Normalized intraplaque hemorrhage signal on MP-RAGE as a marker for acute ischemic neurological events

Abstract

Purpose

This study sought to validate whether the signal intensity ratio (SIR) of carotid intraplaque hemorrhage (IPH) was associated with acute ischemic neurologic events.

Methods

A retrospective review was completed of consecutive patients that underwent neck magnetic resonance angiography using magnetization prepared-rapid gradient echo (MP-RAGE) and T1-CUBE sequences between 2017 and 2020. Patients with magnetic resonance evidence of IPH were included. SIRs were measured by comparing the maximum IPH signal with the mean intramuscular signal from the adjacent sternocleidomastoid. Patients were stratified into ischemic or non-ischemic groups based on the presence of acute ipsilateral ischemic events (stroke, retinal artery occlusion). Logistic regression analysis was performed to determine if increasing IPH SIR was associated with an increased risk of ipsilateral ischemic events.

Results

Of 85 included patients (85 arteries), 66 were male (77.6%). Mean age was 71.0 (SD ± 11.1). There were 70 arteries with IPH that were ipsilateral to an ischemic event, and 15 that belonged to patients without an ischemic event. No association was found between increasing IPH SIR seen on MP-RAGE (odds ratio (OR): 0.82; 95% confidence interval (CI): 0.58–1.4; P = 0.43) or T1-CUBE sequences (OR: 0.85; 95% CI: 0.53–1.5; P = 0.56).

Conclusions

There was no association between the SIR of IPH and acute ischemia on either MP-RAGE or T1-CUBE sequences. Further investigation is required prior to widespread acceptance of SIR as a predictive imaging marker of symptomatic carotid plaque.

Introduction

Carotid artery atherosclerosis is a well-known risk factor for cerebral ischemic events including stroke and ocular ischemia. Multiple magnetic resonance (MR) characteristics of atherosclerotic plaques are known to be markers of instability, including ulceration and lipid-rich necrotic cores. The presence of one or more of these features suggests that a plaque may be vulnerable to rapid changes such as thrombosis. As such, evaluating for vulnerable plaque features is often done to assess the potential risk of future ischemic events. ¹

Intraplaque hemorrhage (IPH) is one such vulnerable plaque MR feature that has a particularly strong association with distal ischemic events.^2–4 A recent meta-analysis which included nearly 700 patients demonstrated that, in addition to severe carotid stenosis, IPH was an independent predictor of ischemic stroke. ⁵ The precise pathophysiological process by which IPH leads to ischemic neurologic events is not completely understood. Some authors have postulated that the mere presence of IPH alone may be inadequate in classifying carotid plaques as culprit lesions in patients with embolic stroke without an obvious source, because IPH may persist for years at a time without progression.^6–9 This has prompted investigators to consider more specific MR-based features of IPH as markers of symptomatic plaques. One feature of recent interest has been the signal intensity ratio (SIR): the signal intensity of the IPH normalized to the signal intensity of the adjacent neck musculature.

Recent reports have suggested that higher SIR in carotid IPH is associated with more frequent ischemic manifestations.^10,11 If such an association did exist, higher SIR may be a clue to the underlying pathomechanics of IPH and ischemic events. Moreover, SIR could also be used as a more specific marker for culprit lesions, as compared with the mere presence of IPH alone. To date, however, there exists a paucity of data to provide sufficient evidence that the SIR of carotid IPH is associated with ipsilateral stroke. As such, this study set out to assess the SIR of IPH in patients with and without acute ischemia in order to interrogate whether a relationship between higher SIR and acute ischemic cerebral events exists.

Methods

Patients and inclusion criteria

This was a retrospective cross-sectional study. Institutional review board approval was obtained prior to the initiation of this study. All patients included in our study provided written informed consent for participation in research activities at our institution. The medical and imaging records of patients who were evaluated at our institution and underwent magnetic resonance angiography (MRA) imaging of the neck that included magnetization prepared-rapid gradient echo (MP-RAGE) (used for evaluation of IPH) and T1-CUBE sequences between 2017 and 2020 were retrospectively reviewed. Only patients with imaging evidence of IPH on MP-RAGE sequences, defined as plaque SIR 1.5 times the signal of the adjacent sternocleidomastoid muscle, were included in the final analysis. Patients were excluded for the following reasons: (a) the MRA protocol did not include MP-RAGE or T1-CUBE sequences; (b) the images obtained were of poor quality and judged to be clinically non-diagnostic; (c) the MP-RAGE sequences did not demonstrate any evidence of carotid IPH; or (d) the SIR was below 1.5 (in which case, imaging criteria for IPH is not met). ³

Data abstracted from each patient chart included age, sex and history of ischemic events. All patients who presented to our institution were evaluated by a neurologist or neurosurgeon. In order to compare SIRs between patients with and without ischemic events, patients were stratified based on history of an acute ischemic event within a period of 6 weeks or less prior to MR imaging of the neck. Ischemic events were only considered if ipsilateral to the side of IPH and were defined as follows: ischemic stroke as evidenced on imaging studies, transient ischemic attack as diagnosed by a staff neurologist at our institution, retinal artery occlusion diagnosed with ophthalmoscopic examination by a staff neurologist or ophthalmologist, and amaurosis fugax as diagnosed by a staff neurologist or ophthalmologist. SIRs of IPH were compared between arteries ipsilateral to an acute ischemic event (any of the aforementioned events, collectively) and those without an ipsilateral event in patients without a documented history of any of the aforementioned events, collectively. A single artery with IPH was used for each patient included. In cases of patients with bilateral IPH, only the side ipsilateral to the ischemic event was considered. In order to determine the role of SIR in the context of the degree of stenosis, carotid arteries were stratified into <50% stenosis (mild) and ≥50% stenosis (moderate to severe) groups based on North American Symptomatic Carotid Endarterectomy Trial criteria. ¹²

Imaging characteristics

Neck plaque MR imaging was performed as previously described. ¹³ The carotid vessel wall imaging was performed on a 3T MRI scanner (GE 750, GE Healthcare, Milwaukee, Wisconsin, USA) utilizing a 16‐channel head/neck/spine coil and consisted of at least the three following sequences: (a) two-dimensional time of flight; (b) three-dimensional (3D) fast spoiled gradient echo (or 3D MP-RAGE) acquired in the coronal plane; and (c) gadolinium bolus carotid MRA acquired in the coronal plane. The 3D MP-RAGE sequence was used as reported previously. ¹⁴ T1-CUBE sequences were performed as previously described. ¹³ The imaging parameters were as follows: TR/TE = 13.2 ms/3.2 ms, flip angle = 15°, in plan spatial resolution = 0.63 mm × 0.63 mm, reconstructed resolution = 0.31 mm × 0.31 mm, slice thickness = 1 mm, number of excitations = 2, TI = 304 ms, TR with respect to the non-selective inversion = 568 ms, acquisition time = 3 min 50 s.

SIR measurements

Examples of SIR measurements are demonstrated in Figure 1. The SIR on both MP-RAGE and T1-CUBE images was calculated by dividing the maximum signal intensity of the IPH by the mean intensity of the adjacent sternocleidomastoid muscle. All measurements were made by a radiology resident and confirmed by a staff neuroradiologist at our institution with greater than 2 years’ experience, both of whom were blinded to the clinical scenario. All measurements were performed using Visage imaging software (Visage Imaging, San Diego, California, USA).

Figure 1.

Example measurement of signal intensity ratio (SIR) on (a) MP-RAGE and (b) T1-CUBE images. To establish each SIR, the maximum signal intensity of the area of the intraplaque hemorrhage (white circle) was divided by the mean signal intensity of an area of the adjacent sternocleidomastoid muscle (yellow circle).

MP-RAGE: magnetization prepared-rapid gradient echo.

Statistical analysis

Percentages were calculated for categorical variables. Means and standard deviations were calculated for continuous variables, including SIRs. Statistical significance between mean values was calculated with Student’s t-test. In order to determine if increasing IPH SIRs were associated with an increased risk of ipsilateral ischemic events, a logistic regression analysis was performed. This was performed by utilizing the IPH SIR as the independent variable and whether or not an acute ipsilateral ischemic event occurred was used as the binary dependent variable. Odds ratios (ORs) with 95% confidence intervals (CIs) were calculated in the logistic regression analysis. All calculations took place in Microsoft Excel and JMP statistical software (SAS Institute, Cary, North Carolina, USA).

Results

Patient characteristics and subgroups

An overview of the patient selection process is outlined in Figure 2. Of 643 patients that underwent neck MRA imaging with MP-RAGE sequences, 533 patients were excluded because no IPH was detected. Four patients with SIR <1.5 were excluded, as these did not meet criteria for MR evidence of IPH. There were a total of 106 patients possessing at least one carotid artery with IPH with SIR >1.5. Ninety-one patients (85.8%) had a history of an ischemic event, whereas 15 patients (14.2%) did not have any prior ischemic events. Of the 91 patients with a prior ischemic event, 70 had IPH within the carotid artery ipsilateral to the ischemic event (70/91 = 76.9%). There were 21 patients (23.1%) with an ischemic event who had IPH found within the contralateral carotid artery. Thus, there were a total of 85 patients with 85 arteries with IPH included: 70 with IPH ipsilateral to an ischemic event and 15 that belonged to asymptomatic patients. All asymptomatic patients had only a single carotid artery with IPH.

Figure 2.

Flow chart of patient selection and stratification for patients with MP-RAGE sequences.

IPH: intraplaque hemorrhage; MP-RAGE: magnetization prepared-rapid gradient echo; MRA: magnetic resonance angiography; SIR: signal intensity ratio.

The mean age of included patients was 71.0 ± 11.1 and 66 were male (77.6%). Of patients with ischemic neurological events with ipsilateral IPH, 62 had ischemic strokes, 16 had transient ischemic attacks, 6 had retinal artery occlusion and 8 had amaurosis fugax (multiple patients had more than one event, thus numbers do not add to the total number of patients). When comparing baseline demographics and comorbidities between patients with and without an ischemic event, patients without an ischemic event had a higher prevalence of coronary artery disease (60.0% versus 34.3%, respectively), although this was not statistically significant (P = 0.08). There were no other significant differences between patient groups (Table 1).

Table 1.

Comparison in comorbidities between patients with and without ischemic events.

Variable	Patients with a cerebral ischemic event and ipsilateral IPH (N = 70)	Patients without an ischemic event (N = 15)	P-value
Age (years), mean ± SD	70.6 ± 11.3	69.7 ± 11.7	0.79
Male, n (%)	55 (78.6)	11 (73.3)	0.73
Coronary artery disease, n (%)	24 (34.3)	9 (60.0)	0.08
Hypertension, n (%)	53 (75.7)	12 (80.0)	0.99
Hyperlipidemia, n (%)	64 (91.4)	13 (86.7)	0.62
Atrial fibrillation, n (%)	10 (14.3)	3 (20.0)	0.69
Diabetes mellitus, n (%)	22 (31.4)	2 (13.3)	0.21
Obstructive sleep apnea, n (%)	14 (20.0)	4 (26.7)	0.73
Ever smoker, n (%)	44 (62.9)	10 (66.7)	0.99

IPH: intraplaque hemorrhage; SD: standard deviation.

Comparison of mean SIRs

These data are summarized in Table 2. On MP-RAGE images, there was no significant difference between the mean SIR of IPH in arteries ipsilateral to an acute event (3.0 ± 1.1) and the mean SIR of IPH in arteries without an acute ipsilateral event (3.3 ± 1.1) (P = 0.43). Similarly, on T1-CUBE images there was no difference between the mean SIR of IPH ipsilateral to an acute event (2.9 ± 0.86) and the mean SIR of IPH without an acute ipsilateral ischemic event (3.0 ± 1.1) (P = 0.73).

Table 2.

Signal intensity ratio values in arteries with and without ipsilateral acute ischemic events.

Sequence	IPH SIR in arteries with an ipsilateral event (mean ± SD)	IPH SIR in patients without an ipsilateral event (mean ± SD)	P-value
MP-RAGE	3.0 ± 1.1	3.3 ± 1.1	0.43
T1-CUBE	2.9 ± 0.86	3.0 ± 1.1	0.73

IPH: intraplaque hemorrhage; MP-RAGE: magnetization prepared-rapid gradient echo; SD: standard deviation; SIR: signal intensity ratio.

Using logistic regression analysis, no statistically significant association was found between increasing IPH SIR and ipsilateral acute ischemic events on either MP-RAGE (OR: 0.82; 95% CI: 0.58–1.4; P = 0.43) or T1-CUBE sequences (OR: 0.85; 95% CI: 0.53–1.5; P = 0.56).

Stratification based on degree of stenosis

Of the 70 arteries with IPH that were ipsilateral to an ischemic neurological event, 26 (37.1%) had mild (<50%) narrowing; 44 (62.9%) had moderate to severe (≥50%) stenosis. Of the 15 arteries with IPH that were not ipsilateral to an ischemic event, 5 (33.3%) had mild narrowing, whereas 10 (66.7%) had moderate to severe narrowing. There were no differences in mean IPH SIR between arteries with and without an ipsilateral ischemic event when stratified into <50% stenosis and ≥50% stenosis groups with MP-RAGE or T1-CUBE sequences (Table 3).

Table 3.

Signal intensity ratio values of intraplaque hemorrhage in carotid arteries with ≥50% and <50% stenosis.

	≥50% stenosis			<50% stenosis
	SIR (mean ± SD)			SIR (mean ± SD)
Sequence	With ipsilateral ischemic event (N = 44)	Without (N = 10)	P-value	With ipsilateral ischemic event (N = 26)	Without (N = 5)	P-value
MP-RAGE	3.1 ± 1.3	3.2 ± 1.1	0.86	2.8 ± 0.86	3.4 ± 1.5	0.24
T1-CUBE	2.9 ± 0.86	3.2 ± 1.0	0.32	2.8 ± 0.88	1.9 ± 0.39	0.16

MP-RAGE: magnetization prepared-rapid gradient echo; SD: standard deviation; SIR: signal intensity ratio.

Discussion

The current study compared the mean IPH SIR between patients with acute ipsilateral ischemic events and those without. None of the differences between IPH SIRs were found to be statistically significant, regardless of the sequence utilized as well as the degree of arterial stenosis. Taken together, our data have two primary implications. First and foremost, higher SIRs measured on both MP-RAGE and T1-CUBE sequences are less likely to be useful in determining whether or not a hemorrhagic carotid plaque is the culprit lesion in instances of acute ischemic events without a known source. Second, increased signal intensity may not play an important role in the pathophysiology of IPH manifesting in neurologic ischemia.

The recent interest in this subject stemmed from a study by Yang et al., ¹⁰ in which the authors found that patients with acute cerebral infarctions had significantly higher SIRs of carotid IPH on T1-weighted images than patients without ischemic events. However, the observed association was not apparent when utilizing MP-RAGE or time-of-flight sequences, confirmatory of the current findings. Furthermore, although logistic regression analysis revealed that SIR was independently associated with ischemic events, the observed association was not apparent when adjusting for volumes of IPH and the presence of a necrotic core. Wang et al. ¹¹ similarly found that among patients with bilateral IPH, SIRs on MP-RAGE sequences were higher on the side ipsilateral to acute ischemic events compared with the contralateral side. Plaques on the symptomatic sides were longer and were also associated with larger necrotic core volumes.

Although higher SIRs were found to be associated with ischemic events in both of these prior reports, the role that other high-risk plaque features (e.g. larger plaque volumes, lipid-rich necrotic cores and ulcerations) play in the context of SIR remains uncertain. The discrepancies observed between these studies and our results may be secondary to these potential confounders, as we did not control for plaque volume or the presence of necrotic cores. In light of these uncertainties, however, it should therefore be emphasized that much work remains to be done before broad recommendations can be made regarding the use of SIR in detecting symptomatic plaques. Although a potential relationship may be elucidated in subsequent studies, the presence of IPH alone should still be considered an important marker of plaque vulnerability and risk factor for ischemic events regardless of SIR values. ⁵

The results of the current study may suggest that the pathomechanism of an IPH segueing into an acute cerebral ischemic event may be unrelated to the temporal nature of plaque hemorrhage. Carotid IPH is purported to be the result of fragile neocapillaries that rupture when exposed to the toxic, pro-inflammatory environment of the atherosclerotic plaque.^15,16 Plaque fissures, too, may contribute to hemorrhage. ¹⁷ Blood products within plaques result in erythrocyte breakdown and formation of methemoglobin, ultimately manifesting as a hyperintense signal on MP-RAGE sequences. ¹⁸ Such hemorrhagic events likely increase the rate of plaque progression with increased risk for fibrous cap rupture, formation of thrombus and distal embolization.^19,20 Although some authors have opined that intraplaque signal characteristics may be used to determine the age of intraplaque blood products, only moderate agreement has been shown between MR signal characteristics and histologic samples. ²¹ If MR is indeed imperfect at characterizing intraplaque blood product age, it is reasonable that the SIR is unrelated to the chronicity, and thus vulnerability, of plaques with IPH. However, our study may be limited in determining whether or not this is the case given our cross-sectional study design, and also that patients underwent neck MRA within a relatively broad period (6 weeks) following an ischemic event.

Despite the controversial nature of increasing IPH SIR as it relates to ipsilateral ischemic events, the presence of IPH alone should continue to be utilized as a marker for carotid plaques that may be symptomatic as it has been shown to be an independent predictor of future stroke. ⁵ The radiographic definition of IPH is based on the signal intensity of the region of interest within a carotid plaque as a ratio to the signal intensity of the adjacent neck musculature. In the current study, we utilized 150% of the adjacent sternocleidomastoid as a cutoff point in defining the presence of IPH, as has been utilized previously.^3,10,22 Other studies, however, have utilized a 200% cutoff in defining IPH. ²³ Although increasing the threshold of signal ratios may improve the specificity in identifying carotid IPH, more work is necessary in order to define the optimal cutoff point. It can therefore be suggested that, in the absence of additional data (ideally radiological-pathological correlational studies), utilizing a lower threshold of 150% as a definition of carotid IPH may be more appropriate so as to not miss any potential symptomatic hemorrhages within a carotid plaque.

The current study possesses limitations that must be considered. This was a single-center retrospective review of patient charts and imaging and our study is therefore at risk for selection bias. We focused solely on the presence of IPH and the corresponding SIR, and we did not address nor control for additional IPH features including volume, degree of carotid occlusion or presence of additional plaque characteristics including lipid-rich necrotic cores. Our sample sizes, particularly for patients with IPH but without ischemic events, were relatively small. We included patients who underwent neck MRA within a 6-week period following a cerebral ischemic event. Future studies should include a narrower window in which neck imaging is performed following an ischemic event. Prospective, longitudinal studies with larger sample sizes are needed to more appropriately characterize the relationship between increasing SIRs and the risk for future ischemic events.

Conclusions

The SIR of carotid plaque IPH as seen on MP-RAGE and T1-CUBE sequences was found to be unrelated to ipsilateral neurologic symptoms. These results call into question the use of SIR as a specific MR marker for plaque vulnerability and as an identifying feature of culprit lesions in the context of an acute ischemic event with an undetermined source. In the absence of more robust data, the presence of IPH alone should still be considered as a crucial marker of symptomatic plaque, regardless of SIR values.