Introduction
The 18-kDa translocator protein (TSPO), also known as the peripheral benzodiazepine receptor, is a protein of the outer mitochondrial membrane expressed in peripheral organs and is particularly enriched in steroidogenic tissue. In the central nervous system (CNS) TSPO expression is low and limited to resident glial cells [
1].
However, cerebral TSPO expression is dramatically increased after glial cell activation and has become a well-characterized marker for neuroinflammation [
2‐
6]. In addition, TSPO levels are also elevated in a number of cancer cell lines and human tumours including breast, ovary, colon and prostate cancer, as well as glioma [
7,
8]. Initial studies on TSPO and TSPO ligands in the brain carried out in the 1980s and 1990s indicated that the density of TSPO was high in malignant gliomas [
9‐
12] and in glioma cell lines [
9‐
14]. Further studies demonstrated that TSPO expression levels positively correlated with the grade of malignancy and showed a negative correlation between TSPO expression and survival [
15,
16]. In the late 1980s positron emission tomography (PET) imaging of human gliomas using TSPO radioligands was suggested [
11‐
13] and first demonstrated with the isoquinoline derivative [
11C]PK11195 [
10,
17]. However, numerous limitations of this radioligand have been reported, including the high level of non-specific binding and the relatively poor signal-to-noise ratio [
2].
In the last few years many efforts have been undertaken in the development of new radioligands targeting the TSPO. Thus, recently developed radioligands with improved in vivo specificity for TSPO may have the potential to improve PET imaging of TSPO expression in human gliomas. One of the promising new radioligands for TSPO imaging is the pyrazolo[1,5]pyrimidine [
18F]DPA-714. [
18F]DPA-714 demonstrated lower non-specific uptake and higher binding potential (BP) as compared to [
11C]PK11195 in a rat model of acute neuroinflammation [
18]. The aim of the present study was to investigate the potential use of [
18F]DPA-714 as a new PET imaging marker for glioma. Here we report the use of [
18F]DPA-714 to monitor TSPO-expressing gliomas in vivo using PET imaging. The specificity of the radioligand binding was confirmed by displacement studies and the level of TSPO expression was assessed using immunohistochemical analysis in syngeneic and allogeneic 9L rat glioma models.
Discussion
The present study evaluated the potential use of the TSPO radioligand [
18F]DPA-714 as a PET imaging marker for non-invasive imaging of intracranial rat glioma models in vivo. [
18F]DPA-714 PET imaging of 9L gliomas grown in three different rat strains (Fischer, Wistar and Sprague Dawley rats) demonstrated significant [
18F]DPA-714 PET uptake at the site of tumour implantation as compared to the contralateral brain hemisphere or cerebellum, and in vivo displacement studies with unlabelled DPA-714 or PK11195 showed high specificity of [
18F]DPA-714 for TSPO. Finally, immunohistochemistry of brain/tumour sections confirmed high TSPO expression within the glioma and demonstrated the presence of TSPO-positive glioma as well as TSPO-positive microglial cells (Fig.
7).
The gold standard imaging method for the diagnosis of human gliomas is T1-weighted MRI with and without gadolinium enhancement in conjunction with T2-weighted MRI. These sequences provide information on the size and localization of the tumour and additional information about secondary phenomena such as disruption of the blood-brain barrier, mass effect, perifocal oedema, haemorrhage, necrosis and signs of increased intracranial pressure [
40]. However, no specific information about the biological or metabolic activity of the tumour is provided by T1- or T2-weighted MRI. Recent reviews point out the usefulness of combining advanced imaging methods, in particular PET in conjunction with MRI, to visualize additional molecular information and biological changes of the tumour [
41,
42]. Overexpression of TSPO has been reported in a variety of cancer cell lines and human tumours including glioma [
8], and clinical studies using the TSPO radioligand [
11C]PK11195 imaged human glioma using PET [
10,
17]. However, this radioligand has various limitations, among others, the labelling with
11C (
T
1/2: 20.4 min) restricts its extensive clinical use [
2]. In the last few years many efforts have been undertaken in the development of new TSPO radioligands. One of the [
11C]PK11195 challengers that showed improved characteristics when compared to the reference compound is the pyrazolo[1,5]pyrimidine [
18F]DPA-714 [
20]. In a rat model of neuroinflammation, [
18F]DPA-714 was directly compared to [
11C]PK11195 and another pyrazolo[1,5]pyrimidine [
11C]DPA-713 [
18] and demonstrated improved signal to noise ratio and increased BP as compared to [
11C]PK11195. Martín et al. used this TSPO radioligand to characterize the time course of TSPO expression in an animal model of focal cerebral ischaemia [
32]. Other new TSPO radioligands reported include the phenoxyarylacetamides [
11C]DAA1106, [
11C]PBR28 or [
18F]PBR06, which have been claimed to perform better than [
11C]PK11195 for in vivo imaging of TSPO, based on higher binding affinity in an animal model of neuroinflammation [
43] and high levels of specific binding in non-human primates and in human brain [
44,
45]. The new TSPO ligands may redraw attention to TSPO PET imaging of gliomas, as suggested by a recent study of in vivo imaging of [
18F]PBR06 in a preclinical model of glioma [
33].
Here we demonstrate that 9L glioma cells as well as intracranial 9L gliomas express TSPO at a high level. In vivo PET imaging of [
18F]DPA-714 in the 9L glioma model resulted in high [
18F]DPA-714 uptake in the tumour as compared to the contralateral brain hemisphere. Comparison with sham-operated animals shows that the PET signal was not influenced by an uptake of [
18F]DPA-714 due to a neuroinflammatory reaction induced by the operation procedure itself (Fig.
2).
TACs for tumour VOIs demonstrate a slow washout of [
18F]DPA-714 as compared to control brain regions, as previously reported in a rat model of focal cerebral ischaemia [
32]. Slow washout may also be a characteristic of the rat glioma model. Using [
18F]PBR06 in a C6 glioma model in Wistar rats, Buck et al. also demonstrated a slower washout of this radioligand from the tumour tissue than from the contralateral brain, thus facilitating a strong tumour to contralateral brain contrast [
33].
The 9L glioma originates from a tumour that arose in Fischer rats [
46]. However, 9L cells can also form tumours in allogeneic Wistar [
35] and Sprague Dawley rats [
47]. Quantitative data analysis demonstrated that [
18F]DPA-714 uptake was significantly enhanced in the tumour as compared to the contralateral brain hemisphere or cerebellum in all of the three strains. Furthermore, differences in the radioactivity concentrations expressed in %ID/cc or SUVs in the tumour were observed between the different rat strains regardless of body weight. An explanation for the differences in the %ID/cc values in the tumour may result from the tumour model itself. Since 9L cells are allogeneic to Wistar or Sprague Dawley rats, the tumours may present different stages of carcinogenesis and thus e.g. express proteins at different levels, as previously shown for K(ATP) and K(Ca) channels when 9L tumours were compared in Wistar and Fischer rats [
35]. Another factor that may influence specific radiotracer binding is the contribution of the TME, in particular the presence of inflammatory cells such as activated microglia, which may be different for 9L tumours in syngeneic or allogeneic rats. Inter-strain differences in radiotracer plasma availability (due to excretion, metabolism or binding to plasma proteins) or differences in cerebral vascularization may also contribute to the differences measured in tumour radioactivity concentration. Wistar rats present a significantly greater number of proximal side branches of the long proximal middle cerebral artery segment than Fischer rats [
48]. Since %ID/cc values represent a combination of free, specifically and non-specifically bound radioligand in a target region, we modelled PET data using the SRTM [
28]. It should be noted that one of the requirements of SRTM is the use of a reference region without specific binding of the ligand. In order to normalize for inter-strain differences we performed our calculations using SRTM, although there is a small displaceable binding in both reference regions analysed in our study. Estimated R
1, k
2 and BP
ND values did not show significant changes between the rat strains independently of the reference region chosen, suggesting that the inter-strain differences are likely due to anatomical and/or radiotracer metabolism variations between strains. However, a difference was found for R
1 values, whether the contralateral side or the cerebellum were taken as reference region (
p < 0.05 and
p < 0.001 for Fischer and Wistar or Sprague Dawley rats, respectively, Table
1). This may be explained by the differences seen in particular for the first minutes of TACs, as R
1 defined as the ratio of tracer delivery (K
1/K
1’) and K
1 and K
1’, respectively, are mainly determined by the first minutes after radiotracer injection.
As R1 values (K1/K1’) calculated with the contralateral side as reference region were higher than 1 (1.24 ± 0.19, 1.33 ± 0.17 and 1.42 ± 0.31 for Fischer, Wistar and Sprague Dawley rats), indicating a facilitated entry of the radioligand into the tumour as compared to the reference region, we examined whether increased [18F]DPA-714 uptake into the tumour may have been due to disruption of the blood-brain barrier. However, imposing a hypothetical R1 value of 1 did not significantly influence the BPND values, indicating that [18F]DPA-714 delivery into the tumour has only a limited effect on the BPND.
Accordingly, in vivo displacement studies demonstrated that [
18F]DPA-714 uptake reflects specific TSPO binding. Increasing amounts of unlabelled compound (DPA-714 or PK11195) yielded an increased displacement of tumour radioactivity concentrations. A dose of 5 mg/kg of unlabelled DPA-714 displaced more than 85% of the in vivo tumour uptake, whereas only 55% of the original tumour radioactivity concentration was displaced after administration of 5 mg/kg unlabelled PK11195 (Fig.
5b, Table
2). This difference in the displacement of the TSPO radioligand using the corresponding unlabelled compound or unlabelled PK11195 has previously been reported [
32] and is likely due to a higher level of specific versus non-specific binding of DPA-714. It is most unlikely due to differences in affinities since in an in vitro binding assay, James et al. showed that affinities of DPA-714 and PK11195 for TSPO are comparable [affinity of DPA-714 (K
i = 7.0 ± 0.4 nM); affinity of PK11195 (K
i = 9.4 ± 0.5 nM)] [
20]. The differences in displacement efficiency may be explained by the fact that PK11195 is in a racemic form while DPA-714 is not. It is known that only the
R-enantiomer of PK11195 is pharmacologically active, thus the displacing dose of PK11195 is half that of DPA-714.
Recently, PET studies using [
11C]PBR28 demonstrated differences in affinity for TSPO in humans [
49‐
51]. In a PET study with healthy human subjects, Fujita et al. reported the existence of non-binders for [
11C]PBR28, whereas no such phenomenon has been described for [
11C]PK11195 [
49]. At a later date Owen et al. reported the presence of three different binding patterns with PBR28 (high-affinity binding, low-affinity binding and mixed-affinity binding), whereas no difference in affinity of PK11195 for TSPO was found in their study [
51]. Furthermore, a recent study by the same group demonstrated that other TSPO ligands, currently used in clinical studies, can distinguish between these three binding patterns, although the differences in affinity do vary depending on the radioligand [
52]. Strong differences were found for the phenoxyphenyl acetamides PBR28 and PBR06 (K
i ratio LAB to HAB: 55 and 17, respectively), whereas others like the phenylimidazopyridine PBR111, the phenylpyrazolopyrimidine DPA-713 or the phenoxyphenyl acetamide DAA1106 demonstrated only smaller differences in affinity (four- to fivefold). Although it is likely that [
18F]DPA-714, which is an
18F-labelled derivative of DPA-713, will demonstrate a similar low difference in affinity ratio (low- versus high-affinity binding), this remains to be demonstrated.
In contrast to [
11C]PBR28, PET studies with [
18F]PBR06, [
18F]PBR111, [
11C]DPA-713 and [
11F]DAA1106 did not report the phenomenon of low-affinity binding, probably due to an affinity for the respective PET radioligand sufficient to obtain a measurable PET signal [
52]. At present the impact of these differences in binding affinity on the interpretation of the PET signal remains to be fully elucidated although it may be expected that low-affinity binding sites should lead to an underestimation of TSPO.
Gliomas are not exclusively composed of cancer cells but also of cells of the TME [
36]. In particular, microglia/macrophages which can make up to 30% of the brain tumour mass [
39] and astrocytes which are found in the proximity of glioma cells [
53]. Increase in TSPO expression is a hallmark of glial cell activation upon neuroinflammation often associated with neoplastic processes. Thus, we tested whether activated microglia/macrophages and astrocytes contributed to the PET signal obtained in the glioma model using triple immunolabelling of brain/tumour sections. CD11b-positive microglia/macrophages were present within the glioma and demonstrated TSPO expression. In contrast, astrocytes were not found within the tumour and, thus, their contribution to the observed TSPO signal may be regarded as limited. Although the majority of the TSPO signal within the tumour is provided by the glioma cells, activated microglia/macrophages contribute to the TSPO signal and thus to [
18F]DPA-714 uptake in the tumour. In their in vivo imaging study using [
18F]PBR06 in a rat glioma model, Buck et al. also demonstrated significantly higher TSPO immunostaining in the glioma as compared to normal brain tissue, and tumour to normal brain ratios were similar to TSPO protein levels in tumour tissue as compared to normal brain tissue [
33]. However, no immunohistochemistry data about other cells possibly contributing to the PET signal were provided. The finding that microglia/macrophages account for the TSPO signal may be of limited importance for diagnostic purposes but may have significant implications when this radioligand is employed for assessment of treatment response to i.e. radiotherapy or combined radio- and chemotherapy.