Temporal Changes in Coronary 18F-Fluoride Plaque Uptake in Patients with Coronary Atherosclerosis

Visual Abstract

Cor onary atherosclerosis is a chronic inflammatory disease that manifests as expansive plaque formation within the intima of the arterial wall and can lead to plaque rupture, coronary thrombosis, and acute myocardial infarction. In response to the atherosclerotic inflammatory cascade, small deposits of microscopic calcification accumulate in the tunica intima and represent markers of plaque activity (1). Coronary 18 F-sodium fluoride ( 18 F-fluoride) PET is a promising noninvasive imaging modality that can detect these focal regions of developing microcalcification in vivo, identifying patients at risk of future coronary atherothrombotic events (2,3). This imaging technique provides a reproducible metric of coronary microcalcification activity (CMA) (4,5) and can be used to monitor the progression of coronary artery calcium in patients with advanced coronary atherosclerosis (6).
Coronary 18 F-fluoride uptake is observed within the territory of the culprit plaque after an acute myocardial infarction (3). Ex vivo histologic validation of coronary artery specimens has confirmed that 18 F-fluoride colocalizes with microcalcification in the tunica intima (7). Moreover, in patients with stable chronic disease, there is increased coronary 18 F-fluoride uptake in atherosclerotic plaques with morphologically high-risk features on intravascular ultrasound and CT, including positive remodeling and low-attenuation plaque (3). However, temporal changes in the development, evolution, and passivation of features of high-risk coronary plaque in patients at increased risk of cardiovascular events remain poorly characterized. In particular, the natural history of 18 F-fluoride uptake within the coronary arteries of patients who have had a recent myocardial infarction is unknown. We hypothesized that coronary 18 F-fluoride uptake would decrease over 1 y in patients with a recent type 1 myocardial infarction but not in those with stable chronic disease. To address this issue, we performed a prospective observational cohort study using serial 18 F-fluoride PET at 3 time points over 1 y in patients with multivessel coronary artery disease and either a recent type 1 myocardial infarction or chronic stable disease.

Study Design
This was an investigator-initiated prespecified prospective observational cohort study nested within 2 clinical trials of investigational medicinal products: the Prediction of Recurrent Events with 18 F-Fluoride to Identify Ruptured and High-Risk Coronary Artery Plaques in Patients with Myocardial Infarction (PRE 18 FFIR, NCT02278211) and the Dual Antiplatelet Therapy to Inhibit Coronary Atherosclerosis and Myocardial Injury in Patients with Necrotic High-Risk Coronary Plaque Disease (DIAMOND, NCT02110303) trials (8). The nested cohort studies were approved by the local institutional review board, the Scottish Research Ethics Committee (REC references 14/SS/0089 and 16/SS/ 0025), the United Kingdom Administration of Radiation Substances Advisory Committee, and the Medicines and Healthcare Products Regulatory Agency. It was performed in accordance with the Declaration of Helsinki. All patients provided written informed consent before undergoing any study procedures.

Study Population
Patients were recruited between March 2015 and July 2019. Inclusion criteria for the natural history cohort studies required the presence of multivessel coronary artery disease on invasive coronary angiography, either within 21 d of an acute type 1 myocardial infarction (NCT02278211) or in the context of advanced chronic coronary artery disease (NCT02110303). Patients were excluded if they were unable to receive iodinated contrast medium, had renal impairment (estimated glomerular filtration rate # 30 mL/min per 1.73 m 2 ), or were female and of child-bearing potential. Full eligibility criteria are provided in Supplemental Table 1 (supplemental materials are available at http://jnm. snmjournals.org). All patients underwent a comprehensive baseline clinical assessment including evaluation of their cardiovascular risk factor profile. The REACH (Reduction of Atherothrombosis for Continued Health) and SMART (Secondary Manifestations of Arterial Disease) risk scores were calculated. Both these scores were created specifically to predict risk in patients with established coronary artery disease (9,10).

Image Acquisition
All patients underwent baseline 18 F-fluoride PET and CT on a hybrid scanner (128-multidetector Biograph mCT; Siemens Medical Systems), along with unenhanced CT for calcium scoring and contrast-enhanced coronary CT angiography using a previously described standardized study protocol (4). In brief, participants with a resting heart rate of more than 65 beats/min were administered oral b-blockade (50-100 mg of metoprolol) unless contraindicated. All participants were administered a target dose of 250 MBq of intravenous 18 F-fluoride and rested in a quiet environment. Sixty minutes after the injection, the PET acquisition was performed. Attenuation correction CT scans were performed before the acquisition of electrocardiogram-gated list-mode PET data using a single 30-min bed position centered on the heart. An electrocardiogram-gated breath-hold unenhanced CT scan (tube voltage, 120 kV; tube current based on body habitus) was performed for coronary CT calcium scoring and reconstructed in the axial plane with a 3-mm slice width and 1.5-mm increments. Finally, electrocardiogram-gated coronary CT angiography (tube voltage, 120 kV; tube current based on body habitus) was performed in mid diastole during held expiration after administration of sublingual glyceryl trinitrate. Serial contrast-enhanced coronary CT angiography and 18 F-fluoride PET and CT were performed using the same standardized imaging protocol and on the same scanner at an interval of 3, 6, or 12 mo after the baseline scan. Unenhanced CT for calcium scoring was conducted at 12 mo (for the subgroup nested in the DIAMOND study) and 24 mo (for the subgroup nested in the PRE 18 FFIR study) (Fig. 1). To minimize exposure to ionizing radiation, all patients underwent a total of only 2 18 F-fluoride PET scans.

Image Analysis
PET Analysis and Quantification. Electrocardiogram-gated PET images were reconstructed in diastole (50%-75% of the R-R interval, 2 iterations, 21 subsets, Siemens Ultra-HD algorithm) and fused with the contrast-enhanced coronary CT angiography images. Qualitative and semiquantitative analyses were performed independently by trained observers using a dedicated software package (FusionQuant; Cedars-Sinai Medical Centre).
CMA was used to quantify 18 F-fluoride uptake across the coronary vasculature as described previously (2). In brief, the proximal and distal sections of the vessel (.2 mm) were identified, and a vessel-tracking algorithm was applied to extract whole-vessel tubular 3-dimensional volumes of interest from the coronary CT angiogram using dedicated semiautomated Autoplaque software (version 2; Cedars-Sinai Medical Center). These encompass all the main epicardial coronary vessels and their immediate surroundings (4-mm radius) and were used to measure the CMA (11).
Coronary 18 F-fluoride uptake was assessed along the entire course of the coronary arteries regardless of the presence of coronary stents, and the left main stem was included in the volume of interest for the left anterior descending artery. To avoid an overspill of aortic root activity, coronary uptake at the orifice of the left main stem was excluded. CMA was defined as the average SUV within the activity volume above a background threshold defined as SUV mean plus 2 SDs measured in the right atrial blood pool as described previously (2). A CMA of 0 indicated no activity, and a CMA of more than 1.56 was indicative of high activity as described previously (2).
Coronary Artery Calcium Score. Coronary artery calcium was quantified on a per-vessel level by an experienced observer using dedicated software (Vitrea Advanced; Toshiba Systems). Calcification was quantified as calcium score (Agatston units [AU]), calcium volume (mm 3 ), and calcium mass (mg). Calcium score was derived using the Agatston method. To calculate calcium mass, a calibration factor was derived using a phantom to calculate equivalent water diameter, adjusted for body mass index and lateral diameter, and applied at a specified x-ray tube voltage (Supplemental Table 2). Because of metal artifacts, only vessels without coronary stenting were selected as part of the comparative analysis.
Coronary CT Angiography. The CT images were analyzed using dedicated software (Vitrea Advanced; Canon Medical Systems), with multiplanar reformatting for plaque analysis applied as necessary. Coronary arteries with diameters of at least 2 mm were assessed according to the 18-segment Society of Cardiovascular CT model (12). Disease severity was evaluated using the Duke Coronary Artery Disease Index (13), with 50% or more stenosis classified as clinically significant. The number of vessels involved, and the location of obstructive lesions (left main and proximal left anterior descending coronary arteries), were weighted according to the Duke Coronary Artery Disease Index criteria (Supplemental Table 3).

Statistical Analysis
Continuous variables are presented as mean 6 SD or as median and interquartile range as appropriate. Change in CMA was defined as the geometric mean difference in the CMA value between the baseline and the follow-up PET/CT scans after logarithmic transformation of the dataset. The Shapiro-Wilk test was used to assess normality for continuous data. Two-sample t testing or Wilcoxon rank-sum testing was applied to compare groups for continuous variables; the Pearson, x 2 , or Fisher exact test was used to compare groups for categoric variables as appropriate. Pearson or Spearman rank correlation was used to assess correlations between continuous variables. All statistical analysis was performed on a per-vessel level. The statistical analyses were performed using R, version 4.0.3 (The R Foundation for Statistical Computing). A 2-sided P value of less than 0.05 was considered statistically significant.
Baseline demographics for both cohorts, including age, sex, traditional cardiovascular risk factors, and history of cerebrovascular disease, are demonstrated in Table 1. Patients with advanced chronic coronary artery disease had a higher Duke score ($4 in 84% vs. 23% in those with recent myocardial infarction, P , 0.001). Patients with advanced chronic coronary artery disease also had higher cardiovascular risk prediction scores (REACH score of 14.0 [interquartile range, 11.5-15.5] vs. 9.0 [interquartile range, 8.0-10.0] in those with recent myocardial infarction, P , 0.001). All participants with advanced chronic coronary artery disease had previously undergone coronary revascularization: 25% had previous coronary artery bypass graft surgery, and 75% had a previous percutaneous coronary intervention. None of the patients with a recent myocardial infarction had prior bypass surgery, and 11% had a prior percutaneous coronary intervention, although 96% of patients underwent coronary revascularization after their index event.

Coronary Artery Calcification
Coronary artery calcium was assessed in all nonstented vessels at baseline. In patients with advanced chronic coronary artery disease, vessels were more calcified, with a higher baseline calcium score, higher calcium volume, and higher calcium mass, than in the patients with recent myocardial infarction (Supplemental Table 4 Fig. 3). Qualitative data are number and percentage; continuous data are median and interquartile range or mean 6 SD.
A regression model to assess change in calcium volume was performed adjusting for baseline calcium volume and demonstrated no significant independent association with coronary microcalcification (b 5 coefficient 25.5, P 5 0.421) (Supplemental Table 6). A regression model to assess change in calcium score (AU) was performed adjusting for baseline calcium score and demonstrated no significant independent association with coronary microcalcification (b 5 coefficient 4.13, P 5 0.426) (Supplemental Table 6).

DISCUSSION
In this prospective observational cohort study of patients with advanced chronic coronary artery disease or acute myocardial infarction, we showed that, using coronary 18 F-fluoride, increased CMA is detectable in 3 of 4 patients and that microcalcification activity remains elevated for up to 12 mo after initial assessment. Coronary 18 F-fluoride uptake correlates with disease burden, in terms of both coronary calcification and baseline coronary artery disease severity. Furthermore, coronary 18 F-fluoride uptake correlated with progression of coronary artery calcification at followup. This correlation was consistent across a range of measures of calcification and for patients with either stable or unstable coronary artery disease. Despite these associations, we demonstrated no marked changes in coronary 18 F-fluoride uptake over 12 mo of follow-up in either population. This finding suggests that although coronary 18 F-fluoride uptake is a marker of disease activity, it does not change rapidly with time, consistent with the slowly evolving nature of coronary atherosclerosis.
Calcification plays an important role in the pathogenesis of atherosclerosis and begins early in the disease process (14). CT calcium scoring quantifies macroscopic deposits of calcification and provides a surrogate of total coronary atherosclerotic burden. The relationship between the coronary artery calcium score and major adverse cardiovascular events, including all-cause mortality, cardiovascular events, and nonfatal myocardial infarction, has been well established (15,16). This strong relationship occurs even though heavily calcified plaques are themselves less likely to rupture or precipitate acute myocardial infarction, the rationale being that the more plaque a patient has, the more likely it is that a clinically relevant plaque rupture will occur. Coronary artery calcification is thus a surrogate for the overall burden of coronary artery disease, which will include noncalcified high-risk plaque elsewhere in the coronary circulation.
We have here shown that both coronary 18 F-fluoride uptake and CMA correlate with disease burden as demonstrated by the coronary artery calcium score and the Duke score. This is consistent with previous studies showing that coronary 18 F-fluoride uptake is associated with both luminal stenosis and coronary calcification (6). We have gone on to demonstrate that coronary 18 F-fluoride uptake predicts disease progression, with increasing uptake correlating with more rapid coronary artery calcium progression. This finding is consistent with prior studies (6,17), as well as those reporting that 18 F-fluoride preferentially binds to pathologic mineralization and identifies areas of microcalcification (1). Indeed, 18 Ffluoride binds more readily to regions of developing calcium and acts as a marker of calcification activity, adding distinct information to calcium scoring, which cannot differentiate between quiescent and active disease. This is supported by prior histologic data showing preferential binding of 18 F-fluoride to developing hydroxyapatite (7).  Recent data suggest that 18 F-fluoride PET is a potentially valuable tool in cardiovascular risk stratification. CMA represents a summary measure of 18 F-fluoride uptake within the entire coronary vasculature and, like coronary artery calcium score, is a predictor of future myocardial infarction (5). Indeed, coronary 18 F-fluoride uptake and CMA are associated with an increased risk of future myocardial infarction independent of age, sex, cardiovascular risk factors, segment involvement, coronary artery calcium score, coronary stents, coronary stenosis, Duke score, and recent myocardial infarction (5). These future myocardial infarction events occur over many years, and we wished to assess the time course of coronary 18 F-fluoride uptake and CMA to determine its temporal stability as a measure of coronary artery disease activity. We therefore assessed these measures over differing timelines over a 1-y period in patients with stable and unstable coronary artery disease. We report that there is no major discernible change over a 3-, 6-, or 12-mo period irrespective of the stability of coronary artery disease. This finding suggests that biologic stabilization and healing of coronary atherosclerotic plaque are slow, that plaque activity and vulnerability may be prolonged, and that active coronary calcification persists for many months or indeed years.
Our findings do not undermine the utility of coronary 18 F-fluoride uptake in the identification of metabolically active plaques in patients with coronary artery disease with ongoing calcifying activity and vascular inflammation. Atherosclerosis starts early in life with a long quiescent phase before the manifestation of overt disease. Before the fourth decade of life, subclinical noncalcified plaque forms in the absence of detectable coronary macrocalcification (18).  Without intervention, noncalcified plaque will accumulate at approximately 1 mm 3 per annum, and although statins can accelerate transformation to a calcified phenotype, the increased rate of calcific progression is only 1.27 mm 3 per annum (19). Over many decades, these small differences are amplified and may in part explain the heterogeneity of coronary artery disease presentations in later life (20). This observation of slow incremental change is supported by intravascular imaging studies that reported 1%-2% volumetric changes in dense calcification over 12 mo (21). However, coronary plaques do not all follow a linear trajectory. Although most thincapped fibroatheromas will heal over time, a smaller proportion of plaques with intensely active atherosclerosis may transform into a more vulnerable phenotype (20,21)-hence the rationale for monitoring disease activity using ligand-specific radiotracers. 18 F-FDG has had limited clinical application in the coronary vasculature due to overspill of activity from the myocardium. In the carotid arteries, 18 F-FDG produces a more diffuse uptake pattern along the course of the vessel, as opposed to the discrete signal of 18 F-fluoride, which colocalizes to regions of disrupted laminar blood flow (22,23). More recently, in vivo models have suggested that 18 F-FDG uptake does not represent merely macrophage infiltration and that this diffuse pattern of activity may be more closely aligned with medial smooth muscle uptake (24). This possibility makes it difficult to discern whether the early reduction in 18 F-FDG signal intensity that follows the initiation of plaque-directed therapy is wholly due to a change in inflammatory cell activity in intimal plaque (25)(26)(27). Preclinical animal studies suggest that vascular inflammation and osteogenesis progress in close proximity to, and increase in parallel with, plaque progression (28,29). 18 F-fluoride colocalizes with the distribution of osteopontin and Runx-2, established markers of early calcification activity and adverse plaque formation (7). Analogously, microcalcification is itself associated with markers of plaque vulnerability, such as intraplaque hemorrhage (30), and its presence in the fibrous cap might promote cavitationinduced plaque rupture (31). Paradoxically, macrocalcification represents the end stages of disease, with the formation of homogeneous or sheetlike calcification that effectively walls off the inflamed necrotic core and stabilizes the plaque.
Our findings are consistent with a slow time course in which active disease changes slowly before activity burns out and the plaque becomes quiescent. Microcalcification is a prolonged process compared with active acute inflammation, which is usually short-lived and changes rapidly over time. 18 F-fluoride cannot track the early remodeling changes that have been observed with 18 F-FDG after acute myocardial infarction (27), and our hypothesis that the coronary 18 F-fluoride signal would decrease after acute myocardial infarction was wrong. The longer duration associated with microcalcifications may enable coronary 18 F-fluoride uptake FIGURE 3. Change in coronary artery calcium score and coronary 18 F-fluoride uptake. Vessels with increased CMA at baseline demonstrated higher baseline calcium scores (A) and calcium volume (B) and more rapid progression of calcium scores (C) and calcium volume (D) than vessels without increased 18 F-fluoride uptake. This was consistent for both chronic coronary artery disease and recent myocardial infarction. CAD 5 coronary artery disease; MI 5 myocardial infarction. to detect high-risk plaques at varying phases of atheromatous progression. Moreover, such qualities do make it a more attractive risk marker for future clinical events, as is consistent with our previous finding that baseline coronary 18 F-fluoride uptake and CMA predicted subsequent myocardial infarction at a median of 5 y of follow-up (5). Such a marker of prolonged downstream events is attractive and negates the need for short-term serial scanning or the possibility of false-positive or -negative findings if there was presence or absence of transient inflammation.
We should acknowledge several limitations of our study. Although, to our knowledge, our study included the largest number of consecutive prospectively enrolled patient cohorts to undergo repeat coronary PET and CT angiography for the dynamic assessment of coronary 18 F-fluoride uptake, we recognize that this was a single-center study comprising a largely White male population. Because of the high level of coronary revascularization and stent implantation in our patient cohorts, quantitative analysis of coronary plaque burden was challenging to perform. Future studies exploring the relationship between quantitative plaque characteristics and burden on coronary CT angiography and coronary 18 Ffluoride uptake on PET would be important to evaluate the added value of CMA. Finally, our patient populations all received guideline-directed medical therapy including high use of antiplatelet, statin, and renin-angiotensin system inhibitor therapies. As such, we cannot exclude the modifying effects of the treatment interventions, which are likely to be conservative, on our findings. CONCLUSION Coronary 18 F-fluoride uptake correlates with both coronary artery calcification and disease severity and is a determinant of coronary artery disease progression, irrespective of the stability of coronary artery disease. Coronary 18 F-fluoride uptake was relatively constant over the short term, with no change in activity over 3-12 mo even in patients with recent myocardial infarction. This finding suggests that coronary 18 F-fluoride uptake identifies established and progressive disease that can take considerable time to change and to modify.