Diagnostic classification of stroke, especially Lacunes.

Evolution and Testing of the Lacunar Hypothesis" by Bamford and Warlow in this issue of Stroke reviews many important aspects of these unique pathologic vascular lesions. I concur with the authors' emphasis on the need to maintain homogeneity in the classification of stroke conditions, specifically lacunes. A stroke classification scheme would ideally be based on 1) risk factor profile such as hypertension, cardiac disease, and diabetes mellitus; 2) clinical features such as transient ischemic attacks (TIAs), time of onset of deficit, pattern of evolution of the deficit, and abnormal neurologic findings; 3) neuroimaging studies such as computed tomography (CT) and magnetic resonance imaging (MRI); 4) vascular anatomy determined by noninvasive carotid studies and angiography; 5) functional studies such as singlephoton emission computed tomography (SPECT), positron emission tomography (PET), and xenon blood flow; 6) natural history such as clinical course and level of recovery; and 7) pathologic anatomy. This scheme of stroke classification would make it possible to devise therapeutic trials involving homogeneous groups of patients and would reduce the risk of two types of errors. One type of error occurs when a specific therapy is reported to be effective because the treated stroke population contained many patients with a benign natural history who would have done well even if they had received no specific treatment. The second type of error occurs when a therapy is reported to be ineffective because the treated stroke population was so large and diverse that one or more homogeneous subgroups in which the treatment was effective were not identified. Many reports have described the "lacunar syndromes" as having nonischemic or even nonvascular etiologies including intracerebral hemorrhage, subdural hematoma, neoplasm, and infectiousinflammatory conditions. The list of nonischemic etiologies continues to proliferate, demonstrating the need for careful neurodiagnostic evaluation of each patient. It is crucial to exclude patients with