Amyloid PET imaging

The extracellular deposition of β-amyloid (Aβ) plaques in the cortical grey matter of the brain is one of the pathological hallmarks of Alzheimer's disease (AD) [1]. However, in most of the non-AD dementias, Aβ plaques increase with age and APOE ε4 carrier status.

In AD, the majority of the neuritic plaques are located in the frontal cortex (particularly the orbital and medial frontal cortex), cingulate gyrus, precuneus, and lateral parietal and temporal regions. However, the presence of Aβ plaques is scarce in the primary sensorimotor and occipital cortex and the mesial temporal areas [4]. Interestingly, Aβ plaques develop many years before the onset of cognitive impairment and dementia [4, 5]. Even though the distribution and density of Aβ plaques do not correlate with the degree of cognitive impairment [4].

Amyloid PET radiotracers have been developed that bind to fibrillar Aβ plaques in the brain, allowing non-invasive in vivo detection of Aβ deposits [5, 6]. Amyloid PET contributes to a more accurate and earlier diagnosis of cognitively impaired patients with suspected AD, allowing the disease to be confirmed or excluded [4,5,6]. It also provides clinicians with information about the extent, location, follow-up changes, and response to anti-amyloid therapy effect on the Aβ deposition [1], and no contraindications have been described.

Currently, the use of amyloid PET imaging is considered clinically appropriate to estimate Aβ plaques in adult patients with cognitive impairment being evaluated for AD [2, 5].

Amyloid-Β-targeting tracers

The first amyloid PET radiotracer was the carbon-11 Pittsburgh Compound B ([11C]PiB), which is an analogue of Thioflavin T (fluorescent amyloid dye). [11C] PiB showed high affinity and high specificity for Aβ plaques, however, the 20-min short radioactive half-life of carbon-11 limits its use to centres with an on-site cyclotron [4, 6, 7].

[11C] PiB often serves as a reference standard to which other amyloid PET agents are compared [2].

For wider access and commercialisation of this technique, fluorine-18 labelled amyloid-β-targeting tracers were developed. Fluorine-18 [18F] has a longer radioactive half-life of 110 min which enables widespread distribution and use of these radiopharmaceuticals beyond the manufacturing site for routine clinical use [2, 6,7,8].

Three [18F] labelled compounds are approved for amyloid PET imaging by the U.S. Food and Drug Administration, European Medicines Agency, and other global regulatory agencies [2, 4, 5]: [18F]florbetapir (commercial name AmyvidTM; company Eli Lilly), [18F]florbetaben (NeuraceqTM; Life Molecular Imaging), and [18F]flutemetamol (VizamylTM; GE Healthcare). They are essentially equivalent in clinical practice, have high reproducibility across centres, and have low individual variability [1]. All three radiotracers have been neuropathologically validated showing a sensitivity of 92%, 98%, and 91%, with a specificity of 91%, 89%, and 100% for [18F]florbetapir, [18F]florbetaben, and [18F]flutemetamol, respectively [6,7,8].

All these tracers exhibit a high non-specific white matter uptake, which results in a distinctive white matter pattern in scans of healthy subjects. For example, Alzheimer’s patients with increased cortical amyloid plaques show a loss of the distinction between grey matter and white matter.

[18F]flutafuranol (formerly NAV4694) is another fluorinated compound also available and used in clinical trials but not currently approved for clinical use, which shows a similar brain distribution that [11C]PiB in terms of tissue contrast [2, 4, 5].

Imaging procedure and interpretation

No specific preparation of the patient or drug withdrawal is required for the amyloid PET scan [5]. Amyloid PET tracers differ in their tracing kinetics, specific binding ratios and optimal imaging parameters. Recommended injected activities, image acquisition times, and image interpretation details of the amyloid PET compounds are shown in Table 1 [2, 4, 5].

Table 1 Recommended injected activities, uptake phase, acquisition duration, recommended colour scale and image interpretation for the three FDA and EMA-approved [18F]amyloid-β-compounds

According to the insert package, PET images using [18F]florbetaben should be displayed in greyscale, with [18F]florbetapir in inverse greyscale, and with [18F]flutemetamol in Rainbow or Sokoloff colour scale. Additionally, image activity should be adjusted by setting the colour scale to different regions depending on the tracer employed [2, 5] (Fig. 1).

Fig. 1

Axial PET images of [18F]florbetapir (A), [18F]flutemetamol (B), and [18F]florbetaben (C) in the colour scale recommended by the tracer manufacturer. Positive brain amyloid scans of Alzheimer’s patients in images 1 (A.1, B.1, and C.1) and negative brain amyloid scans in images 2 (A.2, B.2, C.2)

Visual reading.

Interpretation of amyloid PET images is based on structured visual analysis, according to established recommendations for each of the three approved amyloid tracers. A qualitative binary interpretation as a positive or negative scan is performed. As a prerequisite, nuclear medicine physicians must complete an appropriate training programme provided by the radiotracer manufacturers [2, 5] (Table 1).

In general, the contrast between the high activity in the white matter and the uptake in the grey matter is evaluated. Negative amyloid PET scans show only non-specific tracer retention in white matter, resulting in a pattern resembling numerous concave arboreal ramifications that do not extend into the cortical ribbon (Fig. 1: A.2, B.2, and C.2). Besides, positive amyloid PET scans also show grey matter uptake, i.e. the grey-white junction is blurred due to tracer retention extending into the neocortex forming a smooth and regular boundary at the edge of the cerebral cortex [2, 5] (Fig. 1: A.1, B.1, and C.1).

All cerebral cortical and subcortical regions should be examined, with particular attention to the lateral temporal, frontal, posterior cingulate/precuneus, parietal cortices, and basal ganglia [4, 5]. Tracer uptake in the posterior cingulate, precuneus and frontal cingulate cortex is observed in the early stages, while generalised neocortical uptake is usually seen at later stages [2, 7].

In addition, the correlation of amyloid PET images fused with morphological images (MRI or CT) is recommended to assess sources of misinterpretation such as atrophy, vascular comorbidities, or structural abnormalities, among others [5]. The use of multimodality devices such as PET/CT or even PET/MR facilitates the fusion process, but the different imaging modalities can easily be fused using the software package currently available in the workstations Finally, it is important to keep in mind that a positive amyloid PET indicates the presence of high amyloid plaques (Alzheimer´s pathology), not the diagnosis of Alzheimer´s disease.

Quantification.

Quantification of the amyloid burden by means of PET has been extensively used for inclusion criteria and monitoring the response to the anti-amyloid therapy effect of the Aβ deposits [7]. In clinical practice, it can complement visual analysis, particularly for inexperienced readers, when the confidence in visual analysis is low, or in borderline cases. Quantification of amyloid PET has been included as an adjunct to visual interpretation in European labels, but not in the FDA labels yet [2].

The most common quantitative measure is the standardised uptake value ratio (SUVr) which calculates the ratio between selected cerebral regions of interest (ROI) uptake (target regions that are known to accumulate amyloid plaques) and a reference region uptake (that is relatively spared of pathology) [2, 4, 6]. As ROIs vary depending on the radiotracer used, SUVr thresholds for a positive or negative scan differ from one [18F]-compound to another [6,7,8]. Interestingly, PET manufacturers are providing software to calculate automatically the SUVr of all three approved radiotracers.

The current trend is to use a single measure of amyloid burden, independent of the radiotracer used, that reflects the total pathological burden of amyloid. This will allow better comparison of data between centres and ultimately lead to the application of universal diagnostic and prognostic values.

This metric is named Centiloid, and the value is derived from the SUVr using a linear equation that transforms it into an unbounded scale from 0 to 100. The two parameters of this linear equation (a scale factor and an intercept) are specific to each tracer and pipeline analysis and should be calibrated with a test population to ensure that zero is associated with “high certainty” amyloid negative subjects and an average of 100 is associated with “typical” AD patients [9]. Once the calibration has been established, for example in commercial software or by a previous study, the calculation is straightforward.

In general, centiloid values below 10 excludes the presence of Aβ pathology, values above 40 centiloids correspond well with pathological amounts, while values falling in between these two cutoffs (“intermediate-range” or “grey-zone”) are related to an increased risk of disease progression [10]. Neuropathologic studies have shown that the earliest detectable amyloid PET signal occurs around 12 centiloids, the optimal threshold to predict future significant Aβ accumulation ranged from 15 to 17.5 centiloids in cognitively unimpaired participants, and 19 centiloids reflect the cut point beyond which the amyloid rate of change increases reliably beyond baseline variation in the measure. Centiloids between 21 and 24 best discriminate between subjects with no-to-low Aβ plaque burden and those with intermediate-to-high deposition, a cut-off value of 26 was found to best predict progression to dementia, and 30 centiloids are indicative of established Aβ burden [10, 11].

Furthermore, centiloid has been recently approved by the EMA to quantify brain amyloid deposition using amyloid PET as an adjunct to visual reading [12].

Centiloid quantification has been proposed as a valuable adjunct to the visual assessment of amyloid PET images to provide high confidence in identifying early or emerging Aβ pathology as intermediate values are associated with an increased risk of disease progression, to select patients for anti-amyloid disease-modifying therapies, and to measure the degree of Aβ clearance after treatment [10].

Pitfalls and artefacts

Amyloid PET images should be correlated with a morphological imaging study to identify sources of error and provide useful information for those studies that pose a diagnostic challenge. Atrophy, which is common in ageing patients, complicates interpretation because it can lead to a false-negative scan. For example, in severe atrophy, it is difficult to distinguish a thin band of amyloid-positive grey cortex from non-specific uptake in the adjacent white matter. Moreover, it may lead to a false-positive scan due to overestimation of radiotracer uptake from the white matter uptake to the remaining cortex [4, 5]. Encephalomalacia from prior stroke, brain surgery, head trauma, and hydrocephalus might be some additional causes of misinterpretation [5] (Fig. 2). Finally, head movement during acquisition results in blurring of the image which decreases the diagnostic accuracy [4, 5]. Aβ plaques can be present in other non-AD dementias among which dementia with Lewy bodies has the highest amyloid positivity (51–70% of patients have diffuse Aβ plaque deposition), [2, 4, 5, 8], followed by vascular dementia (30%), frontotemporal dementia (12%), cerebral amyloid angiopathy, and which also increases with age and in APOE ε4 carriers. [2, 8, 13, 14]. In patients with corticobasal syndrome, the overall prevalence of Aβ plaques is 38% but does not increase with age; this is thought to be because AD may be the underlying pathology in young patients with corticobasal syndrome, whereas primary tauopathy becomes more likely with age [14].

Fig. 2

Pitfalls and artefacts in amyloid PET imaging. A [18F]Florbetapir PET/CT scan of a patient with right frontal stroke and brain atrophy. The PET/CT fusion images clearly show no difference between white matter and grey matter (positive amyloid PET scan). B [18F]Florbetapir PET/CT scan of a patient with normal pressure hydrocephalus. In this case, the fusion images showed much higher activity in the white matter than in the grey matter (negative amyloid PET scan)

Tau PET imaging

Tau protein is mainly found in the axons of the central nervous system, where it plays a crucial role in stabilising microtubules. In the human brain, there are six isoforms of tau that are classified according to the number of the microtubule-binding domain repetitions, as three (3R) or four (4R) repeats [15, 16].

There are a number of neurodegenerative diseases called tauopathies that are characterised by the pathological accumulation of tau [15,16,17]. They are subdivided into primary tauopathies in which tau deposits predominate, and secondary tauopathies in which other protein aggregates predominate as in AD with Aβ deposits. In addition, tau protein deposits also vary in each of the neurodegenerative diseases, in terms of their morphology or ultrastructural conformations [16, 17] (Table 2).

Table 2 Histopathological appearance of tau isoforms and conformation in major tauopathiesTau PET tracers

The tau PET compounds initially developed were [18F]THK5317, [18F] THK5351, [18F]AV1451 (also known as [18F]flortaucipir), and [11C]PBB3) [2, 15]. However, [18F]flortaucipir was the first widely used tau agent, and was approved for clinical use by the FDA in 2020 and more recently by EMA in 2024. [18]. This first-generation of tau radiopharmaceutical showed to have a high affinity for paired helical filaments and neurofibrillary tangles for the different tau isoforms, which mainly characterise AD (AD-related tau) [15, 19]. Post-mortem studies have shown that tau PET radiopharmaceuticals bind selectively to phospho-tau and not to amyloid or other aggregates [15]. However, these tracers are limited by off-target binding (binding to intracerebral areas where tau aggregates are not localised) to the basal ganglia and the choroid plexus, brainstem (substantia nigra), and dural venous sinuses. Additionally, the [18F]THK compounds are shown to be displaced by monoamine oxidase (MAO) inhibitors [20].

These limitations prompted the development of a second-generation of tau tracers, including 18F]MK-6240, [18F]RO-948, [18F]RO1643, [18F]RO4693, [18F]PI-2620, [18F]GTP1, [18F]PM-PBB3, [18F]AM-PBB3, [18F]JNJ311, and [18F]JNJ-067, although none have yet received FDA or EMA approval [2, 15, 16]. These compounds were developed to minimise off-target binding, which significantly improves the assessment of tau accumulation, with a better ratio between pathological and normal values.

These new tracers have been shown in post-mortem studies to bind to AD-associated tau [21,22,23], and to correlate with AD dementia severity and symptoms, in contrast to Aβ [13, 16] (Fig. 3).

Fig. 3

[18F]FDG PET (A) and [18F]flortaucipir (B) in an AD patient. Images show temporoparietal and frontal hypometabolism (A), which correlates with AD-related tau deposits

Some recently developed tau PET tracers, such as [18F]PI-2620, have also been shown to bind to 4R tauopathies, such as PSP or corticobasal degeneration (CBD), showing promising results for the diagnosis of other non-AD tau pathologies [15, 23, 24].

According to the FDA and