Introduction
The leading etiological hypothesis of Alzheimer's disease (AD) points to excessive brain β-amyloid (Aβ) that aggregates to form extracellular plaques and vascular wall deposits [
1]. With increasing prevalence and associated cost of care and the likelihood of greater benefit if therapies are applied early, earlier and more accurate identification of AD has become a research priority.
Dementia is usually preceded by a transition period of cognitive decline commonly referred to as mild cognitive impairment (MCI). Characterized by an objective impairment of memory and/or other cognitive domains, MCI is not severe enough to significantly interfere with activities of daily living [
2]. The prevalence of MCI in people aged 65 is believed to be 10 to 20%, with over 10% who have been classified as MCI converting to dementia per year [
3]. Histopathologic studies on brains of MCI subjects have shown characteristic AD pathology including Aβ plaques and neurofibillary tangles in the majority of cases [
4]. MCI has been further classified based on whether memory has been affected (amnestic MCI) or spared (nonamnestic MCI), and whether the cognitive deficit affected is mainly in one cognitive domain (single-domain MCI) or more than one domain (multidomain MCI). Hence, MCI can be classified into four clinical subtypes: nonamnestic single-domain, nonamnestic multiple domains, amnestic single-domain (asMCI), and amnestic multiple domains (amMCI). These subtypes probably differ in etiology and outcome. Impaired episodic memory, which characterizes asMCI and amMCI, is thought to be a prodromal condition for AD [
3,
4].
The new research diagnostic criteria for AD and MCI allow for Aβ imaging in the workup of individuals with cognitive impairment [
5,
6]. Non-invasive Aβ imaging to confirm the presence of AD neuropathology could aid in early differential diagnosis, identify at-risk individuals, help predict or monitor disease progression, and potentially evaluate the response to disease-specific therapy.
11C-Pittsburgh Compound B (PiB) has been the most widely used agent in dementia research to assess Aβ burden
in vivo [
7]. The major disadvantage of PiB is that it is radiolabeled with carbon-11, which has a short decay half-life (20 minutes) that limits its use to centers with an onsite cyclotron and
11C-radiochemistry expertise.
To overcome these limitations, a number of novel fluorine-18 Aβ imaging tracers such as
18F-florbetaben (BAY 94-9172) [
8‐
10],
18F-florbetapir (AV45) [
11,
12] and
18F-flutemetamol (GE067) [
13,
14] have been developed. The 110-minute radioactive decay half-life of fluorine-18 allows centralized synthesis and regional distribution of these tracers as currently practiced worldwide in the supply of
18F-fluorodeoxyglucose for routine clinical positron emission tomography (PET) imaging.
18F-florbetaben (FBB; trans-4-(
N-methyl-amino)-4"(2-(2-(2-[
18F] fluoro-ethoxy)ethoxy)-ethoxy)stilbene), developed by Avid Radiopharmaceuticals (Philadelphia, USA) and Bayer-Schering Pharma (Berlin, Germany), has been shown to bind with high affinity to Aβ in brain homogenates and selectively labeled Aβ plaques and cerebral amyloid angiopathy (CAA) in AD tissue sections [
15]. After injection into Tg2576 transgenic mice,
ex vivo brain sections showed localization of FBB in regions with Aβ plaques as confirmed by thioflavin binding [
16]. At the tracer concentrations achieved during human PET studies, FBB did not show binding to α-synuclein in Lewy bodies or to tau lesions in postmortem cortices from dementia with Lewy bodies, AD or frontotemporal lobar degeneration patients [
17]. In human studies, cortical retention of FBB was significantly higher in AD patients compared with age-matched controls and frontotemporal lobar degeneration patients, with binding matching the reported postmortem distribution of Aβ plaques [
9]. Phase II clinical studies further confirmed these results [
8]. FBB is highly correlated with
11C-PiB (
r = 0.97 with a slope of 0.71) [
18], and was used to detect the presence or absence of AD pathology in the brain in participants with a wide spectrum of neurodegenerative diseases including a few MCI participants [
10]. Phase III studies for FBB have reached completion [
19].
Human postmortem studies have shown that while soluble Aβ oligomers and the density of neurofibrillary tangles strongly correlate with neurodegeneration and cognitive deficits, the density of Aβ insoluble plaques does not [
20‐
24] and Αβ burden as assessed by PET does not strongly correlate with cognitive impairment in AD patients [
25,
26]. The severity of tau pathology in AD patients is closely related to neuronal loss [
27], hippocampal atrophy [
28,
29] and memory impairment [
30,
31]. Amyloid imaging studies in MCI have shown an association between Aβ burden and memory [
32], an association that is believed to be mediated by hippocampal atrophy [
33]. Vascular pathology, as reflected in white matter hyperintensities (WMH), has been shown to be associated with cognitive impairment, particularly affecting working memory and executive function, as well as visuospatial abilities among people with MCI [
34].
The purpose of this study was to characterize FBB binding in a well-characterized MCI cohort, and to explore the relationships of Aβ burden cognitive performance, hippocampal volume (HV), and WMH.
Materials and methods
Participants
Forty-five participants fulfilling Petersen's criteria for MCI [
3] were recruited between June 2008 and December 2009 from memory disorder specialists. Fifteen healthy older controls and 15 patients who met National Institute of Neurological Disorders and Stroke-Alzheimer's Disease and Related Disorders Association criteria for probable AD, which were previously described in an earlier study [
9], were used for comparison against the MCI cohort.
Consistent with the consensus criteria for MCI at the time of enrolment [
3], all participants (and their next of kin) reported a history of cognitive decline and had objective cognitive impairment on neuropsychological assessment but remained generally independent in daily activities. In addition, participants had to be at least 60 years of age, had at least 7 years of formal education, spoke fluent English, were capable of giving informed consent, had a reliable informant capable of giving a collateral history, were able to tolerate a brain magnetic resonance imaging (MRI) scan, did not meet the National Institute of Neurological Disorders and Stroke-Association Internationale pour la Recherché et l'Enseignement en Neurosciences criteria for the diagnosis of vascular dementia, and scored ≥ 24 on the Mini-Mental State Examination (for detailed exclusion criteria, see Table S1 in Additional file
1). These participants were referred from local specialist public and private memory disorders clinics upon being diagnosed with MCI and had no other evidence of significant neurodegenerative disease, moderate or severe psychiatric illness, drug or alcohol dependence, or participated in any anti-Aβ therapeutic trial prior to enrolment.
The recruitment criterion was defined as having at least one test score falling 1.5 standard deviations below published means. For precision, subsequent classification of participants into MCI subtypes by Petersen's criteria [
3] was based instead on test scores falling 1.5 standard deviations below the mean of a carefully screened and demographically well-matched cohort living in the same region as the participants. This cohort consisted of 45 healthy older participants from the Australian Imaging Biomarkers and Lifestyle flagship study of ageing [
35] with no history of cognitive decline who had negative brain PiB scans, normal brain MRI, Clinical Dementia Rating = 0 and Clinical Dementia Rating sum of boxes = 0, and had no psychiatric illness.
Approval for the study was obtained from the Austin Health Human Research Ethics Committee. Written informed consent for participation was obtained from all participants prior to screening. Safety monitoring consisted of clinical observation, baseline ECG, hematology and biochemistry testing and measurement of vital signs before and after tracer injection. Vital signs, hematology, and biochemistry testing were repeated 1 week after injection. Participants were asked about possible adverse events after their PET scan and 1 week after injection.
Neuropsychological evaluation
Neuropsychology evaluation was conducted within 24.5 ± 15.5 days of the FBB PET scan by a licensed neuropsychologist. Evaluation consisted of the Mini-Mental State Examination, the Clinical Dementia Rating, the California Verbal Learning Test Second Edition, the Rey Complex Figure Test (RCFT), Logical Memory I and II (Wechsler Memory Scale; Story A only), the Controlled Oral Word Association Test, Categorical Fluency, the Boston Naming Task (30-item version), Digit Symbol-coding and Digit Span.
Individual composite episodic memory
z scores (EM) were generated in 44 participants by averaging the
z scores for delayed recall trials of the RCFT, the California Verbal Learning Test Second Edition, and Logical Memory II. The RCFT delayed recall score was missing for another participant and was substituted with the RCFT immediate delay score because the relationship between scores on the immediate and the delayed recall trials was very strong (
r = 0.93). Composite nonmemory
z scores in all 45 participants were calculated by averaging the
z scores for the Boston Naming Task, the Controlled Oral Word Association Test, Categorical Fluency, Digit Span, Digit Symbol-coding and RCFT copy [
32].
Image acquisition
Magnetic resonance imaging
A three-dimensional T1-weighted magnetization prepared rapid gradient echo sequence and a fluid-attenuated inversion recovery sequence were performed on either a 1.5 T or a 3 T magnetic resonance scanner prior to the PET scan.
18F-florbetaben imaging
Labeling was carried out in the Austin Health Centre for PET, as previously described [
9]. Mean specific activity at the time of injection for MCI was 60 ± 29 GBq/μmol. Imaging was performed with a three-dimensional GSO Philips Allegro PET camera. A 2-minute transmission scan using a rotating
137Cs source was performed for attenuation correction immediately prior to scanning. Each MCI participant received on average 286 ± 19 MBq FBB intravenously over 38 ± 17 seconds. Images were reconstructed using a three-dimensional RAMLA algorithm (Philips, Cleveland, USA). Images obtained between 90 and 110 minutes post injection were used for the analysis.
Image analysis
All image analysis was performed by experienced operators blind to the clinical status and cognitive test scores of the subjects.
Extraction of HVs from the three-dimensional magnetization prepared rapid gradient echo MRI data in 43 MCI cases was performed using a commercial, US Food and Drug Administration-approved, fully automated volumetric measurement program (NeuroQuant
®) [
36]. Preprocessing of the fluid-attenuated inversion recovery images was performed to correct for bias field effects and remove noise using anisotropic diffusion prior to manual segmentation of deep WMH. Manual segmentation of the WMH (PR) was performed using MRIcro software [
37]. The total WMH volume in each MCI subject was calculated, as well as the number of individual lesions. All volumes were normalized for head size using the total intracranial volume, defined as the sum of gray matter, white matter and cerebrospinal fluid volumes.
Spatial normalization and co-registration of the PET and MRI images was performed using SPM8 [
38]. PET images were processed with a semiautomatic volume of interest method. This method used a preset template of narrow cortical volume of interest that was either applied to the spatially normalized MRI and then transferred to the co-registered FBB scan or applied directly to the spatially normalized FBB scan. Minor manual adjustments were made to ensure that overlap with white matter and cerebrospinal fluid was minimized. Mean radioactivity values were obtained from the volume of interest for the cortical, subcortical and cerebellar regions. The cerebellar cortical volume of interest was placed taking care to avoid cerebellar white matter. All volume of interest placement was performed by a single experienced operator (VLV) blind to the clinical status of the individuals. No correction for partial volume effects was applied to the PET data.
The standardized uptake value, defined as the decay-corrected brain radioactivity concentration normalized for injected dose and body weight, was calculated for all regions. These values were then used to derive the standardized uptake value ratio (SUVR), which was referenced to the cerebellar cortex. Neocortical Aβ deposition was expressed as the average SUVR of the mean for the following cortical regions of interest: frontal (consisting of dorsolateral prefrontal, ventrolateral prefrontal, and orbitofrontal regions), superior parietal, lateral temporal, lateral occipital, and anterior and posterior cingulate.
To identify a SUVR cutoff point, a hierarchical cluster analysis of the neocortical SUVR of FBB scans in healthy control participants was performed similar to that previously described [
10]. The cutoff value for high neocortical SUVR in this study was defined as ≥ 1.45.
Statistical analysis
Independent-sample t-tests were used to compare means of MCI subtypes with healthy controls and AD patients, and to compare means within the MCI subtypes. Categorical differences were assessed using Fisher's exact test. Pearson's or Spearman's rank correlation analyses were conducted to assess the degree of linear relationship between neuroimaging variables (SUVR, HV, WMH) with composite EM and nonmemory z scores, adjusting for age, gender and years of education. Data are presented as mean ± standard deviation unless otherwise stated. Adjustment for multiple testing was not performed.
Role of the funding source
The funding sources had no role in the data analyses and interpretation. The corresponding author had full access to all data presented in this study and had final responsibility for the decision to submit for publication.
Competing interests
VLV and CCR were consultants for Bayer Schering Pharma. RSM received research support from Bayer Schering Pharma. CBR, BP and KR are Bayer Schering Pharma employees. The remaining authors declare that they have no competing interests.
Authors' contributions
KO contributed to participant recruitment, patient screening, data collection, data analysis, and wrote the paper. VLV and CCR contributed to data analysis, and provided comments and critical revision to the paper. AB-F and FL performed neuropsychology assessments and provided comments to the paper. GC provided comments to the paper. PR performed white matter hyperintensity analyses, the method of which was standardized by OS. CBR and BP developed the protocol for clinical trial designs involving FBB. KR performed the statistical analyses. RSM produced FBB in-house. VLV, CLM and CCR are the principal investigators of this study who contributed to trial design. All authors read and approved the manuscript for publication