Can Craniectomy Benefit Potentially Lethal Acute Cerebral Ischemic Stroke?

abstracts & commentary

Sources: Firlik AD, et al. Relationship between cerebral blood flow and the development of swelling and life-threatening herniation in acute ischemic stroke. J Neurosurg 1998;89:243-249. Gujjar AR, et al. Mechanical ventilation for ischemic stroke and intracerebral hemorrhage: Indications, timing and outcome. Neurology 1998;51:447-451. Wijdicks EFM, Diringer MN. Middle cerebral artery territory infarction and early brain swelling: Progression and effect of age on outcome. Mayo Clin Proc 1998;73:829-836. Schwab S, et al. Early hemicraniectomy in patients with complete middle cerebral artery infarction. Stroke 1998;29:1888-1893.

Firlik and associates report on 20 patients with acute, severe, internal carotid distribution (ICD) strokes who received xenon-enhanced, computerized tomography (Xe-CT) to estimate cerebral blood flow (CBF) and its relation to outcome. The first measurement was completed within six hours of stroke onset. Fourteen patients had left hemisphere damage, and six had right. All patients showed clinical and laboratory evidence of severe ICD-territory dysfunction within a mean of 230 minutes after first symptom; at onset, 11 of the 20 had abnormal CT images.

CBF was calculated from 55-65 regions of cerebral interest on Xe-CT perfusion images. Transfemoral arteriography delivered urokinase to the region of ICD-MCA occlusion in 11 patients. All patients received heparin. Estimates of edema were deduced from midline hemisphere shift on CT scans. Displacements of less than 1 cm were considered "mild edema." "Severe edema" was considered if a more than 1 cm midline shift appeared or the ipsilateral ambient cistern was effaced. Calculated Xe-CT regional CBF, expressed in mL/100 gm/min, were considered normal if more than 20. CBF below that level was considered abnormal. Mean hemisphere CBF values on admission for the group equaled 15.2 ± 8.1 in the damaged hemisphere vs. 34.2 ± 9.2 in the non-injured side (P = 0.0005). Supplementary Xe-CT blood flow studies were obtained on 17 patients during their worst clinical condition at 36-96 hours after disease onset. (Three who had recovered were not reanalyzed.) Eight patients with "mild edema" had a mean CBF of 19 mL; three had regional blood flows above 20 mL. Nine patients with severe edema had mean values of 10.4 ± 6 mL. Six of the nine showed CBFs of 8.6 ± 9.5 mL. The remaining three had estimated values of 18 ± 9.5 mL. Overall, severe edema occurred 13.3 times more often in patients whose CBF was less than 15 mL/100gm/min compared to those with higher values.

Firlik et al note that no strong prognostic clues emerged from the first Xe-CT study. The 11 patients who received intraarterial urokinase developed no better CBF values or less edema than the nine patients not receiving such therapy (5 had partial ± 1 had full reperfusion). The critical result is that during a time when patients expressed maximal cerebral dysfunction, only one patient with CBF below 15 mL did well, whereas of the 11 who did badly, eight had a mean CBF of 15 or lower. Regrettably, mortality is not described.

Firlik et al present a discouraging note for current treatment of acute, severe MCA strokes. Neither osmotherapy, invasive delivery of urokinase to the locus of carotid-MCA obstructions, or any other therapy appeared to have any favorable effect. Blood flows below 15 mL/100gm/min in 13 patients, despite all efforts of therapies, were associated with severe brain injury in eight persons and uncal herniation in six. The biologic dilemma is obvious: Does brain swelling in stroke depend largely on inflammatory defense processes in the ischemic zones or is the enlargement due mainly to increased osmols secondary to tissue destruction and immunologic responses? The association of low blood flows with poor outcome is hardly unexpected, but its potential advantage in decision-making about craniectomized decompression will be discussed below.

Gujjar et al present a slightly different example of the frustrations that accompany efforts to revive patients with acute stroke and compare their severe stroke outcome against that occurring with intracranial hemorrhage. Over a three-year period, they applied mechanical ventilation (MV) to 44 patients with ischemic stroke (IS) in the carotid artery watershed and 124 patients with acute cerebral hemorrhage (CH). The 44 IS patients had large infarcts in the internal carotid artery (ICA) distribution, and 36 (82%) were intubated for neurologic deterioration. Twenty-four (55%) of the ischemic-MV cases died, and another 11 had a poor permanent outcome. By contrast, 70 (57%) of the intubated CH group died, and another 20 had a poor outcome. Thus, with both diseases, it appears that reactive tissue bulk of a similar size beyond that of the initial lesion either killed the subjects or produced severe, permanent brain damage. Gujjar et al’s accompanying references support this concept—some listing mortality rates as high as 96% in similar groups of patients.

Wijdicks and Diringer provide data similar to the above reports. Among 42 patients with symptoms of severe cerebral swelling associated with MCA infarcts and verified by CT scans, 33 developed impaired consciousness, impaired diencephalic function, and/or uncal herniation. All 15 patients older than 60 years died; five of seven between ages 46-60 died, but only three of 11 persons ages 18-45 died. The older patients suffered from pre-existing medical problems that might have accentuated their poor outcomes.


The above fatal outcomes of severe brain swelling secondary to internal carotid distribution cerebral infarction will come as no surprise to any neurologist. What is a new finding, however, is Gujjar et al’s finding of similar degrees of hemorrhagic expansion-irritation as it develops in strokes. This suggests that the tissue reaction in stroke may not be a direct consequence of tissue oxygen lack but, rather, to occlusion of microscopic blood flow caused by excess tissue pressure blocking capillary distributions.

Against the above background, Schwab et al update their outcome reports on early hemicraniectomy for patients who suffered acute, complete MCA infarcts.

Schwab et al have now studied 63 patients who had acute unilateral craniectomies ipsilateral to severe MCA infarction. Thirty-two of these patients prior to 1995 received surgery after an average of 39 hours following onset. Thirty-one subsequent patients were operated on a mean of 21 hours after onset. Criteria for surgery in both series of patients was younger than 70 years old with clinical and CT evidence for acute, complete CA-MCA infarcts and direct signs of impending or complete severe hemispheric brain swelling. Severe post-admission neurological worsening provided the critical step to proceed with surgery. Patients with previous, disabling neurological diseases were excluded. Mean age of total surgical cohort was ± 49 years. This was compared to a previous group of 55 severe stroke patients who were treated non-surgically but with a mean age of 56 years. Only 20% of the 63 decompressed group had left hemispheric strokes compared to 62% of the previously non-decompressed group.

The table, taken directly from Schwab et al’s communication gives outcome statistics. Surgical treatment in all such patients consisted of removing a 12 cm frontal-parietal-temporal bone flap followed by opening the dura and covering the exposed, swollen gap by cadaver dura or homologous fascia. As the reader will note, craniectomy was less frequently applied over the left semantic hemisphere in both late and early groups. Nevertheless, mortality after craniectomy appears to be reduced and, anecdotally, some patients appeared more improved than expected, given the severity of their pre-surgical state. (See Table.)

Outcome Statistics 
  Medically Treated Late Hemicraniectomy Early Craniectomy
n = 55 32 31
Hemisphere = R21, L34 R 26, L6 R 26, L5
Time to surgery = 39 h 21 h
Presurgical herniation   75% (24/32) 13% (4/31)
Mortality 78%(43/55) 34%(11/32) 16%(5/31)
Time in ICU(D) 12.6 13.3 7.4
Bartle Index 60 62.6 68.8


These four reports emphasize the cruel challenge that stalks the decisions of neurologists and the outcome of patients with severe cerebral hemispheric stroke. As all references to date indicate, neither clot dissolver, hyperventilation, injections of high osmolal mannitol, nor barbiturate coma improve outcomes of such patients. Most of them develop drowsiness soon after admission along with dense hemiplegia. Fatalities often result in the presence of the following: severe drowsiness, dense hemiplegia, CT early demonstrating hypodensity in the injured hemisphere, a hemisphere shift of more than 0.8-1.0 cm away from the midline, and age older than 50 years. Perhaps even worse, is that most of those who do survive remain permanently dependent. Possibly, successive measurements of cerebral blood flow, such as performed by Firlik et al, may predict ensuing malignant brain swelling in patients with apparently large infarcts. If so, they might add to both selection and outcome for craniectomy as indicated below.

Craniectomy to treat acute, severe stroke has been described occasionally in the literature during recent years. Most reports, however, described few cases and failed to follow a specific protocol. Craniectomy following head trauma has not shown great therapeutic promise. Accordingly, to recommend craniectomy for stroke therapy hardly seemed neurologically appropriate. Two years ago, however, Schwab et al reported on their first group of craniectomized patients (Cerebrovasc Dis 1996;6:325-329). The possible lessening of disability and death given to that first group was apparent, but the small numbers required validation against appropriate controls. Their most current 31 patients appeared to improve even more. The evidence of reducing death and improving post-stroke functional ability is encouraging but must be validated by a controlled study. Firlik et al’s report of finding specific correlations between measured blood flow and degree of danger may turn into a crucial step. In any case, a controlled study is presently being formulated in Europe and the United States. Presumably, all patients must be selected as acceptable before they are deemed appropriate to receive the procedure. It will be a challenge to adhere to who gets which arm of the study, but this is a crucial ethical step. At the moment, it seems that craniectomy may turn out to bring measurably better physical outcomes to patients suffering from severe hemispheric strokes. More important, however, will be the effect on quality of life for both patients and families.