Article Text
Abstract
Background In patients with a middle cerebral artery (MCA) occlusion, the involvement of the cortex may be affected by the presence of leptomeningeal anastomoses between the cerebral arteries.
Methods The authors enrolled consecutive patients with acute infarctions in the MCA territory and MCA occlusion on angiographic studies. Infarct patterns were classified into three categories based on the extent of cortical surface involvement: total cortex (TC), partial cortex (PC) and no cortex (NC). The authors analysed the infarction patterns by stroke subtype, and investigated factors that resulted in cortex sparing.
Results Out of 73 total patients, cortex-sparing infarctions were seen in 53 patients (72.6%, NC in 39 (53.5%) and PC in 14 (19.1%)). The extent of cortical involvement differed according to stroke subtype (p=0.036). TC was more frequent (42.9% vs 22.2%), and PC was less frequent (10.7% vs 27.9%, p=0.037) in cardioembolism than large-artery atherosclerosis. However, the proportion of patients with complete cortical sparing (NC) was similar between cardioembolism and large-artery atherosclerosis (46.4% vs 49.9%). In the upstream of leptomeningeal collateral arteries, the extent of cortical involvement was associated with significant stenosis of the ipsilateral anterior or posterior cerebral artery (p=0.011).
Conclusion This study suggests that pre-existing arteriolar connections, which may cover almost entire cortical surfaces of the MCA territory, exist in many patients. The findings also suggest that the extent of cortical involvement is different between stroke subtypes, and is critically affected by the status of upstream collateral arteries.
- Cerebrovascular disease
- middle cerebral artery
- collateral circulation
Statistics from Altmetric.com
Occlusion of the middle cerebral artery (MCA) usually develops into a large territorial infarction. However, many patients with MCA occlusion have infarctions that spare the cortex, have small subcortical infarctions or may even lack an ischaemic lesion, suggesting the presence of leptomeningeal anastomoses (LMA) between cerebral arteries.1 2
Although previous anatomical and angiographical studies have demonstrated the presence of arteriolar connections between leptomeningeal arteries,3–10 it is not known if they are functionally patent and can deliver sufficient blood to salvage tissue in the oligaemic brain area when a cerebral artery is occluded.2 11 The extent of cortical involvement may reflect the degree to which the leptomeningeal collateral circulation can supply the cerebral cortex.
This study investigated the presence and function of pre-existing connections of the cerebral arteries on the leptomeningeal surface by categorising infarct patterns in patients with MCA occlusion according to the degree of cortical involvement. We further analysed the infarct patterns of different stroke mechanisms for cortical involvement.
Methods
Patient selection
We retrospectively reviewed 3321 patients with acute infarction who were admitted to the neurological department and registered in the Yonsei Stroke Registry (YSR) between January 2000 and October 2008. The YSR is a prospectively collected, hospital-based registry, in which registered patients have had an acute ischaemic stroke or a transient ischaemic attack within 7 days of symptom onset.12 13 This study was approved by the hospital's institutional review board.
Included in the study were patients who had an acute infarction in the MCA territory, as demonstrated by diffusion-weighted imaging (DWI) within 7 days of symptom onset, and occlusion in the M1 segment of the MCA by magnetic resonance angiography (MRA) or digital subtraction angiography (DSA). We excluded patients who had thrombolytic treatment, evidence of a previous infarction larger than 1.5 cm in the ipsilateral side of acute infarction by T2-weighted or fluid-attenuated inversion recovery images, or patients with other aetiological causes, such as Moyamoya disease, arterial dissection and coagulopathies.
MRI assessment
DWI and MRA (both time-of-flight and contrast-enhanced images) were performed using a 1.5 T system (Signa Horizon, GE Medical System, Milwaukee, Wisconsin; or Intera or Achieva, Philips Medical System, Best, The Netherlands) or a 3.0 T system (Achieva, Philips Medical System).
We divided the infarction patterns into three categories based on the extent of cortical surface involvement (figure 1). The total cortex (TC) involvement pattern was defined as DWI lesions involving the cortex of two or more subdivisions and having a cortical margin longer than the subcortical margin. The partial cortex (PC) involvement pattern was defined as DWI lesions involving the cortex of one or more subdivisions and having a cortical margin shorter than the subcortical margin (reversed-wedge pattern), involving one subdivision (branch occlusion pattern), or involving one subdivision with minimal subcortical lesion (cortex dominant pattern). The no cortex (NC) involvement pattern was defined as lesions involving only the subcortex, with or without dot-like cortical lesions. This group included lesions involving the superficial perforator territory (superficial subcortex pattern), the deep perforator territory (deep subcortex pattern) or a combination of the two (combined subcortex pattern). The lesion pattern for each patient was classified based on the consensus of six stroke neurologists who were unaware of the stroke mechanisms of the patients.
Occlusion in the M1 segment of the MCA was determined by DSA or MRA findings. If the patient underwent DSA, these findings were given priority, even if MRA was performed concurrently with the DWI. The presence of significant stenosis (>50%) of the ipsilateral anterior cerebral artery (ACA), posterior cerebral artery (PCA) and tandem internal carotid artery (ICA) was assessed because of their potential effect on the leptomeningeal collaterals. Stenosis of the ACA and PCA was obtained to the A2 and P2 segment levels, respectively. If the ipsilateral ACA was supplied from the contralateral ACA through the anterior communicating artery, stenosis of the contralateral A1 segment was also noted. Stenosis of the ICA was measured using the North American Symptomatic Carotid Endarterectomy Trial (NASCET) method.14
Stroke subtype classification
Stroke subtype was determined using the Trial ORG 10172 in Acute Stroke Treatment (TOAST) classification system.15 Because we excluded patients with other aetiological causes of stroke or incomplete evaluation, patients were categorised into four types. Large-artery atherosclerosis (LAA) was defined as one or more significant (≥50%) stenoses of the relevant intracranial or extracranial artery without potential sources of cardioembolism. Cardioembolism (CE) was defined as at least one cardiac source for embolus in the absence of significant stenosis of the relevant artery. Both high- and medium-risk cardiac sources were considered potential sources of cardioembolism. Undetermined aetiology due to a negative evaluation (UN) was defined when neither significant stenosis of the relevant artery nor a potential source of cardioembolism was identified. Undetermined aetiology due to two or more causes identified (UT) was defined as at least one cardiac source of embolism plus a significant coexisting stenosis in the proximal segment of the relevant artery.
Vascular risk factor
Hypertension was defined by high blood pressure (systolic ≥140 mm Hg or diastolic ≥90 mm Hg) or the taking of antihypertensive agents. Diabetes mellitus was diagnosed by a high fasting plasma glucose level (≥7.0 mmol/l) or the taking of hypoglycaemic agents. Hyperlipidaemia was defined by a high level of fasting serum total cholesterol (≥6.2 mmol/l) or low-density lipoprotein cholesterol (≥4.1 mmol/l), or treatment with antihyperlipidaemic agents after a diagnosis of hyperlipidaemia. A patient was considered a smoker if they were either a current smoker or an ex-smoker of 1 month or less.
Statistical analysis
Categorical variables were compared using the Pearson χ2 test or Fisher exact test, where appropriate, and continuous variables were compared using a one-way analysis of variance (ANOVA) and expressed as mean±SD. A p value of <0.05 was considered to be statistically significant. SPSS for Windows (version 13.0) was used for statistical analysis (SPSS, Chicago, Illinois).
Results
Baseline characteristics
Of 3321 patients, 214 (6.4%) had ischaemic lesions in the MCA territory with occlusion in the M1 segment. Of these, 73 were eligible for study based on our criteria. A detailed description of patient selection is shown in figure 2. Of the 73 patients, 51 (69.9%) were men, with a mean age of 64.6±11.5 years. Hypertension was found in 54 patients (74.0%), diabetes mellitus in 28 (38.4%) and hyperlipidaemia in 13 (17.8%), and 26 (35.6%) were smokers. All patients underwent DWI and MRA within 7 days of symptom onset (mean: 0.9±1.1 days), and 16 (21.9%) patients underwent DSA. The mean time interval between DWI and DSA was 3.3±2.2 days. Transesophageal echocardiography was performed on 28 (38.4%) and transthoracic echocardiography on 12 (16.4%) patients.
Ischaemic lesion patterns according to stroke subtype
Of the 73 patients with symptomatic MCA occlusions, the most common pattern of DWI lesions was NC, seen in 39 (53.5%) patients, followed by TC in 20 (27.4%) and PC in 14 (19.1%) (figure 3). Analysis of the frequency of stroke subtypes showed LAA in 36 (49.3%) patients, CE in 28 (38.4%) and UN in 9 (12.3%). None of the patients were classified as UT. Significant differences in ischaemic lesion patterns were found among stroke subtypes (p=0.036, figure 3). In the LAA group, NC was the most common (49.9%), followed by PC (27.9%) and TC (22.2%). In the CE group, NC (46.4%) and TC (42.9%) were common, with PC accounting for 10.7% of occurrences. Compared with the LAA group, the CE group had more TC and fewer PC patterns (p=0.037). However, the proportions of NC patterns were similar between the two groups. The patients classified as the UN group showed predominantly NC (88.9%), with PC at 11.1%, and no patient with a TC pattern. Among PC patterns, the reversed wedge pattern was infrequent compared with the branch occlusion or cortex dominant patterns (table 1). Superficial and deep subcortical structures appeared to be equal in the NC group.
Factors affecting the extent of cortical involvement
The demographics, risk factors and laboratory findings were not significantly different for the ischaemic lesion patterns (table 2). Significant stenosis of the ipsilateral ACA or PCA was more frequent in patients with the TC pattern than in those with a PC or NC pattern (p=0.011). However, significant stenosis of the ipsilateral ICA was not associated with a difference in infarct pattern among the three groups (p=0.767).
Discussion
Several studies have investigated the patterns of infarctions in patients with stenosis or occlusion of the MCA, to speculate on aetiological mechanisms based on the patterns.16–20 This study is unique because only patients with a complete occlusion of the MCA were enrolled, and the infarct patterns were categorised by degree of cortical surface involvement. This approach allowed us to investigate whether functionally active arteriolar connections were present between leptomingeal arteries, which may substantially affect the extent of infarction following MCA occlusion. Cortex-sparing infarctions appeared to be more common than classic full-territory infarctions in patients with symptomatic MCA occlusions, because 72.6% of patients with an MCA occlusion had a cortex-sparing infarction. The higher frequency of cortex-sparing in patients with an MCA occlusion might be because LMA between the cerebral arteries salvaged the cortical areas in the vicinity of the MCA. We also demonstrated that the extent of cortex involvement in patients with an MCA occlusion may be critically affected by the haemodynamic status of the ACA and PCA, which are routes of leptomenigeal anastomosis to the MCA.
In MCA occlusion, LMA is considered an important factor in maintaining adequate cerebral flow to the penumbral area.3 4 Previously, anatomical and radiological studies indicated the existence of LMA by demonstrating multiple anastomoses between cerebral arteries on the surface of the brain.3–10 However, some researchers questioned the presence of functionally active LMA because their small diameters were assumed to be insufficient for providing adequate communication between the arteries.2 11 For this reason, we assessed the extent of the cortical involvement of ischaemic lesions, presuming this would reflect the compensatory function of LMA, since the cerebral cortex is supplied by leptomeningeal arteries.21 The results showing a high frequency of cortex-sparing infarctions in patients with MCA occlusions supported the hypothesis that the cerebral arteries are functionally interconnected by LMA.
We noted that only 27.4% of patients with symptomatic MCA occlusion demonstrated an infarction involving the full MCA territory. This frequency is similar to a previous study, which found territorial infarction in 30.3% of patients with MCA occlusion.22 In contrast, the cortex was spared completely in almost half of the patients with MCA occlusion in this study, and their proportions were not different between the LAA and CE groups. This finding indicates that collateral channels reacted promptly, regardless of the rate of arterial closure, after the cessation of blood flow, and strongly suggests that sufficient LMA was already present over the surface of the cortex in the MCA territory.
In this study, the PC pattern was more frequently observed in the LAA group than in the CE group, while the TC pattern was more frequent in CE. The higher proportion of TC in the CE group compared with the LAA group may be because the time to develop collateral pathways or ischaemic tolerance in CE was insufficient due to sudden arterial occlusion.18 19 23–25 Furthermore, the higher frequency of PC in the LAA group may be explained by the growth of functionally active formations of collateral circulation. Chronic hypoperfusion is known to enlarge pre-existing arterioles through arteriogenesis.26–28 Significant stenosis of the parent artery in LAA might induce such an arterial remodelling mechanism prior to MCA occlusion, resulting in partial salvage of the cortex after the ischaemic injury.
We found that significant stenosis of the ipsilateral ACA or PCA was associated with the extent of cortical involvement. In MCA occlusion, the main sources of leptomeningeal arteries were ipsilateral ACA and PCA.1 Thus, haemodynamic compromise of these vessels is expected to influence the extent of cortical involvement in patients with MCA occlusion.
Limitations of this study that must be considered are, first, the small sample size. Although the number of patients who were initially considered for this study was large, many were excluded based on a violation of study criteria. Second, the functional status of leptomeningeal arteries was assessed by an indirect method, specifically, comparing the extent of cortical involvement by DWI. The presence of collateral arteries, including leptomeningeal collaterals, may be visualised directly by DSA.28 However, the compensatory function of collateral vessels was difficult to estimate by this method. Salvage from the infarction, which was used as the determinant in this study, may provide definitive proof of the presence of functionally active LMA. Finally, we could not ascertain whether stenosis of the ipsilateral ACA or PCA was pre-existing before an event stroke or occurred concomitantly as the result from multiple emboli.
The presence of cortex-sparing in a stroke patient with MCA occlusion has been occasionally considered as the marker of chronic occlusive disease of the parent artery. This presumption seems partly valid when the cortex is involved because the extent of cortical involvement was different between CE and LAA in our study. However, in cases without cortical surface involvement (totally spared cortex), it could be due to LMA that was already present. From our findings, many patients seem to have sufficiently pre-existing LMA over the surface of the cortex in the MCA territory. Therefore, this point should be considered when one speculates on stroke mechanisms or plans flow augmentation treatments such as extracranial–intracranial bypass surgery.
References
Footnotes
Funding This work was supported by a grant from the Korea Healthcare Technology R&D Project, Ministry for Health, Welfare & Family Affairs, Republic of Korea (A080602).
Competing interests None.
Ethics approval Ethics approval was provided by the Institutional Review Board of Severance Hospital, Yonsei University Health System.
Provenance and peer review Not commissioned; externally peer reviewed.