Basilar Artery Occlusive Disease in the New England Medical Center Posterior Circulation Registry FREE

The basilar artery (BA) is the largest artery in the posterior circulation and forms the central core of this vascular territory. It is most commonly affected by atherosclerosis, and necropsy studies have found it to be one of the first and most severely involved of all intracranial arteries.¹ The BA is also affected by embolism, dissections, aneurysms, migraine, and inflammatory conditions.² Despite the wide range of underlying pathologic features, BA occlusive disease is an uncommon cause of stroke. Most reports have retrospectively described isolated cases or small series of patients.^{3 - 10} Early reports were based on necropsy studies that emphasized clinicoanatomical correlations rather than the underlying vascular lesion, correlating focal brain lesions with symptoms and signs found during life.^{4 - 6 ,11} Until recently, premortem diagnosis of posterior circulation vascular lesions required invasive techniques, such as catheter angiography, a procedure often not performed because it was considered risky and unnecessary to guide therapy. Consequently, many series included only select patients and did not reflect the true spectrum of disease.^{7 - 10} Now new treatment options, such as thrombolysis and angioplasty and noninvasive diagnostic imaging techniques promote premortem diagnosis of BA occlusive disease. However, still few prospective studies are available,^{12 - 15} and disparities remain regarding the natural history, underlying stroke mechanisms, vascular features, and prognosis of BA occlusive disease. We herein report an analysis of patients with moderate to severe stenosis or occlusion of the BA in the New England Medical Center Posterior Circulation Registry (NEMC-PCR), the largest prospective series of consecutively collected patients with posterior circulation ischemia to date.

The patients analyzed herein were identified among 407 patients collected consecutively in the NEMC-PCR between January 1, 1988, and December 31, 1996. Detailed inclusion criteria and characteristics of this registry have been previously described.¹³ Briefly, the NEMC-PCR is a prospective database of patients with symptomatic vertebrobasilar territory ischemia. All patients were evaluated at the New England Medical Center in Boston, Mass, by 1 of 3 stroke neurologists (L.R.C., L.D.D., or M.S.P.). Detailed information on clinical symptoms and signs, neurologic examination, stroke risk factors, hospital course, and outcome was recorded. All patients in the registry with a diagnosis of moderate (50%-70%) to severe (>70%) BA stenosis or occlusion were included in the analysis. The BA lesions were documented by magnetic resonance angiography, catheter angiography, or transcranial Doppler. Infarct location was determined by computed tomography and magnetic resonance imaging using standard axial sequences (T1, T2, and fluid-attenuated inversion recovery). Patients with dolichoectatic BA were excluded if there were no signs of atherothrombosis.

Stroke mechanisms were determined according to previously described criteria.¹³ Large artery occlusive disease–hemodynamic mechanism required demonstration of severe stenosis or occlusion of the vertebral arteries, BA, or posterior cerebral arteries (PCAs) with infarction in the respective territories supplied by these arteries. Cardioembolism was diagnosed when infarction within the territory of multiple, superficial, or single large arteries occurred and an embolic cardiac source was identified. Artery-to-artery embolism was defined as the presence of a severely stenosed or occluded proximal artery with evidence of infarction distal to the occlusive vascular lesion. Branch penetrator disease was diagnosed when infarction was limited to the territory of the penetrating branch(es) without large artery involvement or a cardioembolic source.

We used a previously published classification¹⁶ that divides the intracranial vertebrobasilar territory into 3 levels according to its rostral-caudal distribution to describe the localization of brain infarctions. The levels and arteries included are shown in Figure 1. The proximal posterior circulation territory is fed by the intracranial vertebral arteries (ICVAs) and their posteroinferior cerebellar artery branches and includes the medulla and the posteroinferior cerebellum. The middle posterior circulation territory includes the BA up to its superior cerebellar artery branches and the anteroinferior cerebellar arteries and their pontine and anteroinferior cerebellar artery–cerebellar supply zones. The superior cerebellar arteries, the penetrating distal branches of the BA, and the PCAs, in addition to their midbrain, thalamic, superior cerebellar artery–cerebellar, temporal, and occipital lobe supply zones, constitute the distal posterior circulation territory. When patients had infarcts in more than 1 territory, they were classified in combined groups: proximal and middle; proximal and distal; middle and distal; and proximal, middle, and distal.

Patients were divided into 3 groups. In group A, stenosis or occlusion of the BA was due to intrinsic and isolated disease, usually atherosclerosis, often with superimposed thrombus formation. This included patients in whom thrombosis of the BA was thought to have originated in and propagated from an ICVA. In group B, BA disease was part of severe and extensive atherosclerosis with additional, often multiple, atherosclerotic sites within the posterior circulation. Group C included patients who had embolism to the BA from the heart or vertebral arteries.

Differences in demographic, vascular, and clinical characteristics between groups were compared by univariate analysis, using the t test for age and the χ² test for all categorical variables. To examine the association of predictor variables and the risk of poor outcome (defined as death or major disability), we determined relative risks (RRs) and their corresponding 95% confidence intervals (CIs). All statistical analyses were performed using Sigmastat version 2.0.3 (Sigmastat Corp, SPSS Inc, Chicago, Ill).

Among the 407 NEMC-PCR patients, we identified 109 patients (26.8%) with BA disease. Among these, 87 patients (79.8%), 59 men and 28 women, had moderate to severe BA occlusive disease. This represents 21.4% of the total number of patients in the registry. Ages ranged from 5 to 87 years (mean, 61.2 years). Women were, on average, older than men; half of the women were older than 70 years, compared with 22% of the men (P = .008). Seventy-six patients (87.4%) were white, 7 (8.0%) were of Asian descent, and 2 (2.3%) were African American. The demographic characteristics of the study population are shown in Table 1.

Most patients (39 [44.8%]) belonged to group A and had isolated, intrinsic BA stenosis or occlusion, including 2 patients in whom the BA atherothrombosis originated and propagated from an ICVA. Slightly fewer BA lesions (36 [41.4%]) were due to BA occlusive disease as part of extensive posterior circulation atherosclerosis (group B). The 12 patients (13.8%) in group C had embolism to the BA from the heart or vertebral arteries.

Table 2 shows age, sex, and vascular risk factors for stroke according to BA groups. Patients with embolic BA lesions (group C) were significantly younger and had a wider age range than patients with atherosclerotic BA disease (group B) (P = .01). Men were most likely to have isolated, intrinsic BA lesions (29 [49.2%]), whereas widespread atherothrombosis of posterior circulation arteries predominated among women (13 [46.4%]). All groups of patients with BA occlusive disease had a high prevalence of vascular stroke risk factors, and multiple risk factors were common (62 patients [71.3%]). Hypertension was the most frequent risk factor, present in 58 patients (66.7%), followed by hyperlipidemia (33 patients [37.9%]) and coronary artery disease (29 patients [33.3%]). The prevalence of most risk factors was similar in groups A and B. Five patients (12.8%) with isolated, intrinsic BA lesions and only 1 patient (2.8%) with multivessel atherosclerosis had no recognized risk factors. Most individuals without vascular risk factors had nonatherosclerotic causes of their BA disease: 2 had traumatic BA dissections; 1, basilar migraine; and 1, a hypercoagulable state secondary to AIDS. Among patients with isolated or widespread atherothrombosis (groups A and B), 57 (76%) of 75 had 2 or more vascular risk factors, while more than half (7 [58.3%]) of patients with embolism to the BA had no or 1 risk factor (P = .03). The only 2 patients who used oral contraceptives had cardioembolism, a 29-year-old smoker and user of illicit drugs who had atrial fibrillation and an 18-year-old patient who had a large atrial septal defect. Among patients with a cardiac embolic source, 6 (75%) of 8 had 1 risk factor or none. Only 1 patient in this subgroup had several risk factors. A 68-year-old patient who smoked had hypertension, hyperlipidemia, and coronary artery disease; a left ventricular akinetic segment; and a thrombus noted in the left ventricle on echocardiography. When excluding this patient, hypertension, smoking, and use of oral contraceptives were the only risk factors associated with cardioembolic BA disease.

The course of the BA occlusive disease among groups is shown in Table 3. Two thirds of patients (58 of 87) initially had TIAs, of whom 34 (58.6%) progressed to stroke. The most common clinical course among patients with intrinsic, localized BA or multivessel disease was TIA followed by stroke (32/75 [42.7%]). Except for 2 patients, embolism to the BA was associated with sudden-onset stroke without preceding TIAs (10 patients [83.3%]). These 2 patients had TIAs that preceded strokes by an average of 1.2 months. One was the 18-year-old woman with a large atrial septal defect described earlier; she had a TIA 10 days before the stroke. The other was a 20-year-old man with a small patent foramen ovale with an interatrial shunt; he had progressively frequent and symptomatic TIAs in the 2 months preceding the stroke. Among patients in groups A and B, the frequency of ischemia patterns was similar, with TIAs followed by stroke predominating in both groups. The TIAs occurred for months and even years in some patients (Table 3) and were usually multiple and increased in frequency as the stroke approached. When excluding 4 patients (3 from group A, 1 from group B) with TIAs occurring across 5 years, the average period of occurrence of TIAs was 2.1 months for patients with isolated BA disease and 7.8 months for patients with multivessel disease. Three patients with widespread atherosclerotic disease continued having TIAs after the stroke.

Table 4 lists the most common symptoms and clinical signs. Overall, the most common findings were bulbar and pseudobulbar signs (64 patients [73.6%]), weakness (49 patients [56.3%]), and vertigo or dizziness (47 patients [54.0%]). Only 5 patients (5.7%) had tetraparesis or tetraplegia; 1 patient had locked-in syndrome. Considering long tract motor findings, 2 (16.7%) of 12 patients with BA embolism had tetraparesis compared with only 3 (7.7%) of 39 patients with localized BA disease and none in those with widespread posterior circulation atherosclerosis. Hemiparetic patients often had motor or reflex abnormalities on the nonparalyzed side. Almost half of the patients had cerebellar signs and oculomotor abnormalities. Cerebellar, bulbar and pseudobulbar signs, and sensory abnormalities were significantly more common in groups A and B than among those with embolism to the BA (P = .015, .013, and .02, respectively). Headache was quite frequent (36 patients [41.4%]). Sixteen patients (18.4%) had decreased level of consciousness. Four patients (4.6%) had "top of the basilar syndrome."

The most common mechanism underlying BA territory ischemia was, by far, hemodynamic associated with large artery occlusive disease (65 patients [74.7%]) (Table 5). When excluding patients with BA embolism, this mechanism accounted for 65 (86.7%) of the ischemic events. The remaining stroke mechanisms in groups A and B were artery-to-artery embolism secondary to BA dissection (3 patients), branch artery occlusive disease (3 patients), basilar migraine (2 patients), coagulopathy (1 patient), and unknown (1 patient). Among the 12 patients with embolism to the BA, 8 had a cardiac source. Congenital heart lesions were the underlying source in 4 patients. A 5-year-old boy had cyanotic congenital heart disease due to an atrial septal defect and pulmonary atresia. Atrial septal defect was present in 2 other patients and patent foramen ovale in 1 patient; these patients were aged 18, 42, and 20 years, respectively. Two patients had atrial fibrillation, secondary to thyrotoxicosis in a 29-year-old woman and chronic in an 87-year-old man. One patient had an atrial septal aneurysm and 1, a left ventricular akinetic segment with a mural thrombus. None of these individuals had a potential proximal arterial source of embolism. Two patients had intra-arterial embolism, 1 from a severe extracranial vertebral artery (ECVA) occlusion and 1 from bilateral ICVA occlusive disease. Neither had a potential cardiac embolic source. One patient had both a large, dyskinetic ventricular segment and an ECVA occlusion. No embolic source was identified in 1 individual.

Fifty-five (63.2%) of 87 patients had moderate or severe BA stenosis, and 32 (36.8%) of 87 patients had BA occlusion (Table 6). Stenoses were more common among patients with intrinsic BA disease (50/75 [66.7%]), whereas occlusions predominated in the group with embolism to the BA (7/12 [58.3%]). The BA lesion was most often located in the middle third of the artery (54/87 [62.1%]), followed by the proximal (34/87 [39.1%]) and distal (20/87 [23%]) thirds. In groups A and B, the middle segment was involved in 50 (66%) patients and the distal third in 13 (less than 20%), whereas BA embolism (group C) most often affected the distal third (7/12 [58.3%]). Single segment BA lesions predominated in all groups (68 [78.2%]). Multiple segment lesions were most common in group B (12/36 [33.3%]). Among the 17 patients in groups A and B who had involvement of multiple BA segments, most (16 [94.1%]) had involvement of the middle third, of whom 9 (56.3%) also had proximal lesions and 5 (31.3%) had distal lesions. All 3 BA segments were affected in only 2 patients. Among patients with embolism to the BA, 10 (83.3%) had a single lesion, half of these localized to the distal segment. Both patients with lesions affecting multiple segments had middle and distal BA involvement.

Table 7 lists the additional posterior circulation vascular lesions among patients in group B. The arteries most often affected in association with the BA were the ICVAs, which showed moderate to severe unilateral or bilateral lesions in 27 (75.0%) of 36 patients. Basilar artery atherosclerotic disease combined with bilateral ICVA lesions was the single most frequent pattern of multivessel posterior circulation disease and occurred in 9 patients (25%) in this group. The ECVAs were diseased in 20 patients (55.6%), most of whom had unilateral lesions (14 [70.0%]). Only 2 patients had unilateral PCA involvement, 1 in association with a unilateral ICVA and 1 with a unilateral ECVA lesion. Eleven patients (30.6%) in this group also had internal carotid artery disease, which was moderate to severe in 9 patients (25.0%) and occluded in 2 (5.6%). Among patients with BA stenosis, 75.0% of other affected arteries were stenotic and 25.0% occluded, while among BA occlusion patients, 80.0% of other affected arteries were also occluded (P <.001), showing a significant association in severity of atherosclerosis between posterior circulation arteries.

Localization of infarcts within posterior circulation territories is shown in Table 8. Ischemia distribution varied substantially according to the BA group. Most patients with isolated or multivessel BA occlusive disease had involvement of the middle territory alone or in combination with other territories (33 [84.6%] and 26 [72.2%], respectively). Distal territory was involved in 21 patients (53.8%) in group A and the proximal territory in 20 patients (55.6%) in group B. Basilar artery embolism predominantly affected the distal territory (8 patients [66.7%]), either alone (8 patients [66.7%]) or combined with the middle territory (4 patients [33.3%]). Involvement of multiple intracranial territories was more common in group B (22 [61.1%]), whereas in 21 patients (53.8%) with isolated intrinsic BA lesions and 8 patients (66.7%) with embolism to the BA, only 1 territory was affected. When patients with only TIAs and reversible deficits were included, involvement of the medial and distal territories, either alone or combined, accounted for 43 (68.3%) of 63 patients. Overall, the middle posterior circulation territory including the pons and the anteroinferior cerebellum was most often affected (66/87 [75.9%]), followed by the distal territory supplied by the rostral BA, superior cerebellar arteries, and PCAs (44/87 [50.6%]) and the proximal segment, which includes the medulla and posteroinferior cerebellum (26/87 [29.9%]). Among 26 individuals with proximal territory infarcts, only 4 had no ICVA occlusive lesion. One of these patients had migraine with extensive BA vasoconstriction, 1 had coagulopathy secondary to AIDS, and 2 had ECVA occlusions.

Table 9 shows outcomes at hospital discharge. The mortality rate was surprisingly low; only 2 patients (2.3%) died, 1 in group B because of cerebrovascular disease and 1 in group A from other causes. Overall, 25 patients (28.7%) died or were left with a major deficit. Outcome was the best in patients with widespread atherosclerosis, with 30 (83.3%) having no or only minor disability at hospital discharge, whereas 30 [69.2%] of 39 patients with isolated BA occlusive disease had favorable outcomes. The worst outcomes occurred in group C; although there were no deaths, 7 (58.3%) individuals had major deficits and only 5 (41.7%) had minor or no deficits. This represented a 2.4-fold higher risk of poor outcome in patients with BA embolism compared with patients with diffuse or localized atherosclerosis (58.3% vs 24%; RR, 2.43; 95% CI, 1.30-4.54; P = .03). Even when patients with only TIAs and reversible deficits were excluded from the analysis, overall, 38 (60.3%) of 63 individuals had a good outcome.

When considering clinical symptoms and signs, decreased level of consciousness, tetraparesis or tetraplegia, and pupillary abnormalities were significantly associated with worse outcome. Among the 25 patients who faired poorly, 11 (44%) had a reduced level of consciousness compared with 5 patients (8.1%) among those who had minor or no disability; this yielded a 3.5-fold increase in risk of poor outcome (RR, 3.49; 95% CI, 1.96-6.19; P <.001). Three patients with poor outcome (12%) and 1 patient (1.6%) who had a minor deficit all initially had pupillary abnormalities, showing a trend toward worse outcome in patients with abnormal pupils (RR, 2.93; 95% CI, 1.49-5.75; P = .06). All patients who had tetraparesis had poor outcomes (RR, 4.1; 95% CI, 2.8-6.0; P = .001); 1 patient died and 4 had major disabilities.

Among patients with posterior circulation ischemia due to isolated or multivessel BA atherosclerosis, we analyzed outcome as a function of BA lesion severity and found that patients with BA occlusion had a significantly higher risk of poor outcome than individuals with BA stenosis (40% vs 16%; RR, 2.5; 95% CI, 1.13-5.54; P = .02). When we included patients with BA embolism in this analysis, there was still a trend toward a higher risk of poor outcome associated with occlusions (40.6% vs 21.8%; RR, 1.86; 95% CI, 0.97-3.58; P = .06), even though this was the only group in which occlusions predominated over stenoses, even among patients with favorable outcomes.

We further analyzed outcome stratified by region of infarction. Table 10 presents poor outcome as a function of the posterior circulation territory involved. Involvement of the middle and distal territories alone or combined accounted for 21 poor outcomes (84.0%) (death or major disability) and 100% in the group with BA embolism. All patients who had only the proximal territory affected had favorable outcomes. Among patients with involvement of the middle or distal territories (alone or associated with other territories), 38 (35.0%) of 110 patients had poor outcomes. Involvement of the distal territory was associated with a more than 2-fold increase in the risk of death or major disability (RR, 2.08; 95% CI, 1.0-4.3; P = .04). Risk estimates for the proximal and proximal and middle regions combined could not be calculated accurately because of the small number of patients.

Following the landmark report of Kubik and Adams,⁴ necropsy, imaging, and angiographic studies have attempted to define the common clinical and pathologic features of BA occlusive disease. Our report of 87 patients with moderate to severe BA stenosis or occlusion collected consecutively in the NEMC-PCR is, to our knowledge, the largest prospective series to date of BA occlusive disease collected before death. The previous largest prospective studies include the follow-up analysis of 44 patients with basilar and distal vertebral artery stenosis reported by Moufarrij et al,¹⁷ the Mehler¹⁵ description of 61 patients with rostral BA syndrome, and the report from Bogousslavsky et al¹⁷ of 70 patients with infarcts in the posterior circulation from the Lausanne Stroke Registry, 27 of whom had BA stenosis or occlusion.

The incidence of BA occlusive disease has not been determined because of selection bias and diverse inclusion criteria used in most reports. We found that approximately one fourth of patients in the NEMC-PCR had some degree of BA involvement, of which the majority (87/109 [79.8%]) was moderate to severe or occluded. This frequency lies within the range previously described in most necropsy- or angiography-based reports. In the necropsy study of Castaigne et al,¹¹ 18 BA occlusions were identified among 44 patients (41%) with posterior circulation infarcts. In the Lausanne Stroke Registry, magnetic resonance angiography showed 27 cases of BA stenosis or occlusion among 70 consecutive patients (39%) with posterior circulation strokes¹⁷ ; only patients older than 45 years who had magnetic resonance imaging results that showed lesions appropriate to the clinical syndrome were included. In a recent follow-up of this registry, only 43 patients (1.8%) had documentation of BA occlusive disease among 2332 patients with posterior circulation ischemia admitted consecutively during a 10-year period. In the Joint Study of Extracranial Arterial Occlusions, an arteriographic study of patients with extracranial artery disease, BA stenosis was identified in only 292 (7.7%) and occlusion in 303 (0.8%) of 3788 patients.¹⁸

We described 3 distinct groups of patients that were relatively homogeneous with respect to risk factors, etiology, clinical course, symptoms and signs, localization of the BA lesion, infarct distribution, and outcome. Previous inconsistent data in the literature regarding these characteristics may be owing to the inclusion of variable numbers of patients belonging to these different groups. The largest group (group A) consisted of patients with isolated occlusive BA disease, caused mostly by atherosclerosis, as well as a small number of nonatherosclerotic etiologies. This group made up 44.8% of cases of BA disease in the NEMC-PCR. The prevalence of vascular risk factors was high, and the middle segment of the artery, with the corresponding posterior circulation territory, was most frequently involved. Twenty-nine of these patients initially had TIAs followed in 20 by strokes. Bulbar and pseudobulbar symptoms were present in about 31 patients (80.0%), and more than half (20/39) had hemiparesis or hemiplegia, whereas tetraparesis was infrequent. Outcome was favorable in almost 27 (70.0%) of patients, and mortality was extremely low.

The second most frequent group (group B) included older patients with BA atherothrombosis as part of widespread posterior circulation atherosclerosis. The ICVAs were most often involved in association with the BA, followed by the ECVAs. Despite its frequency, BA disease as part of extensive atherosclerosis has not been recognized as a vascular pattern and has not been well described in the past. However, analysis of some reports in the literature reveals the common presence of multivessel posterior circulation disease. Moufarrij et al,¹² in their long-term follow-up of 44 patients with BA and distal vertebral artery stenosis, included 17 patients (38.6%) with involvement of multiple intracranial arteries. Among 49 patients with BA occlusion described by Ferbert et al,¹⁹ 22 (44.9%) also had ICVA lesions. When Bogousslavsky et al¹⁷ reported the etiology of posterior circulation infarcts in the Lausanne Stroke Registry, they identified by magnetic resonance angiography results 7 patients (25.9%) with vertebral or PCA disease among 27 with BA lesions; 1 of these patients had BA occlusion, unilateral ICVA stenosis, and PCA occlusion. In addition, Doppler ultrasound results identified internal carotid artery stenosis in a few of these patients. In the Devuyst et al¹⁴ report on symptomatic BA stenosis and occlusion, vertebral artery atherosclerotic lesions were found in 46.5% of patients.

Nearly all patients in the group with multivessel atherosclerotic disease had vascular risk factors, usually multiple, and the primary stroke mechanism was hemodynamic. The proximal and middle segments of the artery were most often affected, and occlusions were the most prevalent in this group. Bulbar and pseudobulbar signs, cerebellar signs, and vertigo were most common. Eighteen patients (50%) had long tract motor signs, yet none were tetraparetic. Despite the severity of the atherosclerosis in these patients, the outcome was the best, with fewer than 9 (25%) developing a major deficit or death. In 29 patients (80%), TIAs were the initial symptoms, lasting during periods ranging from a day to several months and even years before progressing to strokes. The frequent occurrence of TIAs in patients with BA occlusive disease, including the phenomenon of herald hemiparesis, has been noted in several previous reports^{9 - 10 ,19} ; Pessin et al¹⁰ found that among patients with middle and distal BA stenosis, the predominant initial clinical symptom was TIA, which occurred in two thirds of patients.

The third and smallest group (group C) included patients with BA embolism from the heart or proximal arteries and constituted only 13.8% of the registry. These patients were, on average, younger and had fewer vascular risk factors; 3 (27.3%) of 12 had no risk factors. Almost all patients had sudden onset of symptoms without previous TIAs, and embolism was most common to the middle and distal segments of the artery. Tetraparesis and tetraplegia, as well as a decreased level of consciousness, occurred the most often in this group. Five patients (40%) also had bulbar and pseudobulbar signs and oculomotor abnormalities. This group had the worst outcome; although no patients died, 7 (58.3%) had major deficits, which represented a 2.4-fold higher risk of adverse outcome than among patients with localized or diffuse atherosclerosis.

Overall, outcomes in this series were substantially better than previously reported. Historically, BA occlusive disease has been considered to convey a severe prognosis and high mortality rate; this since the initial description of the clinical and pathological findings among 18 patients with fatal brainstem infarction due to BA occlusion by Kubik and Adams.⁴ This was followed by studies of select groups of patients with specific vertebrobasilar lesions or clinical symptoms, and the morbidity and mortality rates in most of these reports were high, frequently more than 60% to 70%.^{20 - 23} In the most extensive of these studies, Labauge et al²⁴ reviewed 348 patients with BA occlusion and found a mortality rate of 86%. In a recent analysis by Devuyst et al¹⁴ of 43 patients from the Lausanne Stroke Registry with BA stenosis or occlusion, outcome was poor (54% of patients died or had severe disability, and only 12% had minor or no disability).

In contrast, some authors reported small groups of select patients with severe BA occlusive disease, yet minor or no residual neurologic deficit.^{9 ,25 - 26} Fields et al²⁵ reported 8 patients with angiographically proven BA occlusions of whom only 1 had a severe deficit. Caplan⁹ described 6 patients with BA occlusion confirmed by angiography results, 4 of whom had mild or no long-term deficit. Labauge et al²⁶ reported 4 patients with BA occlusion with favorable outcomes and suggested that long-term survival may occur in patients with occlusions restricted to the lower or middle BA with good collateral supply from the carotid and cerebellar arteries. The findings from these isolated reports were supported by results from prospective studies: Moufarrij et al¹² observed a 5-year survival rate of 78%; in the Mehler¹⁵ description of 61 patients with rostral BA syndrome, 77% had minor or no residual deficits; and in the Lausanne Stroke Registry, the mortality rate among 70 patients with acute posterior circulation infarcts was only 4%.¹⁷ In our series, only 2 patients died, yielding a mortality rate of 2.3%. Overall, 62 (72.2%) of 87 patients had favorable outcomes, suggesting a more benign natural history of the disease than previously assumed. Patients with benign outcome may not have been identified in the past because noninvasive vascular imaging techniques were not available and only patients with severe clinical findings were studied, skewing the results to more adverse outcomes.

We identified several factors that were significantly associated with more severe outcomes by univariate analysis: distal territory involvement, embolism, and BA occlusion. Twenty-one (84.0%) of 25 patients with poor outcome had infarcts that involved the middle territory alone or in combination with another territory. The distal territory was often infarcted along with the middle territory suggesting spread of the lesion to the rostral BA and its distal branches, or embolism to these branches. Involvement of the distal territory was associated with a more than 2-fold increase in risk of death or major disability, in accordance with previous studies.^{12 ,14} This reflects the effects of pontine and tegmental brainstem infarctions. Emboli most often lodge in the rostral BA,^{8 ,15} explaining why patients in group C had the worst outcomes. Furthermore, patients with embolic lesions to the BA have less time to develop collateral circulation. Accordingly, patients with BA stenosis are at lower risk of severe outcomes compared with those with BA occlusion.

Decreased level of consciousness, tetraparesis, and abnormal pupils were identified as clinical predictors of poor outcome. Although these clinical signs were less frequent than previously described, their presence increased the probability of major disability or death 3- to 4-fold. Devuyst et al¹⁴ recently found that decreased consciousness was the single most powerful prognostic factor in BA disease, followed by pupillary disorders, bulbar signs, and dysarthria.

Analysis of this large series of patients contributes to understanding the spectrum of BA occlusive disease and clarifies previous controversies. We described 3 distinct groups of patients with different vascular, clinical, etiological, and prognostic characteristics. A large group of patients, previously underemphasized, who had BA disease as part of widespread posterior circulation atherosclerosis had surprisingly good prognoses. We found several predictive factors of poor outcome. Early identification of these factors in patients and their inclusion into 1 of the BA groups may guide diagnostic evaluation and rational treatment.

Corresponding author: Louis R. Caplan, MD, Division of Cerebrovascular Disease-Palmer 127, West Campus, Beth Israel Deaconess Medical Center, 330 Brookline Ave, DA 779, Boston, MA 02215 (e-mail: lcaplan@bidmc.harvard.edu).

Accepted for publication November 20, 2003.

Author contributions: Study concept and design (Drs Voetsch and Caplan); acquisition of data (Drs DeWitt, Pessin, and Caplan); analysis and interpretation of data (Drs Voetsch and Caplan); drafting of the manuscript (Drs Voetsch and Caplan); critical revision of the manuscript for important intellectual content (Drs Voetsch, DeWitt, Pessin, and Caplan); statistical expertise (Dr Voetsch); administrative, technical, and material support (Dr Caplan); study supervision (Dr Caplan).