Direct Comparison of [18F]F-DPA with [18F]DPA-714 and [11C]PBR28 for Neuroinflammation Imaging in the same Alzheimer’s Disease Model Mice and Healthy Controls

Purpose In this study we compared the recently developed TSPO tracer [18F]F-DPA, with [18F]DPA-714 and [11C]PBR28 by performing in vivo PET imaging on the same Alzheimer’s disease mouse model APP/PS1-21 (TG) and wild-type (WT) mice with all three radiotracers. Procedures To compare the radiotracer uptake, percentage of injected dose/mL (%ID/mL), standardized uptake value ratios to cerebellum (SUVRCB), and voxel-wise analyses were performed. Results The peak uptake of [18F]F-DPA was higher than 4.3% ID/mL, while [18F]DPA-714 reached just over 3% ID/mL, and [11C]PBR28 was over 4% ID/mL in only one brain region in the WT mice. The peak/60-min uptake ratios of [18F]F-DPA were significantly higher (p < 0.001) than those of [18F]DPA-714 and [11C]PBR28. The differences in [18F]F-DPA SUVRCB between WT and TG mice were highly significant (p < 0.001) in the three studied time periods after injection. [18F]DPA-714 uptake was significantly higher in TG mice starting in the 20–40-min timeframe and increased thereafter, whereas [11C]PBR28 uptake became significant at 10–20 min (p < 0.05). The voxel-wise analysis confirmed the differences between the radiotracers. Conclusions [18F]F-DPA displays higher brain uptake, higher TG-to-WT SUVRCB ratios, and faster clearance than [18F]DPA-714 and [11C]PBR28, and could prove useful for detecting low levels of inflammation and allow for shorter dynamic PET scans. Supplementary Information The online version contains supplementary material available at 10.1007/s11307-021-01646-5.


Introduction
Neuroinflammation is associated with several neurological diseases, such as multiple sclerosis (MS), Alzheimer's disease (AD), and stroke. Innate pathology triggers inflammation that induces an increase in the expression of mitochondrial 18-kDa translocator protein (TSPO) in the microglia. TSPO is the main target for the PET tracers currently used for imaging neuroinflammation in vivo [1][2][3][4][5].
The half-life of fluorine-18 (t 1/2 = 109.8 min) makes it a more desirable radioisotope than carbon-11 (t 1/2 = 20.3 min) for the development of radiotracers because of the possibility of distribution to PET centers lacking an on-site cyclotron. Furthermore, fluorine-18 emits positrons with low energy (E β+ max = 0.63 MeV). Consequently, the positrons have a short range in tissue and provide higher-resolution PET images than those afforded by carbon-11 radiotracers. For these reasons, several 18 F-labeled tracers have been developed, notably among them [ 18 F]DPA-714, which presents good binding potential and bioavailability. According to several animal studies, [ 18 F]DPA-714 is better for PET imaging than [ 11 C]PK11195, due to its low nonspecific binding in the brain and the longer half-life of the labeling radionuclide [13,14]. Recent longitudinal studies using mouse and rat models of AD have shown an increase in [ 18 F]DPA-714 uptake with disease progression [15][16][17]; however, the results of human [ 18 F]DPA-714 PET studies in AD patients have been contradictory [18][19][20].
We have previously published the synthesis of [ 18 F] F-DPA, an analogue of [ 18 F]DPA-714, by an electrophilic 18 F-labeling route and showed that the position of the label directly on the aromatic moiety imparts a higher in vivo stability than that of [ 18 F]DPA-714 in Sprague-Dawley rats [12]. More recent studies have demonstrated the specificity of [ 18 F] F-DPA towards TSPO and its usefulness in imaging glial activation in the APP-PS1/21 mouse model of AD [11] and in a model of ischemic stroke [21]. In addition, a study comparing electrophilic and nucleophilic syntheses of [ 18 F]F-DPA demonstrated that a 100-fold difference in injected mass of [ 18 F]F-DPA affected both the tracer kinetics and uptake in the APP-PS1/21 mouse model. The higher injected mass gave a faster washout and more rapid establishment of tracer equilibrium while still providing a significant uptake difference between age-matched TG and WT animals [10].
In this study we chose to compare [ 18 F]F-DPA synthesized by the electrophilic approach with the clinically used TSPO tracers [ 18 F]DPA-714 and [ 11 C]PBR28 in the same TG and WT mice. The relative ease of the electrophilic synthetic PET/computed tomography (CT) scanner (Siemens Medical Solutions, Knoxville, TN, USA). A few drops of Oftagel (2.5 mg/g; Santen, Tampere, Finland) were applied to the eyes of the animals to prevent eye dryness. The scanner has an axial 12.7-cm field of view and 10-cm transaxial field of view, generating images from 159 transaxial slices of voxel size of 0.78 × 0.78 × 0.8 mm 3 . CT preceded the PET modality for attenuation correction and anatomical reference. One of the TG animals imaged with [ 18 F]F-DPA died during the scan, and hence this animal was excluded from the analysis.

Analysis of PET Data
The PET/CT images were pre-processed in MATLAB R2017a (The MathWorks, Natick, Massachusetts, USA) with an in-house semi-automated pipeline for preclinical images that use SPM12 (Wellcome Department of Cognitive Neurology, London, UK) pre-processing functionalities and analysis routines. Images were first cropped to a bounding box containing the heads, and individual PET images were co-registered through a rigid-body transformation to their corresponding CT scan. Subjects were spatially normalized through a two-step registration (a rigid followed by an affine transformation) of each subject's CT to a template CT that was previously constructed as an average of several subjects and was aligned with an atlas T2-weighted MRI template [24]. The combination of transformations was then applied to the PET images, which were also re-sampled to a voxel size of 0.2 × 0.2 × 0.2 mm (trilinear interpolation), matching the anatomical atlas dimensions.
Volume of interest (VOI) analysis of whole brain, cortex, hippocampus (HIPPO), striatum (STR), thalamus (THA), and cerebellum (CB) was performed on each subject by averaging the signal inside a slightly modified version of the Ma et al. [24] atlas-delineated VOIs. Data were obtained as the percentage of injected dose/mL (%ID/mL) or standardized uptake value ratios to cerebellum (SUVR CB ).
Prior to the voxel-wise analysis, single static frames representing three time periods after tracer injection (10-20 min, 20-40 min, and 40-60 min) were constructed by averaging the corresponding frames of the dynamic scans for each subject and each tracer. The resulting static images were intensity-normalized to the CB as a reference region (tissueto-reference ratio images) and smoothed using an isotropic Gaussian kernel of 0.5 mm full width at half maximum. Ratio images were masked using a whole brain mask to remove the extra-cerebral signal prior to the statistical parametric mapping analysis, and the analysis was performed without global normalization due to the use of ratio images. Betweengroup effects were tested for each tracer using a voxel-wise two-sample t-test. An uncorrected voxel-level significance threshold (p < 0.01) was used, and subsequently a clusterlevel family-wise error correction at p < 0.01 was applied.

Data Analysis and Statistics
The results are reported as average ± standard deviations (SDs). All statistical analyses were calculated using Graph-Pad Prism (GraphPad Software, v. 5.01, San Diego, CA, USA).
The differences in [ 18 F]F-DPA, [ 18 F]DPA-714, and [ 11 C] PBR28 SUVR CB between TG and WT animals in the different time frames shown in Fig. 2 were calculated using the Mann-Whitney test.
The differences in peak/60-min ratios shown in Table 1 Table 2 were analyzed using repeated-measures ANOVA with Tukey's post hoc test for multiple comparison.
The residuals followed a normal distribution. Differences were considered significant for values of p < 0.05.

In vivo [ 18 F]F-DPA, [ 18 F]DPA-714, and [ 11 C]PBR28 Uptake in WT Mice
The graphs in Fig. 1 show the 60-min time-activity curves for whole brain, cortex, HIPPO, STR, THA, and CB in WT mice. The results show the higher initial brain uptake and faster washout of [ 18  we used 9-month-old APP/PS1-21 AD model mice. SUVR CB values were calculated for the whole brain, cortex, HIPPO, STR, and THA in WT and TG mice, and the differences in SUVR CB between WT and TG animals were calculated at 10-20, 20-40, and 40-60 min after radiotracer injection (Fig. 3). For [ 18 F]F-DPA, the SUVR CB differences between WT and TG mice were highly significant (p < 0.001) in all the studied areas in the three studied periods (Figs. 3a-e). In the case of [ 18 F]DPA-714, there were no significant differences in the 10-20-min time period, but the differences were significant in the 20-40-min (p < 0.01) and 40-60-min (p < 0.01) periods for all regions (Figs. 3f-j). For [ 11 C]PBR28 at 10-20 min, the SUVR CB differences between WT and TG were already significant for all the analyzed regions (p < 0.05), increasing in significance at 20-40 min (p < 0.001) and at 40-60 min (p < 0.001) (Figs. 3k-o). In the 20-40-min period, high significance (p < 0.001) was still observed in the differences between the TG/WT ratios of [ 18 F]F-DPA and [ 18 F]DPA-714 in the whole brain, HIPPO,    We performed a voxel-wise analysis of the PET images to study the differences in tracer uptake between WT and TG mice at 10-20, 20-40, and 40-60 min after injection at the voxel level. Figure 4 shows coronal and axial images of
In humans, increased microglial activation has been shown in post mortem brain samples of AD patients, although the role of this activation is still controversial [32,33]. The first human PET studies targeting neuroinflammation were performed using [ 11 C]PK11195, with contradictory results [34][35][36][37]. The varied findings with [ 11 C]PK11195 could be explained by the variability in the studied populations, as well as the limitations of the radiotracer itself, including high non-specific binding, high lipophilicity, low blood-brain barrier penetration, and low binding potential [6,7,38]. Studies using newer TSPO radiotracers have also shown heterogeneous results in different populations [19,[39][40][41]. Among these novel tracers, [ 18 F]DPA-714 has a better signal-to-noise ratio and a greater affinity than [ 11 C]PK11195 [13]. The first human studies with [ 18 F]DPA-714 concluded that this tracer cannot be used to distinguish individual AD patients from healthy subjects [19]. In contrast, a more recent human prospective study looked into early and protective microglial activation in AD and concluded that [ 18 F]DPA-714 can be a good tool for assessing neuroinflammation in early and preclinical AD [20].
In the current study, we have shown in WT mice that [ 18 F]F-DPA has better brain penetration and faster washout than [ 18 F]DPA-714 and [ 11 C]PBR28, as shown in Table 1 by the peak tracer uptake and the highly significant differences in peak uptake/60-min ratios. In addition, we compared PBR28 (bottom) brain uptake at different periods after injection (10-20, 20-40, and 40-60 min) the uptake in the brains of 9-month-old APP/PS1-21 mice. We have shown in two longitudinal studies that significant [ 18 F]F-DPA and [ 18 F]DPA-714 uptake can be measured in 9-month-old APP/PS1-21 model mice [11,16] and therefore chose animals of that age for the direct tracer comparison. In the current study, higher SUVR CB values were achieved with  (40-60 min), when the cluster sizes are also more similar. Given that the same animals were scanned with the three radiotracers, the most likely explanation is that this is due to the differences in uptake and washout speeds between the three radiotracers. This is in agreement with the results observed in Fig. 3, where in the last time frame (40-60 min) the smallest differences in SUVR CB between WT and TG mice were observed for the three radiotracers.
The observed uptake differences between the closely related radiotracers [ 18 F]F-DPA and [ 18 F]DPA-714 could be due to the higher in vivo metabolic stability of [ 18 F]F-DPA reported in rats and mice [11,12]. In those studies, we showed that [ 18 F]F-DPA is more metabolically stable than [ 18 F]DPA-714 in the brain; non-metabolized [ 18 F]F-DPA accounted for more than 90% of the remaining radioactivity even 90 min after injection, whereas for [ 18 F]DPA-714, only about 50% of the brain activity was the parent compound. This can be explained by the direct 18 F-labeling of the aromatic ring in [ 18 F]F-DPA imparting a higher stability than the metabolically unstable alkoxy-linked 18 [42][43][44][45].
In this study, we used the CB as the reference region for analyses of the PET images. The use of the CB as a reference region is well established for amyloid quantification with [ 11 C]PIB, but the choice of the CB is more controversial for analyzing the binding of TSPO tracers. We previously observed an age-dependent increase in tracer accumulation in the cerebellum; however, this increase resulted in a significant difference between TG and WT animals only from the age of 12 months onwards, with no significant difference observed at 9 months. The reference region (hypothalamus) employed in the previous study [10] is unsuitable as an in vivo reference region due to its proximity to the pituitary gland. Our use of the CB as the reference region for [ 18 F]F-DPA, [ 18 F]DPA-714, and [ 11 C] PBR28 was based on our PET imaging data showing no differences in tracer uptake between TG and WT mice, as shown by the voxel-wise analysis in Fig. 4, and the cerebellar time-activity curves in Suppl. Figure 1. In addition, in previous studies, the CB has proven to be a reliable reference region, and its use as such also decreases group variability [11,16,46]. From our study we can conclude that the novel TSPO radiotracer [ 18 F]F-DPA shows higher initial brain uptake, faster clearance, and better target-to-background ratios than [ 18 F]DPA-714 and [ 11 C]PBR28 when the comparisons are made with the same AD and WT animals. Furthermore, due to the washout kinetics, higher SUVR CB values are measured with [ 18 F]F-DPA and at earlier time points after injection. With all of these characteristics, the novel tracer [ 18 F]F-DPA could prove very useful for the detection of low levels of microglial activation inflammation and, because of its fast clearance, would permit shorter dynamic scans.

Conflict of Interests
The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.