|Home | About | Journals | Submit | Contact Us | Français|
Inflammation and neovascularization play critical roles in the stability of atherosclerotic plaques. Whole-body quantitative assessment of these plaque features may improve patient risk-stratification for life-threatening thromboembolic events and direct appropriate intervention. Here we determined the utility of the MR contrast agent Gadofluorine-M (GdF) for staging plaque stability and compared this to the conventional agent Gd-DTPA.
5 control and 7 atherosclerotic rabbits were sequentially imaged following administration of Gd-DTPA (0.2 mmol/kg) and GdF (0.1 mmol/kg) using a T1-weighted pulse sequence on a 3T MRI scanner. Diseased aortic wall could be distinguished from normal wall based on wall-to-muscle contrast-to-noise values following GdF administration. RAM-11 (macrophages) and CD-31 (endothelial cells) immunostaining of MR-matched histological sections revealed that GdF accumulation was related to the degree of inflammation at the surface of plaques and the extent of core neovascularization. Importantly, an MR measure of GdF accumulation at both 1 and 24 hours post-injection, but not Gd-DTPA at peak enhancement, was shown to correlate with a quantitative histological morphology index related to these two plaque features.
GdF-enhanced MRI of atherosclerotic plaques allows non-invasive quantitative information about plaque composition to be acquired at multiple time points post-injection (within 1 and up to 24 hours post-injection). This dramatically widens the imaging window for assessing plaque stability that is currently attainable with clinically approved MR agents, therefore opening the possibility of whole-body (including coronary) detection of unstable plaques in the future and potentially improved mitigation of cataclysmic cardiovascular events.
Atherosclerosis is the underlying cause of life-threatening cardiovascular events such as myocardial infarction, and represents the leading cause of morbidity and mortality 1. It is characterized by the formation of inflammatory plaques that occupy the intimal layer of arterial walls. Acute plaque rupture leads to the formation of thrombi that disrupt blood flow to downstream tissues. The current clinical standard for identifying atherosclerotic arteries is x-ray angiography, which assesses the degree of luminal narrowing. However, it is well recognized that plaque composition has more influence on their propensity to rupture than does the degree of luminal narrowing they cause 2–4. Hence, new imaging techniques reporting on plaque composition are urgently needed, which would improve stratification of patients at risk for future thromboembolic events and allow monitoring of anti-atherosclerotic treatment.
Two classic morphological features of atherosclerotic plaque stability are the lipid-rich necrotic core and the collagen-rich fibrous cap 5. In addition, the number of activated macrophages that have infiltrated the cap directly influences plaque stability 6–8. Plaque stability is also influenced largely by the amount of intraplaque neovascularization 9, 10. Neovascularization provides a new pathway for blood constituents to enter the plaque including macrophages, lipids and red blood cells 11–13. Due to the roles that fibrocellular tissue composition and neovascularization have on plaque stability, an imaging technique that is capable of quantitatively evaluating these two components would be of great clinical value.
Magnetic resonance imaging (MRI) is a leading candidate modality for staging plaque stability due to its ability to directly image the vessel wall at high resolutions. In particular, contrast-enhanced MRI using extracellular gadolinium (Gd)-based contrast agents has shown promise for highlighting specific plaque components, allowing determination of the thickness of the fibrous cap, the plaque’s fibrocellular tissue composition, the size of the lipid core, and the degree of plaque neovascularization 14–17. Two drawbacks of these conventional agents for plaque imaging is the short timeframe in which these plaque features can be assessed and the necessity to image single plaques over time during the agent’s rapid wash-in phase to assess neovascular volume 16. An agent that can widen this imaging window and report on the composition of multiple plaques, in multiple vascular beds would be much more beneficial for diagnostic and therapeutic purposes.
One particular agent of interest that has been tested primarily in different animal models of atherosclerosis has been Gadofluorine-M (GdF). GdF first gained attention when it was shown to preferentially accumulate in diseased wall versus normal wall, and allowed detection of plaques that were not detectable using non-contrast MRI 18. Further studies showed a correlation of the degree of GdF enhancement at 24 hours post-injection with the amount of lipid present within plaques 19, and the ability to track plaque progression 20. Finally, GdF has also been shown to have a strong affinity for fibrous (tenascin, proteoglycans and collagen), not lipidic, components of plaques, suggesting that GdF-enhanced high-resolution MRI may allow better evaluation of plaque composition 21.
The aims of this study were to understand the potential mechanisms by which GdF accumulates into plaques, relate these mechanisms to the ability of GdF-enhanced MRI to report on plaque stability, and compare the MR assessment of stability with GdF to a conventional Gd-based agent, Gd-DTPA. This highlighted advantages for the use of GdF for non-invasive assessment of plaque stability over conventional contrast agents, which may allow whole-body assessment of plaque vulnerability in the future.
A total of fourteen male New Zealand white rabbits were used. Eight rabbits were fed 100 g/day of cholesterol (CH)-supplemented rabbit chow for 27–32 months and the CH level was titrated between 0.125 and 0.25% (w/w) to promote aortic atherosclerotic plaque formation as previously described 22–24. Six age-matched rabbits were used as controls and fed normal chow. Twelve of the fourteen rabbits were used in MRI experiments, whereas one control and one CH-fed rabbit were sacrificed without contrast media injection. Animals were cared for in accordance with guidelines of the Canadian Council on Animal Care.
Gadofluorine-M (GdF, Bayer Schering Pharma, Berlin, Germany) is an amphiphilic, macrocyclic, gadolinium-containing complex (1528 g/mol). It is a derivative of Gd-DO3A containing a perfluorooctyl side chain and a mannose moiety. GdF has an r1 relaxivity of 17.5·mM−1s−1 in blood (1.5T and 37°C) 25 and a blood half-life of approximately 10 hours in rabbits 18. Carbocyanine-labelled GdF (cc-GdF) is a formulation where mannose has been replaced by carbocyanine (2% of total GdF content is fluorescently tagged) 21. Gd-DTPA (Magnevist®, Bayer Schering Pharma, Berlin, Germany) has an r1 relaxivity of approximately 3.9 mM−1s−1 in human plasma (1.5T and 37°C) 26 and a plasma half-life of approximately 28 minutes 18.
Anesthetized rabbits were imaged in the supine position using a clinical 3T MRI scanner (GE Signa HD 12x, GE Healthcare, Waukesha, WI) interfaced with a two-channel phased array surface RF coil 24. Each rabbit was scanned sequentially with Gd-DTPA and cc-GdF (or GdF in 2 control rabbits) using a standardized MRI protocol. Please see online supplemental (OS) text for technical specifications of anesthesia and MRI. Briefly, high-resolution (0.156 × 0.156 × 3 mm3) images were collected pre- and during the first 22 minutes following Gd-DTPA injection (0.2 mmol/kg) or pre- and at 1 and 24 hours following GdF administration (0.1 mmol/kg). A T1-weighted (T1w) quadruple inversion recovery fast-spin-echo (QIR-FSE) sequence developed previously for quantitative contrast-enhanced imaging was used 27.
Animals were sacrificed after imaging with an intravenous injection of ketamine (200 mg) and transcardially-perfused under pressure with ~1.5 L of heparinized (1 IU/ml) Hank’s balanced salt solution. One control and one CH-fed rabbit without contrast agent injection were also sacrificed. Following sacrifice, while the animal was in the supine position (similar to MRI) the imaged aortic segments (2, 3 and 4 cm superior to the celiac bifurcation) were carefully isolated, marked on the ventral surface with Evan’s blue dye for matching to MRI, dissected and frozen in OCT. Fresh-frozen sections of each segment were collected and cc-GdF distribution within each section was imaged. Sections were then stained for various plaque components including collagen (picrosirius red), lipid (Oil Red O), macrophages (RAM-11 immunostaining), and endothelial cells (CD-31 immunostaining). Please see OS text for staining procedures. All microscopy was performed using a Zeiss Axioplan 2ie microscope (Carl Zeiss Canada, Toronto, ON).
MR images were analyzed with OsiriX DICOM reader (version 2.7.5, Geneva, Switzerland). The inner and outer vessel wall boundaries were traced to determine average and total wall signal intensity (SIave-wall and SItotal-wall, respectively). Regions of interest were placed in both the paraspinal muscle adjacent to the aorta and in a motion-free region outside the animal (air) to determine average muscle signal intensity (SIave-muscle) and the standard deviation of the noise signal (σAir), respectively. Contrast-to-noise ratio (CNRouter) between the vessel wall and adjacent muscle (CNRouter = (SIave-wall-SIave-muscle)/σAir) was calculated. Finally, the change in total wall SI (ΔSItotal-wall) between various time points post-contrast and baseline values was calculated (ΔSItotal-wall= SItotal-wall-post-contrast − SItotal-wall-baseline).
For histology, collagen stained (picrosirius red) sections were analyzed for the area of loose connective tissue (LCT; AreaLCT) at the surface of the plaque (RAM-11 staining confirmed that these regions were also highly inflamed). Additionally, sections stained for endothelial cells (CD-31) were analyzed for the extent of neovascularization (NV) in the core (AreaNV). AreaNV was defined as the total area circumscribing neovascularized regions within the core (RAM-11 staining confirmed that these areas were also highly inflamed). From this a histological morphology index (HMI) was calculated (HMI = AreaLCT + AreaNV) to give a quantitative measure of the combined extent of these two features of plaque instability.
For comparisons of MR data between control and CH-fed animals where images of the same plaque were collected at multiple time points per animal a two-way repeated measures ANOVA was performed. For this analysis, measurements from multiple plaques in a single animal were averaged to generate a mean MR measurement per animal at each time point. For all other comparisons of data either a one-way ANOVA followed by a post-hoc Tukey’s multiple comparison test or a two-tailed t-test was performed. Univariate Pearson correlational analysis was performed between MRI values (ΔSItotal-wall) and histological index (HMI) using data from individual plaques. Multivariate regression analysis was performed to assess the effects of both HMI and rabbit identity (multiple plaques imaged per rabbit) on MRI values (ΔSItotal-wall). The nominal level of significance for all tests was p<0.05. Both GraphPad Prism 4.0a (GraphPad Software Inc., San Diego, CA) and Statistical Analysis Software (SAS Institute Inc. Cary, NC) were used for statistical analysis.
To quantitate the enhancement of normal and diseased vessel wall following either GdF or Gd-DTPA administration we performed high-resolution (0.156 × 0.156 × 3 mm3) black-blood imaging pre-contrast (baseline) and at various time points post-contrast (first 22 min. for Gd-DTPA; 1 and 24 hours for GdF) (Fig. 1). As shown, blood nulling was successful both before and after administration of either contrast agent, allowing good delineation of inner wall boundaries at all time points. Qualitatively, diseased wall enhanced at both 1 hour and 24 hours post GdF, however the patterns of enhancement appeared different between the two time points (Fig. 1a). At 1 hour, a trilayered structure (bright inner surface, dark core, bright outer surface) was more apparent than at 24 hours. The more obvious bright inner surface at 1 hour versus 24 hours is mostly likely due to the gradual accumulation of GdF deeper within the plaque as time passes. Similar results showing deeper penetration of GdF at 6 hours compared to 0.5 and 2 hours post-administration have been reported 21. The more obvious bright outer surface at 1 hour versus 24 hours may be explained by significant amounts of GdF circulating through neovessels present in both the media and core of the plaque at 1 hour, whereas at 24 hours GdF blood levels are significantly reduced (blood half-life of ~10 hours). As expected diseased wall enhanced with Gd-DTPA, however the enhancement appeared more homogenous within the vessel wall (Fig. 1b). Surprisingly, normal wall also appeared to enhance at both 1 and 24 hours post-GdF (Fig. 1c), contrary to what has been previously reported 19. Enhancement of normal wall was also apparent after Gd-DTPA administration (Fig. 1d).
To measure the ability of either GdF or Gd-DTPA to improve the conspicuity of diseased or normal wall from surrounding muscle the wall-to-muscle contrast-to-noise ratio (CNRouter) was determined (Fig. 2). CNRouter values were significantly increased following GdF administration for normal wall at both 1 and 24 hours, whereas diseased wall conspicuity was significantly increased at 24 hours only (Fig. 2a). Furthermore, diseased wall could be clearly distinguished from normal wall based on the highly positive CNRouter values at 24 hours post-GdF administration versus the negative values for normal wall at all time points, consistent with previous findings at 48 hours 18. Similarly, Gd-DTPA administration improved the conspicuity of both normal and diseased wall shortly after administration (Fig. 2b).
Next, the change in total SI (ΔSItotal-wall; reflecting total contrast agent accumulation) within vessel wall after contrast agent administration was determined (Fig. 3). At 1 hour, a trend towards increased GdF accumulation was seen within diseased wall compared to normal wall (Fig. 3a). At 24 hours, significantly more GdF accumulated within diseased wall than normal wall (Fig. 3a). Finally, a trend for greater GdF accumulation in normal wall at 1 versus 24 hours was seen, whereas this trend was not seen in diseased wall (Fig. 3a), suggesting GdF is retained better in diseased wall than normal wall. As expected for Gd-DTPA use, diseased wall accumulated more contrast agent than normal wall (Fig. 3b).
To detect the distribution of cc-GdF within diseased and normal wall we first maximized our ability to detect cc-GdF fluorescence. Using normal and diseased vessel wall from animals not injected with contrast agent we adjusted our microscope acquisition parameters to a point where autofluorescence from the aortic tissue was just nulled (Fig. S1 in online supplemental (OS) text). By doing this, cc-GdF was visualized in both diseased and normal vessel wall from animals injected with the agent 24 hours prior to sacrifice (Fig. S1), confirming our MR observations that both wall types enhanced using GdF. It was also apparent that the fluorescent signal was stronger in diseased wall also confirming the MR results that diseased wall accumulates more GdF than normal wall (Fig. 3a; significantly higher ΔSItotal-wall).
Similar to previous findings 21, imaging of cc-GdF revealed that within the majority of plaques (15 out of 20 plaques) cc-GdF accumulation was restricted to near the luminal surface of the plaque (Fig. 4a). The highest density of GdF was present in regions composed of dense, organized collagen bundles, consistent with the concept that GdF binds to fibrotic plaque components 21 (Fig. 4b). However, interestingly, the penetration and amount of GdF in these regions appeared to be dictated by the degree of macrophage infiltrate present within the cap (Fig. 5). Areas with minimal macrophages were often composed solely of dense, organized collagen bundles and this appeared to limit the amount of GdF that penetrated into the plaque (Fig. 5a). The presumed mechanism of this is limited plaque permeability and a high number of binding partners directly at the surface of the plaque. However, as the number of macrophages at the surface increased the amount of GdF getting into the plaque also increased, presumably due to increased permeability and decreased fibrotic binding partners directly at the surface (Fig. 5b and c). Finally, whereas the majority of GdF was found to be associated with collagen, in contrast to previous findings which found paracellular GdF accumulation 21, it was also clearly apparent that a small pool of GdF accumulated within macrophage-derived foam cells present at the plaque surface (Fig. S2).
While in most plaques cc-GdF accumulation was restricted to near the plaque surface, 5 of the 20 plaques showed evidence of deep accumulation within the core. Similar to the surface, core cc-GdF accumulation (Fig. 6a) was found in areas rich in lipid-laden macrophages (Fig. 6 b/c) and with less collagenous material present (Fig. 6d). However, for intravenously injected GdF to accumulate into these areas we assumed an additional route (not directly from aorta lumen) of entry into the plaque core would need to be present. Hence we hypothesized that these areas would also be rich in intraplaque neovessels. As shown, this hypothesis held true as positive core CD-31 immunostaining was found only in the locations of those 5 plaques where core GdF was found (Fig. 6e).
As the amount of GdF accumulating within the plaque appeared to be determined by both the plaque’s fibrocellular tissue composition and the amount of neovascularization within the core we correlated the ΔSItotal-wall values obtained from the MR data to a histological measure of the additive extent of these two plaque features, referred to as the plaque’s histological morphology index (HMI). Scatter plots of HMI values versus ΔSItotal-wall values at 1 and 24 post GdF administration or 22 minutes post Gd-DTPA administration are shown in Figure 7a and 7b, respectively. Univariate correlational analysis revealed significant positive correlations between HMI and ΔSItotal-wall values at both 1 (r-value=0.8458, p<0.001) and 24 hours (r-value=0.7962, p<0.0001) following GdF administration (Table 1). In contrast, ΔSItotal-wall from a single time point during the peak enhancement phase following Gd-DTPA administration showed a weaker, non-significant correlation (r-value=0.5335, p=0.0605) (Fig. 7b). Finally, multivariate regression analysis was performed to investigate any effects of rabbit identity on the variability in the MR data since multiple plaques were imaged per rabbit (Table 1). A significant effect (p=0.02) of rabbit identity was noted for MR data collected at 24 hours post GdF administration but this accounted for only 14.6% of the variability (Table 1). No effect of rabbit identity was seen for the MR data collected at 1-hour post GdF administration or 22 minutes post Gd-DTPA administration.
Non-invasive measures of atherosclerotic plaque composition in multiple vascular beds will dramatically improve the ability to identify patients at risk of life-threatening thromboembolic events and will aid in directing appropriate, patient-specific therapies. Gd-based MR contrast agents have been shown to be previously useful for aiding in determination of plaque composition 14–17. However, this typically is limited to one vascular bed (or even one MR slice) due to the rapid wash-in/out plaque kinetics of currently approved agents. In this study we sought to determine the ability of GdF, an agent with a long plaque half-life, to report on plaque composition at multiple time points in a rabbit model of atherosclerosis. We present evidence showing that GdF administration improved the conspicuity of diseased vessel wall to muscle 24 hours following administration and that GdF accumulated more in diseased wall than normal wall. Importantly, we found that total GdF accumulation within diseased wall as measured by MRI (ΔSItotal-wall) at both 1 and 24 hours positively correlated with a histological index of morphological features implicated in plaque instability, namely the plaque’s fibrocellular tissue composition and extent of intraplaque neovascularization. In contrast, accumulation within the same plaques during the peak enhancement phase using the conventional agent Gd-DTPA did not correlate with this measure. This suggests an advantage for using GdF over currently approved agents for quantifying these two plaque features as this agent can be used within a short time period after administration (currently an hour) but also widens the imaging window (up to 24 hours) potentially allowing multiple vascular beds to be examined.
The two primary features of advanced atherosclerotic plaques are the formation of a lipid-rich necrotic core with an overlying fibrous cap 5. The composition of the cap and core is intrinsically linked to plaque stability. This includes the cap’s fibrocellular tissue composition and the degree of neovascularization in the core 6, 11–13, 15, 28. Strong evidence suggests that plaques with a large number of macrophages at the surface and significant neovascularization are unstable and susceptible to rupture 6, 8, 11–13, 28. Hence, an imaging agent that accumulates according to the plaque’s fibrocellular tissue composition and degree of neovascularization, such as GdF, would allow the staging of advanced plaque development and help identify plaques that are destabilizing at an accelerated rate.
As mentioned, Gd-based extracellular (EC) contrast agents have shown promise for identifying plaques with altered fibrocellular density and plaque neovascularization 14–16, 29. While these are significant advances in in vivo imaging of plaque composition, one potential drawback of using conventional agents is that the imaging window after injection of the agent is short (typically first 20 minutes). This will limit the use of these agents for imaging multiple plaques in a single vascular bed or multiple vascular beds in a single patient (or animal). Since atherosclerosis is considered a systemic disease affecting multiple vascular beds the use of an imaging agent capable of extending this imaging window would be of great benefit for clinical assessment of plaque vulnerability on a whole-body basis. GdF may be such an agent since it is retained in plaques for significant periods of time and its diffusion within the plaque is directly related to the plaque’s composition. Another added benefit of using GdF is its higher r1 relaxivity compared to conventional EC agents (17.4 mM−1s−1 vs ~4 mM−1s−1 at 37°C and 1.5T, respectively) 19, 25 and therefore should improve detection of these plaque features at similar doses, or may alternatively be used at lower doses.
Several groups have theorized about how GdF accumulates within plaques. One study showed that the amount of lipid within plaques correlated with the degree of enhancement of the plaques, and it was theorized that GdF binds to lipidic components of plaques via interactions with its hydrophobic tail 19. This theory was tested in another study and it was shown that GdF enters plaques bound to albumin (hydrophobic interaction) and once in plaques it has a high affinity for other hydrophobic partners such as tenascin, proteoglycans and collagen found in more fibrous, not lipidic, regions of plaques 21. These results suggest that GdF should accumulate in highly fibrotic regions of plaques. Clearly a discrepancy exists between the MR imaging results of the first study and the biochemical analysis of GdF affinity in the second study. Our study provides new evidence that helps to resolve this discrepancy. We have shown that while GdF does preferentially bind to collagenous (fibrous) material within plaques, consistent with the Meding study 21, it also appears to accumulate more and penetrate deeper into plaques that contain a large number of lipid-rich macrophages (foam cells), consistent with the Sirol study 19.
Two results in this study were in striking contrast to previously published results using GdF in atherosclerosis models. First, we found evidence of macrophage uptake of GdF, whereas cc-GdF was previously found to paralocalize to macrophages within plaques 21. We believe this result is not unexpected for 3 reasons: 1) monocytes spontaneously phagocytose GdF in vitro 30; 2) GdF enhancement of atherosclerotic wall can last for up to 2 months after a single injection, which may be due to GdF being taken up by phagocytes 20; and 3) recent evidence has shown GdF colocalization with macrophages in optical nerves in rats with experimental autoimmune encephalomyelitis (EAE) rats 31. Our second contrasting result to previous studies18, 19 is the significant enhancement of normal vessel wall using GdF. However, we are confident with our result since we also confirmed the presence of GdF within normal wall using both epifluorescent and confocal (not shown) microscopy. We believe that the discrepancy between our finding and previous literature is partially due to the fact that these previous studies used significantly lower resolution MR imaging protocols than ours, leading to partial voluming effects that likely limited their ability to detect the substantial accumulation of GdF in the very thin normal wall.
In summary we provide novel evidence for the use of the MR contrast agent GdF for non-invasive staging of plaque stability in a rabbit model of atherosclerosis. This agent allowed the combined effects of multiple plaque features related to plaque stability to be assessed simultaneously and is capable of doing so at multiple time points following administration, lessening the imaging timing restrictions typically required with currently approved Gd-based agents. We believe this agent could be useful for testing of the efficacy of current (eg. statins) or novel therapeutics (eg. anti-angiogenic or anti-inflammatory) aimed at affecting plaque stability in individual animals (or patients) over time. Finally, this agent would also allow plaques responsible for thromboembolic events, so-called ruptured or “culprit” plaques, to be localized non-invasively prior to these life-threatening events and guide clinicians to make appropriate medical or surgical decisions that could save lives.
We thank Dr. Andrew Alejski for assistance with this work.
This work was funded in part by the NIH R01-HL078641 and Canadian Institutes of Health Research-Heart and Stroke Foundation of Canada (CIHR-HSFC) CMI-72324 to BKR. BKR holds the Barnett-Ivey Heart and Stroke Foundation of Ontario Research Chair. JAR holds the Great-West Life doctoral research award from the HSFC.
Journal Subject Codes: Atherosclerosis:  Imaging; Imaging of the Brain and Arteries:  Computerized Tomography and Magnetic Resonance Imaging: Diagnostic Testing:  CT and MRI
Dr. B. Misselwitz: As an employee of the contrast media company Bayer Schering Pharma AG (Berlin, Germany), I have financial interest in the products under investigation or subject matter discussed in the manuscript. I authorize the editor to publish this financial disclosure with the article if deemed appropriate.