This study was part and continuation of the ‘Cardiovascular Molecular Calcification Assessed by 18F-NaF PET/CT (CAMONA)’ study, conducted 2012-2014 , approved by the Danish National Committee on Health Research Ethics (s-20120056), and registered at ClinicalTrials.gov (NCT01724749). CAMONA was carried out in accordance with the Declaration of Helsinki. All study participants provided written informed consent.
CAMONA included 89 healthy individuals with low CVD risk who were recruited via a local advertisement or from the blood bank at Odense University Hospital, Odense, Denmark. Individuals with no history of malignant diseases, immunodeficiency syndromes, autoimmune diseases, illicit drug use, alcohol abuse or CVDs were considered healthy and were eligible for inclusion. Adults were preselected by age and gender to make sure a balanced inclusion of both genders aged 20–29, 30–39, 40–49, 50–59, and 60 years or older was guaranteed. Furthermore, 50 patients suspected of having angina pectoris who were referred to the Department of Cardiology at OUH for coronary angiography were included.
The included patients were asked to fill a questionnaire about alcohol consumption, smoking habits, past medical history, familial history, and current medical status. Blood pressure after at least 30 minutes of rest was measured three times in the supine position. The mean of the last two measurements was recorded as the systolic and diastolic blood pressure. Laboratory tests included total serum cholesterol, serum low-density lipoprotein, serum high-density lipoprotein, serum triglycerides, fasting plasma glucose and glycated hemoglobin, and glomerular filtration rate, which was calculated using the Modification of Diet and Renal Disease equation . The 10-year risk of developing CVD was estimated using the Framingham Risk Score (FRS) based on age, gender, systolic blood pressure, total serum cholesterol, serum HDL cholesterol, smoking habit, and treatment for hypertension . Then they were offered a whole-body NaF-PET/CT scan at baseline and after two years of follow-up performed at the same PET/CT scanner and at approximately the same time of the day (morning or noon). Of all initially included patients, 29 healthy individuals and 20 patients with angina pectoris attended the 2-year follow-up. So, this subgroup of patients was examined to inspect the change prospectively during two years in NaF uptake in major arteries, including the carotids and the arch, thoracic, and abdominal parts of the aorta.
NaF-PET/CT imaging was performed according to previously published methods  on hybrid PET/CT systems (General Electric Healthcare using Discovery PET/CT, 690/710, VCT, or XTe) with comparable spatial resolution. All participants underwent PET/CT imaging 90 minutes after they were injected with approximately 2.2 MBq/kg (max 400 MBq) NaF. The acquisition time was 2.5 minutes per bed position. PET/CT system specifications and parameters of image reconstruction are summarized in the Electronic Supplementary Material 1. The 3D acquisition of total-body PET images and reconstruction of them into transverse, coronal and sagittal slices was made by an iterative reconstruction algorithm (VUE Point; GE Healthcare). The correction of PET images for random, scattered coincidences, attenuation and anatomic directions were done by implanting transmission maps produced by a 64-slice CT scan as follows (120 kV, 200 mA, 16 x 2.5 mm collimation, 0.5 seconds per rotation).
All scans at baseline and follow-up were analyzed and quantified independently without the reader being aware of the participants’ demographic and clinical features. ROVER software version 3.0.4 (ABX GmbH, Radeberg, Germany) was used for quantitative analysis. Initially, PET and CT images were reregistered using DICOM information, allowing us to improve the diagnostic accuracy of both modalities and optimizing the outlining of aortic segments, and then imported into the software. If necessary, additional adjustment of images was made by modification of PET images in transverse, coronal and sagittal planes considering the CT images as the reference point.
The volume of interest (VOI) was formed by stacking manually defined regions of interest (ROIs) using a 5 mm width brush in CT images for each participant. The VOIs included left carotid and right carotid, arch of aorta, thoracic aorta, and abdominal aorta. The arch of the aorta was defined as aorta above the lower level of T5 in a transaxial view until the aortic valve. The carotids were defined from the branching initiation (branching from aorta for left carotid and brachiocephalic artery for right carotid) until the bifurcation (including itself). The thoracic aorta was defined as aorta between the inferior edge of T5 to T12. The abdominal aorta was defined as aorta between the lower level of T12 until the beginning of the bifurcation. A sample of segmented NaF-PET/CT images, including VOIs in 3-dimensional planes, is shown in Figure 1. The manual ROI determination was done in a manner that would contain the whole carotid or aortic wall (intima, media and adventitia), excluding the vertebral bones and their uptake halo from inclusion in defined ROIs. Therefore, in some transaxial slices, where the aorta was adjacent to the vertebral body, the ROI was defined with a lunar shape, unlike all other slices in which the ROI was circular.
Performing a quantitative assessment of PET scans was done by generating standardized uptake values (SUVs) of the VOIs, adjusted to body weight. After segmentation of each ROI, the recorded NaF uptake was expressed as SUVmean (average SUV of all voxels within VOI), SUVmax (the highest SUV of all voxels in the VOI), SUVtotal (sum of the SUVs of all voxels), and as the corresponding measures corrected for partial volume effect (i.e., cSUVmean and cSUVtotal) as described by Hofheinz et al.  The measurement of NaF uptake was, therefore, performed in two automated steps, first by approximation of the actual object boundaries with a threshold-based method and determination of the total activity in ROI and then determining activity fraction, which is measured outside the ROI due to spill-out. With this correction approach, accurate knowledge of image resolution is not necessary as it is, for instance, with deconvolution techniques [18-19]. The measurement unit was MBq/ml. Also, the CT-related variable mean density (CTmean) expressed in Hounsfield units was extracted in all corresponding VOIs.
The reproducibility of quantifying arterial wall NaF uptake is reported in a manuscript submitted elsewhere. By repeat determination of aortic uptake performed in 25 randomly selected scans after several months and without knowledge of prior results, one observer (RP) found in the three segments of the aorta a variation in SUV values of maximally 6 percent.
Descriptive statistics were expressed as frequency (percentage), mean ± standard deviation or median (minimum-maximum). Mann-Whitney U and Fisher’s exact test were used to compare demographic, laboratory, and PET/CT variables between healthy and angina groups. Intragroup comparisons over time were performed with Wilcoxon matched pairs signed rank sum test, and change over time was shown as the mean of estimated differences. Aside from the comparison of control and angina groups and in order to compare NaF uptake in different major arteries within groups, Friedman’s test was used, in which the Wilcoxon matched pairs signed rank sum test was performed post hoc. Finally, the non-parametric Spearman’s correlation test was used to examine for correlation between PET/CT variables and age, then Fisher’s r-to-z transformation method was applied to compare them . NaF uptake was plotted against age in a scatter plot, supplemented by fitted lines from linear regression for each group of subjects. A p-value of less than 0.05 was considered statistically significant. All statistical analyses were performed using SPSS 19.0 (SPSS Inc., Chicago, IL, USA).