Coronary Arterial ¹⁸F-FDG Uptake by Fusion of PET and Coronary CT Angiography at Sites of Percutaneous Stenting for Acute Myocardial Infarction and Stable Coronary Artery Disease

Patient Population

Consecutive adult patients who had undergone successful PCS for AMI or stable CAD at our center were assessed as possible study subjects. AMI was defined by presentation of chest pain consistent with new or worsening angina, followed by detection of troponin I enzyme elevation to >5 times the upper limit of normal (0.05 ng/mL at our center) and peaking within 24 h of presentation. Stable CAD was defined by no suspicion for acute coronary syndrome and at least 1 of the following criteria: ≥1 episode of chest pain consistent with angina beginning >2 wk before PCS, ≥1 mm ST segment depression in contiguous regional leads during exercise testing, >1 segment of inducible wall motion abnormality during exercise or dobutamine echocardiography, or >5% reversible defect on radionuclide myocardial perfusion imaging. Exclusion criteria were prior coronary artery bypass grafting, AMI related to a previously placed stent, placement of >1 stent to treat the culprit stenosis (increased ambiguity of localization of ¹⁸F-FDG signal), position of the new stent within 10 mm of a previously placed stent, inability to follow breathing instructions, allergy to iodinated contrast, active hyperthyroidism, pregnancy, diabetes or maximum post-PCS random blood glucose > 150 mg/dL, ongoing infusion of dextrose-based solution, creatinine > 1.5 mg/dL, estimated creatinine clearance < 45 mL/min, any active supraventricular or ventricular tachyarrhythmia, and patient's decision not to participate. This study was approved by our center's institutional review board. All participating patients provided written informed consent.

Enrolled patients underwent cardiac ¹⁸F-FDG PET and coronary CTA within 7 d of PCS (median time, 2 d). Both scans were obtained in the same imaging session, using the general scanning protocol shown in Figure 1.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

¹⁸F-FDG PET and coronary CTA imaging protocol used in study. Ten minutes before administration of intravenous ¹⁸F-FDG, 5,000 units of intravenous heparin were given in attempt to shift myocardial metabolism further toward fatty-acid use. After ¹⁸F-FDG injection, oral metoprolol was given to control heart rate for CTA, which was performed approximately 60 min before PET image acquisition. Total time from ¹⁸F-FDG administration to PET image acquisition was 180 min.

Cardiac ¹⁸F-FDG PET Image Acquisition and Reconstruction

PET was performed on a hybrid 4-ring PET/CT scanner (Biograph 64; Siemens Medical Systems). In preparation for imaging, patients ate a low-carbohydrate dinner the night before, followed by fasting until the scan was complete (all patients fasted for >12 h before ¹⁸F-FDG injection). Our prior work showed that this approach produced the lowest myocardial uptake (13). Whenever possible, patients were first administered 5,000 units of intravenous heparin on the basis of prior reports that intravenous heparin can augment myocardial fatty acid metabolism (14,15). Ten minutes after heparin injection, 480 MBq of intravenous ¹⁸F-FDG were given. Imaging began 3 h afterward (8). Initially, low-dose noncontrast CT for attenuation correction (CTAC) was performed using spiral mode, 3.0-mm slice thickness, 1.5 pitch, 0.5-s rotation time, 24 × 1.2 mm collimation, 120-kVp tube voltage, and 11-mAs tube current. ¹⁸F-FDG emission data were then acquired in list mode at 1 bed position (216-mm coverage) with cardiac gating. Patients were instructed to breathe normally during both CT and PET. Emission data were captured for 25–30 min.

Summed images were reconstructed using the first 10 min of emission data. High-definition reconstruction was performed with 4 iterations and 14 subsets and a 2-mm gaussian postprocessing filter (16,17). The reconstruction matrix was 168 × 168 × 109, with a zoom of 2 and voxel size of 2 × 2 × 2 mm. Scatter correction, decay correction, and random correction were applied. CTAC data were used for attenuation correction. The alignment of CTAC and PET was reviewed by an experienced reader to ensure proper coregistration, using the aorta, myocardium, sternum, liver, and vertebral bodies as major landmarks. In all cases, CTAC and PET alignment was found to be optimal.

Coronary CTA Image Acquisition, Reconstruction, and Analysis

CTA was performed on a dual-source CT scanner (Somatom Definition; Siemens Medical Systems). When needed, oral or intravenous metoprolol (oral, ≤100 mg; intravenous, ≤30 mg) was administered to achieve a heart rate of <70 beats per minute. Three to five minutes before image acquisition, 0.4 mg of sublingual nitroglycerin spray (Sciele Pharma) was administered. Eighty milliliters of intravenous contrast (Omnipaque; GE Healthcare) were then power-injected into the antecubital vein, followed by 50 mL of saline at a rate of 5 mL/s. Once >100 Hounsfield units were detected in the ascending aorta, electrocardiogram-gated scanning from the carina to just below the diaphragm was performed in 1 breath-hold. Dose modulation was used in all patients, with maximal current given at between 40% and 70% of the R-R interval (if heart rate was ≥70) or just at 70% (if heart rate was <70). Scanning parameters included heart rate–dependent pitch (0.2–0.45), 330-ms gantry rotation-time, 120-kVp tube voltage, and 330–350 mAs reference tube current. Acquired data were reconstructed in mid diastole and at end systole (if data available) using a 0.6-mm slice thickness (0.75 mm if body mass index was >35 kg/m²), 0.3-mm slice increment, 250-mm field of view, 512 × 512 matrix, and B26f medium-smooth kernel.

Coronary artery anatomy was evaluated by a blinded expert reader using axial images, oblique multiplanar reformations, and oblique maximum intensity projections. The recently placed coronary artery stent was assessed for location and patency. At the proximal and distal ends of the stent, the true short-axis view of the stent was used to measure the shortest distance from the stent edge to ventricular myocardium.

Coregistration and Analysis of Combined ¹⁸F-FDG PET and Coronary CTA

Coregistration and analysis of all cases were independently performed by 2 experienced blinded readers using software developed at our center (QPET). CTAC images were fused with CTA images to best overlay the 3-dimensional position of the recently placed stent in both datasets by rigid body alignment (translation in 3 dimensions). These registration parameters were then applied to PET data for fusion with CTA images (Fig. 2). The same coregistration steps were repeated for the ascending aorta at the level of the pulmonary artery bifurcation, using the pulmonary arterial bifurcation as principal landmark. To detect maximal ¹⁸F-FDG uptake at the PCS site, a 1 × 1 × 1 cm cubical region of interest centered at the luminal midline of the stented segment was moved from the proximal to distal end of the stent. To explore ¹⁸F-FDG in nonculprit coronary arteries, we examined fused PET and CTAC datasets and applied the cubical region-of-interest technique along the left main and proximal coronary arterial segments, using the interventricular and atrioventricular grooves as guiding landmarks. This approach was selected because nonculprit arteries lacked a clearly visible marker (stent), substantially increasing subjectivity when fusing PET and coronary CTA. Mid and distal segments were not included in this analysis to avoid excessive coregistration error, which was expected to increase as the locations of interest move further from the aortic root. To obtain maximal ¹⁸F-FDG signal from the ascending aorta at the level of the pulmonary artery bifurcation, 6 regions of interest were evenly spaced along the ascending aortic circumference (in cross-section, corresponding to 12-, 2-, 4-, 6-, 8-, and 10-o'clock positions). The average of maximal ¹⁸F-FDG intensity from 4 regions of interest within the left atrial cavity was used to define background signal intensity and facilitate calculation of maximal target-to-background ratio (maxTBR). ¹⁸F-FDG signal from the myocardium near the stent was assigned a visual score of 0–3 (0, no uptake; 1, mild uptake; 2, moderate uptake; and 3, dense uptake); a score ≥ 2 was considered excess uptake (inadequate suppression). After completion of independent data collection, results from both readers were reviewed. In the cases in which the difference in maxTBR from independent measurements was ≤0.2, the average of reader values was used as the consensus maxTBR. When the difference in maxTBR between readers was >0.2 (9 cases for PCS site, 5 cases for ascending aorta), repeated coregistration and analysis of the image data were performed to obtain consensus; this was a completely fresh analysis performed by both readers in concert, and the resultant maxTBR could be lower or higher than initial values by both readers. Consensus evaluation of myocardial ¹⁸F-FDG uptake was performed whenever readers disagreed regarding the presence of excess uptake (3 cases). MaxTBR > 2.0 was considered evidence of definitely increased ¹⁸F-FDG uptake (12).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Case example demonstrating steps used to fuse coronary CTA and PET images at site of stenting. (A) Step 1: default overlay of CTA and PET images. Top row of colored images is from CTA and PET. Bottom row of gray-scale images is from CTA and noncontrast CTAC. CTA clearly shows stent in proximal RCA, marked with thick arrows. There is substantial offset in heart and stent positions of both PET and noncontrast CT relative to CTA (thin arrows denote heart silhouette signal in PET and CTAC). (B) Step 2: CTA and CTAC are superimposed to achieve optimal overlay of stent (thick arrows), and coordinates in which this takes place are recorded. Compared with images in step 1, heart silhouette of CTA and CTAC is much better aligned. (C) Step 3: PET images are superimposed onto CTA images at coordinates determined by step 2. Fused images are then used for quantifying ¹⁸F-FDG uptake at stent site. There is visibly increased signal at stent site after fusion (within white circles). This was 69-y-old woman who had acute ST elevation myocardial infarction from culprit plaque in proximal RCA.

Biomarkers

Before initiation of the imaging protocol (i.e., before administration of intravenous heparin and ¹⁸F-FDG), blood samples were obtained in 22 patients (16 with AMI, 6 with stable CAD) to measure serum biomarkers of inflammation, including c-reactive protein, erythrocyte sedimentation rate, IgM, interleukin-12p70, oxidized low-density lipoprotein, and tumor necrosis factor α.

Radiation Dose to Patients

Effective radiation dose from coronary CTA was calculated by multiplying the dose–length product (obtained from the scanner) and the overall organ weighing factor for the chest (k = 0.014 mSv × [mGy × cm]⁻¹) (18,19). Effective radiation dose from 480 MBq of ¹⁸F-FDG was estimated at 9.1 mSv, using published data (20). Total effective dose was the sum of doses from CTA and PET.

Statistical Methods

Continuous variables were described using median, 25th percentile, and 75th percentile values. Categoric variables were described with frequencies. Comparisons of continuous variables between patients with AMI and patients with stable CAD were made using the Wilcoxon rank-sum test. Comparisons of categoric or binary variables were made using the χ² test or Fischer exact test, where appropriate. Calculation of the κ-statistic was used to evaluate interreader agreement of maxTBR > 2.0. Pre-PCS variables that differed between AMI and stable CAD patients (active smoking and aspirin therapy), time between completion of PCS and ¹⁸F-FDG injection, and predefined postimaging variables (myocardial ¹⁸F-FDG signal, minimal distance from proximal and distal ends of the stent to myocardium) were first examined using univariable logistic regression analysis for any potential relationship to maxTBR > 2.0 at the PCS site. These variables were then included with AMI status into a stepwise multivariable logistic regression model, using a maxTBR > 2.0 as outcome. P values of <0.05 were considered statistically significant. All analyses were performed using Stata software (version 10.0; StataCorp LP).