Diffusion-Weighted Magnetic Resonance Imaging in Internal Carotid Artery Dissection FREE

Analysis of infarct patterns on computed tomography and magnetic resonance (MR) imaging has led to deeper insights into the mechanisms of cerebral ischemia. Diffusion-weighted imaging (DWI) allows the identification of small ischemic lesions, which may not be detected by computed tomography or conventional MR imaging.^1- 2 Multiple acute ischemic lesions, also known as acute multiple brain infarction (AMBI), on DWI are associated with arterial or cardiac sources of embolism.^3- 4 This pattern is believed to arise from repeated embolism or the fragmentation of a single embolus, resulting in at least 2 noncontiguous areas of cerebral infarction.^4- 5

The mechanism of ischemia in internal carotid artery (ICA) dissection remains controversial. Sudden narrowing of the arterial lumen in dissections may result in rapid hemodynamic compromise, leading to infarction. Weiller et al⁶ found an equal role of hemodynamic and embolic mechanisms in ICA dissection. Two other studies^7- 8 highlighted the importance of embolism. If, indeed, embolism is the predominant mechanism of cerebral ischemia in ICA dissection, it is likely that DWI will detect the presence of AMBI in this setting. We therefore undertook this study to review the DWI characteristics of patients with ICA dissection.

Patients admitted to Jackson Memorial Hospital, Miami, Fla, between August 1, 1999, and February 28, 2003, with a diagnosis of ICA dissection were identified by review of an angiographic database and of hospital discharge codes (International Classification of Diseases, Ninth Revision, Clinical Modification, codes 433 and 434). We reviewed the medical charts and radiological data (catheter cerebral angiograms, neck MR angiograms, and neck MR imaging with fat-suppressed images) of patients diagnosed as having ICA dissection by the treating physicians to confirm the primary diagnosis. Patients were included if DWI was obtained within 10 days of symptom onset. Penetrating injury to the neck as a cause of dissection was an exclusion criterion.

Diffusion-weighted imaging studies were reviewed for infarct patterns. Vascular territories were determined according to established vascular maps, and border zones were divided into the anterior, posterior, and deep internal watershed areas.^9- 10 Infarctions were examined for the presence of AMBI and border zone involvement. Acute multiple brain infarction was defined as 2 separate DWI-positive lesions within the same or different vascular territories. All DWIs were independently interpreted by 2 observers blinded to the clinical data. In cases of discrepancy, a consensus reading was obtained.

We identified 26 patients with ICA dissection. Diffusion-weighted imaging was obtained in 21 patients, and 14 had evidence of cerebral infarction. The clinical features, infarct patterns, and radiological data of this patient group are presented in Figure 1. The diagnosis of dissection was confirmed by intra-arterial angiography (n = 12) and MR angiography with axial MR imaging (n = 2). The mean patient age was 46 years (range, 28-59 years). Nine patients (64%) were men. Hypertension was present in 5 (36%) of the 14 patients. Neck pain or headache on presentation was noted in 4 (29%) of 14. Oculosympathetic paresis was found in 2 of 14.

Internal carotid artery occlusion was present in 10 (71%) of the 14 patients. The right ICA was involved in 8 (57%) of 14 patients. One patient (patient 7) experienced bilateral ICA dissections.

Acute multiple brain infarction was found in 10 (71%) of 14 (patients 1-10). A single infarct was detected in 4 (29%) (patients 11-14). Cortical involvement was found in 8 (57%) (patients 1-3, 8-10, 12, and 14). The infarct was restricted to the subcortical region in 6 (43%) (patients 4-7, 11, and 13). In 9 (64%) (patients 1-8 and 10), ischemic lesions were located in 1 of the 3 border zones. The ICA was occluded in 6 (60%) of 10 patients with AMBI pattern and in 4 (100%) of 4 with a single infarct on DWI. In patients with border zone involvement, 7 of 9 patients had an occluded ICA, compared with 4 of 5 patients without border zone lesions.

In our series of patients with ICA dissection, DWI detected AMBI in 10 (71%) of the 14 patients. Border zone involvement was present in 9 (64%) of 14. Two previous studies^7- 8 described computed tomography and conventional MR infarct patterns in patients with ICA dissection. Steinke et al⁸ found evidence of embolic infarcts, defined as territorial middle cerebral artery or large striatocapsular infarcts, in 70% of patients with ICA dissection. In 16% of patients, the area of infarction was located in a border zone. Lucas et al⁷ found areas of low-flow involvement in 8% of patients with ICA dissection and a stroke pattern suggestive of embolism, using criteria identical to those of Steinke et al, in 92%. Both groups of authors speculated that the incidence of border zone infarction was overestimated because of variability in the vascular anatomy and the misclassification of branch artery occlusion as watershed infarcts.

In the present study, the high incidence of AMBI confirms the importance of embolism in ICA dissection–related strokes. However, we found an increased frequency of border zone involvement compared with the previous series.^7- 8 In fact, at least 1 lesion of the AMBI was located in a border zone in 9 (64%) of 14 patients. The discrepancy between this finding and those noted by the other 2 author groups may be partially explained by differences in imaging techniques. The 2 previous studies used computed tomography and conventional MR imaging, which have less sensitivity in the detection of small ischemic lesions compared with DWI.^1- 2 In addition, differences in methods and definition of low-flow infarction, as well as severity of the underlying steno-occlusive vascular disease, with a high number of ICA occlusions being present in this series (71% of patients), may contribute to the discrepancy.

We are not aware of any previous studies analyzing DWI infarct patterns in ICA dissection. Therefore, it may be instructive to compare our findings with those of DWI series of strokes related to atherosclerotic ICA disease. In this condition, an AMBI infarct pattern has been reported to be present in at least 58% to 83% of patients.^11- 13 These findings are similar to the results of the present study.

Szabo et al¹² noted small lesions in hemodynamic risk zones in 33 (32%) of 102 patients. Kang et al¹³ found border zone involvement, with or without the presence of a territorial lesion, in 10 (29%) of 35 patients. The latter study used methods similar to ours, allowing for a more direct comparison of infarct pattern. The incidence of border zone involvement in that study is lower than the 64% noted in our series, despite the presence of high-grade (>90%) ICA stenosis or occlusion in 86% of patients.

It is likely that in ICA dissection hemodynamic factors assume greater importance as diminished perfusion and embolism interact to cause cerebral infarction. This interaction has been well described recently.¹⁴ Decreased perfusion in low-flow zones may fail to wash out emboli. Cerebral infarction may occur more readily in these areas when cerebral embolism takes place.¹⁴ Indeed, unilateral watershed infarcts have been associated with embolism,^15- 16 and following cardiac surgery, territorial and border zone lesions are present simultaneously.¹⁷ In addition, border zone infarcts diagnosed by DWI are associated with transcranial Doppler–detected microemboli in patients with middle cerebral artery stenosis.¹⁸

The higher frequency of border zone involvement in strokes secondary to ICA dissection, compared with ICA atherosclerotic disease, may be related to a more sudden compromise of cerebral blood flow. Collateral flow may not establish itself as rapidly, leading to greater perfusion loss and increased susceptibility to embolism in the border zone areas.

In summary, ICA dissection should be added to the list of stroke etiologies associated with an AMBI pattern on DWI. Acute multiple brain infarction was found in 71% of patients with ICA dissection. At least 1 ischemic lesion was seen in a border zone area in 64% of patients. These findings may be further evidence for an interaction of hemodynamic and embolic mechanisms in strokes related to ICA dissection.

Corresponding author and reprints: Sebastian Koch, MD, Department of Neurology, University of Miami School of Medicine, Professional Arts Center, 1150 NW 14th St, Suite 304, Miami, FL 33136 (e-mail: skoch@med.miami.edu).

Accepted for publication October 31, 2003.

Author contributions: Study concept and design (Drs Koch and Rabinstein); acquisition of data (Drs Koch, Rabinstein, and Forteza); analysis and interpretation of data (Drs Koch, Rabinstein, Romano, and Forteza); drafting of the manuscript (Dr Koch); critical revision of the manuscript for important intellectual content (Drs Koch, Rabinstein, Romano, and Forteza); administrative, technical, and material support (Dr Koch); study supervision (Drs Koch, Romano, and Forteza).