OUP user menu

Non-invasive imaging of atherosclerosis

Francis R. Joshi, Alistair C. Lindsay, Daniel R. Obaid, Erling Falk, James H.F. Rudd
DOI: http://dx.doi.org/10.1093/ehjci/jer319 205-218 First published online: 25 January 2012


Atherosclerosis is an inflammatory disease that causes most myocardial infarctions, strokes, and acute coronary syndromes. Despite the identification of multiple risk factors and widespread use of drug therapies, it still remains a global health concern with associated costs. It is well known that the risks of atherosclerotic plaque rupture are not well correlated with stenosis severity. Lumenography has a central place for defining the site and severity of vascular stenosis as a prelude to intervention for relief of symptoms due to blood flow limitation. Atherosclerosis develops within the arterial wall; this is not imaged by lumenography and hence it provides no information regarding underlying processes that may lead to plaque rupture. For this, we must rely on other imaging modalities such as ultrasound, computed tomography, magnetic resonance imaging, and nuclear imaging methods. These are capable of reporting on the underlying pathology, in particular the presence of inflammation, calcification, neovascularization, and intraplaque haemorrhage. Additionally, non-invasive imaging can now be used to track the effect of anti-atherosclerosis therapy. Each modality alone has positives and negatives and this review will highlight these, as well as speculating on future developments in this area.

  • atherosclerosis
  • imaging
  • non-invasive


It is well known that the risks of atherosclerotic plaque rupture are not well correlated with stenosis severity. Lumenography has a central place for defining the site and severity of vascular stenosis as a prelude to intervention for relief of symptoms due to blood flow limitation. Atherosclerosis develops within the arterial wall; this is not imaged by lumenography and hence it provides no information regarding underlying processes that may lead to plaque rupture. For this, we must rely on other imaging modalities such as ultrasound, computed tomography (CT), magnetic resonance imaging, and nuclear imaging methods. These are capable of reporting on the underlying pathology, in particular the presence of inflammation, calcification, neovascularization, and intraplaque haemorrhage. Each modality alone has positives and negatives and this review will highlight these, as well as speculating on future developments in this area.

Imaging of the processes involved in atherosclerosis serves several purposes. First, many non-invasive imaging techniques have been shown to predict clinical events, both in symptomatic and asymptomatic patients. Secondly, imaging can further our understanding by allowing novel insight into the underlying biology of the disease. Thirdly, it can be used to track the effects of drug therapies. Finally, as the direct and indirect costs related to cardiovascular disease and stroke—estimated at €169 billion in the European Union in 20031 and in the USA in 2007 alone at $286 billion2—continue to rise, imaging techniques are sure to play a vital part in preventing, diagnosing and treating the consequences of cardiovascular disease in the most cost-effective manner possible.

In the interests of space, we have restricted this account to non-invasive imaging modalities that are either already in the clinic or close to it. Although we have included descriptions of pre-clinical imaging studies, these are mentioned largely as validation of subsequent clinical imaging approaches. For those interested in techniques further from immediate use, we suggest the comprehensive reviews from Leuschner et al.,3 Sadeghi et al.,4 or Vancraeynest et al.5 as excellent starting points.

A pathologist's view of atherosclerotic plaque development

Atherosclerosis is a systemic, lipid-driven inflammatory disease of the arterial wall leading to multifocal plaque development.6 The speed of progression varies greatly, but it usually takes decades to develop the advanced atherosclerotic lesions responsible for clinical disease. Most plaques remain asymptomatic (sub-clinical disease), some become obstructive and might cause symptoms because of impaired maximal blood flow (stable angina, intermittent claudication, or mesenteric ischaemia are examples), and a small percentage, in some individuals, become thrombosis-prone (vulnerable) and lead to atherothrombotic events such as heart attack, stroke, and sudden death.7,8 The variable speed of both plaque development and its ultimate pathological course is not well understood; although genetic factors and the extent of exposure to cardiovascular risk factors undoubtedly play a part.9

Vulnerable plaques

The great majority of symptomatic coronary thrombi (∼75%) are caused by ‘plaque rupture’.7 The remaining thrombi are caused by less well-defined mechanisms of which so-called plaque erosion is the most common type.10 By inference, there are two major types of vulnerable plaques, rupture-prone, and erosion-prone, that are presumed to resemble the corresponding thrombosed plaques, just without rupture and thrombosis. The archetypal presumed rupture-prone plaque contains a large and soft lipid-rich necrotic core covered by a thin and inflamed fibrous cap,11 and they are often known as TCFAs (thin-capped fibroatheromas). Associated features include large plaque volume, expansive outward remodelling mitigating luminal obstruction, neovascularization, plaque haemorrhage, inflammation, and a ‘spotty’ pattern of calcifications (Figure 1). Although the macrophage density in ruptured caps is high, whole-plaque macrophage density rarely exceeds a few percentage points because ruptured caps are tiny.12 Vulnerable plaques of the erosion-prone type are heterogeneous and defined only by their fate (thrombosis, mostly mural).10 The surface endothelium is missing, but whether it vanished before or after thrombosis happened remains unknown. No distinct morphological features have been identified but, in general, eroded plaques with thrombosis are not often calcified, rarely associated with expansive remodelling, and only sparsely inflamed.13 This lack of a typical set of features makes them harder to detect than TCFAs. This subtype is commoner among females.

Figure 1

Coronary plaque rupture and rupture-prone vulnerable plaques. For comparison, a ruptured plaque with thrombus (top) and an intact and stable plaque (bottom) are depicted, and vulnerable plaque features are listed to the right. By inference, vulnerable plaques of the rupture-prone type are presumed to look like plaque rupture except for an intact cap without thrombus. *Values obtained from Kolodgie et al.21

Vulnerable plaques, plaque rupture, and thrombosed plaques tend to cluster in ‘hot spots’ within the proximal segments of the major coronary arteries,14,15 and rarely more than one or a few such lesions exist simultaneously.16 Vulnerable plaques are probably outnumbered by plaques with more stable phenotypes by a factor of 10-to-1.7

Determinants of vulnerability

Plaque rupture requires the presence of a lipid-rich necrotic core covered by a thin fibrous cap. The size of the necrotic core and the thickness of the fibrous cap appear to be the two major structural determinants of vulnerability.

Necrotic core

During atherogenesis, atherogenic lipoproteins are retained within the intima, are modified, and accumulate predominantly deeply in the abluminal part of the intima.6 Some of these ‘pools’ of lipids seem to attract macrophages that secrete proteolytic enzymes and engulf lipid until they die, leaving behind a soft and destabilizing lipid-rich cavity containing cholesterol crystals and devoid of supporting collagen and cells, the ‘necrotic core’. Such a plaque is called an ‘atheroma’ or ‘fibroatheroma’.13,17 Later on, plaque neovascularization (angiogenesis) supervenes.18 The new microvessels rarely originate from the lumen but usually from vasa vasorum in the adventitia. They lack supporting cells and are fragile and leaky, giving rise to local extravasation of plasma proteins and erythrocytes. Such intraplaque bleeding is common and may expand the necrotic core, causing rapid progression of the lesion.19 Another not uncommon source of plaque haemorrhage is extravasation of blood through a ruptured fibrous cap.

Fibrous cap

The fibrocellular part of the plaque located between the necrotic core and the lumen is called the ‘fibrous cap’. It is extremely thin in coronary plaque rupture. Assessed by microscopic examination ‘post-mortem’, ruptured caps were usually <65 microns thick.10 Assessed by optical coherence tomography of culprit lesions in acute myocardial infarction, the mean thickness was only 49 microns.20 If the fibrous cap is thin, the plaque is called a TCFA. In TCFA, the necrotic core occupies ∼23% of plaque area.21 Thin fibrous caps are usually heavily inflamed (macrophage density ∼14%), particularly those that have ruptured (macrophage density ∼26%), but because they are thin, their ability to accommodate macrophages are limited.21 Apoptosis is common at the site of fibrous cap rupture, usually confined to macrophages because the vascular smooth muscle cells (SMCs) have largely vanished when rupture occurs. With their ability to synthesize extracellular matrix, including collagen, SMC apoptosis is associated with impaired healing and repair, increasing the risk of plaque rupture.


Atherosclerosis is an innate inflammatory disease in which smouldering inflammatory activity is not confined to just a few atherosclerotic lesions but is present, more or less, in all such lesions throughout the body. In contrast, vulnerable plaques are relatively rare, and inflammation may play a causal role in plaque rupture only if located within a thin fibrous cap, i.e. the microstructure of the plaque needs to be permissive for rupture. Thus, although plaque inflammation may be useful as a marker of disease activity, it is probably not useful as the only marker of plaque vulnerability. For this reason, many imaging approaches are multi-modality, designed to interrogate several aspects of plaque biology.

Expansive remodelling

During atherogenesis, the artery tends to remodel in such a way that the luminal obstruction caused by some plaques is attenuated (expansive remodelling) and by others, accentuated (constrictive remodelling). Although vulnerable plaques of the rupture-prone type (TCFA) are usually high volume overall, they often appear non-obstructive (if detected at all) at angiography because of expansive remodelling.22 In contrast, plaques responsible for stable angina usually are smaller but nevertheless are often associated with more severe luminal narrowing because of concomitant constrictive remodelling. The reasons for the different modes of remodelling remain to be defined, but recent clinical observations indicate that diabetes is accompanied by inadequate compensatory remodelling.23

Atherothrombosis in carotid arteries

Similar to coronary atherosclerosis, plaque rupture is by far the most common cause of symptomatic thrombosis in carotid atherosclerotic disease,8,24 but thrombus formation is more often non-occlusive though prone to embolize, and often prolonged because of slow-healing plaque ulceration.25 Intraplaque haemorrhage is common. Carotid plaques, including those that rupture, are rarely heavily inflamed.26,27 The critical thickness for a rupture-prone cap is greater for carotid plaques than for the much smaller coronary plaques, but a relatively thin cap is most critical for rupture in both arteries.28 In one study, the mean fibrous cap thickness in carotid plaque rupture was found to be nearly three times greater than in coronary plaque rupture (72 ± 15 vs. 23 ± 17 microns; mean ± SD), and the macrophage density within ruptured caps was lower (13.5 ± 10.9 vs. 26 ± 20%).25

Ultrasound imaging of atherosclerotic plaque

Non-invasive ultrasound imaging represents a safe, fast, and comparatively cheap method of assessing atherosclerosis, however its use is largely confined to the carotid and peripheral vasculature. Even before the development of significant atherosclerosis, the high temporal and spatial resolution of carotid ultrasound allows accurate measurement of the distance from the luminal surface to the intima-media boundary, termed carotid intima-media thickness (CIMT). Prospective data have shown CIMT correlates with cardiovascular risk, even in asymptomatic individuals. The Atherosclerosis Risk in the Community (ARIC) study demonstrated that for each 0.19 mm increase in CIMT, the risk of death or myocardial infarction increased by 36% in a middle-aged cohort.29 Furthermore, CIMT has been used to demonstrate the arterial response to therapy; the REGRESS study found that a 0.05 mm reduction in CIMT in patients treated with pravastatin was associated with a 10% decrease of the absolute risk of cardiovascular events over 2 years.30 The measurement of CIMT, however, has remained principally a research application of carotid ultrasound.

Ultrasound remains limited in its ability to distinguish plaque consituents and morphology. More recent developments promise to overcome this limitation. Delayed imaging using microbubble contrast agents (contrast-enhanced ultrasound) can demonstrate, for example, the degree of inflammation31 and neovascularization32 in the arterial wall (Figure 2). Symptomatic plaque can be distinguished from asymptomatic plaque;33 additionally proliferation of the adventitial vasa vasorum—a precursor of atherosclerosis—can be demonstrated in the walls of otherwise normal carotid arteries. Furthermore, such agents can also be conjugated with ligands that target specific pathological processes (e.g. the expression of VCAM-1 on surface endothelium34), thus offering the possibility of molecular imaging using currently available ultrasound technology. Lastly, three-dimensional (3D) ultrasound might potentially be used to monitor the effects of novel drug therapies using smaller sample sizes, and over shorter periods of time than traditional 2D studies.35 Testing of reproducibility and whether changes can be tracked with therapy are awaited.

Figure 2

Contrast ultrasound imaging of neovessels (from Giannarelli et al.32 with permission.). Representative ultrasound images of a Contrast Enhancement-positive atherosclerotic lesion before (A) and after (B) contrast-administration. (C) Contrast enhancement quantification by digital subtraction imaging. Green area depicts contrast enhancement of the representative contrast-enhanced (CE)-positive lesion. (D) Positive strong correlation between the total number of neovessels and the signal enhancement in fibrofatty lesions is shown. At 25× magnification (E) and 200× magnification (F), a histological section of the imaged atherosclerotic plaque stained for lectin (neovessels) is shown. Asterisk indicates the corresponding area in the ultrasound images and in the histological section from the same atherosclerotic lesion. Arrows in (E) and (F) indicate the corresponding luminal surface of the plaques in the ultrasound images and in the histological section from the same animal.

Magnetic resonance imaging

Whereas magnetic resonance angiography (MRA) has been used for several years to produce arterial lumenograms, more recent imaging sequences have permitted direct imaging of the components of atherosclerotic plaque itself. MRI is capable of detecting most of the features of vulnerable atherosclerotic plaque, including the fibrous cap, intraplaque haemorrhage, calcification, lipid core and in situ thrombus36 (Figure 3). The use of gadolinium contrast can improve the detection and quantification of the lipid core, and in particular the presence of a thin fibrous cap.37 More recently, the development of DCE (Dynamic Contrast Enhanced) MRI, where the kinetics of gadolinium transit through the plaque is quantified over time, has allowed assessment of the degree of plaque neovascularization and inflammation to be made.38

Figure 3

Examples of plaque changes and downstream injury (T1 images top, diffusion-weighted imaging images bottom). (A) Upper panel shows an intraplaque haemorrhage associated with minimal diffusion-weighted imaging injury (lower left) in the left anterior lobe. (B) Centre, a large thrombus is seen in the lumen of the right internal carotid artery, which was associated with only minimal damage in the right anterior lobe. (C) Right, clear surface disruption is seen, which in this case was associated with a large infarct in the left cerebral hemispheres.

The ability of MRI to delineate such plaque features has mainly been investigated in large vessels, including the aorta, carotid, and femoral arteries. In particular, carotid artery disease has been extensively studied; Hatsukami et al.39 were the first to show that MRI could be used to examine the integrity of the fibrous cap, and the same group subsequently demonstrated the link between fibrous cap rupture, as seen on MRI, and a recent history of transient ischaemic attack (TIA) or stroke.40 More recent work has highlighted the potential role for carotid MRI in the assessment of acute TIA and ischaemic stroke,41 while current studies are investigating the ability of MRI to detect high-risk plaque that may require further invasive management, such as carotid endarterectomy, despite not reaching current luminal criteria for surgery (i.e. <70%); it is in this subgroup of patients that carotid MRI is most likely to have a significant clinical impact.42

By comparison, MRI of the coronary arteries has failed to advance as rapidly, largely due to the inherent difficulties in imaging a smaller, moving, vessel. While MRA of the coronary arteries can be used to identify three-vessel or left main coronary disease,43 and whole heart magnetic resonance coronary angiography now allows imaging to occur over a much shorter time period,44,45 the ability of MRI to delineate elements of individual plaques in the coronary circulation remains relatively limited. So-called black-blood imaging of the coronaries allows the detection of increased thickness of the coronary arterial wall in patients with angiographically documented coronary artery disease,46 and positive coronary arterial remodelling is described in patients with subclinical atherosclerosis.47 More recently T1-weighted imaging has been shown to identify high-intensity signal in atherosclerotic plaque, associated with a recent plaque rupture event.48

Due to its safety and reproducibility, vascular MRI has been extensively applied in many research studies over the past decade. The ability of MRI to document changes in the vessel wall in response to anti-atherosclerotic therapy has already been described with statins49 and nicotinic acid.50 In a recent study,51 serial carotid MR was able to demonstrate that lipid was removed from plaques early during treatment with atorvastatin, and that this change in plaque composition occurred prior to plaque regression (Figure 4). Consequently, MRI is in use in Phase III drug trials assessing the arterial response to new anti-atherosclerotic therapies. Furthermore, the development of targeted molecular imaging probes has expanded the potential role of MRI. In particular the use of iron oxide particles, which lead to signal dropout, has shown that MRI can assess changes in the inflammatory state of the arterial wall,52 and the number of macrophages in atherosclerotic plaque.53,54 The ongoing development of novel molecular imaging probes, coupled with increases in imaging field strength and continuing sequence development, suggest that vascular MRI is likely to play an important clinical role in the future.

Figure 4

Changes in carotid plaque lipid content and tissue composition with atorvastatin. (from Zhao et al.51 with permission.). (A) An example of significant lipid content reduction (yellow arrows) and plaque regression at 3 years compared with baseline in the left carotid artery. Overall, 11% of study subjects had completed plaque lipid depletion over 3 years. (B) Magnetic resonance imaging example of the plaque lipid depletion time course. Regression in lipid-rich necrotic core size was notable between the baseline, 1-year, and 2-year magnetic resonance imaging scans. Regression in plaque volume seemed to follow plaque lipid depletion and was most pronounced from years 1 to 3. CE, contrast enhanced; T1W, T1-weighted. Right side: Carotid plaque tissue composition change over 3-year lipid therapy. Changes in plaque fibrous tissue (FT), calcium (CA), and loose matrix (LM), both volume (V) (green squares) and composition (pink circles), over 3 years. CA, calcium; FT, fibrous tissue; LM, loose matrix. Compared with baseline, volume of fibrous tissue decreased, but percent of fibrous tissue increased. These changes were significant at each of the first 2 years. Calcium and loose matrix did not change significantly over 3 years. Bars around the estimates are standard error bars.

Molecular imaging with single-photon emission computed tomography and positron emission tomography


Of all the imaging modalities discussed, single-photon emission computed tomography (SPECT) and positron emission tomography (PET) have the highest sensitivity. This means that they are able to detect pico-molar levels of contrast at the target site. PET imaging has greater spatial resolution (4–5 mm vs. 1–1.6 cm) than SPECT,55 though given the size of atherosclerotic plaques, it is still subject to partial volume error.56 Co-registration with CT or MRI is needed to localize PET tracer uptake to underlying anatomy. Both SPECT and PET involve ionizing radiation, limiting their use for widespread screening of asymptomatic individuals.

Imaging inflammation with 18F-fluorodeoxyglucose

18F-fluorodeoxyglucose (FDG) is the most commonly used tracer in PET imaging of atherosclerosis. It exploits the fact that macrophages, key inflammatory cells in plaque, have higher glucose metabolism than both surrounding plaque cells and healthy artery wall.5759 FDG, a glucose analogue, competes with endogenous glucose for facilitated transport sites and after phosphorylation becomes trapped within cells. This accumulation can then be imaged and quantified in the PET scanner. FDG-PET is the gold-standard imaging modality for the detection of tumour metastases in oncology.

Arterial FDG uptake was first noted in patients undergoing PET for cancer staging.60 In a rabbit model of atherosclerosis, FDG uptake in atherosclerotic regions correlated strongly with macrophage content (r = 0.93, P < 0.002),61 findings confirmed by others.62 In the first prospective study in man, levels of inflammation measured as accumulation rate of FDG were 27% greater in the culprit carotid after recent stroke or TIA than in the contralateral vesseI,63 though with contemporary therapy this difference is not apparent beyond 38 days of the event.64 Further work has confirmed FDG uptake correlates with areas of macrophage (CD68+) staining (r = 0.96, P < 0.001), and is independent of plaque thickness, area or luminal stenosis.65 FDG uptake is thought to reflect activation status of macrophages,66 though recent work by Folco et al.67 suggests that hypoxia, known to exist within atherosclerotic plaque,68 may also be an important determinant of glucose (and hence FDG) uptake into these cells. Nevertheless, FDG-PET remains the best-validated tracer in molecular atherosclerosis imaging (Figure 5).

Figure 5

Carotid arterial calcification and 18F-fluorodeoxyglucose uptake on positron emission tomography/computed tomography imaging (from Fayad et al.79 with permission.). Top row: Baseline non-contrast enhanced computed tomography images of a subject with extensive atherosclerosis in the dal-PLAQUE study. Axial (left) and coronal (right) images demonstrating bilateral carotid artery calcification, more marked in the right carotid artery. Bottom row: Baseline fused 18F-fluorodeoxyglucose-positron emission tomography and computed tomography from the same patient shown in the top row. Axial (left) and coronal (right) demonstrating bilateral carotid artery calcification more marked in the right carotid artery. In addition, the red/orange regions represent 18F-fluorodeoxyglucose uptake (yellow arrows). This appears more marked in the right than the left carotid (coronal view).

Arterial FDG signal is linked to male sex, age, the metabolic syndrome,69,70 levels of matrix metalloproteinases71 and genetic markers of plaque instability.72 Plaques with high-risk features on MRI (large lipid core) have higher FDG uptake than more stable phenotypes.73 FDG uptake can distinguish between culprit carotid and vertebral lesions in posterior circulation stroke74 and in the carotid artery correlates with microembolic signals on transcranial Doppler ultrasound after TIA.75

FDG-PET is a reproducible measure, valid across several vascular beds with excellent short-term and interobserver reproducibility.76 Rarely co-localizing with vascular calcification,71 uptake may be widespread,70,71 though especially affecting the carotid arteries and proximal aorta. This is consistent with the hypothesis that atherosclerosis is a generalized (rather than focal) inflammatory condition.

Increasingly, arterial FDG PET/CT imaging is being applied for the early assessment of anti-atherosclerosis therapy. Treatment with statins demonstrated signal reduction after only 3 months' therapy,77 while Mizoguchi et al.78 used FDG-PET/CT to show that pioglitazone reduced vascular inflammation when compared with glimepiride (Figure 6). The recent dal-PLAQUE trial with dalcetrapib79 has further demonstrated the value of FDG-PET for this purpose, providing evidence of vascular safety (i.e. no increase in inflammation) vs. placebo at 6 months with this novel agent. Consistent with the mechanism of action of the drug, the degree of increase in HDL correlated with the reduction in carotid artery inflammation noted on PET. The authors of this study also used black-blood MR imaging to link early drug-induced inflammation reduction with later beneficial structural changes within the artery wall.

Figure 6

Treatment Effects on 18F-fluorodeoxyglucose Uptake in Atherosclerotic Plaques (from Mizoguchi et al.78 with permission.). Treatment effects of pioglitazone and glimepiride on 18F-fluorodeoxyglucose uptake in atherosclerotic plaques. Representative 18F-fluorodeoxyglucose-positron emission tomography/computed tomography with contrast media images (left) at baseline and (right) after 4-month treatment with (bottom) pioglitazone or (top) glimepiride. Note reduction in FDG uptake in the atherosclerotic plaque with pioglitazone treatment (arrows). Changes in target-to-background ratio (TBR) after 4-month treatment with pioglitazone or glimepiride. Top: The TBR was evaluated in individual patients at baseline and after 4-month treatment for quantitative analysis. Bottom: Change (Δ) in TBR from baseline. Bar = 1× SEM. N.S., not significant.

Imaging coronary artery atherosclerosis presents special challenges. These arteries are small, less than the resolution of PET. Imaging is hindered by respiratory and cardiac motion during the time taken to acquire a PET dataset, typically around 20 min. Finally, FDG is taken up avidly by myocardium, which preferentially metabolizes glucose over free fatty acids.

Attempts to switch myocardial metabolism to free fatty acids by using high fat diets before imaging have had varying success.80,81 Rogers et al.82 reported higher FDG uptake in culprit segments in acute coronary syndrome compared with stable angina. Work with dual gating PET data for both respiratory and cardiac motion to improve definition of individual plaques is ongoing,83,84 and recent pre-clinical data suggest that administration of calcium antagonist drugs might improve plaque visibility.85 Further developments are awaited.

There are no prospective data to support FDG-PET atheroma imaging for predicting future cardiovascular events. In retrospective analyses of patients undergoing PET for oncology staging, high levels of baseline vascular FDG uptake were associated with subsequent cardiovascular events.86 The results of prospective event-driven studies are awaited, including the BioImage Study that aims to identify imaging markers (CT, MRI, and FGD-PET/CT) of future cardiovascular risk.87

FDG-PET is limited by its lack of cellular specificity (FDG is accumulated by all cells that metabolize glucose), and the highlighted problems with imaging the coronary circulation. These issues have prompted the search for more macrophage-specific tracers. In a recent prospective clinical study in carotid atherosclerosis, imaging with11 C-PK11195 was found to correlate with the extent of immunohistochemical macrophage infiltration.88 Other tracers showing promise include radio-labelled choline89 and the somatostatin receptor analogue68 Ga-DOTATATE.90

Imaging calcification

18F-sodium fluoride (NaF) is a positron-emitting bone-seeking agent used clinically to identify primary osteoblastic tumours and bone metastases.91 Uptake of NaF within calcified structures such as bone reflects both blood flow and osteoblastic activity.92 It therefore detects areas of new calcification. Retrospective studies, all performed in oncology patients, have shown that NaF can inform the processes of vascular calcification.9396 There are positive correlations between the extent of NaF uptake and the presence of cardiovascular risk factors. A significant correlation exists between a history of cardiovascular events and presence of fluoride uptake in coronary arteries.96 Subjects previously imaged with both FDG and NaF demonstrated that vascular lesions seldom have coincident uptake of both tracers, implying each reflects distinct pathophysiologic processes within atherosclerosis.94 Eighty per cent of calcified arterial sites appear ‘inert’, but NaF uptake in the remainder suggests the tracer may have a role in imaging dynamic micro-calcification in advance of current imaging techniques.97

Other targets

Numerous other metabolic and signalling pathways associated with plaque vulnerability have been evaluated with SPECT in preclinical models of atherosclerosis. 99mtechnetium-labelled annexin-A5 has been used to image macrophage apoptosis in pre-clinical models98 and in a pilot clinical study99 uptake was seen in carotid atherosclerosis after recent TIA. We await further studies in human atherosclerosis.

Imaging atherosclerosis using computed tomography

CT reconstructs an image based on the differing x-ray attenuation of body tissue.100 The addition of intravenous X-ray contrast medium allows visualization of coronary lumen allowing reliable exclusion of significant stenosis with high negative predictive values (97–99%).101,102 This has led to widespread acceptance of the utility of coronary CT angiography (CTA).103 Importantly, CT has the capacity to visualize the vessel wall in addition to the lumen. This opens up the potential for more precise quantification and characterization of atherosclerotic plaque.

Coronary artery calcium

Calcium within atherosclerotic plaque has much higher attenuation values than other plaque elements. The development of electron beam CT (EBCT) allowed quantification of calcified plaque on non-contrast imaging—the Agatston score.104 Coronary artery calcium (CAC) scoring has well-validated prognostic implications. In a registry of over 25 000 patients, a calcium score of 0 conferred a very low event rate with a 12-year survival of 99.4%.105 The measurement of calcified plaque alone does have significant limitations. It is possible to have significant stenosis despite a reassuring CAC score of 0 with plaque composed entirely of non-calcified elements. This occurred in 4% of symptomatic patients in one series.106 In addition, extensive calcification may represent a ‘burnt-out’, less biologically active stage of the atherosclerotic disease process. In spite of impressive reductions in serum lipids and cardiovascular events following administration of statins, progression of CAC scores remains unaffected, making CAC an unsuitable endpoint for treatment trials.107,108 This disconnect is likely due to the fact that statins may lower inflammation as well as cholesterol levels, and may reduce the progression from inflammation to calcification, a process not measured by calcium scoring using CT.

Detection of coronary plaque

Reliable detection of non-calcified coronary plaque was made possible by development of multi-detector CT (MDCT) with faster gantry rotation speeds.109 CTA has excellent diagnostic accuracy for detection of coronary plaque when compared with intravascular ultrasound (IVUS) with a sensitivity of 0.90 and a specificity of 0.92.110 When compared with the gold standard of histopathology, CTA detected 100% of advanced plaques (Stary IV–VIII) but only 29% of early plaques (Stary I–III).111 Broadly speaking, coronary plaque has been divided up into calcified plaque (high attenuation), non-calcified plaque (low attenuation) and partially calcified or ‘mixed’ plaque. Detection of coronary plaque by CTA, even if non-obstructive, has been associated with an increased risk of cardiac events with detection of mixed plaques carrying the worst prognostic significance.112114

Coronary plaque characterization

MDCT allows more opportunity to characterize the components of non-calcified plaque (Figure 7). Studies comparing CT with IVUS and ‘post–mortem’ histology have confirmed that lipid-rich plaque has lower attenuation than fibrous plaque.115,116 CT-defined total plaque area correlates well with histology, but CT overestimates calcified plaque volume117 The very high attenuation values of calcified plaque lead to partial volume effects and overestimation of the size of the calcified plaque (also known as a ‘blooming’ artefact).118 Using pre-defined attenuation ranges to measure volumes of fatty, fibrous and calcified plaque for comparison with virtual histology (VH)–IVUS has previously been unsuccessful.119 However, recent attempts with high-quality CT data sets and meticulous co-registration of CT and VH–IVUS images have shown reasonable correlation for calcified plaque (r = 0.43), fibrous plaque (r = 0.47), and fatty plaque/necrotic core (r = 0.25).120 Serial CT measurement of low-attenuation plaque, representing lipid-rich necrotic core, has revealed that its volume can be seen to decrease following statin treatment.121

Figure 7

Comparison of plaque composition on MDCT and VH-IVUS. Curved MPRs of coronary arteries with transverse sections (inset) at level of the white arrows demonstrating: (A) Non-calcified plaque with corresponding VH-IVUS image showing predominantly fibrous (green) plaque. (B) Calcified plaque (high attenuation) with corresponding VH-IVUS image showing significant calcified (white) plaque. (C) Low attenuation plaque and ring enhancement with corresponding VH-IVUS image showing significant necrotic core (red). MDCT, multi-detector computed tomography; VH-IVUS, virtual histology intravascular ultrasound; MPR, multiplanar reformat.

Identification of vulnerable plaque

Invasive identification of the vulnerable TCFA plaque type in vivo using VH–IVUS has been shown to predict clinical outcomes.122 Prospective non-invasive identification of TCFA would be an important step in risk-stratifying patients. Unfortunately, direct visualization is currently not possible as vulnerable thin caps are beyond the spatial resolution of the most advanced MDCT scanners (∼320 µm). However, CTA can be used to identify other features of vulnerability including large necrotic core, positive remodelling, and the presence of ‘spotty’ calcification within the plaque (calcified plaque <3 mm); these have been shown to be more prevalent in patients presenting with ACS than stable angina.113 Detection of small necrotic cores is again limited by spatial resolution, however, a study of 64-slice CT in large proximal coronary segments allowed the correct identification of 70% of lipid pools pre-defined by IVUS.123 Detection of positively remodelled segments with CT has shown good correlation with IVUS.124 Interestingly, these features were not more prevalent in plaques shown on CT to be causing greater stenosis.125 (Figure 8). In a prospective study of >1000 patients followed for 2 years the presence of positive remodelling or low attenuation (<30 HU) were both associated with subsequent acute coronary syndromes.114

Figure 8

Plaque features on MDCT associated with culprit lesions in acute coronary syndromes (adapted from Motoyama et al.113 with permission.). Top panels: (A) Volume rendering. (B, C) Curved MPR. (D) Coronary angiogram. Bottom panels: (A) Volume rendering. (B) Curved MPR. (C) Coronary angiogram. White arrows show site of maximal luminal stenosis. Yellow arrows demonstrate site of positive remodelling. Green blocks represent non-calcified (fibrous) plaque and red blocks represents low attenuation (<30 HU) plaque. Pink arrows show spotty calcification. MDCT, multi-detector computed tomography; MPR, multi planar reformat.

Future developments

Although CTA is a promising technique for identifying vulnerable plaque there remain significant limitations to be overcome. The use of CT raises concerns over radiation exposure.126 This varies depending on the patient, scanner, and protocol used, but is usually between 2 and 20 mSv.127 Multiple dose reduction strategies have been devised and in favourable patients, using prospective-gating and high pitch protocols, doses of <1 mSv are now possible.128 Further reductions in dose may be possible with the use of new adaptive statistical iterative reconstruction techniques that allowing reduced tube current but preserved signal-to-noise ratios.129

Ultimately, tissue differentiation by CTA is limited by overlap of the attenuation ranges between plaque components, especially between necrotic core and fibrous tissues. Multi-energy CT may provide a solution.130 Plaque has also been interrogated at multiple energies using spectral CT or ‘photon counting’ and this has been shown to permit differentiation of contrast from other tissue types.131 Finally, new contrast agents may be able to target inflammatory components in areas of potential vulnerability. The nanoparticle N1177 has been shown to track macrophage accumulation in plaque in an animal model.132 Further discrimination may be possible using a combination of targeted contrast agents and multi-energy CT. The macrophage-targeted nanoparticle (Au-HDL) contains gold that can be discriminated from iodine-based contrast and calcium using spectral CT133 (Figure 9).

Figure 9

Discrimination of plaque components using targeted contrast agents and spectral computed tomography (from Cormode et al.133 with permission.). Spectral computed tomography (multicolour) showing separation of Calcium, iodine rich contrast and the high-density lipoprotein nanoparticle contrast agent (Au-HDL), for characterization of macrophage burden.


With recent focus moving from the arterial lumen onto the wall, non-invasive imaging of atherosclerosis is crucial to reveal underlying disease, to help evaluate new drugs and improve current prognostic algorithms. Several pathological aspects of atherosclerosis can now be imaged and quantified non-invasively. Both ultrasound and MRI can provide information, without ionizing radiation, on plaque volume and composition in multiple arterial territories.

Atherosclerosis imaging with FDG-PET/CT can quantify inflammation and is now being used to assess the efficacy of novel therapies. Nuclear imaging, however, remains expensive, exposes subjects to ionizing radiation and is currently not well suited for imaging coronary plaque. The results of prospective, outcome-driven trials are needed to define the role and cost-effectiveness of FDG-PET in risk prediction. Besides searching for more macrophage-specific tracers, research is also advanced in using PET imaging to detect other features of high-risk atheroma including hypoxia,67,134 neovascularization135 and dynamic calcification.93 Hardware developments, such as the introduction of combined PET and MRI scanners, should help to harness the advantages of each modality.

With recent advances in CT technology, coronary plaque can now be reliably detected by CTA. However, accurate plaque characterization remains difficult. While some features of plaque vulnerability detected by CT have been prospectively validated, better discrimination of the non-calcified plaque elements is required before CT can be used to confidently identify ‘vulnerable plaque’ and ‘vulnerable patients’. In addition, prospective clinical trials are still needed to demonstrate that such imaging approaches do indeed result in an improvement in patient outcomes, particularly if they are to be applied to asymptomatic subjects.136

Conflict of interest: none declared.


Work described in the review was part-supported by the NIHR Cambridge Biomedical Research Centre (FRJ/DRO/JHFR).


View Abstract