PET Imaging of Neuroinflammation in Alzheimer’s Disease

Neuroinflammation play an important role in Alzheimer’s disease pathogenesis. Advances in molecular imaging using positron emission tomography have provided insights into the time course of neuroinflammation and its relation with Alzheimer’s disease central pathologies in patients and in animal disease models. Recent single-cell sequencing and transcriptomics indicate dynamic disease-associated microglia and astrocyte profiles in Alzheimer’s disease. Mitochondrial 18-kDa translocator protein is the most widely investigated target for neuroinflammation imaging. New generation of translocator protein tracers with improved performance have been developed and evaluated along with tau and amyloid imaging for assessing the disease progression in Alzheimer’s disease continuum. Given that translocator protein is not exclusively expressed in glia, alternative targets are under rapid development, such as monoamine oxidase B, matrix metalloproteinases, colony-stimulating factor 1 receptor, imidazoline-2 binding sites, cyclooxygenase, cannabinoid-2 receptor, purinergic P2X7 receptor, P2Y12 receptor, the fractalkine receptor, triggering receptor expressed on myeloid cells 2, and receptor for advanced glycation end products. Promising targets should demonstrate a higher specificity for cellular locations with exclusive expression in microglia or astrocyte and activation status (pro- or anti-inflammatory) with highly specific ligand to enable in vivo brain imaging. In this review, we summarised recent advances in the development of neuroinflammation imaging tracers and provided an outlook for promising targets in the future.


TSPO Imaging
TSPO is expressed mainly in the outer mitochondrial membrane of steroid-synthesizing cells in the central nervous system (microglia, astrocytes, endothelial cell, etc.) ( Figures 1A, B) and in the peripheral (191). TSPO is involved in many physiological processes including transporting cholesterol into mitochondria, steroid hormone synthesis, and bioenergetics (191,192). Upregulation of TSPO was found in patients with AD and in animal models of AD (92,193).

The First Generation TSPO Tracers
The first-generation tracers exemplified with [ 11 C]PK-11195 have been widely used in preclinical and clinical studies. However, [ 11 C] PK-11195 suffers from several major limitations such as low permeability of the blood-brain barrier and high non-specific plasma binding, leading to a low signal-to-noise ratio in the final reconstructed PET images (194). Careful analysis of plasma metabolites is required to determine the accurate arterial input function for quantitative PET measurement (195). Increased [ 11 C] PK11195 is reported to be associated with Ab accumulation in patients with MCI and AD compared to healthy controls, correlating with the deficits in functional network connectivity, grey matters atrophy, and cognitive decline (37)(38)(39)196). Using [ 11 C]PK11195, recent studies have showed a biphasic trajectory of inflammation with an early microglial activation with increasing Ab load and a later decline when Ab load reaching plateau (AD) levels (40 (45, 46, 52, 61-63, 66, 69-71, 83, 84, 197) (Table 1). However, the binding affinities of second generation TSPO tracers in human brain differ based on the rs6971 polymorphisms, which introduces higher variability between subjects (45, 46, 52, 61-63, 66, 69-71, 197). In addition, the [ 11 C] PBR28 binding appears to be affected by chromosome 1 variant rs2997325 on microglial activation (198 (104). The cellular location of the signal is another major concern for TSPO ligands. Two different binding sites on glial and vascular TSPO were reported for several TSPO ligands, e.g., [

Emerging Targets
Given that TSPO is not exclusively expressed in glia, it is thus imperative to search for new imaging biomarkers that can detect neuroinflammation with higher sensitivity and specificity. Promising targets should have almost exclusive expression in microglia or astrocyte and highly specific ligands to enable in vivo imaging evaluations (32, 170,201,202).

Colony-Stimulating Factor 1 Receptor
CSF1R is expressed mainly on microglia and on infiltrating macrophages/monocytes and dendritic cells in the brain ( Figures 1A, B). CSF1R is important for microglia growth, proliferation, and survival. Two endogenous ligands, the growth factors colony stimulating factor-1 and interleukin-34 (203), have been reported for CSF1R. Upregulation in CSF1R have been reported in response to injury and AD-related neuropathology (204,205    CSF1R with higher sensitivity, associated with increased TSPO pattern in the brain (64) (Figures 1C-E).

Cyclooxygenase-1 and Cyclooxygenase-2
Cyclooxygenase (COX) is an enzyme involved in the production of prostaglandin H2, which is the substrate for molecules including prostaglandins, prostacyclin, and thromboxanes (206). The two isoforms COX-1 and COX-2 are considered to be involved in the neuroinflammation in neurodegenerative diseases including AD. Immunochemical evidence showed that COX-1 and COX-2 are expressed in microglia and neuron in the central nervous system (207).

Cannabinoid Receptor Type 2
Cannabinoid receptor type 2 (CB 2 R) are mainly expressed by immune cells including monocytes, macrophages, and microglia in the brain (151,152) and have low expression levels under physiological conditions (2,4,31 Table 1). Upregulation of brain CB 2 R expression has been demonstrated in acute inflammation such as LPS-injected model and murine stroke model (151)(152)(153) in chronic inflammation senescence-accelerated models (155) and in amyloidosis mouse model associated with Ab deposits (150

Purinergic P2X7 Receptor and P2Y12 Receptor
The expression of purinergic P2X7 receptor is found upregulated specifically in M1 microglia. P2X7 receptor mediates NLRP3 inflammasome activation, cytokine and chemokine release, T lymphocyte survival and differentiation, transcription factor activation, and cell death (213). Microglia monitors and protects neuronal function through purinergic P2Y12 receptor-dependent junctions (214) linked with neuronal mitochondrial activity. Brain injury-induced changes at somatic junctions triggered P2Y12receptor-dependent microglial neuroprotective effect, regulating neuronal calcium load and functional connectivity (215,216). Immunohistochemical staining indicated that the levels of P2Y12 receptor were decreased in the brains derived from patients with multiple sclerosis and AD cases (217 (120,122,123). Maeda et al. showed a distinct response of P2Y12 receptor to tau and amyloid deposits using P2Y12 receptor tracer [ 11 C]AZD1283. The levels of P2Y12 receptor decline in tau-laden region with increased total level of microglia in rTg4510 and PS19 tau mice and increase in APP23 and APP NL-F/NL-F mice (123). However PET imaging using [ 11 C]AZD1283 showed no uptake signal in the wild-type mouse brain. Two other tracers [ 11 C] P2Y12R-ant and [ 11 C]5 have showed sufficient brain uptake and promising results in experimental autoimmune encephalomyelitis model of multiple sclerosis (120) and stroke model for detecting antiinflammatory microglia (122).
[ 18 F]fluorodeprenyl-D 2 showed favorable kinetic properties with relatively fast washout from non-human primate brain and improved sensitivity for MAO-B imaging (165). However, the technical challenges of irreversible inhibitors such as deprenyl hinder the accurate image analysis. Several reversible-binding inhibitors have been developed in recent years such as [ 11 (Figures 2A-C). Livingston et al. demonstrated that increased astrocytosis assessed by [ 11 C]BU99008 in regions of earlier stages with low Ab loads assessed by [ 18 F]florbetaben and reduced astrocytosis in regions of advanced stage with greater Ab load and atrophy (177). In vitro autoradiography and immune-histochemical staining showed the specificity of [ 3 H]BU99008 and the colocalization of with glial fibrillary acidic protein staining of astroctyes in brain tissues from patients with AD.

DISCUSSION
Non-invasive detection of central pathologies is indispensable for understanding the mechanism underlying AD continuum and for facilitating early and differential diagnosis (28, [222][223][224][225]. TSPO-PET is still the most powerful imaging tool for AD-associated neuroinflammation but is currently facing two challenges. First, a human TSPO polymorphism TSPO rs6971 commonly affects the binding affinities of the second generation tracers to a different extent. Classification with polymorphism enables to correct the variability and bias from different binding affinities, but it raises the threshold for sample size of human subjects. Third-generation tracers have been developed for circumventing this limitation. In vitro testing in post-mortem human brain tissues have demonstrated the insensitivity of [ 11 C]GE-180, [ 11 C]GE-387, and [ 11 C]ER176 to TSPO polymorphism (75,106,197). However, recent clinical study with [ 11 C]ER176 (105) and [ 11 C]GE-180 (88) demonstrated a significant decrease in ligand retention in lowaffinity binders, suggesting the necessity of further in vivo examination. Second, the heterogenous cellular sources of TSPO PET tracers have been demonstrated in astrocytes, endothelial cells, and vascular smooth muscle cells, in addition to microglia in both patients with AD and animal models (61,85,86,193,(226)(227)(228)(229) ( Figures 1A, B). Although conventional opinions consider microglia as major cellular source of TSPO in the central nervous system, latest study finds vascular TSPO provides major binding sites for TSPO ligands including most widely used [ 11 C]PK11195 and [ 11 C]PBR28 in normal mouse brains (57). These findings suggest the possibility that changes in TSPO PET signal may be partly due to changes in the levels of vascular TSPO and not purely of glial TSPO. [ 18 F]FEBMP and [ 11 C]AC-5216 showed relatively selectivity for glial-TSPO compared to other ligands such as [ 11 C] PK11195 (200). It remains to be investigated whether the third generation of TSPO tracers shows a portion of vascular TSPO detection similarly. Moreover, further research on next generations of TSPO tracers are needed, with the selection criteria including optimal binding property, insensitivity for TSPO polymorphism, and high glial TSPO selectivity.
The role of neuroinflammation in AD pathogenesis is still not fully elucidated. Early clinical studies with first generation tracer [ 11 C]PK11195 showed conflicting results in the brains from AD patients. Some studies demonstrated significant increases in [ 11 C] PK11195 retention in diseased brain regions in AD (230,231), which was not observed in some other studies (232,233). Albrecht et al. recently reported negative associations between regional Ab and tau PET uptake and CSF inflammatory markers in patients with AD and in non-demented controls and suggested a protective role of neuroinflammation (234). Ewers et al. showed that a higher CSF level of soluble TREM2 is indicative of microglia activation in patients with AD. The CSF level of TREM2 negatively aassociated with the rate of Ab accumulation assessed by using [ 18 F]florbetapir over 2-years follow-up in AD patients (101). Biphasic trajectory with an early increase and a later decline in the level of microglial activation might explain such inconsistency between results from clinical studies (62). The recently reported biphasic trajectory of astrocytosis (177) adds further complexity in the interpretation.
A recent study has showed that microglia is involved in the formation of senile plaque by promoting the diffuse form converting to dense cored form (15). In vitro immunohistochemical analysis found that TSPO-positive microglia were surrounded dense cored plaque, not diffuse plaques (235). These results may explain the complex spatial association between TSPO-PET and amyloid-PET signals. [ 11 C]PBR28 signal correlated with both tau aggregation and Ab deposition (55), suggesting distinct dynamic profiles of microglial activation. Collectively, current clinical studies have not provided a consensus on association between TSPO-associated neuroinflammation and AD-pathological changes. Given the different binding sites in glial and vascular TSPO for different tracers, the divergent results using different TSPO-PET tracers are not unexpected. A multitracer imaging paradigm for detecting the regional patterns of Ab, tau, and microglia activation and astrocytosis is expected to provide better temporal and spatia mapping of disease processes and assessment of immunomodulatory therapeutic interventions in clinical study.
Several promising targets and tracers for neuroinflammation imaging have been reported but not yet been evaluated in AD patients or animal models, such as the ligands for inducible nitric oxide synthase ( (184). More preclinical and clinical evidence are required to indicate the utilities of these emerging ligands in in vivo imaging. An almost exclusive expression of CSF1R and P2X7 receptor and P2Y12 receptor in microglia have demonstrated their potentials as nextgeneration imaging targets for microglia activation. Further evaluation of these tracers in amyloidosis and tauopathy models and patients with MCI and AD will potentially facilitate better phenotyping of microglia activation. The association of these emerging targets with AD pathologies, disease progression, and the improvement in the ligand binding properties and analysis methods for PET data require further investigations (236). With the advances in new techniques, e.g., single-cell analysis of neuroinflammatory responses and plasma biomarkers, the link between neuroinflammation PET with other indicators will likely be studied in a more systematic manner.