Early Recurrent Ischemic Lesions on Diffusion-Weighted Imaging in Symptomatic Intracranial Atherosclerosis

Background  Prior observations have shown that early recurrent ischemic lesions (ERILs) on diffusion-weighted imaging occur frequently within the first week after an index stroke.

Objective  To investigate differential patterns of ERILs among stroke subtypes, particularly intracranial large-artery atherosclerosis (IC-LAA).

Design  Retrospective study.

Setting  Tertiary university hospital.

Patients  We included 133 patients who experienced an acute ischemic stroke and who underwent initial diffusion-weighted imaging within 24 hours and subsequent diffusion-weighted imaging within 7 days after onset, and whose stroke subtype was IC-LAA, extracranial LAA (EC-LAA), or cardioembolism (CE).

Main Outcome Measure  Early recurrent ischemic lesions were defined as new ischemic lesions on follow-up diffusion-weighted imaging, separate from the index stroke lesion.

Results  Early recurrent ischemic lesions were observed in the following proportions: 50.9% (28/55) in the IC-LAA group, 47.4% (9/19) in the EC-LAA group, and 44.1% (26/59) in the CE group. Early recurrent ischemic lesions in the IC-LAA group had the following characteristics: (1) they occurred mostly (27 [96.4%] of 28) in the pial area of the same vascular territory as the index stroke; (2) they were more frequently observed in a higher grade of stenosis than in milder stenosis (P<.001), whereas ERILs in the EC-LAA group were not related to the degree of stenosis; (3) they were not associated with subsequent recanalization, whereas ERILs in the CE group were mostly associated with subsequent recanalization (P<.001); and (4) they were more closely associated with clinical recurrence than in the EC-LAA or CE group (P=.02).

Conclusion  Early recurrent ischemic lesions in the IC-LAA group are relatively frequent and have different patterns than in the EC-LAA or CE group.

Clinically silent early recurrent ischemic lesions (ERILs) on diffusion-weighted imaging (DWI) are much more frequent than the clinical recurrence within the first week and up to 1 to 3 months after the onset of index stroke.^1-3

The risk of early recurrent stroke is likely related to the underlying causative stroke subtypes. Population-based studies^4,5 have reported that patients with large-artery atherosclerosis (LAA) had the highest risk of early clinical recurrent stroke, and the DWI study¹ showed that LAA was the most frequent stroke subtype associated with ERILs. These studies, however, were performed in Western countries, in which the prevalence of intracranial LAA (IC-LAA) is much lower than that of extracranial LAA (EC-LAA), and they did not distinguish between IC-LAA and EC-LAA. Although the classic Trial of Org 10172 in Acute Stroke Treatment classification has lumped IC-LAA and EC-LAA together,⁶ there is evidence that the 2 types of stroke differ in ethnicity, risk factors, natural history, mechanism of index stroke, and long-term recurrence.^7-9 We, therefore, sought to determine the characteristic pattern and mechanism of ERILs in IC-LAA compared with other stroke subtypes, EC-LAA, and cardioembolism (CE).

This is a retrospective analysis of all consecutive stroke patients admitted to the Stroke Center at Asan Medical Center, between November 1, 2002, and August 31, 2004. Patients were included if they had (1) a final diagnosis of ischemic stroke confirmed by DWI, (2) initial DWI performed within 24 hours and subsequent DWI performed within 1 week (mostly 3-5 days) after symptom onset, and (3) IC-LAA, EC-LAA, or CE corresponding to the acute infarct. Patients were excluded for the following reason: (1) if they had small-vessel occlusion (because of their low incidence of ERILs¹), other determined or undetermined stroke cause, or tandem IC-LAA and EC-LAA; or (2) if they underwent diagnostic or therapeutic neurointerventional procedures (diagnostic cerebral angiography, intra-arterial thrombolysis, or stenting) between initial and follow-up DWI scans. However, those who received intravenous thrombolysis were included. The onset time was defined as the time patients were last known to be without ischemic symptoms. During the 1-week study, all patients received secondary stroke prevention therapy (antiplatelet agents, anticoagulants, and/or statins), as standardized by our center's stroke care pathway. This study was approved by the institutional review boards of the Asan Medical Center.

The protocol for this study includes DWI, 3-dimensional time-of-flight magnetic resonance angiography (MRA), and 3-dimensional contrast-enhanced MRA. The detailed magnetic resonance imaging protocol has been previously described.¹⁰

Ischemic lesions on DWI were analyzed by one of us (D.-W.K), who was blinded to clinical and MRA data and stroke subtypes. Early recurrent ischemic lesions were defined as new separate lesions on follow-up DWI outside the region of the acutely symptomatic (index) lesion and which were not detected on initial DWI. We classified each ERIL as occurring in the “same” or “different” vascular territory (VT) or cerebral circulation as that of the index ischemic lesion. One patient who had ERILs in the same and different VT and cerebral circulation was classified as having ERILs in a different VT and cerebral circulation. For IC-LAA, the location of the ERILs was classified as occurring in the pial territory, perforating artery territory, and cortical or internal border zone. The number of recurrent lesions was counted. Ischemic lesion volume was measured by another investigator (K.-Y.K) blinded to clinical information and stroke subtypes, as infarct volume (measured in milliliters) is equal to the sum of the infarct area on each DWI slice (slice thickness + interslice gap).

Magnetic resonance angiograms were independently reviewed by another investigator (S.U.K.) blinded to the clinical and DWI data. The degree of intracranial arterial stenosis was semiquantitatively classified as normal (grade 0), mild (signal reduction, <50%) (grade 1), moderate (signal reduction, ≥50%) (grade 2), severe (focal signal loss with the presence of a distal flow signal) (grade 3), or occluded (grade 4). Recanalization of initially steno-occluded vessels (grades 2-4) was assessed as none (no change), partial (improvement in 1 or 2 grades of steno-occlusion, but still significant stenosis; ie, grade 2 or 3 on follow-up MRA), significant (complete visualization of distal branch arteries with residual stenosis of <50%; ie, grade 1 on follow-up MRA), or complete (grade 0 on follow-up MRA).

The stroke subtypes were determined according to the Trial of Org 10172 in Acute Stroke Treatment classification,⁶ with the following modifications. Intracranial LAA was diagnosed if the initial MRA showed symptomatic atherosclerotic intracranial arterial stenosis of 50% or more or occlusion, if follow-up MRA showed persistent steno-occlusion, and if neither a cardiac nor an extracranial arterial embolic source could be identified. Extracranial LAA was diagnosed if there was symptomatic extracranial arterial stenosis of 50% or more or occlusion without evidence of IC-LAA or CE. Stroke subtype was diagnosed by the consensus of 3 stroke neurologists (D.-W.K., S.U.K., and J.S.K.) blinded to ERIL results.

Clinical stroke recurrence within 1 week was identified by another investigator (S.-H.Y.) blinded to the magnetic resonance imaging data as “any recurrent stroke occurring >24 hours after onset of the index stroke, irrespective of VT.”^11(p1926) Systemic causes of clinical deterioration after an initial stroke (eg, infection) and worsening of initial symptoms because of the progression or hemorrhagic transformation of the initial stroke were not classified as clinical recurrences.

Baseline demographics, time to magnetic resonance imaging, risk factors, intravenous thrombolysis, anticoagulant vs antiplatelet therapy during the first week, and the proportion of patients with initial multiple infarcts were compared between those with and those without ERILs. The previously mentioned variables together with ERILs and clinical recurrence were also compared among groups of patients with stroke subtypes. The association between ERILs and the degree of stenosis in the LAA group, the association between ERILs and recanalization of initially occluded vessels, and the association between lesion volume (of total lesions and ERILs) and clinical recurrence were also explored. The Pearson χ² test with exact method, an analysis of variance, or a nonparametric Kruskal-Wallis test was used where appropriate. All statistical analyses were performed using a commercially available software program (SPSS for Windows, version 12.0; SPSS Inc, Chicago, Ill).

During the study, we identified 133 patients (80 men and 53 women; mean ± SD age, 65.6 ± 11.9 years) who met the eligibility criteria. The median time from onset to the initial DWI scan was 5.85 hours (range, 0.3-23.9 hours); and to follow-up DWI, 4.0 days (range, 1.03-6.82 days). Intracranial LAA was diagnosed in 55 patients, EC-LAA in 19 patients, and CE in 59 patients.

Early recurrent ischemic lesions were observed in 63 (47.4%) of the 133 patients. Baseline characteristics did not differ between patients with and without ERILs, except that the frequency of multiple infarcts on initial DWI was significantly higher in patients with ERILs (40 [63.5%] of 63 patients) than in patients without ERILs (24 [34.3%] of 70 patients) (P=.001). Antiplatelet or anticoagulant treatments did not influence the occurrence of ERILs (P=.94).

Baseline characteristics among stroke subtypes were comparable, except that time to initial DWI from onset was shorter in patients with CE and history of diabetes mellitus was more frequent in patients with EC-LAA. Early recurrent ischemic lesions were observed in 50.9% of the IC-LAA group, 47.4% of the EC-LAA group, and 44.1% of the CE group (percentages are “off” for the EC-LAA and CE groups using values given in Figure 1 because of rounding). The distribution of ERILs differed across stroke subtypes. Relative to index stroke lesions, ERILs in the same VT were most frequent in the IC-LAA group, whereas ERILs in different VTs or in different cerebral circulations were most frequently observed in the CE group (Table and Figure 1).

Of the 28 patients with IC-LAA who had ERILs, 18 had recurrent lesions in the pial territory, 5 in the pial territory and border zone, 3 in the pial and perforator territories, 1 in the pial and perforator territories and the cortical border zone, and 1 in the cortical border zone alone. Overall, 27 (96.4%) of these 28 patients developed recurrent lesions involving the pial territory.

Of the 55 patients with IC-LAA, 13 had moderate stenosis, 12 had severe stenosis, and 30 had occlusion on initial MRA; the frequency of ERILs was 0% (0/13), 75.0% (9/12), and 63.3% (19/30), respectively (P<.001). In contrast, of the 19 patients with EC-LAA, 6 each had moderate and severe stenosis and 7 had occlusion; the frequency of ERILs caused by index arterial lesions was not associated with the degree of stenosis (P=.33), being 66.7% (4/6), 33.3% (2/6), and 28.6% (2/7), respectively (Figure 2).

The association between recanalization and ERILs was compared in patients who showed intracranial steno-occlusive lesions on initial MRA and ERILs in the same VT (ie, in 28 patients with IC-LAA, 11 patients with CE, and 0 patients with EC-LAA). Early recurrent ischemic lesions were accompanied by no (21 [75.0%]) or partial (7 [25.0%]) recanalization in the IC-LAA group, but by complete (8 [72.7%]), significant (1 [9.1%]), partial (1 [9.1%]), and no (1 [9.1%]) recanalization in the CE group (P<.001) (Figure 3 and Figure 4).

Clinical recurrence within 1 week was observed in 8 (6.0%) of the 133 patients (7 with IC-LAA and 1 with EC-LAA). Early recurrent ischemic lesions in the IC-LAA group were more closely associated with clinical recurrence (7 [25.0%] of 28 patients) compared with ERILs in the EC-LAA group (1 [11.1%] of 9 patients) and CE group (0 patients). In patients with ERILs, the mean ± SD total infarct volume on initial (5.9 ± 4.9 vs 8.4 ± 12.0 mL; P=.79) and follow-up (14.7 ± 8.7 vs 17.0 ± 25.1 mL; P=.20) DWI and the mean ± SD volume of ERILs (5.0 ± 9.7 vs 1.8 ± 5.1 mL; P=.12) did not differ between patients with and without clinical recurrence. In addition, the volume of ERILs did not differ across stroke subtypes (Table).

We have shown herein that the patterns of ERILs in patients with IC-LAA had characteristics differentiating them from the patterns of ERILs in patients with EC-LAA or CE. Early recurrent ischemic lesions in those with IC-LAA occurred predominantly in the pial area of the same VT as the index stroke, were more frequently associated with a higher grade of stenosis, and were not accompanied by subsequent recanalization. Moreover, ERILs in those with IC-LAA were more closely associated with clinical recurrence than in those with EC-LAA or CE.

Although the frequency of recurrent stroke relative to stroke subtype has been reported,^4,5 less is known about the pattern and location of recurrent stroke according to stroke subtype. A recent long-term clinical follow-up study⁸ showed that in most patients with IC-LAA, clinically recurrent strokes occurred at the same site as the index stroke, whereas the pattern of recurrence was unpredictable in patients with EC-LAA. Our data are consistent with these findings and provide further evidence that silent recurrent ischemic lesions occur earlier and more frequently than clinically recurrent stroke.

Early recurrent ischemic lesions occurred more frequently in patients with IC-LAA who had a higher grade of stenosis, whereas the degree of stenosis was not related to ERILs in patients with EC-LAA. This finding suggests that different pathogenic mechanisms may be associated with the genesis of ischemic stroke. In those with IC-LAA, the progression of stenosis may be associated with an increased risk of recurrent ischemic stroke,¹² such that patients with high-grade stenosis are at greater risk of artery-to-artery embolization or hemodynamic failure than those with milder stenosis. In contrast, in those with EC-LAA, plaque heterogeneity rather than the degree of stenosis may be more important in pathogenesis.¹³ However, our results are limited by a lack of Doppler data regarding microemboli detection and plaque characteristics to support this hypothesis.

We also found a striking difference between the IC-LAA and CE groups in the development of ERILs within the same VT as the index stroke. In the IC-LAA group, ERILs occurred without subsequent recanalization, whereas in the CE group, ERILs were associated with significant recanalization. These results suggest that in patients with IC-LAA, ERILs may be caused by recurrent ischemic events, whereas in patients with CE, ERILs may result from fragmentation of the initial embolus. A previous study¹ also found that ERILs occurring within the initial perfusion defect (“local lesion recurrence”) were associated with subsequent reperfusion. In this group, ERILs may represent the natural evolution of the initial ischemic event. In the IC-LAA group, however, recurrent artery-to-artery embolism or hemodynamic failure may be a plausible mechanism for ERILs. Additional studies using serial DWI or MRA and microembolic signal monitoring are needed to determine the precise mechanism of ERILs in these patients.

In this study, ERILs in those with IC-LAA were more closely associated with clinical recurrence within the first week compared with other stroke subtypes. Patients with stroke resulting from IC-LAA are at high risk of recurrent stroke (>20% over 2 years), indicating a need for more effective therapies.¹⁴ Our results suggest that ERILs in those with IC-LAA may be a marker of increased risk of clinical recurrence, but additional studies are required to determine whether ERILs in those with IC-LAA are associated with a higher risk of subsequent clinical events.

This study has several limitations, including its retrospective design, the relatively small size of the EC-LAA group, and the lack of long-term magnetic resonance imaging and clinical follow-up data. Although we performed repeated MRA for the diagnosis of IC-LAA, some embolic occlusion might be misclassified as atherosclerotic occlusion.

Correspondence: Dong-Wha Kang, MD, PhD, Department of Neurology, Asan Medical Center, University of Ulsan College of Medicine, 388-1 Pungnap 2-dong, Songpa-gu, Seoul 138-736, Korea (dwkang@amc.seoul.kr).

Accepted for Publication: July 12, 2006.

Author Contributions:Study concept and design: Kang and J. S. Kim. Acquisition of data: Kang, Yoo, K.-Y. Kwon, Choi, S. J. Kim, and Koh. Analysis and interpretation of data: Kang and S. U. Kwon. Drafting of the manuscript: Kang. Critical revision of the manuscript for important intellectual content: S. U. Kwon, Yoo, K.-Y. Kwon, Choi, S. J. Kim, Koh, and J. S. Kim. Statistical analysis: Kang and S. U. Kwon. Obtained funding: Kang. Administrative, technical, and material support: Kang, Choi, S. J. Kim, and Koh. Study supervision: S. U. Kwon and J. S. Kim.

Financial Disclosure: None reported.

Funding/Support: This study was supported by grant A050586 from the Korean Health 21 R&D Project, Ministry of Health and Welfare.