Файл: PLUM AND POSNER S DIAGNOSIS OF STUPOR AND COM-1.pdf

different pathways may fire independently under a variety of different conditions, modulating the functional capacities of cortical neurons during a wide range of behavioral states.

Behavioral State Switching

An important feature of the ascending arousal system is its interconnectivity: the cell groups that contribute to the system also maintain substantial connections with other components of the system. Another important property of the system is that nearly all of these components receive inputs from the ventrolateral preoptic nucleus.82–84 Ventrolateral preoptic neurons

contain the inhibitory transmitters GABA and galanin; they fire fastest during sleep.40,83,85 Le-

sions of the ventrolateral preoptic nucleus cause a state of profound insomnia in animals,86,87 and

such lesions undoubtedly accounted for the insomniac patients described by von Economo19 (see Box 1–1).

The ventrolateral preoptic neurons also receive extensive inhibitory inputs from many components of the ascending arousal system. This mutual inhibition between the ventrolateral preoptic nucleus and the ascending arousal system has interesting implications for the mechanisms of the natural switching from wakefulness to sleep over the course of the day, and from slow-wave to REM sleep over the course

of the night. Electrical engineers call a circuit in which the two sides inhibit each other a ‘‘flipflop’’ switch.84 Each side of a flip-flop circuit is self-reinforcing (i.e., when the neurons are firing, they inhibit neurons that would otherwise turn them off, and hence they are disinhibited by their own activity). As a result, firing by each side of the circuit tends to be self-perpetuating, and the circuit tends to spend nearly all of its time with either one side or the other in ascendancy, and very little time in transition. These sharp boundaries between wakefulness and sleep are a key feature of normal physiology, as it would be maladaptive for animals to walk around half-asleep or to spend long portions of their normal sleep cycle half-awake.

REM sleep is a stage of sleep in which the brain enters a very different state from the highvoltage slow waves that characterize NREM sleep. As indicated in Box 1–3, during REM sleep, the forebrain shows low-voltage, fast EEG activity similar to wakefulness, and the ascending cholinergic system is even more active than during a wakeful state. However, the ascending monoaminergic systems cease firing virtually completely during REM sleep,46–49 so that the increased thalamocortical transmission seen during REM sleep falls upon a cerebral cortex that lacks the priming to maintain a wakeful state. As a result, REM sleep is sometimes called paradoxical sleep because the cortex gives an EEG appearance of wakefulness, and yet the

individual is profoundly unresponsive to external stimuli.

A second flip-flop switch in the pons for switching from NREM to REM sleep (and back again) has recently been identified in the rostral pons. Many GABAergic neurons in the extended part of the ventrolateral preoptic nucleus are specifically active during REM sleep, suggesting that they inhibit a population of REM-off neurons.88 In addition, the orexin neurons in the lateral hypothalamus are excitatory, but their firing inhibits REM sleep, suggesting that they may activate REM-off neurons, as patients or animals with narcolepsy who lack orexin neurons transition into REM sleep exceptionally quickly.70,89 By searching for the intersection of these two pathways, a population of neurons was defined in the rostral pons, including the ventrolateral periaqueductal gray matter and the lateral pontine tegmentum at the level where they are adjacent to the dorsal raphe nucleus. These sites contain many GABAergic neurons, and lesions of this region increase REM sleep, confirming a REM-off influence.53 GABAergic neurons in the REM-off area innervate an adjacent region including the sublaterodorsal nucleus and pre-coeruleus region that contain REM-active neurons. This REM-on region contains two types of neurons. GABAergic neurons, mainly in the sublaterodorsal nucleus, project back to the REM-off area. This produces a flip-

flop switch relationship accounting for the tendency for transitions into and out of REM sleep to be relatively abrupt. A second population of neurons is glutamatergic. Glutamatergic REMon neurons in the sublaterodorsal nucleus project to the brainstem and spinal cord, where they are thought to be responsible for the motor manifestations of REM sleep, including atonia and perhaps the rapid eye movements that are the hallmarks of the state. Glutamatergic REMon neurons in the coeruleus region target the basal forebrain where they appear to be critical for maintaining EEG phenomena associated with REM sleep.

Cholinergic and monoaminergic influences may have a modulatory effect on REM sleep by playing upon this flip-flop switch mechanism. Although lesions of these systems do not have a major effect on REM sleep, overactivity may have quite dramatic effects. For example, injections of cholinomimetic agents into the region containing the REM switch can trigger prolonged bouts of a REM-like state in animals.90 Whether this is due to activating REM-on neurons or inhibiting REM-off neurons (or both) is not known. On the other hand, patients who take antidepressants that are either serotonin or norepinephrine reuptake inhibitors (or both) have very little REM sleep. This effect may be due to the excess monoamines activating the REM-off neurons or inhibiting the REM-on

neurons (or both) and thereby locking the individual out of REM sleep.70,89

Relationship of Coma to Sleep

Because the brain enters a state of quiescence during sleep on a daily basis, it is natural to wonder whether coma may not be a pathologic entrance into the sleep state. In fact, both impaired states of consciousness and NREM sleep are characterized by EEG patterns that include increased amounts of high-voltage slow waves. Both conditions are due, ultimately, to lack of activity by the ascending arousal system. However, in sleep, the lack of activity is due to an intrinsically regulated inhibition of the arousal system, whereas in coma the impairment of the arousal system is due either to damage to the arousal system or to diffuse dysfunction of its diencephalic or forebrain targets.

Because sleep is a regulated state, it has several characteristics that distinguish it from coma. A key feature of sleep is that the subject can be aroused from it to wakefulness. Patients who are obtunded may be aroused briefly, but they require continuous stimulation to maintain a wakeful state, and comatose patients may not be arousable at all. In addition, sleeping subjects undergo a variety of postural adjustments, including yawning, stretching, and turning, which are not seen in patients with pathologic impairment of level of consciousness.

The most important difference, however, is the lack of cycling between NREM and REM sleep in patients in coma. Sleeping subjects undergo a characteristic pattern of waxing and waning depth of NREM sleep during the night, punctuated by bouts of REM sleep, usually beginning when the NREM sleep reaches its lightest phase. The monotonic high-voltage slow waves in the EEG of the comatose patient indicate that although coma may share with NREM sleep the property of a low level of activity in the ascending arousal systems, it is a fundamentally different and pathologic state.

The Cerebral Hemispheres

and Conscious Behavior

The cerebral cortex acts like a massively parallel processor that breaks down the components of

sensory experience into a wide array of abstractions that are analyzed independently and in parallel during normal conscious experience.42 This organizational scheme predicts many of the properties of consciousness, and it sheds light on how these many parallel streams of cortical activity are reassimilated into a single conscious state.

The cerebral neocortex of mammals, from rodents to humans, consists of a sheet of neurons divided into six layers. Inputs from the thalamic relay nuclei arrive mainly in layer IV, which consists of small granule cells. Inputs from other cortical areas arrive into layers II, III, and V. Layers II and III consist of smallto mediumsized pyramidal cells, arrayed with their apical dendrites pointing toward the cortical surface. Layer V contains much larger pyramidal cells, also in the same orientation. The apical dendrites of the pyramidal cells in layers II, III, and V receive afferents from thalamic and cortical axons that course through layer I parallel to the cortical surface. Layer VI comprises a varied collection of neurons of different shapes and sizes (the polymorph layer). Layer III provides most projections to other cortical areas, whereas layer V provides long-range projections to the brainstem and spinal cord. The deep part of layer V projects to the striatum. Layer VI provides the reciprocal output from the cortex back to the thalamus.91

It has been known since the 1960s that the neurons in successive layers along a line drawn through the cerebral cortex perpendicular to

the pial surface all tend to be concerned with similar sensory or motor processes.92,93 These

neurons form columns, of about 0.3 to 0.5 mm in width, in which the nerve cells share incoming signals in a vertically integrated manner. Recordings of neurons in each successive layer of a column of visual cortex, for example, all respond to bars of light in a particular orientation in a particular part of the visual field. Columns of neurons send information to one another and to higher order association areas via projection cells in layer III and, to a lesser extent, layer V.94 In this way, columns of neurons are able to extract progressively more complex and abstract information from an incoming sensory stimulus. For example, neurons in a primary visual cortical area may be primarily concernedwithsimple lines,edges,andcorners, but by integrating their inputs, a neuron in a higher order visual association area may

respond only to a complex shape, such as a hand or a brush.

The organization of the cortical column does not vary much from mammals with the most simple cortex, such as rodents, to primates with much larger and more complex cortical development. The depth or width of a column, for example, is only marginally larger in a primate brain than in a rat brain. What has changed most across evolution has been the number of columns. The hugely enlarged sheet of cortical columns in a human brain provides the massively parallel processing power needed to perform a sonata on the piano, solve a differential equation, or send a rocket to another planet.

An important principle of cortical organization is that neurons in different areas of the cerebral cortex specialize in certain types of operations. In a young brain, before school age, it is possible for cortical functions to reorga-

nize themselves to an astonishing degree if one area of cortex is damaged. However, the organization of cortical information processing goes through a series of critical stages during development, in which the maturing cortex gives up a degree of plasticity but demonstrates improved efficiency of processing.95,96 In adults, the ability to perform a specific cognitive process may be irretrievably assigned to a region of cortex, and when that area is damaged, the individual not only loses the ability to perform that operation, but also loses the very concept that the information of that type exists. Hence, the individual with a large right parietal infarct not only loses the ability to appreciate stimuli from the left side of space, but also loses the concept that there is a left side of space. We have witnessed a patient with a large right parietal lobe tumor who ate only the food on the right side of her plate; when done, she would

get up and turn around to the right, until the remaining food appeared on her right side, as she was entirely unable to conceive that the plate or space itself had a left side. Similarly, a patient with aphasia due to damage to Wernicke’s area in the dominant temporal lobe not only cannot appreciate the language symbol content of speech, but also can no longer comprehend that language symbols are an operative component of speech. Such a patient continues to speak meaningless babble and is surprised that others no longer understand his speech because the very concept that language symbols are embedded in speech eludes him.

This concept of fractional loss of consciousness is critical because it explains confusional states caused by focal cortical lesions. It is also a common observation by clinicians that, if the cerebral cortex is damaged in multiple locations by a multifocal disorder, it can eventually cease to function as a whole, producing a state of such severe cognitive impairment as to give the appearance of a global loss of consciousness. During a Wada test, a patient receives an injection of a short-acting barbiturate into the carotid artery to anesthetize one hemisphere so that its role in language can be assessed prior to cortical surgery. When the left hemisphere is acutely anesthetized, the patient gives the appearance of confusion and is typically placid but difficult to test due to the absence of language skills. When the patient recovers, he or she typically is amnestic for the event, as much of memory is encoded verbally. Following a right hemisphere injection, the patient also typically appears to be confused and is unable to orient to his or her surroundings, but can answer simple questions and perform simple commands. The experience also may not be remembered clearly, perhaps because of the sudden inability to encode visuospatial memory.

However, the patient does not appear to be unconscious when either hemisphere is acutely anesthetized. An important principle of examining patients with impaired consciousness is that the condition is not caused by a lesion whose acute effects are confined to a single hemisphere. A very large space-occupying lesion may simultaneously damage both hemispheres or may compress the diencephalon, causing impairment of consciousness, but an acute infarct of one hemisphere does not. Hence, loss of consciousness is not a typical feature of unilateral carotid disease unless both hemispheres

are supplied by a single carotid artery or the patient has had a subsequent seizure.

The concept of the cerebral cortex as a massively parallel processor introduces the question of how all of these parallel streams of information are eventually integrated into a single con-

sciousness, a conundrum that has been called the binding problem.97,98 Embedded in this

question, however, is a supposition: that it is necessary to reassemble all aspects of our experience into a single whole so that they can be monitored by an internal being, like a small person or homunculus watching a television screen. Although most people believe that they experience consciousness in this way, there is no a priori reason why such a self-experience cannot be the neurophysiologic outcome of the massively parallel processing (i.e., the illusion of reassembly, without the brain actually requiring that to occur in physical space). For example, people experience the visual world as an unbroken scene. However, each of us has a pair of holes in the visual fields where the optic nerves penetrate the retina. This blind spot can be demonstrated by passing a small object along the visual horizon until it disappears. However, the visual field is ‘‘seen’’ by the conscious self as a single unbroken expanse, and this hole is papered over with whatever visual material borders it. If the brain can produce this type of conscious impression in the absence of reality, there is no reason to think that it requires a physiologic reassembly of other stimuli for presentation to a central homunculus. Rather, consciousness may be conceived as a property of the integrated activity of the two cerebral hemispheres and not in need of a separate physical manifestation.

Despite this view of consciousness as an ‘‘emergent’’ property of hemispheric information processing, the hemispheres do require a mechanism for arriving at a singularity of thought and action. If each of the independent information streams in the cortical parallel processor could separately command motor responses, human movement would be a hopeless confusion of mixed activities. A good example is seen in patients in whom the corpus callosum has been transected to prevent spread of epileptic seizures.99 In such ‘‘split-brain’’ patients, the left hand may button a shirt and the right hand follow along right behind it unbuttoning. If independent action of the two hemispheres can be so disconcerting, one could only imagine