Central Nervous System Monitoring

Published on 27/02/2015 by admin

Filed under Anesthesiology

Last modified 27/02/2015

Print this page

rate 1 star rate 2 star rate 3 star rate 4 star rate 5 star
Your rating: none, Average: 5 (1 votes)

This article have been viewed 2685 times

16 Central Nervous System Monitoring

Many important new developments in neuromonitoring have occurred since the previous edition of this textbook went to press. First, professional society practice guidelines have been published for electroencephalographic (EEG), auditory- (AEPs), and somatosensory-evoked potentials (SSEPs), and transcranial Doppler (TCD) surgical monitoring. Second, randomized clinical trials have proved the clinical benefit of EEG and cerebral oximetry monitoring. Third, motor-evoked potential (MEP) monitoring has received U.S. Food and Drug Administration (FDA) clearance and has become a valuable tool to protect descending motor pathways in the brain and spinal cord. In this new edition, each of these developments is discussed, as well as many new studies extending the clinical value of perioperative neuromonitoring for cardiothoracic and vascular surgery.

Nearly half of the 1 million patients undergoing cardiac surgery each year worldwide will likely experience persistent cognitive decline.1 The direct annual cost to U.S. insurers for brain injury from just one type of cardiac surgery, myocardial revascularization, is estimated at $4 billion.2 Furthermore, the same processes that injure the central nervous system (CNS) also appear to cause dysfunction of other vital organs. Thus, there are enormous clinical and economic incentives to improve CNS protection during cardiac surgery.

Historically, there has been little enthusiasm for neurophysiologic monitoring during cardiac surgery because of the presumed key role of macroembolization. It is widely assumed that most brain injuries during adult cardiac surgery result from cerebral embolization of atheromatous or calcified material dislodged from sclerotic blood vessels during their manipulation. Until the introduction of myocardial revascularization without cardiopulmonary bypass (CPB) or aortic clamp application, these injuries often have been viewed as unavoidable and untreatable.

Technical developments have begun to alter this perception. First, CNS injuries still occur despite reductions in aortic manipulation with the new approaches to coronary artery bypass and aortic surgery.3 Second, neurophysiologic studies have implicated hypoperfusion and dysoxygenation as major causative factors in CNS injury4,5 (Box 16-1). Because these functional disturbances are often detectable and correctable, there is an impetus to examine the role of neurophysiologic monitoring in CNS protection (see Chapter 36).

Cardiac anesthesia provider familiarity with neuromonitoring is becoming increasingly important. The introduction of compact and simplified monitors of brain electrical activity, blood flow velocity, and oxygenation promises to integrate these devices and their information into cardiac anesthetic management. The goal of this chapter is to highlight the practical issues involved with these emerging neuromonitoring technologies. This emphasis on practicality limits discussion to FDA-approved devices.

Electroencephalography

EEG monitoring for ischemia detection has been performed since the first CPB procedures, but this long experience is not broad.6 In contrast with its widespread use during carotid endarterectomy, EEG monitoring for cardiac surgery is performed primarily in academic centers or those specializing in pediatric surgery. Limited use appears to have several causes.

First, small, practical, and affordable EEG monitors have only recently become available. For example, using a book-sized box of electronics and a notebook computer, it is now possible to concurrently display multichannel conventional and processed EEG, as well as bilateral TCD ultrasonic spectra and cerebral oxygen saturation. All of the resulting data can be easily transmitted over high-speed data lines for Web-based consultation or archival or post hoc analysis.

Second, the traditional diagnostic approach to EEG analysis depended on complex pattern recognition of 16-channel analog waveforms to identify focal ischemic changes.7 This analytic format necessitated extensive training and constant vigilance. As a result, cardiac surgery EEG monitoring directly by anesthesia providers has often been viewed as impractical. However, Craft et al8 and Edmonds et al9 have shown that a four-channel recording, which included bilateral activity from both the anterior and posterior circulation, was effective in identifying focal ischemia. In addition, computerized processing of EEG signals provides simplified trend displays that have helped to overcome many of the earlier complexities.

Third, EEG analysis during cardiac surgery was often confounded by anesthetics, hypothermia, and roller-pump artifacts.10 Fortunately, these technical problems have now been overcome in the following ways: (1) elimination or replacement of the troublesome roller pumps with centrifugal pumps, (2) routine use of mild hypothermic or normothermic bypass, and (3) adoption of fast-track anesthesia protocols that avoid marked EEG suppression.

image

Physiologic Basis of Electroencephalography

EEG-directed interventions designed to correct cerebral hypoperfusion during cardiac surgery require an appreciation of the underlying neurophysiologic substrate. Scalp-recorded EEG signals reflect the temporal and spatial summation of long-lasting (10 to 100 milliseconds) postsynaptic potentials that arise from columnar cortical pyramidal neurons (Figure 16-1). These potentials are produced by dipoles distributed over soma-dendritic surfaces. Pyramidal neurons have a long, vertically oriented, apical dendrite and shorter basal dendrites radiating from the soma base. Near-synchronous excitation (or inhibition) of neighboring dendritic membranes produces large-amplitude spatially summating vertical dipoles, whereas radial current layers are generated in the somatic region. Simultaneous current generation in the two regions may appear to be self-canceling at distant surface electrodes. In addition, traditional EEG depicts only voltage change, not absolute voltage. Thus, sustained high-frequency neuronal activity may result in a large but nonvarying surface voltage deviation that would be invisible to the conventional EEG. These important EEG characteristics should be appreciated when interpreting low-amplitude signals; they do not necessarily indicate synaptic quiescence.

EEG rhythms represent regularly recurring waveforms of similar shape and duration. These signal oscillations depend on the synchronous excitation of a neuronal population. The descriptive nature of conventional EEG characterizes the oscillations (measured in cycles per second [cps] or Hertz [Hz]) as sinusoids that were classified according to their amplitude and frequency. The terminology used to describe the frequency bands of the most common oscillatory patterns is illustrated (Figure 16-2). In addition, a high-frequency (25 to 55 Hz) gamma band is recognized (Box 16-2).

BOX 16-2 Electroencephalographic Frequency Bands

Delta 0.5 to 2 Hz
Theta 3 to 7 Hz
Alpha 8 to 12 Hz
Beta 13 to 24 Hz
Gamma 25 to 55 Hz

EEG oscillatory patterns are functional manifestations of specific intraneuronal networks. The extent of cortical processing among neighboring neuronal columns influences the extent of scalp-recorded EEG waveform synchronization and is not necessarily dependent on the subcortically mediated arousal level. At a high level of cortical processing, each neuronal palisade may function in relative independence. The resultant EEG signal will be of low amplitude, representing the distance-weighted average of many desynchronized micropotentials. The large number of small potentials is reflected in an EEG pattern characterized by a high dominant frequency (13- to 24-Hz beta waves). Such a pattern may be seen during very different vigilance states, such as awake, mentally alert (see Figure 16-2, top trace) versus rapid eye movement (REM; i.e., dream) sleep (see Figure 16-2, bottom trace). Partial cortical columnar synchronization develops with a reduction in information processing, resulting in higher amplitude and lower frequency EEG oscillations associated with a relaxed, drowsy state (see Figure 16-2; 8- to 12-Hz alpha rhythm). Progressive suppression is associated with lower frequency 3- to 7-Hz theta waves. Minimal processing leads to the very-high-amplitude, low-frequency hypersynchronous 0.5- to 2-Hz delta waves seen during the low vigilance states of deep coma, deepest sleep, hypoxia, ischemia, and some forms of surgical anesthesia.

Synchronization of cortical columns is influenced by subcortical structures, including the thalamus (Figure 16-3) and reticular activating system (Figure 16-4). Reticular inhibition can block the passage of sensory information to the cortex that is routed through thalamic relays. This state of functional deafferentation results in unconsciousness, an essential component of both natural sleep and surgical anesthesia.11 However, the individual components of a modern balanced anesthetic technique may differentially affect the separate control mechanisms for sensory processing and vigilance. Thus, an EEG pattern suggestive of a low vigilance state (i.e., surgical hypnosis) does not necessarily guarantee the absence of subcortical (i.e., unconscious) sensory perception (i.e., reflexive response to painful stimuli).12 Furthermore, because the neuronal basis for the EEG is primarily of cortical origin, it is not surprising that many univariate (i.e., single-variable) EEG amplitude or frequency descriptors are only weakly correlated with clinical measures of anesthetic effect or developing pathology involving primarily or exclusively subcortical structures.

image

Practical Considerations of Electroencephalographic Recording and Signal Processing

This section describes practical issues involved in the conversion of these tiny potentials into interpretable EEG displays. The process begins with choice of scalp electrodes (subdermal needle, metallic disk, or silver-silver chloride gel self-adhesive patch) and their location. All three electrode types provide high-quality signals. Single-use sterile needle electrodes are easy to apply but are invasive, relatively expensive, and not well tolerated by conscious patients. Reusable disk electrodes, held in place with conductive gel, gel-free self-abrading plastic retainers, or built into a nylon mesh cap, may be used on conscious patients and are the least costly option. Adhesive patch electrodes are generally used only on glabrous skin and have a cost midway between the other options.

Standardized electrode placement is based on the International 10-20 System (Figure 16-5). It permits uniform spacing of electrodes, independent of head circumference, in scalp regions known to correlate with specific areas of cerebral cortex. Four anatomic landmarks are used: the nasion, inion, and preauricular points. Electrodes are located at 10% or 20% segments of the distance between two of these landmarks. The alphanumeric label for each site uses an initial uppercase letter to signify the skull region (i.e., frontal, central, temporal, parietal, occipital, auricular, and mastoid). Second and sometimes third letters, in lowercase, further delineate position (e.g., “p” represents frontal pole, whereas “z” indicates zero or midline). Subscript numbers represent left (odd) or right (even) and specific hemispheric location, with the lowest numbers closest to midline. The prime notation (′) is used to signify specialized locations designed for certain evoked potential applications (e.g., C3′ and C4′ represent 2 cm posterior to C3 and C4, directly over upper limb sensory cortex).

The differential amplifiers used in EEG recording measure the voltage difference between two inputs. By convention, a negative voltage at input 1 relative to input 2 results in an upward deflection of the tracing. With a referential arrangement (montage) of recording channel selections (Figure 16-6, left), the input 2 connections from a series of channels are connected to a single electrode, whereas input 1 electrode connections all differ. Alternatively, in bipolar recordings, a common reference is not used (see Figure 16-6, right).

Although an array of scalp electrodes theoretically permits many possible montages to be used, the capability to quickly change montage varies greatly among different EEG monitors. This ability to quickly change recording montage may be important in the detection and characterization of both focal and diffuse abnormalities. With a referential montage (Figure 16-7), the transient will be distorted if the reference lies within the transient electric field. Alternatively, with a bipolar montage, the potential may actually disappear because of in-phase cancellation.

Montage choice also influences susceptibility to artifact. For example, millivolt ECG potentials may contaminate the thousand-fold smaller EEG signal. Contamination is often problematic with an ear or mastoid reference montage but may be invisible with an anterior- to-posterior bipolar montage (Figure 16-8). The extreme lateral placement of ear or mastoid references maximizes contamination by the perpendicularly oriented high-voltage dipole generated by the heart.

The frequency range involved in production of the EEG waveform is termed its bandwidth. The upper and lower bandwidth boundaries are controlled by filters that reject frequencies above and below the EEG bandwidth. Both the appearance of the unprocessed EEG waveform and the value of univariate numeric EEG descriptors such as the mean frequency may be heavily influenced by signal bandwidth. The same cerebral biopotential recorded by different EEG devices may result in dissimilar waveforms and numeric values.

Modern EEG monitors use digital microprocessors to analyze the amplified analog biopotentials. Yet, analog-to-digital conversion imposes limitations on signal processing. Digitization converts a continuously varying biopotential into a series (i.e., sample) of discrete quantal values. At least two samples per period are required to minimize conversion inaccuracies. Sampling (Nyquist) frequency must be greater than twice the highest frequency of interest. For example, with an EEG bandwidth of 50 Hz, the minimum acceptable sampling frequency is 100 Hz (e.g., 10-millisecond sampling interval). Aliasing, the counting of high-frequency signals as low-frequency input, may occur if the complex biopotential contains frequencies above the Nyquist frequency. Therefore, most EEG monitors contain antialiasing filters that sharply attenuate waveform components above the Nyquist frequency. The details of filtering further add to the manufacturer-specific characteristics of processed EEG.

The continuous analog signal is also simplified into a (usually) discontinuous set of segments of a fixed duration (i.e., epoch). Window functions can minimize, but not totally eliminate, digital distortion produced by the abrupt truncation of a continuously varying waveform. These window functions are numerical series containing the same number of elements as the epoch. Their purpose is to reduce the value of epoch terminal elements. In addition to windowing, commercial EEG analyzers often use another form of signal conditioning called whitening. The energy content of the EEG is not uniform at all frequencies, but instead is heavily skewed to the lower range. Whitening mathematically alters the momentary frequency–amplitude relations to achieve nearly equal energy per octave and may improve pattern recognition in processed waveforms. Antialiasing, windowing, and whitening may vary not only among different devices but among software versions used with a single device. The user should be aware that a standard unprocessed analog EEG signal may generate digitally processed displays and numeric descriptors that are unique for each monitor design and software version.

image

Display of Electroencephalographic Information

Time-Domain Analysis

Traditional display of the EEG is a graph of biopotential voltage (y-axis) as a function of time and, consequently, is described as a time-domain process. The objective of a diagnostic EEG is to identify the most likely cause of a detected abnormality at one moment in time. Typically, a diagnostic EEG is obtained under controlled conditions, using precisely defined protocols. Recorded EEG appearance is visually compared with reference patterns. Interpretation is based on recognition of unique waveform patterns that are pathognomonic for specific clinical conditions.13 In contrast, the goal of EEG monitoring is to identify clinically important change from an individualized baseline. Unlike diagnostic EEG interpretation, monitoring requires immediate assessment of continuously fluctuating signals in an electronically hostile, complex, and poorly controlled recording environment. Therefore, of necessity, interpretation relies less on pattern recognition and more on statistical characterization of change. Simple numerical descriptors thus may appropriately form an integral part of EEG monitoring.

Both EEG diagnostic and monitoring interpretations are based, in part, on the “Law of the EEG” (Box 16-3). It states that amplitude and dominant frequency are inversely related. As described earlier, synchronously generated postsynaptic potentials may produce large-amplitude biopotentials. However, long membrane time constants limit the number of changes that may occur per second (e.g., high amplitude, low frequency). Conversely, summation of spatially distributed asynchronous potentials results in EEG signals of low amplitude but relatively high frequency. Thus, the inverse relation between amplitude and frequency generally is maintained during unchanging cerebral metabolic states. Parallel increases in both may occur in some hypermetabolic states such as seizure activity, whereas decreases may be seen in hypometabolic states such as hypothermia. In the absence of these influences, simultaneous decreases in both amplitude and frequency may indicate ischemia or anoxia (Figure 16-9), whereas a parallel increase may be artifact (Figure 16-10).

Time-domain analysis of traditional electroencephalography uses linear amplitude voltage and time scales. The amplitude range of EEG signals is quite large (several hundred microvolts), and univariate statistical measures of its central tendency and dispersion may contain clinically useful information.14 Furthermore, amplitude variation may present clinically significant changes in reactivity that can be obscured by frequency-domain analysis.15 Advances in the technology of EEG amplitude integration have prompted a resurgent interest in this attractively simple approach, particularly in pediatrics.16

Frequency-Domain Analysis

An alternative method, frequency-domain analysis, is exemplified by the prismatic decomposition of white light into its component frequencies (i.e., color spectrum). As the basis of spectral analysis, the Fourier theorem states that a periodic function can be represented, in part, by a sinusoid at the fundamental frequency and an infinite series of integer multiples (i.e., harmonics). The Fourier function at a specific frequency equals the amplitude and phase angle of the associated sinusoid. Graphs of amplitude and phase angle as functions of frequency are called Fourier spectra (i.e., spectral analysis). The EEG amplitude spectral scale (Figure 16-11) squares voltage values to eliminate troublesome negative values. Squaring changes the unit of amplitude measure from microvolts to either picowatts (pW) or nanowatts (nW). However, a power amplitude scale tends to overemphasize large-amplitude changes. Clinically important changes in lower amplitude components that are readily discernible in the linearly scaled unprocessed EEG waveform may become invisible in power spectral displays.

Simplification of the large amount of spectral information generally has been achieved through the use of univariate numeric descriptors. Most commonly, the power contained in a specified traditional EEG frequency band (delta, theta, alpha, or beta) is calculated in absolute, relative, or normalized terms. Relative amplitude represents the fraction of total power contained in a specified frequency band (relative delta power = delta [−0.5 to 2 Hz] power/total [0.5 to 55 Hz] power). Normalization equalizes the total power of successive epochs to that of some arbitrary reference before the calculation of relative power (the delta power is described as Z-score change from a previous individualized baseline). The latter two derived measures are particularly useful in minimizing misinterpretation of spectral changes. For example, during the production of hypothermia, absolute delta power declines in parallel with cerebral metabolism, whereas the fraction of the total power in the delta band remains unchanged. In this circumstance, exclusive focus on absolute delta power may lead to the erroneous conclusion that the hypnotic state is decreasing.

The most widely used univariate frequency descriptors (Box 16-4) are (1) peak power frequency (the single frequency of the spectrum that contains the highest amplitude), (2) median power frequency (frequency below which 50% of the spectral power occurs), (3) mean spectral frequency (sum of power contained at each frequency of the spectrum times its frequency divided by the total power), (4) spectral edge frequency (SEF; frequency below which a predetermined fraction, usually 95%, of the spectral power occurs), and (5) suppression ratio (SR; percentage of flatline EEG contained within sampled epochs). The performance of two descriptors in characterizing clinically important EEG changes is compared (Figure 16-12). Note that at anesthetic induction, the large transient decrease in SEF and reciprocal SR increase. However, later transient SR increases signifying marked EEG suppression were not detected by SEF, apparently because of low-level radiofrequency contamination.

Pronk17 evaluated computer-processed univariate descriptors of EEG changes occurring before, during, and after CPB. Mean spectral frequency alone was sufficient to adequately describe all EEG changes except those occurring at very low amplitudes. Addition of a single-amplitude factor improved agreement with visual interpretation to 90%. Further factor addition did not improve agreement.

Multivariate (i.e., composed of several variables) descriptors have been developed to improve simple numeric characterization of clinically important EEG changes. With this approach, algorithms are used to generate a single number that represents the pattern of amplitude-frequency-phase relations occurring in a single epoch. Several commercially available monitors provide unitless numbers that have been transformed to arbitrary (i.e., 0 to 100) scales. Each monitor provides a different probability estimate of patient response to verbal instruction. Current monitors designed for use by anesthesia providers are listed in Box 16-5. BIS-XP, NT, PSI, and SNAP II are rule-based proprietary indices empirically derived from patient data. In contrast, CSI uses a fuzzy logic-based algorithm, whereas state entropy (SE) applies standard entropy equations to EEG analysis. Each product is designed to require the use of proprietary self-adhesive forehead sensors. Collectively, these products are now in widespread use as objective measures of hypnotic effect.

These hypnotic indices appear to provide clinically useful information. However, their fundamental differences may result in distinctly unique performances. Close agreement among these measures should not be expected. Thus, it is inappropriate and unjustified to apply clinical evidence of improved outcome obtained with one of these measures to a competing index. To date, adequately powered, prospective, randomized evidence of a statistically significant (i.e., 82%) reduction in risk for intraoperative awareness in adult noncardiac surgery has been achieved only with BIS-XP.24 A recent meta-analysis review of the extensive peer-reviewed BIS literature concluded that “BIS could reduce the incidence of perioperative recall in surgical patients with high risk of awareness.”25 In addition, the review authors opined that anesthesia guided by BIS monitoring could decrease anesthetic consumption and enhance recovery from relatively deep anesthesia. These findings suggest that prevention of intraoperative awareness may represent only the tip of the iceberg of potential clinical and economic benefit to be derived from routine quantitative EEG perioperative monitoring.

Scalp-recorded cerebral biopotentials are complex physiologic signals representing the algebraic summation of voltage changes produced from cortical synaptic activity (i.e., EEG), upper facial muscle activity (i.e., facial electromyogram [fEMG]), and eye movement (i.e., electro-oculogram [EOG]). During consciousness and light sedation, high-frequency gamma power (i.e., 25 to 55 Hz) is a mixture of EEG and subcortically influenced facial electromyogram. Muscle activity makes a larger contribution because of the closer proximity of signal generators to the recording electrodes. Hypnotics and analgesics typically suppress both cerebral and muscle activities, resulting in reduced gamma power. Because the upper facial muscles are relatively insensitive to moderate neuromuscular blockade, they may remain reactive to noxious stimuli.26 Nociception results in sudden gamma power increase, independent of activity in the lower frequency classical EEG bands.

The EEG analyzers just described either provide separate quantitative estimates of the high-frequency information or incorporate it into the hypnotic index. For example, the Datex-Ohmeda Entropy Module separately analyzes the 32- to 47-Hz band and terms the signal “response entropy.” Addition of response entropy to the lower frequency SE is claimed by the manufacturer to facilitate distinction between changes in hypnosis and analgesia, although supporting evidence for this proposition awaits carefully designed and adequately powered randomized, prospective studies. EEG suppression decreases both entropy indices because noise-free flatline EEG segments are generally thought to have near-zero entropy. However, during cardiac surgery, EEG signals that appear to be totally suppressed may be associated with paradoxically very high entropy values. To minimize this problem, SE uses a special algorithm that assigns zero entropy to totally suppressed EEG epochs.

In addition to the quantitative EEG numeric indices, many monitors also display pseudo-three-dimensional plots of successive power spectra as a function of time. This frequency-domain approach was originated by Joy27 and popularized by Bickford, who coined the term compressed spectral array (CSA).28 Popularity stems, in part, from enormous data compression. For example, the essential information contained in a 4-hour traditional EEG recording consuming more than 1000 pages of unprocessed waveforms can be displayed in CSA format on a single page.

With CSA (see Figure 16-11), successive power spectra of brief (2- to 60-second) EEG epochs are displayed as smoothed histograms of amplitude as a function of frequency. Spectral compression is achieved by partially overlaying successive spectra, with time represented on the z-axis. Hidden-line suppression improves clarity by avoiding overlap of successive traces. Although the display is esthetically attractive, it has limitations. The extent of data loss caused by spectral overlapping depends on the nonstandard axial rotation that varies among EEG monitors. More important, epoch duration and the frequency with which each is measured (i.e., update rate) may critically affect the presentation of clinically important change. For example, there are three distinctly different burst-suppression CSA patterns: high-amplitude bursts, flat line, or a combination of the two.29

Fleming and Smith30 designed an alternative to the CSA display to reduce data loss. Density-modulated spectral array (DSA) uses a two-dimensional monochrome dot matrix plot of time as a function of frequency (see Figure 16-11). The density of dots indicates the amplitude at a particular time-frequency intersection (e.g., an intense large spot indicates high amplitude). Clinically significant shifts in frequency may be detected earlier and more easily than with CSA. However, the resolution of amplitude changes is reduced. Therefore, the color density-modulated spectral array (CDSA) has been developed to enhance amplitude resolution (see Figure 16-11). The CSA, DSA, and CDSA displays are not well-suited for the detection of nonstationary or transient phenomena like burst-suppression or epileptiform activity.

In summary, a quick assessment of EEG change in either the time or frequency domain focuses on the following: (1) maximal peak-to-peak amplitude, (2) relation of maximal amplitude to dominant frequency, (3) amplitude and frequency variability, and (4) new or growing asymmetry between homotopic (i.e., same position on each cerebral hemisphere) EEG derivations. These objectives are generally best achieved through the viewing of both unprocessed and processed displays with a clear understanding of the characteristics and limitations of each (Box 16-6).

image

Electroencephalography for Injury Prevention during Cardiac Surgery

Since 2005, a great deal of new information has become available on the rationale for perioperative EEG monitoring. The American Society of Neurophysiologic Monitoring has just published its EEG practice guideline.31 It describes the technical features and limitations of current EEG monitors and the available approaches to injury prevention. In addition, a new compendium has described the process of EEG monitoring in detail,32,33 and special journal issues have been devoted specifically to cardiac surgery applications.34,35

The physiologic basis for EEG monitoring is the normally tight coupling among cerebral cortical synaptic activity, metabolism, and blood flow. Such coupling is necessary because of the large energy requirements of interneuronal communication. Indeed, at least 60% of neuronal oxygen and glucose is consumed in the processes of synaptic and axonal transmission, whereas the remainder is used to maintain cellular integrity. Neurons rapidly adjust their signaling capabilities to conserve vital energy stores. Even a slight new imbalance between supply and consumption of energy substrates is manifested by altered synaptic activity. The EEG provides a sensitive measure of this synaptic change and represents an early warning of developing injury. Identification and correction of the physiologic imbalance may then avert serious injury. It must be emphasized that EEG alterations signify imbalance, not necessarily injury. For example, using functional brain imaging with fluorodeoxyglucose, Alkire36 demonstrated EEG relative alpha power to be linearly related to cerebral metabolism during propofol and isoflurane anesthesia. In this case, the synaptic depression manifested by the reduction in high-frequency activity signified anesthetic effect, not ischemia.

Furthermore, because clinically apparent neurologic injury often involves subcortical circuits and structures invisible to the EEG, an expectation of perfect agreement between specific EEG change and neurologic outcome is unwarranted. Imbalance in neuronal homeostasis may lead to either enhanced or diminished excitability. The majority of cortical neurons are small interneurons involved with maintenance of inhibitory tone. Their limited capacity for ionic buffering and energy storage makes them especially susceptible to imbalance. Early signs of energy deficiency may paradoxically result in EEG signs of excitation caused by disinhibition.37 Conversely, with inadequate analgesia, intense sensory stimuli may result in a paradoxically depressed EEG activity resembling ischemia38 (Box 16-7).

Although the EEG is a very sensitive indicator of cerebrocortical synaptic depression, possibly signifying cerebral ischemia or hypoxia, it is not specific. EEG suppression may also be the result of cooling or hypnotic agents. Fortunately, the time course of EEG suppression and information available from other monitors usually permit distinction between potentially harmful (e.g., ischemia) and harmless (e.g., hypothermia) causes. The following clinical situations provide the rationale for routine EEG monitoring during cardiac surgery: documentation of preexisting abnormalities, hypnotic adequacy, perfusion, cooling, rewarming, and seizure detection.

Objective Measurement of Hypnotic Effect

Recording the transition from wakefulness to unresponsiveness permits detection of unusual sensitivity or resistance to general anesthetics (Figure 16-13). Such information is vital during fast-track anesthesia protocols to avoid excessive or inadequate anesthesia. Although no available neurophysiologic monitoring modality is an infallible predictor of anesthetic inadequacy, characteristic changes in the EEG frequency pattern provide an easily recognized warning of potential patient awareness. EEG detection of excessive or inadequate hypnosis during cardiac surgery may be aided by the multivariate EEG descriptors.

Hypocarbia and Hypercarbia

Cerebral arteries normally constrict in response to decreased carbon dioxide tension or hydrogen ion concentration. With reactive arteries, hypocarbia may lead to cerebral ischemia because flow is reduced 4%/mm Hg [CO2]. In this circumstance, return to normocarbia can markedly improve cerebral perfusion.41 Conversely, hypercarbia may steal blood from the dilated vessels of an already underperfused region, resulting in focal EEG slowing. This circumstance may be exacerbated by the use of volatile anesthetics causing cerebral vasodilation.

Cooling

During cardiac surgery requiring deep hypothermic circulatory arrest, the EEG provides an effective method for assessing the effects of cooling. Optimal brain temperature may be viewed as a balance between decreased cerebral oxygen consumption and increased risk for coagulopathy. Actual brain temperature cannot be measured directly and appears to be influenced by many variables including the rate of cooling, as well as acid-base and anesthetic management. The EEG is often used to assess the functional consequences of brain cooling because a flat line pattern signifies cortical synaptic quiescence. The wide interpatient variation in flat line temperature is the rationale for EEG guidance of the cooling process.34 Alternative use of a fixed temperature criterion increases the risks for both excessive and inadequate brain cooling.

As nasopharyngeal or tympanic temperature decreases (Figure 16-15), there is a gradual loss of absolute spectral power across all frequency bands.34 Although the reversible decreases in absolute spectral band power are quite large, computing these changes as relative spectral band power minimizes them suggesting that initial brain cooling exerts a similar decrease in spectral power across all frequency bands and cortical regions.42 Cooling below 28°C leads to progressive slowing of the residual EEG until the EEG waveform becomes a flat line.

Rewarming

After circulatory arrest, the EEG documents the recovery of synaptic function during rewarming. However, rapid rewarming may result in cerebral ischemia because of cold-triggered vasoparesis, which uncouples cerebral blood flow and metabolism. In this case, rewarming-induced increase in metabolic demand may outpace the lagging delivery of essential nutrients.43 This situation is indicated by a loss of EEG high-frequency activity accompanying the increase in cranial temperature. Although the practice is controversial, administration of a metabolic suppressant drug such as propofol, dosed to EEG burst suppression, may facilitate balancing cerebral perfusion with metabolic demand.44

Myocardial Revascularization without Extracorporeal Support

Avoidance of CPB during myocardial revascularization may protect patients from the many potential hazards of this nonphysiologic insult.45 EEG stability during beating-heart coronary revascularization is often observed, despite transient hypotension and bradycardia. Nevertheless, some surgeons believe that neuromonitoring is mandatory because the combination of controlled hypotension and vascular torsion can suddenly disrupt cerebral perfusion. In the absence of the neuroprotective effects of mild hypothermia, even relatively brief episodes of cerebral ischemia may result in injury.

Seizure Detection

Certain anesthetics and adjuvants such as etomidate, sevoflurane, and the opioid analgesics may produce seizure-like EEG activity, although the clinical manifestations may be obscured by neuromuscular blockade (Figure 16-16).4648 The consequences of this anesthetic-induced seizure-like activity have not been established in adults. However, seizures may be deleterious to the developing brain. Bellinger et al49 showed that perioperative EEG seizure activity in infant cardiac surgery patients was associated with a 10-point decline in expected IQ when subsequently measured in the patients’ fifth year after surgery (Box 16-8).

Auditory-evoked potentials

Important new literature on the intraoperative use of auditory-evoked potentials (AEPs) includes the publication of a professional society practice guideline,50 a neuromonitoring textbook,51 and a special journal issue on cardiac surgery neuromonitoring.52

AEPs assess specific areas of the brainstem, midbrain, and auditory cortices. Because of their simplicity, objectivity, and reproducibility, AEPs are suitable for monitoring patients during cardiovascular surgery. Specific applications of AEP monitoring in this environment are the assessment of temperature effects on brainstem function and evaluation of hypnotic effect. Direct involvement of cardiac anesthesia providers with AEP monitoring is likely to increase after the introduction of EEG/AEP modules designed for use with available operating room physiologic monitors.

Acoustic stimuli trigger a neural response integrated by a synchronized neuronal depolarization that travels from the auditory nerve to the cerebral cortex. Scalp-recorded signals, obtained from electrodes located at the vertex and earlobe, contain both the AEPs and other unrelated EEG and EMG activity. Extraction of the relatively low-amplitude AEPs from the larger amplitude background activity requires signal-averaging techniques.52 Because the AEP character remains constant for each stimulus repetition, averaging of many repetitions suppresses the inconstant background. For the AEP sensory stimulus, acoustic clicks are the most commonly used.52 These broadband signals are generated by unidirectional rectangular short pulses (40 to 500 microseconds) with frequency spectra below 10 KHz.

The AEPs comprise a series of biopotentials generated at all levels of the auditory system in response to an acoustic stimulus (Figure 16-17). A dozen peaks have been identified within the first 100 milliseconds after stimulus onset using scalp electrodes.52 Each peak is described by its poststimulus latency and peak-to-peak amplitude. AEPs are commonly classified as early or middle-latency potentials.52 Figure 16-18 is a schematic representation of the AEPs most commonly used for surgical monitoring. Early AEPs are generated from the auditory nerve and the brainstem and include a series of wavelets recorded within the first 10 milliseconds poststimulus. These evoked responses have been designated brainstem auditory-evoked potentials (BAEPs). Seven waves (I to VII) characterize the adult BAEP. Peaks I and II are generally thought to originate from the distal and proximal parts of the eighth nerve, and peak III arises from the cochlear nucleus. Peak IV sources include the superior olivary complex, cochlear nucleus, and nucleus of the lateral lemniscus. Peak V contributors seem to include both the lateral lemniscus and inferior colliculus. Peak VI and VII origins are not well defined but may arise from the medial geniculate body and the acoustic radiations. BAEPs are useful in assessing brainstem and subcortical function during surgery, in part because of their relative resistance to the suppressant effects of most anesthetics.53

The middle-latency auditory-evoked potentials (MLAEPs), with poststimulus latencies between 10 and 100 milliseconds, are generated in the midbrain and primary auditory cortex.52 In an awake adult subject, the MLAEPs usually consist of three main peaks: Na, Pa, and Nb, with respective latencies near 15, 28, and 40 milliseconds (see Figure 16-18). Children under general anesthesia commonly display the trimodal configuration of the adult MLAEP waveform (Na, Pa, Nb waves), although neonates may exhibit only a small Pa wave.54 Many agents with hypnotic effects prolong the latency and suppress the amplitude of Pa and Nb in a concentration-dependent manner.55 It appears that the latency and amplitude changes allow reliable detection of consciousness and nociception during cardiac surgery.55 In addition, parallel monitoring of MLAEP and quantitative EEG descriptors (i.e., BIS) may permit distinction between the hypnotic and antinociceptive anesthetic components.55 This approach has also been used successfully in pediatric cardiac surgery patients to objectively assess postoperative sedation.56

Amplitude and latency from each of the primary MLAEP components have been integrated into a proprietary autoregressive linear function (A-Line; Danmeter A/S, Odense, Denmark) to facilitate continuous perioperative monitoring.57 Subsequently, this metric was expanded to the A-Line Autoregressive Index (AAI), which included the quantitative EEG descriptors percentage burst suppression and β ratio (i.e., percentage of total EEG power contained in the high-frequency β frequency band).58

Part of the benefit of AEPs to cardiac surgery derives from their temperature sensitivity because cooling slows both axonal conduction and synaptic transmission. During cooling, the BAEP wave V Q10 (ratio of two values separated by 10°C) is 2.2.59 A decrease of tympanic or nasopharyngeal temperature from 35°C to 25°C doubles wave V latency. Further cooling will eventually suppress waves III to V completely, signifying the virtual elimination of synaptic transmission within the brainstem auditory circuits. BAEPs document complete deep hypothermic electrocerebral silence before temporary circulatory arrest.59

The critical protective action of hypothermia on the brain cannot be accurately assessed by thermometry because of marked individual differences in thermal compartmentation throughout the body.60 Even within the brain, cooling technique (e.g., rapid vs. slow, alpha-stat vs. pH-stat acid-base balance, α-adrenergic blockade vs. none) may result in substantial thermal inhomogeneity within the cerebrum.5961 Therefore, hypothermia-induced electrocortical silence (i.e., flat EEG) does not necessarily indicate cessation of synaptic activity within deep brain structures. The high metabolic rates of some of these structures (e.g., basal ganglia and inferior colliculus) render them particularly vulnerable to ischemic injury.61 EEG quiescence plus a loss of BAEP waves III to V (Figure 16-19) ensure thorough cooling of the brain core. This approach appears to offer an optimal neuroprotective environment during temporary cessation of cerebral perfusion (Box 16-9).