Decompressive craniectomy for large, life-threatening supratentorial infarcts has been considered a controversial procedure for a long time. In the past 2 decades, authors of numerous single-center series have documented improved survival after decompressive craniectomy compared with maximal medical therapy.2 However, the procedure is not universally accepted. Detractors have been most vocal about the lack of Level A evidence and the danger that surgery could lead to a high proportion of survivors with severe disability. This controversy has sparked a few randomized clinical trials, and a landmark report of a pooled analysis of 3 such trials has answered some of the questions raised by critics.5 Among patients younger than 60 years old with a large middle cerebral artery infarct treated within 48 hours of stroke onset, the mortality rate decreased from 78% in the medically treated group to 29% in the decompressive craniectomy– treated group. This difference was statistically signifi cant. The authors of this report also concluded that 2 pa tients must be treated to prevent 1 death regardless of functional outcome. Four patients must be treated for each patient with a modified Rankin Scale score of 3 or less (a score most would agree corresponds to a decent quality of life). Despite these “high-quality” sci entific data and Level A evidence, several questions remain unanswered. In particular, is there a “neuroprotective” role for therapies (including decompressive cra niectomy) aimed at decreasing the secondary insult from vasogenic edema? In this issue of the Journal of Neurosurgery, Walberer and coworkers report their findings from an experimental model of reversible middle cerebral artery occlusion. These authors aimed to study the effects of vasogenic edema on infarct volume and func tional outcome. To eliminate the space-occupying ef fects of postinfarction edema, a bilateral decompressive craniectomy was prophylactically performed before reversible cerebral ischemia; the control group consisted of sham-operated animals. Infarct size was measured on MR images obtained 5 and 24 hours after ischemia on set. Ischemic lesions were consistently smaller in the craniectomy group at both the 5and 24-hour time points. Clinical scores were also significantly bet ter and the midline shift was less in rats that had un dergone prophylactic craniectomy. These observations lend further support to the notion that vasogenic edema aggravates the extent of infarction. Thus, therapies targeting vasogenic edema and its deleterious effects can prove neuroprotective. Hypothetically, vasogenic edema around the ischemic tissue causes regional compression of leptomeningeal vessels providing critical collateral blood flow to the penumbra. Indeed, experimental studies have suggested that decompressive craniectomy im proves cortical perfusion by increasing local cerebral blood flow through leptomeningeal collateral vessels.1 In the study by Walberer et al., animals were killed after 24 hours. Note, however, that postinfarction edema progresses well beyond that time point, and we don’t know for sure if the protective effect demonstrated in their ex perimental setting is maintained beyond 24 hours. An extension of this study might address this issue in the future. Can we use these observations to propose an ultraearly decompressive craniectomy and very aggressive intracranial pressure–targeted therapies in patients with acute cerebral infarction to improve functional out come? The easy answer is that randomized clinical trials are required. Experience shows that such trials are very difficult, expensive, and impractical to perform and often do not provide definitive answers. The decision to proceed with a decompressive craniectomy in a patient with a life-threatening infarct is always difficult because of the significance and impact (on the quality of life of the patient as well as the caregiver[s]) of the residual neu rological deficits. Nonetheless, when craniectomy is con sidered, data from the following and other studies3,4 sug gest that early is better.
Read full abstract