Introduction
Surgical removal of atherosclerotic plaques from the carotid artery highly reduces the risk of future stroke in symptomatic patients with ≥ 70% stenosis.
1 However, most of these patients will not have a new event when treated with best medical therapy.
2 Furthermore, the role of surgery in moderate symptomatic stenosis (50% to 69%) and asymptomatic stenosis is under debate.
3‐
5 Therefore, taking into account the potential risk for surgical complications, the selection of patients who will benefit most from surgery is challenging.
In order to improve risk stratification, research has been focused on the identification of plaques at risk for rupture, so-called vulnerable plaques.
6‐
8 Currently, plaque thickness and intraplaque processes, such as inflammation and microcalcification, are seen as important contributors to vulnerability. These processes have become targets of various molecular imaging techniques, as they potentially allow non-invasive risk stratification of individual patients with carotid artery stenosis.
9,
10
Recently, several studies have shown the feasibility of
18F-sodium fluoride (
18F-NaF) positron emission tomography (PET) for imaging of atherosclerotic plaques.
11‐
1318F-NaF predominantly binds to areas of microcalcification within the plaque.
14 Appearance of microcalcifications indicates the active formation of calcification and is associated with plaque vulnerability.
15,
16 In contrast, established calcifications are seen as atherosclerotic end stage products and are associated with plaque stability.
17‐
20
It has been suggested that
18F-NaF may additionally be a useful marker for plaque vulnerability.
21 Indeed, a clinical study by Joshi et al. showed that ruptured and high-risk coronary plaques have a significantly higher
18F-NaF uptake than non-culprit and low-risk coronary plaques.
22 However, data on
18F-NaF uptake in carotid plaques are limited and their usefulness for the prediction of future stroke is unclear.
23‐
25 Additionally, limited data have been published on the relation between active microcalcifications and established calcifications in human carotid plaques.
26,
27
The primary objective of this study is to compare in vitro microPET assessed 18F-NaF uptake between culprit and non-culprit carotid plaques from stroke patients, using non-macrocalcified renal arteries from healthy kidney donors as negative controls. The secondary objective is to compare the distribution of 18F-NaF uptake on microPET with calcification visualized on a high-resolution microcomputed tomography (microCT) in carotid plaques.
Materials and Methods
Study Subjects
Carotid plaques were collected from stroke patients who underwent carotid endarterectomy (CEA) at the Department of Surgery (Division of Vascular Surgery) of the University Medical Center Groningen (UMCG), between July 2015 and March 2016. Indication for CEA was decided by a surgeon expert panel and was based on the presence of symptomatic (culprit) stenosis (≥ 50%) or asymptomatic (non-culprit) stenosis (≥ 70%) of the internal carotid artery, according to internal guidelines.
28,
29 One patient with < 50% stenosis was selected for CEA because of an irregular aspect of the plaque surface.
In order to increase the reliability of our measurements, we used renal artery specimens from healthy kidney donors as negative controls. The specimens were obtained during living donor nephrectomy.
Clinical and demographic data from the included patients were collected from medical records. In the group with culprit plaques, medication use and history of cardiovascular diseases prior to the recent event were registered. The study was reviewed by the ethics committee of the UMCG (METc 2015/258). All patients gave written informed consent.
Study Procedure
Immediately after excision, carotid plaques and renal artery specimens were placed into phosphate buffered saline (PBS) and kept on ice. Both were incubated for one hour in 49.4 ± 7.2 MBq 18F-NaF in 20 mL. After incubation, the plaques and renal arteries were carefully rinsed 5 times with 10 mL PBS. Then, tissue samples were weighed and microPET and microCT scans were performed. After the imaging procedure, the carotid plaques were cut transversely into segments of 3 to 4 mL. The renal arteries had a maximum thickness of 5 mL and therefore no cross-sections were made. The segments were embedded in paraffin for histological analysis.
Production of 18F- NaF
18F-NaF was produced by passing a solution of 18F-fluoride in water over a quaternary methyl ammonium (QMA) light anion exchange cartridge (Waters Chromatography B.V., Etten-Leur, The Netherlands). After washing the QMA with water, [18F]-fluoride was eluted with saline and passed over a sterile Millex GS 0.22 µm filter (Millipore B.V., Amsterdam, The Netherlands). The radiochemical purity for all runs was > 95%.
PET and CT Acquisition
Carotid plaques and renal arteries were positioned into a microPET scanner (MicroPET Focus 220, Siemens Medical Solutions USA, Knoxville, TN, USA), and an emission scan of 30 minutes was performed. After the PET scan was finished, the bed of the PET scanner was transferred to a microCT scanner (Inveon CT, Siemens Medical Solutions USA, Knoxville, TN, USA) without moving or touching the tissue samples. The CT exposure settings were 50 keV and 500 µAs, and a 100-ms exposure time for 360 projections during one 360° rotation.
The PET scans were reconstructed into a single frame of 30 minutes, using OSEM2D (4 iterations and 16 subsets), after being normalized and corrected for attenuation and decay of radioactivity. The CT images were reconstructed with the Feldkamp algorithm.
30
Data Analysis
The PET and CT images were automatically registered using PMOD 3.7 (PMOD Technologies LLC, Zürich, Switzerland). The registration was visually inspected and manually corrected when necessary. For quantification of the average 18F-NaF uptake, three-dimensional volumes of interest (VOIs) were drawn around the whole tissue samples. The uptake (in kBq/cc) was corrected for weight of the specimen and the incubation dose, and expressed as percentage uptake of total incubation dose per gram of tissue (%Inc/g). It was assumed that 1 cubic centimeter equals 1 gram of tissue.
VOIs were also automatically drawn around
18F-NaF PET areas with a threshold of ≥ 50% of the maximum
18F-NaF uptake and assigned as
18F-NaF PET VOI. VOIs were automatically drawn around CT areas with a Hounsfield Unit (HU) ≥ 1000 and assigned as CT calcification VOI. The threshold of 50% of the maximum uptake value was chosen in order to select the volume with the highest
18F-NaF uptake, and thereby minimize the bias of a partial volume effect.
27 The HU of 1000 was based on the CT scan of a phantom with various known calcium hydroxyapatite densities, whereby a lower threshold was chosen in order to not miss any calcification. To determine the overlap between the
18F-NaF PET VOIs and CT calcification VOIs, an intersection VOI was automatically drawn. Then, the CT calcification area (HU ≥ 1000) within the
18F-NaF PET VOI was measured and expressed as a percentage of the
18F-NaF PET VOI; and the other way around;
18F-NaF PET uptake area (≥ 50% of maximum
18F-NaF uptake) within the CT calcification VOI was measured and expressed as a percentage of the CT calcification VOI.
Histological Staining
To validate our data, von Kossa and alizarin red stainings for calcification were performed on two plaque segments without any CT calcification, but with clear 18F-NaF uptake. The pattern of 18F-NaF uptake on PET images was compared with the results of histology. Furthermore, the two renal arteries with the highest 18F-NaF uptake were selected for staining (negative controls). Only negligible 18F-NaF uptake was expected and no calcification in the renal arteries, because only healthy kidney donors with a renal vasculature without any signs of atherosclerosis are eligible for transplantation.
To obtain further information on the vulnerability of included carotid plaques, segments of five culprit and three non-culprit plaques were assessed for histological features of vulnerability. The segments were stained with Martius Scarlet Blue (MSB) using histochemistry and for CD68- and CD34-expressing cells using immunohistochemistry.
31 With these markers, the presence of intraplaque thrombus and collagenous fibrous cap (MSB), inflammation (CD68-positive macrophages), and intraplaque microvessels (CD34-positive endothelial cells) could be identified. One observer, highly experienced in vascular pathology, visual inspected and interpreted the histological features, and compared the culprit and non-culprit plaques. For a detailed description of the staining procedures, see Online Resource 1.
Statistical Analysis
Descriptive data are presented as frequencies (percentage), median (interquartile range), or mean ± SD. Based on the distribution of data (tested by normal probability plots), differences between data were analyzed with non-parametric tests. For continuous data the Mann–Whitney U test (two groups) or the Kruskal–Wallis test (≥ two groups) was used. Categorical data were analyzed with the Chi Square test. A Spearman Correlation was used to test the association between continuous data. A two-sided P-value ˂ 0.05 was considered statistically significant. Statistical analyses were performed using SPSS for Windows (version 23.0).
Discussion
The present study investigated in vitro microPET assessed
18F-NaF uptake, in culprit and non-culprit human carotid plaques. We hypothesized that
18F-NaF uptake in culprit plaques was higher than in non-culprit carotid plaques, based on previous results in coronary plaques, and the concept of microcalcification and plaque rupture.
15,
22 Our results, however, demonstrate comparable
18F-NaF uptake in culprit and non-culprit carotid plaques. Interestingly, we found that
18F-NaF uptake was present in regions without evidence of calcification on CT scan. Furthermore, most of the CT calcification VOIs had low
18F-NaF uptake, confirming that both techniques represent a different stage of calcification.
14,
23,
32
Recently, Vesey et al. showed that in vivo
18F-NaF uptake was higher in culprit carotid artery stenosis than in the contralateral non-culprit stenosis in 18 patients with recent CVA (log
10 standardized uptake value, mean 0.29 ± 0.10 vs 0.23 ± 0.11 respectively,
P < 0.001).
23 These findings are consistent with the results of a clinical study of Quirce et al., in nine patients, where
18F-NaF uptake reported as mean target-to-background ratio was higher in culprit plaques (2.12 ± 0.44) than in contralateral non-culprit plaques (1.85 ± 0.46,
P = 0.220).
24 Age and sex distribution were comparable between these two studies and our study. Only Vesey et al. provided information about the cardiovascular history and other cardiovascular risk factors of the included patients. The prevalence of smoking and diabetes mellitus was similar. Furthermore, more than 50% of the patients had other manifestations of atherosclerosis, as in our study. Since the included patients in both studies were comparable, the remaining question is how these contradictive results can be explained.
First, although Vesey et al. did find a significant difference between
18F-NaF uptake in culprit and contralateral non-culprit plaques, the differences were small and a substantial overlap between the uptake values in both groups was present, as was in the study of Quirce et al. Furthermore, the
18F-NaF uptake between patients scheduled for CEA and control patients differed to a larger extent (delta 0.17 SUV
mean) than the difference between culprit and non-culprit uptake (delta 0.07 SUV
mean).
23 Therefore, we believe that no absolute cut off value for the diagnosis of culprit plaques based on
18F-NaF uptake can be determined, only when compared with control patients there is a relevant difference. This is in line with the results of our study. Although the sample size is small, our data show that
18F-NaF uptake between culprit and non-culprit carotid plaques was comparable while the uptake in plaques was significantly higher than in control renal arteries.
Second, there may be differences in the degree of stenosis of the carotid arteries between patients in the aforementioned studies and our patients. Vesey et al. found that 18F-NaF uptake was related to the degree of stenosis on CT, but they did not report the stenosis degree in the separate groups, neither did Quirce et al. In our study, the stenosis degree in both, non-culprit and the culprit plaques, was high and all plaques showed high-risk features based on histology. This could explain the similar 18F-NaF uptake.
In contrast, Joshi et al. did find a difference between culprit and non-culprit
coronary plaques of patients with myocardial infarction.
22 This might be the consequence of local differences in the mechanical forces exerted to the artery wall and the endothelium of the coronary vascular bed, causing local differences in plaque initiation and progression.
33 In contrast to the coronary arteries, both carotid arteries are exposed to similar blood flow patterns and mechanical stress. This might result in the same pattern of plaque development and progression.
18F-NaF activity was increased in areas without calcification on CT and most of the CT calcification VOI showed minimal
18F-NaF uptake. This supports the idea that microcalcification, as visualized with
18F-NaF PET, and established calcification visualized on CT may reflect different stages of the calcification processes in atherosclerotic plaques.
23,
27,
34
Established calcification is a well-known marker of total plaque burden and is strongly associated with the risk for cardiovascular events.
35 However, the amount of established calcification, as detected by CT, only provides information about the processes in the past and not about the actual biological activity of the plaque.
36 Moreover, larger and denser areas of calcification may even stabilize the plaque.
37 This has, for example, been suggested by Shalaan et al., who found a higher CT assessed calcification volume in non-culprit than in culprit carotid plaques.
38 The average percentage of plaque volume that was calcified was comparable with our study. Unfortunately, in our study CT assessed calcification volume could not be compared between culprit and non-culprit plaques, due to limited availability of CT images in the non-culprit group (n = 1) because of image reconstruction failures.
Another important example can be derived from the work of Puri et al., who performed a post-hoc patient-level analysis of 8 prospective trials in which coronary atheroma was measured with intravascular ultrasound. They showed that although statins had a clear plaque-regressive effect, they also promoted coronary atheroma calcification, indicating stabilization.
39
In this study increased
18F-NaF uptake was related to the calcium volume on CT in the majority of the carotid plaques. This is probably caused by binding of
18F-NaF at only the surface of the calcifications.
14 The binding of fluoride to hydroxyapatite is based on ion exchange, rather than incorporation by active transport. This probably also explains why
18F-NaF uptake can still be found in vitro.
However, in our study a few plaques with low calcium volume had a high
18F-NaF uptake and vice versa. This suggests
18F-NaF accumulation in areas without any evidence of calcification, or at least no calcification with a size above the detection limit of the microCT scan.
14 The presence of calcifications smaller than the CT detection limit, i.e. microcalcifications was indeed confirmed by histological staining of
18F-NaF positive and CT negative segements.
18F-NaF probably binds to the relatively large surface of microcalcifications , causing an intense signal on PET images.
14
These observations indicate that
18F-NaF imaging can detect biologically active plaques, before they can be visualized on CT. This implies that
18F-NaF imaging may be useful in evaluating disease progression, as was further shown in patients with aortic stenosis, where baseline
18F-NaF uptake correlated well with the calcium progression after 1 year.
40 Especially,
18F-NaF uptake in areas without established calcification on CT was the best predictor of calcium progression.
Furthermore, Derlin et al. found a positive correlation between
18F-NaF uptake in the carotid arteries and, age and various cardiovascular risk factors in 269 patients with no history of stroke.
32 Derlin et al. included a heterogeneous population, consisting of patients with low or minimal cardiovascular risk as well. In contrast, we included only patients with already a history of cardiovascular disease and, therefore, at a high-risk for a cardiovascular event. These findings further highlight the possibility of
18F-NaF imaging to identify patients at high-risk for cardiovascular disease in a low-risk population. In addition, the finding that
18F-NaF uptake in carotid plaques exceeded that of controls (renal arteries) and no calcification was visible in renal arteries on microCT or with histological staining adds evidence to the hypothesis that the presence of microcalcification identified by
18F-NaF is a feature of atherosclerosis. The relation between the extent of
18F-NaF uptake in bilateral carotid plaques and features of vulnerability needs to be further investigated.
Strengths of our study are the inclusion of a control group, and the scanning of calcium phantoms in order to accurately determine the calcium threshold. Furthermore, by comparing calcification identified by 18F-NaF PET imaging with calcification visualized on microCT, this study adds knowledge to the relatively new field of 18F-NaF imaging in atherosclerosis. Our study has some limitations. First, it should be considered that the number of plaques, especially non-culprit plaques, is small and a type II statistical error might be introduced. However, given the similar distribution of 18F-NaF uptake in both groups, a high number of plaques would need to be recruited to find a statistical significant difference if present. We believe that small differences in uptake will not have any clinical implication.
Second, the tracer uptake could be underestimated due to partial volume effects, as with every imaging study. Third, CT images of 16 plaques (one non-culprit) out of 23 were available for analysis due to practical and technical issues. Fourth, we did not assess the plaque morphology of the right and left side in the same patient. Culprit and non-culprit plaques were derived from different patients.