16 Neurological assessment of the acute cardiac care patient

Many techniques are currently available for cerebral physiological monitoring in the intensive cardiac care unit environment. The ultimate goal of cerebral monitoring applied during the acute care of any patient with/or at risk of a neurological insult is the early detection of regional or global hypoxic/ischaemic cerebral insults. In the most ideal situation, cerebral monitoring should enable the detection of any deterioration before irreversible brain damage occurs or should at least enable the preservation of current brain function (such as in comatose patients after cardiac arrest). Most of the information that affects bedside care of patients with acute neurologic disturbances is now derived from clinical examination and from knowledge of the pathophysiological changes in cerebral perfusion, cerebral oxygenation, and cerebral function. Online monitoring of these changes can be realized by many non-invasive techniques, without neglecting clinical examination and basic physiological variables—with possible impact on optimal cerebral perfusion/oxygenation—such as invasive arterial blood pressure monitoring or arterial blood gas analysis.

Neurological monitoring in the acute care setting should offer an essential window to the patient’s brain. Although the current cerebral monitoring techniques (invasive as well as non-invasive) offer essential and readily interpretable information on a patient’s neurological status, the knowledge on cerebral physiology and pathophysiology remains a keystone to optimally interpret and manage a patient’s neurological condition.

Reasons for cerebral monitoring in at-risk acute cardiac care patients include: (1) the detection of early neurological dysfunction before irreversible brain damage occurs (i.e. in stroke patients or patients suffering from an intracerebral bleeding), (2) to individualize patient care decisions and to guide patient management, (3) to improve neurological outcome in survivors of any acute neurological injury such as patients presenting after cardiac arrest, and finally (4) to allow the prognosis of neurological outcome, especially in the comatose patient after cardiac arrest.

In this chapter, a number of currently available bedside cerebral monitoring techniques will be discussed. Insights into cerebral physiology and pathophysiology are added in order to optimally interpret the obtained information. Implications into patient management (especially concerning the neurological prognosis after cardiac arrest) will be discussed.

The clinical neurological examination assesses arousability, level of consciousness, and, if awake, the eventual presence of specific (focal) deficits. The examination remains most important to address aetiological questions and management. However, in the intensive cardiac care unit (ICCU) environment, this clinical examination may become a challenging task as: (1) patients may have received sedating or paralysing drugs, (2) drug pharmacokinetics may be altered by major organ dysfunction (liver, kidney) or by the use of induced hypothermia, (3) patients may be intubated, and (4) more than one disease process may be ongoing; if multiple organ failure has occurred, this can contribute to the impairment of consciousness (in particular, in the presence of septic encephalopathy).

In general, neurological examination assesses the level of consciousness, the functioning of the cranial nerves, and the patient’s motor function [1]. Arousability is assessed by stimulating the patient by speech. It is better to make a clear assessment of the patient’s activity and responsiveness than to use non-specific terms such as semi-coma. Even within the context of coma (unarousable, unconscious), there may be different types of responses (no response, posturing, purposeful movements, or abnormal posturing) [2].

Coma may obscure the evaluation of all nerves, except cranial nerves II, III, and IV. Direct and consensual pupillary light reflexes are easily assessed with the eyes passively held open, looking for constriction of the ipsilateral and contralateral pupils. With damage to cranial nerve III, the ipsilateral pupil is typically dilated and unreactive; the eye is turned outwards, and there is complete, or nearly complete, ipsilateral ptosis. The presence of a normal pupillary light reflex indicates that its afferent and efferent limbs are intact. An early manifestation of brain herniation is the loss of pupillary reactivity, usually on the side of the mass effect.

Spontaneous movements in the ICCU patient include posturing, purposeful movements, withdrawal movements, seizures, or myoclonus. A ‘purposeful’ motor response is one that avoids the noxious stimulus or attempts to push it away. Decerebrate or decorticate posturing indicates severe bilateral dysfunction, deep in both hemispheres or in the brainstem.

Coma scales and scoring systems were developed for the objective quantification of the severity of acute neurological dysfunction and for the prediction of outcome [3, 4]. The Glasgow coma scale (GCS) was introduced in 1974, aimed at standardizing the assessment of level of consciousness, namely in head-injured patients. It has been used mainly in evaluating prognosis, comparing different groups of patients, and monitoring the neurological status. Today, the GCS is the most universal scale used in emergency rooms and ICCUs. It gives a maximum of 15 points, 6 maximum points for best motor response, 5 points for the best verbal response, and 4 points for eye opening. A theoretical disadvantage is the fact that the total Glasgow coma scale (GCS) score is obtained by adding the values for three motor activities, namely eye opening, best motor response, and best verbal response. These are assumed to be independent variables; however, they are not. The key concept in all literature data is that, although GCS is not a perfect tool and other coma scales have been proposed, it will remain incorporated in clinical decision making regarding coma for many years to come. It should be the standard coma scale applied to all acute cardiac care patients in need of any cerebral monitoring [5, 6].

Although clinical neurological examination (including GCS assessment) remains the ‘gold standard’ for cerebral monitoring in the ICCU environment, by the time any deterioration has been noted, brain injury, that may or may not be reversible, may have already occurred. In addition, the management of any kind of acute neurological failure may urge for sedation (and eventual paralysis) of the patient, seriously disturbing the reliability of bedside clinical examination as a screening tool. Ideally, cerebral physiological monitoring should provide a continuous bedside assessment to monitor various disease processes and to detect abnormalities before irreversible brain damage has occurred, and it should be helpful in assessing therapeutic interventions known to influence outcome [7, 8, 9]. Over the last years, critical care monitoring (and management) of patients with acute severe brain injury (of any origin) has advanced tremendously. Today’s armamentarium consists of invasive and non-invasive monitoring devices to detect secondary brain injury, to guide therapy, and to provide prognostic indications. In general, cerebral monitoring techniques applied in the ICCU environment can be categorized into those measuring cerebral perfusion (or cerebral oxygenation), those measuring cerebral function, those which focus on biochemical markers (by the use of biomarkers), and finally, brain imaging.

The adequacy of cerebral perfusion is usually monitored as a derived value, rather than as a directly measured parameter, because of the complexity of cerebral blood flow (CBF) measurements. In order to optimally interpret any monitoring parameter on cerebral perfusion, a basic knowledge on the physiological regulation of cerebral perfusion is required [10, 11].

CBF is determined by cerebral perfusion to resistance:

(where CPP, cerebral perfusion pressure, is the mathematical difference between MAP and intracranial pressure (ICP); CVR, cerebral vascular resistance)

In human, cerebral autoregulation adjusts CVR automatically and continuously such that global CBF remains constant over a wide range of MAP, from 50 to 150 mmHg. In the presence of any acute brain injury, such as intracranial bleeding, stroke, or post-resuscitation status, however, autoregulation may be regionally or globally impaired, resulting in a pressure-passive CBF regulation and implying the need for meticulous control of MAP. In these conditions, every decrease (or increase) in MAP will induce a reciprocal decrease (or increase) in CBF, thereby reaching critical thresholds (for cerebral ischaemia in cases of severe arterial hypotension, or for severe cerebral hyperaemia with the ensuing risk of blood–brain barrier disruption in cases of severe arterial hypertension). This rigorous control of MAP is best accomplished using an intra-arterial catheter for invasive beat-to-beat monitoring of blood pressure. A CPP threshold of 60–65 mmHg is the minimal goal of blood pressure management. In most ICCU patients, ICP monitoring—as such—is however not applicable due to an increased risk of intracerebral bleeding in the presence of anticoagulant therapy, and at the best, an estimated ICP value can be used. In cases of an estimated normal ICP (10–15 mmHg), this implies an MAP threshold of 70–80 mmHg, but, in cases of an estimated increased ICP, MAP values of at least 85–90 mmHg should be aimed for [12, 13]. Calibration of the arterial pressure transducer to zero should ideally be done at the level of the external auditory meatus to take into account the differences in perfusion pressure between the level of the head and the thorax.

In the absence of critical vascular stenosis, CVR is determined by cerebral autoregulation, the level of neural activity, blood viscosity, and arterial CO₂ (PaCO₂) and arterial O₂ tensions (PaO₂). Acute incremental changes in PaCO₂ cause corresponding directional changes in CBF, in the order of a 4% change in CBF per 1 mmHg change in PaCO₂. If the intracranial compliance is reduced, as in the presence of an intracerebral bleed, any increase in CBF may be accompanied by a rise in cerebral blood volume (CBV) and ICP. Conversely, alterations in PaO₂ cause opposite changes in CBF. CBF increases abruptly as PaO₂ falls below 40 mmHg, whereas a rise in PaO₂ from the normoxic to the hyperoxic range results in a 15% decrease in CBF. This is offset by a corresponding increase in arterial O₂ content. Thus, monitoring of ABG tensions enhances the interpretation of changes in other parameters (e.g. ICP). This is accomplished either intermittently by sampling of arterial blood or continuously by measurement of transcutaneous peripheral oxygenation (i.e. pulse oximetry) or end-tidal CO₂ analysis (i.e. capnography).

In the past, techniques for accurate quantitative measurements of regional CBF were evaluated at the bedside in the ICCU (e.g. based on the clearance of freely diffusible, poorly soluble, inert radioisotope 133-xenon from the brain tissue). However, technical issues made it difficult to include this monitoring technique into the standard monitoring care of neuro-ICCU patients. Moreover, few data are available to correlate CBF with clinical outcome in patients suffering from critical neurological illness. Clinical interpretation of CBF data is further hindered by the complex interaction between CBF, CBV and ICP, intracranial compliance, and functional neuronal integrity.

A continuous, non-invasive, qualitative assessment of cerebral perfusion is available in the form of transcranial Doppler (TCD) imaging [14, 15, 16, 17]. This device utilizes a 2 MHz pulsed Doppler ultrasound, directed through the thin regions of the skull (e.g. the temporal area), to measure flow velocity (FV), rather than flow volume, in the major intracranial arteries (mostly in the middle cerebral artery). Although the absolute measurements of FV correlate poorly with 133-xenon CBF measurements, FV measurements give an accurate indication of the changes in CBF in individual patients. FV (cm/s) varies directly with the absolute level of CBF and inversely with the square of the vessel diameter. In patients with subarachnoid haemorrhage (SAH) and cerebrovascular spasm, it was shown that increases in FV precede the clinical signs of cerebral ischaemia. If TCD monitoring is performed on a frequent basis, the severity and time course of arterial spasm can be monitored non-invasively, thus facilitating optimal timing for interventions, such as cerebral aneurysm surgery (best results if no cerebral vasospasm is already present), or for the evaluation of medical treatment, while obviating the need for repeat angiography. In closed head injury with raised ICP, TCD may aid to differentiate patients with low CBF from those with high CBF and vasoparalysis (i.e. loss of autoregulation). A patient with high ICP and low CBF may suffer from further ischaemia due to excessive hyperventilation (as a reduction in PaCO₂ induces an intense cerebral vasoconstriction). Nevertheless, hyperventilation was a therapy widely and empirically used in the past to reduce CBF (and ICP), but the application of bedside monitoring of cerebral perfusion completely changed this approach into the maintenance of strict normoventilation (with normal PaCO₂ values).

Cerebral O₂ consumption (CMRO₂) is calculated from the product of mean global CBF and the cerebral AV O₂ content difference (a-vDO₂, normal 6 mL/100 g/min). Cerebral venous effluent is sampled from a catheter inserted percutaneously into the internal jugular vein and advanced in a retrograde fashion into the jugular bulb. Using this method, the value generated for a-vDO₂ is not global, because jugular venous O₂ tension is measured from a unilateral sample, and moreover there is considerable mixing of the venous effluent from both hemispheres. Two thirds of the blood supplied to one hemisphere through an internal carotid artery is drained through the ipsilateral jugular vein, whereas one-third drains contralaterally. Extracerebral contamination of blood in the jugular bulb accounts for around 3%.

Jugular bulb oximetry is an invasive technique which consists of the insertion of a fibreoptic catheter retrogradely into the internal jugular vein, until its tip reaches the jugular bulb. It offers a continuous estimation of the adequacy of cerebral perfusion. Normal jugular bulb saturation values (SjO₂) are between 50% and 80%. Especially too low SjO₂ values (i.e. <50%) indicate the presence of inadequate cerebral perfusion. Despite the relative ease of the technique and the clinical value of the obtained information, jugular bulb oximetry is currently used only in the setting of severe head injury [18, 19].

Near-infrared spectroscopy (NIRS) is a non-invasive method to monitor cerebral tissue O₂ saturation [20, 21, 22]. It uses up to five wavelengths of LED or laser light to distinguish oxygenated from deoxygenated haemoglobin, thereby providing the respective O₂ saturation of the cerebral tissue. As the NIRS sensor is applied over the patient’s forehead, it measures the cerebral tissue O₂ saturation at about 1.5 cm depth in the frontal cerebral lobe. This cerebral tissue O₂ saturation reflects arterial as well as venous (as capillary) components. Therefore, a straight comparison with SjO₂ values remains difficult. Nevertheless, validation studies of NIRS against SjO₂ values confirmed the obtained NIRS readings. Recent studies revealed that this non-invasive monitoring technique becomes of interesting value in detecting, for example, inadequate cerebral perfusion in the acute care patient (post-resuscitation status) [23, 24, 25].

As already mentioned, bedside neurological examination (like the GCS) is widely used as a clinical measure of cerebral function in the ICCU environment. Any deterioration in score prompts further investigation or therapy. However, clinical deterioration may not be seen until significant structural damage has occurred. Electrophysiological monitoring seeks to address these deficiencies. In clinical practice, it consists of recordings of the EEG, some form of computer processing of the raw EEG signal, and evoked potential (EP) monitoring.

The standard EEG is the summation of electrical activity generated in the pyramidal cells of the cerebral cortex. The EEG signal recorded at the scalp represents the activity of cortical cells to a depth of 1–2 cm only, because the high electrical impedance in the intervening bone and soft tissues attenuates the energy emanating from deeper brain structures. Typically, 16 leads are used, permitting considerable regional discrimination. These standardized leads are placed according to guidelines known as the International 10–20 System. Amplitude is measured from peak to peak and expressed in microvolts, normally ranging from 10 to 200 microvolts, with peak voltages of approximately 100 microvolts during seizure activity. The frequency spectrum of the EEG is categorized as follows: delta <4 Hz; theta 4–8 Hz; alpha 8–13 Hz; beta >13 Hz. In awake, relaxed adults, with eyes closed, the alpha rhythm predominates. Beta rhythm supervenes when the subject is stimulated (e.g. when the eyes are opened). Theta and delta waves are seen during normal sleep or sedation. Normally, some theta activity is detectable in awake patients, but persistent delta activity is considered abnormal. Visual, online interpretation of the raw EEG is confounded by many sources of electrical artefacts, including body movements, ECG monitors, and nearby electrical devices.

In the ICCU setting, a premorbid tracing is seldom available for comparison, but cortical dysfunction is generally associated with slowing of EEG frequency and a decrease in amplitude of the recorded signal, similar to that produced by cerebral ischaemia. Comparisons between the two hemispheres are usually helpful in clinical interpretation, particularly in those patients with focal cerebral lesions.

Cerebral dysfunctions, accompanied by disturbed consciousness, will result in some specific slowing of EEG frequency and diminution of EEG amplitude, but these changes in EEG cannot give any differentiation in the exact aetiology of the respective cerebral dysfunction (e.g. metabolic, anoxic, toxic, infectious, or degenerative origin). Although no particular EEG pattern is related to a specific aetiology or prognosis, EEG monitoring can be used in comatose patients to assess the depth of coma and the reactivity to stimulation. Observation of rapid eye movement (REM) sleep and spontaneous variability in patterns of EEG activity over extended periods of time are associated with improved outcome in comatose patients, compared with those with a fixed pattern. When performing EEG recordings over time, an improved EEG (higher EEG frequency and increased amplitude) can be one of the first signs pointing to a better neurological outcome.

Additionally, in paralysed patients, EEG monitoring allows the detection of seizures, prompting the institution of anticonvulsant therapy. It is noteworthy that recognition of seizure patterns is still hampered by computer processing of the EEG, emphasizing the need for immediate access to the raw data. In case of barbiturate therapy for intracranial hypertension, the drug infusion rate of barbiturates can be titrated, according to the EEG signal, to produce a burst suppression pattern, rather than a full isoelectric signal, thus allowing maximum potential benefit without an additional risk from drug overdose.

Recently, EEG interpretation was simplified by computer processing of the raw signal, which reduces the amount of data to be examined and facilitates the interpretation by relatively unexperienced personnel. Several forms of EEG processing have been used in critical care practice, including the cerebral function monitor (CFM), compressed spectral array (CSA), and density spectral array (DSA).

The latest computerized processed analysis of the raw EEG consists of a bispectral analysis of the EEG, offering a bispectral index (BIS) with a numerical value between 0 and 100. Literature has revealed the unique value of BIS to assess the depth of hypnosis during surgical anaesthesia and especially to prevent intraoperative awareness. Likewise, BIS was used in the ICCU setting to assess the depth of sedation, revealing satisfactory correlation to all clinical scoring systems (used to assess the depth of ICU sedation, e.g. Ramsay sedation score, agitation sedation scale). Nevertheless, the exact value of BIS monitoring in patients with acute neurological disturbances is not yet established, although its possible prognostic value in comatose post-cardiac arrest patients is currently extensively studied.

Future evolutions in EEG monitoring in the ICCU setting will surely focus on the application of continuous EEG monitoring (cEEG). cEEG monitoring offers many advantages, compared to intermittent EEG monitoring. Its main advantage is the possibility to detect non-clinical seizures, which are frequently reported in any critical care patient. The introduction of future computerized algorithms for cEEG monitoring may render this monitoring even more feasible, as it should allow readily the interpretation of any epileptic activity present in the raw EEG tracing [26, 27, 28, 29, 30].

The major value of EP monitoring is that these signals are resistant to alterations by pharmacological agents such as barbiturates; therefore, changes can be attributed correctly to a genuine functional derangement. In addition, under conditions of cerebral hypoxia, the evoked responses are a more sensitive marker of irreversible hypoxic cellular damage than the EEG. Importantly, EP responses can be ascribed to specific anatomic structures not readily amenable to clinical assessment. The measurement and recording of EPs require computer-assisted analysis, because the amplitude of the EP signal is so low (0.12 microvolts) that it cannot be differentiated from the background EEG activity. Computer analysis is facilitated by the fact that EP signals occur at a predictable interval after a standardized stimulus, so averaging of many responses can be performed. The information derived from the interpretation of an EP response includes the post-stimulus latency (in ms) and the peak amplitude (in microvolts) of the various waveforms in the tracing. Abnormalities are classified as an absence of certain waveforms, prolonged latency, or reduced amplitude. Several neural pathways lend themselves to EP monitoring, but the three most commonly used in clinical practice are: (1) somatosensory evoked potentials (SSEPs), (2) brainstem auditory evoked responses (BAERs), and (3) visual evoked potentials (VEPs).

The stimulus for SSEP monitoring is supplied by the electrical stimulation of nerves in any of the four limbs. Because SSEPs monitor supratentorial elements that are affected early by a shift of intracranial structures, they can be used diagnostically to detect the expansion of space-occupying lesions. Prognostic information based on SSEP is extremely accurate in that the absence of certain waveform components (bilaterally) in comatose patients is uniformly predictive of a persistent vegetative state or death.

Future developments focus on using a variety of electrophysiological monitoring tools to assess patients clinically and to formulate a prognosis. This concept has been termed multimodality evoked motor potential (MEP) monitoring, and it comprises a combination of SSEPs, VEPs, and BAERs. In patients with severe head injury, MEP measurements are highly predictive of outcome and improve the prognostic reliability of information derived from clinical examination [31, 32].

Biomarkers are quantifiable biological markers that can be collected and measured in different fluid compartments. The most convenient markers are those that can be readily obtained at the bedside. Several studies have measured markers of brain injury in serum and cerebrospinal fluid (CSF) [33, 34, 35].

NSE is a glycolytic enzyme found mainly in neurons and neuroectodermal cells. High serum levels have been reported with malignant tumours such as neuroblastomas and small cell carcinoma of the lung. Studies have shown that NSE is a marker for severity of neuronal cell injury and predicts clinical outcomes in stroke, head injury, encephalitis, brain metastasis, and status epilepticus.

The prognostic role of NSE as a biochemical marker of neuronal injury in cardiac arrest was first investigated in the 1980s. Several authors have since measured NSE in serum or CSF to help prognosticate more accurately the neurological outcome following cardiac arrest. This will be more extensively discussed in the last part of this chapter, which focuses on neurological monitoring and prognosis after cardiac arrest.

S-100 beta (S-100B), a member of the calcium-modulated protein family, has both intracellular and extracellular regulatory activities. It is expressed in varying abundance in astrocytes, Schwann cells, adipocytes, melanocytes, chondrocytes, Langerhans skin cells, lymphocyte subpopulations, skeletal muscle cells, and many neuronal populations. It modulates the differentiation and proliferation of neurons and glia. Earlier studies demonstrated a significant correlation between S-100B serum levels, anoxia time, and the degree of coma at days 1 and 2 after cardiac arrest.

Long-term outcomes related to S-100B and NSE levels in cardiac arrest have also been evaluated, which revealed that high levels of serum S-100B predicted poor outcome, according to GCS measured 1 year following cardiac arrest. NSE, on the contrary, does not have this predictive power.

Very few studies actually compared NSE and S-100B head to head. Up till now, the exact role of biomarkers in the whole panel of cerebral monitoring is still explored. They can, however, play an important future role in the multimodal monitoring setting of prognostication after cardiac arrest.

Neuroimaging provides a non-invasive, reliable method of assessing structural brain injury. Grey matter, in particular, the hippocampus, caudate putamen, thalamus, large cell layers of the neocortex, and Purkinje cells of the cerebellar cortex, are more vulnerable to hypoxic injury. White matter injury is believed to occur weeks to months after global hypoxic injury. The imaging findings of diffuse anoxia/hypoxia include diffuse cerebral oedema, loss of grey–white matter distinction, and selective neuronal necrosis affecting the deep grey nuclei and the cortical layers III, IV, and V. CT and conventional MRI could play a role in prognostication following diffuse cerebral anoxia [36, 37, 38].

CT scanning is one of the most frequently used tests in comatose patients. It allows the immediate detection of the presence of a localized intracerebral bleed or SAH. A normal CT scan of the brain shows a clear difference between white matter, with its high lipid content, and grey matter with its high water content. After global cerebral anoxia (e.g. following cardiac arrest), the inadequate production of adenosine triphosphate that accompanies global ischaemia results in an overall increase in water content (cytotoxic oedema). Furthermore, a delayed hyperaemia after resuscitation can lead to increased ICP and occasionally to brain swelling. Initially, the cerebral blood vessels collapse, resulting in a decreased intracranial volume and preventing a further increase in ICP. If systemic hypotension is corrected, this results in distension of the deep medullary veins. As a result, white matter becomes distended with blood and appears denser on unenhanced CT scans. Therefore, a loss of distinction between grey and white matter after cardiac arrest could result from a combination of decreased grey matter intensity, due to cytotoxic oedema, and increased white matter intensity, due to distension of the medullary draining veins.

It has been a general impression that a reduced distinction between grey and white matter on CT predicts poor outcome after cerebral insults. However, this qualitative assessment is not very reliable. Additional studies confirmed that CT scans could be used as a reliable tool to evaluate long-term brain injury after cardiac arrest. CT imaging revealed atrophy in quantitative and qualitative analyses in patients between 1 and 3 years after cardiac arrest, which correlates with impairment in cognitive testing.

MRI during global cerebral hypoxia is possibly a valuable tool for early prognosis of clinical outcome. Although there is no doubt of the accuracy of MRI in identifying early ischaemic changes, the use of this technique is still limited by several factors such as the hostile environment of the MRI setting.

Although earlier studies focused primarily on grey matter injury following cerebral ischaemia, DWI also opened the door to the evaluation of white matter injury. The results of these studies were contradictory to the common belief that white matter injury usually occurs later in the course following cardiac arrest. Investigators were able to find white matter changes in the subacute period of 14–20 days following injury. These studies demonstrated that myelinopathy can occur quite early following anoxic injury and holds potential for evaluation as a prognostic indicator. The fact that this leukoencephalopathy occurred earlier than would be expected for a Wallerian degeneration suggests that it may be a result of the primary insult caused by diffuse cerebral anoxia, impaired cerebral perfusion, or both. As a result, an early axonal injury is produced independently of neuronal injury.

It can be concluded that MRI (and DWI) imaging may be a useful tool to assess prognosis in a multimodal setting in comatose patients after cardiac arrest.

Table 16.1 shows an overview of all current cerebral monitoring techniques and their most relevant and different characteristics.

Monitoring the brain function in the acute cardiac care patient is mainly applied to comatose patients admitted after cardiac arrest. The degree of hypoxic-ischemic brain damage, mainly caused by the insufficient flow during cardiac arrest and CPR and the ensuing global reperfusion injury after ROSC, is widely considered as the determining factor of survival in these patients. Over the past decade, outcome substantially ameliorated due to improvements in advanced life-support, emergency coronary interventions, the implementation of targeted temperature management (TTM) and intensive care management with hemodynamic targets aimed at the optimization of (extra-) cerebral perfusion. Despite these improvements in the post-resuscitation care, about 50% of the patients successfully resuscitated and admitted to the ICCU still decease or end up with a poor neurological prognosis. Following cardiac arrest, it is of utmost importance to distinguish as early as possible between delayed awakening or prolonged coma and irreversible neurologic injury. Recently, a sequential algorithm on neuroprognostication after cardiac arrest has been adopted by international guidelines including four main modalities which - whenever possible - should be used in conjunction with each other, i.e. clinical examination, electrophysiology, blood biomarkers and brain imaging. These guidelines highly recommended to avoid any decisive evaluation of prognosis before at least 72 hours after the return to normothermia [39, 40, 41, 42, 43, 44].

Clinical neurological examination is still the gold standard to assist with neuroprognostication although careful interpretation is paramount, because many patients are affected by sedatives and therapeutic hypothermia. A clinical examination includes the assessment of motor responses to pain and brainstem reflexes, in particular corneal and pupillary light responses. Before the advent of TTM, the absence of a motor response or a presence of an extensor response to pain on day 3 after cardiac arrest was considered as a reliable predictor of poor outcome after cardiac arrest (referring to the 2006 AAN guidelines). In patients undergoing TTM however, motor responses have a reduced accuracy in poor outcome prognostication due to biasing effects of sedatives and neuromuscular blockers (reported FPR around 20%). As such, we recommend not to rely on motor reactions alone after TTM and to integrate full clinical neurological examination in a multimodal assessment. In contrast to motor responses, the bilateral loss of corneal and pupillary light responses remains a robust indicator of poor outcome with the lowest FPR reported at 72 hours after cardiac arrest, independent of the application of TTM. However, the reactivity of constricted pupils, as well as the corneal reflex, may be difficult to assess, and it is vital that the effects of lingering sedation are excluded. Nowadays, automated infrared pupillometry is being introduced as a promising alternative to quantify the pupillary light response. Besides, it has an outstanding accuracy in predicting poor neurological outcome, even outperforming the one of standard clinical examination [45, 46].

Electroencephalographic signals directly reflect synaptic activity in cortical neurons and therefore provide valuable insights into the degree of neuronal injury in the post-ischemic brain of cardiac arrest patients. The interpretation of a routine EEG should cover three main aspects: (a) background activity, (b) EEG reactivity and (c) the presence of epileptiform features or status epilepticus. These features should ideally be assessed in both the early and late stage following cardiac arrest, implying that EEG recordings should be initiated as soon as possible after hospital admission. In the context of neuroprognostication, studies demonstrated that the highest predictive power of EEG lies within 12 to 24 hours after the return of spontaneous circulation. Multiple EEG background patterns can be classified as malign and, at least without timely improvement, are strongly indicative for a lack of neurological recovery, i.e. the presence of low-voltage (< 20µV) or suppressed background at 24 hours; burst suppression patterns with generalized epileptiform discharges and the appearance of a suppression-burst pattern with identical bursts. On the contrary, a normal voltage EEG background with continuous activity at 12 hours and to a lesser extent 24 hours following cardiac arrest is associated with a high likelihood to attain a good neurological outcome. Recent data indicate that absent EEG background reactivity, assessed after TTM, is associated with poor outcome after cardiac arrest. Unreactive EEG background was confirmed as a strong predictor of poor outcome by a large prospective study on 111 patients treated with TTM at 33°C (FPR 7%). This correlation even improved when EEG reactivity was absent during TTM (FPR 2%).

Status epilepticus (defined as prolonged periodic or rhythmic epileptiform discharges) after cardiac arrest is observed in about 30% of patients and is an independent predictor of poor outcome, suggesting that it may per se aggravate prognosis. However, epileptiform EEG patterns are not obligate predictors of dismal prognosis (FPR 9%). Post-anoxic status epilepticus, if diagnosed and treated early, can be compatible with good recovery, at least in a subset of patients. The identification of favourable prognostic markers is essential to guide therapy; these include preserved brainstem reflexes and EEG background reactivity, together with the presence of cortical SSEPs.

One of the limitations of the EEG is that a single conventional EEG is only a snapshot of what is occurring in the brain during a limited amount of time (usually 20–30 min). The obvious application for continuous EEG monitoring is in seizure detection in the sedated, paralysed patient in whom clinical seizures may be concealed. Moreover, relevant monitoring of anticonvulsive treatment mandates cEEG monitoring. Instead of a multichannel EEG, a simplified EEG montage, with a limited number of raw EEG curves, may be used, and, because cardiac arrest leads to a global hypoxic, ischaemic injury, two hemispheric leads should be sufficient to give relevant bedside information on the recovery or deterioration of cortical activity. Continuous processed analysis of the raw EEG into a simple number (such as the BIS) might offer the minimal necessary information, which is currently under investigation.

As long as only outcome prognostication is the main concern, standard EEG recording should ideally be performed in the first 24 hours following hospital admission (if possible, for at least 30 min) and the assessment of EEG reactivity at the bedside is paramount. cEEG may be preferable if treatment of seizures is initiated to monitor the effect of antiepileptic drugs and the eventual need for further drug titration [43, 44, 45].

It has been repeatedly shown that bilaterally absent cortical responses (N20) on SSEP are almost 100% predictive of poor outcome when performed at 48–72 hours after the arrest. Neuroprognostication data derived from the TTM study confirmed that bilaterally absent SSEPs maintained their strong prognostic accuracy for poor outcome after cardiac arrest and reported a FPR of only 2.6%. As such, it remains possible to regain consciousness despite the bilateral absence of cortical N20 responses, at least in exceptional cases. Hence, SSEPs should not be used alone but should be incorporated in a multimodal algorithm. The value of SSEP (when N20 is present) in predicting good outcome on the other hand is rather limited and seems lower than that of EEG reactivity [53, 54, 55].

Biomarkers are quantifiable biologic markers that can be collected and measured in different fluid compartments, most often blood. The best studied biomarkers after cardiac arrest are NSE and S-100B, of which NSE has been incorporated in the 2006 AAN guidelines stating that a peak NSE >33 mg/L at 1–3 days after cardiac arrest was predictive of poor outcome, with an FPR of 0%. However, these guidelines were based on studies performed prior to the implementation of TTM. Although elevated NSE 1–3 days after cardiac arrest remain a reliable marker of the severity of post-anoxic injury, the validity of the previously suggested NSE threshold of 33µg/L has been questioned by studies in patients undergoing TTM. In a substudy of the TTM trial, thresholds between 30 and 50 ng/ml at 48–72 hours after the arrest were strongly correlated with a poor prognosis with tight false positive ratios (FPR<3%) and there was no impact of temperature on NSE within the range of 33–36°C. Besides, serial NSE sampling between 24 and 72 hours seemed to improve the prognostic accuracy as compared to a single measurement. One of the major limitations regarding the use of NSE as prognostic marker is the large variability in laboratory assays, thereby stressing the need for new and standardized measuring techniques for the accurate determination of NSE. Moreover, NSE is found in platelets and RBCs; thus, blood samples need to be carefully handled in order to avoid haemolysis and false positive test results. To summarize, whether a cut-off NSE value exists beyond which neurological recovery is impossible remains to be elucidated. For now, a single NSE cut-off value is not being recommended for outcome prediction on itself but instead should be used as a valuable adjunct within the process of early neuroprognostication after cardiac arrest.

Like NSE, S-100B has been extensively studied although its role as biomarker for neurological prognostication is less well established. Elevated serum astroglial S-100B as early as 24 hours after cardiac arrest are indicative for a dismal prognosis. Nonetheless, S-100B was not considered as a valid prognostic indicator in the revised guidelines since optimal cut-off values for the prediction of poor outcome varied among studies performed in TTM-treated patients [56, 57, 58, 59, 60].

An early CT of the brain is commonly used to rule out an unexpected intracerebral cause of coma in patients who remain unconscious after cardiac arrest. In fact, the admission of a comatose cardiac arrest patient, where the cause of arrest is ambiguous, should always be followed by a brain CT scan in order to rule out any cerebral pathology. Besides, recent studies showed that a brain CT scan performed within 24 to 48 hours after the arrest predicts poor neurological outcome accurately although the reported sensitivity was rather poor. The main CT finding of global cerebral anoxia after cardiac arrest is cerebral swelling and appears as a marked reduction in the interface between grey and white matter (quantified as the ratio (GWR) between grey and white matter densities). An GWR ranging between 1.12 and 1.22 has been associated with a lower likelihood for a good outcome (FPR 0%). Nonetheless, current guidelines state that further research is mandatory to support CT imaging as a prognostic instrument [61, 62, 63].

Compared to CT, diffusion-weighted brain MRI (DWI) is considered to be a more appropriate and robust imaging option to visualize hypoxic–ischemic cerebral lesions after cardiac arrest, and therefore might serve as prognostic adjunct. The majority of studies assessed the prognostic power of DWI based on the calculation of apparent diffusion coefficient values reflecting the degree of diffusion of water molecules in brain tissue. Two large, retrospective multicentre trials recently reported an FPR for poor outcome lying between 7 and 9% based on different DWI findings. Nonetheless, the insufficient power of available prognostication studies—which are based on a limited sample size and are mostly retrospective in nature—clarify why a brain MRI is still considered as optional within the process of neuroprognostication. In clinical practice, guidelines now recommend its use only in those patients who remain persistently comatose (i.e. more than 5 to 7 days after cardiac arrest) without obvious poor outcome signs based on other prognostic markers [64, 65, 66].

Based on expert opinion, a multimodal approach for the continuous evaluation of prognosis in all -TTM-treated cardiac arrest patients is proposed. This includes daily neurological examinations and, whenever possible, continuous monitoring of cortical activity, preferably using a simplified EEG trend monitor. If no such monitor is available, at least one conventional EEG should be performed in all comatose patients early after return to normothermia, in order to rule out potentially treatable epileptic seizures or for the purpose of prognosis. Brain imaging should be used to resolve specific questions in those patients who remain persistently comatose; CT scanning—if indicated—should be performed early (<24 hours) to rule out any cerebral pathology (trauma, bleeding), whereas MRI may be used in patients with prolonged coma to visualize an ischaemic injury at a later time (48–108 hours). SSEP, when available, should be performed in patients to confirm an already presumed poor neurological prognosis. Serial NSE is only recommended as an adjunct to continuous neurological assessment. The current knowledge only supports a decisive evaluation on prognostication, relying on all multimodal information, and a decision on the level of care not earlier than 72 hours after return to normothermia or even still later in a patient still under the influence of sedatives and analgesics.

In the previous section, we discussed the practical use of cerebral monitoring and its value in the particular case of comatose post-cardiac arrest patients. Although post-cardiac arrest patients constitute the vast majority of acute cardiac care patients necessitating any mode of cerebral monitoring, there are still other indications for the use of brain monitoring in these patients. Patients suffering from an acute stroke (ischaemic, e.g. secondary to AF, or haemorrhagic such as after thrombolytic treatment) can present with major changes in their neurological condition, necessitating extreme alertness to preserve brain function or to prevent any deleterious acute deterioration. In every acute cardiac care patient, the presence of abnormal consciousness implicates a thorough clinical neurological examination to assess the actual neurological state. Meanwhile, the need for any further monitoring should be thoroughly considered.