Evaluation of Neurodegenerative Disorders with Amyloid-β, Tau, and Dopaminergic PET Imaging: Interpretation Pitfalls

AMYLOID PET

The guidelines of the Society for Nuclear Medicine and Molecular Imaging and the European Association of Nuclear Medicine define clinically appropriate use of amyloid PET, indicated in the setting of objective evidence of unexplained mild cognitive impairment assessed by a dementia expert. Amyloid PET is also appropriate for those who meet clinical criteria for possible AD but with uncertainty about the diagnosis, as well as patients who have progressive dementia at an atypically early age (<65 y) (16,17). Recently, disease-modifying amyloid antibody therapies have entered clinical practice. Confirmation of abnormal amyloid-β deposition as a therapeutic target may be obtained with amyloid PET (Table 3) (18–21). The clinical trials leading to accelerated U.S. Food and Drug Administration (FDA) approval of the amyloid-targeted therapies aducanumab and lecanemab used amyloid PET as a method to define amyloid-β positivity, a core inclusion criterion for the trials, as well as to evaluate therapeutic efficacy (19–21). Amyloid clearance measured on PET was a trial endpoint for both aducanumab and lecanemab (19,20). Follow-up amyloid PET scans after administration of antiamyloid therapies may be clinically useful for assessing response to treatment, in parallel to the use of amyloid PET as a biomarker in trials, although accessibility for this use is limited by reimbursement. For example, decreased amyloid-β deposition as evaluated by PET after lecanemab was correlated with the therapeutic benefit of delayed cognitive decline (20). Amyloid PET may provide clinically useful prognostic information when weighing the risks and benefits of continued therapy.

The amyloid radiotracer ¹¹C-Pittsburgh compound B was the first available amyloid tracer and has been used extensively in the research setting (22). At present, 3 amyloid radiotracers have FDA approval: ¹⁸F-florbetaben, ¹⁸F-florbetapir, and ¹⁸F-flutemetamol (Table 4). Each of these has specific properties and differences in procedural standards and acceptable interpretation methods, as defined in the FDA package inserts (Table 4) (23–25). For each, the final goal in interpretation is a binary decision of positive or negative. As more experience with antiamyloid therapy accumulates, the ability to monitor changes and therapeutic response with amyloid PET may influence patient management decisions. If amyloid PET is used for monitoring treatment response, as has been done in the clinical trial setting, quantification techniques will also likely become germane to clinical interpretation.

For all FDA-approved amyloid tracers, image interpretation should be performed without consideration of collateral clinical information, mirroring the methods used in clinical studies assessing the radiotracers’ performance (4). Although this approach may be counterintuitive, the aim is not to synthesize comprehensive information into a clinical diagnosis such as AD but rather to categorize the PET as a positive or negative biomarker for amyloid-β in an unbiased manner.

For all amyloid tracers, the underlying principle for image interpretation is that the tracer binds to normal white matter but spares gray matter. In a normal (negative) scan, this results in a clearly visible outline of the branching white matter structures (Fig. 1). Loss of gray–white differentiation, outward extension of radiotracer from the white matter to the cortical surface, or more intense gray matter radiotracer binding relative to white matter are features of abnormal scans.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Amyloid PET. (Top row) Positive amyloid PET scan with loss of gray–white differentiation and areas of abnormally increased cortical uptake in multiple regions of cerebral hemispheres. This is compatible with presence of moderate to frequent amyloid-β neuritic plaques. (Bottom row) In contrast, negative amyloid PET scan has distinct gray–white contrast in all lobes, with distinctly visible branching white matter tracts. SUVR = SUV ratio.

For ¹⁸F-fluorbetapir, a method for systematic image interpretation of grayscale axial images is defined, starting with inspection of the cerebellar gray–white differentiation and proceeding to the occipital lobe, the temporal lobe, the frontal lobe, and finally the parietal lobe. Each of these lobes in both hemispheres count as one region. A total of 2 regions must be abnormal for a scan to qualify as positive on the basis of loss of gray–white differentiation; however, one abnormal region may qualify a scan as positive on the basis of cortical uptake exceeding the adjacent white matter (24).

For ¹⁸F-florbetaben, a similar method is defined scrolling from inferior to superior but with slightly different regions, starting with the cerebellum and proceeding to the temporal lobe, frontal lobe, precuneus and posterior cingulate, and finally the parietal lobe. These regions may contribute to a positive scan designation, with the precuneus and posterior cingulate considered separately from the parietal lobe. Positive scans require abnormal uptake in most slices within a brain region, and this uptake can be further subdivided into moderate amyloid-β deposition (small areas of abnormal uptake within ≥1 region) or pronounced amyloid-β deposition (large and confluent areas of abnormal uptake within ≥1 region) (25).

For ¹⁸F-flutemetamol, a manufacturer-specified color scale should be used with the pons set at 90% of the maximum intensity and a minimum intensity of 0. The following regions are reviewed separately in specified planes, each of which counts toward criteria for a positive scan: the frontal lobe (axial, optional sagittal), the precuneus/posterior cingulate (sagittal, optional coronal), the lateral temporal lobe (axial, optional coronal), the inferolateral parietal lobe (coronal, optional axial), and the striatum (axial, optional sagittal) (23). The striatum is assessed only for ¹⁸F-flutemetamol, and the optional sagittal plane can be helpful for detecting a normal striatal gap between the frontal white matter and the thalamus.

Regionally positive amyloid PET scans can be more difficult to identify, and it is necessary to scrutinize each lobe of the cerebral hemispheres for the integrity of the gray–white contrast (Fig. 2). Amyloid PET interpretation can also be challenging in the setting of brain parenchymal volume loss, a common scenario. Enlargement of the cerebrospinal fluid spaces due to volume loss may mimic a normal branching white matter pattern when the PET images are reviewed alone (Supplemental Fig. 1; supplemental materials are available at http://jnm.snmjournals.org). Use of multiplanar reconstructions and fusion with anatomic images can be most helpful for clarification of the outer borders of the cortex. Familiarity with the major white matter tracts can also be helpful, as the association tracts within a cerebral hemisphere should be clearly visible as distinct radiotracer-avid structures (Fig. 1). Other potential pitfalls that may result in an inaccurate interpretation can be technical in nature. A low-count study can reduce gray–white matter contrast, resulting in an inaccurate categorization of a normal scan as positive. When one is systematically reviewing the images, starting with the cerebellum may provide a reliable internal control for the degree of gray–white contrast to expect in the cerebral hemispheres. Increased image noise may be conspicuous in the extracranial soft tissue, an additional indicator of a low-count study. Increased uptake in an extracranial structure, such as the parotid glands, scalp, or even an osseous or intracranial mass, can impact automated windowing and leveling of the study, leading to an inaccurate interpretation. To correct the windowing and leveling of the examination, the abnormally radiotracer-avid structure can be omitted from a selected representative region of the brain that includes cerebral gray and white matter. Some intracranial masses such as meningiomas are known to bind to some amyloid radiotracers, and it is important to avoid mistaking such lesions for radiotracer-avid cortex (Supplemental Fig. 2) (26). Anatomic fusion images or comparison MRI examinations may be helpful for confirming the presence of a mass lesion and diagnosing such lesions more definitely.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Regional positive amyloid PET: baseline and follow-up ¹¹C-PiB PET examinations. (Top row) Right parietal regionally positive examination with loss of gray–white contrast in right parietal lobe. Regional uptake progressed on follow-up in 7 y (arrows) and correlated to amyloid-β neuritic plaques at autopsy. Cerebrospinal fluid markers for amyloid were negative. (Bottom row) Bilateral frontal regionally positive examination with loss of gray–white contrast in left greater than right frontal lobes (arrows). Uptake progressed over 6 y and correlated with frontal amyloid-β neuritic plaques at autopsy. Cerebrospinal fluid markers for amyloid became positive in second case 7 y after baseline PET.

A limitation of amyloid PET for characterizing AD pathology is that radiotracer binding correlated with both neuritic amyloid-β plaques and diffuse amyloid-β plaques. Diffuse plaques are noncompact deposits that lack neuritic components. Diffuse plaques are commonly found in aged brains and are not specific for AD. A positive amyloid PET study reflective of diffuse plaques may be present in pathologic aging or alternative neurodegenerative diagnoses (Fig. 3) (27). Tau PET may potentially provide additional diagnostic certainty. Cases of low amyloid-β plaque burden may be undetectable with amyloid PET, an additional limitation of the imaging modality (27,28). Follow-up scans may be useful for detecting progressive amyloid accumulation. Amyloid-β accumulation detected on amyloid PET has been observed with other neurodegenerative processes such as dementia with Lewy bodies, atypical AD, and frontotemporal lobar degeneration, and in isolation, a positive amyloid PET study should not be viewed as sufficient for a diagnosis of AD (29,30). Nevertheless, a negative amyloid PET study is able to reliably exclude AD—clinically valuable information in deciding whether to pursue antiamyloid therapy or whether to consider alternative diagnoses.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3.

Pitfall of diffusely positive amyloid PET in dementia with Lewy bodies. (Top row) ¹¹C-PiB PET is positive and shows diffuse loss of gray–white contrast in both cerebral hemispheres. (Bottom row) ¹⁸F-FDG PET z score map (left and center) in same person shows hypometabolism in occipital lobes, atypical region for AD, in pattern suggestive of dementia with Lewy bodies. ¹²³I-ioflupane SPECT (DaTscan; GE Healthcare) is abnormal (right), further supporting diagnosis of dementia with Lewy bodies, which was ultimately confirmed at autopsy. In non-Alzheimer neurodegenerative pathology, presence of diffuse plaques without neuritic components may still be associated with abnormal amyloid PET. SUVR = SUV ratio.

TAU PET

The spatial distribution of pathologic tau hyperphosphorylation and neurofibrillary tangle deposition corresponds to AD pathology, as characterized in Braak neurofibrillary tangle staging as a measure of abnormal tau at autopsy. Antemortem evaluation of pathologic tau distribution can be performed with PET. ¹⁸F-flortaucipir (AV-1451) is the only FDA-approved radiopharmaceutical for tau PET and has been found to closely follow neurofibrillary tangle Braak staging for AD (Table 4) (31,32).

Per the FDA package insert, ¹⁸F-flortaucipir is indicated to assess tau burden in cognitively impaired adults being evaluated for AD. The scan should be interpreted without consideration of collateral information such as clinical data or other biomarkers, which may bias the interpretation. At present, tau PET is not indicated for evaluating non-AD tauopathies or chronic traumatic encephalopathy (32,33).

¹⁸F-flortaucipir binds with high affinity to paired helical filament tau, and abnormal radiotracer binding in the neocortex above background is the basis for identifying a positive tau PET scan (Fig. 4) (34). A threshold level of background uptake is set at 1.65-fold the average cerebellar uptake, and the manufacturer-specified display guidelines for ¹⁸F-flortaucipir are devised to set the color scale to show a transition above background level uptake. Sequentially, neocortical uptake should be assessed in each lobe: temporal, occipital, parietal, and frontal. The temporal lobe should be subdivided into quadrants (Fig. 4) including anterolateral, anterior mesial, posterolateral, and posterior mesial temporal. Only the posterolateral temporal quadrant can contribute to classification of a positive tau PET scan. Uptake in the anterior and medial temporal lobe does not meet visual interpretation criteria for a positive tau PET scan. Abnormal uptake in the parietal lobe/precuneus or in the occipital lobes may also qualify a scan as positive for widely distributed tau pathology. Abnormal frontal uptake may or may not be identified in positive scans. Negative scans may have ¹⁸F-flortaucipir uptake in the medial or anterolateral temporal lobes, frontal lobes, or deep gray nuclei and white matter (32,34).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 4.

Tau PET with ¹⁸F-flortaucipir. (Top row) Positive tau PET with areas of abnormally increased (red and orange color overlay) uptake in lateral temporal, frontal, and parietal lobes, including precuneus and posterior cingulate. Temporal lobes are divided into quadrants (dotted lines). Only uptake in posterolateral quadrant of temporal lobe should be used to consider ¹⁸F-flortaucipir PET scan positive using visual interpretation. (Bottom row) Negative tau PET with no abnormally increased regions of uptake in cerebral hemispheres. SUVR = SUV ratio.

As with amyloid PET radiotracers, volume loss can be a pitfall, rendering distinction of neocortical binding from white matter uptake difficult. Careful correlation with anatomic images can be useful in the setting of parenchymal volume loss. Small noncontiguous foci of uptake should be interpreted with caution, particularly in scans with increased noise, as these can lead to a false-positive assessment (35).

Off-target binding of ¹⁸F-flortaucipir is a limitation (Fig. 5) and can involve structures such as the brain stem nuclei and substantia nigra, striatum, choroid plexus, leptomeninges, and blood vessels (36). Possible mechanisms have been suggested for off-target binding, such as radiotracer affinity for monoamine oxidase A and B, pigmented compounds such as neuromelanin, and mineralized structures (32,36).¹⁸F-flortaucipir binding has also been reported within meningiomas (Supplemental Fig. 2) (37). Binding to the leptomeningeal structures could be mistaken for cortical uptake, a pitfall that may be avoided with careful attention to coregistered anatomic images. Off-target binding may potentially be ameliorated with next-generation tau radiotracers (36).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 5.

Off-target binding on tau and amyloid PET. (Top 2 rows) ¹⁸F-flortaucipir PET shows off-target uptake in substantia nigra (red rectangle), choroid plexus (white arrows), muscles of mastication (yellow rectangles), extraocular muscles (cyan arrows), and multiple osseous structure including calvarium (yellow arrows), occiput (pink arrow), and sphenoid bones (green arrows). (Bottom row) Off-target osseous uptake can also be seen on amyloid PET (¹⁸F-florbetapir PET/CT) showing avid region of hyperostosis frontalis interna (ellipses), which in some cases could mimic cortical uptake without anatomic correlation. SUVR = SUV ratio.

¹⁸F-flortaucipir uptake correlates strongly with the 3R + 4R isoform of tau associated with AD but is not strongly associated with preferential 3R or 4R isoforms of tau, a limitation of the radiotracer’s ability to assess other non-AD tauopathies in cases of Pick disease (3R predominant), corticobasal degeneration (4R predominant), or progressive supranuclear palsy (4R predominant). Weak binding to TDP-43 (transactive response DNA binding protein of 43 kDa) may confound diagnosis in cases of frontotemporal lobar degeneration TDP. Although tau PET is an important biomarker for AD, positive uptake on tau PET can be seen in alternative neurodegenerative conditions such as semantic dementia (Fig. 6) and prion protein defects (Supplemental Fig. 3) (27,28).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 6.

Positive tau PET in semantic dementia. (Top row) Multiple PET examinations from patient with semantic dementia, commonly with underlying TDP-43 proteinopathy. ¹⁸F-flortaucipir PET was positive, more prominently in left than right temporal lobes (white arrows) and left parietal lobe (yellow arrow). (Second row) ¹¹C-PiB PET was negative, with preserved gray–white contrast indicating lack of moderate or frequent amyloid-β neuritic plaques. (Third and fourth rows) ¹⁸F-FDG PET z score maps demonstrated regions of hypometabolism correlating with tau PET, worst in left anterior temporal lobe (arrows). Pattern of hypometabolism is characteristic of semantic dementia. Although tau PET was positive, which can be seen with AD, alternative neurodegenerative entities can also result in abnormal tau PET. In this case, negative amyloid PET and pattern of ¹⁸F-FDG hypometabolism suggest semantic dementia. SUVR = SUV ratio.

A significant limitation of ¹⁸F-flortaucipir PET in evaluation of AD is poor detection of early tau deposition. According to the FDA package insert instruction for ¹⁸F-flortaucipir PET, the regions of early tau deposition corresponding to early Braak stages are to be excluded from visual interpretation (Fig. 4). In general, pathologic tau neurofibrillary tangle accumulation occurs earliest in the mesial temporal structures such as the entorhinal cortex, corresponding with early Braak stage distribution (38). Specificity for diagnosing AD on the basis of medial temporal involvement alone may be limited, as this can occur in cognitively unimpaired patients. Tau uptake in the entorhinal cortex, hippocampal formations, parahippocampal gyrus, and middle temporal lobe gyrus strongly correlates with poor memory performance, even in cognitively unimpaired individuals (39). One factor limiting accurate assessment of early tau on PET is the detection of low levels of uptake above background noise. Technical developments and innovative image processing techniques may help to improve early tau detection, such as the use of the overlap index in sequential scans (40). Future updates in Alzheimer diagnostic criteria may incorporate distinct categories of positive tau PET, distinguishing categories of isolated medial temporal uptake and moderate or high neocortical uptake (11). At present, these distinctions are not made in the clinical visual interpretation of tau PET (32).

Tau PET in cognitively unimpaired individuals has demonstrated uptake in regions that would correspond to early Braak stage involvement but also in extratemporal locations corresponding to more advanced Braak stages (38). The variable distribution of tau pathology in both cognitively unimpaired and cognitively impaired individuals underscores the need for placing a positive tau PET scan within a greater clinical context to support a diagnosis of AD.

DOPAMINERGIC PET

Dopaminergic PET is indicated for the evaluation of parkinsonian syndromes in adults (Table 5) (41). Parkinsonian syndromes result from loss of the dopaminergic neurons projecting from the substantia nigra pars compacta to the striatum and presents as a movement disorder characterized by bradykinesia, rigidity, tremor, or postural instability. Typical parkinsonian syndrome (idiopathic Parkinson disease) is the most common parkinsonian syndrome (13,42). Other distinct neurodegenerative entities that present as parkinsonian syndromes include dementia with Lewy bodies, progressive supranuclear palsy, multisystem atrophy, and corticobasal degeneration and are collectively termed atypical parkinsonian syndromes (43). Typical parkinsonian syndrome is a clinical diagnosis relying of the presence of bradykinesia and either rigidity or rest tremor, as well as other supportive criteria and absence of findings suggesting an alternative diagnosis. Response to dopaminergic therapy is an important supportive criterion for typical parkinsonian syndrome, and poor response is suggestive of an atypical parkinsonian syndrome (44).

¹⁸F-DOPA is a molecular precursor of the dopamine neurotransmitter, and physiologic uptake is expected in viable presynaptic dopaminergic neurons (45). One hour before intravenous radiotracer injection, 150 mg of oral carbidopa should be administered to inhibit peripheral decarboxylation of ¹⁸F-FODA and augment brain radiotracer availability. Multiple drug classes used in treated parkinsonian syndromes should be discontinued 12 h before radiotracer injection, including dopamine agonists, reuptake inhibitors, releasing agents, catechol-O-methyltransferase inhibitors, and monoamine oxidase inhibitors. After ¹⁸F-DOPA injection and uptake, a 3-dimensional PET acquisition is recommended. Axial images should be reconstructed along the anterior–posterior commissure line. Images are to be interpreted on the basis of visual inspection, without collateral clinical information used to influence the classification as normal or abnormal (41).

A normal (negative) scan should demonstrate radiotracer-avid basal ganglia structures including the caudate head and putamen (Fig. 7), together forming a crescent shape. A positive (abnormal) scan may have asymmetric or symmetric reduction of uptake in the putamen, resulting in either diminished intensity of uptake or a truncated region of uptake in the putamen. Uptake may also be reduced in the caudate nuclei in a positive scan.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 7.

Dopaminergic PET. (Left) Normal ¹⁸F-DOPA PET examination with expected uptake in caudate and putamen bilaterally (arrows). (Center) Abnormal ¹⁸F-DOPA PET examination in 43-y-old woman with onset of Parkinson disease at early age. Abnormal examination shows diminished uptake in bilateral putamina, asymmetrically worse on patient’s right, and subtle decreased uptake in patient’s right caudate (arrows). (Right) In same patient, ¹²³I-ioflupane SPECT (DaTscan; GE Healthcare) was also abnormal, with decreased basal ganglia uptake (arrows) causing loss of expected comma-shaped region of uptake, asymmetrically worse on patient’s right. SUVR = SUV ratio.

A positive scan may indicate a typical or atypical parkinsonian syndrome by detecting loss of striatal dopaminergic neurons in Parkinson disease, progressive supranuclear palsy, corticobasal degeneration, and multiple-system atrophy. Dopaminergic PET may also be useful in diagnosing dementia with Lewy bodies (46). These different entities cannot be distinguished on the basis of dopaminergic PET. Conditions resulting in a negative scan may include essential tremor or other causes of parkinsonian syndromes including pharmacologic, psychogenic, or vascular etiologies (15). In vascular parkinsonism, correlation with anatomic imaging may help avoid an interpretation pitfall. A basal ganglia infarct may explain the patient’s symptoms and could result in an abnormal truncated appearance on dopaminergic PET.

¹⁸F-DOPA PET has been reported to have a diagnostic performance similar to that of ¹²³I-ioflupane SPECT/CT; however, some discrepancies have been reported, suggesting that more cases are positive with ¹²³I-ioflupane SPECT (47–49). In the setting of dopaminergic neuron loss, upregulation of dopamine synthesis in the residual neurons could preserve uptake of ¹⁸F-DOPA relative to ¹²³I-ioflupane. Advantages of ¹⁸F-DOPA PET include improved spatial resolution and a shorter uptake time of 1.5 h, compared with 3–6 h for ¹²³I ioflupane SPECT (41,50). At present, ¹²³I-ioflupane SPECT is more widely available and reimbursed.

Multiple investigational agents can image the dopaminergic and monoaminergic systems, including postsynaptic dopamine receptor ligands and presynaptic targets such as the dopamine active transporter. ¹⁸F-DOPA and ¹²³I-ioflupane are the radiopharmaceuticals used clinically in the United States, with ¹²³I-ioflupane being more frequently used (51,52).

In addition to parkinsonian syndromes, dopaminergic PET has been investigated for use in evaluation of schizophrenia, psychosis, and glioma but does not have current FDA regulatory approval for these indications (53,54).