22
and [18F]THK5351 [Harada et al., 2016]. Apart from THK compounds, there are three other tau radiotracers that are applied clinically with results reported in literature including [18F]flortaucipir ([18F]T807/[18F]AV-1451, Eli Lilly, US) [Chien et al., 2013] and [18F]T808 (also known as [18F]AV-680) [Chien et al., 2014] developed by Siemens Healthcare (Germany) and [11C]PBB3 developed by National Institute of Radiological Sciences (Japan) [Maruyama et al., 2013]. The
18F-labelled PBB3, [18F]PM-PBB3 (Aprinoia, Japan) has just completed its first in-human studies but the results have yet been reported in the literature. Although there are other tau radiotracers like [18F]MK6420 (Merck, US) that have been applied clinically, the results and chemical structures were not reported in the literature thus far.
23 showed only binding affinity to BS2. As such, [11C]BF227 might not show high diagnostic capability compared to [18F]florbetapir or [18F]florbetaben due to high binding preference to BS of low concentration. The binding affinities of other amyloid radiotracers to the three different binding sites were not evaluated or reported.
Figure 2.5: Different binding sites (BS) on Aβ protein. [Lockhart et al., 2005]
Tau proteins also showed three possible binding sites [Lemoine et al., 2015] but the concentrations of all binding sites have yet been evaluated. Thus far, only [18F]THK5117 have been evaluated to bind with high binding affinity to one BS and low binding affinity to the second BS, with a possibility of binding to the third BS [Lemoine et al., 2015]. Consideration to the binding preference of the radiotracers to high and low concentration binding sites are preferred but such information is not normally available from the literature. However, the binding affinities of the radiotracers to the types (e.g. Aβ1-40 vs. Aβ1-42) and forms (soluble vs. fibrillary) of amyloid, Aβ) and tau are more important as they are more reflective of the total binding signal in PET measurements. As such, the binding preference of tau radiotracer to Aβ is also important due to the lower concentrations of NFT compared to Aβ plaques.
Differential Binding
In general, the highest concentration of amyloid plaques deposits is found in the frontal cortex [Villemagne et al., 2015]. However, as shown in Braak and Braak’s and Delacourte’s stagings of AD, amyloid deposition does not follow a consistent spatial pattern (section 2.1.3). In addition, amyloid radiotracers have different affinities (Ki2/Ki1 > 10, [Ni et al., 2013]) to different binding sites on both amyloid and tau proteins [Harada et al., 2013].
The highest concentration of PHF-tau deposits is found in temporoparietal cortices [Villemagne et al., 2015]. Although PHF-tau deposition followed a consistent pattern, the tau concentration may vary greatly in subjects regardless of disease severity. Moreover, tau undergoes different post-translational processes in different tauopathies and are present only in certain regions in
24
different tauopathies (e.g. tau is present in the brainstem in PSP and CBD but not AD) [Villemagne et al., 2015]. The binding of tau radiotracers to non-AD tauopathies have only been evaluated in a few clinically-applied tau radiotracers such as [18F]T807, [18F]THK523 and [11C]PBB3 (Chapter 7). Although [18F]THK523 does not bind to some non-AD tauopathies [Fodero-Tavoletti et al., 2014], and [11C]PBB3 showed distinct selectivity to different tauopathies [Ono et al., 2017], [18F]T807 showed great variation in binding in non-AD tauopathies [Lowe et al., 2016].
PET imaging is unable to discriminate binding to different binding sites on amyloid and tau proteins as it measures all the radioactive signal from the radiolabeled isotope. As such, direct comparison of radiotracers using the same region in the PET image of the same subject is limited.
In vitro binding information to different binding sites is also limited and measured values vary greatly for low-affinity binding sites. Comparison of radiotracers using measured in vitro binding affinity to the highest affinity binding sites might lead to more consistent results than the inclusion of binding affinity to all binding sites.
Binding to Other Proteins
The currently-developed amyloid radiotracers target only fibrillary amyloid plaques, which have β-sheet structures (Figure 2.1B). Likewise, neurofibrillary tangles (Figure 2.2B) and α-synuclein proteins also have β-sheet conformation. As such, specific binding to other targets having similar conformation might arise. Amyloid proteins are much smaller than tau proteins (37~43 vs.
352~441 amino acids), they are present at much higher concentrations of 4~20 times that of tau proteins [Villemagne et al., 2015]. Moreover, amyloid proteins exist in both intracellular and extracellular spaces, in particularly fibrillary amyloid plaques are present in extracellular space, while PHF-tau and α-synuclein proteins are present only in intracellular space. Thus, the quantitative evaluation of amyloid PET images is less affected by specific-binding or NSB to other proteins.
Tau radiotracers, on the other hand, needs to be able to cross the blood-brain-barrier (BBB) and the cell membrane in order to reach the target. On reaching its target, tau radiotracers may bind to amyloid plaques present in the extracellular space. Hence, tau radiotracers require high selectivity of tau over amyloid. A simulation study showed that the selectivity of tau over amyloid needs to be over 20 times in order to accurately discriminate the specific binding to tau from that of amyloid [Schafer et al., 2012].
25 Off-Target Binding
Candidate radiotracers may bind non-selectively to off-targets [Bittner et al., 2017]. Off-target refers to receptors or enzymes that the radiotracers binds specifically to but are not the target of interest. These off-target sites can be observed if the radiotracer shows uptake in brain regions that are devoid of the target of interest. For example, although tau radiotracers also show binding to amyloid proteins due to similar structural conformation, tau radiotracers also show binding or high uptake in regions devoid of tau and amyloid. These regions are known as off-targets.
[18F]T807 was reported to show off-target binding in the midbrain, vessels, iron-associated regions (e.g. basal ganglia), substantia nigra, calcifications in the choroid plexus, and leptomeningeal melanin [Lowe et al., 2016]. Similarly, [11C]PBB3 was reported to accumulate in the venous sinuses [Maruyama et al., 2013] and [18F]THK5351 was reported to accumulate in the basal ganglia [Harada et al., 2016].
Depending on the region of off-target binding, the effects of off-target binding may not limit PET quantification due to little or no anatomical overlap of off-target regions with the target regions of interest (ROIs). Accurate PET quantification is also less affected if the radiotracer has high target selectivity or if the concentrations of off-target binding sites are much lower compared to that of the target [Bittner et al., 2017]. Possible binding to off-targets are difficult to predict and systematic screening is required to determine the binding of the candidate compound to a wide range of proteins. This will increase the time and cost of compound screening.
Cerebellar Binding
Standardised uptake values ratio (SUVR) is a semi-quantitative method of PET images and is commonly used for evaluation in amyloid and tau imaging (section 3.2.1). The cerebellum is often used as a reference region using SUVR in amyloid PET quantification, due to the low concentration of amyloid in the cerebellum. However, amyloid may accumulate in the cerebellum in the late stages of AD and this may lead to small changes in SUVR values in quantitative longitudinal studies. White matter region was reported to lead to more accurate SUVR quantification than cerebellum [Landau et al., 2015; Chen et al., 2015], whereby longitudinal increase in SUVR was observed using white matter as reference region but not with cerebellum, in MCI and AD [Chen et al., 2015]. The method used for PET quantification is thus important in evaluating the diagnostic capability of the radiotracers.
26
Reduced Cerebral Blood Flow
Cerebral blood flow (CBF) measurement was proposed as a biomarker for AD in the 1990s as reduced CBF were consistently observed in the temporoparietal regions in AD subjects [Jagust et al., 1997]. However, great variations were observed both within subjects, across the different brain regions, and across different subjects. There were also complications that reduced CBF was due to reduced glucose consumption in the brain. This was later shown to be due to the underlying AD pathological changes, which led to increased oxygen extraction fraction to maintain the cerebral metabolic needs for the brain processes [Nagata et al., 1997]. Yet at the same time, CBF was reported to be preserved despite changes in blood pressure in AD population [van Beek et al., 2012]. Thus far, conflicting changes in CBF and oxygen consumption in early reports and later reports have been reported [Jagust et al., 1997].
Changes in blood flow were reported to affect PET quantification in terms of SUVR in longitudinal studies in AD, as the target and reference regions may have different rates of change in CBF [canBerck et al., 2013; Cselenyi et al., 2015]. As such, distribution volume ratio (DVR) (section 3.2.3) was proposed as a more consistent method of quantification of PET images than SUVR.
However, simulation studies showed that the CBF-dependent component in SUVR quantification was small and could only explain about 1.5% reduction in longitudinal SUVR measurements [Cselenyi et al., 2015]. In these studies, the cerebellum was chosen as the reference region, which might be subjected to changes in amyloid load in dementia subjects. The use of white matter region as the reference region yielded an increase in SUVR quantification while cerebellum resulted in a decrease in SUVR values in longitudinal studies [Chen et al., 2015]. As such, changes in CBF will not significantly affect PET quantification using SUVR.
Non-Specific White Matter Retention
White matter retention led to inaccuracies in cortical SUVR measurements [Landau et al., 2014;
Villemagne et al., 2012]. The retention was said to occur due to slower white matter clearance compared to gray matter regions [Vandenberghe et al., 2010; Heurling et al., 2015; Villemagne et al., 2012]. Slower washout may be due to the lipophilicity of the radiotracers or due to non-specific binding to myelin sheath [Vandenberghe et al., 2010; Furumoto et al., 2013]. White matter retention was also said to be affected by the spill-over of higher gray matter uptake on neighbouring white matter regions [Landau et al., 2014]. Although some reported higher white matter retention using 18F-labelled radiotracers[Landau et al., 2014; Vandenberghe et al., 2010], others reported no difference in white matter retention from their 11C-equivalent compounds
27 [Shidahara et al., 2015]. However, some studies have showed that the amount of white matter retention was independent on the amount of amyloid load present in the subjects as supported by the lack of differences in white matter retention between HC and AD [Vandenberghe et al., 2010;
Furumoto et al., 2013; Cselenyi et al., 2012; Villemagne et al., 2012].
Studies have also reported that white matter retention did not limit the quantification of cortical uptake of some amyloid PET radiotracer [Vandenberghe et al., 2010; Barthel et al., 2011;
Villemagne et al., 2012]. The SUVR measurements in the white matter do not correlate with cortical SUVR measurements and white matter modifications may be due to normal ageing or other diseases [Nemmi et al., 2014]. Similarly, although PHF-tau is also found in high concentrations in subcortical white matter region in AD [Villemagne et al., 2012], quantification of tau PET images was not shown to be limited.
Metabolites Crossing BBB
The parent radiotracer is metabolised in the body and its resulting metabolites may cross the BBB.
In such cases, the presence of metabolites will affect PET quantification if the radioisotope is attached on the metabolite, as the measured PET signal comes from both the parent and the metabolites. Moreover, the binding selectivity and affinity of the metabolite to the target binding sites may be different from that of the parent. Metabolites analysis is thus important for any new radiotracers to ensure accurate quantification of PET signal. Thus far, only the metabolite of one clinically-applied amyloid PET radiotracer, [18F]FDDNP was reported to cross the BBB [Yaqub et al., 2009]. For tau radiotracer, radiolabeled metabolites of [11C]PBB3 [Kimura et al., 2015] was reported to cross the BBB. Although metabolite analysis is important, reliable methods of predicting possible metabolites crossing the BBB has yet been identified or proposed.