Introduction
Abdominal aortic aneurysm (AAA) is a focal dilatation of the abdominal aorta that progresses to rupture, which confers high mortality.
1 AAA is detected incidentally or through ultrasound scanning (USS)-based screening programmes.
2,
3 Once detected, USS monitoring is used to track AAA diameter until it reaches the 5.5-cm intervention threshold; this can take > 10 years and places a discernible burden on patients.
4,
5 A non-invasive stratification biomarker applied at the time of AAA detection to personalize monitoring regimens and identify those who may benefit from a specific medical therapy or early intervention would be beneficial. Positron emission tomography (PET) can provide information about the molecular processes beyond the anatomical characteristics of USS or computed tomography (CT) and may therefore be useful for this purpose.
Fluorine-18 fluorodeoxyglucose ([
18F]FDG) is a glucose uptake marker commonly used in oncology as an indicator of high metabolic activity, such as at sites of inflammation, and is the most commonly administered PET radiotracer in clinical practice. The data on [
18F]FDG uptake in AAA are mixed and conflicting.
6-
9 Some studies suggest that increased [
18F]FDG uptake positively correlates with AAA progression,
10-
12, while others indicate an inverse relationship or lack of correlation between [
18F]FDG uptake and aortic size.
13,
14 The Sodium Fluoride Imaging of AAA (SoFIA3) trial recently demonstrated that fluorine-18 sodium fluoride AAA uptake (reflecting microcalcification) predicts AAA progression and rupture, providing the first proof-of-concept data for PET/CT as a AAA stratification biomarker
15; however, the mechanism remains unclear.
Maegdefessel et al. showed that modulating microRNA-21 expression reduces cell proliferation, protecting mice from AAA expansion.
16 Smooth muscle cell de-differentiation and proliferation is an early event in AAA.
17 Preventing this event by deleting Kruppel-like factor-4 prevents AAA formation in mice.
18 These data point to a period of pathological cellular remodeling in early AAA and suggest that anti-proliferative therapy might be beneficial. Therefore, detecting cell proliferation in vivo may be useful. Our hypothesis is that this proliferative remodeling is detectable using PET/CT with an analogue of the pyrimidine deoxynucleoside thymidine, fluorine-18 fluorothymidine ([
18F]FLT). This tracer accumulates in proliferating cells and reflects thymidine kinase-1 (TK-1) activity. Here, we investigate [
18F]FLT as a PET/CT radiotracer in the progressive angiotensin II (AngII) infusion pre-clinical model of AAA.
Methods
Animals
All animal work was conducted in accordance with the UK Home Office, Animals (Scientific Procedures) Act 1986 under Project Licence P606320FB. Male JaxTM apolipoprotein E-knockout mice (ApoE−/−; B6.129P2-Apoetm1Unc/J; Charles River, UK) were used for experiments at 14 weeks of age. Male C57BL6/J mice (Charles River, UK) were used at 8 weeks of age for baseline biodistribution studies. All animals were part of the Jackson Laboratories Genetic Stability Programme to limit cumulative genetic drift. Mice were maintained at 21 °C with a 12-hour light/dark cycle and 50% to 70% humidity in GM500 individually ventilated cages (Techniplast, Italy) and fed standard chow diet and Hydropac water ad libitum. All mice were provided with a housing dome and two chew sticks as environmental enrichment.
For the AngII AAA model, 14-week-old male ApoE
−/− mice (Jackson Laboratories) received human AngII (Sigma A9525) infusions at 750 ng·kg
−1·min
−1 via Alzet
® 1002 (14 days) or 1004 (28 days) osmotic mini-pumps that were implanted subcutaneously by posterior neck incision under recovery isoflurane anaesthesia, as described previously.
19 ApoE
−/− controls received saline infusions via identical pumps.
[18F]FLT Production
[18F]FLT was prepared from fluorine-18 fluoride produced using an on-site 7.5-MeV ABT Biomarker Generator cyclotron and purified using an in-house-developed microfluidic electrochemical cell for electrode trapping, through which irradiated target water (0.5 mL, oxygen-18 H2O) containing ca. 1 GBq (27.03 mCi) of fluorine-18 fluoride was pumped at 0.2 mL·min−1 while applying a 20 V potential, and the cell was flushed with 2 mL of MeCN at 1 mL·min−1 with no potential applied. Subsequently, 0.4 mL of a solution containing KHCO3 (30 mM) and K222 (37 mM) in MeCN was pumped through the cell at 40°C at 0.1 mL·min−1. Ten mg of fluorothymidine precursor 3′-N-Boc-5′-dimethoxytrityl-3′-O-nosyl-thymidine was added to the released solution to perform a radio-labeling reaction at 100 °C for 10 minutes. After the fluorination, the unreacted fluorine-18 fluoride was trapped on the neutral alumina cartridge (light) and reaction mixture treated for 5 minutes at room temperature with an equivalent volume of 2 N HCl solution. The mixture was then neutralized with a stoichiometric amount of 8 N NaOH solution and purified by semi-preparative high-performance liquid chromatography on an ACE 5 C18 10 × 250 5A column eluted with 35% acetonitrile in water (both 0.1% trifluoroacetic acid) (flow rate= 4.7 mL·min−1, Rt= 12 minutes). The high-performance liquid chromatography fraction containing partially protected product was diluted two to three-fold with water and passed over an HBL Oasis C18 cartridge. Trapped product was eluted with 0.5 mL of ethanol and 3 mL of diethyl ether and dried at 60 °C under an inert gas stream. The heat applied for drying generated the fully deprotected [18F]FLT (confirmed by analytical high-performance liquid chromatography with a cold fluorine-19 FLT standard), which was re-dissolved in 10% ethanol/phosphate-buffered saline solution, filtered through a 0.22-µm filter for sterility, and delivered for animal administration. Tracer preparation started with 1.0 to 1.3 GBq (27.03 to 35.14 mCi) of cyclotron-produced fluorine-18 fluoride and yielded 40 ± 8 MBq (1.08 ± 0.22 mCi) (n.d.c.) [18F]FLT intravenous-injectable formulation in 136 ± 15 minutes (n = 5) (RCY (decay-corrected)= 8% ± 2%).
In Vivo PET/CT
Mice were induced with 5% isoflurane/oxygen (v/v) anaesthesia before maintenance at 2% at 1 L·min−1. Mice were cannulated in the tail vein using a bespoke catheter before being placed into an imaging cell where temperature and respiration were monitored (Minerve, France). [18F]FLT was injected with the following means ± standard deviations (SDs) of activity in 200 µL of 0.9% saline solution (Aqupharm No1, Animalcare Ltd., York, UK) through the pre-cannulated lateral tail vein at the beginning of a 90-minute dynamic imaging sequence: 7.27 ± 2.89 MBq (0.20 ± 0.08 mCi) (14-day scans; n = 12) and 10.08 ± 1.77 MBq (0.27 ± 0.05 mCi) (28-day scans; n = 7). Images were acquired using the Super Argus (Sedecal) small-animal PET/CT scanner, with animals placed prone and the field-of-view centered on the abdominal aorta. Mice were maintained at 1% anaesthesia during scanning, with temperature and respiration monitored throughout. Following the PET scan, a CT image was acquired for anatomic co-registration (40 kV, 140 µA, 360 projections, 8 shots). PET images were histogrammed into 15 2-seconds, 2 15-seconds, 4 60-seconds, 1 300-seconds, and 8 600-seconds frames and reconstructed using the three-dimensional ordered subsets expectation maximization algorithm with 2 iterations and 16 subsets with attenuation correction, yielding voxel dimensions of 0.39 × 0.39 × 0.78 mm3.
PET/CT Image Analysis
Animals with failed or paravenous radiotracer injection (2/5 in the saline control group, 2/5 in the 14-day AngII AAA group, and 1/4 in the 28-day AngII AAA group) were excluded from the analysis. Three-dimensional isocontour regions of interest (ROIs) incorporating the supra-renal abdominal aortic area (between the renal arteries and diaphragm) were manually drawn on the reconstructed images using AMIDE v1.0.4 software. The aortic ROI was constructed between the kidneys and anterior to the anterior border of the vertebral column based on the CT images including the most prominent PET signal in the 80 to 90-minute timeframe. For saline controls, the aortic ROI was constructed between the kidneys, superior to the bladder, and anterior to the anterior border of the vertebral column based on the CT images. These ROIs had dimensions of 1.08 mm3 and contained 15 voxels (see Supplementary Figure 1 for example visual representation of ROIs). Incorporating mouse weights and injected radioactivity doses, the output statistics revealed the maximum and mean standardized uptake values (SUVs-) across all frames. These values were used to generate time-activity curves for ROIs in the abdominal aorta (as shown in Supplementary Figure 2), spleen, kidneys, and urinary bladder. The maximum SUV (SUVmax) was chosen as the metric of choice to quantify the aortic ROI to avoid partial volume effects of the adjacent kidneys and bladder, where high [18F]FLT signals were observed consistently.
Histological Staining
Murine aortae were fixed in situ by perfuse fixation with 10 mL of phosphate-buffered saline followed by 5 mL of 4% paraformaldehyde in terminally anaesthetized animals and then fixed at 4 °C for 48 hours before they were embedded in Cellwax (Cellpath Ltd., UK) using a Leica EG1150H embedding station. Four-µm thickness sections were cut onto Plus Frost slides (Solmedia, UK) using a Leica RM2235 microtome. Proliferating cells were stained with rabbit anti-mouse Ki67 antibody (ab15580, Abcam; 1:750 dilution) for 1 hour and Menapath Polymer HRP secondary antibody (Menarini Diagnostics Ltd., Winnersh, UK) for 30 minutes. Slides were counterstained with haematoxylin and imaged using the Aperio
® AT2 (Leica Biosystems, Wetzlar, Germany) digital pathology slide scanner with × 20 maximal magnification. All quantitative analyses were performed using ImageJ 1.51k software.
20 For quantification of the Ki67 staining, the proportion of Ki67-positive nuclei was determined in three 50-µm
2 ROIs and averaged for each animal.
Protein Quantification
An independent cohort of animals was used to harvest tissue at baseline and after 28-day saline, 14-day AngII, and 28-day AngII infusions. Aortic tissue (supra-renal abdominal aortic segment only) was harvested under terminal isoflurane anaesthesia, snap-frozen in liquid nitrogen after phosphate-buffered saline perfusion, and stored at − 80 °C until use. Proteins were extracted using cell extraction buffer (FNN0011, Invitrogen) supplemented with Protease Inhibitor Cocktail (P8340, Sigma). The DC Protein assay kit (5000112, Bio-Rad) was used to quantify isolated proteins. The average yield was more than 3 mg of protein per aorta. Western blotting was performed for TK-1, a key enzyme expressed during DNA synthesis and the substrate for [
18F]FLT, and the specific carriers responsible for the transport of both [
18F]FLT and thymidine across the cell membrane: equilibrative nucleoside transporter (ENT)-1 and -2 and concentrative nucleoside transporter (CNT)-1 and -3. For each protein, blots were conducted as per the manufacturer’s suggestions using 100 µg of protein per well. Antibodies and dilutions used were anti-TK-1, 1:500 (GTX113281; GeneTex, CA, USA); ENT-1, 1:1000 (ab135756); ENT-2, 1:1000 (ab181192); CNT-1, 1:500 (ab192438); CNT-3, 1:500 (ab223085); and β-actin, 1:1000 (Ab8226; all Abcam, Cambridge, UK). Proteins were detected using SuperSignal West Femto Maximum Sensitivity enhanced chemiluminescent substrate (ThermoFisher Scientific, Loughborough, UK). ImageJ 1.51k software was used to quantify the optical intensity of the bands relative to that of corresponding β-actin bands.
20
In Vivo USS
USS was performed under 5% recovery isoflurane anaesthesia before maintenance at 2% using the Vevo2100 high-resolution (30 µm), high-frequency pre-clinical µUSS system (Visualsonics, FUJIFILM VisualSonics, Inc., Toronto, ON, Canada) with an MS-550D transducer at a 40-MHz frequency on a heated platform maintaining a core body temperature of 35 to 37 °C. Prior to imaging, the abdomen area was shaved and hair removed using depilatory cream. Imaging was performed using Aquasonic® clear gel (Parker Labs). Transverse imaging was acquired with a motor along an 11.96-mm ROI from the right renal artery in a cranial direction with 157 frames at 0.076-mm intervals gated for respiration with electrocardiographic triggering 50 ms after the r-wave. Images were reconstructed and measured using Vevo Lab v1.7.0 (VisualSonics, FUJIFILM VisualSonics, Inc., Toronto, ON, Canada) to generate three-dimensional lumen volume measurements. For the co-registration analysis, bony landmarks on CT and USS were aligned and displayed using ImageJ 1.51k.
Ex Vivo Gamma Counting
Animals were anaesthetised with 5% isoflurane with an oxygen flow rate of 2 L·min−1. [18F]FLT in 100 µL of 0.9% saline solution (Aqupharm No1, Animalcare Ltd., York, UK) was injected into the tail vein at the following doses (mean ± SD): C57BL6/J study, 6.07 ± 2.57 MBq (0.16 ± 0.07 mCi); AngII AAA study, 0.25 ± 0.32 MBq (0.01 ± 0.01 mCi). For the AngII AAA ex vivo gamma counting study, we reduced the injected dose to allow more animals to be included in a single experiment with a single delivery of [18F]FLT. This was possible due to the high sensitivity of the Hidex gamma counter. Animals were recovered for 90 minutes and subsequently humanely culled by cervical dislocation under Schedule 1 of the UK (Scientific Procedures) Act 1986. Samples (blood, plasma, heart, spleen, kidneys, small and large intestines, supra-renal abdominal aorta, bone, and tail) were collected, weighed, and measured for radioactivity using a gamma counter (Hidex). Importantly, blood was removed from the aortic lumen before it was placed in the gamma counter. All ex vivo biodistribution data were decay-corrected, and radioactivity counts were expressed as the percentage of injected dose per gram of tissue (%ID/g).
Ex Vivo Autoradiography
Following a 90-minute delay for biodistribution and gamma counting, spleen, control aorta, and 14-day AngII AAA whole-organ samples were placed on a phosphor screen (PerkinElmer) covered with clear cellophane wrap. After overnight exposure, the imaging plates were scanned using the Cyclone Plus imager (PerkinElmer) at a resolution of 300 DPI. Images were processed using OptiQuant software (PerkinElmer).
Statistical Analysis
All statistical analyses were performed using Origin 2017 software (OriginLab Corporation, Northampton, MA, USA). All graphs show the mean ± the standard error of the mean (SE) unless otherwise stated. Histology, USS, and some gamma counting results were analyzed using two-sampled t tests with Welch’s correction. PET/CT, western blotting, and some gamma counting results were analyzed using one-way analysis of variance with post hoc Bonferroni-Holm correction. The threshold for statistical significance was set at P < 0.05.
Discussion
This is the first study to explore the feasibility of [18F]FLT PET/CT in an experimental murine model of progressive AAA. We demonstrated that [18F]FLT uptake in AngII-induced AAA increased after 14 days of infusion compared to saline controls with subsequent signal fall-off after 28 days. This was corroborated by a similar pattern of TK-1 protein expression in an independent cohort of animals. We confirmed that the [18F]FLT signal was originating from the aortic tissue by performing ex vivo gamma counting of supra-renal aortic tissue (without blood) and co-registered the PET/CT and USS data. Furthermore, ENT-1, ENT-2, CNT-1, and CNT-3 levels were observed to have increased in AngII AAA.
Initially, both [
18F]FLT and thymidine are transported across the cell membrane via passive diffusion or specific carriers (ENT-1/2, CNT-1/3). The fluorine atom replacing the 3’ hydroxyl group prevents [
18F]FLT from incorporating into DNA and proceeding along the thymidine mechanistic pathway. The cytosolic accumulation of phosphorylated [
18F]FLT is detected using PET, reflecting TK-1 activity and thus proliferation.
24,
25 Phosphorylated [
18F]FLT is unable to escape from cells and acts selectively as a substrate for TK-1, but not TK-2, contributing to the tracer’s specificity.
26
[
18F]FLT has been identified as a suitable PET radiotracer to assess proliferative activity in tumors.
24,
27,
28 More recently, Ye
et al. demonstrated [
18F]FLT PET/CT uptake in atherosclerosis and a weak association between [
18F]FLT and [
18F]FDG uptake, suggesting that glucose metabolism and cell proliferation may be weakly associated in this context.
29 Our results support the idea that [
18F]FLT uptake might be useful in AAA as opposed to [
18F]FDG uptake, for which the existing data are conflicting.
Selective [
18F]FLT phosphorylation, and thus its uptake and retention in the cytosol, is limited by TK-1 during cell synthesis.
30 It is therefore important that we observed significantly increased TK-1 expression after 14-day AngII infusion, when we observed the greatest [
18F]FLT uptake on PET. TK-1 overexpression is similarly observed in cancer models,
30,
31 and its association with the Ki67 proliferation index has been demonstrated in various cancers.
32-
34 Interestingly, AAA [
18F]FLT uptake was observed after 14 days of AngII infusion, when aneurysms expand, but this effect was weaker after 28 days, when the growth phase begins to plateau, suggesting decreased cell proliferation in late-stage AAA. This supports the idea of early-stage cellular remodeling leading to late-stage replicative senescence, which has been reported in end-stage human AAA samples.
17,
35,
36 The early proliferative period may drive disease progression, but this theory requires further pre-clinical and clinical evaluation.
The nucleoside transporters that transport [
18F]FLT into cells may also influence [
18F]FLT uptake: ENT-1, ENT-2, CNT-1, and CNT-3. Paproski
et al. characterized [
18F]FLT transport by these transporters in cancer cells and demonstrated that they—particularly ENT-1 in murine tumor models—facilitate [
18F]FLT uptake.
22,
23 Consistent with increased 14-day AngII AAA [
18F]FLT uptake on in vivo PET/CT and ex vivo gamma counting, these transporters were significantly upregulated in 14-day AngII AAA tissues. The established association between [
18F]FLT uptake and thymidine-associated marker expression in cancer is interesting to draw parallels with that in AAA. While concomitant malignancies occur in up to 14% of AAA cases
37 and smoking is an important shared risk factor between the two pathologies, we do not suggest a shared mechanism to the proliferation observed in our study in AAA and that previously reported in cancer. We merely observe that both pathologies involve cell proliferation.
Some limitations are associated with our study. The AngII AAA mouse model exhibits a high rate of mortality due to aneurysm rupture.
38 Importantly, rupture in this model is not typically due to AAA progression, but is an early event in AAA initiation. Regarding PET/CT image analysis, the abdominal aorta and spleen are both anatomically near the bladder and kidneys, which readily take up [
18F]FLT during excretion and consequently raise the possibility of signal spill-over into the aortic ROI. The ability to definitively study the aortic wall was also limited by the PET scanner’s spatial resolution (1 mm), making it difficult to evaluate tracer uptake in aneurysmal regions. To mitigate this risk, we used SUV
max for PET/CT analysis so that partial volume effects were minimal. We also took efforts to confirm that the PET/CT signal was originating from the aortic wall by conducting ex vivo gamma counting and performing co-registration of the 3D USS and PET datasets. Finally, the cell type(s) contributing to this proliferative [
18F]FLT signal are insufficiently clear from our study, warranting further investigation.
New Knowledge Gained
In 14-day AngII-induced AAA compared to 28-day AngII-induced AAA and saline controls, the [18F]FLT PET signal is significantly enhanced and [18F]FLT counts are greater, which themselves correlate with aortic volume. The expression of the [18F]FLT substrate, TK-1, and [18F]FLT transporter proteins ENT-1, ENT-2, CNT-1, and CNT-3 are also increased in AngII AAA. These findings suggest an early period of cell proliferation in the AngII AAA murine model, which is detectable using PET/CT.
Acknowledgements
We thank Melanie Reay, Laura Goodlass, and Danielle Webster for their technical work maintaining the animals during the study.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.