Aβ Imaging in Aging, Alzheimer’s Disease, and Other Neurodegenerative Conditions

Abstract

In vivo imaging of β-amyloid (Αβ) has transformed the assessment of Αβ pathology and its changes over time, extending our insight into Aβ deposition in the brain by providing highly accurate, reliable, and reproducible quantitative statements of regional or global Aβ burden in the brain, proving essential in anti-Αβ therapeutic trials. Although cross-sectional evaluation of Αβ burden does not strongly correlate with cognitive impairment in AD, it does correlate with memory impairment and a higher risk for cognitive decline in the aging population and MCI subjects. This correlation with memory impairment, one of the earliest symptoms of AD, suggests that Αβ deposition is not part of normal aging. Longitudinal observations, coupled with different disease-specific biomarkers to assess potential downstream effects of Aβ, have confirmed that Αβ deposition in the brain starts decades before the onset of symptoms. Aβ imaging studies continue to refine our understanding of the role of Αβ deposition in Alzheimer’s disease.

1 Introduction

Alzheimer’s disease (AD) is a progressive and irreversible neurodegenerative disorder clinically characterized by memory loss and cognitive decline that severely affect the activities of daily living (Masters et al. 2006). AD is the leading cause of dementia in the elderly, leading invariably to death, usually within 7–10 years after diagnosis (Khachaturian 1985). The progressive nature of the neurodegeneration suggests an age-dependent process that ultimately leads to synaptic failure and neuronal damage in cortical areas of the brain essential for memory and other cognitive domains (Isacson et al. 2002). AD not only has devastating effects on the sufferers and their caregivers, but it also has a tremendous socioeconomic impact on families and the health system, a burden which will only increase in the upcoming years as the population of most countries ages (Johnson et al. 2000). In the absence of reliable biomarkers, direct pathologic examination of brain tissue derived from either biopsy or autopsy remains the only definitive method for establishing a diagnosis of AD (O'Brien et al. 2000). The typical macroscopic picture is gross cortical atrophy, while, microscopically, there are widespread cellular degeneration and diffuse synaptic and neuronal loss, accompanied by reactive gliosis and the presence of the pathological hallmarks of the disease: intracellular neurofibrillary tangles (NFT) and extracellular amyloid plaques (Jellinger 1990; Masters 2005; Masters and Beyreuther 2006). While NFT are intraneuronal bundles of paired helical filaments mainly composed of the aggregates of an abnormally phosphorylated form of tau protein (Jellinger and Bancher 1998; Michaelis et al. 2002), neuritic plaques consist of dense extracellular aggregates of β-amyloid (Αβ) (Masters et al. 1985), surrounded by reactive gliosis and dystrophic neurites. Αβ is a 4 kDa 39–43 amino acid metalloprotein derived from the proteolytic cleavage of the amyloid precursor protein (APP), by β- and γ-secretases (Cappai and White 1999). To date, all available genetic, pathological, biochemical, and cellular evidence strongly supports the notion that an imbalance between the production and removal of Αβ leading to its progressive accumulation is central to the pathogenesis of AD (Villemagne et al. 2006). The “Aβcentric theory” (Masters et al. 2006) postulates that Aβ plaque deposition is the primary event in a cascade of effects that lead to neurofibrillary degeneration and dementia (Hardy 1997).

Despite new diagnostic criteria (McKhann et al. 2011), clinical diagnosis of AD is still largely based on progressive impairment of memory and decline in at least one other cognitive domain and by excluding other diseases that might also present with dementia such as frontotemporal lobar degeneration (FTLD), dementia with Lewy bodies (DLB), stroke, brain tumor, normal pressure hydrocephalus, or depression (Larson et al. 1996; Cummings et al. 1998). A variable period of up to 5 years of prodromal decline in cognition characterized by a relatively isolated impairment in short-term memory that may also be accompanied by impairments of working memory, known as amnestic mild cognitive impairment (MCI), usually precedes the formal diagnosis of AD (Petersen et al. 1999; Petersen 2000). At this point, there is no cure for AD nor proven way to slow the rate of disease progression. Symptomatic treatment with acetylcholinesterase inhibitors or a glutamatergic moderator provides modest benefit in some patients usually by temporary stabilization rather than a noticeable improvement in memory function (Masters et al. 2006).

2 Aβ Imaging Radiotracers

Αβ plaques and NFTs are the pathological hallmark brain lesions of AD. These microscopic aggregates are still well beyond the resolution of conventional neuroimaging techniques used for the clinical evaluation of patients with AD. Positron emission tomography (PET) is a sensitive molecular imaging technique that allows in vivo quantification of radiotracer concentrations in the picomolar range, allowing the non-invasive assessment of molecular processes at their sites of action, detecting disease processes at asymptomatic stages when there is no evidence of anatomic changes on computed tomography (CT) and magnetic resonance imaging (MRI) (Phelps 2000). Since Αβ is at the center of AD pathogenesis, and given that several pharmacological agents aimed at reducing Αβ levels in the brain are being developed and tested, many efforts have been focused on generating radiotracers for imaging Αβ in vivo (Sair et al. 2004; Mathis et al. 2005; Villemagne et al. 2005). Several applications of Aβ imaging have been proposed (Table 10.1).

Table 10.1 Applications of Aβ imaging

As a quantitative neuroimaging probe, the Aβ radiotracer must possess a number of key general properties: it should be a lipophilic, non-toxic small molecule with a high specificity and selectivity for Αβ, amenable for high specific activity labeling with ¹⁸F or other long-lived radioisotope, with no radiolabeled metabolites that enter the brain, while reversibly binding to Αβ in a specific and selective fashion (Laruelle et al. 2003; Nordberg 2004; Pike 2009). Overall, binding affinity and lipophilicity are the most crucial properties for in vivo neuroimaging tracers. While high affinity for Αβ is desirable to provide an adequate signal-to-noise ratio, this high affinity might also delay reaching binding equilibrium requiring an extended scanning time. And while lipophilicity is necessary for the tracer to cross the blood-brain barrier (BBB), if the radiotracer is too lipophilic, its non-specific binding might be too high (Laruelle et al. 2003; Pike 2009).

Through the years, several compounds have been evaluated with different fortune as potential Aβ probes: monoclonal and anti-Aβ antibodies fragments, serum amyloid P, basic fibroblast growth factor, Aβ fragments, and derivatives of histopathological dyes such as Congo red, Chrysamine-G, and Thioflavin S and T (for review, see (Villemagne & Rowe 2010)). While selective tau imaging for in vivo NFT quantification is still in the early stages of development (Okamura et al. 2005; Maruyama et al. 2009; Ojida et al. 2009; Fodero-Tavoletti et al. 2011), several Αβ radiotracers found their way into human clinical trials.

Almost a decade after unsuccessful trials with anti-Αβ antibodies (Majocha et al. 1992), Αβ imaging came to fruition with the first report of successful imaging in an AD patient with ¹⁸F-FDDNP, a tracer claimed to bind both plaques and NFT (Shoghi-Jadid et al. 2002). Since then, human Αβ imaging studies have been conducted in AD patients, normal controls, and patients with other dementias using ¹¹C-PiB (Klunk et al. 2004), ¹¹C-SB13 (Verhoeff et al. 2004), ¹¹C-ST1859 (Bauer et al. 2006), ¹¹C-BF227 (Kudo et al. 2007), ¹¹C-AZD2138 (Nyberg et al. 2009), ¹⁸F-florbetaben (Rowe et al. 2008), ¹⁸F-flutemetamol (Serdons et al. 2009a; b), ¹⁸F-florbetapir (Wong et al. 2010), and ¹⁸F-NAV4694—a.k.a. AZD4694 (Cselenyi et al. 2012; Rowe et al. 2013a; b)—with PET.

In order to test compounds that could have more widespread application, preliminary studies with SPECT Aβ radiotracers ¹²³I-CQ (Opazo et al. 2006) and ¹²³I-IMPY (Newberg et al. 2006) showed limited utility for the evaluation of Aβ burden in AD, although ¹²³I-IMPY might be useful in the evaluation of transmissible spongiform encephalopathies (Song et al. 2008). New SPECT radiotracers labeled with ¹²³I or those that could be potentially labeled with ^99mTc are being evaluated (Qu et al. 2007a; b; Serdons et al. 2007; Lin et al. 2009).

While all of the aforementioned tracers bind with varying degrees of success to Αβ fibrils and brain homogenates of AD patients, Congo Red and Thioflavin T—and some of their derivatives—have recently been shown to also bind to soluble oligomeric forms of Αβ (Maezawa et al. 2008). On the other hand, Αβ-soluble species represent less than 1% of the total brain Αβ (McLean et al. 1999; Roberts et al. 2017), and the reported affinity of PiB for these soluble oligomers seems to be significantly lower than for Αβ fibrils (Maezawa et al. 2008). Until highly selective radiotracers are developed to bind the Αβ-soluble species, the contribution of these oligomers to the PET signal in sporadic AD from tracers such as ¹¹C-PiB is considered to be negligible (Mathis et al. 2007).

2.1 ¹¹C-Labeled Radiotracers

2.1.1 ¹¹C-PiB

¹¹C-PiB (PiB) was the most successful and, early on, the most widely used of available Αβ tracers. PiB has been shown to possess high affinity and high selectivity for fibrillar Αβ in plaques and in other Αβ-containing lesions (Klunk et al. 2001; Mathis et al. 2002; Mathis et al. 2005; Price et al. 2005; Ye et al. 2005; Cohen et al. 2012). In vitro studies with high specific activity ³H-PiB demonstrated two binding sites for PiB (Klunk et al. 2005), and in in vitro studies, it has been used to establish the different binding sites in frontal cortex and hippocampus (Ni et al. 2013). Recent studies have shown that ¹¹C-PiB binds with high affinity to the N-terminally truncated and modified Αβ and AβN3-pyroglutamate species in senile plaques (Maeda et al. 2007). In vitro assessment of ³H-PiB binding to white matter homogenates failed to show any specific binding (Fodero-Tavoletti et al. 2009a; b). PiB is a derivative of Thioflavin T, a fluorescent dye commonly used to assess fibrillization into β-sheet conformation (LeVine 3rd 1999), and as such PiB has been shown to bind to a range of additional Αβ-containing lesions, including diffuse plaques and cerebral amyloid angiopathy (CAA) (Lockhart et al. 2007), as well as to Αβ oligomers—albeit with lower affinity (Maezawa et al. 2008). PiB also displayed lower affinity toward other misfolded proteins with a similar β-sheet secondary structure such as α-synuclein (Fodero-Tavoletti et al. 2007; Ye et al. 2008) and tau (Lockhart et al. 2007; Ikonomovic et al. 2008). This is relevant particularly since AD has been described as a “triple brain amyloidosis” (Trojanowski 2002). Most importantly, these studies have shown that, at the concentrations achieved during a PET scan, ¹¹C-PiB cortical retention in AD or DLB primarily reflects Αβ-related cerebral amyloidosis and not binding to Lewy bodies (LB) or NFT (Klunk et al. 2003; Fodero-Tavoletti et al. 2007; Lockhart et al. 2007; Ikonomovic et al. 2008). ¹¹C-PiB has consistently provided quantitative information on Αβ burden in vivo, contributing new insights into Αβ deposition in the brain, allowing earlier detection of AD pathology (Klunk et al. 2004; Mintun et al. 2006; Rowe et al. 2007; Cohen et al. 2012) and accurate differential diagnosis of the dementias (Ng et al. 2007a; b; Rabinovici et al. 2007; Rowe et al. 2007).

2.1.2 ¹¹C-BF227

¹¹C-BF-227 (BF227), a benzoxazole derivative, is also a promising tool as an Aβ imaging tracer (Okamura et al. 2004). In vitro binding studies with BF-227 demonstrated a K_i value of 4.3 nM for the compound to Aβ_1–42 fibrils (Kudo 2006). A PET study using BF-227 showed AD patients were clearly distinguishable from age-matched controls with AD patients displaying significantly higher tracer retention in the cerebral cortex than controls (Kudo et al. 2007). Regional parametric analysis of the images further demonstrated a higher BF-227 retention in the posterior association cortex in AD patients. Unlike PiB, BF-227 presents with similar binding affinities for Aβ and α-synuclein (Fodero-Tavoletti et al. 2009a; b).

A few other C-11 Aβ radiotracers have been tested in humans. The evaluation of the stilbene derivative, ¹¹C-SB13, showed that it could differentiate between five AD and six controls but had a lower effect size values when compared with ¹¹C-PiB (Verhoeff et al. 2004). A preliminary study with the anti-Aβ agent ¹¹C-ST1859 revealed small differences in radiotracer retention between nine AD patients and three healthy controls, but showed that the biodistribution and specificity of therapeutic agents can be assessed with PET (Bauer et al. 2006). Another Aβ radiotracer, ¹¹C-AZD2138, not only showed reversible binding but also displayed very low non-specific binding to white matter (Johnson et al. 2009; Nyberg et al. 2009).

2.2 ¹⁸F-Labeled Radiotracers

Unfortunately, the 20-min radioactive decay half-life of carbon-11 (¹¹C) limits the use of ¹¹C-PiB to centers with an on-site cyclotron and ¹¹C radiochemistry expertise, making the cost of studies prohibitive for routine clinical use. To overcome these limitations, several Aβ tracers labeled with fluorine-18 (¹⁸F; half-life of 110 min) that permits centralized production and regional distribution have been developed and tested. Recent studies with the newly introduced ⁸F-labeled Aβ-specific radiotracers, ¹⁸F-florbetapir (Wong et al. 2010; Clark et al. 2011), ¹⁸F-florbetaben (Rowe et al. 2008; Barthel et al. 2011; Villemagne et al. 2011a; b), ¹⁸F-flutemetamol (Nelissen et al. 2009; Vandenberghe et al. 2010), and ¹⁸F-NAV4694 (Rowe et al. 2013a; b), have been successfully replicating the results obtained with ¹¹C-PiB. The latest entry into the list of fluorinated Aβ radiotracers is ¹⁸F-MK-3328, developed by Merck, that has the disadvantage that >17% of its cortical signal was attributed to binding to monoamine oxidase-B (MAO-B) (Sur et al. 2010).

2.2.1 ¹⁸F-FDDNP

A marked progression in the development of Aβ imaging tracers was the synthesis and characterization by Barrio and colleagues of a very lipophilic radiofluorinated 6-dialkylamino-2-naphthyethylidene derivative that presents nanomolar affinity to Aβ fibrils (Barrio et al. 1999; Agdeppa et al. 2001a; b). ¹⁸F-FDDNP is reported to bind both the extracellular Aβ plaques and the intracellular NFT in AD (Shoghi-Jadid et al. 2002) while also binding to prion plaques in Creutzfeldt-Jakob disease (CJD) brain tissue (Bresjanac et al. 2003). However, in vitro evaluation of FDDNP in concentrations similar to those achieved during a PET scan showed limited binding to both NFT and Aβ plaques (Thompson et al. 2009). ¹⁸F-FDDNP was used to obtain the first human PET images of Aβ in an 82-year-old woman with AD. AD patients present with higher accumulation and slower clearance of ¹⁸F-FDDNP than controls in brain areas such as the hippocampus (Shoghi-Jadid et al. 2002). Retention time of ¹⁸F-FDDNP in these brain regions was correlated with lower memory performance scores, regional glucose hypometabolism, and brain atrophy (Small et al. 2002; Small et al. 2006). However, the dynamic range of ¹⁸F-FDDNP cortical uptake values between HC and AD patients is small (9%). Direct comparison of ¹⁸F-FDDNP with ¹¹C-PIB in monkeys (Noda et al. 2008) and in human subjects showed very limited dynamic range of ¹⁸F-FDDNP (Shih et al. 1987; Tolboom et al. 2009a; b) and in a longitudinal study was found to be less useful than ¹¹C-PiB and ¹⁸F-fluorodeoxyglucose (FDG) for examining disease progression (Ossenkoppele et al. 2012). ¹⁸F-FDDNP has also been shown to bind to pure tauopathies, such as patients with progressive supranuclear palsy (Kepe et al. 2013) and in individuals suspected of chronic traumatic encephalopathy (Barrio et al. 2015; Chen et al. 2018; Omalu et al. 2018). Interestingly, ¹⁸F-FDDNP still remains the only Aβ tracer showing retention in the medial temporal cortex of AD patients (Shih et al. 1987) and correlation with cerebrospinal fluid (CSF) tau (Tolboom et al. 2009a; b).

2.2.2 ¹⁸F-NAV4694

Developed by AstraZeneca, the most salient features of ¹⁸F-NAV4694 are, as with ¹¹C-AZD2138, fast tracer kinetics and low non-specific binding to white matter, similar to that one observed with ¹¹C-PiB, which is helpful for the detection of small Aβ cortical deposits at very early stages of the disease process (Sundgren-Andersson et al. 2009; Jureus et al. 2010). Head-to-head comparison with PIB suggests ¹⁸F-NAV4694 has a better binding profile than PiB (Rowe et al. 2013a; b; Rowe et al. 2016). Clinical studies with ¹⁸F-NAV4694 showed a clear distinction in tracer retention between healthy controls and AD patients (Rowe et al. 2013a; b; Rowe et al. 2016). To this date, ¹⁸F-NAV4694 remains the only main F-18 Aβ radiotracer that has not yet been presented to the Food and Drug Administration (FDA) for approval.

2.2.3 ¹⁸F-Florbetaben

¹⁸F-florbetaben (a.k.a. Neuraceq™; AV1; BAY94–9172), synthesized by Kung and colleagues (Zhang et al. 2005a; b), developed by Bayer Healthcare, and nowadays commercialized by Life Molecular Imaging Inc., has been shown to bind with high affinity to Aβ in brain homogenates and selectively labeled Aβ plaques and CAA in AD tissue sections (Zhang et al. 2005a; b). After injection into Tg2576 transgenic mice, ex vivo brain sections showed localization of ¹⁸F-florbetaben in regions with Aβ plaques as confirmed by thioflavin binding (Zhang et al. 2005a, b). At tracer concentrations achieved during human PET studies, florbetaben did not show binding to LB or NFT-tau in post-mortem cortices from DLB or FTLD patients (Fodero-Tavoletti et al. 2012). In human studies, cortical retention of ¹⁸F-florbetaben was higher at 90 min post-injection in all AD subjects compared to age-matched controls and FTLD patients, with binding matching the reported post-mortem distribution of Aβ plaques (Rowe et al. 2008). Recently completed phase II clinical studies further confirmed these results (Barthel et al. 2011). A longitudinal study in MCI subjects established the usefulness of FBB PET in predicting progression to AD (Ong et al. 2013; Ong et al. 2015). ¹⁸F-Florbetaben is highly correlated with ¹¹C-PiB (Villemagne et al. 2012a; b; c; d; Rowe et al. 2017) and was used to detect the presence or absence of AD pathology in the brain in subjects from a wide spectrum of neurodegenerative diseases (Villemagne et al. 2011a; b) aiding in the differential diagnosis of dementia (Ceccaldi et al. 2018) and clarifying diagnosis in patients with ambiguous CSF results (Manca et al. 2019). It has been shown that ¹⁸F-florbetaben can make accurate ante-mortem differential diagnosis of brain amyloidosis (Sabbagh et al. 2017), and phase III confirmed ¹⁸F-florbetaben is highly associated with Aβ plaque pathology (Sabri et al. 2015), leading to its approval for clinical use by the FDA. ¹⁸F-Florbetaben has been recently incorporated as an Aβ radiotracer to ADNI-3.

2.2.4 ¹⁸F-Florbetapir

As ¹⁸F-florbetaben, ¹⁸F-florbetapir is a stilbene derivative that was also synthesized by Kung and colleagues at the University of Pennsylvania (Zhang et al. 2005a, b) and has been developed by Avid Radiopharmaceuticals. Initial in vitro evaluation showed binding to Aβ plaques in AD brain sections (Choi et al. 2009; Lin et al. 2009). The most salient feature of this tracer is its rapid reversible binding characteristics allowing scanning only at 45–50 min after injection, similar to ¹¹C-PiB (Wong et al. 2010). ¹⁸F-Florbetapir (Lister-James et al. 2011) has become the most widely used Aβ radiotracer after ¹¹C-PiB and has been adopted as the Aβ radiotracer for ADNI 2 and also used in patient selection and to evaluate treatment response in a large number of multicenter therapeutic trials around the world. Several multicenter phase I and II studies in AD, MCI, and HC showed the ability of ¹⁸F-florbetapir to discriminate between AD and age-matched controls, with ~50% of the MCI subjects presenting with high ¹⁸F-florbetapir retention (Fleisher et al. 2011). Multicenter studies using ¹⁸F-florbetapir to assess the relationship between Aβ burden and cognition showed that in clinically normal older individuals, Aβ burden in the brain is associated with poorer memory performance (Sperling et al. 2012), while MCI subjects with high Aβ burden in the brain are at a significantly higher risk of cognitive decline over 18 months (Doraiswamy et al. 2012). An initial phase III study in 35 volunteers demonstrated a 96% agreement between ¹⁸F-florbetapir and neuropathology for the detection of brain Aβ in vivo and no retention in young controls (Clark et al. 2011). An extension phase III in 59 volunteers established that ¹⁸F-florbetapir has a sensitivity of 92% and a specificity of 100% for the detection of Aβ pathology (Clark et al. 2012). The use of ¹⁸F-florbetapir changes patient management (Pontecorvo et al. 2017a; b), and using a combination of visual inspection with a semiquantitative approach improves diagnostic accuracy (Pontecorvo et al. 2017a; b). No significant difference in the rates of Aβ accumulation was observed when compared against PiB in longitudinal studies (Su et al. 2019), where a slow but steady increase in florbetapir signal was observed in young controls (Gonneaud et al. 2017), which was associated with incipient cognitive decline (Bischof et al. 2016) even at low Aβ levels (Landau et al. 2018). ¹⁸F-Florbetapir (as Amyvid™) was the first radiotracer approved by the FDA for detection of Aβ in vivo and the first ¹⁸F-labeled radiotracer approved by the FDA since FDG.

2.2.5 ¹⁸F-Flutemetamol

Another fluorinated tracer, developed by GE Healthcare, that also completed a phase III study (Curtis et al. 2015) is ¹⁸F-flutemetamol (a.k.a. GE067; Vizamyl™) (Serdons et al. 2009a; b). Phase I and II studies demonstrated that ¹⁸F-flutemetamol can clearly differentiate between AD and HC (Nelissen et al. 2009; Vandenberghe et al. 2010) and that when combined with brain atrophy, it could be predictive of disease progression in MCI subjects (Thurfjell et al. 2012). ¹⁸F-Flutemetamol brain retention is highly correlated with ¹¹C-PIB (Vandenberghe et al. 2010; Mountz et al. 2015; Lowe et al. 2017) and with neuropathology (Wolk et al. 2011; Wong et al. 2012; Thal et al. 2015; Thal et al. 2018) where no difference was found in the ¹⁸F-flutemetamol regional brain distribution between MCI subjects and end-stage AD patients (Farrar et al. 2019). ¹⁸F-Flutemetamol has been shown to aid in the clinical diagnosis and management of individuals presenting with memory decline (Zwan et al. 2017; Leuzy et al. 2019). ¹⁸F-Flutemetamol has been approved for clinical use by the FDA.

All Aβ imaging radiotracers present a similar pattern of tracer retention in AD, with the highest retention in frontal, temporal, and posterior cingulate cortices, reflecting Aβ plaque burden (Fig. 10.1). While the cortical retention of ¹⁸F-florbetapir, ¹⁸F-florbetaben, and ¹⁸F-flutemetamol provides a clear separation of AD patients from HC subjects, the degree of cortical retention with ¹⁸F-florbetapir and ¹⁸F-florbetaben is lower than with ¹¹C-PiB (Villemagne et al. 2012a; b; c; d; Wolk et al. 2012a; b), showing a narrower dynamic range of SUVR values that visually appears as a relatively higher degree of non-specific binding to white matter. While the cortical retention of ¹⁸F-flutemetamol is similar to that of ¹¹C-PiB, the non-specific retention in white matter is much higher (Vandenberghe et al. 2010). Due to this lower signal-to-noise ratio, visual read-outs of the images are more challenging than with ¹¹C-PiB (Rowe and Villemagne 2011). These issues are not observed when using ¹⁸F-NAV4694, which is an Aβ tracer that yields high contrast images with relatively low tracer retention in white matter and has proven to perform slightly better than ¹¹C-PiB (Rowe et al. 2016). While ¹¹C-PiB PET and ¹⁸F-NAV4694 images in subjects with AD pathology usually clearly show high radiotracer retention in the gray matter in excess of that in subjacent white matter, these FDA-approved ¹⁸F tracers frequently show loss of the normal gray-white matter demarcation as the predominant evidence of cortical Aβ deposition (Rowe and Villemagne 2011).

Fig. 10.1

Representative sagittal (top row), transaxial (middle row), and coronal (bottom row) PET images overlaid on MRI from Alzheimer’s disease patients obtained with different Aβ imaging radiotracers. From left to right, ¹¹C-PiB, ¹⁸F-florbetaben, ¹⁸F-flutemetamol, ¹⁸F-florbetapir, and ¹⁸F-NAV4694. The images show the typical pattern of tracer retention in AD, with the highest retention in frontal, temporal, and posterior cingulate cortices, reflecting Aβ plaque burden

3 Aβ Imaging in Alzheimer’s Disease

On visual inspection, cortical retention of Aβ tracers is higher in AD, with a regional brain distribution that is highest in frontal, cingulate, precuneus, striatum, parietal, and lateral temporal cortices, while occipital, sensorimotor, and mesial temporal cortices are much less affected (Villemagne et al. 2018). Both quantitative and visual assessment of PET images present a pattern of radiotracer retention that seems to replicate the sequence of Αβ deposition found at autopsy (Braak and Braak 1997), with initial deposition in the orbitofrontal cortex, inferior temporal, cingulate gyrus, and precuneus, followed by the remaining prefrontal cortex, lateral temporal, and parietal cortices (Fig. 10.2). Multimodality studies in early AD have shown that the regional pattern of Αβ deposition is similar to the anatomy of the “default network” (Buckner et al. 2005; Sperling et al. 2009), a specific, anatomically defined brain system responsible for internal modes of cognition, such as self-reflection processes, conscious resting state, or episodic memory retrieval (Buckner et al. 2008; Wermke et al. 2008), an association that is present even in non-demented subjects (Drzezga et al. 2011). The regional distribution of Αβ tracers in the brain sometimes varies with the specific Αβ distribution characteristic of the underlying pathology, for example, carriers of mutations associated with familial AD (Klunk et al. 2007; Koivunen et al. 2008a; b; Villemagne et al. 2009a; b) and subjects with posterior cortical atrophy (Ng et al. 2007a; b; Tenovuo et al. 2008; Kambe et al. 2010; Formaglio et al. 2011) or CAA (Johnson et al. 2007; Dierksen et al. 2010) each presenting a different regional ¹¹C-PiB distribution of retention to the one usually observed in sporadic AD (Klunk et al. 2004; Rowe et al. 2007).

Fig. 10.2

Top row. Schematics showing the stages of Aβ deposition in the human brain as proposed by Braak and Braak (Braak and Braak 1997). Bottom row. Representative sagittal PET images showing the regional brain distribution of ¹¹C-PIB in two asymptomatic healthy age-matched controls, one with low (PiB- HC) and one with high Aβ burden (PiB+ HC), a subject classified as mild cognitive impairment (MCI) with significant Aβ deposition in the brain, and an Alzheimer’s disease (AD) patient with even a higher Aβ burden in the brain. The pattern of PiB retention replicates the sequence of Αβ deposition described from post-mortem studies (Braak and Braak 1997)

Cross-sectional semiquantitative PET studies using diverse Αβ radiotracers have shown a robust difference in retention between AD patients and age-matched controls (Klunk et al. 2004; Price et al. 2005; Mintun et al. 2006; Kudo et al. 2007; Pike et al. 2007; Rabinovici et al. 2007; Rowe et al. 2007; Jack Jr. et al. 2008a; b; Morris et al. 2009; Furst et al. 2010; Rowe et al. 2010; Barthel et al. 2011; Fleisher et al. 2011; Villemagne et al. 2011a; b; Camus et al. 2012; Cselenyi et al. 2012). It should be noted that about 20–30% of clinically diagnosed probable AD patients have been shown to have a low Aβ imaging scan, which effectively rules out AD pathology as the etiology of the dementia.

Longitudinal studies have shown that significant, albeit small, increases in Αβ deposition can be measured but also that these increases in Αβ deposition are present across the whole spectrum of cognitive stages, from cognitively unimpaired individuals to AD dementia patients (Engler et al. 2006; Jack Jr. et al. 2009; Okello et al. 2009a; b; Resnick et al. 2010; Rinne et al. 2010; Sojkova et al. 2011a; b; Villemagne et al. 2011a; b; Villain et al. 2012; Jack Jr. et al. 2013; Villemagne et al. 2013; Landau et al. 2015). Αβ accumulation is observed in individuals considered to have low Αβ loads with about 7% of them progressing above the threshold in about 2.5 years (Vlassenko et al. 2011), and subtle cognitive decline is observed in those subthreshold individuals that accumulate Αβ over time (Landau et al. 2018). Αβ accumulation is a slow and protracted process that precedes by about two decades the onset of the typical phenotype of AD dementia (Jack Jr. et al. 2013; Villemagne et al. 2013; Landau et al. 2015). In individuals with the highest Aβ burdens independent of clinical classification and also as Alzheimer’s dementia progresses, the rates of Aβ deposition start to slow down (Villain et al. 2012; Jack Jr. et al. 2013; Villemagne et al. 2013; Landau et al. 2015). This slower rate of Αβ deposition observed at the late stages of the disease might be attributed to a saturation of the process or, much more likely, extensive neuronal death that precludes production of Αβ.

4 Ante-mortem Post-mortem Correlations

While clinicopathological studies have shown that the accuracy of clinical assessments for the diagnosis of AD ranges between 70 and 90%, depending on the specialization of the center, the regional retention of radiotracers is highly correlated with regional Αβ plaques as reported at autopsy or biopsy (Bacskai et al. 2007; Ikonomovic et al. 2008; Leinonen et al. 2008a; b; Villemagne et al. 2009a; b; Burack et al. 2010; Clark et al. 2011; Kadir et al. 2011; Sabbagh et al. 2011; Sojkova et al. 2011a; b; Wolk et al. 2011; Clark et al. 2012; Wong et al. 2012; Curtis et al. 2015; Sabri et al. 2015), with a higher Aβ concentration in the frontal cortex than in the hippocampus, consistent with previous reports (Arnold et al. 2000; Naslund et al. 2000). As mentioned before, and in contrast with CSF studies, Αβ imaging studies can assess both the quantity and regional brain distribution of Αβ, as observed in the different patterns of regional distribution of Αβ observed in sporadic AD (Klunk et al. 2004; Nordberg 2007), familial AD (Klunk et al. 2007; Koivunen et al. 2008a; b; Villemagne et al. 2009a; b), familial British dementia (Villemagne et al. 2010), and CAA subjects (Johnson et al. 2007; Dierksen et al. 2010). The results from Αβ imaging might be considered for the brain regions to be sampled and stained for the neuropathological diagnosis of AD (Montine et al. 2012).

There have been a handful of cases reported with a discrepancy between a positive neuropathological finding or a strong biomarker profile of AD and a low PET signal (Leinonen et al. 2008a; b; Tomiyama et al. 2008; Cairns et al. 2009; Schöll et al. 2011; Sojkova et al. 2011a; b; Ikonomovic et al. 2012). For example, Tomiyama and colleagues reported a case of a novel APP mutation (E693Delta) where the mutant Αβ does not fibrillize in the same way as wild-type Aβ (Inayathullah and Teplow 2011) in which the ¹¹C-PiB study showed low cortical retention (Tomiyama et al. 2008). Schöll and colleagues report on two subject carriers of the Artic APP mutation (Schöll et al. 2011). These two mutation carriers, while presenting with low ¹¹C-PiB retention, had a clear “AD pattern” of brain hypometabolism as measured by FDG, low Aβ_1–42, and elevated t-tau and p-tau in CSF (Schöll et al. 2011). On the one hand, these cases illustrate the possibility that different conformations of Aβ deposits (Levine 3rd and Walker 2010) may affect the binding of tracers and that Aβ imaging may not recognize all types of Aβ pathologies with equal sensitivity (Walker et al. 2008; Rosen et al. 2010). On the other, it raises issues not only in regard to the sensitivity of Aβ imaging (e.g., what is the minimal amount of Aβ detected by PET?) but also in regard to the evaluation of brain tissue (e.g., type of tissue dye, type of antibodies used for immunohistochemistry and ELISA, diagnostic criteria, etc.).

Second-generation radiofluorinated Aβ imaging radiotracers underwent phase III studies comparing ante-mortem PET with post-mortem findings. All these tracers showed a very high accuracy against neuropathology, with a sensitivity of 92% and a specificity of 100% for florbetapir (Clark et al. 2012), a sensitivity of 86% and a specificity of 92% for flutemetamol (Curtis et al. 2015), and a sensitivity of 98% and a specificity of 89% for florbetaben (Sabri et al. 2015). Based on the results of these phase III studies, all three tracers received Food and Drug Administration (FDA) and European Medicines Agency (EMA) approval for clinical use, specifically to rule out the presence of Aβ-amyloid in the brain.

5 Aβ Deposition in Non-demented Individuals and Its Relation with Cognition

About 25–35% of elderly subjects performing within normal limits on cognitive tests present with high cortical ¹¹C-PiB retention, predominantly in the prefrontal and posterior cingulate/precuneus regions (Mintun et al. 2006; Rowe et al. 2007; Aizenstein et al. 2008; Villemagne et al. 2008a; b; Mormino et al. 2009; Reiman et al. 2009a; b; Rowe et al. 2010). These findings are in perfect agreement with post-mortem reports that ~25% of non-demented older individuals over the age of 75 have Aβ plaques in the brain (Davies et al. 1988; Morris and Price 2001; Forman et al. 2007). Aβ deposition in these non-demented individuals (Mintun et al. 2006; Rowe et al. 2007; Aizenstein et al. 2008; Villemagne et al. 2008a; b) might reflect preclinical AD (Price and Morris 1999). Furthermore, the prevalence of subjects with high ¹¹C-PiB retention increases each decade at the same rate as the reported prevalence of plaques in non-demented subjects in autopsy studies (Rowe et al. 2010). While cross-sectional examination of these controls essentially show no significant differences in cognition (Mintun et al. 2006; Aizenstein et al. 2008; Jack Jr. et al. 2008a; b; Villemagne et al. 2008a; b; Rowe et al. 2010), significant cognitive differences in females were reported (Pike et al. 2011). More recently, Sperling and colleagues reported that high Aβ deposition in the brain was correlated with lower immediate and delayed memory scores in otherwise healthy controls (Sperling et al. 2012). Interestingly, 23% of healthy elderly controls showed signs of neurodegeneration but no signs of Aβ pathology: This group of healthy elderly controls was defined by the presence of neurodegeneration in the absence of Aβ pathology and was proposed to have suspected non-Alzheimer’s pathophysiology (SNAP) (Jack Jr. et al. 2012). While Aβ is a strong predictor of cognitive decline, neurodegeneration alone did not differ from those controls without neurodegeneration or Aβ in the brain (Burnham et al. 2016). When this basic approach was applied to the general population, it was found that almost all 50-year-old cases had no signs of either Aβ or neurodegeneration, while by 80 years old ~75% of the cohort presented signs of one or both markers (Jack Jr. et al. 2014). The detection of Aβ pathology at the presymptomatic stage is also of crucial importance because if this group truly represents preclinical AD (Price and Morris 1999; Thal et al. 2004; Backman et al. 2005; Small et al. 2007), it is precisely the group that may benefit the most from therapies aimed at reducing or eliminating Αβ from the brain before irreversible neuronal or synaptic loss occurs (Sperling et al. 2011a; b; c). Moreover, some individuals remain cognitively intact, while others develop cognitive impairment, despite carrying the same pathological burden, so factors that convey resilience or susceptibility to Αβ should also be considered.

Individuals fulfilling criteria for MCI are a heterogeneous group with a wide spectrum of underlying pathologies (Petersen 2000; Winblad et al. 2004). About 40–60% of carefully characterized subjects with MCI will subsequently progress to meet criteria for AD over a 3–4-year period (Petersen et al. 1995; Petersen et al. 1999; Petersen 2000). Αβ imaging has proven useful in identifying MCI individuals with and without AD pathology. Approximately 50–70% of the subjects classified as MCI present with high cortical ¹¹C-PiB retention (Price et al. 2005; Kemppainen et al. 2007; Pike et al. 2007; Forsberg et al. 2008; Lowe et al. 2009; Mormino et al. 2009). While most studies found that subjects classified as non-amnestic MCI show low Αβ levels in the brain (Pike et al. 2007; Lowe et al. 2009), other reports did not (Wolk et al. 2009). More specifically, subjects classified as non-amnestic single domain MCI usually do not show Αβ deposition, consistent with a non-AD underlying pathological process (Pike et al. 2007; Villemagne et al. 2011a; b).

Subjective cognitive impairment (SCI) or subjective cognitive decline (SCD) is believed to be an early behavioral manifestation along the AD spectrum that may precede objective MCI and is associated with up to a sixfold increased risk of developing AD (Jessen et al. 2010). In order to better characterize those SCI/SCD individuals that might have confounding personality or affective factors unrelated to AD, several studies evaluated the relationship between SCI/SCD and specific markers of AD pathology. While only a few individuals with SCI/SCD develop AD within a few years (Slot et al. 2019), these individuals are likely to already harbor Aβ in the brain (Buckley et al. 2016; Perrotin et al. 2017; Hu et al. 2019), but Aβ burden has a stronger association to worries than to the severity of the cognitive complaints (Verfaillie et al. 2019).

To date, the relationship between Αβ deposition and the development of clinical symptoms is not fully understood. Although Aβ burden as assessed by PET does not correlate with measures of memory impairment in established clinical AD, it does correlate with memory impairment in MCI and healthy older subjects (Pike et al. 2007; Villemagne et al. 2008a; b). In subjects with MCI, this Aβ-related impairment is thought to be mediated through tau and hippocampal atrophy (Mormino et al. 2009; Sperling et al. 2019) while modulated by cognitive reserve (Roe et al. 2008; Rentz et al. 2010) and glucose metabolism (Cohen et al. 2009; Hanseeuw et al. 2017), although at the AD stage there is no longer correlation between glucose metabolism, cognition, and Αβ burden (Furst et al. 2010).

Longitudinal studies are helping clarify the relationship between Aβ burden and cognitive decline in healthy individuals and SCI/SCD and MCI subjects. High ¹¹C-PiB cortical retention in non-demented elderly subjects is associated with a greater risk of cognitive decline (Villemagne et al. 2008a; b; Resnick et al. 2010; Villemagne et al. 2011a; b). For example, the evaluation of either the cognitive trajectories (Villemagne et al. 2008a; b; Resnick et al. 2010) or changes in cognition prospectively (Morris et al. 2009; Villemagne et al. 2011a; b; Doraiswamy et al. 2012) shows that individuals with substantial Αβ deposition are more likely to present with cognitive decline. Aggregating the results from recently reported longitudinal studies reveals that 65% of MCI subjects with marked ¹¹C-PiB retention progressed to AD over 2–3 years, showing that MCI subjects with high Aβ deposition in the brain are >10 times more likely to progress to AD than those with low Aβ deposition, highlighting the clinical relevance and potential impact in patient management of Aβ imaging (Forsberg et al. 2008; Okello et al. 2009a; b; Wolk et al. 2009; Villemagne et al. 2011a; b; Rowe et al. 2013a; b). The observations that ¹¹C-PiB retention in non-demented individuals relates to episodic memory impairment, one of the earliest clinical symptoms of AD, and that extensive Aβ deposition is associated with a significant higher risk of cognitive decline in HC and with progression from MCI to AD emphasize the non-benign nature of Aβ deposition and support the hypothesis that Aβ deposition occurs well before the onset of symptoms, further suggesting that early disease-specific therapeutic intervention at the presymptomatic stage might be the most promising approach to either delay onset or halt disease progression (Sperling et al. 2011a; b; c).

The high prevalence of Αβ deposition in non-demented individuals raises questions in regard to Αβ toxicity and deposition. The evidence points to the fact that Αβ toxicity in the form of oligomers and Αβ deposition in the form of aggregates precede by many years the appearance of clinical symptoms (Price and Morris 1999). These neurotoxic processes eventually lead to synaptic and neuronal dysfunction, manifested as cognitive impairment, gray matter atrophy, and glucose hypometabolism (Eckert et al. 2003; Suo et al. 2004; Leuner et al. 2007). Therefore, it is highly likely that ¹¹C-PiB retention in non-demented individuals reflects preclinical AD (Mintun et al. 2006; Rowe et al. 2007; Aizenstein et al. 2008; Villemagne et al. 2008a; b). This may be associated or attributed to a different susceptibility/vulnerability to Αβ, at either a cellular (frontal neurons seem to be more resistant to Αβ than hippocampal neurons) (Roder et al. 2003; Resende et al. 2007), a regional (compensatory upregulation of choline acetyltransferase (ChAT) activity in the frontal cortex and hippocampi of MCI and early AD subjects) (DeKosky et al. 2002), or an individual or personal level, either due to a particular cognitive reserve (Mortimer 1997; Stern 2002; Kemppainen et al. 2008; Roe et al. 2008), due to differences in Aβ conformation affecting toxicity and/or aggregation (Lockhart et al. 2005; Deshpande et al. 2006; Levine 3rd and Walker 2008; Walker et al. 2008), or because an idiosyncratic threshold must be exceeded for synaptic failure and neuronal death to ensue (Suo et al. 2004). These factors would help explain why some older individuals with a significant Αβ burden are cognitively unimpaired, while others with lower Αβ burden and no genetic predisposing factors have already developed the full clinical AD phenotype. Ongoing longitudinal studies will help identify those subjects that despite substantial Aβ deposition do not develop the AD phenotype, thus allowing to isolate genetic or environmental factors that may convey resistance to Aβ (Sojkova et al. 2011a; b).

The lack of a strong association between Aβ deposition and measures of cognition, synaptic activity, and neurodegeneration in AD, in addition to the evidence of Αβ deposition in a high percentage of MCI and asymptomatic HC, suggests that Αβ is an early and necessary, though not sufficient, cause for cognitive decline in AD (Villemagne et al. 2008a; b; Rabinovici et al. 2010; Sojkova et al. 2011a; b), indicating the involvement of other downstream mechanisms, likely triggered by Αβ such as NFT formation, synaptic failure, and eventually neuronal loss. Also at play are factors as age, years of education, occupational level, and brain volume among others that, under the umbrella of “cognitive reserve,” appear to have a modulatory role between cognition and Αβ deposition (Roe et al. 2008; Chetelat et al. 2010a; b; Tucker and Stern 2011). Selective tau imaging studies are helping elucidate the bridge between Aβ deposition and the persistent decline in cognitive function observed in AD (Fodero-Tavoletti et al. 2011, Villemagne et al. 2015, Villemagne et al. 2018).

6 Relation of Aβ Imaging with Other Biomarkers

Another rapidly growing area is the exploration of the potential association between biomarkers of Aβ deposition or neurodegeneration (Wahlund and Blennow 2003; Morris et al. 2005; Sunderland et al. 2005; de Leon et al. 2006; Thal et al. 2006; Blennow et al. 2007; Shaw et al. 2007; Clark et al. 2008; Jack Jr. et al. 2010; Albert et al. 2011; McKhann et al. 2011; Storandt et al. 2012) and Aβ burden as measured by PET (Jack Jr. et al. 2008a; b; Jack Jr. et al. 2010). The new National Institute on Aging and Alzheimer’s Association (NIA-AA) Research Framework—AT(N) (Jack Jr. et al. 2016; Jack et al. 2018)—is based on a biological definition of Alzheimer’s disease and uses three kinds of biomarkers, markers of Aβ (A) and tau (T) pathology and markers of neuronal injury/neurodegeneration (N), for characterizing the physiopathology of AD. The implementation of the AT(N) classification is helping to refine prognosis in clinical practice and identify those at-risk individuals more likely to clinically progress, for their inclusion in therapeutic trials (Cummings 2019).

6.1 FDG

While the typical pattern of Aβ deposition in sporadic AD is highest in frontal, cingulate, precuneus, striatum, parietal, and lateral temporal cortices (Klunk et al. 2004), in FDG studies, the typical “AD pattern” is of temporoparietal and posterior cingulate hypometabolism with sparing of the basal ganglia, thalamus, cerebellum, and primary sensorimotor cortex (Salmon et al. 1994; Devanand et al. 1997; Coleman 2005; Jagust et al. 2007). Due to its high sensitivity (>90%) for detecting temporoparietal and posterior cingulate hypometabolism, FDG-PET has improved diagnostic and prognostic accuracy in patients with probable AD (Salmon et al. 1994; Kennedy et al. 1995; Small et al. 1995; Silverman et al. 2001; Silverman et al. 2002). A similar pattern of hypometabolism has been reported in normal elderly ApoE ε4 carriers (Reiman et al. 1996), MCI subjects (Chetelat et al. 2003, Chetelat et al. 2005, Mosconi et al. 2006), asymptomatic subjects with mutations associated with familial AD (Kennedy et al. 1995; Rossor et al. 1996), and subjects with a strong family history of AD (Mosconi et al. 2006a; b). FDG hypometabolism is correlated with cognition (Landau et al. 2009; Furst et al. 2010) and is predictive of future cognitive decline (Drzezga et al. 2003; Mosconi et al. 2004; Drzezga et al. 2005).

Visual assessment of PET images by clinicians blinded to the clinical status has demonstrated that ¹¹C-PiB was more accurate than FDG to distinguish AD from HC (Ng et al. 2007a; b) and slightly better than FDG in differentiating AD from FTLD (Rabinovici et al. 2011). Similarly, Aβ imaging outperforms FDG at identifying MCI subtypes (Lowe et al. 2009).

While some reports found no association between FDG and ¹¹C-PiB in AD (Furst et al. 2010), others found an inverse correlation between them in temporal and parietal cortices (Cohen et al. 2009), but no correlation has been shown in the frontal lobe (Klunk et al. 2004; Edison et al. 2007). A possible explanation for this dissociation is that there are compensatory mechanisms such as upregulation of ChAT in the frontal lobe of MCI individuals and mild AD patients (DeKosky et al. 2002; Ikonomovic et al. 2007) that might delay the manifestation of synaptic dysfunction; therefore, no hypometabolism is observed on FDG in the frontal lobe. Glucose metabolism in the frontal lobe eventually decreases as the disease progresses, and the characteristic temporoparietal pattern of AD becomes less apparent (Mielke et al. 1992; Herholz 1995). Conversely to what is observed in the frontal area, in AD patients ChAT is significantly decreased in the posterior cingulated gyrus (Ikonomovic et al. 2011), which is characterized by being one of the regions where marked and early Aβ deposition is associated with substantial hypometabolism (Drzezga et al. 2011; Yokokura et al. 2011).

6.2 CSF

Analysis of CSF allows simultaneous measurement of the concentration of the two main hallmarks of AD, Aβ and tau. It has been reported to be highly accurate in the diagnosis of AD as well of predicting cognitive decline (Blennow et al. 2007; de Leon et al. 2007; Hansson et al. 2007; Mattsson et al. 2009; Mattsson et al. 2011). The typical CSF profile in AD is low Aβ_1–42 and high total (t-tau) and phosphorylated tau (p-tau) (Strozyk et al. 2003; Hampel et al. 2010). Slower Aβ clearance from the brain (Bateman et al. 2006; Mawuenyega et al. 2010) is likely to lead to Aβ deposition in the brain and lower CSF Aβ_1–42 concentrations. While there is no correlation between ¹¹C-PiB retention and CSF t-tau or p-tau (Tolboom et al. 2009a; b; Forsberg et al. 2010), several studies have reported a strong inverse correlation between Aβ deposition in the brain and CSF Aβ_1–42 (Fagan et al. 2006; Fagan et al. 2007; Forsberg et al. 2008; Koivunen et al. 2008a; b; Fagan et al. 2009; Grimmer et al. 2009; Landau et al. 2013; Mattsson et al. 2017). Both high Aβ burden and low CSF Aβ_1–42 have been observed in cognitively unimpaired individuals probably reflecting Aβ deposition years before the manifestation of the AD phenotype (Mintun et al. 2006; Fagan et al. 2007; Rowe et al. 2007; Aizenstein et al. 2008; Fagan et al. 2009). Large population studies, such as ADNI, AIBL, and DIAN (Shaw et al. 2007; Apostolova et al. 2010; Li et al. 2015; McDade et al. 2018), have been crucial for the validation of Aβ imaging and CSF assessments as antecedent biomarkers of AD as well as determining the longitudinal trajectories of these fluid biomarkers (Toledo et al. 2013; Fagan et al. 2014; Wang et al. 2018), while the introduction of automated immunoassay platforms has allowed reproducibility of results across time and across centers (Rozga et al. 2017; Schindler et al. 2018; Shaw et al. 2019). Given that CSF and Aβ imaging measure different biochemical pools of Aβ (Roberts et al. 2017), the inverse relationship between them is not linear (Toledo et al. 2015). This is also reflected in how these biomarkers change over time. In contrast to the sigmoid shape curve observed for longitudinal Aβ imaging studies in the brain (Jack Jr. et al. 2013; Villemagne et al. 2013), CSF Aβ_1–42 decreases over time tend to adopt a logarithmical shape (Toledo et al. 2013).

6.3 Plasma Aβ

In recent years, there has been a revolution in the ability to measure Aβ peptides and APP fragments in plasma. Several reports demonstrate that plasma Aβ levels can predict both CSF Aβ and brain Aβ with an accuracy ~90% (Fandos et al. 2017; Ovod et al. 2017; Nakamura et al. 2018; Palmqvist et al. 2019). The introduction of these tests will have immediate effects as an inexpensive way to identify individuals at risk of developing AD, lowering the costs of screening for therapeutic AD trials by reducing number of lumbar punctures and Aβ PET scans.

6.4 MRI Volumetrics

While Aβ imaging or CSF Aβ provides information in regard to Aβ pathology, CSF tau quantification, FDG, and structural MRI provide information related to the neurodegenerative process (Jack Jr. et al. 2010). Hippocampal and cortical gray matter atrophy and ventricular enlargement are typical, albeit not specific, MRI findings in AD and MCI (Frisoni et al. 2010; Drago et al. 2011). Furthermore, it has been shown that the rates of hippocampal atrophy might be predictive of conversion from MCI to AD (Jack Jr. et al. 2008a; b).

Aβ deposition is associated with regional cerebral atrophy as measured by MRI (Archer et al. 2006; Bourgeat et al. 2010; Chetelat et al. 2010a; b; Becker et al. 2011) and correlates with the rates of cerebral atrophy (Becker et al. 2011; Tosun et al. 2011). Moreover, the data suggest that the relationship between Aβ deposition and cortical atrophy is sequential, where Aβ deposition precedes synaptic dysfunction and neuronal loss (Chetelat et al. 2010a, b, Drzezga et al. 2011, Forster et al. 2012), and then manifested as structural changes (Becker et al. 2011). While some reports did not find a correlation between atrophy trajectories and Aβ deposition (Josephs et al. 2008; Driscoll et al. 2010), most reported that those with high Aβ deposition showed a significantly higher rate of cortical atrophy compared to those with low Aβ burden (Chetelat et al. 2012; Sarro et al. 2016; Fletcher et al. 2018; Dang et al. 2019). These discrepancies might be due to degree of tau deposition in the brain, which is believed to mediate the effects of Aβ on brain volumetrics (Ossenkoppele et al. 2019; La Joie et al. 2020).

6.5 Neuroinflammation

Microglia are the primary resident immune surveillance cells in the brain and are thought to play a significant role in the pathogenesis of several neurodegenerative disorders including AD (Venneti et al. 2006). Activated microglia can be measured using radioligands for the translocator protein (TSPO), formerly known as the peripheral benzodiazepine receptor. The TSPO is upregulated on activated macrophages and microglia and has been established as a biomarker of neuroinflammation in the central nervous system. Most reports found that in AD there is no association between Aβ deposition and microglial activation measured by [¹¹C]PK11195 (Edison et al. 2008a; b; Okello et al. 2009a; b; Wiley et al. 2009) despite being an association between [¹¹C]PK11195 binding and both glucose hypometabolism and severity of dementia (Edison et al. 2008a; b; Yokokura et al. 2011). Higher TSPO tracer binding is observed in AD patients compared to elderly controls (Kreisl et al. 2018), but TSPO polymorphism affects the binding of these tracers in the brain (Kreisl et al. 2013). New developed radiotracers for activated microglia are focusing on cell membrane markers like P2X7 receptors (Berdyyeva et al. 2019; Hagens et al. 2020), as well as renewed interest on tracers binding to COX-1 and COX-2 (Kim et al. 2018). Also, PET studies assessing the relationship between Aβ deposition and neuroinflammation have been conducted using the MAO-B inhibitor ¹¹C-l-d-deprenyl as a marker for reactive astrocytosis (Carter et al. 2012; Rodriguez-Vieitez and Nordberg 2018). The most interesting issue in regard to imaging reactive astrocytosis is that, in contrast with activated microglia, it is an early phenomenon, suggesting it might be a better neuroinflammatory target at the preclinical and prodromal stages of AD (Scholl et al. 2015; Rodriguez-Vieitez et al. 2016).

7 Relation of Aβ Deposition with Genetic Risks and Predisposing Factors

Age is the strongest risk factor in sporadic AD, with the prevalence of the disease increasing exponentially with age. These risks are increased in the presence of the APOE ε4 allele (Farrer et al. 1997). Both risk factors have been directly associated with Aβ burden as measured by PET (Reiman et al. 2009a; b; Morris et al. 2010; Rowe et al. 2010).

To date, APOE ε4 is the most consistent genetic risk factor associated with sporadic AD, and its presence has been associated with an earlier age of onset and a dose-dependent—either one or two APOE ε4 alleles—higher risk of developing AD (Farrer et al. 1997). Examination of APOE ε4 allele status revealed that, independent of clinical classification, ε4 carriers present with significantly higher ¹¹C-PiB retention than non-ε4 carriers, further emphasizing the crucial role that APOE plays in the metabolism of Αβ (Reiman et al. 2009a; b; Morris et al. 2010; Rowe et al. 2010). When specific clinical groups are examined, the results— from either autopsy or Αβ imaging studies— are not that clear-cut. While some studies found significant differences in Αβ burden between AD ε4 carriers and non-ε4 carriers (Gomez-Isla et al. 1996; Drzezga et al. 2008; Grimmer et al. 2010), others did not (Berg et al. 1998; Klunk et al. 2004; Rowe et al. 2007; Rabinovici et al. 2010).

Both symptomatic and asymptomatic individual carriers of autosomal mutations within the APP or presenilin 1 or 2 genes associated with familial AD present with very high and early ¹¹C-PiB retention in the caudate nuclei, retention that seems to be independent of mutation type or disease severity (Klunk et al. 2007; Koivunen et al. 2008a; b; Remes et al. 2008; Villemagne et al. 2009a; b). Larger multicenter studies are helping elucidate the value of several biomarkers in these autosomal mutation carriers, assessment that might lead to better targeted disease-specific therapeutic strategies (Bateman et al. 2012; Benzinger et al. 2013).

While a similar pattern of Aβ deposition as seen in familial AD has been reported in adults with Down syndrome (Cohen et al. 2018), a completely different pattern of ¹¹C-PiB retention was observed in a mutation carrier of familial British dementia (Villemagne et al. 2010).

There are several other factors that have been associated with AD. Among these, the most relevant are vascular risk factors (VRF), such as diabetes, hypercholesterolemia, hypertension, hyperhomocysteinemia, etc. The influence of these VRF on AD pathology is an area that, despite the long and intense scientific debate, has not been clearly established (Viswanathan et al. 2009). Studies using MRI white matter hyperintense lesions as surrogate markers of vascular disease in association with assessment of Aβ deposition and cognition (Kuczynski et al. 2008; Marchant et al. 2011) suggest that cerebrovascular lesions may accelerate the clinical presentation of AD (Riekse et al. 2004) by having an independent effect on cognition (Vemuri et al. 2017; DeCarli et al. 2019).

As with AD, VRF are strongly age-dependent factors that share similar pathogenic pathways, such as oxidative stress or inflammation, so it is not easy to extricate one from the other in order to ascertain a causal, casual, or comorbid association between them (Rosendorff et al. 2007). On the other hand, in light of the absence of disease-modifying therapies for AD, the relevance of these risk factors rests in the fact that early intervention, especially at middle age, might have an impact on delaying either the onset or progression of AD, as suggested by recent reports of a lowering incidence of AD (Satizabal et al. 2016; Pase et al. 2017).

8 Aβ Imaging in Other Neurodegenerative Conditions

Aβ imaging in a wide spectrum of neurodegenerative conditions has yielded diverse results. Although lower than in AD, similar patterns of ¹¹C-PiB retention are usually observed in DLB (Rowe et al. 2007; Gomperts et al. 2008; Maetzler et al. 2009). Cortical ¹¹C-PiB retention is also higher in subjects diagnosed with CAA (Johnson et al. 2007), while there is usually no cortical ¹¹C-PiB retention in patients with FTLD (Rabinovici et al. 2007; Rowe et al. 2007; Drzezga et al. 2008; Engler et al. 2008). Aβ imaging has also facilitated differential diagnosis in cases of patients with atypical presentations of dementia (Ng et al. 2007a; b; Wolk et al. 2012a; b).

8.1 Cerebral Amyloid Angiopathy

CAA, characterized by Aβ deposits in and around the media of small arteries and arterioles of the cerebral cortex and leptomeninges, has been found to be present in most patients with AD (Jellinger and Attems 2005). To this date, neuropathologic examination of the brain remains the only definitive method for diagnostic confirmation of CAA. However, the combination of Aβ imaging with ¹¹C-PiB PET and T2* susceptibility-weighted (SWI) MR imaging (Knudsen et al. 2001) allows for the concomitant assessment of molecular and structural changes in vivo. The combination of cerebral microhemorrhages (MH) and superficial siderosis (SS) has been suggested as a radiological marker for CAA (Feldman et al. 2008). Aβ imaging studies have shown both processes may be intimately associated with Aβ deposition (Dierksen et al. 2010; Dhollander et al. 2011), preferentially in posterior areas of the brain, showing a distinct pattern of ¹¹C-PiB retention than the one observed in sporadic AD (Johnson et al. 2007). Furthermore, focal Aβ deposition might be useful in predicting new MH in CAA patients (Gurol et al. 2012; Raniga et al. 2017). While deep subcortical MH are generally associated with VRF, lobar MH (particularly posterior) (Pettersen et al. 2008) are usually attributed to CAA (Knudsen et al. 2001; Vernooij et al. 2008). Lobar MH are a frequent finding in AD patients or even in cognitively normal older individuals, and they are strongly associated with increasing age and Aβ deposition (Yates et al. 2011). This association between Aβ and vascular lesions has crucial implications not only for the selection and risk stratification of individuals undergoing anticoagulant therapies but also in those enrolled in anti-Aβ therapeutic trials (Weller et al. 2009).

8.2 Lewy Body Diseases

While AD is the most common cause of dementia in the elderly, DLB accounts for 20% of the cases (McKeith et al. 2005). The pathological hallmark of DLB is the presence of α-synuclein-containing LB within the neocortical and limbic regions (McKeith et al. 2005; Kantarci et al. 2020), as well as substantial loss of pigmented dopaminergic neurons in the substantia nigra, reflected in a marked dopaminergic terminal denervation in the striatum (McKeith and Mosimann 2004). More refined neuropathological techniques have revealed extensive synaptic deposits of α-synuclein, more concordant with the DLB clinical phenotype than the usually sparse cortical LB (Schulz-Schaeffer 2010). Overlap of cognitive symptoms early in the disease course offers a challenge for clinicians making it difficult to distinguish between patients that will develop AD and those that will ultimately develop DLB. Post-mortem histopathological studies have identified that 50–80% of DLB cases often have cortical Aβ deposits with characteristics and a distribution similar to AD patients (McKeith et al. 2005; Rowe et al. 2007; Edison et al. 2008a; b; Kantarci et al. 2020). This has also been described as “mixed” DLB/AD or “Lewy body variant of AD.” So-called “pure” DLB is much less common, and it is not clear if the clinical characteristics and prognosis differ from the mixed-pathology cases. On visual inspection, cortical Aβ tracer retention in DLB is generally lower and more variable than in AD, probably reflecting a larger spectrum of Αβ deposition (Rowe et al. 2007; Gomperts et al. 2008; Maetzler et al. 2009). Therefore Aβ imaging is not a useful tool in differentiating between AD and DLB, but might assist in the differential diagnosis of those cases (“pure” DLB) without Aβ deposition in the brain (Armstrong et al. 2000; Landau and Villemagne 2020).

Taking advantage of its high affinity for α-synuclein (Fodero-Tavoletti et al. 2009a; b), ¹¹C-BF227 has been used for the assessment of multiple system atrophy (MSA), a disease characterized by the glial accumulation of α-synuclein in the absence of Aβ plaques, showing a higher retention in MSA patients compared to age-matched controls (Kikuchi et al. 2010). Interestingly, while ¹¹C-BF227 retention in DLB patients was lower compared to AD in most cortical regions, DLB presented with significantly higher ¹¹C-BF227 retention in the amygdala (Furukawa et al. 2011).

In contrast to Aβ imaging, assessing the integrity of dopaminergic terminals via dopamine transporter (DAT) or vesicular monoamine transporter type 2 (VMAT2) imaging can robustly detect reductions of dopaminergic nigrostriatal afferents in DLB patients, assisting in the differential diagnosis from AD (O'Brien et al. 2004; Koeppe et al. 2008; Villemagne et al. 2012a; b; c; d; Gomperts et al. 2016).

Cortical Aβ is not usually present in cognitively intact Parkinson’s disease (PD) patients, and no significant ¹¹C-PiB cortical retention is usually observed in these patients (Johansson et al. 2008; Mashima et al. 2017). Aβ deposits, both vascular and parenchymal, are frequent, however, in PD patients who develop dementia (PDD) (Edison et al. 2008a; b; Gomperts et al. 2008; Maetzler et al. 2008; Maetzler et al. 2009; Burack et al. 2010; Foster et al. 2010; Kalaitzakis et al. 2011; Gomperts et al. 2016).

The contribution of Aβ to the development of Lewy body diseases remains unclear, but cortical Aβ deposits are associated with extensive α-synuclein lesions and higher levels of insoluble α-synuclein lesions (Pletnikova et al. 2005), as well as exacerbation of neuronal injury (Masliah et al. 2001). Furthermore, it has been shown that in PD, CSF Aβ is predictive of cognitive decline (Siderowf et al. 2010). Along these lines, while Aβ imaging cannot contribute to the differential diagnosis between AD and DLB, it may have prognostic relevance in these Lewy body diseases. For example, similar to what has been reported for PDD, where higher Αβ burden was associated with a shorter prodromal phase before dementia (Halliday et al. 2008), Αβ burden in DLB patients was inversely correlated with the interval from onset of cognitive impairment to the full development of the DLB phenotype, faster rates of cortical atrophy, and faster rates of clinical progression (Rowe et al. 2007; Sarro et al. 2016; Donaghy et al. 2019).

8.3 Frontotemporal Lobar Degeneration

FTLD is a syndrome that can also be clinically difficult to distinguish from early-onset AD, especially at the initial stages of the disease. Based on clinical phenotype, FTLD has been categorized mainly into two classes: behavioral variant of FTLD (bvFTLD) with distinct changes in behavior and personality with little effect on language functions and progressive aphasias (comprised by semantic dementia (svFTLD), progressive non-fluent aphasia (nfvFTLD), and logopenic aphasia (lvFTLD)) showing progressive language deficits, with less obvious personality and behavioral changes (Rabinovici & Miller 2010).

Aβ deposition is not a pathological trait of FTLD, and definitive differential diagnosis can only be established after post-mortem examination of the human brain. Neuropathological examination of the brain in FTLD shows variable frontal and temporal atrophy (Cairns et al. 2007; Hodges and Patterson 2007; Snowden et al. 2007), microvacuolation, and neuronal loss, with white matter myelin loss, astrocytic gliosis, as well as neuronal and glial inclusions (Cairns et al. 2007). Three intraneuronal inclusion types have been identified in FTLD: (1) hyperphosphorylated tau (FTLD-tau), (2) ubiquitin-hyperphosphorylated and proteolysed TAR-DNA binding protein 43 (FTLD-TDP-43), and (3) fused sarcoma positive (FUS) (Taniguchi et al. 2004; Mott et al. 2005; Neumann et al. 2006; Neumann et al. 2009). Tau-positive inclusions account for approximately 40% of all FTLD cases, the remainder being tau negative and TDP-43/ubiquitin positive (Josephs et al. 2004). In bvFTLD, both tau and TDP-43 are equally prevalent; and while 90% of the SD cases show TDP-43 pathology, PNFA presents predominantly (70%) tau pathology (Josephs et al. 2004; Mackenzie et al. 2008). In contrast, logopenic progressive aphasia (LPA) is thought to be a language presentation of AD, with typical Aβ and NFT (Leyton et al. 2011). Absence of Aβ and tau deposition is considered one of the features of limbic-predominant age-related TDP-43 encephalopathy (LATE) (Nelson et al. 2019), a neurodegenerative condition that presents with an amnestic dementia similar to AD, which also presents a typical signature on FDG studies (Botha et al. 2018).

Aβ imaging has been helpful in the differential diagnosis between FTLD and AD (Rabinovici et al. 2007; Rowe et al. 2007; Drzezga et al. 2008; Engler et al. 2008). Furthermore, Aβ imaging has been used to ascertain the absence of AD pathology in the different FTLD aphasias (Drzezga et al. 2008; Rabinovici et al. 2008; Leyton et al. 2011). Despite displaying similar specificities, the diagnostic performance of Aβ imaging proved to be more sensitive for the diagnosis of FTLD than FDG (Rabinovici et al. 2011).

8.4 Prion Diseases

No cortical Aβ deposition is usually observed in sporadic Creutzfeldt-Jakob disease (CJD) cases (Villemagne et al. 2009a; b; Dulamea and Solomon 2016; Matias-Guiu et al. 2017). Due to the rapid progression and relatively short duration of symptomatic illness, sporadic CJD patients do not demonstrate appreciable PrP-plaque deposition at autopsy (Liberski 2004). Although some sporadic CJD molecular subtypes can manifest larger PrP plaques (Hill et al. 2003), the most common subtype usually demonstrates only very fine “synaptic” and small perivacuolar deposits (Hill et al. 2003; Liberski 2004). A study in transmissible spongiform encephalopathies showed that Aβ imaging with ¹¹C-BF227 might be able to distinguish between Gerstmann-Straussler-Scheinker disease (GSS) and sporadic CJD disease with clear tracer retention in brain regions of PrP deposition (Okamura et al. 2010).

8.5 Traumatic Brain Injury and Chronic Traumatic Encephalopathy

Traumatic brain injury (TBI) has been associated with a number of long-term sequelae, including an increased risk for dementia (Plassman et al. 2000; Barnes et al. 2014); however, the long-term effects of a single TBI event in regard to triggering AD-related proteinopathies remain controversial (Furst and Bigler 2016; LoBue et al. 2019). While some studies have reported higher cortical Aβ deposition in TBI (Malkki 2014; Gatson et al. 2016), others have found no evidence of it (Weiner et al. 2017). These discrepancies might be in part explained by slightly different definitions of TBI, as well as in the cause or severity of the event. A more careful selection of and better clinical and neuropsychological characterization of participants is needed to elucidate this issue.

Chronic traumatic encephalopathy (CTE) is a neurodegenerative 3R/4R tauopathy associated with a history of repetitive head trauma (McKee et al. 2018). However, in about 50% of CTE cases, other pathological protein deposits such as Aβ, α-synuclein, and TDP-43 are also present (McKee et al. 2013), which, added to a variable degree of cortical gray matter atrophy, make the interpretation of molecular imaging studies complicated.

9 Clinical and Research Applications of Amyloid Imaging

Aβ imaging has been widely used in research settings as well as in a host of disease-specific therapeutic trails while ascertaining its place as a diagnostic and prognostic tool in the clinical evaluation and management of dementia patients.

The application of Aβ imaging in research and clinical settings raises issues referring to the quantitative or semiquantitative approaches used to determine Aβ burden. Given that the studies are performed in individuals or populations who might not tolerate long scanning sessions, the majority of studies use a short 20- or 30-min PET scan when Aβ tracers reach, at different periods after injection, apparent steady state (Lopresti et al. 2005; Rowe et al. 2007; Yee et al. 2007; Edison et al. 2008a; b). While these image acquisition intervals are more than enough to obtain clinically useful images, it is unlikely they adequately or reliably establish Aβ burden in therapeutic trials (Rinne et al. 2010; Ostrowitzki et al. 2011; Ossenkoppele et al. 2012), where drug-induced changes in tracer metabolism, blood-brain barrier permeability, cerebral blood flow, gray matter atrophy, target conformation, etc. might affect Aβ tracer binding (Lopresti et al. 2005; Price et al. 2005). Finally, the issue of partial volume correction (PVC) raises several other questions: How much of the intrinsic noise of the PET signal is augmented by PVC? Should longitudinal studies be evaluated without PVC (Jack Jr. et al. 2013; Villemagne et al. 2013; Landau et al. 2015)? How to correct for spillover radioactivity from white matter? All these issues remain contentious.

9.1 Aβ Imaging in Large Observational Cohorts

Since the launch of the Alzheimer’s Disease Neuroimaging Initiative (ADNI https://adni.loni.usc.edu) in the USA in 2003 ever-growing, similarly oriented consortia such as the Australian Imaging, Biomarker and Lifestyle (AIBL, https://aibl.csiro.au) Study of Ageing (Ellis et al. 2009) and Europe’s Amyloid Imaging to Prevent Alzheimer’s Disease (AMYPAD, http://amypad.eu) (Frisoni et al. 2019) have been organized and assembled around the world, setting the groundwork for biomarker discovery and the establishing of standards for early diagnosis of sporadic AD (Villemagne et al. 2014). A similar worldwide consortium, the Dominantly Inherited Alzheimer’s Network (DIAN, https://dian.wustl.edu) (Morris et al. 2012), focuses on autosomal mutations leading to AD. The various accomplishments of these large cohort observational studies, especially in regard to Aβ imaging, have contributed substantially to a better understanding of the underlying physiopathology of aging and AD. These accomplishments are basically predicated in the trinity of multimodality, standardization, and sharing. This multimodality approach can now better identify those subjects with AD-specific traits that are more likely to present cognitive decline in the near future and that might represent the best candidates for smaller but more efficient therapeutic trials, trials that, by the gained and shared knowledge of some of the pathophysiological mechanisms underpinning aging and AD pathology, can be more focused on a specific target or a specific stage of the disease process.

9.2 Aβ Imaging in Disease-Specific Therapeutic Trials

Aβ imaging with PET is also contributing to the development of more effective therapies by allowing better selection of patients for multicenter anti-Aβ therapy trials around the world (Long and Holtzman 2019) and providing a means to ascertain target engagement and measure their impact on Aβ burden (Rinne et al. 2010; Ostrowitzki et al. 2011; Liu et al. 2015) as well as outcome measure of their effectiveness (Doody et al. 2014; Salloway et al. 2014; Sevigny et al. 2016).

While these studies represent one of the principal applications of Aβ imaging today, a key challenge for these trials is that reductions in Aβ burden, as measured by either CSF or PET, have not been accompanied with cognitive stabilization or improvement (Doody et al. 2014; Salloway et al. 2014; Sevigny et al. 2016). Furthermore, some of these therapies were associated with side effects such as edema and microhemorrhages. Other Aβ-centered therapeutic mechanisms have been attempted, such as inhibition or modulation of β- and γ-secretases, but most have been discontinued as well (Egan et al. 2019, Long and Holtzman 2019). Since study participants were, while mild, symptomatic AD patients, these setbacks have been attributed to treatment occurring too late in the course of disease, so trials like A4 have enrolled cognitively normal elderly individuals with high Aβ, aiming at preventing the cognitive decline expected in this cohort (Sperling et al. 2014). Others have argued that these failures are clear evidence that Aβ reduction will not be an effective strategy for treating AD and that Aβ as a target for treatment or preventing AD should be abandoned.

The contribution of Aβ imaging to the selection of patients for therapeutic trials needs to be complemented with adequate MRI sequences such as SWI or GRE for the detection of vascular pathology to rule out the possibility of vascular complications such as Aβ imaging-related abnormalities (ARIA), either as “vasogenic edema” and/or sulcal effusion (ARIA-E) or as hemosiderin deposits including MH and SS (ARIA-H) (Cordonnier 2010; Sperling et al. 2011a; b; c). The presence of MH is impacting on patient selection in therapeutic trials, and some anti-Aβ immunotherapy trials are already excluding individuals with MH. To make matters worse, about one third of non-demented individuals with high Aβ burden and no prior history of cerebrovascular disease show MH (Yates et al. 2011). While focal Aβ deposition might be a better predictor of a vascular event (Gurol et al. 2012), only longitudinal follow-up of large cohorts of patients will allow elucidation of the relevant risk factors for lobar MH.

9.3 The Centiloid Scale

One of the obstacles for the widespread implementation of Aβ imaging in therapeutic trials or comparison of results in different cohorts across different centers is that Aβ tracers present with differing pharmacological and pharmacokinetic properties, yielding results in different dynamic ranges. These issues, summed to the use of a diverse arsenal of quantitative approaches and contrasting criteria for the selection of internal scaling region and different thresholds to separate high from low Aβ burden in the brain, pose a challenge when trying to compare results from different cohorts or different therapeutic trials that use different Aβ tracers as Aβ quantitative tools. Therefore, a method was developed to standardized all Aβ imaging results under a single common semiquantitative scale—called the Centiloid (Klunk et al. 2015)—to improve the clinical and research application of these Aβ tracers. Since then, all F-18-labeled radiotracers Aβ tracers currently in use have been cross-calibrated against PiB (Rowe et al. 2016; Rowe et al. 2017; Battle et al. 2018; Navitsky et al. 2018) to enable translation into Centiloids.

The other obstacle is to define a threshold. One of the issues is in regard to the utility of a cut-off given the continuous nature of Aβ deposition (Jagust 2011; Mormino et al. 2012). While thresholds are arbitrary, in order to adopt one, it needs to be shown that it is relevant and accurate from a diagnostic and/or prognostic point of view. From a clinical point of view, a visual binary read will help separate those with a significant Aβ burden in the brain from those with a null or low Aβ burden. Similar dilemmas arise in research settings. In order to avoid being trapped in this dichotomy, some researchers use tertiles (Resnick et al. 2010; Rowe et al. 2013a; b), while others are proposing to distinguish two thresholds for “positivity,” one more sensitive, to detect the early stages of Aβ deposition usually in still cognitively unimpaired elderly individuals, and another, a higher one, that can more specifically ascribe the clinical phenotype of dementia to Aβ pathology (Klunk et al. 2012).

The introduction of the Centiloid scale has tried to solve this problem, and while it transforms all Aβ tracers’ semiquantitative results into a single universal scale, the idiosyncratic properties of these Aβ tracers remain so they might be more or less sensitive or accurate to provide a similar statement to a similar Aβ burden in the brain. Most of the studies have used neuropathology to establish Centiloid thresholds. For example, a large multicenter study using neuropathology from pre-mortem ¹¹C-PiB studies found that below a threshold of 12.2 Centiloids can distinguish between plaques and no plaques, while a threshold of 24.4 Centiloids identified intermediate-to-high AD neuropathological changes (La Joie et al. 2018). Similar neuropathology-based thresholds were determined with other Aβ tracers such as ¹⁸F-florbetaben yielding 21 Centiloids (Dore et al. 2019) or ¹⁸F-florbetapir yielding 24 Centiloids (Navitsky et al. 2018).

9.4 Appropriate Use Criteria and Clinical Impact of Aβ Imaging

Also, clinical criteria for the appropriate use of Aβ imaging has been established highlighting the need to integrate Aβ imaging with a comprehensive clinical and cognitive evaluation performed by a clinician experienced in the evaluation of dementia, ensuring a positive impact on patient management (Johnson et al. 2013). These criteria clearly stipulate the specific cases Aβ imaging should be used, such as patients with persistent or progressive unexplained cognitive impairment, progressive atypical or unclear clinical presentation of dementia, or dementia onset in individuals 65 years or younger (Johnson et al. 2013; Apostolova et al. 2016). It also outlines the inappropriate use of Aβ imaging, such as cases of probable AD with typical age of onset or to determine dementia severity, asymptomatic individuals or unconfirmed cognitive complaint, a family history of dementia or presence of the APOE ε4 allele, as well as non-medical use such as litigation for insurance purposes (Johnson et al. 2013; Apostolova et al. 2016).

These criteria are being tested in multicentre trials such as the Imaging Dementia—Evidence for Amyloid Scanning (IDEAS, https://www.ideas-study.org), assessing the clinical utility of Aβ imaging in the workup of patients with MCI or dementia of uncertain origin. The initial report from this trial showed that results from Aβ imaging were associated with more than a 60% change in the clinical management of these patients within 90 days, usually reflected in a change in acetylcholine esterase inhibitor medication (Rabinovici et al. 2019). Overall, Aβ imaging leads to change in diagnostic confidence and patient management in up to 60% of cases attending memory clinics (de Wilde et al. 2018; Rabinovici et al. 2019).

10 CODA

The clinical diagnosis of AD is typically based on progressive cognitive impairments while excluding other diseases. Clinical diagnosis of sporadic disease is challenging, however, often presenting mild and non-specific symptoms attributable to diverse and overlapping pathology presenting similar phenotypes. Overall, the accuracy of clinical diagnosis of AD compared to neuropathological examination ranges between 70% and 90% (Beach et al. 2012). While clinical criteria, together with current structural neuroimaging techniques (CT or MRI), are sufficiently sensitive and specific for the diagnosis of AD at the mid to late stages of the disease, they are focused on relatively non-specific features such as memory and functional activity decline or brain atrophy that are late features in the course of the disease. Confirmation of diagnosis of AD still relies on autopsy. But how reliable is pathology? How much of a “gold standard” is autopsy (Jellinger 2010; Scheltens and Rockwood 2011)? A more sensible paradigm, adopting a more fluid model that integrates all cognitive, biochemical, and imaging biomarkers that can provide a better predictive framework, has been proposed in the new diagnostic criteria for AD (Dubois et al. 2010; Sperling and Johnson 2010; McKhann et al. 2011), MCI (Albert et al. 2011), and preclinical AD (Sperling et al. 2011a; b; c). Along the same lines, Scheltens and Rockwood suggest that “a shift in focus from why people with AD have plaques and tangles to why not all people with plaques and tangles have AD would be welcome” (Scheltens and Rockwood 2011). While most people with AD pathology develop AD dementia, there are some that while having the pathology do not. What makes them different? What conveys resistance to some or, conversely, what makes others more vulnerable to the same pathological burden? Identifying these factors will allow a more discerning selection of individuals that will benefit from early intervention with anti-Aβ medication while also providing insights into potential preventive strategies or even new therapeutic approaches (Sojkova and Resnick 2011; Sperling et al. 2011a; b; c). While the new diagnostic criteria has blurred once again the boundaries between MCI and AD (Morris 2012), it also allows for earlier diagnosis and therapeutic intervention, arguing that when there is a clear history of progressive cognitive decline, objective evidence from psychometric tests of episodic memory impairment and characteristic abnormalities in the CSF, and/or neuroimaging studies such as Aβ imaging, dementia is no more required for the diagnosis of probable AD (Dubois et al. 2010; Sperling and Johnson 2010; McKhann et al. 2011; Jack Jr. et al. 2016; Jack et al. 2018). Thus, as the criteria for the diagnosis of AD changes, Aβ imaging is likely, given its impact on patient management (Rabinovici et al. 2019), to play an increasingly important role in clinical practice provided it is accessible and affordable (Villemagne et al. 2010).

At this stage Aβ imaging, in combination with other modalities, is probably the best diagnostic tool to rule out Aβ pathology in individuals being evaluated for cognitive decline. The data to date clearly show that Aβ production and deposition is not a benign process with a more prominent role at the early, mostly asymptomatic, stages of the disease (Hyman 2011; Karran et al. 2011); therefore, anti-Aβ therapy at these stages seems a logical approach (Sperling et al. 2011a, b, c). On the other hand, we still do not have the full grasp of the long-term cognitive consequences of Aβ deposition. Longitudinal Aβ imaging, especially of cognitively unimpaired elderly individuals, is allowing the assessment of pathology almost simultaneously with tau, cognition, and other biomarkers of neurodegeneration. This will provide not only a better picture of how these different factors interact but also the weight or relevance this complementary and/or supplementary information plays at different stages of the disease.

The pathological process usually begins decades before symptoms are evident, making early identification difficult. This in turn precludes early intervention with disease-modifying medications during the presymptomatic period, which by arresting neuronal loss would presumably achieve the maximum benefits of such therapies (Villemagne et al. 2008a; b; Sperling et al. 2011a; b; c). Therefore, a change in the diagnostic paradigm is warranted, where diagnosis moves away from identification of signs and symptoms of neuronal failure—indicating that central compensatory mechanisms have been exhausted and extensive synaptic and neuronal damage is present—to the non-invasive detection of specific biomarkers for particular traits underlying the pathological process (Clark et al. 2008). Given the complexity and sometimes overlapping characteristics of these disorders, and despite recent advances in molecular neurosciences, it is unlikely that a single biomarker will be able to provide the diagnostic certainty required for the early detection of neurodegenerative diseases like AD, especially for the identification of at-risk individuals before the development of the typical phenotype. Therefore, a multimodality approach combining biochemical and neuroimaging methods is warranted (Shaw et al. 2007).

Biomarkers are important as they can be used to identify the risk of developing a disease (antecedent biomarkers), aid in identifying disease (diagnostic biomarkers), assess disease evolution (progression biomarkers), predict future disease course (prognostic biomarkers), or evaluate and/or customize therapy (theranostic biomarkers). While CSF Aβ and tau, structural, and molecular neuroimaging (both FDG and Aβ imaging) are all good—specially if combined—diagnostic biomarkers for AD, evaluation of the data available to date suggests that Aβ imaging and CSF assessments of Aβ are good antecedent biomarkers at the preclinical and prodromal stages of the disease (Jack Jr. et al. 2010; Ossenkoppele et al. 2011; Villemagne et al. 2011a; b). Conversely, CSF assessments of tau and p-tau, structural neuroimaging, and FDG seem to be excellent biomarkers of disease progression (Jack Jr. et al. 2010). Despite the fact that Aβ accumulation is still ongoing at the AD stage (Villemagne et al. 2011a; b; Villain et al. 2012), and in contrast to what is observed at the preclinical and prodromal phases (Pike et al. 2007; Villemagne et al. 2008a; b; Cohen et al. 2012), Aβ burden at this stage is not correlated with markers of neurodegeneration, disease severity, or cognitive impairment (Hyman 2011; Karran et al. 2011). A review of the literature shows that quantification of Aβ_1–42, total tau, and p-tau in CSF and FDG and a long list of assessments of global or regional brain atrophy as well as white matter hyperintensities with MRI have all been proposed as predictors of conversion to AD (Chang and Silverman 2004; Smith et al. 2008; Jack Jr. et al. 2010). While larger longitudinal studies will further clarify the role of Aβ imaging in predicting conversion from MCI to AD, one of the most important conclusions to be drawn from these studies is that the likelihood of developing AD in a cognitively unimpaired individual with a low Aβ burden scan is extremely small (Villemagne et al. 2011a; b).

Another aspect to be considered is prediction of therapeutic response. As new therapies enter clinical trials, the role of Aβ imaging in vivo is becoming increasingly crucial. Αβ imaging allows the in vivo assessment of brain Αβ pathology and its changes over time, providing highly accurate, reliable, and reproducible quantitative statements of regional or global Aβ burden in the brain. As mentioned before, therapies, especially those targeting irreversible neurodegenerative processes, have a better chance to succeed if applied early, making early detection of the underlying pathological process so important. Therefore, Aβ imaging is an ideal tool for the selection of adequate candidates for anti-Aβ therapeutic trials while also monitoring—and potentially predicting—treatment response, thus aiding in reducing sample size and minimizing cost while maximizing outcomes. Studies are already showing the utility of this technique (Rinne et al. 2010; Ostrowitzki et al. 2011; Liu et al. 2015). These results, taken together with the results from longitudinal studies, support the growing consensus that anti-Aβ therapy to be effective may need to be given early in the course of the disease, perhaps even before symptoms appear (Sperling et al. 2011a, b, c), and that downstream mechanisms may also need to be addressed to successfully prevent the development of AD. Furthermore, the slow rate of Aβ deposition indicates that the time window for altering Aβ accumulation prior to the full manifestation of the clinical phenotype may be very wide (Villemagne et al. 2011a, b).

The introduction of radiotracers for the non-invasive in vivo quantification of Αβ burden in the brain has revolutionized the approach to the evaluation of AD. Αβ imaging has extended our insight into Aβ deposition in the brain, where Aβ burden as measured by PET matches histopathological reports of Αβ distribution in aging and dementia, it appears more accurate than FDG for the diagnosis of AD, and it is an excellent aid in the differential diagnosis of AD from FTLD (Mathis et al. 2005; Johnson 2006; Mathis et al. 2007; Villemagne et al. 2008a; b; Jagust 2009; Rabinovici and Jagust 2009; Drzezga 2010; Klunk 2011; Villemagne et al. 2012a; b; c; d). APOE ε4 carriers, independent of diagnosis or disease severity, present with higher Aβ burden than non-ε4 carriers, and high retention of Aβ radiotracers closely corresponds with low CSF Aβ levels and increasing age (Reiman et al. 2009a; b; Fagan and Holtzman 2010; Morris et al. 2010; Rowe et al. 2010). Although Αβ burden as assessed by PET does not strongly correlate with cognitive impairment in AD, it does correlate with memory impairment and a higher risk for cognitive decline in the aging population and MCI subjects. This correlation with memory impairment, one of the earliest symptoms of AD, suggests that Αβ deposition is not part of normal aging, supporting the hypothesis that Αβ deposition occurs well before the onset of symptoms and it likely represents preclinical AD in asymptomatic individuals and prodromal AD in MCI. Further longitudinal observations, coupled with different disease-specific biomarkers to assess potential downstream effects of Aβ, are required to confirm this hypothesis and further elucidate the role of Αβ deposition in the course of Alzheimer’s disease.