninvasive method of assessing regional cerebral arteriovenous oxygenation similar to SjVO2 (Figure 7). The NIRS noninvasive method of measuring cerebral oxygen saturation is of increasing scientific interest, but has a largely undefined role in the daily management of most ICU stroke patients. The high prevelance of VSP after SAH and secondary ischemic brain injury may increase the need for brain monitoring in this patient population. However, current modalities have significant limitations. Modalities such as SjVO2 indicate changes in arterial blood flow without direct brain tissue measurement; other more standardized methods such as PET, MRI with perfusion weighted imaging, CT perfusion and xenon CT offer excellent spatial resolution, but require transportation of critically ill patients and potential exposure to harmful radiation. NIRS offers the advantage of noninvasive, radiation-free, continuous and real-time estimations of brain tissue arteriovenous oxygenation ( Table 4 ).
(Enlarge Image)
Figure 7.
Near-infrared spectroscopy portrays correlation in oxygenation parameters as defined by jugular venous oxygen saturation and pulse oximetry blood sampling. NIRS: Near-infrared spectroscopy; SaO2: Arterial oxygen saturation measured on arterial blood gas; SctO2: Cerebral tissue oxygen saturation; SjvO2: Jugular venous oxygen saturation; SO2: Oxygen saturation. Adapted with permission from [103].
NIRS optodes (two) are placed bifrontal like that of an adhesive bandage separated by an interoptode distance of 4–5 cm. Simultaneous measurement of the two optodes enables left and right frontal hemispheric quantification. Cerebral oximetry measurements by the NIRS are regional and measure approximately 2.5 cm below the NIRS optodes from the level of the skin. NIRS data may be further limited by the thickness of the skull or swelling after craniotomy. Regional brain sensor information from NIRS, Licox pBtO2 and cerebral microdialysis (CMD; see below) must be taken with caution since the information they provide is prone to 'sample bias', meaning that the region sampled may provide information useful to only the small region sampled and not inferences from ipsilateral hemispheric or global brain states.
Cerebral NIRS is able to simultaneously monitor oxygenated and deoxygenated hemoglobin based on the strong chromophoric or light-absorbing properties associated with the hemoglobin molecule. This theory was first described by Frans F Jöbsis in 1977, by testing the transmission of near-infrared light (~600–900 nm) through a feline's head. Takuo Aoyagi further established the theory of using spectroscopy to measure oxygen saturation by the isolation of pulsatile changes in light transmission through living tissue as dictated by the shifts in arterial blood volume. This would enable the device to eliminate contamination of skin, bone, tissue and other elements, thus isolating arterial blood flow, and by 1985, the first pulse oximetry device was manufactured. NIRS mimics the technology of pulse oximetry in the sense of using near-infrared light to detect O2 saturation based on total hemoglobin content. A main difference between cerebral NIRS and pulse oximetry on the fingers of ICU patients is that the pulse oximeter is able to isolate the arterial waveform and thus isolate the arterial O2; whereas frontal cerebral NIRS is mixed venous O2 and felt to be a 30:70 ratio of mixed arterial–venous O2. The detection of oxy- and deoxy-hemoglobin is due to the strong chromophoric or light-absorbing properties associated with the hemoglobin molecule. The significant difference between the optical absorbing properties of these two hemoglobin forms allows for its detection by NIRS. Oxygenated hemoglobin has a maximum absorption at approximately 900 nm and deoxygenated hemoglobin at approximately 760 nm. By means of the Lambert–Beer law, the relation of oxygenated versus deoxygenated hemoglobin enables this device to simultaneously monitor the transmittance of light across brain tissue at two or more wavelengths, detecting combined cerebral arterial–venous oxygen saturation levels, which may be used to represent real-time metabolic states via supply and demand physiology.
The ability to understand NIRS and the minimal training necessary to use the device appealing for stroke ICU monitoring. Physicians can establish baseline NIRS values, which are arbitrary units (AU) since they are derived from optical density units and converted mathematically. These baseline values can then be trended visually with changes in relative NIRS values over time. A 20% drop in NIRS O2 values, for example, correlates with a symptomatic carotid-balloon occlusion testing. The interpretation of NIRS is therefore similar to various other modalities for cerebral hemodynamic measurements like SjvO2 due to supply and demand physiologic correlations. Extremes in NIRS values may be useful in indicating changes in hemoglobin, systemic arterial O2 saturation (SPO2; via finger arterial NIRS), cardiac output, DO2, CPP, oxygen extraction fraction and body temperature; through these indications, VSP and subsequent ischemia/infarction may be predicted and treated ( Table 3 ). Further research is needed to compare these numbers against validated measurements like CT perfusion.
The US FDA approved at least two cerebral oximetry devices, the INVOS Somanetics device and CAS Med FORESIGHT. The FDA approved these NIRS devices for adjunct trend monitoring of regional hemoglobin oxygen saturation of blood in the brain or in other tissue beneath the sensor in patients at risk for reduced-flow or no-flow ischemic states. The FDA also cautioned that neither of the NIRS devices should be used as the sole basis for diagnosis or therapy. However, assessment of regional oxygen saturation by NIRS has been seen as valuable in the detection of VSP, and has been referenced to correlation with the various other modalities currently being used to monitor brain-injured patients. NIRS provides the advantage of radiation-free, noninvasive, portable and real time application to cerebral oxygenation parameters, which may indirectly assess CBF due to the supply and demand physiology of cerebral tissue. A small pilot study conducted by Taussky et al. demonstrated the values of NIRS with CT perfusion-derived CBF. Although this research contained a small sample size, data may provide a glimpse that NIRS may have utility in critically ill brain-injured patients too unstable to be transported to the CT perfusion scanner. In addition, NIRS monitoring may be of use in perioperative cardiac surgery monitoring for cerebral desaturations, and may provide more useful information than TCD. In a case study of a 60-year-old Asian female with modified Fisher 4 SAH with right frontal intraparenchymal hemorrhage (IPH), 4 mg of intrathecal nicardipine was administered due to TCD-proven VSP, which resulted in a transient increase in EEG alpha–delta ratio (ADR), CPP and NIRS values (Figure 8). This case study demonstrates that NIRS, as part of multimodal monitoring (MMM) for stroke patients, can potentially help with physiological interpretation and treatment decisions. In this case, the nicardipine appears to cause an upward trend in CPP, improved CBF (via ADR values, which are not shown on this figure) and NIRS cerebral oxygenation in comparison to TCD monitoring. The intraparenchymal hemorrhage on the right caused a baseline asymmetry on NIRS values. The case helps illustrate the ability of NIRS as a safe bedside tool for helping interpret physiological changes with other means of MMM techniques.
(Enlarge Image)
Figure 8.
Near-infrared spectroscopy cerebral oximetry monitoring, after administering intrathecal nicardipine for middle cerebral artery vasospasm, shows upwards trends in cerebral perfusion pressure. AU: Arbitrary unit; CPP: Cerebral perfusion pressure; ICP: Intracranial pressure; IT: Intrathecal. Adapted from [76] with permission from the Mayo Foundation for Medical Education and Research.
The Licox system may be a useful modality in the continuous monitoring of partial oxygenation pressure in brain tissue oxygenation (pBtO2) as well as direct brain tissue temperature (Bt) within the area of the injured brain. The partial pBtO2 is a valuable parameter due to the correlation of CPP as defined by the purpose of direct cerebral oxygenation tension necessary by the variance of autoregulation in the brain-injured patient. Cerebral hypoxia has been seen as a primary indicator of poor outcome in severe traumatic brain injury (TBI); thus, management of cerebral oxygenation is an important factor in the reduction of ischemic events. Invasive pBtO2 monitoring is not indicated in most ischemic stroke patients. However, this type of monitoring has been shown in TBI patients to prevent secondary ischemic damage or brain injury as described below.
The Licox system uses an invasive probe inserted directly into the parenchymal brain tissue using a Clark-type electrode. Data values are quantified as the oxygen molecules diffuse directly though the probe, which are then reduced at the device's electrode to produce a pulsatile electrical current directly proportional to the partial pressure of the oxygen tension as blood passes from arterial to venous circuits. The location of the Licox catheter is determined postscan. Licox probes enable direct pBtO2 measurement with tiny area of brain tissue, which may be useful for regional measurements. However, like that of NIRS and CMD, regional information may not reflect global cerebral hemodynamic states. The Licox system is limited due to the intricacy of the physiological interpretation of this data, which may or may not represent varying degrees of oxygen delivery, extraction and demands among the injured brain, or brain regions outside the scope of the probe.
Recent studies have shown the correlation of pBtO2 by the Licox system to CBF derived by xenon CT, Hemedex derived CBF and SjvO2. The management of CPP has been seen as a key indicator of outcome in patients suffering from TBI, to which the Licox system may be of value. The system is used in CPP/ICP-directed therapy for direct measurement of pBtO2, and has shown increased specificity to SjvO2 and ICP monitoring. Low pBtO2 <20 mmHg has been seen as possibly indicative of secondary ischemia or reduced brain metabolism, and can sometimes be placed before an ICP monitor. ICP is a very useful methodology in monitoring cerebral pathophysiology. However, ICP does have limitations; for example, ICP values may not always correlate with poor CBF or tissue oxygenation deficits. ICP increase may be useful in the prediction of ischemia; however, this data may be misleading. In the case of a febrile patient, an increase in ICP and normal pBtO2 may conclude dysfunction of the cerebral autoregulatory system, but these raised ICP values may not always be indicative of ischemic events. CBF is a primary indicator of the status of cerebral autoregulation; however, this value may be independent of current ICP/CPP modalities. The Licox system is able to measure direct brain tissue pBtO2, and most importantly has shown correlation with CBF-derived modalities. Finally, pBtO2 is not used in the routine ischemic stroke patient, but has literature in SAH and TBI to help detect and manage secondary ischemic insults.
Similar to pBtO2 invasive monitoring, accurate and continuous CBF measurements can be a key factor in the detection and correction of secondary ischemic events. The Hemedex CBF invasive tissue probe may be useful in providing real-time and absolute (ml/100 g/min) CBF values primarily in critically ill SAH patients who are comatose and require invasive ICP monitor placement. ICP and CBF (or another invasive) devices are sometimes placed during the same procedure to provide additional physiological information to manage the patient. Aided by the technology of thermal diffusion, placement of an invasive probe has been approved by the US FDA for up to 10 days of continuous use. The thermal diffusion technology within this modality makes use of two thermistors, one at the tip of the probe and the other 8 mm distal from the tip. The distal thermistor measures baseline brain tissue temperature, while the thermistor at the tip generates a spherical thermal field consisting of a 2°C increase in brain tissue temperature derived from baseline values. The power necessary to generate this increase in temperature is related to the tissue's ability to transfer heat, which is directly proportional to the rate of CBF. Hemedex values are based on the thermal dissipation of the temperature increase, and calculated by two components: the thermal convection component (CBF) and a thermal conduction component (K value). Thermal convection in this device is measured by the rate of thermal diffusion as determined by the rate of CBF. Thermal conduction (K value) is directly proportional to the water content established within the tissue. The Hemedex probe is able to quantify CBF data in absolute values (ml/100 g/min), and graphically viewed by the Bowman Perfusion Monitor.
However, there are serious limitations that must be addressed. An invasive procedure is necessary for the placement of probes 2.0–2.5 cm below the dura mater, which may place the patient at risk of infection, and further complicate treatment of the brain-injured patient. Also, the ability for the Hemedex device to accurately measure CBF is dictated by the thermal stability of the environment during the calibration needed to establish the K value. Incorrect calibration of the K value has been shown to produce artifacts. The pulsatility of CBF has shown to be a factor in the problem of inefficient calibration. The Hemedex system has answered this by the provision of Probe Placement Assistance (PPA), which can help minimize the effects of pulsatility.
Limited research has been conducted on the reliability of this relatively new device. A small study conducted by the University of San Francisco showed the feasibility of using the Hemedex system to accurately assess cerebral autoregulation status in patients suffering from TBI. Other centers like our own use this probe in patients to measure CBF in dysautoregulated comatose SAH patients in which CPP may not behave in a 'plateau' or regulated fashion to prevent secondary ischemia. However, more research is needed before more definitive conclusions can be drawn about this monitoring system and daily management.
A continuous flow of a fluid (perfusate) through a semipermeable probe in the cortical white matter gives a collection of fluid (microdialysate) containing equivalent proportions of specific water-soluble substances compared to the concentration in the extracellular fluid. The microdialysate is dependent on the membrane's pore size, area, rate of flow of perfusate, diffusion speed of the substance and the placement of the CMD. Levels of lactate, pyruvate and, more specifically, the lactate/pyruvate ratio can indicate aerobic/anaerobic glycolysis due to hypoxia and ischemia or energy crisis. Low brain glucose levels compared to the systemic blood glucose may be caused by impaired glucose uptake, mitochondrial dysfunction or hypermetabolism in critically ill SAH patients or TBI. In addition, glutamate and glycerol are biochemical markers of cell damage and cell death. CMD monitoring has therefore been suggested to assist individual clinical evaluation and therapy, such as management of CPP, hyperventilation, blood glucose level management, and surgical/endovascular intervention. However, CMD remains a tool perhaps best fit within the realm of research or at centers that have the capacity to perform 'MMM' in a sophisticated neurointensive care setting. Therefore, more research is need for CMD as a bedside stroke-monitoring application, and its values should be taken into context among the trends of other modalities. In additon, CMD's invasive nature limits the device to the sickest or comatose stroke patients who already require some indication for invasive probe such as ICP monitoring.
Other Protocols
Patients with intracranial hemorrhage have may have seizures or abnormal EEG, which can be associated with worsened neurological outcomes and are not related to underlying structural changes seen on noncontrast head computed tomography (CT), such as periodic lateralized epileptiform discharges. Finally, 19% of medical and neurologically critically ill patients who underwent diagnostic CEEG had abnormal EEG (Figure 4). Of these patients, 95% had NCSs. Therefore, EEG may help guide anticonvulsant prophylaxis or treatment as shown in SAH patients, in whom routine prophylaxis was associated with worse cognitive outcomes.
Table 1
(Enlarge Image)
Figure 4.
Continuous EEG and quantitative EEG example. (A) The regular EEG with suppression-burst pattern in a patient with cardiac arrest and (B) the corresponding image of the quantitative EEG over a longer period of time with discrete 'spikes' on Rosetta seizure probability index (red upper part of figure) and corresponding discrete dark blue areas of Rhythmic Run Detection Display for each hemisphere (middle panels below Rosetta), which correspond to high electrical bursts as seen in regular EEG. The asymmetry index next to the bottom is another quantitative EEG panel that can be useful in stroke if there is hemispheric asymmetry to trend in terms of electrical power. The bottom Q-EEG panel 'aEEG 25–75%' is useful if there is a normal a pattern to trend for differences between hemispheres. In this patient's case the amplitude-integrated EEG is confounded by large electrical bursts shown in the upper panel followed by near isoelectric EEG pattern. Copyright Mayo Foundation for Medical Education and Research.
Transcranial Doppler Monitoring
Transcranial Doppler (TCD) uses sound waves typically around the range of 2–5 MHz through the skull to find intracranial flow of the main circle of Willis arteries such as the first segments of the middle cerebral artery (i.e., MCA-M1 segment), the anterior cerebral artery first segment (ACA-A1) and posterior cerebral artery first segment (PCA-P1). Evaluation of these proximal arterial segments has been well studied in the stroke subtype of SAH in obtaining baseline mean flow velocities (MFVs) and following trends. This is based on the physics of 'flow', which is perhaps best derived conceptually from Ohm's law, V = IR, where V = voltage or pressure difference, I = current or flow, and R is resistance. TCD only measures flow (i.e., the velocity of MCA-M1 blood flow in cm/s). However, as flow changes over time, based on baseline MFV trends, one can infer that resistance might be changing and in the case of VSP (or narrowing of the MCA-M1 due to luminal stenosis) the MFV increases. One must also understand that flow can artifically increase with measures that increase cardiac output to treat VSP, such as use of vasopressors and dopamine. Such increases in systemic cardiac output will increase MCA-M1 MFV as well. Therefore, the Lindegaard ratio (also known as hemispheric ratio) is used to account for hemodynamic augmentation. The Lindegaard ratio is the MCA-M1 MFV divided by the ipsilateral internal carotid artery MFV (i.e., MCA/ICA). Thus, true VSP or luminal narrowing will typically have a Lindegaard ratio of three or greater, whereas hyperdynamic states with increased flow are typically less than three.
TCD VSP sensitivity varies between 50 and 100%, and is vessel-dependent due to location and size, but has a specificity of >90% as compared to the 'gold standard' of digital subtraction angiography (Table 1). TCD is also heavily operator-dependent in terms of education and experience. The ACA and PCA arteries are also more difficult to obtain compared with MCA, no matter which machine is utilized to get a continuous wave (Figure 5B) versus transcranial color Doppler imaging (TCDI) or pulse wave (PW) (Figure 5A), hence sensitivity is reduced for these vessels (Table 2).
(Enlarge Image)
Figure 5.
Sensitivity and specificity of tests for vasospasm. (A) Example of transcranial color Doppler imaging (TCDI) using Sonosite with labeling of the anterior cerebral artery and middle cerebral artery (MCA)-M1 vessels. The hypoechoic structure off to the right of the red and blue color is the midbrain in cross-section and is one of the intracranial landmarks that can be seen with TCDI that is not possible to visualize with routine 'blind' transcranial color Doppler. (B) The 'blind' or non-imaging-based transcranial color Doppler via transtemporal window. The upper part of the image is the depth-based screen with red colors indicating intracranial flow headed toward the probe, whereas blue goes away from the probe and is the anterior cerebral artery. Therefore, the upper flow pattern with the horizontal (yellow) line is the MCA. This can only be known by an experienced examiner who knows the depth of the MCA and other intracranial artery segments, which is also seen on the left around 5 cm from the temporal bone probe location. The advantage of TCDI is you can see the anatomic landmarks and vessels as shown in (A), which is essentially the Circle of Willis. Copyright Mayo Foundation for Medical Education and Research.
There is some concern whether TCDI should not include angle correction, given conflicting studies between TCD continuous wave and pulse wave or TCDI-obtained MFV values, which are perhaps best studied in sickle cell disease. The summary of data in patients with sickle cell disease comparing TCDI and continuous PW TCD has shown that TCDI velocities are lower by approximately 10–20% compared to TCD velocities. A Mayo Clinic Florida institutional review of select patients in which angle correction was carried out and compared against a nonangle correction showed a 10–20% difference, comparable with larger studies. This 'corrected MFV' of approximately 20–30% typically affected the ACAs and PCAs since the MCA had little need for angle correction similar to Krejza. However, TCDI might be able to identify the major cerebral arteries more effectively than TCD. It has also been our experience that TCDI may have difficulty with those with thick temporal bone (Figure 6), approximately >8 mm or a combination of bone thickness and high Hounsfield units (HU; bone density).
(Enlarge Image)
Figure 6.
In patients where thickness of the skull bone is an obstacle to obtain a window for transcranial Doppler pulse waves, near-infrared spectroscopy may be an alternative noninvasive method for continuous screening for early signs of vasospasm. HU: Hounsfield units; ICA: Internal carotid artery; MCA: Middle cerebral artery; NIRS: Near-infrared spectroscopy; TCD: Transcranial Doppler. Copyright Mayo Foundation for Medical Education and Research.
Overall, for SAH VSP stroke detection, the American Academy of Neurology has published guidelines suggesting TCD monitoring of the basal cerebral arteries (MCA-M1, PCA-P1 and ACA-A1). This obviously depends on the skill and resources of each individual institution. For example, some institutions perform this monitoring 7 days a week, 365 days a year, like ours, whereas others perform this monitoring only from Monday to Friday, which can lead to decreased detection of VSP.
Jugular Venous Oxygen Saturation
An established method to measure hemispheric brain mixed arteriovenous oxygenation is via the jugular venous oxygen saturation (SjvO2). Cannulation of the jugular venous bulb allows continuous monitoring of the delivery and consumption of oxygen in the brain. The catheter is equipped with an oximeter at the tip, and accuracy depends on correct positioning. This method requires systemic oxygen saturation, hematocrit and calculations of arteriovenous O2 saturation differences (AVDO2). As 70% of the cerebral venous blood drains via the ipsilateral jugular veins, some clinicians advocate cannulating at the side of injury. In the case of diffuse cerebral injury, most clinicians would monitor the right side, as it is commonly dominant, whereas some clinicians would advocate monitoring the side of dominant flow in all situations. The interpretation of SjvO2 values remains nevertheless complex, with a constant cerebral oxygen consumption, and with interpretation in the context of the hematocrit and SjvO2 variations correlates with CBF variations. The interpretation is complex, based on 'supply' and 'demand' physiology such as cardiac output, hemoglobin, and factors such as cerebral O2 demand, and the impact of oxygen extraction (e.g., fever, diseased or normal brain O2 uptake) (Table 3). Also, interpretation should correct for reduced systemic oxygen saturation and reduced oxygen demand due to hypothermia, sedatives, pathologic arterial–venous communications and brain death, which may result in increased SjvO2. After brain trauma, SjvO2 <50 or >75% is associated with a poor prognosis. To maintain SjvO2 >50% constitutes a reasonable therapeutic goal, but the benefit associated with such a strategy has not yet been validated. The jugular venous SjvO2 method is invasive and there are risks of complications including pneumothorax, arterial puncture, thrombosis and infection.
Near-infrared Spectroscopy
Near-infrared spectroscopy (NIRS) is a relatively new bedside application, which facilitates clinicians with a noninvasive method of assessing regional cerebral arteriovenous oxygenation similar to SjVO2 (Figure 7). The NIRS noninvasive method of measuring cerebral oxygen saturation is of increasing scientific interest, but has a largely undefined role in the daily management of most ICU stroke patients. The high prevelance of VSP after SAH and secondary ischemic brain injury may increase the need for brain monitoring in this patient population. However, current modalities have significant limitations. Modalities such as SjVO2 indicate changes in arterial blood flow without direct brain tissue measurement; other more standardized methods such as PET, MRI with perfusion weighted imaging, CT perfusion and xenon CT offer excellent spatial resolution, but require transportation of critically ill patients and potential exposure to harmful radiation. NIRS offers the advantage of noninvasive, radiation-free, continuous and real-time estimations of brain tissue arteriovenous oxygenation ( Table 4 ).
(Enlarge Image)
Figure 7.
Near-infrared spectroscopy portrays correlation in oxygenation parameters as defined by jugular venous oxygen saturation and pulse oximetry blood sampling. NIRS: Near-infrared spectroscopy; SaO2: Arterial oxygen saturation measured on arterial blood gas; SctO2: Cerebral tissue oxygen saturation; SjvO2: Jugular venous oxygen saturation; SO2: Oxygen saturation. Adapted with permission from [103].
NIRS optodes (two) are placed bifrontal like that of an adhesive bandage separated by an interoptode distance of 4–5 cm. Simultaneous measurement of the two optodes enables left and right frontal hemispheric quantification. Cerebral oximetry measurements by the NIRS are regional and measure approximately 2.5 cm below the NIRS optodes from the level of the skin. NIRS data may be further limited by the thickness of the skull or swelling after craniotomy. Regional brain sensor information from NIRS, Licox pBtO2 and cerebral microdialysis (CMD; see below) must be taken with caution since the information they provide is prone to 'sample bias', meaning that the region sampled may provide information useful to only the small region sampled and not inferences from ipsilateral hemispheric or global brain states.
Cerebral NIRS is able to simultaneously monitor oxygenated and deoxygenated hemoglobin based on the strong chromophoric or light-absorbing properties associated with the hemoglobin molecule. This theory was first described by Frans F Jöbsis in 1977, by testing the transmission of near-infrared light (~600–900 nm) through a feline's head. Takuo Aoyagi further established the theory of using spectroscopy to measure oxygen saturation by the isolation of pulsatile changes in light transmission through living tissue as dictated by the shifts in arterial blood volume. This would enable the device to eliminate contamination of skin, bone, tissue and other elements, thus isolating arterial blood flow, and by 1985, the first pulse oximetry device was manufactured. NIRS mimics the technology of pulse oximetry in the sense of using near-infrared light to detect O2 saturation based on total hemoglobin content. A main difference between cerebral NIRS and pulse oximetry on the fingers of ICU patients is that the pulse oximeter is able to isolate the arterial waveform and thus isolate the arterial O2; whereas frontal cerebral NIRS is mixed venous O2 and felt to be a 30:70 ratio of mixed arterial–venous O2. The detection of oxy- and deoxy-hemoglobin is due to the strong chromophoric or light-absorbing properties associated with the hemoglobin molecule. The significant difference between the optical absorbing properties of these two hemoglobin forms allows for its detection by NIRS. Oxygenated hemoglobin has a maximum absorption at approximately 900 nm and deoxygenated hemoglobin at approximately 760 nm. By means of the Lambert–Beer law, the relation of oxygenated versus deoxygenated hemoglobin enables this device to simultaneously monitor the transmittance of light across brain tissue at two or more wavelengths, detecting combined cerebral arterial–venous oxygen saturation levels, which may be used to represent real-time metabolic states via supply and demand physiology.
The ability to understand NIRS and the minimal training necessary to use the device appealing for stroke ICU monitoring. Physicians can establish baseline NIRS values, which are arbitrary units (AU) since they are derived from optical density units and converted mathematically. These baseline values can then be trended visually with changes in relative NIRS values over time. A 20% drop in NIRS O2 values, for example, correlates with a symptomatic carotid-balloon occlusion testing. The interpretation of NIRS is therefore similar to various other modalities for cerebral hemodynamic measurements like SjvO2 due to supply and demand physiologic correlations. Extremes in NIRS values may be useful in indicating changes in hemoglobin, systemic arterial O2 saturation (SPO2; via finger arterial NIRS), cardiac output, DO2, CPP, oxygen extraction fraction and body temperature; through these indications, VSP and subsequent ischemia/infarction may be predicted and treated ( Table 3 ). Further research is needed to compare these numbers against validated measurements like CT perfusion.
The US FDA approved at least two cerebral oximetry devices, the INVOS Somanetics device and CAS Med FORESIGHT. The FDA approved these NIRS devices for adjunct trend monitoring of regional hemoglobin oxygen saturation of blood in the brain or in other tissue beneath the sensor in patients at risk for reduced-flow or no-flow ischemic states. The FDA also cautioned that neither of the NIRS devices should be used as the sole basis for diagnosis or therapy. However, assessment of regional oxygen saturation by NIRS has been seen as valuable in the detection of VSP, and has been referenced to correlation with the various other modalities currently being used to monitor brain-injured patients. NIRS provides the advantage of radiation-free, noninvasive, portable and real time application to cerebral oxygenation parameters, which may indirectly assess CBF due to the supply and demand physiology of cerebral tissue. A small pilot study conducted by Taussky et al. demonstrated the values of NIRS with CT perfusion-derived CBF. Although this research contained a small sample size, data may provide a glimpse that NIRS may have utility in critically ill brain-injured patients too unstable to be transported to the CT perfusion scanner. In addition, NIRS monitoring may be of use in perioperative cardiac surgery monitoring for cerebral desaturations, and may provide more useful information than TCD. In a case study of a 60-year-old Asian female with modified Fisher 4 SAH with right frontal intraparenchymal hemorrhage (IPH), 4 mg of intrathecal nicardipine was administered due to TCD-proven VSP, which resulted in a transient increase in EEG alpha–delta ratio (ADR), CPP and NIRS values (Figure 8). This case study demonstrates that NIRS, as part of multimodal monitoring (MMM) for stroke patients, can potentially help with physiological interpretation and treatment decisions. In this case, the nicardipine appears to cause an upward trend in CPP, improved CBF (via ADR values, which are not shown on this figure) and NIRS cerebral oxygenation in comparison to TCD monitoring. The intraparenchymal hemorrhage on the right caused a baseline asymmetry on NIRS values. The case helps illustrate the ability of NIRS as a safe bedside tool for helping interpret physiological changes with other means of MMM techniques.
(Enlarge Image)
Figure 8.
Near-infrared spectroscopy cerebral oximetry monitoring, after administering intrathecal nicardipine for middle cerebral artery vasospasm, shows upwards trends in cerebral perfusion pressure. AU: Arbitrary unit; CPP: Cerebral perfusion pressure; ICP: Intracranial pressure; IT: Intrathecal. Adapted from [76] with permission from the Mayo Foundation for Medical Education and Research.
Licox or Brain Tissue Oxygen Monitoring
The Licox system may be a useful modality in the continuous monitoring of partial oxygenation pressure in brain tissue oxygenation (pBtO2) as well as direct brain tissue temperature (Bt) within the area of the injured brain. The partial pBtO2 is a valuable parameter due to the correlation of CPP as defined by the purpose of direct cerebral oxygenation tension necessary by the variance of autoregulation in the brain-injured patient. Cerebral hypoxia has been seen as a primary indicator of poor outcome in severe traumatic brain injury (TBI); thus, management of cerebral oxygenation is an important factor in the reduction of ischemic events. Invasive pBtO2 monitoring is not indicated in most ischemic stroke patients. However, this type of monitoring has been shown in TBI patients to prevent secondary ischemic damage or brain injury as described below.
The Licox system uses an invasive probe inserted directly into the parenchymal brain tissue using a Clark-type electrode. Data values are quantified as the oxygen molecules diffuse directly though the probe, which are then reduced at the device's electrode to produce a pulsatile electrical current directly proportional to the partial pressure of the oxygen tension as blood passes from arterial to venous circuits. The location of the Licox catheter is determined postscan. Licox probes enable direct pBtO2 measurement with tiny area of brain tissue, which may be useful for regional measurements. However, like that of NIRS and CMD, regional information may not reflect global cerebral hemodynamic states. The Licox system is limited due to the intricacy of the physiological interpretation of this data, which may or may not represent varying degrees of oxygen delivery, extraction and demands among the injured brain, or brain regions outside the scope of the probe.
Recent studies have shown the correlation of pBtO2 by the Licox system to CBF derived by xenon CT, Hemedex derived CBF and SjvO2. The management of CPP has been seen as a key indicator of outcome in patients suffering from TBI, to which the Licox system may be of value. The system is used in CPP/ICP-directed therapy for direct measurement of pBtO2, and has shown increased specificity to SjvO2 and ICP monitoring. Low pBtO2 <20 mmHg has been seen as possibly indicative of secondary ischemia or reduced brain metabolism, and can sometimes be placed before an ICP monitor. ICP is a very useful methodology in monitoring cerebral pathophysiology. However, ICP does have limitations; for example, ICP values may not always correlate with poor CBF or tissue oxygenation deficits. ICP increase may be useful in the prediction of ischemia; however, this data may be misleading. In the case of a febrile patient, an increase in ICP and normal pBtO2 may conclude dysfunction of the cerebral autoregulatory system, but these raised ICP values may not always be indicative of ischemic events. CBF is a primary indicator of the status of cerebral autoregulation; however, this value may be independent of current ICP/CPP modalities. The Licox system is able to measure direct brain tissue pBtO2, and most importantly has shown correlation with CBF-derived modalities. Finally, pBtO2 is not used in the routine ischemic stroke patient, but has literature in SAH and TBI to help detect and manage secondary ischemic insults.
CBF Hemedex
Similar to pBtO2 invasive monitoring, accurate and continuous CBF measurements can be a key factor in the detection and correction of secondary ischemic events. The Hemedex CBF invasive tissue probe may be useful in providing real-time and absolute (ml/100 g/min) CBF values primarily in critically ill SAH patients who are comatose and require invasive ICP monitor placement. ICP and CBF (or another invasive) devices are sometimes placed during the same procedure to provide additional physiological information to manage the patient. Aided by the technology of thermal diffusion, placement of an invasive probe has been approved by the US FDA for up to 10 days of continuous use. The thermal diffusion technology within this modality makes use of two thermistors, one at the tip of the probe and the other 8 mm distal from the tip. The distal thermistor measures baseline brain tissue temperature, while the thermistor at the tip generates a spherical thermal field consisting of a 2°C increase in brain tissue temperature derived from baseline values. The power necessary to generate this increase in temperature is related to the tissue's ability to transfer heat, which is directly proportional to the rate of CBF. Hemedex values are based on the thermal dissipation of the temperature increase, and calculated by two components: the thermal convection component (CBF) and a thermal conduction component (K value). Thermal convection in this device is measured by the rate of thermal diffusion as determined by the rate of CBF. Thermal conduction (K value) is directly proportional to the water content established within the tissue. The Hemedex probe is able to quantify CBF data in absolute values (ml/100 g/min), and graphically viewed by the Bowman Perfusion Monitor.
However, there are serious limitations that must be addressed. An invasive procedure is necessary for the placement of probes 2.0–2.5 cm below the dura mater, which may place the patient at risk of infection, and further complicate treatment of the brain-injured patient. Also, the ability for the Hemedex device to accurately measure CBF is dictated by the thermal stability of the environment during the calibration needed to establish the K value. Incorrect calibration of the K value has been shown to produce artifacts. The pulsatility of CBF has shown to be a factor in the problem of inefficient calibration. The Hemedex system has answered this by the provision of Probe Placement Assistance (PPA), which can help minimize the effects of pulsatility.
Limited research has been conducted on the reliability of this relatively new device. A small study conducted by the University of San Francisco showed the feasibility of using the Hemedex system to accurately assess cerebral autoregulation status in patients suffering from TBI. Other centers like our own use this probe in patients to measure CBF in dysautoregulated comatose SAH patients in which CPP may not behave in a 'plateau' or regulated fashion to prevent secondary ischemia. However, more research is needed before more definitive conclusions can be drawn about this monitoring system and daily management.
Cerebral Microdialysis
A continuous flow of a fluid (perfusate) through a semipermeable probe in the cortical white matter gives a collection of fluid (microdialysate) containing equivalent proportions of specific water-soluble substances compared to the concentration in the extracellular fluid. The microdialysate is dependent on the membrane's pore size, area, rate of flow of perfusate, diffusion speed of the substance and the placement of the CMD. Levels of lactate, pyruvate and, more specifically, the lactate/pyruvate ratio can indicate aerobic/anaerobic glycolysis due to hypoxia and ischemia or energy crisis. Low brain glucose levels compared to the systemic blood glucose may be caused by impaired glucose uptake, mitochondrial dysfunction or hypermetabolism in critically ill SAH patients or TBI. In addition, glutamate and glycerol are biochemical markers of cell damage and cell death. CMD monitoring has therefore been suggested to assist individual clinical evaluation and therapy, such as management of CPP, hyperventilation, blood glucose level management, and surgical/endovascular intervention. However, CMD remains a tool perhaps best fit within the realm of research or at centers that have the capacity to perform 'MMM' in a sophisticated neurointensive care setting. Therefore, more research is need for CMD as a bedside stroke-monitoring application, and its values should be taken into context among the trends of other modalities. In additon, CMD's invasive nature limits the device to the sickest or comatose stroke patients who already require some indication for invasive probe such as ICP monitoring.