Книги по МРТ КТ на английском языке / PLUM AND POSNER S DIAGNOSIS OF STUPOR AND COM-1

60 years.16 In a prospective study of 372 patients with a GCS score of less than 13, age older than 50 years and lower GCS scores correlated with higher mortality.16 A prospective series with 2,664 patients found an essentially linear correlation of age and outcome following severe brain injury.17 The odds of an outcome less than 4 on the GOS increased 40% to 50% for every 10 years of age as a continuous variable. A meta-analysis of 5,600 patients identified a continuously worsening prognosis with increasing age without a sharp stepwise drop at any point.17 Several factors, other than age alone, may play a role in the association of age with outcome in TBI. Data from the Traumatic Coma Data Bank8 reveal an increased incidence of intracranial hemorrhage with age and premorbid medical illnesses, but did not demonstrate a significant statistical association.

NEURO-OPHTHALMOLOGIC SIGNS

The Brain Trauma Foundation review identified class I evidence that loss of pupillary light reflexes has at least a 70% PPV for a poor prognosis following TBI. Bilateral absence of pupillary or oculocephalic responses or both at any point in the illness predicts an outcome less than 4 on the GOS. In one series, 95% of patients who had either bilaterally nonreactive pupils or absent oculocephalic responses at 6 hours after injury died.18

SECONDARY INJURIES

Hypotension, hypoxia, and uncontrolled intracranial hypertension are independent predictors of poor outcome. Class I evidence supports a high likelihood of an outcome less than 4 on the GOS in comatose TBI patients who suffer either hypoxia or hypotension (defined as a systolic blood pressure of less than 90 mm Hg) early in the course. A single episode of hypotension (arterial line reading) is associated with a doubling of mortality and a significant increase in morbidity.8

NEUROIMAGING

Several neuroimaging findings correlate with outcome following TBI. Class I and strong class II evidence identifies several computed tomography (CT) findings that predict outcome8; accurate interpretation requires consideration of

the type of brain injury (e.g., focal brain injuries vs. diffuse axonal injury). The majority of patients with TBI have an abnormal CT scan, but certain findings carry a stronger predictive value for an outcome less than 4 on the GOS. Compression of the basal cisterns, a reliable indicator of increased ICP, is a strong negative predictor in several studies19 (see 8 for review). Midline shift of brain structures, another indicator of increased ICP, is also a predictor of an outcome less than 4 on the GOS.20 A midline shift of greater than 1.5 cm has a 70% PPV of death.8 Other CT findings that predict an outcome less than 4 on the GOS include traumatic subarachnoid hemorrhage in the suprasellar or ambient cisterns, and mass lesions (intracerebral hematoma, variable density CT abnormalities, epidural and subdural hematomas).

DURATION OF COMA

Figure 9–2 reproduces Carlsson and colleagues’21 classic diagram (1968) of the effect of duration of TBI-induced coma on outcomes at different ages. Not surprisingly, the longer the coma lasts, the worse the outcome is. Although length of coma provides a good indication of severity of brain damage, it can be determined only retrospectively when the patient awakens and thus cannot be used for early prognosis of outcome. On the other hand, it can be predicted with some confidence that a patient in prolonged coma is unlikely to recover. The same limitation applies to efforts to correlate outcomes of recovery of cognitive functions with the duration of posttraumatic amnesia.

ELECTROPHYSIOLOGIC MARKERS

Electrophysiologic measures have limited effectiveness in assessing TBI outcome. Several electroencephalographic (EEG) abnormalities are seen following TBI,22 and although EEG is useful for the identification of treatable complications of head trauma such as seizures, it does not predict outcome. Somatosensoryevoked potentials (SSEPs) are a better indicator.23 Bilateral absence of cortical components of SSEPs strongly correlates with a GOS below 424; in one small study, bilateral loss of SSEPs predicted outcomes of death or VS in all patients,25 but other reports indicate that bilateral loss of cortical response in posttraumatic coma may, on rare occasions, be associated with

favorable outcome.26,27 In these published reports, the measurements may have been confounded by sedating medications or the very early testing of the evoked potentials. Logi and associates24 prospectively studied 131 comatose patients of varying etiologies, including head trauma patients (N ¼ 22), and found 100% specificity for bilateral absence of cortical responses predicting nonawakening when sedating medications had been withdrawn and there were no other metabolic disturbances. Other electrophysiologic markers, including cognitive event-related potentials,28 might provide better prognostic value in future studies. Lew and colleagues25 suggested that the P300 response elicited by spoken words such as ‘‘mommy’’ may find use as an early predictor of outcomes greater than 3 on the GOS for comatose TBI patients. However, Perrin and associates29 found that similar P300 paradigms could not differentiate patients remaining in VS studied months after injury from other patients recovering to higher functional levels.

BIOCHEMICAL MARKERS

Elevated serum levels of glial fibrillary acidic protein (GFAP), part of the astroglial skeleton, and S100B, an astroglial protein, have been reported to predict mortality.30 In 42 severely

injured adults studied within 7 days of injury, the ratio of glutamate/glutamine (Glx) and choline (Cho) was significantly elevated in occipital gray and parietal white matter in patients who showed long-term (6- to 12-month) outcomes of less than 4 on the GOS.31

Nontraumatic Coma

PROSPECTIVE ANALYSES

OF OUTCOME FROM

NONTRAUMATIC COMA

In the late 1960s, a team of investigators at The New York Hospital, led by Dr. Plum and coworkers, in close association with Dr. Jennett and colleagues in Glasgow, undertook prospective studies of the outcome from coma as caused by medical disorders.4 Collaborating with the Royal Victoria Hospital, Newcastle- upon-Tyne, United Kingdom, and the San Francisco General Hospital, the investigators ultimately evaluated 500 patients in acute nontraumatic coma. All patients over 12 years old, save those with head trauma or exogenous intoxication in acute coma, were identified and repeatedly examined. Meticulous efforts were made to examine every patient in coma using examining techniques that guaranteed consistency of observation. To avoid bias, the

examiners refrained from either making recommendations for therapy or disclosing preliminary results to the treating staffs. The patients were followed for a minimum of 12 months (unless death occurred first) and many for much longer (only two of the 500 patients were lost to follow-up). This large population provided landmark data on substantial numbers of individuals in each of the major disease categories, permitting correlations between outcome and both the severity of early signs of neurologic dysfunction and the specific etiology of coma. Subsequent studies have largely confirmed the conclusions drawn from this patient population, including larger prospective studies of coma following cardiac arrest.2

The results of the medical coma study indicate that loss of consciousness lasting 6 hours or more bestows a poor prognosis. Of the 500 patients, 379 (76%) died within the first month and 88% had died by the end of a year. Threequarters of those dying by 1 month never regained consciousness, and within that month, only 15% of the entire 500 recovered to a GOS of 4 or 5.

Table 9–4 charts the best 1-month recovery by disease state. Some of the patients died during that first month of nonneurologic causes, but the table is constructed so as to indicate the highest possible chance of recovery by the brain. (Actual outcome from the illness in many instances was worse than this best neurologic state, because some patients who temporarily recovered neurologically died from complications, such as recurrent cardiac arrhythmias, infections, and pulmonary embolism.)

Nontraumatic coma, while always serious, has a better outcome in some diseases than in others. About 30% of patients with hepatic and miscellaneous causes of coma recovered to a GOS of 4 or 5, three times the recovery rate of patients with vascular-ischemic neurologic injuries (subarachnoid hemorrhage, cerebral vascular diseases, and hypoxia-ischemia). The difference is explained by most of the hepatic and miscellaneous patients having reversible biochemical, infectious, or extracerebral intracranial (e.g., subdural hematoma) lesions that may have transiently depressed brain function, but nevertheless left the structure of the brain intact. By contrast, many patients with stroke or global cerebral ischemia suffered destruction of brain structures crucial for consciousness. Reflecting this difference, the metabolic-miscellaneous group of patients showed significantly fewer signs of severe brainstem dysfunction than did those with vascular-ischemic disorders. For example, corneal responses were absent in fewer than 20% of the metabolic group, but in more than 30% of the remaining patients. Furthermore, when patients with hepatic-miscellaneous causes of coma did show abnormal neuro-ophthalmo- logic signs (see below), their prognosis was as poor as that of patients in the other disease groups with similar signs.

Patients who survived medical coma had achieved most of their improvement by the end of the first month. Among the 121 patients still living at 1 month, 61 died within the next year, usually from progression or complication of the illness that caused coma in the first place.

There were seven moderately disabled patients who improved to a good recovery. Of 39 patients severely disabled at 1 month, nine later improved to a good recovery or moderate disability rating. At the end of the year, three patients remained vegetative and four severely disabled. While current patients may have a greater chance of survival with modern therapies, it is unfortunately not likely that they would have a significantly different natural history after 1 month, suggesting that the data from this series remain relevant.

The outcome was influenced by three major clinical factors: the duration of coma, neuroophthalmologic signs, and motor function. Of somewhat lesser importance was the course of recovery; a history of steady improvement was generally more favorable than was initially better function that remained unchanged for the next several days. Only one patient who remained in coma for a week recovered to a GOS of 5 at 1 month. Conversely, the earlier consciousness returned, the better was the outcome. Among patients who awakened and regained their mental faculties within 1 day, nearly one-half achieved a GOS of 4 or 5, compared with only 14% among those who at 1 day remained vegetative or in coma. Among patients who survived three days, 60% who were awake and talked made a satisfactory recovery within the first month, compared with only 5% of those still vegetative or in a coma. Contrary to initial expectations, no consistent relationship emerged between age and prognosis either for the study as a whole or for individual illnesses. The sex of the patient had no apparent influence on outcome. Coma of 6 hours or more turned out to be such an innately serious state that in most cases it became difficult to predict accurately who would do well (i.e., make a moderate or good recovery) much before the third day of illness. By contrast, about one-third of patients destined to achieve a GOS of 1 or 2 showed overwhelmingly strong indications of that outcome on admission.

As Table 9–5 immediately discloses, a potentially bewildering amount of early clinical information showed an association with outcomes in patients with medical coma. To reduce this mass of data to manageable proportions and thereby sharpen the accuracy of prognosis for physicians working at the bedside, Levy and associates32 constructed logic diagrams based on the actual outcomes of patients showing

certain signs at various time intervals (Figure 9–3). In constructing these decision trees, which give an estimate of prognosis based on actual experience, the most important concern was to be sure that signs denoted as implying a GOS prognosis of 1 or 2 described virtually no one (less than 3%) who achieved an ultimate GOS of 4 or 5. One can immediately recognize that an inaccurate estimate of prognosis could result in the curtailing of potentially useful treatment, a step to be avoided at almost all costs. Chi-square testing of the decision criteria given in Figure 9–3 against the actual findings and outcomes of the 500 patients indicates that all the discriminations have an accuracy of association with p < 0.001.

Even as early as 6 hours after the onset of coma, clinical signs identified 120 patients as having virtually no chance of regaining independent function (Figure 9–3A). Only one of 120 patients achieved even a brief functional return equivalent to a moderate level of disability, a 19-year-old woman with cardiac arrest associated with uremia who briefly improved before dying the following week. The remaining 380 patients could be divided on the basis of their clinical findings into groups with relatively better prognoses, the best having a 41% chance of attaining independent function. Similar discrimination was possible at 1 day (Figure 9–3B). At this time, 29 of the 87 patients with the poorest prognosis survived 2 more days and 10 survived at least a week; on the other hand, 24 patients could be predicted as recovering to an outcome of GOS 4 or 5, and two-thirds of these actually regained independent function. With the further passage of time (Figure 9–3C, D), success at identifying patients with a prognosis of GOS 4 or 5 improved even further.

Subsequent prospective evaluations of outcome in medical coma have generally confirmed the accuracy of these original studies. A prospective cohort study of 596 patients with nontraumatic coma identified five clinical variables that predicted 2-month mortality (Table 9–6).33 This population reflected mostly patients in coma following cardiac arrest (31%), cerebral infarction, or intracerebral hemorrhage (36%) (other etiologies included subarachnoid hemorrhage, sepsis, neoplasm, and infections). Patients with four of five clinical findings of abnormal brainstem responses (absent pupillary responses, absent corneal

reflexes, and absent or dysconjugate roving eye movements), absent verbal response, absent withdrawal to pain, age older than 70 years, or a creatinine of greater than or equal to 1.5 mg/dL (132 mmol/L) had a 97% mortality at 2 months. An age-related worsening of prognosis was identified in distinction from the Plum and Levy

study,4 but may be partly confounded by comorbid systemic conditions. A prospective study of 169 patients older than 10 years with nontraumatic coma admitted to an intensive care unit found that 75% of those with hypoxic or ischemic injuries had died or remained comatose at 2 weeks34 (Table 9–7).

A 500 PATIENTS at ADMISSION

B 387 PATIENTS at 1 DAY

How is one to act on these predictions? The physician, together with the patient’s health care proxy and family, must decide. A patient who has been in coma for 6 hours from a known nonpharmacologic cause, without pupillary responses or eye movements, has essentially no chance of making a satisfactory recovery. Knowledge of this prognosis will deter many physicians from applying heroic and extraordinary measures of care. (Nevertheless, such patients may be candidates for well-controlled new or unconventional treatments, as conventional therapy offers such a dismal out-

come.) Conversely, a seriously ill and still unresponsive patient who shows normal eye or motor signs at 1 to 3 days following cardiac arrest has about a 30% chance of recovering to a GOS of 4 or 5. This information should provide strong encouragement to intensive care staff members. The latter individuals often feel they are working blindly and with little chance of success when caring for patients who have suffered brain injury. Knowledge of a potentially favorable outcome greatly improves morale and the associated level of care.

CARDIOPULMONARY ARREST/

HYPOXIC-ISCHEMIC

ENCEPHALOPATHY

Several large studies have examined outcome in coma specifically following cardiac arrest. Data from 942 patients prospectively enrolled in the Brain Resuscitation Clinical Trials35 (circa 1979 to 1994) demonstrated that loss of any of the cranial nerve reflexes following cardiac arrest significantly predicted poor outcome. Booth and associates2 reviewed all available large studies of coma following cardiac arrest from 1966 to 2003 to assess the precision and accuracy of the physical examination in prognosis. They found that five clinical signs were strongly predictive of death, VS, or severe disability (GOS 1, 2, or 3): absent corneal reflexes, absent pupillary reflexes, absent withdrawal to painful stimuli, absent motor response at 24 hours, and absent motor response at 72 hours. Notably, no clinical examination finding strongly predicted a GOS of 4 or 5. In

the aggregate, the data shown in Table 9–8 support the algorithms shown in Figure 9–3 and add further details as well as time points. It should be recognized that the Booth et al. predictors aggregate severely disabled outcomes (GOS 3) with outcomes of death or permanent VS (GOS 1 and 2). Thus, careful explanation of the predicted outcomes is required if the physician uses these data to counsel families, as choices concerning severe disability may differ widely (see family dynamics and philosophic considerations, page 379).

ELECTROPHYSIOLOGIC TESTING IN HYPOXIC-ISCHEMIC ENCEPHALOPATHY

Although the physical examination gives a strong prediction of poor outcome, it does not accurately assess the extent of cortical injury. Electrophysiologic testing adds valuable data. SSEPs provide the best predictors of poor

outcomes and are relatively insensitive to metabolic derangements and drug effects.36 Bilateral loss of primary cortical somatosensory responses has been repeatedly confirmed to have a 100% specificity for outcomes no better

than a permanent VS following anoxic injuries.37,38 A recent review23 found that of 176

patients with absent bilateral primary somatosensory responses (N20), none recovered past a permanent VS (Table 9–9). The robust correlation of bilateral loss of SSEPs and poor outcome reflects a close connection with the underlying degree of anoxic injury as indicated by autopsy studies.39 Of 10 patients examined at autopsy who had SSEP measurements obtained within 48 hours of cardiac arrest, all seven with bilateral absence of the SSEPs had extensive anoxic-ischemic destruction of the cerebral cortex (with acute ischemic changes in patients with short survival, and frank necrosis of the pseudolaminar type in those patients with longer survival times). Two additional patients (one with delayed SSEPs and one with normal-latency SSEPs) showed patchy neuronal loss in the cerebral cortex. Importantly, although an index of better outcomes, preservation of normal-latency SSEPs following cardiac arrest is not a definite predictor of positive outcomes. Death or vegetative outcomes may occur in as many as 40% of cases where a normal N20 response is measured.38

Other electrophysiologic techniques, including EEG, brainstem auditory-evoked responses (BAERs), and transcranial motor-evoked responses, also have predictive value (see 23 for detailed review). EEG patterns are often suppressed early following anoxic injuries and a variety of signal abnormalities22 correlate with poor outcomes; these include burst suppres-

sion, alpha-theta patterns, and generalized suppression or periodic patterns. The BAER test can identify severe brainstem injury, but does not address the outcome of cerebral cortical injury. Preservation of longer latency auditoryevoked responses that involve contributions from larger cerebral cortical networks may predict recovery of cerebral function with greater specificity. Both a late auditory response (N100) and the mismatch negativity (MMN) response have value in predicting outcome from coma following anoxic injury.40 Other longer latency evoked responses such as the P300 and N400 have also been studied (see 22 for review).

PITFALLS IN THE EVALUATION

OF COMA FOLLOWING

CARDIOPULMONARY ARREST

Although prognosis in coma following cardiopulmonary arrest is generally accurate, pitfalls do exist. The following case illustrates an extreme, although not isolated, example from the literature.41

Patient 9–1

A 25-year-old asthmatic man collapsed at home and stopped breathing. The patient received cardiopulmonary resuscitation (CPR) from a family member for 6 minutes until emergency medical personnel arrived to find the patient without respiratory effort or palpable pulse. Electrocardiogram (ECG) showed a rate of 24 bpm; CPR and tracheal intubation were performed. Three minutes later the pulse was 107 bpm and spontaneous respirations were noted. Initial GCS was 3. In the

emergency room the patient was unresponsive with dilated pupils that were responsive to light; spontaneous decorticate posturing was noted. The patient was sedated with propofol, given atracurium, and transferred to the intensive care unit (ICU). In the ICU the patient required mild pressor support and was noted to exhibit frequent myoclonic jerks of the head and all four limbs. EEG recordings revealed generalized status epilepticus. Theophylline levels were within the normal therapeutic range. Seizures were uncontrolled with phenytoin, midazolam, clonazepam, valproate, and MgSO4, so that thiopental infusion producing burst suppression was required. After cessation of the thiopental drip, generalized alpha frequency activity was noted. On the sixth day the patient was extubated, given a Do Not Resuscitate (DNR) status, and transferred to the general neurology floor still with a GCS of 3. He subsequently gradually improved and had a GCS of 10 by day 16 with the recovery of head nodding and verbalization. His GCS reached 15 by the 19th week following the respiratory arrest. While EEG examinations showed progressive improvements, the patient continued to exhibit frequent myoclonic jerks and epileptiform activity despite multiple antiepileptic medication trials. Ultimately, this patient regained independent function.

This patient’s case highlights the potential complexity of prognosis in coma even in circumstances that appear to predict poor outcome following cardiac arrest and severe hypoxic injury. A retrospective review of the history suggests several points for consideration. While the patient’s young age, initial presence of pupillary light responses, and early return of spontaneous respiration were positive predictors, the presence of myoclonus and seizures with no history of epilepsy suggested severe hypoxic injury. As reviewed above, postanoxic myoclonus usually predicts a dismal prognosis,42 but this is not invariably the case.43 The early sedation and paralysis of the patient due to the seizure activity may have masked improvement in level of consciousness within the first 6 hours, and the extensive use of different antiepileptic medications may have mimicked the pattern of alpha coma, a finding that otherwise carries a greater than 90% mortality in the setting of anoxic injury.44

This patient demonstrates the limitations of obtaining complete information from events

in the field and unequivocal separation of the effects of primary injury versus potential confounds introduced by methods of treatment. A pulseless patient may still have some undetected circulatory activity, or have lost perfusion just prior to evaluation, making accurate estimate of duration of hypoxia problematic. A similar case involving seizures and myoclonus following a cardiac arrest has been reported, withlateimprovementonday16afterremaining at a GCS of 5 until that point.45

Finally, a postictal state can severely depress brainstem function, and tonic seizures can simulate flexion or extension posturing, whereas single epileptic jerks can be difficult to distinguish from myoclonus. Cardiac arrest from a seizure-induced cardiac arrhythmia46 can further complicate the picture.

Vascular Disease

STROKE

Prognosis in coma following stroke depends on the arterial territory affected by the stroke that produces bilateral hemispheric dysfunction as detailed in Chapter 4. Wijdicks and Rabinstein47 surveyed the literature of prognostic factors for severe stroke from 1966 to 2003. They found no evidence-based studies better than class III to indicate prognosis, although several suggestive clinical and radiologic features were identified. Large proximal vessel occlusions causing diffuse hemispheric edema and midline shift carry a grave prognosis with a nearly 90% mortality when the shift of the septum pellucidum was greater than 12 mm.48 Patients with coma caused by acute basilar occlusions may recover49 (see Chapter 2), whereas those with coma due to hypertensive pontine hemorrhages usually do not.50

SUBARACHNOID HEMORRHAGE

Coma resulting from spontaneous subarachnoid hemorrhage (SAH) has a grave prognosis. The World Federation of Neurological Surgeons (WFNS) grades SAH using the GCS51 (see also 52) (Table 9–10). Although brief loss of consciousness is common, coma is a relatively uncommon sign in patients who reach the hospital with SAH; two-thirds present with WFNS grade III examinations or better.