The majority of myocardial infarctions (MI) and sudden cardiac deaths result from the rupture of plaques that in most cases did not cause significant flow limitations before the acute event (1- 2). Although stenosis severity has proven a poor predictor of plaque rupture (3), several other plaque characteristics, notably plaque inflammation, have been strongly associated with plaque rupture and thrombosis (4- 7). Imaging with fluorodeoxyglucose (FDG) positron emission tomography (PET) provides a reproducible measure of glycolysis (8), which has proven invaluable for the identification of small, metabolically active tissues such as tumors and inflamed lesions (9). Furthermore, FDG accumulation has been shown to provide an index of atherosclerotic inflammation, as demonstrated in animal models and humans (10- 12). Several studies have shown that FDG activity within plaques correlates with inflammation in atherosclerosis (13- 15). Furthermore, the vascular PET signal is reproducible (16- 17) and is modifiable by interventions that have an anti-inflammatory effect (18).

Prior investigators have reported FDG accumulation in association with coronary vessels (19- 22). However, the lack of a standard for coronary inflammation within those studies provided little evidence that the FDG uptake was related to plaque activity. Several factors impart formidable obstacles to coronary FDG PET imaging, including uptake of FDG by adjacent myocardium, the limited spatial resolution of PET, and the substantial motion of the coronary arteries. In the present study, we attempted to determine whether FDG uptake can be measured in the ascending aorta and in the relatively immobile left main coronary artery (LM) despite the known obstacles and whether the signal is increased in patients presenting with acute coronary syndrome (ACS). Moreover, we sought to test the hypothesis that FDG uptake is greatest within lesions that are expected to be inflamed: those plaques that are deemed to have recently caused an ACS. We took advantage of concurrent computed tomographic (CT) angiography for delineation of coronary activity and to identify the plaques associated with stents that were deployed to treat ACS. We sought to test the hypothesis that FDG uptake within plaques locations deemed responsible for ACS is greater than in remotely stented plaques, which are presumably less inflamed. Additionally, to exclude FDG uptake secondary to lesion manipulation, we compared FDG uptake in plaques recently stented for ACS with plaques recently stented for stable coronary syndromes.

The study was performed in 25 patients (average age 58 ± 10 years, 72% men), 10 of whom presented with ACS requiring stent placement (ACS, n = 10). Of the remaining 15 patients with stable angina (n = 15), 5 presented with stable angina requiring stent placement (n = 5) and 10 presented with atypical or typical chest pain for which a coronary angiographic evaluation was required in the clinical judgment of the referring physicians. All subject cases were adjudicated for study group classification after eligibility determination and receiving informed consent. Subjects were classified as having experienced ACS (i.e., ST-segment elevation MI [n = 1], non–ST-segment elevation MI [n = 4], or unstable angina [n = 5]) or as having experienced stable angina (n = 15) based on the nature and chronicity of their chest pain, medical history, and available diagnostics (such as electrocardiograms [ECGs] and biomarkers), according to the American College of Cardiology/American Heart Association guideline definitions (23), and as supported by findings on coronary arteriography.

Patients were not eligible if they had diabetes mellitus, an allergy to iodinated contrast, renal dysfunction, hemodynamic instability, substantial arrhythmias, or decompensated heart failure. The study protocol was approved by the human research committee at the Massachusetts General Hospital, and informed consent was obtained from each subject.

After catheterization, subjects underwent cardiac CT angiography and cardiac FDG PET imaging. To minimize myocardial glucose uptake in order to better visualize coronary FDG uptake, we took advantage of the myocardium's preference for free fatty acid over glucose as an energy source. On the day before imaging, subjects were asked to adhere to a very-low-carbohydrate, high-fat diet and then to fast overnight (24), followed by the administration of a high-fat, low-carbohydrate beverage 45 min before FDG injection (approximately 150 ml Atkins Advantage Shake, Atkins Nutritionals, Ronkonkoma, New York).

The PET imaging was performed 3 h after administration of 13 mCi of ¹⁸F-FDG. The choice of time interval was based on previous work (16,25). Despite the widespread availability of PET CT systems, we used a dedicated PET system (Siemens ECAT Exact HR+, Knoxville, Tennessee) to avoid the potential misregistration that can occur when CT attenuation maps are used to correct the PET volume data. The system provides 63 planes, a 15.5-cm field of view, and a maximum 4.2-mm intrinsic resolution at the center of the field of view. Data were acquired in 2-dimensional mode over 25 min, with subjects in the supine position. Image data were attenuation-corrected using PET transmission data; reconstructed using a conventional, filtered back-projection algorithm; and corrected for nonuniformity of detector responses, dead time, random coincidences, and scattered radiation.

The CT imaging was performed on the same day as PET acquisition (Siemens Sensation 64, Forchheim, Germany). Subjects were pre-medicated with intravenous metoprolol and sublingual nitroglycerin unless medically contraindicated. Imaging was performed during an inspiratory breath hold, and ECG-triggered tube current modulation was used to reduce radiation exposure. Parameters included a slice collimation of 64 × 0.6 mm, temporal resolution of 165 ms, tube voltage of 120 kVp, and tube current of 850 mAs. Images were retrospectively gated to the ECG and reconstructed at 65% of the R-R interval using a medium-smooth kernel and slice thickness/increment of 0.75/0.4 mm.

The PET images were manually coregistered with CT images (Leonardo TrueD, Siemens Forchheim, Germany) by an investigator who was blinded to the patients' clinical history and classification. Perfect coregistration of chest structures between the PET and CT images was not feasible given the differences in cardiovascular and respirophasic motion affecting the major structures within the PET images (which were acquired over 25 min of free breathing) and the CT (which was ECG gated and acquired during a breath hold). Accordingly, we used a strategy of prioritizing the registration of the ascending aorta given that it is discernable on both PET and CT image sets and that its proximity to the tethered LM increases the likelihood of excellent LM coregistration in turn.

Once coregistered, the maximum standardized uptake value (SUV) of FDG was measured within the ascending aorta and within several pre-defined coronary locations. To accomplish this, CT images were used to guide placement of 5-mm² regions of interest (ROIs) over the vascular segment being measured and the maximum SUV value within each ROI was recorded. First, measurement of aortic activity was acquired as the average of 6 ROIs placed along the wall of the ascending aorta 20 mm above the aortic annulus. Next, SUV measurements were obtained in several pre-defined coronary locations in all subjects: the LM, the proximal and mid–left anterior descending (LAD), the left circumflex artery (LCx), and the right coronary artery (RCA). Activity in the LM was measured approximately 8 mm from the LM ostium to avoid spillover of activity from the aorta. Activity in the proximal and mid-LAD and LCx was measured approximately 8 and 16 mm from the bifurcation or trifurcation, and activity in the proximal and mid-RCA was measured approximately 8 and 16 mm from the RCA ostium. Next, FDG uptake was measured at the site of the culprit coronary lesion, which was defined as the coronary segment deemed by the clinical team to be responsible for the syndrome, for which a stent was deployed at presentation. The culprit lesion was in turn identified on the CT images by locating the stent. The ROI was placed at the location of highest plaque activity overlying the stent, and the maximum SUV was recorded.

To obtain a background value for FDG uptake, blood SUV was determined by placement of 6 1-cm³ ROIs within the left atrial cavity in locations devoid of significant spillover activity. The mean SUV value for each ROI was measured and the average of the 6 arterial values was recorded. Afterward, a target-to-background ratio (TBR) was calculated for each vascular segment measured as the segment maximum SUV divided by atrial blood mean SUV.

Suppression of myocardial FDG uptake was assessed according to the qualitative visual grading scale proposed by Williams and Kolodny (24): 0 = minimal uptake; 1 = mostly minimal or mild uptake; 2 = mostly intense or moderate uptake; 3 = homogeneously intense.

To evaluate interobserver reliability of the coregistration and measurement process, images from 10 randomly selected patients were independently coregistered by 2 investigators (A.T., I.R.) on 2 separate occasions. The investigators were blinded to all information regarding the case, including clinical history and group classification of each subject. Aortic and coronary measurements were repeated as previously described.

Blood was collected from subjects on the day of imaging before FDG injection or image acquisition. Several inflammatory biomarkers were measured, including high-sensitivity C-reactive protein (CRP), CD40, intercellular adhesion molecule-1, interleukin-12 p70 (IL-12 p70), tumor necrosis factor-alpha, immunoglobulin M, and vascular cell adhesion molecule-1, via multiplexed immunoassay (Rules-Based Medicine, Inc., Austin, Texas).

Analysis of the baseline characteristics of the ACS and stable groups was conducted through the use of t tests for normal continuous variables, Wilcoxon rank sum tests for nonnormal continuous variables, and Fisher exact tests for binary variables. The TBR data are presented as median and interquartile range (IQR), and the differences between the groups were assessed by Wilcoxon rank sum tests. Correlations are reported as Spearman correlation coefficients. Intraclass correlation coefficients were calculated to determine interobserver reliability using the mean SUV values obtained by the readers during the reliability analysis. All analyses were performed with SAS (version 9.1, SAS Institute, Inc., Cary, North Carolina). A value of p < 0.05 was considered significant.

Demographic information regarding the enrolled patients are summarized in (Table 1). The PET/CT examination was performed within a median of 8 days (IQR 3 to 21 days) in patients with recent stent placement for ACS and within a median of 16 days (IQR 8 to 22 days) in patients with recent stent placement for stable angina. The difference in time from stent placement between groups was not significant (p = 0.62).

Myocardial uptake of FDG was minimal in 10 of 25 (40%) subjects, mostly minimal or mild in 11 of 25 (44%) subjects, mostly intense or moderate in 3 of 25 (12%) subjects, and homogeneously intense in 1 (4%) subject. There was no difference in the degree of myocardial uptake among subject groups (p = 0.93). In the 25 patients studied, 100% (25) of ascending aortas and left main segments were evaluable. A total of 175 total coronary segments were investigated, 96% (168 of 175) of which were evaluable by PET. Seven segments (4%) could not be evaluated because of spillover of adjacent myocardial uptake. The blinded interobserver reliability analysis for the coregistration process revealed intraclass correlation coefficients between readers of 0.92 in the ascending aorta, 0.97 in the LM, 0.84 in the proximal LAD, 0.99 in the proximal LCx, and 0.29 in the proximal RCA.

Analysis of the PET signal within stented lesions was feasible in 89% (24 of 27) of all stented lesions, including 8 of 10 in the ACS group and 16 of 17 in the stable angina group (Table 2). The PET signal (given as median TBR [IQR]) within the culprit lesions in patients with ACS (2.61 [2.27 to 3.45]) was higher than in stable angina patients (1.74 [1.14 to 2.0], p = 0.02) ((Figure 1) and Figure 2). In contrast, there was no significant difference in plaque activity between newly stented stable angina plaques and remotely stented plaques (p = 0.49). Additionally, plaque TBR was significantly higher within the culprit lesions of patients presenting with ACS than within the proximal vessels of these same patients (2.61 [2.27 to 3.45] vs. 1.88 [1.3 to 2.4], p = 0.04).

Of the 8 culprit lesions evaluated in ACS, 5 received bare-metal stents and 3 received drug-eluting stents. Of the 16 stented segments in the stable angina group, 6 received bare-metal stents and 10 received drug-eluting stents. There was no difference in TBR between plaques stented with bare-metal stents versus drug-eluting stents in the ACS group (p = 0.70) or among all subjects (p = 0.55). Similarly, there was no correlation between TBR and: 1) stent diameter in either the ACS group (Spearman r = 0.08, p = 0.86) or among all subjects (r = −0.02, p = 0.95); or 2) stent length in either the ACS group (r = 0.39, p = 0.39) or among all subjects (r = −0.31, p = 0.17).

The PET signal within the ascending aorta was higher in subjects presenting with ACS versus stable angina (values given as median TBR [IQR]) (3.30 [2.73 to 4.0] vs. 2.43 [2.00 to 2.9], p = 0.02). Examples of a subject with increased FDG uptake in the ascending aorta are shown in (Figure 3). Similarly, the PET signal was significantly greater within the LM of subjects presenting with ACS versus stable angina (2.48 [2.30 to 2.93] vs. 2.00 [1.71 to 2.44], p = 0.03) (Figure 4). Although the signal within the proximal and mid-LAD segments was higher in ACS versus stable groups (2.20 vs. 1.75 and 1.50 vs. 1.0, p = 0.03 and p = 0.04, respectively), the signals in the proximal LCx and proximal RCA were not significantly different between the 2 groups (2.00 vs. 1.75 and 2.17 vs. 1.60, for median TBR in ACS vs. stable angina in LCx and RCA, p = 0.30 and p = 0.059, respectively).

Subjects with ACS had a higher blood level of log high-sensitivity CRP and IL-12 p70 than stable angina subjects (p = 0.02 and p = 0.02, respectively). Moreover, several blood inflammatory biomarkers correlated with the activity measured within stented segments. The TBR in the plaques treated for ACS or stable angina correlated with levels of CRP (r = 0.58, p = 0.04), immunoglobulin M (r = 0.77, p = 0.002), IL-12 p70 (r = 0.62, p = 0.02), and tumor necrosis factor-α (r = 0.56, p = 0.048).

Within this feasibility study, we found for the first time that FDG uptake in the ascending aorta and LM is increased in patients with ACS compared with patients with stable angina. Furthermore, we found that FDG uptake is greater in lesions judged to be responsible for ACS compared with lesions associated with stable coronary syndromes.

The correlation of ¹⁸F-FDG uptake and atherosclerotic inflammation has been previously demonstrated in animal and human studies. Studies in humans show that FDG uptake correlates strongly with plaque inflammation (20) and that the measurement of atherosclerotic FDG signal is highly reproducible with intraclass correlation coefficients between 0.90 and 0.98 in medium to large arteries (16). Furthermore, the plaque signal is modifiable, as demonstrated in animal models and clinical trials (18,26- 28). Consistent with the concept that ACS is associated with an enhanced inflammatory milieu, we found that FDG uptake in the ascending aorta was greater in patients presenting with ACS than in stable angina patients. Similarly, we observed an increase in the PET signal within the LM of patients presenting with ACS.

One important prior obstacle to demonstrating feasibility coronary plaque imaging with PET had been the lack of a standard to support the contention that the observed hot spots were indeed derived from coronary plaques. In this study, we developed a reference point for clinically significant (culprit) lesions with a position that is conveniently marked by a stent placed at the location deemed to be the culprit lesion at the time of catheterization. We found that FDG uptake within culprit lesions ACS, which as a group are known to substantially inflamed (29), was significantly higher than within stable angina plaques. Furthermore, a significant difference remained when activity within recently stented stable angina plaques and remotely stable angina plaques were compared, suggesting that the increased signal observed in ACS was related to the underlying inflammation and not solely a consequence of vascular injury after stent placement.

To minimize the impact of coronary motion on our image analysis, we prioritized the coregistration of the ascending aorta, which conveniently tends to show increased FDG uptake (16,30) in patients with atherosclerosis. Coregistration of the ascending aorta in turn yielded excellent registration of the proximal coronary tree. Additionally, we directed our attention to the LM, which is tethered to the aorta and has a nonepicardial location, thus it experiences less motion within the cardiac cycle (31) and is less susceptible to interference by myocardial FDG uptake. Even in cases of excellent myocardial suppression of FDG activity, enough residual activity remained to permit registration of the cardiac structures. It should be noted that ECG gating of the PET acquisition was initially attempted in the development of this protocol but was not ultimately used because in our experience the loss of PET data associated with gating offset its advantage of reduced motion. However, we believe that refinements in gating methods, such as those reported by Suzuki et al. (32), might prove useful for coronary PET imaging.

We observed that the coregistration and measurement process was highly reproducible within the ascending aorta, LM, proximal LAD, and proximal LCx, whereas measurements in the RCA and in the mid- and distal coronary tree were substantially less reproducible. We posit that the limited reproducibility in the RCA and the distal coronary tree likely relates to the well-described relative increase in coronary motion in those locations. Although efficiency in the coregistration of images improved significantly with experience, the process remains time consuming and will require continued development before implementation outside of a research setting.

It is important to note that the inflammatory signal measured in our patients with ACS might not reflect the state of the signal just before the clinical presentation. Either the systemic inflammatory milieu associated with the acute ACS, the additional inflammation associated with stenting, or both might have contributed to the high PET signals that we measured in our patients with ACS after their stent implantations. It is possible that if the ACS subjects were hypothetically imaged just before their events, the measurable vascular PET signal might not have been different from that of stable subjects. Ongoing studies examining the impact of therapies on the coronary PET signals should yield further insights. Moreover, natural history studies will be needed to evaluate whether the aortic and coronary PET signals have clinical utility.

The findings of this study show the feasibility of FDG imaging of the LM and that vascular activity is increased in the aorta and LM of patients presenting with ACS. Furthermore, we observed increased FDG uptake associated with culprit coronary lesions in patients presenting with ACS. Because FDG uptake serves as an index of plaque inflammation, this imaging approach may provide insights regarding the state of inflammation within the coronary vasculature. These observations support further efforts to refine PET methods to image coronary plaque inflammation.

For (6), please see the online version of this article.

Feasibility of 18-Fluorodeoxyglucose Imaging of the Coronary Arteries: Comparison Between Patients With Acute Coronary Syndrome and Stable Angina