Future imaging of atherosclerosis: molecular imaging of coronary atherosclerosis with ¹⁸F positron emission tomography

Abstract

Atherosclerosis is characterized by the formation of complex atheroma lesions (plaques) in arteries that pose risk by their flow-limiting nature and propensity for rupture and thrombotic occlusion. It develops in the context of disturbances to lipid metabolism and immune response, with inflammation underpinning all stages of plaque formation, progression and rupture. As the primary disease process responsible for myocardial infarction, stroke and peripheral vascular disease, atherosclerosis is a leading cause of morbidity and mortality on a global scale. A precise understanding of its pathogenic mechanisms is therefore critically important. Integral to this is the role of vascular wall imaging. Over recent years, the rapidly evolving field of molecular imaging has begun to revolutionize our ability to image beyond just the anatomical substrate of vascular disease, and more dynamically assess its pathobiology. Nuclear imaging by positron emission tomography (PET) can target specific molecular and biological pathways involved in atherosclerosis, with the application of ¹⁸Fluoride PET imaging being widely studied for its potential to identify plaques that are vulnerable or high risk. In this review, we discuss the emergence of ¹⁸Fluoride PET as a promising modality for the assessment of coronary atherosclerosis, focusing on the strengths and limitations of the two main radionuclide tracers that have been investigated to date: 2-deoxy-2-(¹⁸F)fluoro-D-glucose (¹⁸F-FDG) and sodium ¹⁸F-fluoride (¹⁸F-NaF).

Introduction

Atherosclerosis is a slowly progressive, chronic disease process characterised by the formation of plaque within arterial walls. It is the underlying pathology of ischaemic heart disease, which is a leading cause of death worldwide accounting for 7.4 million deaths worldwide in 2012 (1). With the increasing rates of obesity and type 2 diabetes, the burden of coronary artery disease (CAD) is expected to continue to rise in future years (2).

Atherosclerosis occurs in the setting of endothelial dysfunction of arteries causing increased endothelial permeability, allowing uptake and retention of apolipoprotein B containing lipids within the subendothelial space. There the lipoproteins can become oxidised and provide an increased stimulation for the overlying endothelium to release chemokines that attract leukocytes. Monocytes transmigrate through the endothelium to enter the subendothelial space where they differentiate into macrophages and take up the retained and modified lipoproteins, eventually becoming lipid laden foam cells. Macrophages can undergo apoptosis but in the early stages of atherosclerosis these apoptotic cells are cleared by efferocytosis that has an anti-inflammatory and stabilizing effect. Over an extended period of time though the need for recurrent efferocytosis and prolonged exposure of cells within the plaque environment to oxidized low density lipoprotein (LDL) can lead to endoplasmic reticulum stress within macrophages and cause secondary necrosis, releasing their cellular contents and further stimulating the cycle and contributing to the formation of lipid rich plaques within the arterial wall (3-5).

Therefore the underlying pathogenic substrate of atherosclerosis is complex, comprising multiple extracellular, cellular and molecular mediators that participate in a cascade of disordered lipid metabolism, chronic inflammation, cell death, angiogenesis and thrombosis all taking place within the vessel wall.

Despite seminal advances in the field of cardiovascular medicine, the risk of complications from coronary atherosclerosis remains considerable, culminating in the sequelae of myocardial infarction. The ongoing need to optimize risk assessment, prevention and treatment strategies hinges on an improved understanding of the pathogenic basis of CAD. In the clinical setting, there are also the specific needs to be able to identify high risk or “vulnerable” coronary plaques and monitor their responsiveness to therapeutic interventions.

Coronary imaging is central to this. Traditional assessment of coronary atherosclerosis has focused solely on anatomical imaging, evolving over recent decades from conventional coronary angiography to intravascular modalities, such as intravascular ultrasound (IVUS) and optical coherence tomography (OCT) and noninvasive techniques, such as a computed tomography (CT). However, there is now also increasing recognition of the scope for molecular imaging to provide assessment of the biological composition of plaque and characterize the dynamic nature of more vulnerable lesions. In this review, we will direct our focus to the use of positron emission tomography (PET) and the applicability of ¹⁸Fluoride-based radionuclide tracers for molecular imaging of coronary atherosclerosis.

“Non-molecular coronary imaging”

Invasive contrast X-ray angiography has been the gold standard for imaging of coronary atherosclerosis over the past 50 years since Dr. F. Mason Sones Jr performed the first selective coronary angiogram by accident in 1958 (6). While remaining the imaging standard and the most studied in regards to patient outcomes, coronary angiography does not provide imaging of atherosclerosis itself, but rather its end result. It has the ability to show the residual coronary lumen and subsequently the degree of stenosis when compared with a reference segment of unaffected vessel, but does not directly visualize plaque or the vessel wall. As a consequence, preservation of the luminal appearance on coronary angiography does not equate to the absence of plaque, and this is particularly true where the vessel wall has undergone positive (or outward) remodeling, as often takes place especially in the early stages of atherogenesis.

The advent of IVUS and subsequent intravascular imaging modalities, such as OCT and near-infrared spectroscopy (NIRS) delivered the ability to directly image the vessel wall and atherosclerotic plaque. These intravascular imaging modalities have progressed our understanding of atherosclerosis significantly and have helped to image plaque at all stages of its development, while also defining its compositional features that are associated with plaque vulnerability. Although invasive, they have also enabled a means by which to serially monitor the natural history of plaque and its modulation with anti-atherosclerotic therapies. While having the ability to identify vulnerable characteristics, such as high plaque volume, thin fibrous cap, lipid-rich core, spotty calcification and intraplaque neovascularization and hemorrhage, they are as yet unable to determine the activity of plaque, in terms of pathogenic molecular pathways. Ideal modalities for imaging of coronary atherosclerosis should combine non-invasiveness so that patients can be assessed at repeat intervals with minimum risk, and the accurate and reproducible ability to identify early signals of plaque vulnerability that predict a high risk of progression and complication.

Molecular imaging

Molecular imaging was christened as a term in the mid to late 1990s to encompass in vivo functional imaging modalities, which go beyond anatomical tissue assessment to also visualize and quantify specific biological processes down to a cellular and molecular level. The early progress and focus of development centered on imaging in oncology but has expanded to use throughout medicine (7).

There are a large number of different molecular imaging agents that are able to target a diverse range of biological activity across the spectrum of pathologies in medicine. These typically combine a targeting component, that ideally interacts specifically with the biochemical process being investigated, and an imaging component that can attach to the targeting component without affecting its interaction with the targeted biochemical process (8). In studies of atherosclerosis unique molecular imaging agents have been created for assessment of a wide variety of processes that contribute to atherogenesis. These have included targeting of vascular cell adhesion molecule-1, monocyte recruitment, macrophage phagocytic activity, apoptosis, oxidative stress, matrix metalloproteinases, intraplaque hemorrhage, and neoangiogenesis (9-11). There are numerous imaging modalities that are used for identification of these imaging agents including PET, single-photon emission CT, magnetic resonance imaging (MRI), and ultrasound, as well as optical imaging modalities including bioluminescence and fluorescence that can be used for in vivo animal imaging. Of these the most promising at a clinical level are MRI and PET.

Positron emission tomography (PET)

PET scanning was first described clinically in the 1950s. Positron emission decay occurs when a proton is turned into a neutron with the release of a positron and a neutrino. The combination of a positron and an electron results in an annihilation reaction which produces two 511 keV photons that are emitted in approximate opposite direction to each other. This release allows identification of positron emission when detectors on opposite sides of the body record simultaneous detection, only registering those that co-detect within a few nanoseconds (12). PET and CT scans have been combined since the 1990s allowing the functional information obtained by PET to be more precisely located anatomically and three dimensionally reconstructed when co-localized with CT images (13).

There are a number of positron emitting isotopes including Carbon 11, Nitrogen 13, Oxygen 15, Gallium 68, Rubidium 82 and Fluorine 18 (¹⁸F). The most common in use in medical imaging is ¹⁸F, which is produced in a cyclotron by bombarding Oxygen 18 enriched water with high energy protons and has a half-life of 109.77 minutes (12). ¹⁸F labeling is used as the imaging component in a number of investigational molecular imaging agents, including [¹⁸F]Galacto-RGD in angiogenesis, [¹⁸F]P2,P3-monochloromethylene diadenosine-5',5"'-P1,P4-tetraphosphate in plaque inflammation, and ¹⁸F-Fluoromisonadazole in hypoxia (11). This review will focus however on the two most commonly used ¹⁸F labelled isotopes, ¹⁸F-FDG and sodium ¹⁸F-fluoride (¹⁸F-NaF).

2-deoxy-2- (¹⁸F) fluoro-D-glucose (¹⁸F-FDG)

2-deoxy-2- (¹⁸F)fluoro-D-glucose or ¹⁸F-FDG has become the most commonly used radiotracer in the world and a mainstay of imaging in oncology. ¹⁸F-FDG is a radiolabelled analogue of glucose that enters cells via the same receptors, the GLUT family of transporters. Once ¹⁸F-FDG has entered cells it is phosphorylated by hexokinase, metabolically trapping it within the cell. Accumulation of intracellular ¹⁸F-FDG is also contributed to by the absence of the hydroxyl group at the 2’ position affecting its ability to undergo further metabolism along the glycolytic pathway (14).

The use of ¹⁸F-FDG in oncology is based on the increased metabolic demand and subsequent glycolysis of tumor cells that use glucose as their source of energy. ¹⁸F-FDG PET was adopted rapidly for cancer imaging and was initially thought to potentially be a specific imaging modality that would allow not only monitoring of malignancy but also a tool to help differentiate it from other pathologies. However, this proved not to be the case, with ¹⁸F-FDG uptake also seen in both benign conditions and in the setting of inflammation (15-18).

Inflammatory basis of atherosclerosis

Inflammation plays a central role in atherosclerosis. Activated intimal endothelial cells express P-selectin and cell adhesion molecules, which in turn recruit circulating leukocytes, such as monocytes that express P-selectin glycoprotein ligand-1 (PSGL-1). These cells adhere to and roll on the endothelium (19). PSGL-1 expressing leukocytes are thought to be a pro-inflammatory subset containing large amounts of pro-inflammatory cytokines. Macrophages differentiate from monocytes at the sites of plaque, and are capable of releasing a large array of cytokines including IL-1, IL-6, IL-12, IL-15, IL-18, and TNF alpha (20), propagating further inflammation at these sites of early atherogenesis. This continues throughout the natural history of the plaque with attraction and accumulation of monocytes at the site of plaques continuing throughout and in a manner proportional to plaque size (21). Monocytes and macrophages are the most abundant leukocyte involved in atherogenesis but others, including neutrophils, T and B cells, also contribute to promoting inflammation both directly and by co-stimulatory interaction between themselves (20,22). Oxidized subendocardial LDL contributes to the formation of foam cells from macrophages, and to the inflammatory milieu of atherosclerosis. Its presence also induces the secretion of potent chemokines, such as CCL2, in endothelial and smooth muscle cells (23,24) and stimulates the production of reactive oxygen species and prostaglandins by macrophages, and other inflammatory cells (25,26).

Acute rupture of atherosclerotic plaque is a common cause of acute myocardial infarction, occurring in 60-80% of acute coronary syndrome (ACS) cases, and is found in the majority of cases of sudden cardiac death with intraluminal thrombi (27,28). Plaque vulnerability and rupture has a strong association with inflammation (29). Macrophages are found in highest concentration in vulnerable atherosclerotic plaques and contribute to the thinning of the fibrous plaque cap by release of matrix metalloproteinases, making them more likely to rupture (30-35). The critical involvement of macrophages in plaque development and disruption makes them an obvious imaging target for the early identification of vulnerable plaques before complications arise.

¹⁸F-FDG detection of vascular inflammation

Both macrophages and foam cells have been shown in vitro to take up ¹⁸F-FDG and at a comparable level to some cancer cell lines (34,36,37). Initially the uptake of ¹⁸F-FDG in atherosclerotic plaque was thought to be due to the respiratory burst of macrophages when phagocytizing the subendothelial oxidized LDL particles (38,39). However, Folco et al. showed that hypoxia is the main driver that modulates glucose uptake in macrophages, mediated by increased expression of hexokinase-2 rather than upregulation of the GLUT1 transporter (40). Hypoxia is a prominent feature of the atherosclerotic milieu and associates strongly with the burden of macrophage infiltration and inflammation (41). Animal (42-48) and human (33,49-51) studies have consistently demonstrated a correlation between ¹⁸F-FDG uptake and the presence and density of macrophages in atherosclerosis.

¹⁸F-FDG PET vascular imaging in humans was first described in 1999 in Takayasu’s arteritis, before being investigated in other vasculitides (52-54). However, case studies subsequently noted vascular uptake even in the absence of vasculitis (55). Retrospective analysis of ¹⁸F-FDG PET scans in patients undergoing oncological workup identified vessel wall uptake in as many as 50% (56). In a number of cases this uptake was nonuniform along the length of the vessel and in one case where uptake extended from the abdominal aorta to the femoral arteries, it was recognized that this corresponded to extensive atherosclerotic disease seen also on CT imaging. A study by Yun et al. then drew associations between ¹⁸F-FDG uptake in large arteries and the presence of cardiovascular risk factors and documented CAD (57).

These observational results paved the way for more dedicated assessment of ¹⁸F-FDG imaging for atherosclerosis. The potential of ¹⁸F-FDG PET was demonstrated early on in a rabbit model of atherosclerosis in which an intravascular positron sensitive probe was used (43). In 2002 Rudd and colleagues then reported a ¹⁸F-FDG PET study performed in eight patients with symptomatic carotid artery disease (49). All patients displayed ¹⁸F-FDG uptake in the symptomatic carotid stenosis, and the mean signal was 27% higher than that observed in the asymptomatic contralateral stenosis present in six of them. Histological assessment of the endarterectomized arteries also confirmed heavy macrophage infiltration at the ¹⁸F-FDG positive sites.

Patients with known CAD have a high prevalence of carotid artery uptake of ¹⁸F-FDG and known cardiovascular risk factors including elevated body mass index, age >65 years, hypertension, and smoking, have been shown to be both strong determinants and independent predictors of vascular ¹⁸F-FDG uptake (58). Retrospective analyses have suggested correlation of ¹⁸F-FDG vascular uptake with increased risk for cardiovascular events (59). Meanwhile prospective studies in patients with symptomatic carotid artery disease have shown that high levels of ¹⁸F-FDG uptake predict early stroke recurrence for both symptomatic and asymptomatic carotid lesions (60-63).

¹⁸F-FDG imaging of coronary atherosclerosis

Coronary artery uptake had been noted in studies looking retrospectively at ¹⁸F-FDG PET scans. In 2009 the first prospective study of coronary uptake was performed, comparing culprit lesion stent sites in 10 ACS and 15 stable CAD patients, as well as in recent to remote stent sites in the stable CAD group (64). Uptake in ACS plaques quantified by the target-to-background ratio (TBR) was higher than in stable CAD plaques (2.61 vs. 1.74, P=0.02), while no significant difference was found between recent and remote stent sites (1.74 vs. 1.43, P=0.49), suggesting that ¹⁸F-FDG signal correlated with acute plaque features, and not simply to post-angioplasty changes (64).

¹⁸F-FDG imaging has been shown to be highly reproducible over short interscan time periods (65) and also to be able to identify plaque changes in plaque regression studies with serial imaging in rabbits (66). It has been used as a serial imaging tool in multiple small intervention trials in humans using statins, and pioglitazone (67,68) Another of its applications has been to confirm that new pharmaceutical agents undergoing clinical evaluation do not cause adverse inflammatory effect on blood vessels, as implemented in the dal-PLAQUE study where dalcetrapib was shown not to be proinflammatory or proatherogenic after concerns had been raised about cholesteryl ester transfer protein inhibitors as a group after the failure of torcetrapib due to off target effects (69,70).

Together these longitudinal studies that have been conducted over relatively short intervals, 3 months, have shown the potential for ¹⁸F-FDG PET coronary imaging to provide surrogate assessment of the effectiveness of anti-atherogenic interventions. This in turn can help guide decision making for clinical research programs to determine which therapies are worth investigating in larger and longer phase III trials which measure hard clinical outcomes. Some retrospective analyses of oncological patients undergoing multiple ¹⁸F-FDG PET scans over longer interscan periods with no specific plaque stabilising intervention however have shown zero correlation between plaque uptake between scans, potentially questioning its ability to serially monitor atherosclerotic plaque in longer duration studies (71).

A major limitation of ¹⁸F-FDG based imaging of coronary atherosclerosis relates to the background noise created by ¹⁸F-FDG uptake in the adjacent myocardium. When combined with PET’s fundamental limitation of spatial resolution (4.2–6.3 millimeters) contributed to by detector size, positron emission range, and non-collinearity, the ability for PET to be able to differentiate between coronary artery and myocardium significantly effects its ability to identify true coronary signal (72,73). While efforts have been made to reduce background myocardial uptake, for example by providing a low carbohydrate-high fat diet prior to scan acquisition (74), the rate of uninterpretable scans in human patients remains prohibitively high. Thus far this has significantly limited the accuracy and reproducibility of coronary ¹⁸F-FDG PET in the rigorous setting of clinical trials, while also precluding its wider application in routine clinical practice.

Another issue that needs to be considered in assessing the PET/CT assessment of coronary atherosclerosis is the radiation exposure to the patient with this imaging modality. With modern day combination scanners a PET/CT cardiac scan exposes patients to a radiation dose in the order of 8-11mSv, with the majority of this accounted for by the internal exposure from the ¹⁸F (75).

Sodium ¹⁸F-fluoride (¹⁸F-NaF)

¹⁸F-Fluoride has been used as a radionuclide for bone scanning since the 1960’s (76). Fluoride is well known to be taken up by bone mineral with the main body store of fluoride thought to be the skeleton (77), with exchange of fluoride for one of the hydroxyl ions of hydroxyapatite forming fluorapatite the purported mechanism of skeletal ¹⁸F-NaF uptake (78). For a period of time ¹⁸F-NaF was superseded by technetium-based radiotracer compounds that had superior characteristics for imaging with gamma cameras. However, the advent of PET scanner technology and its well-established place in oncological practice has seen the re-emergence of ¹⁸F-NaF. In particular, it has been widely used for the assessment of multiple types of bone pathology and increasingly for bone metastases (79). Notably, its uptake on PET scanning has been shown to correlate with dynamic indexes of bone formation (80).

¹⁸F-NaF, vascular calcification and cardiovascular risk

Vascular calcification occurs as an active process in atherosclerosis and is seen with increasing frequency with advancing age and also with increasing atherosclerotic burden (81). Osteogenic differentiation of different cell populations present in the vessel wall, including medial smooth muscle cells and adventitial fibroblasts, pericytes, and mesenchymal stem cells, is stimulated in the inflammatory milieu of atherosclerosis. Macrophages within atherosclerotic plaque promote osteogenesis by the release of cytokines (e.g., IL-1, IL-6, IL-8, tumor necrosis factor alpha, insulin-like growth factor-1, and transforming growth factor-beta) (82-85). Once established the presence of microcalcification crystals contributes to a positive feedback loop by inducing a pro-inflammatory response in macrophages that further propagates the pro-calcific stimulus in the vessel wall (86). Formation and progression of vascular calcification is very similar to bone formation, with the generation of hydroxyapatite crystals in the vessel wall. Given the similarities to bone formation, it is intuitive that vascular calcification may be an amenable target for molecular imaging with ¹⁸F-NaF.

Several retrospective studies have observed ¹⁸F-NaF uptake within the vasculature of patients who have undergone full body ¹⁸F-NaF PET imaging to detect bony metastases in the setting of malignancy (Table 1). Derlin et al. first reported this phenomenon in 76% of 75 patients undergoing radionuclide imaging for bone metastases, with the most common arterial sites of ¹⁸F-NaF uptake being the femoral arteries and aorta (87). In 95% of these cases, arterial wall calcification was confirmed by CT assessment also (Figure 1).

Table 1. Clinical studies of atherosclerosis with ¹⁸F-NaF PET.

Study	Site	Design	Patient number	Dose	Uptake period	Prevalence	Comments	Results
Derlin et al. (87)	Major arteries	Retrospective	75	350±50 MBq	60 min	57/75 (76%)	First feasibility study	Correlation between 18F-NaF uptake and AC
Derlin et al. (88)	Carotid artery	Retrospective	269	350±50 MBq	60 min	94/269 (35%)	Comparison of 18F-NaF uptake to CV RFs	Male, age, HT, ↑ chol correlated with 18F-NaF uptake; ↑ prevalence of 18F-NaF uptake with ↑ RFs
Derlin et al. (89)	Major arteries	Retrospective	45	350±50 MBq	60 min	27/45 (60%)	Comparison of 18F-NaF and 18F-FDG	18F-NaF more likely to co-localize with AC; co-uptake only seen in 6.5% of lesions
Li et al. (90)	Major arteries + coronary arteries	Retrospective	61	370±74 MBq	40 min	59/61 (97%)	First study to assess coronary arteries	↑ Coronary 18F-NaF uptake with Hx of CV event (SUV 1.70 vs. 1.39, P=0.029)
Beheshti et al. (91)	Combined myocardium/coronary arteries + aorta	Retrospective	59	370-550 MBq	60 min	Not described	Trialled new technique assessing ROI including both myocardium & coronary arteries (GMCS)	↑ 18F-NaF uptake with age
Dweck et al. (92)	Coronary arteries	Prospective	119	125 MBq	60 min	34%	Comparison of 18F-NaF and 18F-FDG	18F-NaF but not 18F-FDG correlated with ↑ Framingham risk scores; no 18F-NaF uptake in 41% with CAC >1,000
Kurata et al. (93)	Major arteries	Retrospective	29	~185 MBq	60 min	8/29 (28%)	Uptake was not corrected for blood pooling	91% of 18F-NaF co-localized to AC but only 10% of all AC had 18F-NaF uptake
Janssen et al. (94)	Femoral arteries	Retrospective	409	350±50 MBq	60 mins	159/409 (39%)		18F-NaF correlated with Age, HT, ↑ Chol, DM, Hx of smoking, Hx of CV events; ↑ prevalence of 18F-NaF uptake with ↑ RFs
Blomberg et al. (95)	Coronary arteries + thoracic aorta	Prospective	38	2.2 MBq/kg	45, 90 & 180 mins	Not described	Assessment of delayed image acquisition points	Blood pool corrected uptake not affected over time
Joshi et al. (96)	Coronary arteries (carotid arteries)	Prospective	80 (+9 carotid patients)	125 MBq	60 mins	37/40 (93%) in ACS group; 32/40 (80%) in stable CAD group	Comparison between 18F-NaF vs. 18F-FDG and ACS vs. stable CAD	34% higher TBR in ACS plaques; 93% (NaF) vs. 33% (FDG) uptake in culprit vessels
Morbelli et al. (97)	Major arteries	Retrospective	80	370 MBq	60 mins	Not described	Comparison between 18F-NaF uptake and CT calcification as assessed by Framingham risk factors	18F-NaF uptake dependent on age, HT; smoking & DM; calcification dependent only on age
Fiz et al. (98)	Infra-renal abdominal aorta	Prospective	64	4.8–5.2 MBq/kg	60–75 mins	Not described	Comparison between 18F-NaF uptake to density of calcification	Inverse correlation between 18F-NaF uptake and density of calcification

AC, arterial calcification; ACS, acute coronary syndrome; CAC, coronary artery calcification score; CAD, coronary artery disease; CT, computed tomography; CV, cardiovascular; DM, diabetes mellitus; GMCS, global molecular calcification score; HT, hypertension; RFs, risk factors; ROI, region of interest.

Figure 1.

¹⁸F-sodium fluoride PET/CT images of the aortic arch. (A) CT image; (B) PET image; (C) fused PET/CT image. ¹⁸F-sodium fluoride uptake in this atherosclerotic lesion coincided with calcification on CT imaging (arrows). Reproduced from “Feasibility of ¹⁸F-sodium fluoride PET/CT for imaging of atherosclerotic plaque.” Derlin T, Richter U, Bannas P, et al. J Nucl Med 2010;51:862-5. © by the Society of Nuclear Medicine and Molecular Imaging, Inc. PET, positron emission tomography; CT, computed tomography.

Other retrospective studies of similar patient populations have described association between the incidence of ¹⁸F-NaF uptake in the peripheral vasculature and the number of cardiovascular disease risk factors and variations of Framingham cardiovascular risk scores (88,94,97). In one particular study ¹⁸F-FDG and ¹⁸F-NaF PET were compared retrospectively in a group of 45 patients who had undergone both types of assessment for oncological work up (89). ¹⁸F-NaF showed a markedly increased propensity for co-localization with vascular calcification than did ¹⁸F-FDG (77.1% vs. 14.5%), with co-localization of both radiotracers at the same site present in only 6.5% of cases.

¹⁸F-NaF and coronary atherosclerosis

Dweck et al. performed the first prospective trial focused specifically on assessment of ¹⁸F-NaF uptake in coronary arteries as a substudy of a trial that investigated PET imaging of aortic stenosis (99). 119 patients were studied, including a control group of 13 who had no history of CAD or CT evidence of coronary artery calcification. Each patient underwent PET imaging with both ¹⁸F-FDG and ¹⁸F-NaF as a radiotracer in separate scans. Uptake of ¹⁸F-NaF but not ¹⁸F-FDG was found to be higher in patients with CAD compared to the control group (1.64±0.49 vs. 1.23±0.24, P=0.003 for ¹⁸F-NaF; 1.18±0.31 vs. 1.23±0.20, P=0.498 for ¹⁸F-FDG). Although significant correlations were found between ¹⁸F-NaF uptake and both Framingham risk score and coronary artery calcification, 41% of patients with coronary artery calcium scores >1,000 had no significant radiotracer uptake. Of particular interest were the findings in one patient who had suffered a recent myocardial infarction. Despite the presence of coronary artery calcification of all three coronary arteries, ¹⁸F-NaF was seen only at the site of the culprit lesion (92).

This finding prompted the same investigators to perform a comparison of ¹⁸F-NaF and ¹⁸F-FDG PET imaging of coronary arteries in patients presenting with ACS and those undergoing elective percutaneous coronary intervention for stable angina (96). Measuring TBR for ¹⁸F-NaF, they showed a 34% higher activity in culprit versus non-culprit plaques (maximum TBR 1.66 vs. 1.24, P<0.0001), with 93% of the culprit plaques showing ¹⁸F-NaF uptake. By comparison assessment of ¹⁸F-FDG signal in the coronary arteries was affected by background myocardial noise in half the patients, despite preparation with a low carbohydrate-high fat diet, such that the two could not be distinguished. In those patients in whom this was not the case (n=18), ¹⁸F-FDG uptake could only be detected in a third of culprit vessels, and there was no difference in TBR compared to non-culprit lesion sites (1.71 vs. 1.58, P=0.34). Although the cohort of patients with stable angina had more extensive CAD than the ACS group, with more arterial calcification, ¹⁸F-NaF uptake was seen in less than half of this group and did not correlate with the sites of stent deployment but rather more often coincided with non-obstructive lesions of <70% stenotic severity. Further analysis with VH-IVUS in the patients revealed that coronary plaques with ¹⁸F-NaF uptake were more likely to contain vulnerable plaque characteristics, such as positive remodeling, microcalcification, and necrotic core (100,101).

Taken together, these results indicate that the expectation that ¹⁸F-NaF uptake correlates with gross arterial calcification does not hold true. Another recent study that set out to specifically address this hypothesis in fact showed that ¹⁸F-NaF coronary signal correlated inversely with coronary calcium density, as classified in tertiles of light (130 to 210 HU), medium (211 to 510 HU), or heavy (>510 HU) calcification on CT imaging (98). Rather, the above data start to build a compelling case that ¹⁸F-NaF signal in the coronary vasculature more frequently associates with plaques that are actively inflamed and responsible for ACS events. Importantly, it does so to a far greater extent than ¹⁸F-FDG, without its detection being compromised by the same limitations caused by background myocardial noise.

Basis for ¹⁸F-NaF uptake in unstable plaque: microcalcification

Given what is known about uptake of ¹⁸F-NaF in bone, active microcalcification in atherosclerotic plaque appears to be the most likely substrate for ¹⁸F-NaF uptake in the vasculature. However, owing to a paucity of cellular and animal model studies, the exact molecular mechanisms for this remain unclear.

The most insightful information regarding the mechanism of ¹⁸F-NaF vascular uptake comes from an elegant study, recently published by Irkle et al. who performed comprehensive examinations of human carotid artery lesions in vivo and ex vivo (102). Pharmacokinetic and pharmacodynamic analysis confirmed favourable properties of ¹⁸F-NaF for arterial plaque assessment, including its high affinity for calcification, no significant dissociation from its target within standard time to scans, and low plasma activity at the time of scanning meaning minimal background noise. With the use of a microprobe and scanning electron microscope, fluoride and calcification were identified in carotid tissue. The presence of fluoride was found to be highly specific to calcification and more commonly co-localized with microcalcifications than macrocalcifications.

CT scanning (both micro and clinical) was able to reliably identify macrocalcification but not areas of microcalcification. In contrast, ¹⁸F-NaF PET showed uptake in areas that were not identified on CT and these corresponded to sites of microcalcification detected by Alizarin Red staining of contiguous histological sections (Figure 2). The same study also assessed the possibility that ¹⁸F-NaF uptake could be due to other inflammatory processes at the site of atherosclerosis that precede and contribute to active calcification. However, unlike the strong correlation seen with Alizarin Red staining for mineralisation, ¹⁸F-NaF signal correlated poorly with immunohistochemical detection of macrophage infiltration, or the presence of endothelial or smooth muscle cells (Figure 3).

Figure 2.

Sections showing macrocalcification (solid arrow) and microcalcification (hollow arrow) in: (A) histology section with Alizarin Red staining; (B) micro CT; (C) autoradiography; (D) ¹⁸F-NaF microPET. Adapted from original publication in Nature Communications “Identifying active vascular microcalcification by 18F-sodium fluoride positron emission tomography.” Irkle A, Vesey AT, Lewis DY, et al. Nat Commun 2015;6:7495. Licensed under CC BY 4.0. PET, positron emission tomography; CT, computed tomography.

Figure 3.

¹⁸F-NaF uptake correlates with calcification but none of the histological inflammatory markers. (A) Representative images of ¹⁸F-NaF autoradiography signal overlap with IHC-stained sequential sections. Green: ¹⁸F-NaF signal; red: histology signal; yellow: overlap. Scale bar, 1 mm; (B) high correlation is observed between ¹⁸F-NaF and alizarin red calcification staining, while low correlation is seen between ¹⁸F-NaF autoradiography and inflammatory marker IHC signals. Reproduced from “Identifying active vascular microcalcification by ¹⁸F-sodium fluoride positron emission tomography.” Irkle A, Vesey AT, Lewis DY, et al. Nat Commun 2015;6:7495. Licensed under CC BY 4.0.

Vascular macrocalcification on CT scanning is well established as a predictor of cardiovascular events but does not specifically identify vulnerable plaques (103). However, recent data from CT and IVUS-based studies have indicated that macrocalcification may in fact be protective against acute coronary events (104,105). Conversely, microcalcification is increasingly well understood to be a specific feature of vulnerable plaque, making its detection highly desirable in atherosclerosis imaging (106). As evident from the aforementioned studies, ¹⁸F-NaF PET is more sensitive and accurate than either ¹⁸F-FDG scanning or CTCA in the identification of microcalcification. Therefore it may have utility for the early detection of vulnerable plaque in high risk patients.

Future directions

Further studies are needed to more comprehensively delineate the biological and clinical significance of ¹⁸F-NaF uptake in coronary atheroma and confirm its accuracy and reproducibility for identifying vulnerable lesions at high risk of rupture and thrombotic complication. To this regard Newby and collaborators are due to commence recruitment in the Prediction of Recurrent Events With ¹⁸F-Fluoride (PREFFIR) study in August 2015 (www.ClinicalTrials.gov NCT02278211). This will aim to recruit 700 ACS patients and undertake both ¹⁸F-NaF PET and CTCA for an observational study with cardiac death or recurrent MI as the primary outcome measure assessed for 4 years from study commencement.

There also remains the prospect with ¹⁸F-NaF PET imaging for serial, noninvasive monitoring of atherosclerotic lesions over time to delineate the natural history of plaques and assess their responsiveness to different anti-atherosclerotic treatments.

Conclusions

The use of molecular imaging for assessment of coronary atherosclerosis with ¹⁸F PET appears a promising tool for the non-invasive identification of vulnerable plaque. This, coupled with ¹⁸F-NaF PET’s newly described ability to identify microcalcification in plaque, indicates its potential to provide biological characterization of coronary atherosclerosis that is complementary to other non-invasive imaging modalities, such as CTCA. The findings of recent ¹⁸F-NaF studies suggest its superiority over ¹⁸F-FDG and indicate that it is the more likely radioisotope to be incorporated into future trials. While further investigations are required to more clearly define the utility of ¹⁸F PET in routine clinical practice, it potentially will play a major role in future imaging of coronary atherosclerosis.