We investigated the extent to which decline in memory and working memory in beta-amyloid (Aβ) positive non-demented individuals was related to hippocampal atrophy and Aβ accumulation over 36 months. Cognitively normal older adults (CN) (n = 178) and adults with mild cognitive impairment (MCI) (n = 49) underwent positron emission tomography neuroimaging, magnetic resonance imaging, and cognitive assessments at baseline, 18- and 36-months. Relative to Aβ− CNs, Aβ+ CNs and Aβ+ MCIs showed greater rates of cognitive decline, Aβ accumulation, and hippocampal atrophy. Analysis of interrelationships between these Alzheimer's disease markers in Aβ+ CNs and MCIs indicated that rate of Aβ accumulation was associated with rate of hippocampal atrophy (β = −0.05, p = .037), which was in turn associated independently with rate of decline in memory (β = −0.03, p = .032). This suggests that Aβ accumulation precedes any neurodegeneration or clinical symptoms, and that the relationship between Aβ and cognitive decline is mediated by hippocampal atrophy.

Introduction

Prospective studies indicate that in non-demented older adults, abnormally high levels of beta-amyloid (Aβ) at baseline, detected using positron emission tomography (PET), are associated with an increased rate of further accumulation of Aβ, greater loss of brain volume, and a more rapid decline in cognitive function compared with appropriately matched individuals with low Aβ (Bateman et al., 2012; Chételat et al., 2012; Dore et al., 2013; Jack et al., 2013,, 2014; Villemagne et al., 2013). Data from studies that have quantified the temporal sequence of changes in Aβ accumulation, hippocampal neurodegeneration, and memory loss in humans show that there is an initial but slow accumulation of Aβ, which, upon reaching abnormal levels, increases in rate and leads to an acceleration of neurodegeneration. This in turn leads to a decline in cognitive function, initially in episodic and working memory and ultimately to clinical progression (Bateman et al., 2012; Jack & Holtzman, 2013; Villemagne et al., 2013).

Empirical data from studies of healthy, cognitively normal adults (CN), and adults with mild cognitive impairment (MCI) who are Aβ+ suggest that although the rate of Aβ accumulation, hippocampal neurodegeneration, and cognitive decline may be different, the temporal progression of each can be characterized by a linear trend, which is often associated with increased risk for AD (Bateman et al., 2012; Villemagne et al., 2013). Once Alzheimer's disease (AD) is classified clinically, the rate of change in each marker slows (Jack & Holtzman, 2013; Villemagne et al., 2013). Studies exploring the effect of Aβ+ on different cognitive domains in CNs have shown that Aβ-related cognitive decline manifests first and most reliably on measures of episodic memory as well as in aspects of working memory and executive function (Doraiswamy et al., 2012; Johnson et al., 2013; Lim, Ellis, Pietrzak, et al., 2012; Lim, Maruff, Pietrzak, Ames, et al., 2014). Further, impairment on neuropsychological measures of episodic memory, relative to matched controls (Mormino et al., 2008; Oh, Madison, Villeneuve, Markley, & Jagust, 2014) or objectively defined decline on neuropsychological measures of episodic memory within the same individuals (Dore et al., 2014; Villemagne et al., 2013; Wirth et al., 2013) have been found to be associated strongly with indices of hippocampal neurodegeneration, although these relationships are stronger when analyses are restricted to Aβ+ individuals (Oh et al., 2014; Villemagne et al., 2013; Wirth et al., 2013).

Performance on the episodic and working memory tasks from the Cogstate Brief Battery (CBB), as well as a composite measure of both, has been shown to be sensitive to cognitive decline in both Aβ+ CN older adults and MCI patients. In contrast, performance on the psychomotor and attention tasks from the CBB, and on the composite measure of both, remains stable over time in these groups (Lim, Ellis, Pietrzak, et al., 2012; Lim, Maruff, Pietrzak, Ellis, et al., 2014). Therefore, decline in episodic and working memory as measured using the CBB provides a useful measure against which rates of Aβ accumulation and hippocampal volume (HV) can be compared. Prospective studies of Aβ+ non-demented older adults therefore provide an appropriate context for comparing the rate of decline in episodic memory, working memory, and other cognitive domains with the rates of neurodegeneration and Aβ accumulation. Relationships between cognitive decline, neurodegeneration, and Aβ accumulation can then be modeled using joint latent growth curve modeling for parallel processes (Preacher, Wichman, MacCallum, & Briggs, 2008). These models can elucidate the nature of interrelationships among these variables in a single model instead of conducting series of simple bivariate correlations of change scores, as has been done in most studies to date (Dore et al., 2014; Mormino et al., 2008; Villemagne et al., 2013).

The first aim of this study was to evaluate the relationship between decline in cognition and rates of change in HV and Aβ accumulation over a 36-month period in CNs. The second aim was to examine interrelationships of Aβ, HV, and cognition across the preclinical and prodromal phases of AD. The first hypothesis was that, compared with Aβ− individuals, Aβ+ non-demented individuals would show decline in measures of episodic and working memory, greater loss of HV and faster accumulation of Aβ. The second hypothesis was that Aβ accumulation would be associated with HV loss, which would in turn be associated with greater decline in cognition.

Materials and Methods

Participants

Participants were recruited from the CN and MCI groups enrolled in the Australian Imaging, Biomarkers and Lifestyle (AIBL) Study. The process of recruitment and diagnostic classification of CNs and adults with MCI enrolled in AIBL has been described in detail elsewhere (Ellis et al., 2009; Rowe et al., 2010). Participants were excluded from the AIBL study if they had any of the following: schizophrenia, depression (Geriatric Depression Score [GDS] ≥6), Parkinson's disease, cancer (except basal cell skin carcinoma) within the last 2 years, symptomatic stroke, uncontrolled diabetes, or current regular alcohol use exceeding two standard drinks per day for women or four per day for men. A clinical review panel chaired by DA reviewed all available medical, psychiatric, and neuropsychological information to confirm the cognitive health of individuals enrolled in the CN group. Similarly, all data for participants with MCI were reviewed to ensure that their clinical classification was consistent with international criteria. Clinical classification was blinded to Aβ imaging, and magnetic resonance imaging (MRI) data. Participants were classified according to their Aβ levels and clinical classification at entry into the AIBL study. The demographic characteristics of the sample at baseline are shown in Table 1. All participants who provided cognitive data at the baseline assessment also provided data for the 18- and 36-month assessments. For the neuroimaging measures, data were included in the prospective analyses only if it was available for the three assessments or for the baseline and 36-month assessment. Hence, sample sizes for prospective analysis of neuroimaging measures were slightly smaller than for the cognitive measures. The sample sizes used in the prospective analyses of standardized uptake value ratio (SUVR) and HV are shown in Table 3. The study was approved by and complied with the regulations of the institutional research and ethics committees of Austin Health, St. Vincent's Health, Hollywood Private Hospital, and Edith Cowan University. All participants provided written informed consent prior to participating in the study.

Table 1.

Demographic means (SD) for MMSE, CDR-SB, premorbid IQ, and HADS scores for each group by Aβ status at baseline

 Cognitively normal older adults
 
MCI
 
Aβ− (n = 123) Aβ+ (n = 55) Aβ− (n = 13) Aβ+ (n = 36) 
N (%) female 61 (49.6%) 28 (50.9%) 7 (50.0%) 18 (50.0%) 
N (%) APOE ε4 37 (30.1%) 36 (65.5%) 1 (7.1%) 27 (75.0%) 
Age (years) 69.92 (6.99) 75.20 (7.19) 79.79 (9.18) 80.78 (6.17) 
SUVR neocortex 1.16 (0.09) 1.95 (0.26) 1.17 (0.14) 2.21 (0.41) 
MMSE 28.81 (1.23) 28.58 (1.23) 27.43 (2.47) 26.94 (2.16) 
CDR Sum of Boxes 0.30 (0.52) 0.55 (0.87) 0.96 (0.60) 1.01 (0.73) 
Premorbid IQ 107.93 (7.82) 110.18 (5.96) 103.07 (12.15) 109.28 (7.00) 
HADS Depression 2.79 (2.17) 2.82 (2.60) 3.62 (2.06) 3.50 (2.46) 
HADS Anxiety 4.06 (2.64) 4.33 (3.26) 5.23 (2.49) 4.58 (2.51) 
Cogstate Psychomotor/Attention Composite 99.83 (10.93) 99.74 (8.35) 91.96 (19.37) 95.69 (10.65) 
Cogstate Learning/Working Memory Composite 98.91 (10.84) 95.80 (9.59) 95.45 (10.56) 87.99 (11.65) 
Detection 2.52 (0.13) 2.41 (0.09) 2.58 (0.20) 2.55 (0.10) 
Identification 2.71 (0.07) 2.72 (0.07) 2.78 (0.12) 2.75 (0.08) 
One Card Learning 0.99 (0.11) 0.98 (0.11) 0.99 (0.08) 0.91 (0.11) 
One Back 1.32 (0.20) 1.24 (0.18) 1.23 (0.22) 1.14 (0.22) 
 Cognitively normal older adults
 
MCI
 
Aβ− (n = 123) Aβ+ (n = 55) Aβ− (n = 13) Aβ+ (n = 36) 
N (%) female 61 (49.6%) 28 (50.9%) 7 (50.0%) 18 (50.0%) 
N (%) APOE ε4 37 (30.1%) 36 (65.5%) 1 (7.1%) 27 (75.0%) 
Age (years) 69.92 (6.99) 75.20 (7.19) 79.79 (9.18) 80.78 (6.17) 
SUVR neocortex 1.16 (0.09) 1.95 (0.26) 1.17 (0.14) 2.21 (0.41) 
MMSE 28.81 (1.23) 28.58 (1.23) 27.43 (2.47) 26.94 (2.16) 
CDR Sum of Boxes 0.30 (0.52) 0.55 (0.87) 0.96 (0.60) 1.01 (0.73) 
Premorbid IQ 107.93 (7.82) 110.18 (5.96) 103.07 (12.15) 109.28 (7.00) 
HADS Depression 2.79 (2.17) 2.82 (2.60) 3.62 (2.06) 3.50 (2.46) 
HADS Anxiety 4.06 (2.64) 4.33 (3.26) 5.23 (2.49) 4.58 (2.51) 
Cogstate Psychomotor/Attention Composite 99.83 (10.93) 99.74 (8.35) 91.96 (19.37) 95.69 (10.65) 
Cogstate Learning/Working Memory Composite 98.91 (10.84) 95.80 (9.59) 95.45 (10.56) 87.99 (11.65) 
Detection 2.52 (0.13) 2.41 (0.09) 2.58 (0.20) 2.55 (0.10) 
Identification 2.71 (0.07) 2.72 (0.07) 2.78 (0.12) 2.75 (0.08) 
One Card Learning 0.99 (0.11) 0.98 (0.11) 0.99 (0.08) 0.91 (0.11) 
One Back 1.32 (0.20) 1.24 (0.18) 1.23 (0.22) 1.14 (0.22) 

Notes: One-way ANOVA indicated significant differences in age and premorbid IQ between groups, all p's < .05; but no significant difference between HADS Depression and HADS Anxiety, all p's > .05. χ2 indicated a significant difference in number of APOE ɛ4 carriers between groups.

SUVR = standardized uptake value ratio; MMSE = mini-mental state examination; CDR = clinical dementia rating scale; HADS = hospital anxiety and depression scale.

Measures

Neuroimaging and genotyping

Pittsburgh Compound B (PiB)-PET neuroimaging methodology has been described in detail elsewhere (Rowe et al., 2010). Each subject received ∼370 MBq 11C-PiB intravenously over 1 min. A 30-min acquisition of PET standardized uptake value (SUV) data in 3D mode, consisting of 6 frames each of 5 min, acquired 40–70 min post-PiB injection were summed and normalized to the cerebellar cortex SUV, resulting in a region-to-cerebellar ratio termed SUVR. Following convention, participants with SUVR <1.5 were classified as Aβ−, and those with SUVR ≥1.5 were classified as Aβ+ from the baseline assessment (Rowe et al., 2010).

Participants underwent a clinical MRI for screening and subsequent co-registration with the PET images. A fluid attenuated inversion recovery sequence was obtained for exclusion of subjects with cortical stroke. MRI imaging analysis has been described in detail elsewhere (Dore et al., 2013). MR images were spatially normalized to the Montreal Neurological Institute (MNI) single-subject MRI brain template (Collins et al., 1998) using MilxView®, software developed by the Australian e-Health Research Centre—BioMedIA (Brisbane, Australia). As described elsewhere, T1W MR images for each subject were classified into gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) using an implementation of the expectation maximization segmentation algorithm (Ourselin, Roche, Subsol, Pennec, & Ayache, 2001). The algorithm computed probability maps for each tissue type and was used to assign each voxel to its most likely tissue type and subsequent segmentation. To improve the accuracy of analysis of the hippocampus, a separate, manually delineated template was drawn on the MNI single-subject every 1 mm on coronal slices, and was subsequently used for HV. The average HVs were normalized for head size using the total intracranial volume, defined as the sum of GM, WM, and CSF volumes. An 80 ml blood sample was taken from each participant, 0.5 ml of which was forwarded for apolipoprotein E (APOE) genotyping.

Clinical assessments

The clinical status of all participants on all three visits was assessed using the standard clinical rating scales and neuropsychological battery from the AIBL study (Ellis et al., 2009). All of these measures have been described in detail elsewhere and were administered according to standard protocols by trained research assistants. Performance on the CBB tasks was not used in the classification of individuals' clinical status. On each visit, the clinical status of participants was determined using information which included the Mini-Mental State Examination (MMSE) and Clinical Dementia Rating (CDR) scale. Levels of depressive and anxiety symptoms were assessed using the Hospital Anxiety and Depression Scale (HADS). Premorbid intelligence was estimated using the Wechsler Test of Adult Reading.

The Cogstate Brief Battery

The CBB has been described in detail elsewhere. Briefly, the CBB includes measures of psychomotor function (Detection; DET), attention (Identification; IDN), visual episodic memory (One Card Learning; OCL), and visual working memory (One Back; OBK) (Lim, Ellis, Harrington, et al., 2012; Lim, Ellis, Pietrzak, et al., 2012). The main outcome measures for the DET and IDN tasks is reaction time in milliseconds (speed), which was normalized using a logarithmic base 10 (log10) transformation. The main outcome measures for the OCL and OBK tasks is the proportion of correct responses (accuracy), which was normalized using an arcsine square-root transformation. The normative database for the current study consisted of 650 healthy older adults aged >50 who were enrolled in the AIBL study. The process for classification of normal cognition in the AIBL healthy group has been described (see above and Ellis et al., 2009). These individuals had been classified as healthy on the basis of neuropsychological and medical workup and case analysis by a clinical team lead by a gerontologist (DA). For this group, Aβ neuroimaging data had not been taken into account. The group means for the individual outcome measures from the CBB and the two composites were organized according to 10-year age cohorts (i.e., 51–60, 61–70, 71–80, 81–90, and 90+) with the minimum number of patients for any cohort being 74 (Maruff et al., 2013).

For each participant, each performance measure from the four tasks in the CBB was computed as reported previously (Lim, Ellis, Harrington, et al., 2012; Maruff et al., 2013). For each performance measure, the mean and standard deviation (SD) was estimated for the CN group according to their age. The age-adjusted mean and SD of the HC group was then used to standardize scores on each of the four cognitive tasks for each participant. A Learning/Working Memory composite score was computed by averaging the standardized scores for the OCL and OBK tasks, and an Attention/Psychomotor function composite score was computed by averaging the standardized scores for the DET and IDN tasks. For each individual, both composite scores were then re-standardized using the mean and SD for each composite score computed from the HC group and then transformed once more so that each had a mean of 100 and an SD of 10. This was achieved by first multiplying each standardized score by 10 and then adding 100.

Data Analysis

To examine the first hypothesis, a series of repeated measures linear mixed model (LMM) analyses using maximum likelihood estimation and an unstructured covariance matrix were conducted to examine the relation between Aβ group (CN Aβ−, CN Aβ+, MCI Aβ−, MCI Aβ+) and time (baseline, 18 months, and 36 months) on the rate of change in performance on the CBB tasks, Aβ, and HV. In these analyses, Aβ group, time, and the Aβ group × time interaction were entered as fixed factors; participant as a random factor; age, premorbid IQ, and APOE as covariates; and the CBB performance score, SUVR, or HV entered as dependent variables in separate analyses. As age-corrected normative data were used to compute the CBB composites (i.e., Psychomotor/Attention and Learning/Working Memory), age was not entered as a covariate in these analyses. The magnitude of difference in the rates of change (i.e., slopes) of the CN Aβ+, MCI Aβ−, and MCI Aβ+ groups in relation to the CN Aβ− group was expressed using Cohen's d (Cohen, 1992). To allow appreciation of the magnitude of changes observed, the group means from the raw data (i.e., not modeled data) are provided in Table 4 for the main outcome measures.

As the LMM showed that the Learning/Working Memory composite was most sensitive to Aβ-related decline, this composite measure was used in the growth curve analysis. To determine whether rates of episodic memory decline in individuals over the preclinical and prodromal stages of AD would be associated more strongly with rates of reduction in HV than with rates of Aβ accumulation, joint latent growth curve models were conducted using the data from the combined group of Aβ+ individuals. In these models, associations among slopes of Aβ, HV, and memory decline were evaluated using a latent growth curve model for parallel processes in Mplus version 7.11. In these models, we specified direct paths between Aβ, HV, and Learning/Working Memory slopes; and covariance terms among Aβ, HV, and Learning/Working Memory slopes. Model fit was evaluated using the following fit statistics: S-Bχ2, comparative fit index (CFI), Tucker Lewis Index (TLI), root mean square error of approximation (RMSEA), and standardized root mean square residual (SRMR). By convention, higher CFI and TLI values, and lower χ2, RMSEA, and SRMR values, indicate better fit. Empirically defined benchmarks suggest that CFI and TLI ≥0.90 is indicative of adequate fit, and ≥0.95 indicative of excellent fit. RMSEA and SRMR ≤0.08 indicative of adequate fit and ≤0.05 indicative of excellent fit (Hu & Bentler, 1998, 1999).

Results

Demographic Differences Between Aβ− and Aβ+ Subgroups in CN and MCI Groups

Demographic and clinical characteristics of the CN and MCI groups by baseline Aβ classification are shown in Table 1. Statistically significant differences between Aβ− and Aβ+ subgroups were observed only for age and premorbid IQ and there were significantly more APOE ɛ4 carriers in both the CN and MCI Aβ+ groups. Groups did not differ with respect to any other demographic characteristics.

Rates of Change in Cognition, Reduction in HV, and Aβ Accumulation

Group × time interactions were statistically significant for the Learning/Working Memory composite, the OCL task, reductions in HV and Aβ accumulation (Table 2). The mean slope for each clinical group for each cognitive outcome measure is given in Table 3. The magnitude of the difference in slopes for each group from that of the CN Aβ− group is presented for each CBB outcome measure, SUVR, and HV in Fig. 1.

Table 2.

Results of linear mixed model analyses examining change in cognition, hippocampal volume, and Aβ accumulation in healthy older adults and MCI over 36 months

 Aβ group
 
Time
 
Aβ group × time
 
(df) F p (df) F p (df) F p 
Neuropsychology 
 Psychomotor/Attention (3,200) 2.55 .06 (1,192) 2.44 .12 (3,191) 1.59 .19 
 Detection (3,218) 0.63 .60 (1,198) 2.70 .10 (3,198) 0.89 .45 
 Identification (3,200) 2.83 .04 (1,182) 1.78 .18 (3,182) 2.36 .07 
 Learning/Working Memory (3,219) 2.76 .04 (1,206) 0.03 .86 (3,206) 3.87 .01 
 One Card Learning (3,223) 1.81 .15 (1,204) 1.82 .18 (3,206) 4.32 .00 
 One Back (3,220) 2.00 .12 (1,196) 1.66 .20 (3,198) 2.44 .06 
Neuroimaging 
 Aβ accumulation (3,155) 149.47 .00 (1,155) 81.10 .00 (3,155) 19.86 .00 
 Hippocampal volume (3,141) 1.03 .38 (1,141) 92.12 .00 (3,141) 18.00 .00 
 Aβ group
 
Time
 
Aβ group × time
 
(df) F p (df) F p (df) F p 
Neuropsychology 
 Psychomotor/Attention (3,200) 2.55 .06 (1,192) 2.44 .12 (3,191) 1.59 .19 
 Detection (3,218) 0.63 .60 (1,198) 2.70 .10 (3,198) 0.89 .45 
 Identification (3,200) 2.83 .04 (1,182) 1.78 .18 (3,182) 2.36 .07 
 Learning/Working Memory (3,219) 2.76 .04 (1,206) 0.03 .86 (3,206) 3.87 .01 
 One Card Learning (3,223) 1.81 .15 (1,204) 1.82 .18 (3,206) 4.32 .00 
 One Back (3,220) 2.00 .12 (1,196) 1.66 .20 (3,198) 2.44 .06 
Neuroimaging 
 Aβ accumulation (3,155) 149.47 .00 (1,155) 81.10 .00 (3,155) 19.86 .00 
 Hippocampal volume (3,141) 1.03 .38 (1,141) 92.12 .00 (3,141) 18.00 .00 

Notes: Aβ group = group membership as CN Aβ−, CN Aβ+, MCI Aβ−, or MCI Aβ+; bolded values are significant at the p < .05 or < .001 level; age, FSIQ, and APOE have been entered as covariates in all analyses, except for Psychomotor/Attention and Learning/Working Memory, where FSIQ and APOE were the only covariates.

Table 3.

Mean slopes (SD) of Aβ− and Aβ+ cognitively normal older adults (CN) and adults with MCI

 CN Aβ−
 
CN Aβ+
 
MCI Aβ−
 
MCI Aβ+
 
Mean slope (SDn Mean slope (SDn Mean slope (SDn Mean slope (SDn 
Neuropsychology 
 Psychomotor/Attention −1.321 (6.251) 123 −1.367 (6.056) 55 1.806 (4.433) 13 −2.858 (5.524) 36 
 Detection 0.018 (0.076) 123 0.006 (0.074) 55 −0.005 (0.054) 13 0.029 (0.067) 36 
 Identification 0.006 (0.041) 123 0.014 (0.040) 55 −0.017 (0.029) 13 0.017 (0.038) 36 
 Learning/Working Memory 1.456 (6.882) 123 −1.963 (6.656) 55 2.424 (4.869) 13 −1.435 (6.112) 36 
 One Card Learning 0.007 (0.072) 123 −0.035 (0.071) 55 0.012 (0.051) 13 −0.022 (0.065) 36 
 One Back 0.030 (0.132) 123 −0.018 (0.129) 55 0.065 (0.093) 13 −0.012 (0.117) 36 
Neuroimaging 
 Aβ accumulation 0.015 (0.041) 98 0.061 (0.036) 33 0.018 (0.027) 0.079 (0.032) 20 
 Hippocampal volume −0.032 (0.047) 86 −0.054 (0.040) 28 −0.018 (0.031) −0.119 (0.036) 17 
 CN Aβ−
 
CN Aβ+
 
MCI Aβ−
 
MCI Aβ+
 
Mean slope (SDn Mean slope (SDn Mean slope (SDn Mean slope (SDn 
Neuropsychology 
 Psychomotor/Attention −1.321 (6.251) 123 −1.367 (6.056) 55 1.806 (4.433) 13 −2.858 (5.524) 36 
 Detection 0.018 (0.076) 123 0.006 (0.074) 55 −0.005 (0.054) 13 0.029 (0.067) 36 
 Identification 0.006 (0.041) 123 0.014 (0.040) 55 −0.017 (0.029) 13 0.017 (0.038) 36 
 Learning/Working Memory 1.456 (6.882) 123 −1.963 (6.656) 55 2.424 (4.869) 13 −1.435 (6.112) 36 
 One Card Learning 0.007 (0.072) 123 −0.035 (0.071) 55 0.012 (0.051) 13 −0.022 (0.065) 36 
 One Back 0.030 (0.132) 123 −0.018 (0.129) 55 0.065 (0.093) 13 −0.012 (0.117) 36 
Neuroimaging 
 Aβ accumulation 0.015 (0.041) 98 0.061 (0.036) 33 0.018 (0.027) 0.079 (0.032) 20 
 Hippocampal volume −0.032 (0.047) 86 −0.054 (0.040) 28 −0.018 (0.031) −0.119 (0.036) 17 
Fig. 1.

Magnitude of difference (Cohen's d) in the rate of change in each cognitive outcome measure, SUVR, and hippocampal volume between CN Aβ+, MCI Aβ−, and MCI Aβ+ groups relative to the CN Aβ− group (represented by the “0” line). Error bars represent 95% confidence intervals.

Fig. 1.

Magnitude of difference (Cohen's d) in the rate of change in each cognitive outcome measure, SUVR, and hippocampal volume between CN Aβ+, MCI Aβ−, and MCI Aβ+ groups relative to the CN Aβ− group (represented by the “0” line). Error bars represent 95% confidence intervals.

Post hoc comparison of group mean slopes over 36 months indicated that, relative to the CN Aβ− group, the CN Aβ+ group showed a significantly greater rate of decline over 36 months on the Learning/Working Memory composite, the OCL task, and the OBK task, with the magnitude of these differences, by convention, and moderate (Cohen, 1992). No differences in group mean slopes were observed between the CN Aβ− and CN Aβ+ groups on the Psychomotor/Attention composite, the DET task or the IDN task (Table 2). Relative to the CN Aβ− group, the MCI Aβ+ group also showed a greater rate of decline over 36 months on the Learning/Working Memory composite, the OCL task, and the OBK task, with the magnitude of these differences, by convention, and moderate (Fig. 1). Furthermore, effect sizes reflecting differences in slope from controls for each CBB outcome measure were equivalent in the CN Aβ+ group and the MCI Aβ+ group, as there was overlap between the 95% confidence intervals for these effect sizes (Fig. 1, Table 3). Of note, when compared with the CN Aβ− group, the MCI Aβ− group showed no statistically significant decline or improvement on any cognitive composite measure.

Relative to the CN Aβ− group, the MCI Aβ+ group showed statistically significant reductions in HV and faster rate of Aβ accumulation with this difference being large in magnitude (Fig. 1). The CN Aβ+ group also showed a faster rate of Aβ accumulation and HV loss relative to the CN Aβ− group that was statistically significant. As was observed for the cognitive measures, the MCI Aβ− group showed no significant HV loss, or significant increases in Aβ relative to the CN Aβ− group (Fig. 1).

The increased rate of cognitive decline associated with Aβ positivity in the CN and MCI groups was also reflected in the rates of clinical progression to MCI/AD in these groups. Specifically, in the CN Aβ− group, 8 of 115 (8 withdrew/deceased) progressed to MCI/AD at 36 months (7%); in the CN Aβ+ group, 6 of 38 (17 withdrew/deceased) progressed to MCI/AD at 36 months (16%); in the MCI Aβ− group, 2 of 10 (3 withdrew/deceased) progressed to AD at 36 months (20%); and in the MCI Aβ+ group, 19 of 29 (7 withdrew/deceased) progressed to AD at 36 months (66%).

Association Between Rates of Change on Learning/Working Memory Composite, HV and Aβ Accumulation in Combined Aβ+ CN and MCI Groups

A joint latent growth curve analysis evaluating interrelationships among Aβ, HV, and the Learning/Working Memory composite in the combined CN and MCI Aβ+ groups fit the data well, χ(45)2=45.95, p = 0.43, RMSEA = 0.014, 95% CI = 0.000–0.064; CFI = 1.00, TLI = 1.00, and SRMR = 0.028. Examination of associations among slopes of memory scores, HV, and Aβ accumulation revealed that after accounting for all other interrelationships among intercepts and slopes, greater Aβ at baseline was associated with greater decline in HV (β =− 0.05, p = .037), which was in turn associated with greater decline in the Learning/Working Memory composite measure (β =− 0.03, p = .032). Aβ slope was not significantly related to HV slope (β =− 0.001, p = .47).

Discussion

The first hypothesis that when compared with Aβ− individuals, Aβ+ CN older individuals and individuals with MCI, would show decline on the composite measures of episodic and working memory from the CBB, greater loss of HV and faster accumulation of Aβ was partially supported. Both the Aβ+ CN and MCI groups showed reliable decline on the individual episodic and working memory measures from the CBB and therefore also on the Learning/Working Memory composite score (Table 3). Neither the Aβ+ CN nor MCI groups showed any deterioration over 36 months on the measures of psychomotor function or attention from the CBB. This suggests that the decline in learning and working memory observed in both Aβ+ groups was not an indirect consequence of decline in attentional function. Rather the Aβ-related change in memory is consistent with findings from other studies using different cognitive measures which also report that early Aβ changes are greatest for aspects of memory (Doraiswamy et al., 2012; Johnson et al., 2013; Lim, Maruff, Pietrzak, Ames, et al., 2014).

In accord with the observed cognitive changes, compared with the CN Aβ− group, Aβ+ CN, and MCI groups also showed greater rates of hippocampal neurodegeneration (Table 2; Fig. 1). The rate of hippocampal neurodegeneration over the 36-month study period in the Aβ+ CN group was modest in magnitude, and large for the Aβ+ MCI group. This finding is consistent with observations of increased loss HV and other gray matter in Aβ+ CN and MCI groups from the AIBL and other cohorts (Chételat et al., 2012; Dore et al., 2013; Mormino et al., 2008; Villemagne et al., 2013; Wirth et al., 2013). Finally, in the CN and MCI Aβ+ groups, the rate of Aβ accumulation was also increased relative to the CN Aβ− group and although this increase in accumulation was statistically significant in both Aβ+ groups, it was modest in magnitude. Importantly though, for measures of both HV and Aβ accumulation, the magnitudes of change over 36 months were greater than that observed for the measures of cognition. The data presented here can also contribute to knowledge about expected change when considered in raw values. For example, the data in Table 4 show that in healthy older adults the percent increase in Aβ accumulation from baseline to the 18-month assessment was 0.03%. The same issue applies to loss of HV, where for the same group over the same time interval the loss was also very small (0.03%). Table 4 therefore allows researchers to compute these estimates for differences between other assessments.

Table 4.

Mean (SD) of raw scores for the main neuropsychology and neuroimaging measures for Aβ− and Aβ+ groups

Outcome Baseline 18 months 36 months 
Mean (SDMean (SDMean (SD
Healthy adults 
 Attention/psychomotor    
  Aβ− 99.83 (10.93) 99.26 (9.68) 97.66 (10.33) 
  Aβ+ 99.74 (8.35) 98.92 (8.26) 96.54 (9.82) 
Learning/Working Memory    
  Aβ− 98.91 (10.84) 100.37 (13.82) 101.88 (8.12) 
  Aβ+ 95.80 (9.59) 95.67 (10.70) 90.38 (11.89) 
Hippocampal volume    
  Aβ− 4.16 (0.29) 4.12 (0.30) 4.11 (0.30) 
  Aβ+ 4.10 (0.31) 3.99 (0.35) 3.99 (0.34) 
SUVR    
  Aβ− 1.16 (0.08) 1.16 (0.10) 1.20 (0.12) 
  Aβ+ 1.93 (0.27) 1.99 (0.26) 2.09 (0.30) 
MCI 
 Attention/Psychomotor    
  Aβ− 91.96 (19.37) 91.86 (16.06) 95.96 (9.50) 
  Aβ+ 95.69 (10.65) 94.58 (11.43) 92.94 (12.19) 
 Learning/Working Memory    
  Aβ− 95.45 (10.56) 93.50 (10.23) 101.18 (11.76) 
  Aβ+ 87.99 (11.65) 85.03 (11.76) 80.92 (12.63) 
 Hippocampal volume    
  Aβ− 3.90 (0.48) 4.04 (0.43) 3.96 (0.42) 
  Aβ+ 3.82 (0.45) 3.69 (0.48) 3.52 (0.59) 
 SUVR    
  Aβ− 1.16 (0.14) 1.18 (0.10) 1.17 (0.13) 
  Aβ+ 2.19 (0.43) 2.26 (0.45) 2.38 (0.44) 
Outcome Baseline 18 months 36 months 
Mean (SDMean (SDMean (SD
Healthy adults 
 Attention/psychomotor    
  Aβ− 99.83 (10.93) 99.26 (9.68) 97.66 (10.33) 
  Aβ+ 99.74 (8.35) 98.92 (8.26) 96.54 (9.82) 
Learning/Working Memory    
  Aβ− 98.91 (10.84) 100.37 (13.82) 101.88 (8.12) 
  Aβ+ 95.80 (9.59) 95.67 (10.70) 90.38 (11.89) 
Hippocampal volume    
  Aβ− 4.16 (0.29) 4.12 (0.30) 4.11 (0.30) 
  Aβ+ 4.10 (0.31) 3.99 (0.35) 3.99 (0.34) 
SUVR    
  Aβ− 1.16 (0.08) 1.16 (0.10) 1.20 (0.12) 
  Aβ+ 1.93 (0.27) 1.99 (0.26) 2.09 (0.30) 
MCI 
 Attention/Psychomotor    
  Aβ− 91.96 (19.37) 91.86 (16.06) 95.96 (9.50) 
  Aβ+ 95.69 (10.65) 94.58 (11.43) 92.94 (12.19) 
 Learning/Working Memory    
  Aβ− 95.45 (10.56) 93.50 (10.23) 101.18 (11.76) 
  Aβ+ 87.99 (11.65) 85.03 (11.76) 80.92 (12.63) 
 Hippocampal volume    
  Aβ− 3.90 (0.48) 4.04 (0.43) 3.96 (0.42) 
  Aβ+ 3.82 (0.45) 3.69 (0.48) 3.52 (0.59) 
 SUVR    
  Aβ− 1.16 (0.14) 1.18 (0.10) 1.17 (0.13) 
  Aβ+ 2.19 (0.43) 2.26 (0.45) 2.38 (0.44) 

The finding that Aβ accumulates faster in individuals whose baseline Aβ levels were already abnormal is consistent with models that show the rate of Aβ accumulation accelerates once Aβ levels have reached the threshold for abnormality (Bateman et al., 2012; Jack et al., 2014; Villemagne et al., 2013). In contrast to the Aβ+ groups, no increase in the rate of hippocampal neurodegeneration or Aβ accumulation was observed for the MCI Aβ− group (Fig. 1). Thus, despite meeting rigorous clinical criteria for MCI applied by expert clinicians in large metropolitan medical centers (Ellis et al., 2009), the Aβ− MCI group showed no greater decline in memory, loss of HV, or faster Aβ accumulation. This finding suggests that while the cognitive impairment in these older adults was both qualitatively and quantitatively consistent with the clinical criteria of MCI, it is highly unlikely to reflect AD pathology. Other studies have reported that a proportion of individuals classified clinically with MCI also do not have high Aβ levels (Doraiswamy et al., 2014; Jack et al., 2013). For the current MCI Aβ− group, rigorous study inclusion/exclusion criteria make it unlikely that the stable cognitive impairment identified reflected some other psychiatric (e.g., depression) or systemic illness (e.g., cardiovascular disease). However, we have previously observed that a proportion of Aβ− individuals with MCI do progress to other non-AD dementias (e.g., vascular dementia) (Rowe et al., 2014), although it remains to be determined as to whether Aβ levels in this group will ultimately become abnormal or whether they progress to meet clinical criteria for AD without ever showing abnormal Aβ.

The second hypothesis that rates of Aβ accumulation, loss of HV, and decline in memory, as measured by the CBB, would be associated was also supported. Linear growth curve modeling of relationships between rates of change on the composite measure of episodic and working memory, HV and Aβ in the Aβ+ CN and MCI groups confirmed the order-related relationship, derived from the Aβ cascade model (Masters & Selkoe, 2012), where the rate of memory decline was associated most strongly with the rate of HV loss, which itself was most strongly associated with the rate of Aβ accumulation. A latent growth curve analysis for parallel processes identified no direct relationship between rate of decline on the composite of episodic and working memory and rate of Aβ accumulation. The finding that decline on the episodic and working memory tasks from the CBB was related more directly to decreases in HV, which itself is related to Aβ accumulation, is consistent with the data from smaller cross-sectional studies and with data from prospective studies which also show that changes in memory are associated more strongly with indices of hippocampal neurodegeneration in Aβ+ individuals (Dore et al., 2013; Mormino et al., 2008; Ong et al., 2014; Villemagne et al., 2013). As the sample studied here was larger and measured for longer than those used in previous studies, we had sufficient statistical power to examine relationships between changes in the indices of AD using methods more sophisticated and informative than the simple bivariate correlations used to date. Thus, when relationships between decline in episodic and working memory, hippocampal neurodegeneration, and Aβ accumulation were studied simultaneously in a multivariate model, increasing hippocampal neurodegeneration, but not faster Aβ accumulation, explained ∼15% of the variance in cognitive decline.

When considered together, the results of the current study are consistent with those of previous studies, which have suggested that, in non-demented older individuals, Aβ+ is not a benign process and is associated with increased decline in episodic and working memory and HV (Chételat et al., 2012; Doraiswamy et al., 2014; Lim, Maruff, Pietrzak, Ames, et al., 2014; Villemagne et al., 2013). Importantly though, results of the current study suggest that the relationship between decline in episodic memory, working memory and accumulation in Aβ was mediated by hippocampal neurodegeneration. This indirect relationship between decline in episodic memory and Aβ has also been observed post-mortem, where neurofibrillary tangles have been found to mediate the relationship between Aβ plaques and the level of cognitive function in individuals with AD (Bennett, Schneider, Wilson, Bienias, & Arnold, 2004). Further, the results of this study confirm reports from our group and others (Dore et al., 2013; Jack et al., 2014; Mormino et al., 2008) that Aβ deposition is associated with hippocampal neurodegeneration, which is in turn associated with memory decline. By restricting the interval of investigation to the preclinical and prodromal stages of AD, where changes in AD markers are virtually linear (Bateman et al., 2012; Villemagne et al., 2013), it is also possible to exploit parametric statistical models, that when based on prospective data, require relatively small samples for adequate statistical power and which are robust to missing data. It is however, important to note that the relationships observed in the current study between Aβ, HV, and memory do not reflect disease processes that occur once individuals meet clinical criteria for AD or before Aβ levels reach the threshold for abnormality (e.g., PiB SUVR <1.5).

An important caveat when interpreting the results of this study is that the AIBL study is not an epidemiological but a convenience sample. The selection of MCI groups was biased towards the inclusion of individuals with amnestic MCI. Further, in the recruitment of CNs, participants in AIBL were highly educated, and few had existing or untreated medical, neurological, or psychiatric illnesses. As such, it would be important for these findings to be replicated in Aβ+ individuals in population-based samples, such as the Mayo Clinic Study of Aging (Roberts et al., 2008), where it is possible that Aβ-related decline in cognition and neurodegeneration may be greater than that observed here. A second caveat is that the use of the CBB, and the Learning/Working Memory composite score, means that the cognitive functions studied here were limited. This was because our primary aim was to understand the sequential relationship between decline in cognition, hippocampal atrophy, and accumulation of Aβ. The CBB itself, and the Learning/Working Memory composite score, was selected for the analyses of relationships with Aβ accumulation and neurodegeneration because the results of this study, as well as data from previous studies, demonstrated the sensitivity of this measure to Aβ-related cognitive decline in both preclinical and prodromal AD. Therefore, relationships between cognitive decline, neurodegeneration, and Aβ accumulation observed in the current study must now be extended to consider other aspects of episodic memory as well as other cognitive domains known to be abnormal in early AD (e.g., executive function and language).

Taken together, the findings of these studies converge to accord with theoretical models of Aβ and AD, which have proposed that accumulation in Aβ precedes neuronal degeneration or clinical symptoms. The direct comparison of these biological and cognitive markers of AD over the same fixed period of observation may provide more accurate estimates of the nature and magnitude of change in memory decline, HV, and Aβ in the earliest stages of AD, and may serve as a guide to determine the extent to which Aβ− modifying drugs can halt or delay the accumulation of Aβ, and in turn reductions in HV and decline in memory.

Funding

Funding for the study was provided in part by the study partners [Commonwealth Scientific Industrial and research Organization (CSIRO), Edith Cowan University (ECU), Mental Health Research institute (MHRI), National Ageing Research Institute (NARI), Austin Health, CogState Ltd.]. The study also received support from the National Health and Medical Research Council (NHMRC) and the Dementia Collaborative Research Centres program (DCRC2), as well as funding from the Science and Industry Endowment Fund (SIEF) and the Cooperative Research Centre for Mental Health (CRCMH).

Conflicts of Interest

Y.Y.L., P.B., K.A.E., K.H., and O.S. report no relevant disclosures. C.L.M. is an advisor to Prana Biotechnology Ltd and a consultant to Eli Lilly. R.H.P. and P.J.S. are scientific consultants to Cogstate Ltd. D.A. has served on scientific advisory boards for Novartis, Eli Lilly, Janssen, and Pfizer Inc. R.N.M. is a consultant to Alzhyme. C.C.R. has served on scientific advisory boards for Bayer Pharma, Elan Corporation, GE Healthcare, and AstraZeneca, has received speaker honoraria from Bayer Pharma and GE Healthcare, and has received research support from Bayer Pharma, GE Healthcare, Piramal Lifesciences, and Avid Radiopharmaceuticals. V.L.V. served as a consultant for Bayer Pharma and received research support from an NEDO grant from Japan. P.M. is a full-time employee of Cogstate Ltd.

Acknowledgements

Alzheimer's Australia (Victoria and Western Australia) assisted with promotion of the study and the screening of telephone calls from volunteers. The AIBL team wishes to thank the clinicians who referred patients with AD to the study: Associate Professor Brian Chambers, Professor Edmond Chiu, Dr Roger Clarnette, Associate Professor David Darby, Dr Mary Davison, Dr John Drago, Dr Peter Drysdale, Dr Jacqueline Gilbert, Dr Kwang Lim, Professor Nicola Lautenschlager, Dr Dina LoGiudice, Dr Peter McCardle, Dr Steve McFarlane, Dr Alastair Mander, Dr John Merory, Professor Daniel O'Connor, Dr Ron Scholes, Dr Mathew Samuel, Dr Darshan Trivedi, and Associate Professor Michael Woodward. We thank all those who participated in the study for their commitment and dedication to helping advance research into the early detection and causation of AD.

References

Bateman
R. J.
Xiong
C.
Benzinger
T. L. S.
Fagan
A. M.
Goate
A.
Fox
N. C.
et al.  
Clinical and biomarker changes in Dominantly Inherited Alzheimer's disease
The New England Journal of Medicine
 
2012
367
9
795
804
Bennett
D. A.
Schneider
J. A.
Wilson
R. S.
Bienias
J. L.
Arnold
S. E.
Neurofibrillary tangles mediate the association of amyloid load with clinical Alzheimer disease and level of cognitive function
Archives of Neurology
 
2004
61
3
378
384
Chételat
G.
Villemagne
V. L.
Villain
N.
Jones
G.
Ellis
K. A.
Ames
D.
et al.  
Accelerated cortical atrophy in cognitively normal elderly with high β-amyloid deposition
Neurology
 
2012
78
477
484
Cohen
J.
A power primer
Psychological Bulletin
 
1992
112
155
159
Collins
D. L.
Zijdenbos
A. P.
Kollokian
V.
Sled
J. G.
Kabani
N. J.
Holmes
C. J.
et al.  
Design and construction of a realistic digital brain phantom
IEEE Transactions on Medical Imaging
 
1998
17
3
463
468
Doraiswamy
P. M.
Sperling
R. A.
Coleman
R. E.
Johnson
K. A.
Reiman
E. M.
Davis
M. D.
et al.  
Amyloid-β assessed by florbetapir F 18 PET and 18-month cognitive decline
Neurology
 
2012
79
1636
1644
Doraiswamy
P. M.
Sperling
R. A.
Johnson
K.
Reiman
E. M.
Wong
T. Z.
Sabbagh
M. N.
et al.  
Florbetapir F 18 amyloid PET and 36-month cognitive decline: A prospective multicenter study
Molecular Psychiatry
 
2014
19
9
1044
1051
Dore
V.
Villemagne
V. L.
Bourgeat
P.
Fripp
J.
Acosta
O.
Chetelat
G.
et al.  
Cross-sectional and longitudinal analysis of the relationship between Aβ deposition, cortical thickness, and memory in cognitively unimpaired individuals and in Alzheimer's disease
JAMA Neurology
 
2013
70
7
903
911
Ellis
K. A.
Bush
A. I.
Darby
D.
De Fazio
D.
Foster
J.
Hudson
P.
et al.  
The Australian imaging, biomarkers and lifestyle (AIBL) study of aging: Methodology and baseline characteristics of 1112 individuals recruited for a longitudinal study of Alzheimer's disease
International Psychogeriatrics
 
2009
21
672
687
Hu
L.
Bentler
P. M.
Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives
Structural Equation Modeling
 
1999
6
1
55
Hu
L.
Bentler
P. M.
Fit indices in covariance structure modeling: Sensitivity to under-parameterized model misspecification
Psychological Methods
 
1998
3
424
453
Jack
C. R.
Holtzman
D. M.
Biomarker modeling of Alzheimer's disease
Neuron
 
2013
80
6
1347
1358
Jack
C. R.
Wiste
H. J.
Knopman
D. S.
Vemuri
P.
Mielke
M. M.
Weigand
S. D.
et al.  
Rates of B-amyloid accumulation are independent of hippocampal neurodegeneration
Neurology
 
2014
82
1605
1612
Jack
C. R.
Wiste
H. J.
Lesnick
T. G.
Weigand
S. D.
Knopman
D. S.
Vemuri
P.
et al.  
Brain β-amyloid load approaches a plateau
Neurology
 
2013
80
1
7
Johnson
K. A.
Sperling
R. A.
Gidicsin
C. M.
Carmasin
J. S.
Maye
J. E.
Coleman
R. E.
et al.  
Florbetapir (F18-AV-45) PET to assess amyloid burden in Alzheimer's disease dementia, mild cognitive impairment, and normal aging
Alzheimer's & Dementia
 
2013
12
S1552
S5260
Lim
Y. Y.
Ellis
K. A.
Harrington
K.
Ames
D.
Martins
R. N.
Masters
C. L.
et al.  
Use of the cogstate brief battery in the assessment of Alzheimer's disease related cognitive impairment in the Australian imaging, biomarker and lifestyle (AIBL) study
Journal of Clinical and Experimental Neuropsychology
 
2012
34
4
345
358
Lim
Y. Y.
Ellis
K. A.
Pietrzak
R. H.
Ames
D.
Darby
D.
Harrington
K.
et al.  
Stronger effect of amyloid load than APOE genotype on cognitive decline in healthy older adults
Neurology
 
2012
79
1645
1652
Lim
Y. Y.
Maruff
P.
Pietrzak
R. H.
Ames
D.
Ellis
K. A.
Harrington
K.
et al.  
Effect of amyloid on memory and non-memory decline from preclinical to clinical Alzheimer's disease
Brain
 
2014
137
221
231
Lim
Y. Y.
Maruff
P.
Pietrzak
R. H.
Ellis
K. A.
Darby
D.
Ames
D.
et al.  
Aβ amyloid and cognitive change: Examining the preclinical and prodromal stages of Alzheimer's disease
Alzheimer's & Dementia, S1552–S5260
 
2014
13
02939
02937
Maruff
P.
Lim
Y. Y.
Darby
D.
Ellis
K. A.
Pietrzak
R. H.
Snyder
P. J.
et al.  
Clinical utility of the cogstate brief battery in identifying cognitive impairment in mild cognitive impairment and Alzheimer's disease
BMC Pharmacology & Toxicology
 
2013
1
30
1
11
Masters
C. L.
Selkoe
D. J.
Selkoe
D. J.
Mandelkow
E.
Holtzman
D. M.
Biochemistry of amyloid β-protein and amyloid deposits in Alzheimer Disease
The biology of Alzheimer disease
 
2012
New York
Cold Spring Harbor Laboratory Press
181
204
Mormino
E. C.
Kluth
J. T.
Madison
C. M.
Rabinovici
G. D.
Baker
S. L.
Miller
B. L.
et al.  
Episodic memory loss is related to hippocampal-mediated b-amyloid deposition in elderly subjects
Brain
 
2008
132
1310
1323
Oh
H.
Madison
C. M.
Villeneuve
S.
Markley
C.
Jagust
W. J.
Association of gray matter atrophy with age, β-amyloid, and cognition in aging
Cerebral Cortex
 
2014
24
6
1609
1618
Ong
K. T.
Villemagne
V. L.
Bahar-Fuchs
A.
Lamb
F.
Langdon
N.
Catafau
A. M.
et al.  
Aβ imaging with 18F-florbetaben in prodromal Alzheimer's disease: A prospective outcome study
Journal of Neurology, Neurosurgery, & Psychiatry
 
2014
Jun 26, pii: jnnp-2014-308094
Ourselin
S.
Roche
A.
Subsol
G.
Pennec
X.
Ayache
N.
Reconstructing a 3D structure from serial histological sections
Image and Vision Computing
 
2001
19
(1–2)
25
31
Preacher
K. J.
Wichman
A. L.
MacCallum
R. C.
Briggs
N. E.
Latent growth curve modeling
 
2008
CA, USA: SAGE Publications
Roberts
R. O.
Geda
Y. E.
Knopman
D. S.
Cha
R. H.
Pankratz
V. S.
Boeve
B. F.
et al.  
The Mayo Clinic Study of Aging: Design and sampling, participation, baseline measures and sample characteristics
Neuroepidemiology
 
2008
30
1
58
69
Rowe
C. C.
Bourgeat
P.
Ellis
K. A.
Brown
B.
Lim
Y. Y.
Mulligan
R.
et al.  
Predicting Alzheimer disease with β-amyloid imaging: Results from the Australian imaging, biomarkers, and lifestyle study of ageing
Annals of Neurology
 
2014
74
6
905
913
Rowe
C. C.
Ellis
K. A.
Rimajova
M.
Bourgeat
P.
Pike
K. E.
Jones
G.
et al.  
Amyloid imaging results from the Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging
Neurobiology of Aging
 
2010
31
1275
1283
Villemagne
V. L.
Burnham
S.
Bourgeat
P.
Brown
B.
Ellis
K. A.
Salvado
O.
et al.  
Amyloid β deposition, neurodegeneration and cognitive decline in sporadic Alzheimer's disease: A prospective cohort study
Lancet Neurology
 
2013
12
357
367
Wirth
M.
Villeneuve
S.
Haase
C. M.
Madison
C. M.
Oh
H.
Landau
S. M.
et al.  
Associations between Alzheimer disease biomarkers, neurodegeneration, and cognition in cognitively normal older people
JAMA Neurology
 
2013
70
12
1512
1519