Portal de Periódicos da CAPES

Vascular imaging with positron emission tomography

2011; Wiley; Volume: 270; Issue: 2 Linguagem: Inglês

10.1111/j.1365-2796.2011.02392.x

1365-2796

Francis R. Joshi, David Rosenbaum, Sara Bordes, James H.F. Rudd,

Cardiovascular Disease and Adiposity

Joshi F, Rosenbaum D, Bordes S, Rudd JHF (University of Cambridge, Cambridge, UK; Groupe Hospitalier Pitié-Salpêtrière, Assistance Publique-Hôpitaux de Paris, Paris, France; and Instituto Cardiovascular, Madrid, Spain). Vascular imaging with positron emission tomography (Review). J Intern Med 2011; 270: 99–109. Atherosclerosis is an inflammatory disease that causes most myocardial infarctions, strokes and acute coronary syndromes. Despite the identification of multiple risk factors and widespread use of drug therapies, it still remains a global health concern with associated costs. Although angiography is established as the gold standard means of detecting coronary artery stenosis, it does not image the vessel wall itself, reporting only on its consequences such as luminal narrowing and obstruction. MRI and computed tomography provide more information about the plaque structure, but recently positron emission tomography (PET) imaging using [18F]-fluorodeoxyglucose (FDG) has been advocated as a means of measuring arterial inflammation. This results from the ability of FDG-PET to highlight areas of high glucose metabolism, a feature of macrophages within atherosclerosis, particularly in high-risk plaques. It is suggested that the degree of FDG accumulation in the vessel wall reflects underlying inflammation levels and that tracking any changes in FDG uptake over time or with drug therapy might be a way of getting an early efficacy readout for novel anti-atherosclerotic drugs. Early reports also demonstrate that FDG uptake is correlated with the number of cardiovascular risk factors and possibly even the risk of future cardiovascular events. This review will outline the evidence base, shortcomings and emerging applications for FDG-PET in vascular imaging. Alternative PET tracers and other candidate imaging modalities for measuring vascular inflammation will also be discussed. Atherosclerotic cardiovascular disease, including myocardial infarction and stroke, remains the most common cause of death and disability in the developed world and is associated with spiralling healthcare costs. The disease is usually present in sub-clinical form from adolescence onwards. Atherosclerosis is characterized by inflammation [1], a response to the deposition of low-density lipoproteins in the vascular wall. Endothelial cells express adhesion molecules such as vascular cell adhesion molecule (VCAM)-1 and selectins (of both E- and P-type). As a result, monocytes and T-lymphocytes are recruited into the subendothelial space. Macrophages, derived from monocytes, ingest oxidized lipids to form foam cells. The release of cytokines such as interleukin 1, tumour necrosis factor alpha and monocyte chemotactic protein 1 promotes further inflammatory cell recruitment. A lipid core is formed from foam cells and apoptosis macrophage debris. Over time, and in the continued presence of vascular risk factors, a mature plaque develops with a fibrous cap isolating the core from the vessel lumen. Macrophages secrete enzymes such as matrix metalloproteinases capable of directly degrading the fibrous cap, and this can be sufficient to cause the plaque to rupture, exposing the lipid core and tissue factor to the circulating blood stream. Intraluminal thrombosis may result, causing arterial obstruction and clinical syndromes (stroke, acute coronary syndrome or sudden death). The inflammatory process is driven by multiple genetic and environmental factors [2] and can be reduced by intervention with drugs such as statins [3]. To detect atherosclerosis, one can visualize the artery lumen by either invasive or noninvasive angiography. These provide information about site and severity of plaques causing stenosis but do not provide information about plaque composition [4]. These tests are often carried out after the development of symptoms, although deaths attributed to coronary disease often occur in those without prior warning [2]. Culprit plaques may not be obstructive to blood flow [5] because of positive (outward) remodelling, and the above tests are only modest predictors of future vascular events [6]. Many patients presenting with acute events will have multiple ruptured atherosclerotic plaques [7, 8]. Magnetic resonance imaging (MRI) and intravascular ultrasound may give more detail of the vessel wall and plaque morphology (fibrous cap thickness and size of lipid core) but are at present of limited utility in predicting events. There is some emerging evidence for the use of computed tomography (CT) in prediction of future culprit lesions in the coronary vasculature [9]. Broadly, plaques at high risk of rupture have similar features both histologically [4] and as assessed by contemporary imaging techniques. These include a thin fibrous cap, a necrotic core, heavy macrophage infiltration, positive remodelling of the vessel wall and spotty calcification. Noninvasive assessment of atherosclerotic plaques is needed to help predict events in the at-risk but asymptomatic patient group. Such imaging may also provide information about the underlying vascular biology of the disease and track the effect of drug therapy. Molecular imaging with positron emission tomography (PET) is sensitive enough to detect the degree of inflammation within atherosclerotic plaques. The tracer [18F]-fluorodeoxyglucose (FDG) is an analogue of glucose that is taken up by macrophages, key inflammatory cells of the plaque. This review will outline the evidence base, shortcomings and other emerging applications for FDG-PET in vascular imaging. Alternative imaging techniques and PET tracers will also be discussed. Positron emission tomography imaging with FDG remains the gold standard for the detection of cancer metastasis and for monitoring the response to therapy in patients with cancer [10]. It is also used in the assessment of myocardial viability as a guide to appropriate revascularization [11]. When applied to atherosclerosis, to measure inflammation, FDG-PET is attractive because of its very high sensitivity for molecular targets, in comparison with those required in alternate imaging techniques such as MRI and CT. The technique involves intravenous injection of FDG, which is allowed to circulate for about 90 min prior to imaging using a PET scanner. The circulation time must be long enough to allow for FDG uptake in areas of interest and for background FDG levels (predominantly blood in vascular imaging) to be reduced. PET has limited spatial resolution (3–4 mm) and therefore requires co-registration with CT or MR to localize FDG uptake to the underlying anatomy. Additionally, given the small size of atheromatous plaques, imaging with PET is subject to partial volume errors [12, 13]. Finally, PET imaging does require the use of ionizing radiation, meaning that widespread screening of asymptomatic subjects for vascular inflammation is not feasible. [18F]-fluorodeoxyglucose is a glucose analogue with a half-life of 110 min and is taken up by cells that metabolize glucose. Glucose transport is via two classes of transporter: facilitated glucose transporters (GLUT) and sodium glucose co-transporters [14]. FDG directly competes with glucose for uptake into metabolically active cells, using the same GLUT transporter proteins. It is then phosphorylated to FDG-6-phosphate by the hexokinase enzyme but cannot be metabolized further and thus accumulates within the cell. In theory, it can be dephosphorylated by glucose-6-phosphatase, but this enzyme is not greatly expressed outside skeletal muscle and the liver, and so this effect is considered negligible in atherosclerosis. The rates at which glucose and FDG are phosphorylated are proportional to each other and reflect glucose usage of the cell in question [15]. Macrophages play an active role in the atherosclerotic plaque. Glucose uptake of macrophages is significantly higher than that of neighbouring cell types [16–18]. They can also metabolize free fatty acids but in the anaerobic environment of the plaque [19], preferentially use glucose, as this does not require oxygen to produce ATP [20]. Compared with quiescent cells, activated macrophages show an increased expression of GLUT type 1 and type 3 receptors, along with hexokinase [21–23]. FDG uptake is therefore a function of macrophage density and activation, GLUT transporter expression, hexokinase activity and dephosphorylation. In vitro, culture duration and glucose concentrations are also relevant [22] although moderate hyperglycaemia (up to 250 mg mL−1) does not affect uptake into macrophages [24, 25]. In humans, reproducibility in those with elevated blood glucose levels at the time of FDG imaging is likely to be reduced [26]. Finally, in vitro FDG can be taken up into adjacent cell types including endothelial cells and lymphocytes [27]. FDG uptake in preclinical models of atherosclerosis was first documented in 1996[28]. In a rabbit model [29], uptake has been shown to strongly correlate (r = 0.81, P < 0.0001) with plaque macrophage content, and this has been confirmed by others [30–32]. After balloon aortic injury in New Zealand white rabbits, Tawakol et al. [33] demonstrated again a very strong correlation (r = 0.93, P < 0.002) between FDG uptake into atherosclerotic regions and plaque macrophage content. FDG uptake was 19 times greater in the region of greatest plaque formation as compared to controls. Uptake was independent of plaque size, smooth muscle content or aortic wall thickness. In rabbit models of progressive atherosclerosis, after balloon aortic injury and high-fat, atherogenic diets, FDG uptake has been shown to be higher in diseased regions as compared to healthy arterial wall [34, 35] and can be reduced with reversion to a normal diet [34]. Aziz et al. [36] demonstrated in a similar model FDG uptake correlated with the duration of atherogenic dietary intake. After triggering plaque rupture using Russell's viper venom and histamine, only those plaques with the highest uptake (and macrophage density) progressed to develop superimposed thrombus. This suggested a role for FDG-PET in predicting plaques most likely to rupture. In mice models, using both atherosclerotic and wild-type mice, it has been shown that aortic FDG uptake increased with atherogenic diets [37]. In autoradiography of aorta from atherosclerotic LDLR/ApoB48 mice, Laitinen et al. [38] demonstrated higher uptake in plaques as compared to healthy vascular wall. Only Laurberg et al. [39] failed to demonstrate an increase in aortic FDG uptake in ApoE-deficient mice after atherogenic diets and carotid ligation to accelerate atherosclerosis, perhaps because of an early scanning point [40]. After anecdotal reports, aortic FDG uptake in patients undergoing FDG-PET imaging for oncology staging was described in 2001[41], and later noted as increasing with age [42], male sex and the presence of vascular risk factors [43–45]. In the first prospective clinical study, Rudd et al. [46] imaged eight patients shortly after transient ischaemic attack (TIA) with FDG-PET and reported an FDG accumulation rate 27% higher in the symptomatic carotid compared to the contralateral asymptomatic side. In autoradiography reported in the same paper, tritiated deoxyglucose accumulated in macrophage-rich areas, confirming uptake was in macrophages and reflective of underlying inflammation. This notion is supported by in vitro studies as described earlier [22, 47] and confirmed by Tawakol et al. [48] in patients with severe carotid stenosis undergoing endarterectomy. Mean FDG uptake in each patient was seen to strongly correlate (r = 0.96, P < 0.001) with mean CD68+ macrophage staining. No correlation was found between FDG uptake and plaque thickness, area or smooth muscle staining. It has since been shown that FDG can quantify inflammation in the aorta, iliac and peripheral arteries, with excellent short-term and interobserver reproducibility [49–52]. This has proved important in the search for quantitative measures of inflammation that can be used to test the impact of new therapies. Uptake may be widespread [44, 54], although especially affecting the carotid arteries and proximal aorta (1, 2). This is consistent with the hypothesis that atherosclerosis is a generalized inflammatory condition. FDG uptake can distinguish between culprit carotid and vertebral lesions in patients presenting with posterior circulation stroke [52]. Carotid artery FDG uptake has been shown to correlate with microembolic signals on transcranial Doppler ultrasound after TIA [55]. Transaxial FDG-PET/CT image of a subject with atherosclerosis. Arrows indicate FDG uptake in the ascending and descending aorta. Coronal computed tomography (CT), PET and fused FDG-PET/CT image of a subject with atherosclerosis. Arrows indicate FDG uptake in the aortic wall corresponding to plaque inflammation. The above studies provide evidence that FDG uptake does reflect macrophage density and activation in atherosclerosis and that recently symptomatic plaques are more inflamed than stable lesions. Given its reproducibility, work has since shown that it is possible to monitor the effect of anti-atheroma interventions on levels of FDG uptake and by inference inflammation within atherosclerotic plaque. In a rabbit model of atherosclerosis [56], FDG-PET revealed a reduction in plaque inflammation after treatment with probucol, a lipid-lowering antioxidant. Consistent with the work of Crisby et al. [3]. Tahara et al. [57] showed a reduction in FDG uptake after treatment with simvastatin compared with placebo; the reduction correlated with the degree of increase in HDL cholesterol. In another clinical study, the introduction of lifestyle and dietary modifications to asymptomatic subjects was shown to lead over 17 months to a 65% reduction in the number of vascular regions that accumulated FDG [58]. Another class of drug investigated using vascular FDG-PET was the B vitamins. In agreement with previous randomized controlled trials [59], it was shown that dietary supplementation with B vitamins did not reduce vascular inflammation as assessed by FDG-PET in either the carotid, aorta or femoral arteries [60]. More recently still, in a preclinical rabbit model of atherosclerosis, FDG-PET was used to test the performance of a novel nanomedicinal liposomal formulation of corticosteroid to reduce plaque inflammation [32]. This proved feasible as a method of tracking the lowering of aortic inflammation as early as 2 days after administration of the modified steroid preparation. This method of drug delivery awaits confirmatory trials in patients with atherosclerosis. Vascular FDG uptake seems to be greater in those subjects with pre-existing metabolic syndrome [44], more than 1 cardiovascular risk factor [54], a history of coronary artery disease or diabetes [54, 61]. It rarely co-localizes with vascular calcification [54, 62], an independent risk factor for cardiovascular events, perhaps because calcification tends to occur late in the course of the disease, with inflammation earlier. Arterial uptake in one vascular region correlates strongly with its anatomical pair, and more strongly with neighbouring regions than more distally. Plaques with high-risk features on MRI (large lipid core [49] or intraplaque haemorrhage [63]) or ultrasound [64,65] (echolucency) have higher FDG than more stable phenotypes. Aortic uptake has been shown to correlate with circulating biomarkers matrix metalloproteases 1, 3 and 9, and an inverse correlation has been reported with the atheroprotective biomarkers adiponectin and plasminogen activator inhibitor 1[54, 66]. There are mixed results when comparisons are made between the circulating biomarker of inflammation C-reactive protein (CRP) and FDG uptake. The largest study to date, that examined over 100 subjects, found that those in the highest quartile of CRP had the greatest degree of vascular FDG uptake, regardless of whether they also had elevated LDL cholesterol or not [67]. Finally, in a study of carotid disease in patients undergoing endarterectomy, gene expression markers of vulnerability GLUT-1, CD68, cathepsin K and HK2 were found to predict the degree of arterial FDG accumulation [68]. Although FDG-PET seems capable of assessing the anti-inflammatory effects of therapy, as yet there are no prospective data to support its value in predicting the risk of cardiovascular events. There is circumstantial evidence from retrospective series of patients undergoing PET scanning for oncology staging however; the presence of high levels of vascular FDG uptake predicted cardiovascular events both in the 6 months before and subsequent to imaging [69]. A role for risk prediction of myocardial infarction, stroke or revascularization was also suggested in a larger retrospective analysis (Fig. 3)[70]. Subsequent stroke correlated with foci of FDG uptake in the carotid arteries or aortic arch in a recent case–control study [71]. A small pilot study suggested that higher baseline FDG uptake (standard uptake value >2.7) after symptomatic carotid artery stenosis predicted worse outcomes (recurrent stroke, death or stent re-stenosis) at 6 months [72]. Prospective evaluation of FDG-PET in event-driven studies remains an important step in the evaluation of this imaging technology as a clinically useful tool. The BioImage Study [73], part of the High Risk Plaque Initiative (http://www.hrpinitiative.com), aims to identify imaging markers of risk for cardiovascular events over 3 years of follow-up. The patients recruited to this study will be imaged with a combination of MRI to identify carotid and aortic plaques, CT for arterial calcification and stenosis and FDG-PET-CT to measure carotid and aortic inflammation. The predictive ability of advanced imaging tests such as these will be compared with Framingham risk scoring alone. Increased arterial FDG uptake and calcification is associated with subsequent cardiovascular events. Kaplan–Meier survival curves for subjects with and without FDG uptake (TBR > 1.7) and calcification (plaque score > 15). Adapted from Rominger et al. [70] with permission. The mortality and morbidity associated with acute coronary syndromes has driven attempts to develop imaging of unstable plaque in this vascular bed. This noninvasive approach does present special challenges. Firstly, arteries are small, less than the typical resolution of PET (around 4 mm). Imaging is hindered further by respiratory and cardiac motion during the time taken to acquire a PET data set (around 15 min). There may be spatial shifts between PET and CT image acquisition, even in combined PET/CT scanners. Finally, FDG is taken up avidly by myocardial cells, which preferentially metabolize glucose over free fatty acids. Coronary artery FDG uptake was described in 2005[62], again in a series of patients undergoing tumour staging. Of the 78 patients, 34 were excluded from analysis because of intense myocardial tracer uptake. Hepatic uptake often obscured evaluation of the diaphragmatic aspect of the heart. Coronary FDG accumulation correlated with risk factors and previous coronary disease and was often proximal and multifocal. Recent attempts have tried to switch myocardial metabolism to free fatty acids by using high-fat diets before imaging, with varying success in demonstrating relationships between coronary disease and tracer uptake [74–78]. Rogers et al. [79] reported the use of dietary manipulation in patients treated by percutaneous coronary intervention for stable angina and acute coronary syndromes, defining culprit lesions anatomically as the stented segment. They demonstrated higher FDG uptake in the stented segments in acute coronary syndrome, as well as in the relatively fixed left main coronary artery and the ascending aorta, consistent with a multifocal inflammatory response in unstable atherosclerosis. Published in 2010, Saam [80] measured FDG uptake in the coronary arteries of close to 300 subjects with cancer. Evaluation of coronary artery FDG uptake was only possible in half of those imaged; the remainder had myocardial FDG uptake that swamped any signal from the coronary plaque. In those with coronary FDG uptake, there was a correlation between the number of cardiovascular risk factors, the volume of pericardial fat and the quantity of calcified plaque present in the coronary arteries. In summary, although progress with coronary artery FDG imaging has been made, using pre-imaging fatty diets and the possibility of respiratory and ECG gating of scans, it is likely that novel PET tracers, not accumulated by the myocardium itself, will be required. Aortic plaque inflammation has been well described in clinical FDG-PET imaging studies [54, 81]. Recent studies have extended this work into the area of abdominal aortic aneurysm, to determine whether inflammation might be a predictor of rapid aneurysm expansion or heightened rupture risk. Current guidelines for stable patients recommend surgical or endovascular intervention based largely on aortic dimensions [82, 83], although it is well known that rupture can occur in small aneurysms as well. It has been shown that abdominal aortic FDG uptake is associated with higher wall stress [84] and correlates with instability, symptoms and with macrophage infiltration [85]. Whether FDG uptake can identify patients who will benefit from earlier intervention than that dictated by size remains the subject of several ongoing studies. Thoracic aortic dissection is rare (3 per 100 000 person-years) but carries high mortality of between 10% and 30% at 30 days [86]. For those with type B lesions, initial management is usually medical, with efforts targeted at profound lowering of systolic blood pressure to reduce wall stress. However, a proportion of subjects with expanding aortas after dissection seem to benefit from surgical intervention. Inflammation plays a role in the pathophysiology of the disease [87, 88]. In patients presenting with acute aortic syndrome who underwent acute FDG-PET imaging, there was a trend to less progression of disease in those with lower degrees of FDG uptake [89]. A study by Kato et al. [90]., in subjects with both acute and chronic dissection, suggested worse outcomes in those with the highest aortic FDG uptake at the point of maximal dilatation. Whilst these are small studies conducted on potentially unstable populations, they suggest a possible role in risk stratification and planning of intervention in abdominal aortic aneurysm and type B aortic dissection. Further validation is required in larger event-driven studies [91]. Case reports describe the use of FDG in the clinical arena for the diagnosis and treatment of large vessel vasculitides [92]. Based initially on case series in giant cell arteritis and polymyalgia rheumatica [93], a recent review has quoted sensitivities of 77–92% and specificities of 80–100% in diagnosis and a role in the assessment of response to treatment [94]. There is potential for false-positive results in those with risk factors for atherosclerotic vascular disease. Patients with COPD are known to be at excess risk of cardiovascular disease, even after controlling for their smoking history [95]. It has been suggested that vascular inflammation might drive this increased susceptibility, particularly to myocardial infarction. A recent study used imaging with FDG-PET/CT to compare aortic inflammation in patients with COPD, ex-smoker controls and subjects with metabolic syndrome. FDG uptake levels as a surrogate for inflammation were significantly higher in those with COPD in the aortic arch and abdominal aorta than those of the ex-smokers without COPD, but lesser than those with the metabolic syndrome, suggesting that inflammation may indeed explain this increased risk [96]. Whilst the majority of work in this field has used FDG as a PET tracer, its limitations are well recognized. Whilst it is widely available commercially, it is accumulated by all cells that metabolize glucose and is not specific for cells responsible for inflammation. This has led to the search for more specific tracers. The translocator protein/peripheral benzodiazepine receptor (TSPO) is expressed on macrophages at 20 times greater density than vascular smooth muscle cells. In vitro, the TSPO ligands PK11195 and DAA1106 will bind to macrophage-rich regions in human carotid plaque [97]. However, it has become clear that for in vivo PET imaging of their distribution, there appears to be a wide spectrum of binding affinity of the receptors for these ligands in patients [98, 99]. This may limit their clinical application, although there have been recent promising studies in vasculitis [100]. Other potential PET tracers for atherosclerosis include [11C] or [18F]-choline, metabolized and incorporated into cell membranes in tumour cells and macrophages. Labelled choline was not taken up into normal vascular wall, or purely calcified lesions, in a retrospective analysis of 93 male patients undergoing cancer imaging [101]. In ex vivo imaging of aortae of ApoE-deficient mice, [18F]-fluorocholine uptake better correlated with fat staining and macrophage-positive areas than FDG [102]. Computed tomography imaging of molecular targets is hampered by a significantly lower sensitivity than PET, but provides much better spatial and temporal resolution. The nanoparticulate, iodinated contrast agent N1177 accumulates in macrophages (by phagocytosis) 2 h after intravenous injection in a rabbit model of atherosclerosis [103]. A further study demonstrated aortic uptake that correlated with FDG-PET uptake (r = 0.61, P < 0.001) and macrophage infiltration (r = 0.63, P < 0.001) on immunohistology [104]. The relative molecular insensitivity of CT contrast raises questions regarding tolerability of these agents at the doses required in the clinical area. MR imaging is attractive because of the absence of ionizing radiation in an area in which patients may have to undergo serial scanning. Ultrasmall superparamagnetic particles of iron oxide (USPIO) have shown promise as markers of macrophage activity within plaques. Experimental studies have confirmed accumulation in macrophages after intravenous injection of these agents [105, 106]. Inducing signal loss on T2-weighted images, they have since been used to image macrophage burden in human carotid arteries [107, 108]. Limitations include differentiating induced signal loss ('negative' contrast) from artefact and the 36-h delay between contrast injection and subsequent MR imaging. Nevertheless, they have been successfully used to demonstrate reduction in inflammation after statin therapy [109]. Two recent publications have also illustrated aortic aneurysm inflammation using the USPIO MRI approach [110, 111]. Vascular imaging of atherosclerosis with FDG-PET/CT has shown promise for quantifying plaque inflammation noninvasively and can be used to track changes in inflammation occurring with therapy, something that is difficult using other noninvasive modalities. However, it remains relatively expensive, exposes subjects to radiation and is currently not well suited for imaging the coronary arteries. The results of prospective, outcome-driven trials such as the BioImage study are needed to define its role and cost-effectiveness in risk-prediction relative to other imaging modalities and circulating biomarkers. PET tracers more specific for macrophages within plaques are nearing the clinical assessment stage [112], and research is also well advanced in using PET imaging to detect other features of high-risk atheroma including hypoxia [113], neovascularization [114, 115] and dynamic calcification [116]. Finally, hardware developments, such as the introduction of combined PET and MRI scanners, should help to harness the most attractive aspects of both modalities, high sensitivity and spatial resolution, which are at a premium when searching for insights into the biology of structures as small as the atherosclerotic plaque (Table 1) [117]. No conflicts of interest to declare. Work described in this review was supported by the Cambridge NIHR Biomedical Research Centre. Dr. Rudd is supported by the British Heart Foundation, the Evelyn Trust and HEFCE.