Using Positron Emission Tomography and Carbon 11–Labeled Pittsburgh Compound B to Image Brain Fibrillar β-Amyloid in Adults With Down Syndrome: Safety, Acceptability, and Feasibility

Objective To investigate the safety, acceptability, and feasibility of positron emission tomography (PET) using carbon 11–labeled Pittsburgh Compound B ([¹¹C]PiB) to measure cerebral β-amyloid in adults with Down syndrome (DS) and to explore if the technique differentiates between participants with and without Alzheimer disease (AD).

Design Proof-of-principle case-controlled study of a nonrandomly selected cohort of participants with DS (with or without AD) compared within group and with healthy controls without DS. All had dynamic [¹¹C]PiB PET and magnetic resonance imaging. Carbon 11–labeled PiB binding in the regions of interest associated with AD was quantitatively analyzed.

Setting Wolfson Brain Imaging Centre, Cambridge, England.

Participants Nine with DS (aged 25-64 years), of whom 5 had a diagnosis of AD, and 14 healthy controls without DS (aged 33-69 years).

Main Outcome Measure Positive [¹¹C]PiB binding in regions of interest.

Results The scanning process was feasible and acceptable with no adverse events or safety concerns. Maps and regional values of nondisplaceable binding potential were produced using the reference tissue–input Logan plot, with the cerebellum used as the reference tissue. When compared with the healthy control group without DS, only participants with DS older than 45 years had significant [¹¹C]PiB binding in regions of interest usually associated with AD, whether or not they had clinical evidence of dementia.

Conclusions Dynamic [¹¹C]PiB PET can be used successfully to measure cerebral β-amyloid deposition in DS. A clinical diagnosis of AD and age appear to be predictors of [¹¹C]PiB binding in regions of interest, but given the small numbers, we cannot generalize the results.

Down syndrome (DS) or trisomy 21 is the most common identified cause of a mild to moderate intellectual disability, with an incidence of 1 in 650 to 1000 live births worldwide.¹ Clinical studies have found an increasing age-related prevalence of dementia of the Alzheimer type in people with DS with an age at onset typically significantly earlier than in the general population. The rates reported are a few percent in those aged 30 to 39 years, between 10% and 25% in the 40 to 49 years age group, between 20% and 50% in the 50 to 59 years age group, and between 30% and 75% in those older than 60 years.^2-9

Independent of the presence of the clinical symptoms of Alzheimer disease (AD), postmortem studies indicate that people with DS, who have died when 40 years or older, invariably have evidence of Alzheimer-like neuropathology, in particular an early accumulation of β-amyloid deposits (as senile plaques).^2,10-17 One explanation to account for the striking association between DS and the early development of Alzheimer-like neuropathology is that the amyloid precursor protein (APP) is present in triplicate in the great majority of people with DS because it is located on chromosome 21.¹⁸ For this reason, there is the cerebral accumulation of β-amyloid, which is derived from the proteolytic processing of APP. The subsequent observation that mutations in this gene are associated with the early onset of AD in specific families¹⁹ led to a resurgence of the “amyloid cascade hypothesis” of AD.²⁰ This hypothesis proposes that there is a progressive (extracellular) accumulation of fibrillar aggregates of the peptide products of APP. These β-amyloid deposits consist of β-amyloid peptides of various residue sizes (eg, 42 and 40), and the plaque-associated fibrillar forms of β-amyloid are thought to trigger cellular processes that lead to synaptic dysfunction and cell death.

The development of β-amyloid ligand imaging using positron emission tomography (PET) and carbon 11–labeled Pittsburgh Compound B ([¹¹C]PiB) has now made possible the investigation of the temporal dynamics of β-amyloid accumulation in vivo,²¹ providing a distinct advantage over autopsy studies. Carbon 11-labeled PiB is a radio-labeled, chemically modified version of Thioflavin T with a very high specific binding affinity for fibrillar β-amyloid plaques.^21-26 Despite some immunological and chemical heterogeneity,^27,28 fibrillar β-amyloid is the species of amyloid common to mature neurotic plaques in AD in both DS and the general population.²⁹ In conjunction with magnetic resonance imaging (MRI), to aid tissue localization, it has been possible to detect the β-amyloid distribution and load in human participants.²¹ This technique has been shown to be robust and reproducible.^30-35 In AD in the general population, the relative distribution of β-amyloid as measured by [¹¹C]PiB-PET scanning correlates well with neuropathological findings;^25,26,36,37 strong frontal, temporoparietal, and striatal binding; and relatively low but still significant mesial temporal plaque burden.²¹

Fibrillar β-amyloid deposits (detected by [¹¹C]PiB-PET) have been found to predate clinical symptoms of AD by many years and also to predate markers of neuronal loss (eg, glucose metabolism or regional brain atrophy),^30,33,34,38 thus supporting the hypothesis that β-amyloid accumulation is likely to be one of the earliest events prior to the development of the clinical manifestations of the illness.³⁹

The now well-established link between APP, β-amyloid, neuropathology, and dementia makes DS a unique model for the study of the role of the APP and β-amyloid in the pathogenesis of AD.⁴⁰ It would therefore be reasonable to expect that people with DS who are older than 40 years would have positive amyloid binding as measured by [¹¹C]PiB-PET imaging.

The advantage of involving people with DS in similar studies, as opposed to the general population, is that adults with DS in their 30s have a very high risk of developing dementia that can be followed up as they “convert” to disease, and the likely cause of the dementia is much more likely to be purely “β-amyloid driven.” Furthermore, because atherosclerosis and hypertension are rarer in DS,^41-43 data interpretation is less likely to be complicated by cases of vascular dementia. The process of amyloid deposition in DS therefore provides a valuable setting, not just to study the natural history of possible amyloid-induced brain atrophy, but also an excellent context for proof-of-principle studies with agents aimed at slowing amyloid deposition or increasing amyloid clearance.

The main aims of this proof-of-principle study were to determine the acceptability, safety, and feasibility of using [¹¹C]PiB-PET imaging in adults with DS. Only if all of these factors had a positive outcome could a larger project be scientifically and ethically justified. Given the fact that people with DS frequently have intellectual disabilities, we also developed and evaluated the use of special information material to aid the decision-making process of people with DS asked to take part.

The study was approved by Cambridgeshire 3 Research Ethics Committee on the condition that for this proof-of-principle study only participants with DS with the capacity to consent to the research were included. The UK Administration of Radioactive Substances Advisory Committee also approved the study.

People with DS as potential participants were either recruited through local clinical services for people with intellectual disabilities or by advertisement through the Down's Syndrome Association. We developed materials to provide participants with sufficient information tailored to their level of comprehension and designed to assist them in making an informed decision about the nature and purpose of the study. This included specially designed materials (eg, booklets with photographs) explaining the reasons behind the study and practical aspects of the scanning procedures. The potential participants were visited on several occasions to discuss the study and to build rapport and trust.

Interested persons with DS had the chance to habituate to the scanners by structured orientation visits to the Wolfson Brain Imaging Centre with their support workers. The participants were free to withdraw at any time. Anxiolytic medications prior to the scan were not offered. Following the actual scans, all participants were asked to provide structured feedback about their experiences. Participants were not paid for participating but their expenses were reimbursed.

All participants with DS were in the mild to moderate intellectual ability range (Table). The participants with DS were assessed for dementia using the procedure described in Holland et al,⁷ based on information gained using a modified version of the CAMDEX informant interview,⁴⁴ which was carried out with each participant's main carer. The diagnostic criteria used included CAMDEX-R criteria for dementia of the Alzheimer type⁴⁵ and Gregory and Hodges⁴⁶ criteria for dementia of the frontal type. All participants meeting clinical criteria for dementia (including those with a “frontal-like” presentation resembling that of dementia of the frontal type, reported to precede the clinical features of AD in DS⁴⁷) were included within the dementia-positive group in the analysis.

Fourteen healthy volunteers who provided informed consent aged 33 to 69 years (eTable) underwent clinical assessments, MRI, and PET scans.

Carbon 11–labeled PiB was produced at the Wolfson Brain Imaging Centre radiochemistry laboratories where methods have been developed for regularly producing [¹¹C]PiB with high radiochemical purity (>95%) and specific activity (>150 GBq/μmol) under Good Manufacturing Practices conditions. The PET scanning was performed on a GE Advance PET scanner (GE Medical Systems, Waukesha, Wisconsin). Each participant underwent a germanium 68 transmission scan (15 minutes) for attenuation correction, followed by 550 MBq of [¹¹C]PiB injected as a bolus through an antecubital venous catheter. Dynamic PET emission scanning in 3-dimensional mode was undertaken for 90 minutes following injection (58 frames: 18 × 5 seconds, 6 × 15 seconds, 10 × 30 seconds, 7 × 1 minutes, 4 × 2.5 minutes, 13 × 5 minutes); this acquisition period was used in the seminal [¹¹C]PiB methods study by Price et al.⁴⁸

Participants also had an anatomical MRI scan on a 3-T Siemens Tim-Trio (Siemens Medical Solutions, Erlangen, Germany) that took approximately 30 minutes. Images were acquired using a 3-dimensional magnetization-prepared rapid gradient-echo sequence (repetition time/echo time/inversion time = 2300 milliseconds/2.98 milliseconds/900 milliseconds; flip angle, 9°; 1 average; 176 slices; 256 × 256 matrix size; 1 × 1 × 1-mm voxel size). These data were used for coregistration to the PET data and enabled anatomical delineation of regions of interest (ROIs).

Each dynamic [¹¹C]PiB-PET scan was reconstructed using the PROMIS 3-dimensional filtered back projection algorithm^48,49 installed on the GE Advance scanner, with corrections applied for randoms, dead time, normalization, scatter, attenuation, and sensitivity. Each image had the dimensions of 128 × 128 × 35, with a voxel size of 2.34 × 2.34 × 4.25 mm. To reduce any deterioration due to head motion, images were realigned using SPM8 (www.fil.ion.ucl.ac.uk/spm/software/), which was also used to coregister the mean realigned image to the MRI and reslice all the PET images to the MRI. To allow the use of standard-space ROIs, the MRI, and thence the coregistered PET images, was normalized to the Montreal Neurological Institute/ICBM152 T1–weighted template⁵⁰ using SPM8. The following neocortical ROIs were chosen from the automated anatomical labeling atlas⁵¹—anterior cingulate, posterior cingulate, and prefrontal and superior parietal—because these are preferentially affected in AD, as well as the calcarine cortex and the hippocampus, which are less affected by [¹¹C]PiB deposition in AD.^21,30,48,49

For each participant, a reference tissue ROI was drawn in the superior gray matter of the cerebellum using a 90% threshold on the gray matter segment produced from the individual's MRI by SPM8, which was then smoothed to the PET resolution. This approach ensured consistent tissue was sampled by the ROI irrespective of the cerebellum size, which is known to be smaller in those with DS. For each participant, the cerebellum ROI was applied to the dynamic PET image sequence in MRI space to produce a reference tissue time-activity curve. The cerebellum was chosen as the reference tissue because it is relatively free of fibrillar β-amyloid (and thus specific [¹¹C]PiB binding) across a range of ages in people with DS,^50,52 including individuals as old as 65 years.^51,53

Maps of nondisplaceable binding potential (BP_ND)⁵⁴ and regional BP_ND estimates were produced from the realigned dynamic PET images and the reference tissue time-activity curve using the reference tissue–input Logan plot.^52,55 Regional modeling was applied to time-activity curves produced from the average of left and right regions. The BP_ND was measured in these ROIs in all 14 healthy participants without DS and the averages were plotted against age (eFigure), which showed a dramatic rise in BP_ND in those older than 60 years; we therefore used all those younger than 60 years as a control group (n = 6). The BP_ND measurements in each participant with DS in each ROI were compared with the upper 95% confidence limit computed from the data in the 6 controls, using the sample size–adjusted t = 2.571. The BP_ND was considered significant (PiB positive) if it lay outside this confidence interval.

In total, 21 people with DS were approached to seek their consent for inclusion in the study. Of those seen, 13 were interested enough to visit the Wolfson Brain Imaging Centre and 11 subsequently agreed to take part. One person found the elevation of the scanning table anxiety provoking and asked to withdraw. Another person was discovered to have a preexisting structural cerebral abnormality on her MRI scan and so she did not have the PET scan. One participant (subject 9) was excluded from the MRI scan because of metal screws, but we proceeded with the PET because he was the only participant older than 60 years and hence we were able to image amyloid across the full spectrum of ages. His data were excluded from the final analysis. We achieved PET and MRI scans of sufficient quality for 9 people with DS in total (4 women and 5 men aged between 25 and 64 years); patient demographics are given in the Table. Five participants had a clinical diagnosis of AD as determined at time of recruitment. One participant (subject 8) was initially deemed not to have any signs of AD but on informant interview had convincing evidence of functional decline.

People with DS were successfully able to undergo MRI and PET imaging with the aid of visual information materials and through supported experience of the scanners prior to actual scanning. All viewed the experience favorably. Intravenous cannulation was well tolerated, and no adverse effects to the radiotracer were reported or observed. The MRI scanning was generally well tolerated; however, mild discomfort in relation to the noise and confined space was reported by 4 of 8 of the participants.

We found that those healthy individuals without DS younger than 60 years (n = 6) had BP_ND close to zero in all ROIs, whereas BP_ND increased greatly in most subjects without DS older than 60 years (eFigure). These data are entirely consistent with previous work.^53,56

Figure 1 shows the spatially normalized PiB BP_ND map overlaid on MRI for a subject with DS and AD (subject 6, male, 51 years old) alongside corresponding images for a control without DS (male, 62 years old). We found significant PiB binding in all 6 participants with DS 45 years and older. In these participants, PiB binding greater than the 95% confidence interval of controls was demonstrated in all ROIs except the hippocampus, which was PiB positive only in participants 4 and 6 (Figure 2).

This proof-of-principle study demonstrates for the first time, to our knowledge, that brain fibrillar β-amyloid can be successfully imaged using [¹¹C]PiB PET in people with DS in vivo. In agreement with postmortem studies, there was significant [¹¹C]PiB binding in all participants with DS who were 45 years and older. Carbon 11–labeled PiB was found to bind in all brain regions predicted by patterns of cerebral distribution of amyloid in DS,^10,13,16 except in the hippocampus, which showed less binding than the other selected ROIs (Figure 2). This is consistent with the low binding seen in sporadic AD. One participant in our study did not have clinical evidence for dementia (Table, subject 4), despite extensive positive PiB binding, suggesting that PiB binding alone is not discriminatory for the diagnosis of AD.

In the general population, there is evidence for a relatively rapid accumulation of amyloid many years before symptom onset that then reaches a steady state following which glucose/oxygen metabolism, atrophy, and cognitive decline ensue as later events.³⁴ It is also apparent that the degree of [¹¹C]PiB binding correlates poorly with cognitive performance in AD.^54,57 Taken together, this supports the hypothesis for a “catalytic” role for amyloid such that it initiates a process whereby amyloid accumulates, reaches a steady state, and then sets off a chain of events leading to synaptic dysfunction, neuronal loss, and dementia. Understanding the “accumulation stage” is presently limited to speculation. Longitudinal studies of the general population to address this question are problematic because of the large number of subjects across a range of ages required to capture individuals as they make the transition from PiB negative to PiB positive. The key strength of studying AD in DS is that virtually all participants with DS will be PiB positive by age 50 years.

Major challenges to recruitment included identifying participants with the capacity to consent and finding older participants without any signs of dementia or cognitive decline. Those participants successfully scanned invariably had the support of their carers. Anxiolytic premedication is offered to the general population before undergoing MRI; we did not have ethical permission to provide this. Although our feedback suggested that all participants enjoyed the experience of participation, some expressed reservations about the MRI. We recommend that oral anxiolytics should be offered in future studies to decrease anxiety associated with the procedure.

This proof-of-principle study has demonstrated the safety, efficacy, and acceptability of this technique; larger and systematic investigations are now indicated and are ethically and scientifically defensible, given that people with DS have the most to gain by such investigations. Because everyone with DS will eventually develop extensive cerebral amyloid deposits, medications that either vaccinate against or remove amyloid in its early stages could be universally administered to this high-risk group, with the possibility that the onset of clinical dementia could be prevented or at least delayed.

In our small study, we found that no younger participants were binders, unlike all of the older age group. In future studies, it will be informative to capture the conversion from PiB negative to PiB positive and to evaluate the relationship between cerebral amyloid and cognition, brain atrophy, and AD. This would potentially shed light on the natural history of AD in this population and might provide us with key insights into the role of amyloid in AD in the general population.

Correspondence: Shahid H. Zaman, MBChB, MD, PhD, MRCP, MRCPsych, University of Cambridge, Douglas House, 18B Trumpington Rd, Cambridge CB1 7TX, England (shz10@medschl.cam.ac.uk).

Accepted for Publication: January 5, 2011.

Published Online: March 14, 2011. doi:10.1001/archneurol.2011.36

Author Contributions:Study concept and design: Landt, Holland, Aigbirhio, Menon, Baron, and Zaman. Acquisition of data: Landt, D'Abrera, Aigbirhio, Fryer, and Canales. Analysis and interpretation of data: Holland, Fryer, Hong, Menon, Baron, and Zaman. Drafting of the manuscript: Landt, D'Abrera, and Zaman. Critical revision of the manuscript for important intellectual content: D'Abrera, Holland, Aigbirhio, Fryer, Canales, Hong, Menon, Baron, and Zaman. Obtained funding: Holland, Aigbirhio, and Zaman. Administrative, technical, and material support: Landt, D'Abrera, Aigbirhio, Fryer, Canales, Menon, and Zaman. Study supervision: Holland and Zaman.

Funding/Support: This work was supported by the Medical Research Council; Cambridge National Institute for Health Research Biomedical Research Centre; Wellcome Trust; Royal College of Anaesthetists; the Down's Syndrome Association; and the Health Foundation.

Additional Contributions: We are grateful to people with DS and their families and support workers for their commitment and to the staff at the Wolfson Brain Imaging Centre for their help. We thank the Down's Syndrome Association, Medical Research Council, Cambridge National Institute for Health Research Biomedical Research Centre, the National Institute for Health Research Collaborations in Leadership for Applied Health Research and Care for Cambridgeshire and Peterborough, the Wellcome Trust, the Royal College of Anaesthetists, and the Health Foundation. We thank William E. Klunk, MD, PhD, and Chester A. Mathis, PhD, for kindly providing dosimetry and toxicology data for [¹¹C]PiB.