Evidence from the first prospective studies of coronary atherosclerosis demonstrated that intravascular imaging has limited accuracy in detecting lesions that are likely to progress and cause future events, and divided the scientific community into experts who advocate abandoning this quest and others who suggest intensifying our efforts improve and optimise the available imaging techniques. Although the current evidence may not justify the use of invasive or non-invasive imaging in the clinical setting for the detection of vulnerable, high-risk lesions, it is apparent that imaging has provided unique insights about plaque pathophysiology and evolution. Recent evidence indicates that both invasive and non-invasive imaging also provides useful prognostic information in patients with established coronary artery disease and in asymptomatic individuals and is likely to enable more accurate risk stratification. Future studies are anticipated to provide further insights about the value of novel hybrid imaging techniques, which are expected to enable complete assessment of plaque pathophysiology, in detecting vulnerable lesions and identifying high-risk patients that would benefit from new aggressive treatments targeting coronary atherosclerosis.
Statistics from Altmetric.com
The advances in the treatment of patients with ischaemic heart disease (IHD) have reduced mortality and improved their quality of life. However, the number of patients suffering and dying from IHD has increased in the recent years, and today IHD constitutes the leading cause of death worldwide. This paradox, which can be attributed to lifestyle changes and the increased life expectancy, underscores the need to develop effective prevention strategies that will reduce the incidence of IHD. The clinical scores, however, developed to predict outcomes had a low accuracy in detecting high-risk individuals;1 therefore, efforts have been made over the recent years to understand the mechanisms regulating plaque growth, to stratify more accurately risk and predict lesions that will cause events.
The in vivo assessment of plaque morphology became feasible in the beginning of the 1990s with the development of intravascular ultrasound (IVUS) that enabled identification of plaque characteristics associated with increased vulnerability. The first, small-scale IVUS studies provided promise that intravascular imaging can detect plaques that are prone to progress and cause events.2 However, the recently published results of prospective, large-scale, invasive imaging-based studies of coronary atherosclerosis raised concerns about the value of invasive imaging to predict lesions that will progress and cause events, and divided the scientific community.3–5 Today, a considerable number of researchers express their scepticism about the value of imaging to detect vulnerable lesions and argue that we should shift our focus to more realistic targets such as the accurate risk stratification and the identification of high-risk patients; while others advocate that the in vivo imaging of plaque pathology has provided unique insights about plaque growth and suggest that the improvement of the available imaging techniques and the design of hybrid modalities would allow more detailed and accurate assessment of plaque pathophysiology.6 ,7 In this article, we review the current evidence stemming from invasive and non-invasive-based imaging studies of coronary atherosclerosis, present the arguments raised by both opponents and discuss the future role of imaging in the study of atherosclerosis.
Insights and challenges of intracoronary plaque imaging
Three prospective intravascular imaging-based studies of coronary atherosclerosis have examined the potential value of imaging in detecting lesions that will progress and cause events in patients with IHD.3–5 The PROSPECT study and the VIVA study used radiofrequency IVUS (RF-IVUS) imaging to identify lesions that are likely to progress and cause cardiovascular events and showed that a minimum lumen area <4 mm2, an increased plaque burden (>70%) and a thin-cap fibroatheroma (TCFA) phenotype were predictors of lesions that progressed and caused events. Nevertheless, the positive predictive value of this high-risk triplet in detecting culprit lesions was only 18.2% in the PROSPECT study, while in the VIVA study the number of the events was too small to allow identification of RF-IVUS-derived independent predictors of culprit lesions. Finally, the PREDICTION study investigated the implications of the local haemodynamic forces on plaque progression and on the prediction of lesions that will cause events in 506 patients, and showed that low endothelial shear stress (ESS) and an increased (>70%) plaque burden can predict with positive predictive value of 41% lesions that will progress and require treatment with percutaneous coronary intervention (PCI).5
The above-mentioned studies not only provided evidence that intravascular imaging-derived plaque characteristics allow prediction of culprit lesions, but also highlighted the limitations of invasive imaging in detecting high-risk plaques. First, the predictive value of invasive imaging was limited (ie, 18% in the PROSPECT and 41% in the PREDICTION study). Second, a considerable number of the recruited patients were excluded from the analysis because of incomplete data (10.6% in the PROSPECT and 33% in the PREDICTON). Third, as has been shown in the PROSPECT study, intravascular imaging did not allow assessment of the entire coronary tree and was able to image only 55 from the 104 (53%) lesions that caused future cardiovascular events. Fourth, most of the revascularisations performed in the PROSPECT (94%) were due to progressive or unstable angina, while in the PREDICTION study only 26% of the PCIs were performed because of angina symptoms, and only 7.5% to treat lesions that caused an acute coronary syndrome (ACS). Therefore, there is a lack of robust evidence supporting the role of invasive imaging in detecting lesions that will progress and cause death or myocardial infarction, that is, events that would have irreversible implications on the prognosis and patients’ well-being. Finally, an important limitation of intravascular imaging is the risk of complications, which was 0.6% in the PREDICTION study but raised to 1.6% in the PROSPECT study.
The scepticism about the value of intravascular imaging to detect vulnerable plaques is further supported by the fact that 1/3 of the events are due to plaque erosion—the underlying mechanism of which is poorly understood—occurring in plaques that have a different phenotypic characteristics (ie, pathological intimal thickening (PIT) or FA) and to calcific nodules.8 Moreover, recent histology-based studies demonstrated that RF-IVUS is unable to provide reliable plaque characterisation especially in complex calcified lesions and in stented segments.9 In addition, cumulative evidence has shown that plaque rupture is not always associated with a cardiovascular event as quite often it tends to heal and transform to stable lesions.10 Finally, there are data to suggest that plaque morphology has a dynamic nature and changes over time. Kubo et al,11 in a landmark analysis that included 99 patients, demonstrated that most of the RF-IVUS-derived TCFAs (15 out of the 20 TCFAs) regressed to more stable phenotypes at 1year follow-up and that in the same follow-up period new TCFAs (12 lesions) were developed. These findings raise concerns about the importance of detecting and treating invasively vulnerable plaques as their phenotype constantly change and thus are likely to regress by themselves to more stable forms.
However, it should be stressed that these results are not supported by other recent reports. A similar analysis performed by the same research group in patients admitted with an ST-elevation myocardial infarction (STEMI) showed that in most high-risk lesions (78%) the phenotype of the plaque does not change within 13 months follow-up.12 Similar were the findings of the recently reported IBIS-4 study that examined the effect of a high dose of statins on the compositional characteristics of the plaque in patients admitted with a STEMI.13 In total, 103 patients underwent RF-IVUS imaging at baseline and 13 months follow-up. Most of the non-culprit lesions (140 out of 165) had a high-risk phenotype (116 TCFAs and 24 FAs) at baseline. At follow-up, only 9 TCFAs regressed to more stable plaques, 10 FAs progressed to TCFA and 1 FA regressed to PIT. In addition, only three lesions with a stable phenotype at baseline evolved to TCFAs at follow-up. These findings were supported by the analysis of Diletti et al,14 who used combined RF-IVUS and optical coherence tomographic (OCT) imaging to characterise plaque morphology in coronary bifurcations and assess for changes in the phenotype of the atheroma at 6 months follow-up and showed that most of the lipid-rich plaques (24 out of the 27 FAs and TCFAs) did not change their morphology at follow-up.
The results of these studies contradict the findings of Kubo et al, indicating that most of the high-risk lesions do not change their morphology at short-term follow-up. Of note, a more careful review of the analysis of Kubo et al shows that most (87%) of the TCFAs that changed their phenotype regressed to FAs at follow-up, while almost all the FAs did not change at 1 year. Therefore, it is likely the dynamic changes in plaque morphology reported in this study to be—at least partially—attributed to the limited reproducibility of RF-IVUS to identify TCFA.15
The above paradigm, however, is not the only case where recent invasive imaging data increase our understanding about plaque evolution. Recently, Tian et al16 used combined IVUS-OCT imaging to assess the anatomical characteristics of ruptured plaques that caused events aiming to identify predictors that would allow differentiation of culprit ruptured from non-culprit ruptured TCFAs. OCT imaging showed that ruptured TCFAs had more thrombus and a thinner fibrous cap compared with the non-ruptured TCFAs. In addition, IVUS analysis demonstrated that the ruptured plaques that caused events had smaller lumen area and increased plaque burden compared with the plaques that ruptured and did not cause events. These results support the findings of previous IVUS-based studies, which demonstrated morphological differences between culprit, ruptured lesions and ruptured plaques that remained clinically silent.17 More importantly, this study highlights the value of combined invasive imaging for the identification of lesions that are likely to progress, rupture and cause an event, questioning arguments stating that even if imaging enables prediction of plaque rupture it cannot differentiate silent ruptures from those that will cause an event (figure 1).
The value of multimodality intravascular imaging is also underscored by another study of the same research group that used combined IVUS and OCT imaging to investigate the association between plaque characteristics and the severity of luminal stenosis.18 The studied lesions were classified into three groups according to the angiographic stenosis severity (>50%, 50–70% and >70% diameter stenosis). Lesions with >70% diameter stenosis were more likely to have a TCFA phenotype and exhibit plaque characteristics associated with increased vulnerability such as microvessels and cholesterol crystals. These findings, which are in agreement with histology studies, contradict retrospective analyses of angiographically based studies, reported in the 1980s, which support that lesions with increased vulnerability have only minor or moderate diameter stenosis on coronary angiography.19
From the above, it is obvious that although intravascular imaging techniques have limited accuracy in identifying plaques likely to become culprit lesions, they have provided unique insights about plaque pathophysiology, especially when two modalities were combined, and often changed our understanding about plaque vulnerability.
Non-invasive imaging modalities: current evidence
In contrast to invasive imaging studies that focus on patients with known IHD, non-invasive imaging provides unique opportunities for the study of coronary atherosclerosis in asymptomatic individuals. It has lower resolution than intravascular imaging, but it is safer and allows assessment of plaque characteristics in the entire coronary tree. MRI appears able to characterise the composition of the plaque, but the motion artefacts and the increased time required for high-resolution imaging constitute significant limitations, which have not permitted its use for the study of coronary plaque pathology.
On the other hand, computed tomography coronary angiography (CTCA) has been established as an attractive modality for the study of coronary atherosclerosis. Several reports have shown that it allows accurate evaluation of the lumen and outer vessel wall borders, and characterisation of the composition of the plaque,20 while CTCA-based studies in asymptomatic patients and in patients with suspected coronary artery disease have provided convincing evidence that it provides useful prognostic information and accurate risk stratification.21–23 In addition, histology-based studies indicate that CTCA also provides information about the phenotype of the plaque, as a napkin-ring sign constitutes a typical signature of high-risk atherosclerotic lesions with an increased lipid core component (figure 2).24 A retrospective analysis that included 895 patients who underwent CTCA for clinical purposes demonstrated that CTCA-derived plaque morphology permits identification of lesions, which are prone to progress and cause adverse cardiovascular events.25 Interestingly, the sensitivity and specificity of the napkin-ring sign in detecting these lesions was 41% and 97%, respectively, which are far higher than those reported by RF-IVUS-based studies, a fact that could be attributed to the different study design (eg, retrospective vs prospective design) and different clinical characteristics of the patients recruited into these studies.3 ,4 The value of CTCA in identifying high-risk plaques is also highlighted by the findings of a large retrospective analysis, which included 1650 patients and showed that the lesions that caused events during follow-up had different morphological characteristics on baseline CTCA (an increased plaque burden, a lower attenuation and a smaller lumen area) from those that did not progress.22
The predictive value of CTCA in patients with known IHD has been examined only in a small-scale retrospective study that included 169 patients admitted with an ACS who had non-invasive imaging after PCI.26 The studied population was followed up for 5 years. The presence of obstructive lesions (diameter stenosis >50%) and an increased plaque burden were independent predictors of future events. The addition of CTCA-derived variables onto the model developed from the clinical variables improved its prognostic accuracy considerably (from 0.68 to 0.76). However, significant limitations of this analysis are the small number of patients and the fact that it did not take into account the angiographic-derived metrics, which have a well-established predictive value.27
Novel modalities for assessing plaque pathophysiology
Advances in intravascular imaging
Over the recent years, efforts were made to overcome the inherent limitations of the first intravascular imaging techniques, that are, their low accuracy in assessing plaque morphology and their inability to assess all the plaque characteristics that appear to determine plaque growth, namely the luminal dimensions, plaque burden, plaque composition, the local ESS and the presence of inflammation. New modalities for more accurate characterisation of the phenotype of the plaque are under development. These include near-infrared spectroscopy (NIRS), photoacoustic imaging, near-infrared fluorescence imaging (NIRF), Raman spectroscopy and time-resolved fluorescence spectroscopy. Hybrid dual-probe catheters (ie, combined IVUS-OCT, NIRS-IVUS, IVUS-photoacoustic imaging, OCT-NIRF, IVUS-NIRF), which are currently undergoing preclinical evaluation, appear able to enable more complete evaluation of plaque pathophysiology.28 Although several feasibility and validation studies in animal models have provided evidence about the potential value of these novel modalities in the study of atherosclerosis, clinical data are available today only for NIRS and NIRS-IVUS imaging. Two NIRS-IVUS studies conducted by Madder et al29 ,30 in patients admitted with a STEMI, non-STEMI and unstable angina demonstrated that the culprit lesions have an increased NIRS-derived lipid component and plaque burden, which differentiate them from other non-culprit plaques. These findings support the concept that vulnerable plaques have a specific morphological signature that distinguish them from other lesions, regenerating hope that a reliable imaging modality will enable detection of high-risk lesions that will cause an event. Of note, there is no prospective evidence to substantiate this argument and currently two studies are ongoing, the PROSPECT II (NCT02171065) and the Lipid Rich Plaque Study (NCT02033694), that examine the potential of combined NIRS-IVUS imaging in detecting lesions that will progress and cause events.
Hybrid CTCA-based approaches
Similar to intravascular imaging, CTCA does not allow evaluation of all the plaque characteristics associated with increased vulnerability as it is unable to detect inflammation and quantify the local haemodynamic forces. To overcome the first drawback, combined positron emission tomography (PET)-CTCA imaging has been proposed. The first tracer used to detect inflammation was fludeoxyglucose F18 (18F-FDG), but this is also taken up by the myocardium and thus its distribution cannot accurately reflect vessel wall biology.31 18F-fluoride has been introduced recently to address this pitfall. Small-scale, feasibility studies have provided evidence about the efficacy of this tracer, and recently the Prediction of Recurrent Events with 18F-fluoride (NCT02278211) study has commenced with the aim to investigate the prognostic value of PET-CTCA imaging in 700 patients admitted with an ACS and examine the ability of this technique to detect lesions that will progress and cause events.32
CTCA-derived models processed with fluid dynamics techniques have been used to assess the local haemodynamic forces, and recently a small-scale study demonstrated that CTCA-derived plaque characteristics including the local ESS can predict lesions that will exhibit plaque progression at 3 years follow-up with an accuracy of 59%.33 However, further research is required in the context of a prospective large-scale clinical study in order to examine the potential value of the local ESS and CTCA-derived plaque characteristics in detecting vulnerable plaques (figure 3).
Role of imaging in risk stratification: focus on vulnerable plaques or on vulnerable patients?
Although recent advances in image processing and the miniaturisation of medical devices have provided a plethora of modalities that opened new horizons for assessing plaque pathophysiology, there is scepticism about their value in the research arena. This is based not only on the findings of previous studies, but also to the fact that the final act of atherosclerosis is regulated by numerous systemic factors such as blood viscosity, platelet activity, fibrinogen levels and the interaction between the coagulation and fibrinolytic system.34 Today, the evidence about the prognostic implications of the presence of vulnerable plaques in the coronary tree in patients with no cardiac history comes mainly from retrospective CTCA-based analyses (table 1).23 ,25 ,35–37 Two CTCA studies have demonstrated that patients with high-risk plaque (ie, attenuated plaques with positive remodelling and luminal stenosis >70%) were at an increased risk of having cardiovascular events at mid-term follow-up.36 ,37 In the study of Motoyama et al,36 the composition of the plaque appeared to increase the prognostic accuracy of the model created from the baseline demographics in predicting future events (area under the curve: 0.87 vs 0.81).
Nevertheless, the additive prognostic value of plaque morphology over the genetic profile and/or circulating biomarkers is yet unclear. The Atheroremo, which is the only prospective study designed to answer this question, has significant limitations as imaging was restricted in only one coronary artery per patient, and the number of the included patients and the events reported at 1 year follow-up were not sufficient so as to assess independent predictors in multivariate analyses.38–40
The BioImage study that was designed to evaluate the predictive value of clinical demographics, biomarkers and imaging markers in identifying high-risk patients among asymptomatic individuals is anticipated to provide answers about the prognostic value of imaging in this group of patients. The 3-year follow-up of the study has been reported recently and demonstrated that assessment of carotid atherosclerosis or coronary calcium score considerably increases the accuracy of the model created from patients’ demographics; however, the additive prognostic value of non-invasive imaging over circulating biomarkers is yet unclear.41 In addition, the BioImage study did not focus on the evaluation of plaque composition and burden.
Apart from the lack of convincing evidence, the cost of intravascular imaging (ie, in the UK, IVUS and IVUS-NIRS costs £500, and OCT £600) and, more importantly, the increased time required to analyse the acquired data are additional significant limitations that restrict its broad application as a prognostic tool. For example, the analysis of the RF-IVUS imaging data is approximately 3 h per vessel; this time increases considerably to process data acquired in patients who had IVUS and OCT imaging, and increases to 13 h in cases in which these data are used to reconstruct coronary anatomy, perform blood flow simulation and estimate the ESS. From this perspective, CTCA (cost £550) appears as a more attractive alternative as the time needed for the segmentation of CTCA data is <2 h per vessel, while recent advances in image processing have reduced considerably the processing time allowing border detection, blood flow simulation and estimation of the fractional flow reserve in <1 h.42
Future role of imaging in the study of atherosclerosis
Unfortunately, the models proposed to stratify risk in asymptomatic individuals have a moderate accuracy (see online supplementary table S1). Genetic-based risk scores, novel biomarkers and imaging-based variables are anticipated to be introduced in the future to allow more accurate identification of high-risk individuals. It should be stressed, however, that in asymptomatic individuals the event rate is low (range 0.60–1.58%), and therefore, even if future studies provide robust evidence about the prognostic value of imaging, further analyses will be needed to examine its cost-effectiveness before advocating its broad use to predict outcomes. An attractive option might be a stepwise approach, similar to that proposed in the BioImage study. Risk models that rely on clinical variables and circulating biomarkers can initially be used to identify intermediate-risk/high-risk subjects who will then undergo non-invasive imaging to enable identification of those individuals who are likely to sustain a cardiovascular event and who should be treated more aggressively.43
In contrast to the event rate in asymptomatic subjects, the 1-year cardiovascular event rate in patients admitted with an ACS is high (>15%).44 Of note, the traditional risk scores proposed to stratify risk in this population have a moderate accuracy, and several invasive imaging studies have shown that plaque burden and composition provide useful prognostic information (see online supplementary table S2).38 ,39 ,45 Nevertheless, the evaluation of plaque pathophysiology can be recommended only if it provides additional prognostic information over clinical demographics, circulating biomarkers and coronary angiography, and this should be examined in the context of a prospective large-scale clinical study of coronary atherosclerosis (figure 4).
The identification of higher-risk individuals among patients with known IHD is of upmost importance in the current era where several novel interventions have been proposed. The PEGASUS study has recently demonstrated that prolonged dual-antiplatelet treatment is associated with better clinical outcomes at the cost of an increased risk of bleeding.46 In addition, numerous medications that target vascular inflammation or aim to optimise patients’ lipid profile are currently undergoing preclinical or clinical evaluation and are anticipated to increase our therapeutic arsenal for the treatment of coronary artery disease.47 However, most of these new therapies have significant limitations as some of them are expensive (ie, PCSK9 inhibitors), can be administered only subcutaneously (ie, Apo-A1 Milano, evolocumab), while others are associated with an increased risk of complications (eg, immune-suppressants such as methotrexate can increase the risk of inflammation or cause bone marrow suppression). The invasive or non-invasive assessment of plaque morphology, composition and burden is anticipated to enable more accurate identification of high-risk patients that would have a net benefit from these treatments.
In addition, nanotechnology has introduced revolutionary therapies targeting specific plaque characteristics and is expected to permit focal treatment of high-risk lesions.48 ,49 Recent advances in stent technology and the introduction of the bioresorbable scaffolds (BRSs) have provided an attractive option for the invasive passivation of high-risk vulnerable plaques. Evidence from the first applications of these devices in the clinical arena has shown that BRSs facilitate the development of a thick layer of fibromuscuclar tissue that seals the underlying high-risk plaques without compromising the luminal dimensions; they enable restoration of vessel vasomotion and physiological function and create an atheroprotective ESS environment, while in the long term they have the ability to resolve liberating the vessel from its ‘cage’.7 Today, there are no data to support the use of BRSs in this setting; evidence is expected to derive from the PROSPECT ABSORB study that has recently commenced and aims to investigate the feasibility of BRSs in sealing vulnerable plaques.
Although the first intravascular imaging techniques have significant limitations in assessing plaque pathology and biology, they have provided unique insights about plaque vulnerability and shed light into the mechanisms associated with plaque evolution, demonstrating links between the plaque phenotypic characteristics, clinical presentation and prognosis.39 ,45 The limited data that are available today do not allow us to appreciate the clinical importance of vulnerable plaque detection. Further research is required to investigate the independent prognostic value of plaque assessment in different groups of patients and identify imaging-derived independent predictors (ie, plaque burden, plaque composition and/or plaque morphology) that determine clinical outcomes. Acknowledging the fact that the available imaging modalities are unable to completely assess plaque morphology and biology, we should not be discouraged from the results of the first studies, but intensify our efforts to optimise imaging techniques and assess more reliably plaque pathophysiology. A precise identification of vulnerable plaques is also likely to permit identification of high-risk individuals that would benefit from an aggressive pharmaceutical treatment. The debate on the value of vulnerable plaque detection is open and answers can be given only through research (figure 5).
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online supplement
Contributors CVB drafted the manuscript. The other authors revised it critically for important intellectual content and gave final approval of the version to be published.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.