OUP user menu

Determinants of magnetic resonance imaging detected carotid plaque components: the Rotterdam Study

Quirijn J.A. van den Bouwhuijsen, Meike W. Vernooij, Albert Hofman, Gabriel P. Krestin, Aad van der Lugt, Jacqueline C.M. Witteman
DOI: http://dx.doi.org/10.1093/eurheartj/ehr227 221-229 First published online: 6 August 2011


Aims Components of carotid atherosclerotic plaque such as intraplaque haemorrhage and lipid core are important determinants of plaque progression and destabilization. The association between plaque components and risk factors for cardiovascular disease is not well studied.

Methods and results Participants from the population-based Rotterdam Study with carotid wall thickening on ultrasound (n = 1006) underwent high-resolution magnetic resonance imaging for carotid plaque characterization. Maximum wall thickening, the degree of stenosis, and the presence of intraplaque haemorrhage, lipid core, and calcification were assessed in both carotid arteries and their associations with cardiovascular risk factors were investigated. Intraplaque haemorrhage and lipid core were present in almost 25% of plaques, respectively, and occurred simultaneously in 9% of plaques. In men, intraplaque haemorrhage and lipid core were more prevalent compared with women (28.8 vs. 18.3 and 28.9 vs. 21.7%, respectively). Intraplaque haemorrhage occurred more frequently at older age [odds ratio (OR) per 10 years 1.8, 95% confidence interval 1.6–2.1], in men (OR 2.2, 1.7–2.9), in persons with hypertension (multivariate adjusted OR 1.4, 1.1–1.8), and in current smokers (multivariate adjusted OR 1.6, 1.2–2.3). Men (OR 1.5, 1.2–1.9) and subjects with hypercholesterolaemia (multivariate adjusted OR 1.4, 1.1–1.7) more often exhibited a lipid core.

Conclusion In subjects from the general population with carotid wall thickening, intraplaque haemorrhage and lipid core—both considered indicators of unstable plaque—are highly frequent and more prevalent in men compared with women. Furthermore, different risk factors are associated with these plaque components: hypertension and current smoking were risk factors for the presence of intraplaque haemorrhage, and hypercholesterolaemia was the only risk factor for lipid core presence.

  • Carotid artery
  • Magnetic resonance imaging
  • Plaque components
  • Risk factors

See page 160 for the editorial comment on this article (doi:10.1093/eurheartj/ehr243)


A primary aim in the prevention of cerebrovascular disease is to identify subjects who are at high risk of ischaemic events.1 A major cause of cerebral ischaemia is atherosclerotic disease in the carotid artery.2 Destabilization of atherosclerotic plaque can lead to plaque rupture with thrombus formation on the disrupted plaque surface and subsequent embolization of thrombus and/or plaque material into the distal vessels.3 Components of the atherosclerotic plaque such as haemorrhage and lipid core are presumed to be important determinants of progression and destabilization of the plaque.4,5 Ultrasound and computed tomography (CT) have been used extensively to study the non-calcified plaque components of carotid atherosclerosis, but both modalities cannot distinguish between lipid or haemorrhagic components in plaques.6,7 Magnetic resonance imaging (MRI) is feasible to discriminate between these plaque components.811 Clinical studies in patients with symptomatic carotid stenosis have documented an association between the presence of intraplaque haemorrhage and/or lipid core and subsequent risk of cerebrovascular disease; these studies further indicate that both components may be associated with a different risk for future events.12,13

In the formation of atherosclerotic plaque, classical cardiovascular risk factors are presumed to play an important role.14 Except for the association of hypercholesterolaemia and the presence of a lipid core,15 it is unclear whether cardiovascular risk factors are related to the presence of specific plaque components. Especially factors determining haemorrhagic plaque components are unknown. Knowledge about such risk factors will increase our insight into the biological pathways, which could be suitable targets for the prevention of plaque growth and destabilization.

In a large population-based study, we used high-resolution MRI of the carotid arteries to investigate associations between cardiovascular risk factors and plaque composition.


Study population

This study was embedded within the Rotterdam Study, a prospective, population-based study among subjects aged ≥45 years, aimed at investigating determinants of various diseases among the elderly.16 The study started in 1990 and all participants are invited every 3–4 years to the research centre for follow-up examinations, including carotid ultrasonography. The ultrasonography protocol and reading has been described in detail previously.17 From October 2007 onwards, carotid MRI is included in the Rotterdam Study.

Until October 2009, in 1417 out of 5651 participants (25.1%) from the Rotterdam Study, ultrasonography revealed carotid wall thickening ≥2.5 mm in the left, right, or both carotid arteries. These subjects were selected for carotid MRI scanning. Participants were excluded if they had contraindications for MRI (n = 34), suffered from dementia (n = 9) or physical immobility (n = 34), lived in a nursing home (n = 47), had moved outside the area (n = 42), or had undergone a carotid endarterectomy procedure (n = 4). Of the 1247 eligible subjects for carotid MRI, 1131 agreed to participate (response 90.7%). Due to physical inabilities (e.g. back pain) or claustrophobia, imaging could not be performed or completed in 59 individuals (5.1%), so in total 1072 participants underwent a complete scan. The population consisted for more than 95% of persons of Caucasian descent. The medical Ethics Committee approved the study and all participants gave written informed consent.

Magnetic resonance imaging scan protocol

Magnetic resonance imaging of the carotid arteries was performed on a 1.5 T MR scanner (GE Healthcare, Milwaukee, WI, USA) with a bilateral phased-array surface coil (Machnet, Eelde, The Netherlands). Participants were stabilized in a custom-designed head holder to reduce motion artefacts. High-resolution images were obtained using a standardized protocol. First, both carotid bifurcations were identified by means of two-dimensional (2D) time-of-flight MR angiography. Thereafter, high-resolution MRI sequences were planned to image the carotid bifurcations on both sides: four sequences in the axial plane: a proton density weighted (PDw)-fast spin echo (FSE)-black blood (BB) sequence; a PDw-FSE-BB with an increased in-plane resolution; a PDw-echo planar imaging (EPI) sequence, and a T2w-EPI sequence; and two 3D sequences: a 3D-T1w-gradient echo (GRE) sequence; and a 3D phased-contrast MR angiography. A detailed technical description is provided in Supplementary material online. Total scanning time was ∼30 min.

Image review

Before evaluation of the plaque composition, the quality of all sequences in each MRI scan was rated on a five-point scale (1 = worst; 5 = best).9 Scans were included in the analyses if the image quality was scored ≥3 on all MRI sequences (n = 1006; 93.8%). Carotid plaque size was quantified by obtaining maximum carotid wall thickness and degree of luminal stenosis using the NASCET criteria18 on the PDw-FSE images. We assessed plaque characteristics in all plaques with a maximum thickness of ≥2.0 mm on MRI (n = 1866 carotid arteries). Plaques were reviewed for the presence of three different plaque components (presence of calcification, intraplaque haemorrhage, and lipid core) with a standardized evaluation protocol by one trained observer with 3 years of experience, blinded to all participant characteristics. Detailed information on the assessment, including results for intra-subject and inter-observer reliability (Cohen's κ values ranging from 0.85 to 0.95), is provided in Supplementary material online. Calcification was defined as the presence of a hypointense region in the plaque on all sequences.10,19 Intraplaque haemorrhage was defined as the presence of a hyperintense region in the atherosclerotic plaque on 3D-T1w-GRE (Figure 1).20,21 Lipid core presence was defined as a hypointense region, not classified as intraplaque haemorrhage or calcification, in the plaque on PDw-FSE or PDw-EPI and T2w-EPI images, or a region of relative signal intensity drop in the T2w-EPI images compared with the PDw-EPI images (Figure 2).9,10,19 Subjects were recorded as positive for the presence of any plaque component if the component was identified in one or both carotid arteries.

Figure 1

Plaque with intraplaque haemorrhage in the bifurcation of the left carotid artery. The intraplaque haemorrhage (indicated by arrow) shows high signal intensity on the T1w-gradient echo image and is high- to isointense on PDw and T2w images.

Figure 2

Example of plaque with a large lipid core in the right internal carotid artery. The lipid core (indicated by arrow) shows low signal intensity on the T2w-echo planar imaging, PDw-echo planar imaging, and PDw-fast spin echo images and is isointense on the T1w-gradient echo image.

Risk factor measurements

Information on medical history and smoking behaviour was collected using home interviews. Smoking status was classified as current, past, and never. Medication use was assessed through automated linkage to pharmacies with computerized records. Blood pressure, total cholesterol, HDL cholesterol, and non-HDL cholesterol were measured at study centre visits as described previously.22 Hypertension was identified as the use of antihypertensive medication and/or an average systolic blood pressure of 160 mmHg or above and/or an average diastolic blood pressure of 100 mmHg or above (Grade 2 according to European Society of Cardiology criteria).23 Diabetes mellitus was defined as the use of blood glucose-lowering medication and/or a non-fasting serum glucose level of ≥11.1 mmol/L and/or fasting serum glucose levels ≥7 mmol/L (≥126 mg/dL).24 Impaired glucose intolerance was defined as glucose levels between 5.6 mmol/L (>100 mg/dL) and 7 mmol/L (<126 mg/dL). Hypercholesterolaemia was defined as a serum total cholesterol ≥6.2 mmol/L (≥239 mg/dL).25 Cut-off values for low HDL cholesterol were ≤1.0 mmol/L (≤40 mg/dL)25 and for high non-HDL cholesterol ≥4.1 mmol/L (≥160 mg/dL).26 History of stroke27 and coronary heart disease [myocardial infarction, percutaneous transluminal coronary angioplasty, or coronary artery bypass graft]28 were assessed till date of inclusion as described previously.

Statistical analysis

Differences in plaque characteristics between sexes were studied using logistic regression adjusted for age and, for analyses of plaque components, also for wall thickness. Associations between cardiovascular risk factors and the presence of plaque components were studied with logistic regression, adjusting for age, sex, and maximum carotid wall thickness. We additionally adjusted each association for the other cardiovascular risk factors and for lipid-lowering medication and antithrombotic medication. We did not adjust for blood pressure-lowering medication and glucose-lowering medication because these are part of the definitions of hypertension and diabetes, respectively. Furthermore, we examined combinations of risk factors that were individually associated with carotid plaque components. To adjust for the correlation between both carotid arteries within each participant, a generalized estimation equation approach was followed, with an independent or unstructured working correlation matrix. The unit in the cross-sectional analyses was the presence of plaque component in the left or right carotid artery. Each participant yielded a maximum of 2 U in the data set. We repeated all analyses adjusting for history of stroke or coronary heart disease. We examined for interaction of gender and each of the cardiovascular risk factors but none of the interaction coefficients were statistically significant. Missing values for the cardiovascular risk factors (6.1%) were imputed using the expectation maximization method. All analyses were carried out using the Statistical Package for Social Sciences (SPSS) for Windows version 15.0 (Chicago, IL, USA). The P-value threshold used for significance was 0.05 and all tests were two-sided.


Baseline characteristics of the study population (n = 1006) are presented in Table 1. The mean age of participants was 70.3 ± 10.2 years and 52.0% were men. Table 2 presents MRI plaque characteristics overall and by gender for 1866 carotid arteries with a wall thickness of ≥2.0 mm. In 13.5% of the carotid arteries, a stenosis of more than 30% was present. Intraplaque haemorrhage was present in 23.8% of plaques (34.5% of participants), a lipid core in 25.5% (39.6% of participants), and calcification in 66.7% of the plaques (78.7% of participants). Intraplaque haemorrhage and lipid core were present simultaneously in 9.3% of the plaques (17.7% of participants), and in 40.0% of the plaques (56.4% of participants), only one of these components was present. The presence of intraplaque haemorrhage, lipid core, and calcification were all associated with larger maximum carotid wall thickness (Figure 3).

View this table:
Table 1

Baseline characteristics of the population (n = 1006)

View this table:
Table 2

Plaque characteristics in 1866 carotid plaques in 1006 participants, men and women combined and by gender

Figure 3

Carotid plaque components in relation to plaque thickness. For intraplaque haemorrhage, lipid core, and calcification, prevalences are shown per quartile of maximum wall thickness. All three components are significantly more frequently present in thicker plaques (age- and sex-adjusted association; P < 0.001 for all three components). The relation is strongest for intraplaque haemorrhage, with the frequency increasing from 5.6% in the lowest quartile to 44.5% in the highest quartile of maximum wall thickness.

The maximum wall thickness (3.4 ± 1.0 vs. 3.1 ± 0.8 mm; P < 0.001) was larger and stenosis >30% was more frequent (15.2 vs. 11.9%; P = 0.043) in men compared with women. Carotid plaque composition was also significantly different between sexes. In men, a higher prevalence of intraplaque haemorrhage (28.8 vs. 18.3%; P = <0.001) and lipid core (28.9 vs. 21.7%; P = 0.045) was found compared with women (Table 2). The presence of calcification did not differ significantly between men and women (66.6 vs. 67.8%; P = 0.996).

Table 3 presents the associations of cardiovascular risk factors with different plaque components. As there was no interaction between sex and cardiovascular risk factors, we did not stratify for sex. After adjustment for age and sex, hypertension and current smoking were both associated with the presence of intraplaque haemorrhage [odds ratio (OR) 1.5; 95% confidence interval 1.1–1.9 and 1.6, 1.2–2.3, respectively]. Additional adjustment for maximum wall thickness, the other risk factors, and lipid-lowering and antithrombotic medication in a multivariate adjusted model did not materially change these associations. Subjects in which both risk factors for intraplaque haemorrhage were present (hypertension and smoking) had an almost three times increased risk of intraplaque haemorrhage presence compared with non-hypertensive, non-smoking subjects (OR 2.9, 1.7–4.9).

View this table:
Table 3

Cardiovascular risk factors and presence of plaque components (in 1866 carotid plaques)

Lipid core was more frequent in subjects with hypercholesterolaemia (OR 1.4, 1.1–1.8). Lipid core presence was associated with high non-HDL cholesterol (OR 1.4, 1.1–1.8), but not with low HDL cholesterol (OR 0.9, 0.7–1.1). Lipid core presence was less prevalent in subjects with diabetes mellitus (OR 0.6, 0.5–0.9), but this association was no longer statistically significant in multivariate analysis. Impaired glucose intolerance was not associated with lipid core presence (0.8, 0.7–1.1).

Hypertension was also significantly associated with the presence of calcifications (OR 1.4, 1.1–1.7); after additional adjustments, the association was borderline significant.

Adjusting for prior symptomatic stroke or coronary heart disease (n = 183) or exclusion of subjects with missing values on any of the risk factors did not materially change the above-mentioned results.


Using high-resolution MRI in a large sample of the general population, we investigated associations of plaque components with classical cardiovascular risk factors. Intraplaque haemorrhage and lipid core were each present in ∼25% of plaques and more prevalent in plaques with a larger wall thickness. The prevalence of intraplaque haemorrhage was almost twice as high in men compared with women, independent of wall thickness and cardiovascular risk. Current smoking and hypertension were associated with the presence of intraplaque haemorrhage, and hypercholesterolaemia was associated with the presence of a lipid core.

A major strength of our study is the large sample of middle-aged and elderly subjects from the general population. Compared with small clinical studies of patients with ruptured and symptomatic carotid plaques, the use of an asymptomatic population reduces selection bias and provides estimates of the association between risk factor and plaque components that are not affected by medical intervention or other lifestyle modifications in symptomatic carotid disease. Furthermore, we used a high-resolution validated MRI sequence29 with improved inter-observer agreement, sensitivity, and specificity for intraplaque haemorrhage detection compared with conventional sequences.30

There are two other population-based studies that examined the correlates of carotid plaque characteristics using MRI. The Multi-Ethnic Study of Atherosclerosis (MESA)15 scanned 214 subjects randomly selected from the top 15th percentile of carotid maximum intima–media thickness of 6814 cardiovascular asymptomatic study participants. In the Atherosclerosis Risk in Communities Study (ARIC), 1769 subjects were scanned, of whom 60% had ultrasonography-defined carotid wall thickening.31,32 We found that the prevalence of intraplaque haemorrhage and/or lipid core was higher in men compared with women, which is in agreement with results from the ARIC study.32 The MESA study did not find an association between gender and lipid core presence.15 As the overall percentages of plaque components in MESA and ARIC were measured in the artery with the largest plaque, we translated our prevalence figures per plaque to prevalences per subject. The prevalence of a lipid core per subject was lower in our study compared with MESA (39.6 vs. 70.6%), probably because we used less stringent criteria for MRI scanning and thereby included persons with smaller, less mature plaques. Furthermore, unlike the MESA study, we did not administer contrast material, while a lipid core is more easily detected with a contrast-enhanced MRI sequence.33 In ARIC, the prevalence of a lipid core appeared comparable to our study (42.4% in the ARIC study participants with wall thickness ≥1.5 mm vs. 39.6% in our study sample). A comparison of intraplaque haemorrhage prevalences was not possible because in the ARIC study, these were not given for subjects with increased wall thickness separately.32

Hypertension and current smoking were major risk factors of intraplaque haemorrhage in our study. Intraplaque haemorrhage is presumed to be caused by erythrocyte leakage from dysfunctional intraplaque microvessels.34 These erythrocytes attract macrophages into the plaque. The consequential highly reactive milieu causes formation of new immature leaking microvessels resulting in an increase in plaque volume. Endothelial nicotine receptors play an important stimulating role in angiogenesis, which may explain the higher prevalence of intraplaque haemorrhage that we found in smokers,35 through elevated levels of vascular endothelial cell growth factor,36 a potent regulator of angiogenesis37 and also considered the key regulator of vascular permeability.38

Our finding that hypercholesterolaemia is associated with the presence of a lipid core in the carotid plaque is consistent with results from the MESA study.15 Plasma cholesterol, in particular LDL cholesterol, was found to be the most important determinant of the lipid core in MESA. Our study results were comparable, in that we found plasma cholesterol and non-HDL cholesterol, but not HDL cholesterol, to be associated with lipid core presence. In ARIC, a borderline significant association was found between total cholesterol level and presence of a lipid core.31 Replicating this analysis with cholesterol as a continuous variable in our study yielded a similar effect estimate as in MESA (data not shown in the Results section; OR of 1.13; P = 0.045).

Associations between cardiovascular risk factors and arterial calcification have been studied before, predominantly in CT studies. Our result that age and hypertension (after adjustment of age and sex) are related to the presence of carotid calcification is in line with population-based CT studies.39,40

Diabetes mellitus was not associated with a higher risk of any of the plaque components we studied. Esposito et al.41 studied the association between diabetes mellitus and vulnerable plaque components in 191 patients with a carotid artery stenosis and a history of stroke. In contrast to our results, they found diabetes mellitus to be related to vulnerable plaque composition, after combining components such as intraplaque haemorrhage, lipid core, and calcification as one vulnerable plaque characteristic. An explanation for the difference in findings may be that Esposito et al. studied patients with more advanced atherosclerotic disease. We found an inverse association between diabetes mellitus and lipid core presence. General practitioners in the Netherlands are encouraged by protocol to subscribe lipid-lowering medication to diabetics. As a result, in our study, subjects with diabetes mellitus had significantly lower levels of total cholesterol compared with non-diabetics (5.7 vs. 5.2 mmol/L; P < 0.001). After adjustment for hypercholesterolaemia and the use of lipid-lowering drugs in multivariate analysis, the association was no longer statistically significant. The remaining inverse direction may be the result of residual confounding.

So far, no studies have been published on the predictive value of intraplaque haemorrhage and lipid core presence for clinical events in the general population. In a recent study by Underhill et al.4 in 108 asymptomatic patients with 50–79% carotid stenosis, the size of the lipid core was reported as a stronger predictor for carotid plaque rupture than the presence of intraplaque haemorrhage. In a study of Takaya et al.,13 among 154 initially asymptomatic patients with 50–79% carotid stenosis, intraplaque haemorrhage and the size of the lipid core were both related to risk of a cerebrovascular event, but not the presence of a lipid core. In a study using carotid specimens from patients undergoing carotid endarterectomy by Hellings et al.,12 an increased risk for cardiovascular events was found for the presence of plaque haemorrhage but not for the presence of lipid core.12 From a clinical point of view, the separate assessment of various plaque components could prove to be important. For example, screening of persons for vulnerable plaque components using MRI could be proposed for better risk stratification of persons at risk for cardiovascular disease. Selection of persons for such an MRI screening could be based on risk factors for intraplaque haemorrhage and lipid core, although the modest strength of individual risk factors as found in our study does not lend strong support for such a preselection. Choosing persons based on combinations of risk factors yields a moderately higher gain as subjects who had hypertension and who smoked had an almost three times increased risk compared with subjects without these risk factors. However, proper evaluation of a stepwise selection for MRI screening should probably be based on risk factor assessment before the more elaborate sonographic assessment of plaque thickness. Furthermore, knowledge about risk factors for specific plaque components may provide a basis for further research into suitable targets for the prevention of plaque growth or plaque destabilization.

Limitations of our study need to be discussed. MRI identification of fibrous cap rupture is an important factor in assessing plaque stability.42 This requires contrast-enhanced MRI, as reproducibility in identifying the fibrous cap of carotid artery plaques by non-contrast-enhanced MRI has been shown to be poor.43 Based on our study population (the general population), we chose to not administer contrast material to the participants, thus prohibiting us from studying the fibrotic cap in carotid plaques. As noted above, lipid core is also more easily detected with a contrast-enhanced MRI sequence.33 However, our non-contrast-enhanced MRI sequences were shown to have good accuracy and reproducibility in validation studies.9,10,19 Another potential limitation is that in our study, plaques with wall thickness <2.0 mm were excluded because MRI differentiation between plaque components was not feasible in these small plaques and the presence of components could be missed.

In conclusion, in subjects from the general population with carotid plaques, intraplaque haemorrhage and lipid core—both considered indicators of unstable plaque—are highly frequent and more prevalent in men than in women. Hypertension and current smoking were risk factors for the presence of intraplaque haemorrhage, while hypercholesterolaemia was the only risk factor associated with lipid core presence.


This work was supported by the Netherlands Heart Foundation (grant no. 2006B206) and the Netherlands Organization for Scientific Research (NWO) (Vici, grant no. 918-76-619).

Conflict of interest: none declared.


The authors thank the participants of the Rotterdam Study and Jolande Verkroost, Maarten Leening, Marno van Lieshout, Germaine Verwoert, and the MRI technicians at the research centre for their contributions to the study.


View Abstract