Prospective risk of hemorrhage in patients with vertebrobasilar nonsaccular intracranial aneurysm

Transcription

1 J Neurosurg 101:82 87, 2004 Prospective risk of hemorrhage in patients with vertebrobasilar nonsaccular intracranial aneurysm KELLY D. FLEMMING, M.D., DAVID O. WIEBERS, M.D., ROBERT D. BROWN JR., M.D., MICHAEL J. LINK, M.D., HIROFUMI NAKATOMI, M.D., JOHN HUSTON III, M.D., ROBYN MCCLELLAND, PH.D., AND TERESA J. H. CHRISTIANSON, B.S. Departments of Neurology, Neurologic Surgery, and Diagnostic Radiology; and Division of Biostatistics, Mayo Clinic, Rochester, Minnesota Object. Nonsaccular intracranial aneurysms (NIAs) are characterized by dilation, elongation, and tortuosity of intracranial arteries. Dilemmas in management exist due to the limited regarding the natural history of this disease entity. The objective of this study was to determine the prospective risk of subarachnoid hemorrhage (SAH) in patients with vertebrobasilar NIAs. Methods. All patients with vertebrobasilar fusiform or dolichoectatic aneurysms that had been radiographically demonstrated between 1989 and 2001 were identified. These patients medical records were retrospectively reviewed. A prospective follow-up survey was sent and death certificates were requested. Based on results of neuroimaging studies, the maximal diameter of the involved artery, presence of SAH, and measurements of arterial tortuosity were recorded. Nonsaccular intracranial aneurysms were classified according to their radiographic appearance: fusiform, dolichoectatic, and transitional. Dissecting aneurysms were excluded. The aneurysm rupture rate was calculated based on person-years of follow up. Predictive factors for rupture were evaluated using univariate analysis (p 0.05). One hundred fifty-nine patients, 74% of whom were men, were identified. The mean age at diagnosis was 64 years (range years). Five patients (3%) initially presented with hemorrhage; four of these patients died during follow up. The mean duration of follow up was 4.4 years (692 person-years). Nine patients (6%) experienced hemorrhage after presentation; six hemorrhages were definitely related to the NIA. The prospective annual rupture rate was 0.9% (six patients/692 person-years) overall and 2.3% in those with transitional or fusiform aneurysm subtypes. Evidence of aneurysm enlargement or transitional type of NIA was a significant predictor of lesion rupture. Six patients died within 1 week of experiencing lesion rupture. Conclusions. Risk of hemorrhage in patients harboring vertebrobasilar NIAs is more common in those with evidence of aneurysm enlargement or a transitional type of aneurysm and carries a significant risk of death. KEY WORDS dolichoectatic aneurysm aneurysm fusiform aneurysm subarachnoid hemorrhage N ONSACCULAR intracranial aneurysms are characterized by dilation, elongation, and tortuosity of an artery. They have been designated by several names (Table 1); however, few of these terms have been well-defined pathologically or radiographically. In addition, these are morphological terms and do not necessarily reflect any underlying origin. Small patient numbers and a lack of clinically useful definitions of the spectrum of nonsaccular aneurysms have limited attempts to further define an NIA and its natural history. Symptoms related to posterior circulation NIAs most commonly result from ischemia or compression of surrounding structures. Little is known about the long-term risk and factors predicting ischemia, compression, and rupture, which has led to dilemmas in case management. Current data are based on small series of patients, and available studies are difficult to compare because of the variability in definitions, radiographic results, and treatment. Only case Abbreviations used in this paper: BA = basilar artery; CT = computerized tomography; ECA = external carotid artery; MR = magnetic resonance; NIA = nonsaccular intracranial aneurysm; PCA = posterior cerebral artery; SAH = subarachnoid hemorrhage; VA = vertebral artery. 82 reports regarding hemorrhage exist and no annual risk of aneurysm rupture has been described. The objective in this study was to determine the prospective hemorrhage rate in a large cohort of patients with posterior circulation NIAs. Clinical Material and Methods Patient Selection The records of all patients seen at the Mayo Clinic between 1989 and 2001 with the radiographically based diagnosis (cerebral angiography, MR imaging, MR angiography, or CT) of dolichoectatic, fusiform aneurysm, or enlarged or tortuous VA or BA were reviewed. Patients with arteriographic evidence of a dissecting aneurysm, 10 such as an irregular stenosis proximal to the aneurysm, or evidence of a false lumen or an intimal flap, were excluded, because these aneurysms are known to have a distinctive natural history. 10,11 Data Collection The primary investigator reviewed all medical records. Extensive demographic and clinical information was col-

2 Rupture risk in vertebrobasilar nonsaccular intracranial aneurysm TABLE 1 Synonyms for vertebrobasilar NIA fusiform aneurysm dolichoectasia ectasia S-shaped aneurysm wandering BA tortuous BA cirsoid aneurysm megadolichovascular malformation megadolichobasilar artery dolichomegavertebralis anomaly la dolicho mega basilare aneurysmal malformation lected from medical records and transferred to a standardized form. Each patient s age, sex, presenting symptoms, clinical diagnosis, cerebrovascular risk factors, and initial treatment were noted. Ischemia was ascribed to the NIA only if the criteria of Cohen and colleagues 2 were fulfilled. Results of any follow-up evaluation were obtained from the medical record if the patient had been seen within the last 6 months or from letter and telephone correspondence with the patient. Recorded follow-up information included the following: recurrent ischemic events, intracranial hemorrhage and SAH, compressive symptoms, and death. Hemorrhages were categorized as those likely caused by the NIA or those with an unclear association to the lesion. Interim medical and surgical interventions were recorded. Autopsy information was obtained when available, and death certificates were requested for all patients who had died since the initial evaluation. Radiographic Data Radiological studies were reviewed by the primary investigator and by personnel from the neuroradiology department. An NIA was defined as any intracranial arterial dilation greater than 1.5 times normal 9 without a clearly definable neck. The category of NIA was further divided into three subtypes (fusiform, dolichoectatic, or transitional) based on either cross-sectional imaging or cerebral angiography study results. Some of these aneurysms could not be subtyped based on available imaging studies (cross-sectional only) and thus were labeled indeterminate. The characteristics for each subtype are featured in Fig. 1. The only other classification system for fusiform aneurysms, proposed by Mizutani and colleagues, 11 is based on pathological and not radiographic appearance. Furthermore, that series was limited to the study of nonatherosclerotic aneurysms. The aneurysm location and size, degree of arterial tortuosity, and presence of intraluminal thrombus were documented. The maximal diameter (anteroposterior and mediolateral) of the aneurysm was measured both on cerebral arteriography and cross-sectional imaging results when available. Tortuosity was graded in the BA by using CT criteria described by Smoker, et al., 15 and modified for MR imaging by Giang, et al. 5 Evidence of hydrocephalus, cerebral infarction, and compression of surrounding structures was recorded. When available, CT scans documenting SAH were reviewed. Radiographic data obtained at initial diagnosis and at subsequent follow-up visits, if available, FIG. 1. Classification of NIAs. Upper Left and Right: Illustration and angiogram demonstrating aneurysm dilation 1.5 times the normal diameter without a definable neck, involving a portion of an arterial segment (either the VA or BA) with any degree of tortuosity. Center Left and Right: Illustration and arteriogram demonstrating uniform dilation of an artery greater than 1.5 times normal, involving the entire BA, VA, or both with any degree of tortuosity. Lower Left and Right: Illustration and arteriogram revealing a uniform aneurysm dilation of an entire arterial segment greater than 1.5 times normal, involving the VA, BA, or both with a superimposed dilation of the involved arterial segment. were collected. We noted any enlargement of the aneurysm greater than 2 mm in the anteroposterior diameter, which was determined based on cross-sectional images if serial images were available or the neuroradiologist s interpretation. Statistical Analysis Descriptive statistics including means, standard deviations, and frequencies were calculated for patient characteristics and presenting symptoms. A prospective hemorrhage rate was determined based on the number of hemorrhages that had occurred during the follow-up period divided by the number of prospective person-years of follow up. Fac- 83

3 K. D. Flemming, et al. TABLE 2 Comorbid conditions in 159 patients with NIA Feature No. (%) hypertension 109 (69) diabetes 16 (10) hyperlipidemia 63 (40) coronary artery disease 44 (28) peripheral vascular disease 11 (7) current or former smoker 82 (52) abdominal aortic aneurysm* 21 (13) tortuous aorta 63 (40) * Forty-eight patients had undergone abdominal studies; 111 had not. Based on results of chest x-ray films. Note that 120 patients had undergone chest x-ray studies, whereas 39 had not. TABLE 3 Initial presentation and treatment of patients with prospective aneurysm rupture* Age Treatment Case (yrs), Symptoms at on Initial No. Sex Presentation Presentation 1 71, M 1997, gradual progressive compres- VA occlusion, sive pontine syndrome; 5/1998, warfarin acute brainstem stroke; 7/1998, death from aneurysm rupture 2 49, M 1990, brainstem stroke; 11/1996, attempted VA ocrecurrent stroke; 9/1997, acute clusion; warunresponsiveness & aneurysm farin & aspirin rupture 3 60, M 4/1997, brainstem stroke; 1998, warfarin gradual progressive compressive symptoms during the year; 10/22/1998, recurrent brainstem stroke; 11/4/1998, third brainstem stroke; 11/10/1998, sudden loss of consciousness due to SAH 4 66, F 1989 & 4/1998, vertebrobasilar warfarin distribution stroke; 9/1998, developed compressive pontine syndrome; 1/1999, ruptured 5 63, M 3/2002, transient diplopia w/ aspirin imaging evidence of mild compression; 8/2002, sudden headache & unresponsiveness 6 71, M 6/1989, compressive symptoms; aspirin 12/1990, sudden ataxia & nystagmus w/ negative CSF; sudden death 2 days later w/ evidence of rupture on autopsy 7 40, F abrupt headache w/ evidence of BA warfarin & thrombus; 1998, brainstem stroke clopidogrel following angiography; developed significant headaches & possible compressive symptoms; rupture during attempted surgery 8 65, M 4/1998, brainstem TIA; 10/1999, warfarin & sudden headache & confusion due aspirin to intraventricular hemorrhage 9 63, F developed visual defect from carotid warfarin & cavernous aneurysm, BA NIA ECA PCA discovered at same time; 1990, lt bypass pontine stroke; bypass procedure; SAH post-bypass * CSF = cerebrospinal fluid; TIA = transient ischemic attack. Excluded from calculation of hemorrhage risk. TABLE 4 Radiographic data in patients with prospective aneurysm rupture* Aneurysm Diameter (mm) Aneu- Initial Initial Serial rysm En- Case Cross- Angiog- MR Aneurysm largement No. Section raphy Imaging Type Observed transitional no 2 NA NA 39 transitional yes 3 NA 7 31 transitional yes 4 NA NA NA NA no serial study transitional yes NA yes fusiform no transitional no NA transitional no serial study * NA = not available. Excluded from calculation of hemorrhage risk. tors predictive of aneurym rupture were evaluated using univariate analysis (p 0.05). Proportions were compared with the aid of the Fisher exact test, with significance set at a probability level less than Results We identified 159 patients (74% men) with a radiographically based diagnosis of dolichoectatic or fusiform aneurysm, or enlarged or tortuous BA and/or VA. The mean age at diagnosis was 64 years (range years). Demographic features and medical comorbidities in this group are listed in Table 2. Patients most commonly presented because of symptoms associated with the NIA s mass effect (22%), related ischemic stroke (28%), or hemorrhage (3%). Forty percent of patients had unrelated symptoms and 7% had symptoms with an unclear but possible relationship. In 81% of patients, neuroimaging results were available for review within 6 months of the first diagnosis. In the remaining patients, an initial diagnosis was determined by personnel at another institution and any film available for review had been obtained more than 6 months after initial diagnosis. According to radiographic results, 40% of patients had an NIA involving the BA only; 52%, both the VA and BA; and 8%, the VA only. The aneurysms consisted of 14% fusiform, 45% dolichoectatic, 20% transitional, and 21% indeterminate types. The mean initial (within 6 months of diagnosis) anteroposterior diameter of all NIAs involving the BA segment was 10.9 mm (range 5 35 mm). Eightytwo patients had serial imaging studies. The mean length of follow up was 4.4 years (range years) or 692 person-years. There were 610 personyears of follow up counted at the time of death or surgical/ endovascular intervention. Sixty-five patients (41%) died during follow up. The median survival was 7.8 years. Death was aneurysm-related in 45% of patients and was most commonly caused by cerebral infarction or thrombosis. An additional 8% of patients had death certificates with listed diagnoses that may or may not have been related to the aneurysm (for example, stroke without details of localiza- 84

4 Rupture risk in vertebrobasilar nonsaccular intracranial aneurysm tion). Thirty-eight percent of patients died of causes completely unrelated to the NIA, and in 9% death certificates were unavailable. Twenty-four patients underwent 26 of the following surgical or endovascular procedures: occlusion of a single VA (five patients), bilateral VA occlusion alone (two patients), bilateral VA occlusion with ECA PCA bypass (three patients), attempted but failed VA trial occlusions (three patients), trap with or without bypass (three primarily VA aneurysms, two BA aneurysms, and one aneurysm of the vertebrobasilar junction), placement of a ventriculoperitoneal shunt for hydrocephalus (five patients), and microvascular decompression for trigeminal neuralgia (one patient). Nine patients (6%) experienced hemorrhage after their initial presentation; six hemorrhages were definitely related to the NIA, as noted on the death certificates, autopsy records, or imaging findings. The three remaining hemorrhages had an unclear relationship to the NIA or aneurysm rupture occurred on intervention. One patient had two saccular aneurysms in addition to the NIA, which ruptured within days of an ECA PCA bypass. Another patient had intraventricular hemorrhage with an unclear relationship to the NIA. The third patient experienced SAH during attempted aneurysmorrhaphy. Thus, the annual risk of hemorrhage was 0.9% (six patients/692 person-years). The Kaplan Meier 1-year estimate of an aneurysm rupture was 1.64%. One person (Case 1; Table 3) had undergone coil occlusion of the VA 1 year before lesion rupture. Therefore, if we considered patients at the time of death or intervention (surgical/endovascular procedure excluding ventriculoperitoneal shunt placement), the annual risk of hemorrhage would be 0.8% (five patients/610 person-years). All patients with hemorrhage postdiagnosis had previously demonstrated symptoms of either cerebral infarction or compression due to mass effect (Table 3). The time from initial diagnosis to lesion rupture varied from 22 days to 10 years. Seven aneurysms ruptured within 1.5 years of initial diagnosis. Two ruptured at 10 years from initial diagnosis (Cases 7 and 8). According to radiographic study results, all nine patients had BA NIAs, and six of the aneurysms were the transitional type (Table 4). In the six patients with definite prospective hemorrhage related to the NIA, evidence of enlargement according to serial imaging results (p = 0.007) and a transitional type of aneurysm (p = 0.006) were significant predictors of future lesion rupture. The annual risk of hemorrhage in patients with the fusiform or transitional type of aneurysm is 2.3% (four patients/176 person-years). An initial cross-sectional aneurysm diameter greater than or equal to 10 mm appears to be strongly related to future rupture. Initial (within 6 months of diagnosis) cross-sectional lesion diameter was ascertained in only three of the six aforementioned patients, but was more than 10 mm in all three. Furthermore, in those six patients the cross-sectional aneurysm diameters were greater than or equal to 10 mm at a point prior to rupture. Of those lesions that did not rupture during follow up, 34% were at least 10 mm in diameter. In addition, the ratio of the cross-sectional to angiographic diameter was approaching 1 in most patients, thus indicating that few of those ruptured aneurysms had significant thrombus. Of the six patients with definite prospective hemorrhages, all died within 1 week of the lesion rupture. Representative images obtained in selected patients are featured in Figs. 2 and 3. FIG. 2. Case 5. This patient was a 63-year-old man who had presented with brief diplopia following a motor vehicle accident. His initial angiogram (upper) reveals a diffusely enlarged BA with a superimposed aneurysm in the proximal segment. Axial cross-sectional MR images (center) demonstrating evidence of enlargement during a 5-month period. During that time, the patient developed dysphagia, dysarthria, and a mild left-sided weakness. Sagittal MR images (lower) demonstrating enlargement of the aneurysm, which causes significant compression of the pons and cervicomedullary junction. The aneurysm ruptured 1 day after the second MR image was obtained. Discussion The annual prospective risk of hemorrhage from a vertebrobasilar NIA is 0.9% overall, but can be as high as 2.3% per year in patients with the transitional or fusiform types of aneurysms. No prior estimates of risk have been made because only case reports of hemorrhage exist in the literature and textbooks simply state that hemorrhage is uncommon. Predictors of future aneurysm rupture based on the results of this study include evidence of aneurysm enlargement and transitional morphology. An aneurysm diameter of at least 10 mm was also strongly indicative of future rupture, but did not reach statistical significance because of incomplete data on initial imaging (initial cross-sectional imaging within 6 months of diagnosis). Intuitively, a large 85

5 K. D. Flemming, et al. FIG. 3. Case 1. Angiogram obtained in a 70-year-old man who had presented with a several-year history of gradual, progressive gait decline. He subsequently suffered a pontine ischemic infarction. The featured aneurysm ruptured 2 months following the pontine infarction. dilated vessel or an enlarged vessel with a thin wall would most likely rupture. Mathematical models of aneurysm size in relation to blood pressure also support this. 6 In other case reports of patients with hemorrhage due to vertebrobasilar NIA, the diameters were also typically 10 to 20 mm. 4,7,8,13,14 It remains unclear whether there is an absolute aneurysm size that leads to rupture or if rupture relates to the rate and degree of enlargement. In four of six patients with definite prospective hemorrhage, we found evidence of lesion enlargement over time. Two patients in this study (one whose lesion prospectively ruptured and the other whose aneurysm did not rupture) had evidence of enlargement over a period of less than 6 months. Similarly, Mizutani and colleagues 11 describe three patients with enlarging dolichoectatic dissecting aneurysms that ultimately ruptured. Although our calculations produced values of no statistical significance, the absence of thrombus in patients whose lesions later ruptured is an interesting trend. Among other patients in the cohort with BA diameters as large as 30 and 40 mm, the lesions did not rupture but had a significant amount of thrombus, thus implying a protective effect. All patients whose aneurysm prospectively ruptured had presented with symptoms of compression or ischemia prior to hemorrhage, making management decisions difficult. Many of these patients presented with recurrent posterior circulation ischemia and were placed on a regimen of warfarin as well as other antiplatelet agents. The risks of recurrent ischemia must be carefully weighed against the risk of rupture. Cerebral ischemia and BA thrombosis were the most common causes of death among the entire cohort. Furthermore, the annual risk of posterior circulation ischemia is approximately 5%, that is, higher than the rate of rupture (unpublished data). In three patients, ischemic symptoms immediately preceded aneurysm rupture. The presentation of ischemic symptoms immediately prior to rupture has been previously reported. Several investigators 7,8,12,13 have also described patients with vertebrobasilar NIA presenting with ischemic symptoms; these patients received anticoagulant agents and subsequently experienced aneurysm rupture within 72 hours. Why would ischemic symptoms herald impending rupture? First, it is possible that the acute symptoms resulted from arterial compression due to the enlarging aneurysm, which has reached a critical size, although this has never been demonstrated on angiography, imaging, or pathological studies. Second, the ischemia may have resulted from thromboembolism. Patients with the largest diameter of aneurysms and highest degree of arterial tortuosity are at a higher risk for thrombus formation. Smoker and colleagues 15 found that patients with compressive symptoms and ischemia had larger mean aneurysm diameters (10 mm) compared with those with cranial nerve compression symptoms alone (4.5 mm). Thus, larger NIAs may be prone both to rupture and to cause stroke. Another possibility involves intimal dissection. Dolichoectatic dissecting aneurysms as described by Mizutani, et al., 11 are defined by fragmentation of the internal elastic lamina combined with multiple dissections of the thickened intima. Thrombus formation was common at the site of intimal dissection. This may indicate that ischemic strokes result from thrombus related to the intimal dissection and that intimal dissection may cause fragility of the vessel wall and lead to rupture. Treatment choices are limited in patients with vertebrobasilar NIA. Most often, patients are treated medically with antiplatelet or anticoagulant agents because stroke recurrence is the leading cause of morbidity and death. When patients have recurrent symptoms despite medical treatment or demonstrate evidence of an enlarging aneurysm together with neurological deficits, surgery should be considered. Surgical options include end-to-end anastomosis, aneurysm trapping or wrapping, or aneurysmorrhaphy. 1,3 Surgical options result in good outcomes in 57 to 65% of selected cases of BA fusiform aneurysms. Occluding the dominant VA can also be considered when aneurysms are enlarging, although little clinical and no radiographic imaging data exist to prove that this method is beneficial. In the current study, three patients who had undergone surgical intervention subsequently experienced lesion rupture. The patient in Case 1 had undergone VA occlusion, but nonetheless developed brainstem stroke and lesion rupture within 1 year. The patient in Case 7 underwent attempted aneurysmorrhaphy with resultant rupture. The patient in Case 9 had an NIA that ruptured within several days of an ECA PCA bypass. This study is limited by retrospective data collection and its inherent biases. Tertiary referral bias is also present in the study and may lead to an overestimation of the severity of outcome. Nonetheless, because of the low incidence of vertebrobasilar NIAs, we did not exclude referral patients. The measurement of NIA diameters by cerebral arteriography alone may cause underestimation of the lesion diameter due to intraluminal thrombus; however, just eight patients (5%) underwent only an angiography study. Although we tried to exclude dissecting aneurysms based on radiographic imaging results, this may not always be possible. Perhaps those vertebrobasilar NIAs that initially occurred with hemorrhage were in fact dissecting aneurysms. Mizutani and colleagues 11 describe a pathological classification of fusiform and dissecting aneurysms. A classic dissecting aneurysm pathologically reveals acute disruption of the internal elastic lamina without compensatory intimal thickening, indicating a different origin with a different natural histo- 86

6 Rupture risk in vertebrobasilar nonsaccular intracranial aneurysm ry than the so-called dolichoectatic dissecting aneurysm or atherosclerotic aneurysms. For this reason, only the prospective hemorrhages were included in the estimation of the annual rupture rate. Conclusions We provide the first useful estimate of the prospective annual risk of rupture in patients with vertebrobasilar NIAs. Although limited surgical and endovascular options exist for the treatment of this entity, the risk of cerebral ischemia is greater than that of lesion rupture. Concerns for possible rupture should be raised in symptomatic patients with the transitional type of vertebrobasilar NIA greater than or equal to 10 mm in diameter, particularly in those with evidence of a lesion enlarging over time and a lack of thrombus. References 1. Anson JA, Lawton MT, Spetzler RF: Characteristics and surgical treatment of dolichoectatic and fusiform aneurysms. J Neurosurg 84: , Cohen MM, Hemalatha CP, D Addario RT, et al: Embolization from a fusiform middle cerebral artery aneurysm. Stroke 11: , Drake CG, Peerless SJ: Giant fusiform intracranial aneurysms: review of 120 patients treated surgically from 1965 to J Neurosurg 87: , Flemming KD, Josephs K, Wijdicks EF: Enlarging vertebrobasilar dolichoectasia with subarachnoid hemorrhage heralded by recurrent ischemia. Case illustration. J Neurosurg 92:504, Giang DW, Perlin SJ, Monajati A, et al: Vertebrobasilar dolichoectasia: assessment using MR. Neuroradiology 30: , Hademenos GJ, Massoud T, Valentino DJ, et al: A nonlinear mathematical model for the development and rupture of intracranial fusiform aneurysms. Neurol Res 16: , Hayes WT, Bernhardt HM, Young JM: Fusiform arteriosclerotic aneurysm of the basilar artery. Five cases including two ruptures. Vasc Surg 1: , Hirsch CS, Roessmann U: Arterial dysplasia with ruptured basilar artery aneurysm: report of a case. Hum Pathol 6: , Huber P: Cerebral Angiography. New York: Georg Thieme Verlag, Mizutani T, Aruga T: Dolichoectatic intracranial vertebrobasilar dissecting aneurysm. Neurosurgery 31: , Mizutani T, Miki Y, Kojima H, et al: Proposed classification of nonatherosclerotic cerebral fusiform and dissecting aneurysms. Neurosurgery 45: , Pessin MS, Chimowitz MI, Levine SR, et al: Stroke in patients with fusiform vertebrobasilar aneurysms. Neurology 39:16 21, Rabb CH, Barnwell SL: Catastrophic subarachnoid hemorrhage resulting from ruptured vertebrobasilar dolichoectasia: case report. Neurosurgery 42: , Read D, Esiri M: Fusiform basilar artery aneurysm in a child. Neurology 29: , Smoker WR, Corbett JJ, Gentry LR, et al: High-resolution computed tomography of the basilar artery: 2. Vertebrobasilar dolichoectasia: clinical-pathologic correlation and review. AJNR 7: 61 72, 1986 Manuscript received May 8, Accepted in final form March 11, Address reprint requests to: Kelly D. Flemming, M.D., Department of Neurology, Mayo Clinic, 200 First Street NW, Rochester, Minnesota flemming.kelly@mayo.edu. 87