Introduction

Cardiovascular diseases are the leading cause of morbidity and mortality in most countries around the world. Despite significant progress in the prevention and treatment strategies, an estimated 1.1 million Americans will suffer a major adverse coronary event in 2014 and 34 % of those will die of it. The most common cause of coronary heart disease is the atherosclerosis of the coronary arteries [1]. Coronary atherosclerosis is a chronic, multifocal disease in which the complex interactions of pathological processes precipitate the accumulation of atherosclerotic plaques [2–4]. According to autopsy studies the majority of cardiac events are caused by plaque rupture and subsequent thrombosis [5–7]. The morphology and structure of rupture-prone plaques is different from stable lesions, which provide potential target for non-invasive imaging techniques to identify high-risk plaques before they cause clinical event [8]. The cardiovascular event rate is strongly associated to the presence and extent of atherosclerosis therefore the characterization and quantification of coronary artery disease is of utmost importance in order to identify the patients at highest risk to suffer an acute coronary event.

Coronary CT angiography (CCTA) and its ability to detect and quantify coronary artery disease (CAD) have evolved rapidly during the past decade. Large clinical studies and mate-analyses have proved that CCTA has an unparalleled diagnostic performance to identify obstructive CAD, with a sensitivity of 85–100 %, and specificity of 85–99 % [9–11]. To date CCTA is the only non-invasive diagnostic imaging modality that allows the assessment of luminal narrowing and provides direct information regarding the atherosclerotic plaques. The excellent image quality of state-of-the-art CT scanners enables robust plaque quantification and characterization, which opens new avenues in cardiovascular risk stratification and anti-atherosclerotic therapy response monitoring. The goal of CCTA plaque characterization is to identify the vulnerable plaque and ultimately the vulnerable patient. Therefore, at the dawn of personalized medicine CCTA might have an important role to guide stratified therapeutic approaches based on coronary plaque characteristics. In this review article we provide detailed description of the current status of CCTA on morphologic and functional plaque characterization, in order to identify vulnerable patients, who are at highest risk to develop acute coronary syndromes.

The Concept and Morphologic Features of Vulnerable Plaque

Histopathologic investigations have demonstrated that most of the acute coronary events are caused by the sudden thrombotic occlusion of the coronary lumen due to atherosclerotic plaque rupture [12]. These atherosclerotic lesions are characterized by large necrotic core, which is covered by a ruptured thin layer of fibrous cap. Precursor lesions of plaque rupture are termed as thin cap fibroatheroma (TCFA) in histology, with a cap thickness of <65 μm. Importantly, the current CT technology cannot depict and measure fibrous cap thickness due to the limited spatial resolution of the scanners (≈400 μm). However, high-risk plaques tend to be large, with a necrotic core length of 2 to 17 mm (mean 8 mm). In addition the area of lipid rich, necrotic core is larger than 1 mm2 in 80 % of cases [13]. In an ex vivo study CCTA safely identified lesions with at least 1 mm thickness [14]. Thus, it is reasonable to assume that in some extent CCTA is able to visualize and quantify coronary plaque features that are characteristic of vulnerable lesions. The concept of rupture prone vulnerable plaque underwent remarkable evolution during the past decade. From a sole morphometric plaque feature it became a probabilistic concept that refers to plaques that may be associated with future coronary events. CCTA can detect certain compositional and geometrical features of plaques that have been associated with cardiac events. In the following sections we provide an overview regarding the potential targets of CCTA imaging to identify vulnerable plaques with an increased risk to rupture and cause acute coronary event.

Plaque Burden

Intravascular ultrasound (IVUS) is the clinical reference standard for the quantitative evaluation of coronary atherosclerosis. According to the results of the IVUS based Providing Regional Observations to Study Predictors of Events in the Coronary Tree (PROSPECT) trial increased plaque burden is the strongest predictor of recurrent coronary events in patients who suffered acute coronary syndrome (ACS) [15]. Importantly, the use of IVUS is limited in patients with severe coronary stenosis or occlusion. In addition, IVUS-based plaque measurements are typically limited to 1 or 2 coronary segments, which might limit its applicability to assess the overall plaque growth [16]. The state-of-the-art CT scanners with a sub-millimeter spatial resolution provide a unique opportunity to detect, quantify and characterize coronary plaques globally, along the entire coronary tree (Figs. 1 and 2).

Fig. 1
figure 1

Double oblique reformations of CCTA images represent longitudinal sections of proximal LAD segments (upper image panels) and cross-sections of the same segments (lower panels). Panels a and d: non-calcified plaque with positive remodelling; Panels b and e: partially calcified plaque with spotty calcification; Panels c and f: calcified plaque. CCTA coronary computed tomography angiography. LAD left anterior descending coronary artery

Full size image
Fig. 2
figure 2

Representation of plaque quantification using dedicated semiautomated software tool in CCTA data set. Panel a: LAD curved multiplanar reformation, white arrowheads show a non-calcified plaque with high risk features (positive remodelling and low mean plaque attenuation); white arrows indicate a partially calcified plaque with high risk features (positive remodelling, low mean plaque attenuation and spotty calcification). Panel b and c: plaque cross-section at the site indicated with dashed line. The plaque cross-section shows a large non-calcified plaque (arrow heads). The colour overlay in panel c indicates areas of low CT attenuation (red colour). Panel d: The graph illustrates the areas of different plaque components, x axis: distance fromthe LM orifice; y axis: area in mm2. CCTA coronary computed tomography angiography, LAD left anterior descending coronary artery, LM left main stem, Ca calcium

Full size image

The diagnostic performance of CCTA to detect coronary atherosclerotic plaques compared to IVUS has improved over the past decade. Pioneer studies using 16-slice scanners demonstrated an overall moderate diagnostic accuracy. The sensitivity to detect non-calcified plaque (NCP) was 53-81 % and the specificity 80-87 % as compared to IVUS. On the other hand, studies using 64-slice scanners yielded sensitivity values of 83-97 % with specificity around 89 % [17–20]. Despite the improved sensitivity values, the specificity has not changed significantly with the newer scanner technologies, probably because of the minor improvements in the in-plane resolution and contrast resolution. According to a recent meta-analysis the combined sensitivity and specificity of CCTA to detect coronary plaque is 90 % (95 % CI: 83 % to 94 %) and 92 % (95 % CI: 90 % to 93 %), respectively [21••].

During recent years several studies investigated the accuracy of CCTA against IVUS regarding the assessment of plaque burden. The accuracy was influenced by the plaque composition. The non-calcified and mixed plaque volumes were underestimated, whereas the volume of calcified plaque was overestimated [22, 23]. Brodoefel et al. used a semiautomated plaque quantification tool and demonstrated a good correlation between CCTA and IVUS regarding the volume of non-calcified plaque, however CT overestimated the total plaque volume and the volume of calcification [18]. Recently, the accuracy of automated quantification of coronary plaque by CCTA performed with modern CT scanners was successfully validated against grey scale IVUS and virtual histology IVUS [21••, 24••, 25]. Moreover, the reproducibility of such automated 3-dimensional quantification software for plaque burden was excellent with an intra-class correlation value of 0.88 (95 % CI 0.74 – 0.95) [24••]. Notably, the automated coronary plaque assessment tools demonstrated a good intraplatform reproducibility, however the interplatform variability proved to be significant [26]. Therefore, it is advisable to use the same software tool in longitudinal plaque quantification studies until industry standards are developed to guarantee reproducible plaque assessment regardless of the type of scanner and software used.

Plaque assessment with CCTA in patients with stable angina pectoris and patients who suffered acute coronary syndromes demonstrated significant differences in plaque burden between these two patient populations. Recent cross-sectional clinical investigations found that the culprit lesions in patients with unstable angina or ACS have a larger volume than stable plaques in SAP patients (193-313 mm3 vs. 104-118 mm3, p < 0.001) [27, 28] These findings are in line with previous histological observations that described larger dimensions for culprit lesions compared to stable fibrotic plaques [13]. Moreover, several CCTA studies have demonstrated a strong prognostic value of increased plaque volumes. Motoyama et al. showed that SAP patients who suffered ACS during a follow-up period of 27 ± 10 months had larger coronary plaque volume at baseline (134.9 ± 14.1 mm3 vs. 57.8 ± 5.7 mm3, p < 0.001) [29]. In a subsequent prospective study Kristensen et al. described that total volume of non-obstructive non-calcified plaque is independently associated with recurrent acute coronary event in patients who suffered non-ST-segment elevation myocardial infarction (NSTEMI). During the 16 months follow-up period NCP was independently associated with the recurrent acute events with a hazard ratio of 1.18 per 100 mm3 non-calcified plaque volume increase. Notably, the amount of calcified lesions was not associated with an increased risk. [30••] Similarly, Versteylen et al. found that plaque burden as assessed by a semiautomated software tool provided incremental prognostic value over traditional risk factors and conventional CT readings. Patients who suffered ACS had a larger total NCP volume at baseline compared to patients who did not develop ACS (28 mm3 vs. 4 mm3, p < 0.001) [31].

Plaques with Low CT Attenuation

Plaques prone to rupture contain large, lipid-rich necrotic core, whereas stable lesions are characterized by the fibrous plaque components [32]. Therefore, the differentiation of lipid-rich versus fibrous plaques is desirable, as it might provide valuable information regarding plaque vulnerability (Fig. 2). A study conducted by Kopp et al. was the first to demonstrate the feasibility of plaque attenuation measurements and limited plaque differentiation based on intra-plaque CT numbers [33]. They have demonstrated that plaques described as “soft” by intravascular ultrasound (IVUS) had lower CT numbers as compared to “intermediate” and “calcified” lesions [33]. Subsequent studies differentiated three types of plaques depending on the presence and amount of calcification: calcified, non-calcified and partially calcified (mixed) plaques (Fig. 1). A significant difference in Hounsfield units (HU) was observed between calcified and non-calcified plaque components (mean values across studies, 490 HU and 75 HU, respectively) [20, 23, 34–38]. In an ex vivo IVUS and CCTA study Becker et al. showed that the mean CT attenuation of lipid-rich lesions was 47 ± 9 HU and predominantly fibrotic plaques showed a 104 ± 28 HU mean attenuation [35]. Marwan et al. have used histogram analysis of the intraplaque pixel HU distribution and found significantly lower mean CT numbers of lipid-rich plaques (67 ± 31 HU) compared to fibrotic plaques (96 ± 40 HU) [39]. The pixel based plaque CT number analysis was recently validated by an ex vivo study which demonstrated that intraplaque area of >25 % of pixels with <60 HU had a good accuracy to detect lipid rich atherosclerotic lesions (sensitivity = 73 %, specificity = 71 %) [40]. Vulnerable plaques with thin fibrotic cap and large necrotic core showed lower CT numbers compared to stable, fibrotic lesions in two recently published clinical studies comparing CCTA assessment with optical coherence tomography (OCT, 35-45 HU vs. 62-79 HU, p < 0.001) [41–43].

Nevertheless, despite the significant difference in the average CT numbers, there is wide variability and substantial overlap between the CT numbers of these two plaque components [22, 23, 36, 44]. This can be explained by the similar chemical composition of lipid-rich and fibrous plaques, furthermore by the limited spatial and contrast resolutions of CT scanners. In addition, the intraluminal contrast enhancement, the tube voltage and the reconstruction filter also influence the CT numbers of coronary plaques [38, 45–47]. Therefore, the differentiation between lipid-rich and fibrous plaques based on CT numbers remains to be challenging [48]. Despite the limitations associated with plaque HU assessment, longitudinal studies have consistently demonstrated that plaques with low CT attenuation have an increased risk to cause acute coronary events. Motoyoma et al. has demonstrated that plaques with <30HU mean attenuation are more frequent in patients who suffered ACS compared to patients with SAP (79 % versus 9 %, p < 0.0001) [49]. Subsequent clinical investigations have confirmed these findings, and reported that the mean plaque attenuation was lower in patients with ACS versus patients with SAP (40-86 HU vs, 97-144 HU, p < 0.01) [27, 50, 51].

Positive Remodelling

The compensatory expansion of the vessel wall with the enlargement of the atherosclerotic lesion called positive remodelling, and results a delayed onset of luminal narrowing. This phenomenon was described by Glagov et al. in 1987 [52]. Histopathologic investigations have demonstrated that ruptured coronary plaques of patients who died suddenly tend to be large and occupy more than half of the vascular cross-sectional area [53]. However, due to positive remodelling these lesions do not cause significant narrowing in approximately two-thirds of the cases [54••]. Remodelling index (RI) is the ratio of the vessel area at maximal luminal narrowing and the reference vessel area (average of the proximal and distal reference areas), (Fig. 1, panel a) [55, 56]. The cut-off value for positive remodelling was defined by IVUS studies as ≥1.05 or >1.0, however, due to the limited spatial resolution of CCTA, Gauss et al. suggested to use a threshold of ≥1.1 to define positive remodelling in CT [39, 56]. The CCTA assessment of remodelling index was validated in clinical studies using IVUS as a reference standard [24••, 39, 55]. Coronary plaques with positive remodelling on CCTA were demonstrated to have larger amount of necrotic core and a higher prevalence of TCFA as assessed by VH-IVUS [57]. Moreover, CCTA derived remodelling index was higher in vulnerable lesions defined as TCFA as compared to stable lesions in two recently published clinical investigations where OCT was used as the reference standard (1.14 vs. 0.95 and 1.14 vs. 1.02, both p < 0.0001) [41, 42]. In addition, the authors suggested a remodelling index threshold of 1.08, which had the best diagnostic performance to identify TCFA [45]. In a study by Motoyama et al. positive remodelling was strongly associated with culprit plaques in patients with ACS (ACS 87 %, SAP 12 %, p < 0.0001). A landmark study –conducted by the same investigators– has enrolled 1059 patients and demonstrated that plaques with positive remodelling and/or with low plaque attenuation were independent predictors of ACS during the 27 ± 10 months follow-up period (hazard ratio: 22.8, 95 % confidence interval: 6.9 to 75.2, p < 0.001) [29]. To put these results in a perspective, one in five patients with at least on of these high-risk CT plaque features (positive remodelling, low CT attenuation plaque) will have an event in 1-3 years, which is very similar to the patients with high risk plaques as defined by VH-IVUS in the PROSPECT trial [15, 29].

Spotty Calcium

The presence of calcification indicates advanced coronary atherosclerosis [58]. A strong correlation was observed between coronary calcification and coronary atherosclerotic plaque burden, furthermore, high calcium score is a strong independent cardiovascular risk factor [59–61]. However, the local effect of calcification on plaque stability is controversial [62–64]. In IVUS studies spotty calcium was associated with accelerated disease progression in patients with SAP and it was related to culprit plaques in ACS patients [65, 66]. The definition of spotty calcium in CCTA is based on its small size (<3 mm) and high CT attenuation (>130HU), (Figs. 1, panels b and e and 2, panel a) [29, 49, 67].

Van Velzel el at. have studied 112 patients (53 with ACS, 59 with SAP) further stratified potty calcium into small (<1 mm), intermediate (1-3 mm) and large spotty calcium (>3 mm) [68]. According to their observations small spotty calcium showed the strongest association with vulnerable plaque features as defined by VH-IVUS [68]. Numerous clinical investigations have demonstrated that spotty calcium was more frequent in plaques associated with ACS as compared to plaques in patients with SAP [27, 49, 50, 69]. However, the frequency of spotty calcium varies widely across these studies, demonstrating uncertainty in the relationship between rupture prone lesions and spotty calcium. With further improvements in spatial resolution of the CT scanners, the detection of microcalcification might become feasible, which have the potential to improve the detection of high-risk plaques with CCTA.

Napkin-ring Sign

A recent histopathologic investiagtion by Narula et al. has demonstrated that among plaque features that are potentially accessible by non-invasive imaging the size of necrotic core and the local inflammation (macrophage infiltration) are the two best discriminators between ruptured plaque/TCFA and stable plaques [70••]. A ruptured plaque and TCFA could be separated from fibroatheromas by the presence of a large necrotic core (>3.5 mm2) [70••]. Thus it is safe to assume that state-of-the-art CT scanners might be able to depict large necrotic cores. The necrotic core contains abundant lipid-rich material, therefore a plaque cross-section with a sizable area of low CT numbers adjacent to the lumen might be indicative of a plaque with a large necrotic core. An intraplaque attenuation pattern with low CT numbers in the central part of the plaque cross-section surrounded by a ring-like higher attenuation plaque area was observed in culprit plaques of patients with ACS [27, 41, 42, 59, 71, 72]. The term napkin-ring sign (NRS) was coined specifically for this plaque attenuation pattern [73]. The NRS is a qualitative plaque feature. No CT attenuation measurement is needed to identify a NRS lesion. As noted previously HU measurements are highly influenced by several factors, among other intraluminal iodinated contrast agent concentration, tube voltage and reconstruction setting used. However, the typical NRS intraplaque CT attenuation pattern can be easily identified in a non-calcified plaque cross-section by the presence of two features: (1) a central area of low CT attenuation that is apparently in contact with the lumen; (2) this same area is surrounded by a ring-like higher attenuation plaque tissue (Fig. 3) [73–75]. An ex vivo investiagtion of human hearts demonstrated that the NRS is both visible in non-contast enhanced (native) and in contrast-enhanced images, which suggests that this plaque pattern is caused by the difference in CT attenuation between the large, lipid laden necrotic core and the surrounding ring-like fibrous tissue [73–75]. However, it is important to note that in vivo the NRS appearance might be influenced by additional factors such as vasa vasorum [59]. A subsequent ex vivo study demonstrated that the NRS plaque pattern had an excellent specificity (98.9 %) to identify advanced atherosclerotic coronary plaques and a poor sensitivity (24.4 %) [74]. The excellent specificity is useful to rule-in high-risk lesions, in order to initiate therapeutic interventions (e.g. intensified statin therapy).

Fig. 3
figure 3

Double oblique reformations of CCTA images represent longitudinal sections of a partially calcified plaque (arrowheads) in LM trifurcation (upper image panels) and cross-sections of the same plaque (lower panels). In the cross-sectional view the plaque shows a NRS attenuation pattern. The low attenuation central region (star) of the plaque is surrounded by a ring like area of higher attenuation plaque components (arrowheads, panel d). Panels a and d: filtered back projection. Panels B and E: hybrid iterative reconstruction (hybrid-IR). Panels c and f: model based iterative reconstruction (model-based-IR). CCTA coronary computed tomography angiography, LAD left anterior descending coronary artery, LM left main stem, IM ramus intermedius, Cx left circumflex coronary artery, Ca calcium, SB side branch, L lumen

Full size image

In line with the ex vivo observations the NRS pattern showed a 96-100 % specificity to identify TCFA or culprit lesions in a clinical investigation of patients with ACS [27, 28]. A recently published longitudinal study by Otsuka et al. has enrolled 895 patients who underwent CCTA examination. During the mean follow-up period of 2.3 years, the presence of NRS pattern was an independent predictor of acute coronary events (hazard ratio of 5.6, p < 0.0001) [76••].

Functional Plaque Characteristics

CCTA allows ruling out coronary artery disease with a high negative predictive value [77]. However, the number of false positive findings is relatively high, as the majority of significant stenoses detected by CCTA do not cause ischemia [78]. Large randomized controlled trials have demonstrated that fractional flow reserve (FFR) is able to determine the lesion-specific ischemia, and it is a valuable adjunct to anatomic assessment of coronary atherosclerosis using invasive coronary angiography [79–82].

Adding functional data to anatomical information may help to increase the specificity of CCTA to identify lesions with hemodynamic significance. Computational fluid dynamics (CFD) is a novel method that permits the simulation of coronary blood flow and luminal pressure alterations based on CCTA derived 3-dimensional coronary geometry, which allows the calculation of lesion specific FFR values [83]. Importantly, CFD calculations can be applied on a typically acquired CCTA scan, without any modification of CCTA protocols, additional image acquisition, or administration of medications [83]. In the Diagnosis of Ischemia-Causing Stenoses Obtained Via Noninvasive Fractional Flow Reserve (DISCOVER-FLOW) trial the CT derived FFR calculation (FFR-CT) was compared with invasive FFR, and yielded a per-vessel accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for lesions causing ischemia of 84.3 %, 87.9 %, 82.2 %, 73.9 %, and 92.2 %, respectively [83]. In addition, FFR-CT had better diagnostic performance than CCTA alone in the identification of significant coronary lesions [83]. More recently, the Determination of Fractional Flow Reserve by Anatomic Computed Tomographic Angiography (DeFACTO ) trial has enrolled 252 patients and demonstrated that FFR-CT was superior to CCTA in identifying ischemic lesions (accuracy [73 % vs. 64 %], sensitivity [90 % vs. 84 %], specificity [54 % vs. 42 %], PPV [67 % vs. 61 %], and NPV [84 % vs. 72 %]).

Recently, a histopathologic study has demonstrated that around 50 % of TCFAs produce an intermediate (50 % to 75 %) luminal area stenosis [70••]. Notably, about half of the intermediate lesions cause significant ischemia and the other half not, which demonstrates that the relationship between stenosis and ischemia is unreliable [84]. It has been proposed that in a setting of an intermediate lesion with abnormal FFR, the flow perturbations and the altered endothelial shear stress might be responsible for the development of rupture prone lesion [85, 86]. In fact, this is in line with the vast body of evidence regarding the poor prognosis of patients with ischemic lesions [87, 88]. Therefore, we can hypothesize it is safe to assume that FFR as derived from CCTA (FFR-CT) will not only help to identify luminal narrowing with significant ischemia, but it will improve the accuracy of CT to detect lesions with an increased risk to cause acute coronary event.

Plaques tend to develop at specific locations in the coronary arteries, even though promoting cardiovascular risk factors for coronary plaque formation affect the vascular bed systemically. The inner curvature of coronaries, the outer waist of bifurcations and the side-branches are predilection sites for atherogenesis. At these locations the endothelial shear stress (ESS) is low and the flow is disturbed or turbulent [89–93]. Where ESS is low, the endothelial cell gene expression shifts towards a pro-atherogenic pattern, which subsequently leads to the development of high-risk lesions [92, 94]. In the recently published Prediction of Progression of Coronary Artery Disease and Clinical Outcome Using Vascular Profiling of Shear Stress and Wall Morphology (PREDICTION), a total of 506 patients underwent three vessel IVUS examination and were followed-up for 1 year [95]. Stone et al. demonstrated that low ESS and large plaque burden independently predict plaque progression [96]. It has been shown that the application of CFD on CCTA images enables the calculation of ESS and the generation of 3D ESS-CT maps [97–100]. In a recent clinical investigation CCTA demonstrated sufficient accuracy to study the ESS distribution in the main vessels and in the bifurcation regions as compared to IVUS vascular profiling [101]. The assessment of ESS-CT might be an important addition to morphological CCTA plaque features to improve risk stratification. The comprehensive assessment of ESS-CT and FFR-CT might provide a novel functional dimension in plaque vulnerability assessment in CCTA. This combined anatomic-physiological evaluation of coronary plaques with CCTA may allow to develop novel stratified therapeutic approaches to treat vulnerable plaques and ultimately to improve event-free survival.

Novel Technical Developments in CT Plaque Assessment

Improvements in CT technology, the use of hybrid techniques and novel image reconstruction algorithms may improve non-invasive coronary plaque characterization through providing images with improved spatial and contrast resolution and adding metabolic information to morphological plaque assessment. Dual-energy CT (DECT) is a promising technique for advanced tissue characterization, first applied in the 1970s, and is based on the physical principle that tissue attenuation values change at different energy levels of X-ray [102, 103]. The recording of spectrally different attenuation datasets enable the visualization and quantification of blood supply in the myocardium and may also benefit the discrimination of atherosclerotic plaque components [104]. Vendors have developed different approaches for spectral imaging. The most clinical data regarding the cardiac applications of DECT is available with simultaneous application of two X-ray tubes at different peak voltage [105, 106]. Ruzsics et al. investigated 35 patients and were able to detect myocardial ischemia with a high degree of accuracy [105]. Another approach is the use of rapid kV switching between two different peak voltage levels by one X-ray tube during the scan [107]. The latest, photon-counting technique (also known as spectral CT) applies panels of energy sensitive detectors to capture X-ray photons both at low and high energy levels, whereas only one X-ray tube is operated at a distinct tube voltage [104]. Recently, the feasibility of atherosclerotic plaque characterization was demonstrated with phase-contrast CT in an animal study [108]. Phase contrast CT uses phase shift of x-rays passing through matter to generate tissue contras [109, 110]. This experimental imaging techniques has a great potential to improve contrast resolution as its sensitivity to light elements is almost 1000 times greater than that of the conventional absorption-contrast X-ray method [105].

Novel nanotechnology contrast agents have a great potential to identify inflammatory cell infiltration of the fibrous cap, which is a strong marker of plaque vulnerability [53]. Hyafil et al. demonstrated that the CCTA inflammatory cell detection is feasible in atherosclerotic plaques with iodinated nanoparticle contrast agent N1117 [111]. Gold-labeled high density lipoprotein (Au-HDL) nanoparticles designed to target activated macrophages imaged by spectral CT showed promising results in atherosclerotic mice model [112].

Macrophages have a high metabolic activity and they depend on exogenous glucose for their metabolism, therefore using radiolabeled glucose analog, fluorine-18-fluorodeoxyglucose (18F-FDG) to non-invasively detect vulnerable, inflamed plaques might be feasible with hybrid positron emission tomography (PET) and CT imaging (PET-CT). Rogers et al. observed an increased 18F-FDG accumulation at the culprit lesion site of patients with recent ACS as compared to lesions in patients with SAP [113]. In a recently published clinical trial Joshi et al. demonstrated that 18F-NaF PET-CT imaging allows the detection of metabolically active plaques by identifying areas of ongoing calcification activity [114••]. Metabolic imaging with hybrid techniques or nanoparticle contrast agents might shed light on the molecular aspects of coronary atherosclerosis and vulnerable plaque.

In the 1980s filtered back projection (FBP) became the standard of image reconstruction technique, because it is fast, reliable and can be performed even with limited computer processing capabilities. However, the concerns regarding ionizing radiation and the available high computational power pushed vendors to develop novel image reconstruction algorithms. While in the reconstruction process with FBP assumptions are made regarding system geometry and image reliability weakens with increased noise, iterative reconstruction (IR) techniques were developed by modelling system geometry, physics and noise statistics to achieve improved image quality [115–117]. Through repetitive reconstruction cycles (iterations), the image reconstructed from the actually measured data projection is compared to a forward projected image, simulating an ideal data acquisition. The differences among the true and simulated reconstruction help to update the original image with optimal noise reduction, while image contrast is preserved [118]. The hybrid iterative reconstruction uses data reconstructed with FBP, thus the iterations are performed in the image space, which is less time consuming [119, 120]. The reduced image noise allows scanning with reduced tube current and tube voltage, which results in significant dose reduction (up to 44-63 % reduction in effective radiation dose) [121–123]. In addition, it has recently been demonstrated that IR significantly improves objective image quality [124, 125]. Furthermore, Renker et al. investigated 55 patients with Agatston scores >400, who underwent CCTA and invasive coronary angiography. They found that IR has reduced blooming artifacts in heavily calcified lesions and improved diagnostic performance of CCTA to identify obstructive coronary plaques [126]. More advanced, model-based iterative reconstruction techniques have been introduced recently [118, 127]. Model based IR is a true 3D, raw-data based image reconstruction technique, which improves image resolution through increased point spread function [118]. Moreover, it has shown superior image quality and increased contrast-to-noise ratio as compared to FBP and hybrid-IR (Fig. 3) [118]. Model based image reconstruction improved the performance of automated coronary plaque assessment tools [127], and it has a potential to improve plaque visualization and characterization in the clinical setting (Fig. 4).

Fig. 4
figure 4

CCTA images of an obese patient (BMI: 41 kg/m2). The figure shows volume rendered reformations of CCTA images (upper panels) and axial images at the height of the LM bifurcation (lower panels). The white arrow indicates non-calcified plaque component of the partially calcified plaque in the LM bifurcation. The plaque components are clearly visible with model based iterative reconstruction (panel f), whereas the differentiation between different plaque components is limited with FBP and Hybrid-IR due to the low CNR (panels d and e). CCTA coronary computed tomography angiography, LAD left anterior descending coronary artery, LM left main stem, Cx left circumflex coronary artery, FBP filtered back projection recosntruction, IR iterative reconstruction, CNR contrast-to-noise ratio

Full size image

Conclusion

CCTA is a unique non-invasive imaging technique that allows the visualization of all main epicardial coronary arteries. It can assess individual coronary plaques as well as the global coronary plaque burden. The quantitative and qualitative high-risk plaque features in CCTA are associated with acute coronary events. Positive remodelling, low CT attenuation, napkin-ring sign, spotty calcification are morphological features, whereas ESS-CT and FFR-CT provide functional data regarding individual atherosclerotic lesions. Nanoparticle contrast agents and hybrid imaging techniques have a great potential to add valuable information regarding inflammation and metabolic activity at the site of atherosclerotic plaques. Large prospective trials are warranted to confirm the observations of smaller and to assess the effectiveness of image guided therapeutic approaches based on CCTA.