Quantitative assessment of angioplasty-induced vascular inflammation with ¹⁹F cardiovascular magnetic resonance imaging

The experiments were performed in eight adult Aachen minipigs [13] with an mean age of 2 years (± 5 months) a mean thorax girth of 110 ± 35 cm with a body weight of 72 ± 15 kg, which were bred, housed and extensively characterized [12] at the central animal facility center of Heinrich-Heine-University, Düsseldorf, Germany. All study protocols were carried out in accordance to the national guidelines of animal care and approved by the state authority ‘Landesamt für Natur-, Umwelt- und Verbraucherschutz’ (84-02.04.2018.A154 und 81-02.04.2019.A379). Eight heparinized pigs underwent graded carotid artery injury. Four pigs were subjected to 15 min of oversized balloon angioplasty only (Passeo-35, Biotronik, Berlin, Germany) with a balloon to artery ratio of 1.3:1. According to recent evidence, this injury leads to only subclinical vascular changes with an area stenosis of 10–20% after 4 weeks (BA, n = 4) [14]. The other four pigs underwent a combination of oversized balloon angioplasty and endothelial denudation using a Fogarty catheter. According to recent evidence, this injury leads a significant further decrease in lumen area and increased plaque size (BA + ECDN, n = 4) [15]. To this end, a 10 mm Fogarty over-the-wire embolectomy catheter (LeMaitre Vascular Inc., Burlington, MA, USA) was placed in one of the carotid arteries about 5 cm above the carotid artery trunk. The Fogarty balloon was then fully inflated and endothelial denudation was performed by pulling the inflated balloon down to the carotid artery trunk and repeating this procedure for a total of five times. The individual contralateral side served as internal control for each individual pig. Models were modified according to previous protocols [14, 15] and described in detail in the data supplement.

Perfluorooctyl bromide-nanoemulsion (PFOB-NE) was produced according to established protocols [16]. In detail, 18.4 g purified egg lecithin (E 80 S, 4% wt/wt, Lipoid GmbH, Ludwigshafen, Germany) was dispersed in 285 g phosphate buffer (10 mM, pH 7.4) with 2.5% (v/v) glycerol by magnetic stirring at room temperature. Then 322 g PFOB (AtoChem, Puteaux, France) was added. Emulsions were stabilized by adding a semifluorinated alkane, which is a mixed fluorocarbon/hydrocarbon diblock compound (C₆F₁₃C₁₀H₂₁, F6H10; AtoChem, Puteaux, France) equimolar to the E 80 S lipid. Afterwards, the dispersion was pretreated with a high-performance disperser (T18 basic ULTRA TURRAX, IKA Werke GmbH & CO. KG, Staufen, Germany) at 14,000 rpm for 2 min. This pre-emulsion was further emulsified by high-pressure/shear homogenization (1000 bar, 30 min) using a microfluidizer (Microfluidizer M‐110P, Microfluidics Corp., Newton, MA, USA). Particle size was determined using photon correlation spectroscopy on a Zetatrac (Betatek, Toronto, Canada) device. Afterwards the PFOB-NE was autoclaved (30 min at 121 °C) using a program to autoclave pure liquids and stored at 4 °C for maximal 12 weeks until use. In addition, the emulsion was checked for the presence of particles ≥ 5 µm by light microscopy and each charge was tested for bacterial contamination.

PFOB-NE was administered intravenously at body weight adjusted dose of 5 ml/kg at day 3 after vascular injury according to the maximum circulating monocyte count and protocols adopted from experiments in myocardial infarction [12]. To this end, pigs were anaesthetized again as described above. PFOB-NE was applicated via a cannula placed in a superficial ear vein with an infusion rate of 100 ml/h. The mean infusion time was 4 h ± 31 min.

During infusion and in the course (24 h) following infusion, animals were monitored for potential side effects. Adverse reactions were defined as tachycardia, tachypnea, vomitus, rushing or abnormal behavior. Severe adverse reactions were defined as respiratory insufficiency with the need for re-intubation or death due to any cause.

Before and seven days after vascular injury pigs were subjected to invasive assessment of carotid arteries. A detailed experimental flow chart and protocol is given in Fig. 1 and the data supplement. Briefly, carotid artery blood flow and vessel diameter were evaluated at baseline and 6 days after vascular injury by invasive angiography and intravascular ultrasound (IVUS). Serial blood samplings at day 1, 3 and 6 were obtained to investigate circulating inflammatory response to injury.

Immediately after vascular assessment on day 6, non-invasive ¹H- and ¹⁹F MRI was performed using a whole-body 3.0 T Achieva X-series scanner (Philips Healthcare, Best, the Netherlands). Anesthesia was maintained with a mixture of isoflurane (< 2.0% (v/v), Piramal Critical Care Deutschland GmbH, Hallbergmoos, Germany) in 100% oxygen. The MRI scan consisted of a ¹H protocol including 3D time of flight (TOF) angiography, phase-contrast velocity encoded (VENC) measurements and high-resolution T2-weighted black blood sequences for vessel wall assessment. TOF angiography was performed using the following parameters: repetition time (TR) = 21.84 ms, echo time (TE) = 4.33 ms, flip angle (FA) = 16°. Furthermore, segmented gradient-echo phase contrast MRI was performed at the proximal, middle, and distal parts of the carotid artery (TR = 8.87 ms; TE = 5.22 ms; FA = 10°). The velocity encoding range was set at 150 cm/s in a through-plane direction. For vessel wall assessment a high-resolution T2-weighted black blood turbo spin echo sequences (TR = 1558 ms; TE = 60 ms; FA = 90°) was used. This was followed by a ¹⁹F protocol for monocyte/macrophage imaging with tested and optimized sequences [11]. After acquisition of ¹H reference scans, the pigs were removed from the bore, and the ¹⁹F coil (7 × 12 cm dual tunable ¹H/¹⁹F ellipsoidal coil (Philips Healthcare, Best, the Netherlands) was placed on the neck directly above the injured artery. The ¹⁹F ellipsoidal coil was tuned and matched for both ¹⁹F and ¹H. ¹H and ¹⁹F transmission could be switched by a small extra loop. Pigs were repositioned for optimal dataset overlay as previously established for local signal precision [11]. ¹⁹F imaging was performed using a balanced steady state free precession (bSSFP) sequence centered at 58 ppm (3D acquisition, with a 3 × 3 × 3 mm³ isotropic voxel size; TR = 2.88 ms; TE = 1.00 ms; FA = 30°), as described in a previous applicability study [10, 11]. Due to the frequency selective excitation at 58 ppm and the low isoflurane concentration applied (< 2.0% v/v) no relevant ¹⁹F signals from isoflurane could be observed in the lung or adipose tissue. After in vivo ¹H- and ¹⁹F MRI acquisitions, pigs were euthanized with potassium chloride and Pentobarbital sodium (Narcoren^®, Boehringer Ingelheim Vetmedica, Ingelheim, Germany).

Autopsy was performed and both carotid arteries and the carotid trunk were excised in toto and stored in 4% paraformaldehyde (PFA) for at least 7 days. To optimize image conditions, but using the same technical equipment and sequences, ex vivo scans were performed at 3.0 Tesla with optimized coil distance of only a few mm and an identical isotropic image resolution for ¹H and ¹⁹F (1 × 11 × mm³) with an image acquisition time of 19 min. To precisely localize ¹⁹F signals in carotid arteries for a direct histological validation, high resolution images of total excised carotid arteries were acquired at 9.4 T using a Bruker AVANCEIII 400 MHz Wide Bore NMR spectrometer (Bruker, Rheinstetten, Germany). Here, ¹⁹F datasets were recorded using a 3D RARE sequence (RARE factor 32, TR 3.5 s, 330 × 450 µm in-plane resolution, Slice thickness 1 mm, 465 averages, Scan time 48 h). For exact anatomic localization, the ¹⁹F datasets were merged with corresponding 3D 1H RARE scans (RARE factor 16, TR 5 s, 120 × 120 µm in-plane resolution, Slice thickness 1 mm, 10 averages, Scan time 16 h).

Subsequently paraffin embedded vessels were sectioned (5 µm) and stained with hematoxylin/eosin (H/E). For immunoflorescence and immunohistology the following antibodies were used: anti-CD163 antibody (clone 2A10/11, Bio-Rad, Hercules, CA, USA) for identification of M2-like macrophages; anti-CD68 antibody (ab125212, 7.5 µg/ml, abcam, Cambridge, UK) for identification of M1-like macrophages. Tissue processing is described in detail in the data supplement. Histological examination was conducted with a Leica microscope (DM 4000 M, Leica Microsystems, Wetzlar, Germany) with 10 ×, 20 × and for detailed images 40 × objectives. H/E- or CD163/CD68-stained cells were counted semi-automatically using ImageJ (LOCI, WI, USA) with a protocol provided in the data supplement. Early Remodeling Index (RI) was calculated by using histology: RI was defined as lesion external elastic membrane cross-sectional area (EEM CSA) divided by mean reference EEM CSA.

MRI datasets were visualized and analyzed for signal overlays using HOROS (Nimble Co LLC, Annapolis, MD, USA) and Circle CVI (Circle Cardiovascular Imaging Inc., Calgary, Canada) for analysis of anatomic dimensions and velocities. For assessment of in vivo ¹⁹F datasets, exact anatomical co-localization was performed by image fusion of ¹H and ¹⁹F datasets using HOROS. For quantification of ¹⁹F signal intensity, the primary signal at a respective region was corrected according to the coil sensitivity profile: \(SNR_{Z \,direction} = 0.45 \cdot SNR_{measured} /e^{ - Distance \,in \,Z \,direction/75 mm}\) [12]. Then, SNRs were calculated from the ratio of the mean of a region of interest (ROI) and the standard deviation of the noise of a ROI outside the body [11, 12]. ¹⁹F volumes were calculated by applying background subtraction with SNR = 5 which was sufficient to subtract all unspecific technical background signals (outside the body). For validation of ¹⁹F signals with histology at least 10 segments of each excised carotid artery were analyzed for ¹⁹F SNR and histology (n = 80 segments). The regions were identified by measuring the distance from bifurcation to the respective region in cm.

As minimal detection limit, we defined the minimal number of monocytes/macrophages per mm² that correlates to a visible ¹⁹F Signal (as defined as signal-to-noise ration > 5) in every single sample. However, as there are samples with even lower monocyte/macrophage cell count leading to a ¹⁹F SNR > 5, the actual detection limit might even be lower.

The detection limit of inflammatory cells using ¹⁹F MRI at 3.0 Tesla was previously determined 400 macrophages/mm² in a model of myocardial infarction [5]. This detection limit was considered suitable for imaging vascular inflammation in models of mechanical injury and even human atherosclerotic plaques [17].

For statistical analysis of histology, each evaluated carotid artery was cut into at least 10 cross sections. Histological analysis was performed separately in all resulting sections. Afterwards, a mean value for each carotid artery was calculated. The final statistical analysis (vessel wall thickness, monocyte/macrophage cell count, remodeling index) was then carried out using one average value per carotid artery (n = 4 for BA and BA + ECDN, n = 8 for control carotid artery).

Statistical analysis was performed using GraphPad Prism (Version 9, IBM, San Diego, CA, USA). Unless otherwise stated, continuous variables are presented as mean ± standard deviation (SD). Normal distribution was tested using the Shapiro–Wilk test. Data between the two different groups (BA and BA + ECDN) were analyzed by 2-sided unpaired Student’s t-tests for normally distributed data and Mann–Whitney U-test for not normally distributed data. Data comparing three groups (Control vs. BA vs. ECDN) were analyzed using a one-way ANOVA followed by Šidák multiple comparison test. Pearson’s correlation was used to assess the relationship between different ¹⁹F signal intensity and monocyte/macrophage cell count.

Eight minipigs underwent graded carotid artery injury by either 15 min of oversized balloon angioplasty known to induce only subclinical stenosis [14] (BA, n = 4) or a combination of 15 min oversized balloon angioplasty and mechanical denudation (BA + ECDN, n = 4) using a Fogarty catheter known to cause clinical significant stenosis (Fig. 1) [15]. All animals survived the procedure. Blood sampling and angiography was performed before and after vascular injury. At day three after injury, PFOB-NE was administered intravenously. In 4 of 8 pigs, adverse reactions were observed, but no severe adverse reaction occurred (Additional file 1: Table S1). At day 6 after injury blood sampling as well as invasive angiography and IVUS was performed followed by in vivo and ex vivo ¹H- and ¹⁹F-MRI and histology. Numbers of circulating leukocyte increased mildly from baseline to day 6 after vascular injury and hemoglobin levels remained constant (Additional file 1: Table S2).

At day 6 after injury, angiographic assessment and IVUS revealed no relevant stenosis after BA. However, there was a 56% and 33% diameter stenosis with maintained blood flow in two pigs after BA + ECDN while two pigs displayed nearly total carotid artery occlusion (Fig. 2A). Carotid artery blood flow was assessed non-invasively by VENC MRI (Fig. 2B). Following BA + ECDN, a decrease in forward flow and post stenotic maximum velocity could be detected in all four pigs. Two pigs showed signs of nearly total artery occlusion with a decrease in forward flow and maximal velocity to nearly zero. Nevertheless, histological analysis of the vessel wall including thickness, remodeling index and tissue composition with inflammation was possible. While vessel wall thickness was approximately 0.8 mm after BA, BA + ECDN led to an increase to 1.4 mm at day 6 as quantified non-invasively using high-resolution T2-weighted bright blood imaging (Fig. 2C).

H/E staining was performed to quantify early vascular remodeling after graded vessel injury. Mean diameter of lumen, intima, media, and adventitia was measured in at least 10 cross-sections starting from the carotid artery bifurcation and expressed as relative distance from bifurcation (Fig. 3A and B). Adventitial diameter increased compared to contralateral control carotid arteries after BA and BA + ECDN while diameter of media and intima remained unaffected. Only after BA + ECDN a relevant reduction of luminal radius was observed. Quantification of vessel wall diameter across a range of histological sections obtained from all 8 pigs verified these findings in the individual arteries (Fig. 3C). Compared to contralateral control carotid arteries, adventitial wall thickness increased more than twofold after BA + ECDN, while media thickness was not significantly changed. Remodeling index (external elastic lamina area diseased/control) showed no significant difference after BA + ECDN compared to BA (Fig. 3D).

To assess inflammatory processes in addition to anatomical ¹H MR images, ¹⁹F images were displayed on a ‘hot iron’ color scale and fused with ¹H series to ensure exact anatomical localization of ¹⁹F signal. These analyses were performed in vivo and validated by high-field MRI ex vivo (Fig. 4).

Following BA + ECDN, a robust in vivo ¹⁹F signal with a SNR of 14.7 ± 4.8 after 19 min of image acquisition was detected in the treated carotid artery of all 4 pigs (Fig. 4). No signal was found in the contralateral control carotid artery. Vascular ¹⁹F signal was clearly distinct from other tissue signals, mainly derived from the bone marrow and subcutaneous fat. No vascular ¹⁹F signal was found in vivo after BA-treatment alone. In the ex vivo setting at 3 T, ¹⁹F signal could be detected after 20 min in all 8 pigs with a strong SNR of 23.4 ± 3.8 after BA + ECDN and weak SNR of 6.2 ± 1.2 following BA. In vivo and ex vivo imaging revealed a patchy distribution of ¹⁹F signal along the carotid artery.

To validate the patchy appearance of ¹⁹F signals and to define minimal ¹⁹F detection limits for monocytes/macrophages in vascular inflammation, we conducted ex vivo optimized ¹H- and ¹⁹F-imaging of excised carotid arteries at 3 T and high-resolution ex vivo ¹H- and ¹⁹F-imaging at 9.4 Tesla alongside histological monocytes/macrophages staining (Fig. 5A). Fusion of high resolution ¹H /¹⁹F images clearly demonstrated that PFOB-derived ¹⁹F signal stems from the vascular wall and that the ¹⁹F signal appears in a patchy fashion. Immunofluorescence staining confirmed a corresponding adventitial presence of monocytes/macrophages with predominantly CD163⁻CD68⁺ and to a lesser extent CD163⁺CD68⁻ appearance.

Semi-automated counting of monocytes/macrophages revealed a 4.6-fold increase of monocytes/macrophages in the adventitia after BA + ECDN while only a minor increase could be detected after BA (Fig. 5B). In the media, monocytes/macrophages were sparsely detected after both BA and BA + ECDN, suggesting that monocyte/macrophage infiltration early after vascular injury mainly occurs via the adventitia. The adventitial monocytes/macrophages phenotype was a mixture of CD68⁺ CD163⁻ and CD68⁻ CD163⁺ cells with predominance of CD68⁺ CD163⁻ in BA + ECDN.

For histological validation carotid arteries were cut into at least 10 segments. Co-registration of local ex vivo ¹⁹F signal at 3.0 T with segmental monocytes/macrophages cell count was performed over the entire length of the carotid arteries of every included pig (n = 8) leading to about 80 data points. This analysis of individual arteries revealing a close relationship between local ¹⁹F signal intensity as detected by MRI and spots of monocyte/macrophage accumulation (Fig. 5C). A significant correlation (R² 0.47, p < 0.001) of local ¹⁹F signal with monocyte/macrophage cell density was found when analyzing every segment of the injured carotid arteries after BA as well as BA + ECDN. The detection limit was about 410 monocytes/macrophages per mm².

Experimental approaches in mice have proven feasibility of ¹⁹F MRI to quantitatively visualize macrophage accumulation in a wide range of experimental set-ups and disease models, including vascular inflammation, [5, 6, 18]. However, those experiments were conducted with Perfluorpolyether (PFPE), which confers excellent imaging properties. Yet, a long body half-life renders it less attractive for a potential clinical application. In contrast, other PFCs (e.g., PFOB) have a shorter biological half-life and have already proven safe and efficient applicability in clinical trials for testing their oxygen transport capacities [19]. Thus, these compounds might have greater potential for clinical translation. Furthermore, ¹⁹F MRI using PFOB nanoemulsion is also feasible in pre-clinical models after myocardial infarction using clinical scanners [10‐12] with good agreement between hot spots of ¹⁹F signal and macrophage accumulation [12]. In the present study, we used PFOB for ¹⁹F MRI in an animal model with a body size and an inflammatory reaction comparable to humans to study macrophage visualisation after vascular injury. The resulting ¹⁹F signal in carotid arteries after balloon injury was highly reproducible with a robust in vivo and ex vivo SNR three or five times above the common detection threshold (SNR = 5). Accordingly, our study provides evidence that ¹⁹F MRI using PFOB-nanoemulsion could be a valuable tool to monitor vascular monocytes/macrophages accumulation after angioplasty.

The application of PFOB-NE appeared overall safe corroborating a previous study in a model of myocardial infarction that did not reveal severe adverse reaction (Additional file 1: Table S1). In line with this, the infusion of a significant larger amount of the emulsion did not lead to increased mortality in a clinical phase 3 trial [19]. In the present study we conducted a rigorous quality control of each emulsion charge with careful assessment of particle size. Moreover, the analysis of circulating leukocyte indicated no severe systemic inflammatory reaction upon PFOB infusion and no drop in hemoglobin levels (Additional file 1: Table S2).

There is accumulating evidence, that tissue macrophages are the main source of PFOB-derived ¹⁹F signals, as shown in mice [5], pigs [10‐12] and humans [7]. Untargeted perfluorocarbon nanoemulsions are taken up by classical and alternatively-activated macrophages to an equal extent [20]. Minimal detection limits were calculated to be around 400 macrophages/mm² tissue utilizing a model of myocardial infarction [5]. In the present study, we used a model of oversized balloon angioplasty with or without additional Fogarty-induced mechanical denudation (BA vs. BA + ECDN) according to established protocols [15]. A reliable and reproducible in vivo ¹⁹F signal could only be detected after BA + ECDN. In our study, the minimal detection threshold for detection of macrophages was about 410 cells/mm² corroborating previous estimation [5]. The signal was clearly restricted to the vascular wall. Although 2 animals developed a luminal thrombosis, all 4 animals of this group could be analyzed for the PFOB-derived ¹⁹F signal that was confirmed histologically to stem from macrophage infiltrates.

Untargeted perfluorocarbon nanoemulsions are taken up by classically- (M1-like) and alternatively-activated (M2-like) tissue macrophages to an equal extent [20]. Given the fact, that M1-like macrophages were present in larger numbers in our model on day 6 after angioplasty, the observed ¹⁹F-signal reflected most likely M1-like macrophages. This finding mirrors recent evidence in pigs after myocardial infarction [12].

In addition to profound macrophage infiltration, BA + ECDN-induced injury was consistently paralleled by adventitial wall thickening. In contrast, BA induced only a sparse infiltration of monocytes/macrophages in the vascular wall detected by immunohistology. This was paralleled by a mild increase in adventitial wall thickening. The signal of those inflammatory cells was only measurable under optimal coil distance conditions using ex vivo ¹⁹F scans. In both groups (BA vs. BA + ECDN), there was no significant difference in the vascular remodeling index, underlining the predominant adventitial remodeling, as this index mainly reflects changes in the intima and media.

Other studies using balloon oversize injury in pigs mainly focused on intimal hyperplasia, neointima formation and lumen area after a follow up period of 2–4 weeks [14, 15]. Nevertheless, these studies confirm subclinical early vascular remodeling after overstretch injury with minimal reduction in lumen area [14]. In contrast, balloon oversize injury followed by endothelial denudation caused significantly reduced lumen area and enhanced early vascular remodeling [15]. Therefore, mild monocyte/macrophage infiltration below the ¹⁹F detection limit may also be of minor significance for early vascular remodeling and subsequent development of re-stenosis.

There is evidence, that identification of vessel wall surrogates of inflammation with PET (glucose uptake (¹⁸F-FDG) or microcalcification (¹⁸F-NaF)) can predict restenosis after angioplasty [21]. We have shown, that ¹⁹F MRI with an untargeted nanoemulsion can directly image and quantify macrophage infiltration in vascular inflammation [4, 22]. The specificity of this approach to target classically- and alternatively-activated macrophages could even be improved by functionalizing the nanoemulsion [4, 22]. As previously demonstrated in mice, multitargeted nanoemulsions with different ¹⁹F-agents identified by multi-chemical shift-selective imaging are a powerful tool to monitor simultaneously a broad range of antigens important in the development and exacerbation of atherosclerosis (“multi-color-imaging”) [4]. Thus, the present imaging platform may provide the basic methodology for a comprehensive phenotyping of vascular healing after angioplasty with the major advantage of a positive, specific and quantifiable signal. In the present study, using an untargeted nanoemulsion, we could directly and quantitatively visualize the monocyte/macrophage lesion burden with a resolution sufficient for improved vascular-wall mapping of the signal.

Interestingly, the minimal detection limit of ¹⁹F MRI was comparable to cell density of human carotid atherosclerotic macrophage-rich vulnerable plaques [3, 23]. As shown for primary prevention, the indirect visualization of inflammatory foci by enhanced glucose uptake using ¹⁸F-fluorodeoxyglucose (¹⁸F-FDG) predicts subsequent vascular events in carotid artery atherosclerotic lesions [24, 25]. Inflammatory microcalcification as identified by ¹⁸F-sodium-fluoride (¹⁸F-NaF) can predict future cardiovascular events in the coronary arteries [26, 27]. Given the fact, that the number of monocytes/macrophages is directly related to plaque vulnerability [3, 28], quantitative imaging of monocytes/macrophages may provide additional value for identifying the patient at risk for imminent clinical complication.

Promising advances in MRI-technology such as improved coil designs and MRI sequence optimization have the potential to enhance the sensitivity of this methodology. The current timing of PFOB-NE administration and ¹⁹F MRI imaging was adopted from our protocol of acute myocardial infarction [11, 12]. Since the time course of cellular response to vascular injury may differ from that in myocardial infarction, optimizing time point and duration of PFOB-NE administration and ¹⁹F MRI imaging might even increase sensitivity of the method. In addition, the PFOB-load of the individual nano-droplet could be increased. Furthermore, human monocytes are even more capable of PFOB phagocytosis than those from mice or pigs [7, 10, 12] and remain functional thereafter. Finally, targeted nanoemulsions against specific epitopes can be PEGylated, thus rendering them more potent to pass the bone marrow [4]. Accordingly, they may be longer available to the target cells, thus leading to an additional signal increase. Future studies should investigate the dynamic signal flux that is expected by the dynamic monocyte/macrophage turnover.

The present study only evaluated acute vascular inflammation. Due to the validating nature of the present study, there were no sequential scans which would have possibly given insight in the signal dynamics over time. We cannot make a conclusive statement regarding the significance of ¹⁹F signal intensity for long-term vascular remodeling or chronic inflammatory processes in atherosclerosis-driven peripheral artery disease. Further long-term studies with graded vascular injury and large animal dietary or genetic models of atherosclerosis are needed to investigate vascular remodeling and plaque progression in relation to ¹⁹F signal intensity.