Carotid intraplaque hemorrhage detected by magnetic resonance imaging predicts embolization during carotid endarterectomy

Methods 

Magnetic resonance imaging scanning and analysis 

After a baseline assessment at the clinic, all participants underwent MRI scanning of the carotid arteries to determine the presence and extent of intraplaque hemorrhage (Fig). MRI was performed on one of three 1.5T scanners: Vision (Siemens Medical, Erlangen, Germany), Intera (Philips, Best, Netherlands), or Signa (General Electric [GE], Waukesha, Wis). All participants were scanned using a receive-only quadrature neck array cervical spine coil. Carotid imaging was performed using a coronal T1-weighted magnetization-prepared three-dimensional gradient echo sequence with a water excitation pulse to remove signal from fat in the Siemens and Philips scanners; and fat suppression was achieved by using the Spectral Inversion at Lipids (SPECIAL) option in the GE scanner: Siemens (TR 10.3 ms, TE 4.0 ms, FA 15°, effective TI 740 ms, voxel size 1.17 × 1.17 × 1.14 mm, acquisition matrix 256 × 256 with 62.5% rectangular FOV, 140 partitions, acquisition time 235 seconds); Philips (TR 8.4 ms, TE 3.1 ms, FA 10°, effective TI 740 ms, voxel size 1.1 × 1.1 × 1.1 mm, acquisition matrix 256 × 256 with 72.5% rectangular FOV, 140 partitions, acquisition time 255 seconds); and General Electric (TR 5.8 ms, TE 1.5 ms, FA 15°, TI 19 ms, interpolated voxel size 0.66 × 0.66 × 1 mm, acquisition matrix 256 × 256 [interpolated to 512 × 512] with 100% rectangular FOV, 72 partitions, acquisition time 251 seconds).

View Large Image Download to PowerPoint

Fig. Axial section of bilateral carotid plaques with evidence of intraplaque hemorrhage (arrows).

Image analysis was performed using standard image reconstruction techniques as provided by JAVA Imaging software (Xinapse Systems Ltd, Thorpe Waterville, UK) to assess axial reformats and perform signal intensity assessments of the intraplaque hemorrhage signal. The investigators were blinded to the clinical and the transcranial Doppler data.

Intraplaque hemorrhage was identified by two trained experienced researchers if the maximal signal intensity within the carotid plaque was >150% of the maximum adjacent muscle intensity. The presence or absence of a high signal was recorded in the arteries by the researchers blinded to the symptomatic artery and microemboli findings. The researchers considered six scans were equivocal, and a consensus was then reached between the two trained researchers and an experienced neuroradiologist blinded to the clinical data. Intraobserver and interobserver variability assessed on 60 arteries using this procedure has excellent agreement (Cohen κ = 0.88 and 0.81, respectively).

Results 

Initially, 86 participants were recruited, but six were unable to complete the MRI scan due to claustrophobia and one was withdrawn when a subsequent assessment indicated that his presenting symptoms were unrelated to the carotid artery disease. A further 11 participants did not undergo an operation: five had carotid artery occlusions either at the time of MRI or whilst waiting for carotid endarterectomy, three had large strokes pending the operation, and three opted for conservative management. Details of the clinical outcomes of this cohort are published elsewhere.19 A further eight patients had a poor temporal acoustic window for transcranial Doppler signal acquisition. The MRI scans of the remaining 60 participants were of suitable quality for assessment of intraplaque hemorrhage and were analyzed. The mean age of this group was 69.8 ± 10.5 years, and 17 (28.3%) were women.

MRI signs of intraplaque hemorrhage were found in 36 (60%) of the symptomatic carotid arteries but not in 24 (40%). There were no significant differences in age, atherosclerotic risk factors, and delay between symptom and MRI between those who demonstrated intraplaque hemorrhage and those who did not, but the proportion of men was higher in the intraplaque hemorrhage group (Table I).

Table I. Risk factors and demographics of patients with and without intraplaque hemorrhage in the ipsilateral carotid plaque

	Factor	Intraplaque hemorrhage⁎		P
		Present, n (%)	Absent, n (%)
	Patient total	36(60)	24(40)
	Age, years	69.0±8.7	66.9±11.4	.4
	Sex (females)	6(16.7)	11(45.8)	.02
	Hypertension	31(86.1)	12(50)	.3
	Ischemic heart disease	9(25)	7(29.2)	.5
	Previous MI	7(19.4)	3(12.5)	.5
	Diabetes mellitus	5(13.9)	1(4.2)	.2
	Smoker	10(27.8)	11(45.8)	.4
	Statin use	37(75.0)	20(83.3)	.3
	Antiplatelet drugs
	None	0	1(4.2)	.4
	Aspirin	26(72.2)	13(54.2)
	Aspirin + dipyridamole	7(19.4)	5(20.8)
	Clopidogrel	1(2.8)	1(4.2)
	Heparin	1(2.8)	1(4.2)
	Presenting symptom
	Stroke	12(33.3)	7(29.2)	.8
	TIA	17(47.2)	9(37.5)
	Amaurosis fugax	7(19.4)	8(33.3)
	CEA dissection time, minutes	41.4±9.7	40.5±9.7	.8
	Time between
	Last symptom and MRI	26.0(12.3to62.0)	33(22.5to75.5)	.2
	MRI and CEA	19.0(8.0to34.5)	16.5(8.3to31.3)	1

MI, Myocardial infarction; TIA, transient ischemic attack; CEA, carotid endarterectomy; MRI, magnetic resonance imaging.

⁎ Continuous variables are presented as mean±standard deviation or median (interquartile range).

During the dissection phase (41.1 ± 11.0 minutes), microembolic signals (range, 1 to 64 signals) were detected in the ipsilateral middle cerebral artery in 23 (38.3%) participants. Nineteen of the plaques with intraoperative microembolic signals had MRI intraplaque hemorrhage and four did not (OR, 5.6; 95% CI, 1.6 to 19.7, P = .007). By contrast, none of the patient factors were significantly associated with microembolic signals (Table II). The sensitivity of predicting microembolic signals using the MRI assessment of intraplaque hemorrhage was 82.6%, the specificity was 54.1%, the positive predictive value was 52.8%, and the negative predictive value was 83.3%.

Table II. Univariate logistic regression analysis results of factors affecting the presence of dissection phase microemboli

	Factor	OR (95% CI)	P
	Age	1.0(1.0to1.1)	.6
	Sex (females)	0.6(0.2to1.9)	.4
	Hypertension	1.2(0.3to4.7)	.8
	Ischemic heart disease	0.9(0.3to3.1)	.9
	Diabetes mellitus	0.8(0.1to4.7)	.8
	Smoker	0.9(0.3to2.9)	.8
	Degree of stenosis	0.9(0.5to1.9)	.9
	Time from last symptom and TCD recording during CEA	0.99(0.98to1.00)	.1
	Presence of intraplaque hemorrhage	5.6(1.6to19.6)	.007

OR, Odds ratio for embolization; CI, confidence interval; TCD, transcranial Doppler; CEA, carotid endarterectomy.

In the multivariate logistic regression analysis, intraplaque hemorrhage conferred a significant independent risk for microembolic signals (adjusted OR, 5.8; 95% CI, 1.1 to 30.4, P = .037) after controlling for age, sex, degree of stenosis, individual surgeon, and delay from symptom to carotid endarterectomy.

In the 36 patients with intraplaque hemorrhage, no additional factors (including age, sex, stenosis, ischemic heart disease and hypertension) significantly altered the risk of microembolization (P varied from .3 to .7).

Postoperatively, only one participant developed a minor stroke. This person had intraplaque hemorrhage and demonstrated intraoperative microembolic signals.

Discussion 

Patients with high-grade symptomatic carotid stenosis and preoperatively determined carotid intraplaque hemorrhage had an elevated risk of microembolization during the dissection phase of CEA. This relationship remained significant after adjusting for other known risk factors.

The results of this study provide further evidence that carotid intraplaque hemorrhage, as detected by MRI, can identify active thromboembolic plaque. There has been uncertainty over whether and how intraplaque hemorrhage is associated with plaque instability.25 Lusby et al26 initially noted that plaque hemorrhage was more prevalent in symptomatic than in asymptomatic carotid plaques. In the largest pathologic study of symptomatic carotid plaques, Redgrave et al13 demonstrated that intraplaque hemorrhage was associated with morphologic features of plaque instability. Intraplaque hemorrhage has been suggested to stimulate atherogenic activity by being a source of free cholesterol to the necrotic core and macrophage activation.12 A small study by Sitzer et al27 stated that plaque ulceration and not intraplaque hemorrhage was associated with significant spontaneous embolization (defined as >5 microembolic signals per hour), but in fact, seven of eight patients in that study with plaque hemorrhage had some microembolization (>1 microembolic signal).

By using this MRI technique to assess the carotid plaque, we were able to demonstrate a high negative predictive value of microembolization during the dissection phase. The absence of intraplaque hemorrhage therefore seems to be a strong indicator of a carotid plaque that is unlikely to cause embolization during surgery and, hence, is at a lower perioperative risk. Only about half (53%) of carotid plaques with intraplaque hemorrhage had any embolization, however. Presumably, additional factors (such as fibrous cap integrity of the plaque) or platelet/fibrin aggregation may also trigger intraoperative microembolization. Further studies using the more time-consuming technique of multisequence MRI28 to detect other features of plaque, or using ultrasound scanning, are warranted to see if other factors add to prediction based upon the presence of intraplaque hemorrhage alone. As in the rest of the United Kingdom,29 the delay from the symptomatic event to surgery in our unit was significant. The MRI findings may have changed in this interim period; however, intraplaque hemorrhage appears to remain stable over some time after a symptomatic event.30

Because only one clinical perioperative event occurred in this series, we cannot claim that intraplaque hemorrhage predicts higher perioperative stroke. Perioperative microembolic signals in the dissection phase are known to predict postoperative ischemia, however,8, 9, 10, 31 and so our findings are likely to be of clinical importance. By being forewarned by the presence of intraplaque hemorrhage of the possibility of perioperative morbidity, appropriate measures can be taken to optimize the management of the patient. Preoperative measures include the use of anticoagulants, combinations of antiplatelet agents,32 and intravenous dextran,33 which decrease microembolic signal counts and thereby potentially decrease the preoperative and perioperative risk of stroke.

A technique that allows preoperative identification of those plaques at high risk of embolization could guide the choice of treatment modality and technique. For example, it may be prudent for trainee surgeons not to undertake procedures in patients with intraplaque hemorrhage until they are adequately experienced. For experienced surgeons, earlier clamping of the internal carotid artery before completion of the dissection of the carotid arteries, which increases the carotid clamp time and may increase the risk of iatrogenic cerebral ischemia, may be warranted in those with intraplaque hemorrhage who are at higher risk of embolism during the dissection procedure. It remains to be seen whether this technique is of any value in patients undergoing carotid artery stenting, during which microembolization is of some concern.

A change of operative technique might have occurred once the surgeon was alerted to the presence of microembolic signals, but this would influence the number of subsequent microembolic signals and not their presence in the first instance and, hence, does not invalidate our conclusions. In this study we have not reported spontaneous preoperative or postoperative microembolic signals, both of which are also of clinical significance. However, in choosing to report the relationship between MRI intraplaque hemorrhage and microembolic signals during the dissection phase of carotid endarterectomy, we studied outcomes that are both clinically important and sufficiently frequent for statistical testing to be done on the results.

Conclusion 

Carotid intraplaque hemorrhage, as detected by MRI, is associated with an increased risk of microembolization during the dissection phase of carotid endarterectomy. Determination of intraplaque hemorrhage using MRI may be useful to stratify risk when considering interventions for carotid artery disease and to optimize antithrombotic and other preventive strategies during surgery to reduce operative ischemia.

Author contributions