Background
Alzheimer's disease (AD) is the most common cause of dementia. The main pathological findings in a typical AD brain are neurofibrillary tangles and extracellular neuritic plaques, mainly composed of a fibrillar form of beta-amyloid (Aβ) peptide [
1]. Methods based on molecular imaging have recently made it possible to assess brain Aβ deposition in patients with AD. Positron emission tomography (PET) with the Aβ imaging agent [
11C]Pittsburgh compound B ([
11C]PIB) first established this as a valuable biomarker approach for evaluating changes in Aβ deposition [
2],[
3]. Several small animal PET imaging studies have also been performed with [
11C]PIB, but the results have not been consistent, probably reflecting differences in specific activity of the tracer and in the employed animal models [
4]-[
8]. Due to the short half-life (
T½ = 20.4 min) of
11C,
18
F-labelled ligands for Aβ imaging have been developed to extend the use of amyloid PET to the centres without on-site radionuclide production and tracer synthesis capacity. Due to their practical benefits, the use of
18
F-labelled tracers has also raised interest in small animal imaging [
9]-[
11].
[
18
F]flutemetamol (2-(3-[
18
F]fluoro-4-(methylamino)phenyl)-1,3-benzothiazol-6-ol, [
18
F]3′F-PiB, [
18
F]GE067) is a
18
F-labelled analogue of [
11C]PIB developed and marketed by GE Healthcare (Buckinghamshire, UK). [
18
F]flutemetamol PET has been shown to successfully differentiate between patients with AD and healthy control subjects and to perform similarly to [
11C]PIB in this respect [
12],[
13]. [
18
F]flutemetamol has shown high test-retest reliability and high specificity (96%, [
12]) and sensitivity (93%, [
12]) in the detection of AD, and regional imaging results obtained with it were consistent with AD plaque pathology in cortical biopsy samples [
14]-[
16]. Based on these findings, [
18
F]flutemetamol received FDA approval as a diagnostic agent for the assessments of Aβ deposition in the brains of adults evaluated for AD (FDA application number (NDA) 203137, GE Healthcare).
The present demonstration of principle study evaluated the applicability of [
18
F]flutemetamol to detect and quantitate the changes in brain Aβ deposition in APP23, Tg2576 and APPswe-PS1dE9 transgenic (TG) mouse models of AD. We expected that the very high specific activity of [
18
F]flutemetamol (>1 TBq/μmol) and thus lower injected mass to transgenic animals would be beneficial in small animal Aβ imaging. We also expected physical characteristics and practical benefits of the
18
F-radionuclide to make [
18
F]flutemetamol an attractive preclinical Aβ imaging agent for preclinical studies.
Discussion
In the present study, we evaluated the applicability of [
18
F]flutemetamol to assess brain Aβ deposition in small animal
in vivo PET imaging. [
18
F]Flutemetamol has already been extensively investigated in humans and has received marketing approval as a radiopharmaceutical for the detection of Aβ deposition in subjects with possible AD [
12],[
23]-[
25]. In our previous animal study, we found that the pharmacokinetic properties of [
18
F]flutemetamol in WT rats and mice were suitable for preclinical imaging, and [
18
F]flutemetamol was found to bind to Aβ deposits in the Tg2576 mouse brain, both
in vitro and
ex vivo[
9]
. To our knowledge, the present demonstration of principle study is the first to show that it is possible to detect increase in [
18
F]flutemetamol retention
in vivo in the mouse brain parallel with increasing age. However, in the present study, increased [
18
F]flutemetamol retention was evident only in the transgenic APP23 mice and not in Tg2576 or APPswe-PS1dE9 mice. APP23 mouse model is known to have relatively slow plaque deposition and large Aβ deposits with dense amyloid cores [
4],[
18]. Tg2576 mice have late onset and even slower rate of Aβ deposition than APP23 mice [
4]. In APPswe-PS1dE9 mice, small Aβ deposits are present already at earlier age, and deposition is abundant throughout the brain in older animals [
4].
Varying results have been found in several small animal imaging studies with [
11C]PIB, a structural analogue of [
18
F]flutemetamol which is still considered the golden standard of brain amyloid imaging. A possible explanation for the variation is that different TG mice are likely to express different Aβ isoforms with different amounts of high-affinity binding sites for amyloid tracers [
26]. We hypothesized that very high specific activity and lower injected mass of [
18
F]flutemetamol might provide an advantage in some animal models that did not show increased specific binding with [
11C]PIB in our previous study, perhaps because of low levels of high-affinity binding sites [
4]. In this study, the injected mass of [
18
F]flutemetamol was less than 5.5 ng in all
in vivo cases (
N = 48). This is even at its worst ten times less than previously reported with [
11C]PIB (68 ± 23 ng;
N = 50 [
4]). However, the results revealed that the use of a tracer with high specific radioactivity did not as such provide any advantage in PET imaging of APPswe-PS1dE9 mice
in vivo. In APPswe-PS1dE9 mice, plaque deposition was abundant throughout the brain (including the CB) at 19 months, but plaque composition appeared to be different from the other TG models, with only small fibrillar deposits that could bind ThS and related Aβ imaging agents. In Tg2576 mice, modest increases in DVR and FC/CB
50-60 ratio were seen as the individual transgenic mice aged. In a previous study with [
11C]PIB and Tg2576 mice, no such trend was seen [
4]. In addition, [
18
F]flutemetamol did bind to Aβ in the cortical sections from Tg2576 animals, both
ex vivo and
in vitro, consistent with previous
ex vivo results [
9]. However, the amount of plaques was modest even at 22-month-old animals, explaining the modest increases in
in vivo binding ratios. We concluded that Tg2576 and APPswe-PS1dE9 mouse models are not suited to small animal imaging studies with tracers that shared binding sites with [
11C]PIB and [
18
F]flutemetamol, due to their plaque structure or low plaque load.
Our study is difficult to compare with previous small animal amyloid imaging PET studies that used other
18
F-labelled tracers, due to the differences in tracers, quantitation methods and animal models used. One previous study did not observe increased cortical tracer retention with [
18
F]FDDNP in 13 to 15-month-old Tg2576 mice [
27]. The authors concluded that specific [
18
F]FDDNP binding was insufficient and that insufficient spatial resolution and partial volume effects (PVE) limited the precision of measurements in small VOIs [
27]. However, in a more recent study with [
18
F]florbetaben, Aβ deposition was longitudinally followed in the APPswe mouse model from age 13 to 20 months, and the results were consistent with our results with [
18
F]flutemetamol in APP23 mice. In that study, the PET results agreed well with histopathological brain Aβ, and differentiation was further improved when a PVE correction was applied [
10],[
28]. Using the APP-PS1-21 mouse model, also specific binding of [
18
F]florbetapir was shown to increase from age 3 to 8 months; however, no further increase was detected at 12 months compared to the baseline scan at 3 months [
11]. At the end of synthesis, the specific activities of [
18
F]florbetaben (80 GBq/μmol, [
10]) and [
18
F]florbetapir (150 to 220 GBq/μmol, [
11]) were much lower than the specific activity of [
18
F]flutemetamol in this study (>1 TBq/μmol). This difference further supports the higher importance of the chosen mouse model over high specific activity of the tracer for successful imaging results.
Similarly to human studies, the CB has usually been used as a reference region in small animal amyloid imaging studies. However, several limitations should be considered when applying this approach. In the mouse brain, the white and grey matter cannot be distinguished when the reference region VOIs are drawn over the CB, due to its small size. As a result, because all
18
F-labelled β-amyloid imaging tracers typically exhibit prominent white matter binding, the concentration of radioactivity in the CB can be higher than that in the cortical grey matter ROIs. In our study, DVRs were less than unity in WT animals and in young TG animals, which presumably had only modest plaque loads. Negative BP
ND values for WT and young TG mice were also previously reported for [
18
F]florbetaben. Only TG mice that were 16 and 20 months old showed positive BP
NDs (0.04 ± 0.07 and 0.10 ± 0.11, respectively) [
10]. For small animal imaging, tracers with low non-specific binding to white matter would be beneficial, such as the novel
18
F-labelled fluoropyridyl derivative, BAY 1008472, or the benzofuran ligand [
18
F]AZD4694 [
29],[
30]. Moreover, for APPswe-PS1dE9 mice, the CB is not a suitable reference region because Aβ deposits are present at 19 months of age.
Imaging of small structures in the mouse brain, of a size close to the spatial resolution of the PET scanner, will inevitably be affected by PVEs; thus, measurable signals will be less precise than those achievable with digital autoradiography. Consistent with previous [
18
F]florbetaben and [
18
F]florbetapir studies, the FC/CB ratios observed in our study with [
18
F]flutemetamol with
ex vivo autoradiography (1.8 and 1.9, respectively, at 30 min p.i.) were higher than that observed with
in vivo PET (1.3 at 30 to 35 min p.i.) for old APP23 mice. Adding PVE correction could further improve quantitation of [
18
F]flutemetamol PET data, especially in animals with still only modest plaque load. Also, spillover from tissues close to the brain could introduce errors to the VOI-based analysis method in mice. We observed high radioactivity in close proximity to the brain
in vivo, and subsequent
ex vivo tissue counting experiments confirmed that significant radioactivity was located outside of the brain, inside the nasal cavity. In addition,
18
F-radioactivity concentrations detected in the mouse cranial bone (0.83 ± 0.13% ID/g at 30 min) were higher than those previously reported in rats (0.04 ± 0.01% ID/g at 30 min [
9]); this suggests that more [
18
F]flutemetamol defluorination could be taking place in mice than in rats, reflecting more active metabolism. However, spillover from these structures would presumably lead to similar overestimation in the brains of all three TG mouse models.
One limitation of the present study was the small number of transgenic animals, especially the ones that could be imaged repeatedly over multiple time points. However, we wanted to evaluate the applicability of [
18
F]flutemetamol PET to small animal imaging in three different transgenic models even with limited statistical power, rather than in only one transgenic model with higher sample size and better statistical power. This demonstration of principle type of study was still able to show clear increasing trend in [
18
F]flutemetamol binding in aging APP23 mice and the lack of binding in Tg2576 and APPswe-PS1dE9 mice. Another limitation was that we could not perform
ex vivo studies for all of the female mice imaged
in vivo; however, the brain sections from these mice were used for
in vitro binding experiments, and additional
ex vivo studies were performed later, with aged males. Male APP23 mice were reported to show slightly slower deposition of Aβ plaques compared to females [
17], but these mice were 27 months old; thus, Aβ deposition was abundant. In addition, a hybrid PET/MRI scanner with better spatial resolution and MR details of the brain would better serve the purposes of similar experiments.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AS contributed to the design of the study, acquired, analysed and interpreted the PET data and histological and immunohistochemical stainings. AS drafted the manuscript. JR participated in the design of the study and carried out in vitro PET experiments. FL participated in the immunohistochemical stainings and performance of PET experiments. OE performed the radiochemical synthesis of [
18
F]flutemetamol. MS and GF participated in the design of the animal studies and revised the manuscript critically. MS drafted and revised the manuscript critically for its intellectual content. OS, JR and MHS contributed to the conception, design and coordination of the study, revised the manuscript critically for its intellectual content and gave final approval for the published version. MHS helped to draft the manuscript. All authors read and approved the final manuscript.