Stroke or Transient Ischemic Attacks With Basilar Artery Stenosis or Occlusion: Title and subTitle BreakClinical Patterns and Outcome FREE

BASILAR ARTERY occlusion (BAO) is an uncommon cause of stroke generally associated with a high mortality rate and a poor functional outcome in survivors.^{1 - 7} Labauge et al⁸ reported only 31 survivors of BAO. Even with thrombolytic therapy, the survival rate in BAO is only approximately 50%.^{9 - 16} However, in the past, most studies of BAO were based on autopsy material and thus were naturally biased toward fatal outcome.^{1 ,4 - 6} More recently, cases of survival with more benign outcomes were reported in patients with BAO,^{17 - 21} suggesting that the prognosis of BAO is more diverse than generally admitted. Moreover, only limited data relating to basilar artery stenosis (BAS) have been published.^{22 - 23} However, the development of new, safe, noninvasive diagnostic tools, such as magnetic resonance angiography (MRA), as well as the advent of new treatments, such as thrombolysis and angioplasty, have facilitated diagnosis and treatment of BAO and BAS. We present a retrospective study to provide information about spontaneous outcome and clinical predictive factors in symptomatic patients with BAS greater than 50% or BAO not treated by thrombolysis.

Group 1 consisted of 43 patients selected from a pool of 2332 patients with brain ischemia involving posterior circulation who were consecutively admitted to our center between 1985 and 1995. These 43 patients were part of the Lausanne Stroke Registry (LSR), the characteristics of which have been presented in detail elsewhere.²⁴ The diagnostic methods used to assess BAO or BAS were arteriography (1985-1990) and MRA²⁵ (1991-1995). We obtained computed tomographic scans from 6 and performed magnetic resonance imaging on the 37 remaining patients. The etiology of the ischemic event was classified as previously described.^{24 - 26} Group 2 consisted of 45 of 4620 brain autopsies performed between 1980 and 1995 in our center. Of the 45 autopsied patients, 37 came from Centre Hospitalier Universitaire Vaudois (CHUV) (Lausanne, Switzerland) and 8 came from outside hospitals. Thirty-seven patients of CHUV who died before being assessed by MRA or arteriography were included in group 2: 20 patients from 1980 to 1985 and 17 patients from 1985 to 1995. A case was eligible for the study when the neuropathologic examination included a sufficiently detailed study of the cerebral vasculature, demonstrating BAS greater than 50% together with a well-delineated infarction in the vertebrobasilar territory. All patients died of stroke within 2 weeks.

Magnetic resonance angiography has been performed at our institution since 1991 using a Magnetom SP 63 1.5-Tesla imager (Siemens, Siemens, Germany), which consists of 3-dimensional time-of-flight sequences sensitive to arterial flow. To assess BAS, the vessel was measured at its point of maximal narrowing and compared via MRA with a normal section of the BA adjacent to the stenosis to determine the degree of stenosis (normal lumen diameter − residual lumen / normal lumen diameter). Basilar artery stenosis was graded as follows: no stenosis, stenosis less than 50%, stenosis greater than 50%, and occlusion.^{25 ,27 - 28} The segment of the BAS was defined according to the same criteria used in the New England Medical Center Posterior Circulation Registry.²⁸ Arteriography graded BAS on the basis of the same criteria previously cited for MRA.

The outcome of patients at 1 month was measured by a 5-level scale, with I indicating disability; II, slight disability but able to return to most previous activities; III, moderate disability, with limited return to certain previous activities; IV, severe disability impeaching the return to most previous activities; and V, death. We have used the 1-month outcome since the LSR was first set up.²⁴ Our 5-level scale was very similar to the Rankin scale. We defined poor outcome as levels IV and V to be compatible with our previous studies.^{24 - 26}

At autopsy, the degree of BAS was estimated visually on transverse sections of formalin-fixed arteries and graded as less than 50%, greater than 50%, or occlusion, defined as a complete block of the arterial lumen. The diagnosis of atherosclerosis was based on the usual macroscopic and microscopic criteria. The diagnosis of cerebral embolism was based on arterial occlusion as shown by an antemortem aggregation of platelets, fibrin, and red and white blood cells that differed from a postmortem clot; no adherence to the vessel wall; and occurrence in the absence of local atherosclerotic stenosis, which might have produced thrombosis.

The outcome was dichotomized by comparing patients with stroke having a functional outcome of I through III (favorable outcome) with those having a functional outcome of IV or V (poor outcome). Because a large number of comparisons were performed, a Bonferroni correction was applied. Results were considered statistically significant when the P value was <.001 and marginally statistically significant when the P value was <.05. All variables were analyzed with the Fisher exact test and a multivariate logistic regression of poor outcome was performed on the factors identified with the univariate approach. To validate these findings, the prevalence of the factors determined to be associated with poor outcomes in patients in group 1 was assessed in autopsied patients from group 2.

The 88 patients consisted of 43 in group 1 (34 men and 9 women; mean [SD] age, 64 [12]) and 45 in group 2 (28 men and 17 women; mean [SD] age, 71 [11] years). The diagnosis of BAS in group 1 was obtained by MRA in 37 and arteriography in 6 while in group 2 it was obtained by necropsy. No patient received thrombolysis. Forty-three of 88 patients were treated with anticoagulants and the remaining 45 patients received antiplatelets. In group 1, of 35 patients with stroke, a score of I, II, III, IV, and V was reported in 0 (0 %), 4 (12%), 12 (34 %), 18 (52 %), and 1 (3 %), respectively, resulting in 16 cases with favorable outcomes and 19 cases with poor outcomes. The outcomes of patients enrolled in LSR from 1995 to 2000 are presented in Table 1. Degree and localization of BAS are presented in Table 2. Eight of the 8 patients with transient ischemic attack (TIA) in group 1 had BAS without BAO. The distribution of the degree of BAS between patients with stroke with poor outcomes in group 1 and group 2 was not statistically different (P>.05). Transient ischemic attacks preceding stroke were recorded in 27 (34%) of 80 patients with stroke. The number of TIAs preceding infarct was multiple (>2) in 19 (70%) of 27 cases and the interval between the first TIA and stroke in the posterior circulation was less than 2 weeks in 16 (60%) of the 27 cases. The symptoms included cerebellar ataxia in 52%, vertigo in 44%, dysarthria in 41%, diplopia in 33%, and consciousness disorders in 33%. Stroke patterns for the 80 patients with stroke included, in order of decreasing frequency, motor disorders in 57%, consciousness disorders in 52%, dysarthria and pupillary disorders in 49%, diplopia in 47%, lower cranial nerve involvement in 40%, cerebellar disorders in 36%, sensory impairment in 26%, headache in 25%, nystagmus in 20%, tetraparesia in 12%, epilepsy in 8%, locked-in syndrome in 7%, decerebration in 6%, and "top of the basilar syndrome" in 3.5%. In addition, coma at the onset was reported in 18% of all cases. A bilateral neurologic deficit was observed in 42% of patients on admission. Eight patients in group 1 (19%) had no other clinical manifestation than TIA and all experienced multiple TIAs (3-7) characterized by ataxia in 6, ocular disorders in 4, consciousness disorders in 3, dysarthria in 3, motor disorders in 3, and sensory impairment in 2. Parenchymal lesions by neuropathology (group 2) and neuroimaging (group 1) among the 80 patients with stroke were as follows: bilateral pontomesencephalic (n = 18), unilateral pontine (n = 14), bilateral pontine (n = 13), unilateral mesencephalic (n = 11), unilateral bulbar (n = 9), unilateral pontomesencephalic (n = 7), bilateral mesencephalic (n = 3), bulbopontine (n = 2), unilateral cerebellar (n = 2), and bilateral bulbar (n = 1). Bilateral cerebellar lesions were associated with bilateral pontomesencephalic (n = 6), bilateral pontine (n = 3), bilateral mesencephalic (n = 2), unilateral bulbar (n = 2), unilateral mesencephalic (n = 1), and unilateral pontine (n = 1) lesions. Unilateral cerebellar lesions were linked to bilateral pontomesencephalic (n = 3), unilateral pontomesencephalic (n = 1), unilateral mesencephalic (n = 1), and unilateral bulbar (n = 1) lesions. Predictive factors for outcome in group 1 were determined by univariate analysis. Definitely and marginally statistically significant factors associated with poor outcomes in group 1 are presented in Table 3. The following factors were found to be statistically nonsignificant in terms of outcome: sex, high blood pressure, diabetes mellitus, smoking, cardiac failure, ischemic cardiopathy, atrial fibrillation, family history of stroke, hypercholesterolemia, alcoholism, anticoagulant treatment, localization of the BA segment involved, and the localization of the parenchymatous ischemic lesions. The distinction between BAS greater than 50% and BAO did not reach statistical significance for the outcome; a tendency toward a worse outcome among patients with BAO was noticed. Multivariate logistic regression showed that the 4 definitely significant factors isolated in the univariate analysis remained significant and also revealed that consciousness deficit or the presence of the other 3 factors were associated with 100% poor outcome; conversely, their absence was associated with only 11% poor outcome in nonautopsied cases (Table 4). In group 2, the presence of consciousness disorders alone or the combination of the 3 remaining factors listed in Table 3—dysarthria, pupillary abnormalities, and bulbar symptoms—was found in 87% of patients (Table 5).

The etiology of BAO or BAS was cardioembolic in 18% of patients, due to large atherosclerotic BA disease in 82%, associated with atherosclerotic disorders of the vertebral artery in 46%, and due to the dissection of both the vertebral artery and BA in 0.5%.

As recently suggested,²¹ we found that the spontaneous prognosis of patients with stroke caused by BAO or BAS was more diverse (see Table 1 for 1995-2000 period) than generally believed. Our results show mortality or severe disability in 54%, moderate disability in 34%, and minor or no disability in 12% of nonautopsied patients (group 1), compared with of 20%, 20%, and 60%, respectively, in the series by Brandt et al.²¹ We must keep in mind, however, that our patients who died before being assessed by MRA or arteriography were included among the autopsied cases (group 2): 20 patients from 1980 to 1985 and 17 patients from 1985 to 1995. Our results are also in agreement with those of Hacke et al,²⁹ who found that among 45 patients with BA embolism, 27% and 33%, respectively, had no clinical signs or minor neurologic deficits on discharge. In the New England Medical Center Posterior Circulation Registry,²⁸ posterior circulation ischemia was also found to have a more benign outcome than previously thought: most patients had no or only minor disability after 30 days, while the frequency of death due to stroke or total mortality was 1.9% and 3.6%, respectively, and mortality and major disability were present in 21.3% of patients. In 1993, we reported a mortality rate of only 4% in 70 patients with acute posterior infarcts investigated by MR imaging and MRA.²⁵ These data^{25 ,28} included a variety of vertebrobasilar occlusive lesions but Caplan²⁸ concluded that the outcomes in patients with BA disease was surprisingly good and better than previously reported. Moreover, we found that 8 patients with BAS had TIA as the only presenting event. Patients with benign outcomes were probably not identified in the past because angiography was not performed and noninvasive diagnostic tools such as MRA were not available.²¹ Angiography was considered risky in the posterior circulation²⁸ ; when conducted, clinicians often did not rely on the results to guide therapy because until very recently no effective treatment existed.^{30 - 31}

Slightly more than one third of all patients with stroke reported vertebrobasilar TIAs preceding stroke; these were generally multiple and usually occurred within the 2 weeks before stroke. The high prevalence of prior TIAs in our study is not surprising since TIAs preceding stroke caused by BAS or BAO have been reported with a frequency of approximately 60% in 2 recent angiographic studies^{7 ,21} as well as in the past.^{1 ,4 ,8 ,17 ,19} The frequent occurrence of prior TIAs in our series may suggest local thrombotic-atherosclerotic BA stenotic disease, which was considered to be the cause of ischemia in 81% of patients. However, the frequent combination of BA stenotic disease with vertebral artery atherosclerotic lesions in 46.5% of patients must also evoke a possible artery-to-artery embolization.^{7 ,30 - 33} The significance of prior TIAs is debated, interpreted as a factor of good prognosis by some^{8 ,17 ,19} in contrast with other researchers.³² Most of the time, the significance of prior TIAs was biased toward a fatal outcome because these studies were based on necropsy results.^{1 ,4} In our series, having prior TIAs tended to be associated with a favorable outcome; thus, we believe that prior TIAs may be a sign of adequate collateral circulation as mentioned in previous studies^{8 ,14 ,17} or a protective effect described as "focal ischemic tolerance phenomenon."³⁴

We used MRA (Figure 1, Figure 2, Figure 3, and Figure 4) to diagnose and quantify BAS in our patients because other studies have suggested that MR imaging and MRA are reliable diagnostic tools to define the brain and vascular lesions in patients with posterior circulation infarcts. Recent studies,^{35 - 38} including our own,²⁵ have shown MRA to be very sensitive and accurate in imaging the BA but not the small branches. Caplan et al^{28 ,30 - 31 ,33} have used MRA to image BAS in patients with stroke and advocate this diagnostic tool for stroke in the posterior circulation. While MRA can overestimate the degree of BAS, we observed no statistical difference in the distribution of BAS between autopsied patients and nonautopsied patients with stroke with poor outcomes. Moreover, we found that a poor outcome was not statistically correlated with the degree of BAS even though there was a tendency toward a greater incidence of BAO in nonautopsied patients with stroke with poor outcomes. At the time of our study, MRA was not sufficiently approved for researching collateral circulation.

In the past, predictive factors of a good prognosis for BAO included having a short segmental occlusion of the BA involving the inferior or middle segments of the BA,^{21 ,28} a carotido-basilar reflux^{8 ,21} if the distal portion of the BA—the main supply of the tegmentum of the pons—remained open,²⁸ and the rapidity with which adequate collateral circulation develops.²¹ In contrast to distal embolization at the time of occlusion, a large extent of brainstem infarction, rapid course, long duration of BA occlusion, and embolic origin were correlated with a worse prognosis.¹ However, there is still a lack of clinical data that could allow a prediction of the prognosis based on early clinical findings. Our study shows that 4 clinical features present on admission of the patient—dysarthria, pupillary disorders, bulbar symptoms, and consciousness disorders—are highly significantly (P<.001) associated with a poor outcome. The presence of consciousness disorders alone—the most powerful factor—or the combination of the 3 remaining factors was associated with a poor outcome in 100% of nonautopsied cases and in 87% of autopsied cases. These 4 prognostic clinical factors are more relevant than the etiology of BA disease and the topography of BA stenotic disease (Table 2). Although the study was not powerful enough to demonstrate a statistically significant difference because BA stenotic disease is very heterogeneous according to the segment of the BA involved as well as to the degree of BAS, our results show that stenotic disease of the upper third of the BA is associated with a poor outcome (9/9 cases). A possible explanation for this observation is that the involvement of the superior third of the BA is most often caused by an embolus rather than local thrombotic lesions; consequently, there is not sufficient time to develop adequate collateral circulation. The 4 predictive clinical features mentioned take into account dysfunction of the pons and the midbrain, specifically the paramedian and tegmental portions, which are more sensitive to basilar ischemia.^{39 - 43}

Even though we have not taken into account the collateral circulation due to the limitations of MRA at the time of our study, we present evidence that the outcome of BAS greater than 50% or BAO is much more diverse than assumed from previous reports. In addition, it seems that some clinical patterns are associated with a lower risk of poor outcome.

Accepted for publication December 28, 2001.

Author contributions: Study concept and design (Dr Devuyst); acquisition of data (Drs Devuyst, Meuli, and de Freitas); analysis and interpretation of data (Drs Devuyst, Bogousslavsky, Meuli, and van Melle); drafting of the manuscript (Dr Devuyst); critical revision of the manuscript for important intellectual content (Drs Bogousslavsky, Meuli, de Freitas, and van Melle); statistical expertise (Dr van Melle).

Corresponding author and reprints: J. Bogousslavsky, MD, Department of Neurology, CHUV, rue du Bugnon 46, 1011 Lausanne, Switzerland (e-mail: julien.bogousslavsky@chuv.hospvd.ch).