Utility of an Alzheimer’s Disease Risk-Weighted Polygenic Risk Score for Predicting Rates of Cognitive Decline in Preclinical Alzheimer’s Disease: A Prospective Longitudinal Study

Abstract

Background:

With the exception of APOE, genetic variants associated with increased Alzheimer’s disease (AD) risk are characterized by small effect sizes. Polygenic risk scores (PRS) have shown utility in predicting AD risk; however, their utility for predicting decline in cognition at preclinical stages of AD is poorly understood.

Objective:

To validate associations of a 22-variant AD-risk-weighted PRS with AD risk and related biomarkers and to assess its utility to predict cognitive decline.

Methods:

The PRS was evaluated with respect to brain amyloid-β (Aβ) burden, cerebrospinal fluid (CSF) Aβ₄₂, total-tau, and phospho-tau, and decline in cognition in 643 (570 cognitively normal (CN), 73 AD) PET-imaged participants from the longitudinal Australian Imaging, Biomarkers and Lifestyle (AIBL) Study of Ageing. Cognition was assessed using three composite measures; global cognition, verbal episodic memory, and a Pre-Alzheimer’s Cognitive Composite (PACC).

Results:

PRS, both with and without APOE, were positively correlated with brain Aβ burden, CSF total-tau, and phospho-tau in CN older adults. Further, in CN biomarker positive (Aβ ^high ) participants, significant associations were observed with baseline and longitudinal cognition. However, this association was not observed after the removal of APOE. Partitioning the PRS into quartiles revealed that the PRS associations with cognitive decline in Aβ ^high CN older adults is due to a saturating effect of APOE genotype.

Conclusions:

An AD-risk-weighted PRS is associated with cognitive decline in CN older adults. However, this association is absent when APOE genotype is excluded from the PRS, suggesting that associations with cognitive decline in this model of polygenic risk are driven by APOE genotype alone. Further research is needed to define appropriate PRSs with greater utility for predicting preclinical AD cognitive decline.

Keywords

Alzheimer’s disease amyloid-β cognitive decline episodic memory polygenic risk scores

INTRODUCTION

Alzheimer’s disease (AD) has a strong yet complex genetic component with estimates of heritability ranging from 60–80% [1]. The main genetic risk factor for AD, the ɛ4 allele of apolipoprotein E (APOE), has alone been shown to account for approximately one third of this genetic variance [2]. However, it is established that half of AD cases do not carry APOE ɛ4 (reviewed in [3]). As such, a substantial proportion of genetic hereditability is yet to be explained. Genome wide association studies (GWAS) have been applied with the aim of increasing understanding of genetic risk for AD [4 –8]. While GWAS have identified common variants in several genes associated with the clinical diagnosis of AD, no individual common variant that improves risk estimates beyond that associated with the APOE ɛ4 allele has been identified [4 –8]. While independently these AD risk variants have low effect sizes, with odds ratios between 1 and 1.5, they have been studied in combination to assess cumulative risk for AD, generally through the calculation of polygenic risk scores (PRSs) (reviewed in [9]).

Though PRSs have previously been calculated by the sum of the risk alleles alone [10, 11], most often risk alleles are weighted according to their influence on disease risk (reviewed in [9, 12]). This is particularly common in AD due to the importance of APOE ɛ4. For example, APOE ɛ4 has been observed to provide odds ratios of progression to disease of more than double that of the next most associated gene [4 –8]. A number of published PRS consisting of genes identified in case/control AD GWAS, weighted by their odds ratios [13 –28] beta coefficients [29 –31] or effect size [32 –34], have been associated with AD risk and endophenotypes. Studies have reported PRSs associated with clinical measures including clinical status [16 , 33], progression to AD [14, 22], rate of progression to AD [26], and age at onset [13, 35]. In addition, PRSs have been associated with pathological biomarkers including neocortical amyloid-β (Aβ) [14], cerebrospinal fluid (CSF) Aβ₄₂ [13 , 27], CSF tau [23], hippocampal volume [19 , 30], patterns of AD related brain atrophy [20], and cortical thickness [27, 29].

A number of studies have identified associations between PRSs and measures of cross-sectional and longitudinal cognitive performance in cognitively normal (CN) older adults and those with mild cognitive impairment or dementia. This included global memory [14 , 31], episodic memory [15], and executive function [14]. Further, there have been a number of studies that have investigated PRSs and cognition and observed no significant associations [10 , 28], or which lose any association after the exclusion of APOE from the score calculation [15, 17]. This discrepancy could be in part due to a lack of knowledge regarding healthy participants’ Aβ burden assessed with positron emission tomography (PET). It has previously been shown that higher levels of brain Aβ in a CN sample is associated with accelerated cognitive decline [36]. Further, an increased risk for cognitive decline is only observed in individuals with an above threshold brain Aβ burden [37], which is compounded by neurodegeneration. However, below threshold brain Aβ burden, even with neurodegeneration, confers no increased risk for cognitive decline [37]. As such, a measure of brain Aβ burden should be included in studies that evaluate models of AD risk as it will provide greater accuracy than when clinical diagnosis alone is used. This would include the assessment of the preclinical efficacy of a PRS where it could also assist in the clarification of previous results.

With a long transitional phase between normal aging and clinical AD, characterized by the accumulation of pathognomonic features up to 20 years prior to clinical diagnosis [38], there is considerable inter-individual variability in the rates of cognitive decline. The ability to identify which individuals will have the fastest rates of cognitive decline during pre-symptomatic disease stages would be of importance to preclinical disease trial design [39]. The study presented here aimed to validate, in a highly characterized longitudinal cohort, previous studies that have calculated PRSs, weighted by AD risk effect-sizes. Specifically, a 22 variant PRS was first evaluated against AD prediction and endophenotypes before its utility for predicting cognitive decline was assessed in CN individuals at risk for AD, based on brain Aβ imaging. Investigation into the feasibility of using PRSs, weighted by AD risk, is important in understanding the influence of genes, previously identified as risk genes for AD, in the early prediction of cognitive decline.

METHODS

Study participants

This study reports on data from 643 participants enrolled in the AIBL Study, a prospective longitudinal study of aging. Ellis et al. previously described the AIBL study design, enrollment process, neuropsychological assessments, and diagnostic criteria [40], including the process of classification into clinical groups. Briefly, participants were classified as CN (n = 570) by a clinical review panel, blinded to Aβ status, when the clinical criteria for diagnosis of mild cognitive impairment [41] or dementia [42] were not met. Likewise, the clinical diagnosis of AD (n = 73) was made when the criteria for dementia [42] was met. Ethics approval for the study was granted by committees from Austin Health, St Vincent’s Health, Hollywood Private Hospital, and Edith Cowan University, and informed written consent was given by all participants.

Cognitive measures

The AIBL neuropsychological test battery consists of the Mini-Mental State Examination (MMSE), Clock Drawing Test, California Verbal Learning Test-Second edition (CVLT-II), Logical Memory I and II (LMI; LMII; Story A only), D-KEFS verbal fluency, a 30-item version of the Boston Naming Test (BNT), Wechsler Test of Adult Reading (WTAR), Digit Span and Digit Symbol-Coding subtests of the Wechsler Adult Intelligence Scale-Third edition (WAIS-III), the Stroop task (Victoria version), and the Rey Complex Figure Test (RCFT) [40]. Previously calculated cognitive composite scores [43, 44], derived from this test battery and the Clinical Dementia Rating (CDR) scale, were used to assess cognitive performance. Three cognitive composites were investigated: verbal episodic memory (CDR sum of boxes (CDR_SB), LMII, CVLT-II recognition false positives (CVLT-II_FP), and long delay free recall (CVLT-II_LDFR)) [43], a statistically driven global cognition composite (CDR_SB, MMSE, LMII, CVLT_FP, and Clock) [43], and the Pre-Alzheimer’s Cognitive Composite (PACC (CVLT-II_LDFR, LMII, MMSE, Digit Symbol-Coding)) [44]. In the calculation of the verbal episodic memory and global cognition composites there were corrections for age, sex, years of education, premorbid IQ (WTAR-estimated WAIS-III Full Scale Intelligence Quotient (FSIQ)) and depressive symptoms (Geriatric Depression Scale (GDS)) [45]. Cognitive assessment data were utilized from 0, 18, 36, 54, 72, and 90-month time points.

Brain imaging

Positron emission tomography was utilized for Aβ imaging using ¹¹C-Pittsburgh Compound B (PiB), ¹⁸F-florbetapir, or ¹⁸F-flutemetamol radiolabeled tracers as previously described [46 –48]. CapAIBL ^®, a freely available web-based magnetic resonance (MR)-less methodology, was used to analyze images and generate PET standardized uptake value (SUV) ratios (SUVR) for all tracers [49]. SUVs were summed and normalized to tracer specific brain regions (PiB (cerebellar cortex), florbetapir (whole cerebellum), flutemetamol (pons)) to generate target-region to reference-region SUVR. For the combined analysis of multiple tracers as a continuous variable, linear regression transformation was applied to ¹⁸F-labelled tracers to generate PiB-like SUVR units termed the “Before the Centiloid Kernel Transformation” (BeCKeT) scale [50]. Further, participants were classified as high (Aβ ^high ) or low (Aβ ^low ) Aβ burden, based on a PiB/BeCKeT threshold of≥1.4 at any time point [51].

CSF biomarkers

Participants (n = 58) underwent lumbar puncture for CSF collection at the baseline time-point of the study. CSF was collected and Aβ₄₂, total-tau, and phospho-tau quantified as previously described [52]. Briefly, overnight fasting was followed by the collection of 10–14 ml of CSF by routine lumbar puncture directly into one 15 ml polypropylene tube (Greiner Bio-One188271) [53]. Aliquots of all CSF samples were prepared and stored as previously described [52]. Enzyme-linked immunosorbent assays (ELISA): INNOTEST Aβ (1–42; Aβ₄₂), INNOTEST hTAU Ag (total-tau), and INNOTEST Phospho-tau (phospho-tau181P; P-tau181P) (Fujirebio, Belgium), were used to analyze samples as per published standard methods. All samples were analyzed in duplicate and mean intra-assay coefficients of variation for these assays are as previously published [52].

Genotyping

DNA was extracted from 5 mL of whole blood by QIAamp DNA Blood Maxi Kits (Qiagen, Hilden, Germany) following manufacturer’s instructions. APOE genotyping was performed, as previously described [54], using TaqMan ^® genotyping assays (rs7412, assay ID: C____904973_10; rs429358, assay ID: C___3084793_20), (Life Technologies, Carlsbad, CA) on a QuantStudio 12K Flex™ Real-Time-PCR system (Applied Biosystems, Foster City, CA), using TaqMan ^® GTXpress™ Master Mix (Life Technologies) as per manufacturer instructions. In addition to APOE, 21 SNPs (Table 1) were selected for the construction of the PRS based on their previously identified association with AD risk, identified by GWAS [4 –8] and additional association studies [55]. Genotypes were extracted from a genome-wide SNP array conducted on the Illumina OmniExpressHumanExome +BeadChip with subsequent imputation using impute2 ver2.3, with the 1000 genome reference panel (2015 release).

Table 1

SNP information

						MAF
Gene	SNP	Chromosome	Position	Minor Allele	Risk Allele	CN older adults	AD	Published OR	AIBL OR (95% CI)
APOE	rs7412	19	44908822	ɛ4	ɛ4	0.16	0.43	2.53	6.79 (3.94–11.72)
	rs429358		44908684
CR1	rs3818361	1	207611623	A	A	0.16	0.33	1.21	2.85 (1.74–4.68)
BIN1	rs744373	2	127137039	G	G	0.31	0.33	1.17	1.39 (0.85–2.29)
INPP5D	rs35349669	2	233159830	T	T	0.48	0.49	1.07	0.92 (0.54–1.57)
MEF2C	rs190982	5	88927603	G	A	0.41	0.40	1.08	0.85 (0.46–1.59)
HLA-DRB5	rs9271192	6	32610753	C	C	0.29	0.27	1.11	1.00 (0.61–1.63)
CD2AP	rs9349407	6	47485642	C	C	0.27	0.30	1.14	1.31 (0.81–2.14)
TREM2	rs75932628	6	41161514	T	T	0.004	0.01	4.50	3.18 (0.61–16.71)
NME8	rs2718058	7	37801932	G	A	0.34	0.39	1.08	0.76 (0.38–1.52)
ZCWPW1	rs1476679	7	100406823	C	T	0.29	0.30	1.09	1.14 (0.47–2.76)
EPHA1	rs11767557	7	143412046	C	T	0.21	0.20	1.18	1.16 (0.34–3.92)
CLU	rs11136000	8	27607002	T	C	0.43	0.42	1.20	1.13 (0.59–2.18)
PTK2B	rs28834970	8	27337604	C	C	0.38	0.38	1.10	1.09 (0.66–1.80)
CELF1	rs10838725	11	47536319	C	C	0.30	0.34	1.08	0.98 (0.60–1.60)
MS4A6A	rs610932	11	60171834	T	G	0.44	0.29	1.14	2.85 (1.21–6.74)
PICALM	rs3851179	11	86157598	T	C	0.37	0.36	1.15	0.68 (0.36–1.27)
SORL1	rs11218343	11	121564878	C	T	0.04	0.03	1.32	0.65 (0.03–13.60)
FERMT2	rs17125944	14	52933911	C	C	0.09	0.08	1.13	0.89 (0.45–1.75)
SLC24A4	rs10498633	14	92460608	T	G	0.25	0.25	1.11	1.62 (0.49–5.39)
ABCA7	rs3764650	19	1046521	G	G	0.09	0.09	1.22	1.06 (0.56–2.00)
CD33	rs3865444	19	51224706	A	C	0.30	0.29	1.14	1.11 (0.49–2.53)
CASS4	rs7274581	20	56443204	C	T	0.08	0.12	1.15	0.38 (0.04–3.71)

Information on the 22 gene variants included in PRS calculation. Minor allele frequencies presented are for cognitively normal (CN) older adults and Alzheimer’s disease (AD) patients in the AIBL cohort. Published OR and AIBL OR (95% CI) are given for the risk allele as per published information [4–8 , 55]. SNP, single nucleotide polymorphism; AIBL, Australian Imaging, Biomarkers and Lifestyle Study of Ageing; MAF, minor allele frequency; OR, odds ratio; CI, confidence interval.

PRS calculation

The PRS was calculated by weighting SNPs based on their published association with AD risk. The number (0, 1, or 2) of risk alleles for each SNP was multiplied by the natural logarithm of its respective published odds ratio for the clinical diagnosis of AD (Table 1) [4–8 , 55]. The results of these calculations were then summed for each sample giving the individual sample APOE inclusive PRS (PRS ${\bar{c}}^{APOE}$ ). Additionally, a PRS was calculated excluding APOE genotype (PRS ${\bar{s}}^{APOE}$ ) as the significantly higher effect-size of APOE may have potential saturating effect on the PRS.

Statistical analysis

All statistical analyses were performed using Rstudio (Rstudio Team 2015) Version 0.98.1103 for Macintosh [56]. Means, standard deviations, and percentages for demographic measures, are given for AD patients, CN older adults and β ^high CN older adults. ANOVA (age, premorbid IQ, depressive symptoms) and chi-squared tests (gender, years of education, APOE ɛ4 + ve, high Aβ burden) were used to determine the significance of differences between AD patients and CN older adults. Minor allele frequencies (MAF) for all SNPs were calculated in both the CN older adult and AD patient populations. Odds ratios and 95% confidence intervals were calculated in the AIBL cohort, with respect to the clinical diagnosis of AD, using the “epitools” package in R, with reference to the SNP’s risk allele.

To investigate the ability of both PRS to predict AD, the “ROCR” package in R was used to generate a Receiver Operating Characteristic (ROC) curve and calculate Area Under the Curve (AUC). Partial correlations between PRS and AD biomarkers (Aβ imaging, CSF Aβ₄₂ and CSF total-tau) were performed using the “ppcor” package in R, covarying for sex and age at measurement. PRS and AD biomarker values residualized by covariates are graphically presented with regression lines. Random intercepts linear mixed-effects (LME) models, utilizing the “nlme” package in R, were used to determine associations between PRSs and cognitive measures. Initially, a PRS×Time interaction was modelled, with cognitive composites as the dependent variables. These LME models were presented graphically as PRS against both baseline composite score and composite score change over time, with PRS dependent standard error. Additionally, PRSs were partitioned into quartiles, and a PRS quartile group×Time interaction was modelled. PRS quartile group LME models were presented graphically with time point dependent standard error. All analyses for the AIBL-PACC co-varied for age.

RESULTS

Baseline demographics

Demographic data for the clinical groups (AD patients, CN older adults) are presented in Table 2. Significant differences were observed when comparing AD patients and CN older adults (Table 2) in age (p < 0.001), premorbid IQ (p < 0.001), depressive symptoms (p < 0.001), APOE ɛ4 allele carriage (p < 0.001), and high Aβ burden (p < 0.001). Further, demographic information is presented for CN older adults with high Aβ burden (n = 570). Minor allele frequencies and odds ratios for all SNPs included in the PRSs are presented in Table 1.

Table 2

Baseline demographic information

		AD n = 73	CN (all) n = 570	CN (Aβ ^high ) n = 247
Age, y		74.31±7.67	70.86±6.42	72.44±6.54
Female (%)		54.79	55.44	51.82
Education, y (%)	0–8	16.90	7.75	8.57
	9–12	32.39	36.80	37.15
	13–15	22.54	20.77	22.45
	15+	28.17	34.68	31.84
Premorbid IQ (FSIQ)		103.33±9.25	108.05±7.21	107.96±7.46
Depressive Symptoms (GDS)		2.51±2.68	1.04±1.27	1.01±1.23
APOEɛ4 + ve (%)		72.60	28.07	42.91
CSF Collection (n)		16	42	20
Aβ ^high (%)		94.52	43.33	100

Baseline demographic information for Alzheimer’s disease (AD) patients (n = 73) and cognitively normal older adults (CN, n = 570) in the AIBL Study. CN older adults with high Aβ (Aβ ^high , n = 247) were used to investigate changes in baseline and longitudinal cognition. All values represented as mean±standard deviation, unless otherwise stated. Italicized text denotes p < 0.001 when comparing AD and CN older adults. GDS, Geriatric Depression Scale; FSIQ, WTAR-estimated FSIQ.

PRSs are associated with risk for AD and related biomarkers

As detailed in the methods section, two PRSs were calculated based on the allelic frequency and effect sizes of 22 AD-associated genetic variants. One PRS was calculated that included APOE genotype (PRS ${\bar{c}}^{APOE}$ ; range of scores: 1.812168 to 5.416897) and another that excluded it (PRS ${\bar{s}}^{APOE}$ ; range of scores: 1.746423 to 4.474692). The respective utility of these PRSs was then assessed with respect to AD risk and related imaging and CSF biomarkers. The inclusion of APOE improved the AUC for AD discrimination from 0.5975 to 0.7472 (Fig. 1). In those who underwent brain imaging at baseline (n = 347), the PRS ${\bar{c}}^{APOE}$ was positively correlated with brain Aβ burden (r = 0.4454, r² = 0.1984, p < 0.0001, Fig. 2a). This correlation remained significant when analyzed in CN older adults only (n = 282, r = 0.3856, r² = 0.1487, p < 0.0001, Fig. 2c). Similarly, the PRS ${\bar{s}}^{APOE}$ was observed to be positively correlated with brain Aβ burden in all (r = 0.1829, r² = 0.0334, p = 0.0006, Fig. 2b) and CN older adults (r = 0.1259, r² = 0.0159, p = 0.0352, Fig. 2d).

Fig. 1.

Receiver operating characteristic (ROC) curves. ROC curves displaying the ability of the respective PRSs to discriminate between cognitively normal (CN) older adults (n = 570) and Alzheimer’s disease (AD) patients (n = 73) in the AIBL study. PRS ${\bar{c}}^{APOE}$ (solid black line), PRS ${\bar{s}}^{APOE}$ (dotted black line), APOE only (broken black line). AUC, area under the curve; PRS ${\bar{c}}^{APOE}$ /PRS ${\bar{s}}^{APOE}$ , polygenic risk score with and without APOE, respectively.

Fig. 2.

Correlation between PRSs brain amyloid-β. Correlation between PRSs and brain amyloid-β (Aβ) burden in the whole cohort (a, b) and cognitively normal older adults (c, d). Plotted values represent residuals after model correction for age and sex. Regression lines are presented with 95% confidence interval error shading. AD, Alzheimer’s disease; CN, cognitively normal; PRS ${\bar{s}}^{APOE}$ ^OE, polygenic risk score with APOE genotype; PRS ${\bar{s}}^{APOE}$ ^OE, polygenic risk score without APOE genotype.

Correlations of PRSs with CSF biomarkers were assessed in both a combined group (AD and CN; n = 58) and considering CN older adults only (n = 42). The PRS ${\bar{c}}^{APOE}$ was inversely correlated with CSF Aβ₄₂ (r = –0.4092, r² = 0.1675, p = 0.0017; Fig. 3a) and Aβ₄₂:total-tau ratio (r = –0.4572, r² = 0.2077, p = 0.0007; Fig. 3d), but was not associated with either total-tau or phospho-tau (Fig. 3b, c). These associations, although slightly weaker, held when considering only CN older adults (Aβ₄₂ r = –0.2444, r² = 0.0778, p = 0.0374; Fig. 3e; Aβ₄₂:total-tau ratio r = –0.4589, r² = 0.2105, p = 0.0038; Fig. 3h) and remained non-significant for total-tau and phospho-tau (Fig. 3f, g). When investigating the PRS ${\bar{s}}^{APOE}$ different patterns of correlations were observed (Fig. 4). The PRS ${\bar{s}}^{APOE}$ was not correlated with Aβ₄₂ in either the combined group (Fig. 4a) or when considering CN older adults only (Fig. 4e). However, the PRS ${\bar{s}}^{APOE}$ was positively correlated with total-tau (r = 0.1787, r² = 0.1120, p = 0.0348, Fig. 4b) and phospho-tau (r = 0.2044, r² = 0.0828, p = 0.0719, Fig. 4c) though this only trended toward significance in the latter. When CN older adults were only considered these correlations were both significant (total-tau: r = 0.3704, r² = 0.1372, p = 0.0186, Fig. 4f; phospho-tau: r = 0.3167, r² = 0.1003, p = 0.0465, Fig. 4g). In both the combined and CN only analyses PRS ${\bar{s}}^{APOE}$ was inversely correlated with Aβ₄₂:total-tau ratio (AD and CN: r = –0.3776, r² = 0.1419, p = 0.0058, Fig. 4d; CN: r = –0.3818, r² = 0.1454, p = 0.0180, Fig. 4h).

Fig. 3.

Correlation between PRS ${\bar{c}}^{APOE}$ and cerebrospinal fluid biomarkers. Correlation between PRS and Cerebrospinal Fluid (CSF) biomarkers including Aβ₄₂ (a, e), total-tau (b, f), phospho-tau (c, g), and Aβ₄₂:total-tau ratio (d, h) in the whole cohort (a-d) and cognitively normal (e-h). Plotted values represent residuals after model correction for age and sex. Regression lines are presented with 95% confidence interval error shading. AD, Alzheimer’s disease; CN, cognitively normal; PRS ${\bar{c}}^{APOE}$ , polygenic risk score with APOE genotype.

Fig. 4.

Correlation between PRS ${\bar{s}}^{APOE}$ and cerebrospinal fluid biomarkers. Correlation between PRS and cerebrospinal fluid (CSF) biomarkers including Aβ₄₂ (a, e), total-tau (b, f), phospho-tau (c, g), and Aβ₄₂:total-tau ratio (d, h) in the whole cohort (a-d) and cognitively normal (e-h). Plotted values represent residuals after model correction for age and sex. Regression lines are presented with 95% confidence interval error shading. AD, Alzheimer’s disease; CN, cognitively normal; PRS ${\bar{s}}^{APOE}$ , polygenic risk score without APOE genotype.

PRS

${\bar{c}}^{APOE}$ is associated with cognitive decline in Aβ^high CN older adults but is primarily driven by APOE

To assess the utility of a PRS incorporating APOE genotype (PRS ${\bar{c}}^{APOE}$ ) for predicting rates of cognitive decline in the preclinical stages of AD we investigated the relationship between PRS ${\bar{c}}^{APOE}$ and baseline and longitudinal cognitive composite scores in Aβ ^high CN older adults. Significant differences were observed both at baseline and with longitudinal performance. With increasing PRS ${\bar{c}}^{APOE}$ lower baseline results (Fig. 5a–c, with α (baseline) and β (longitudinal change) weights presented in Table 3) in the statistically driven global composite (p = 0.0022), verbal episodic memory (p = 0.0016), and PACC (p = 0.0449) were observed. Further, an increasing PRS ${\bar{c}}^{APOE}$ was also associated with increasingly worsening rates of cognitive change (Fig. 5d–f, Table 3) in the statistically driven global composite (p = 0.0006), verbal episodic memory (p = 0.0007) and PACC (p = 0.0084).

Fig. 5.

Association between PRS ${\bar{c}}^{APOE}$ and baseline and longitudinal change in cognition in Aβ ^high CN older adults. Association between PRS ${\bar{c}}^{APOE}$ and baseline (a-c) and longitudinal (d-f) change in composite measures of cognition including a statistically driven global composite (a, d), verbal episodic memory composite (b, e), and Pre-Alzheimer’s Cognitive Composite (AIBL-PACC; c, f) in cognitively normal (CN) older adults with high Aβ (Aβ^high; n = 247). AIBL-PACC controlled for age. Shaded regions represent PRS dependent standard error. PRS ${\bar{c}}^{APOE}$ , polygenic risk score with APOE genotype.

To apportion a level of genetic risk the PRS

{\bar{c}}^{APOE}

was partitioned into quartiles, with risk increasing (low-to-high) with increasing quartile (Q1 through Q4), i.e., PRS

{\bar{c}}^{APOE}

-Q1(range, 1.97–2.71) < PRS

{\bar{c}}^{APOE}

-Q2(range, 2.72–3.11) <PRS

{\bar{c}}^{APOE}

-Q3(range, 3.12–3.69) < PRS

{\bar{c}}^{APOE}

-Q4(range, 3.70–5.42). There was no significant between quartile differences in mean age at baseline (Q1, 73.07±6.49 years; Q2, 72.80±6.88 years; Q3, 72.57±6.33 years; Q4, 71.40±6.47 years; F = 0.796, p = 0.497). PRS

{\bar{c}}^{APOE}

-Q4 had significantly lower baseline scores when compared to PRS

{\bar{c}}^{APOE}

-Q1 and Q2 in the statistically driven global composite (p = 0.0087 and p = 0.0019, respectively) and verbal episodic memory (p = 0.0067 and p = 0.0020, respectively), and AIBL-PACC (PRS

{\bar{c}}^{APOE}

-Q2 only, p = 0.0230; Fig. 6, Table 4). Additionally, PRS

{\bar{c}}^{APOE}

-Q4 showed greater rates of decline when compared to all other groups, on the statistically driven global composite (PRS

{\bar{c}}^{APOE}

-Q1 p = 0.0006; PRS

{\bar{c}}^{APOE}

-Q2 p = 0.0170; PRS

{\bar{c}}^{APOE}

-Q3 p = 0.0006), verbal episodic memory (PRS

{\bar{c}}^{APOE}

-Q1 p = 0.0019; PRS

{\bar{c}}^{APOE}

-Q2 p = 0.0383; PRS

{\bar{c}}^{APOE}

-Q3 p = 0.0155), and AIBL-PACC (PRS

{\bar{c}}^{APOE}

-Q1 p = 0.0035; PRS

{\bar{c}}^{APOE}

-Q2 p = 0.0310; PRS

{\bar{c}}^{APOE}

-Q3 p < 0.0001) over

Table 3

Association between PRS ${\bar{c}}^{APOE}$ and baseline and longitudinal change in cognition in Aβ ^high CN older adults

	α	p	β	p
Global	–0.3879	0.0022	–0.0692	0.0006
Verbal Episodic Memory	–0.3940	0.0016	–0.0659	0.0007
AIBL-PACC	–0.7245	0.0449	–0.1451	0.0084

Models assessing the association between PRS ${\bar{c}}^{APOE}$ and baseline (α) and longitudinal (β) change in cognitive composite measures in cognitively normal (CN) older adults with high Aβ (Aβ^high; n = 247). AIBL-PACC controlled for age. PRS ${\bar{c}}^{APOE}$ , polygenic risk score with APOE genotype.

Fig. 6.

Cognitive rates of change in Aβ ^high CN older adults by PRS ${\bar{c}}^{APOE}$ quartile. Rates of change in a statistically driven global composite (a), verbal episodic memory composite (b), and Pre-Alzheimer’s Cognitive Composite (AIBL-PACC; c) in cognitively normal (CN) older adults with high Aβ (Aβ^high; n = 247). PRS ${\bar{c}}^{APOE}$ -Q1 (green), PRS ${\bar{c}}^{APOE}$ -Q2 (blue), PRS ${\bar{c}}^{APOE}$ -Q3 (orange), PRS ${\bar{c}}^{APOE}$ -Q4 (red), APOE ɛ4 negative (–ve; grey dotted), APOE ɛ4 positive (+ve; black broken). AIBL-PACC controlled for age. *p < 0.05 compared to PRS ${\bar{c}}^{APOE}$ -Q1 slope, p < 0.05 compared to PRS ${\bar{c}}^{APOE}$ -Q2 slope, ^φp < 0.05 compared to PRS ${\bar{c}}^{APOE}$ -Q3 slope, ${\bar{c}}^{APOE}$ p < 0.05 compared to PRS ${\bar{c}}^{APOE}$ -Q1 baseline score, ^Ωp < 0.05 compared to PRS ${\bar{c}}^{APOE}$ -Q2 baseline score, ^γp < 0.05 compared to PRS ${\bar{c}}^{APOE}$ -Q3 baseline score. Shaded regions represent time dependent standard error. PRS ${\bar{c}}^{APOE}$ , polygenic risk score with APOE genotype.

Table 4

Group intercepts and slopes for cognitive composites in Aβ ^high CN older adults by PRS ${\bar{c}}^{APOE}$ quartile

	Quartile 1		Quartile 2		Quartile 3		Quartile 4
	α	β	α	β	α	β	α	β
Global	0.040	0.023	0.149	–0.016	–0.198	0.024	–0.696 ${\bar{c}}^{APOE}$ ^Ωγ	–0.103 ^*∧φ
Verbal Episodic Memory	0.021	0.032	0.102	–0.005	–0.169	0.008	–0.631 ${\bar{c}}^{APOE}$ ^Ω	–0.078 ^*∧φ
AIBL-PACC	–0.597	0.027	–0.097	–0.051	–0.656	0.126	–1.645^Ω	–0.264 ^*∧φ

Group intercepts (α) and slopes (β) for cognitive composites (presented in SD and SD/year) and AIBL-PACC (presented in SD×4 and SD×4/year) in imaged cognitively normal (CN) older adults with high Aβ (Aβ ^high ; n = 247). AIBL-PACC controlled for age. *p < 0.05 compared to PRS ${\bar{c}}^{APOE}$ -Q1 slope, p < 0.05 compared to PRS ${\bar{c}}^{APOE}$ -Q2 slope, ^φp < 0.05 compared to PRS ${\bar{c}}^{APOE}$ -Q3 slope, ${\bar{c}}^{APOE}$ p < 0.05 compared to PRS ${\bar{c}}^{APOE}$ -Q1 baseline score, ^Ωp < 0.05 compared to PRS ${\bar{c}}^{APOE}$ -Q2 baseline score, ^γp < 0.05 compared to PRS ${\bar{c}}^{APOE}$ -Q3 baseline score. PRS ${\bar{c}}^{APOE}$ , polygenic risk score with APOE genotype.

7.5 years (Fig. 6, with group intercepts and slopes presented in Table 4). For comparative purposes both APOE carriers and non-carriers are also presented in Fig. 6. Negligible differences in group slopes for APOE carriers compared to PRS^APOE-Q4 were observed (APOE carriers versus PRS^APOE-Q4: Global, –0.071SD/year versus –0.103; verbal episodic memory, –0.057 versus –0.078; AIBL-PACC, –0.181 versus –0.264), while baseline cognition across the three composites tended to be higher (APOE carriers versus PRS^APOE-Q4: Global, –0.352 versus –0.696; verbal episodic memory, –0.371 versus –0.631; AIBL-PACC, –1.092 versus –1.645). However, when APOE carriers were split into ɛ4 homozygotes and heterozygotes, PRS^APOE-Q4 was observed to have a baseline cognitive performance that fell between these two groups (see Supplementary Figure 1).

While a PRS incorporating APOE genotype showed utility for predicting rates of cognitive decline in the preclinical stages of AD no such utility was observed for a PRS excluding APOE genotype (PRS ${\bar{s}}^{APOE}$ ). No significant association between the PRS ${\bar{s}}^{APOE}$ and cognition was observed at baseline or longitudinally, for any composite measure when investigating Aβ ^high CN older adults (Table 5). To ascertain whether the PRS ${\bar{s}}^{APOE}$ was associated with cognitive decline in either the presence or absence of APOE ɛ4, Aβ ^high CN older adults were stratified by carrier status. No significant association was observed when individually investigating Aβ ^high CN older adults carrying no APOE alleles (n = 141; ɛ3/ɛ2, ɛ3/ɛ3), one APOE allele (n = 86; ɛ4/ɛ2, ɛ4/ɛ3), or two APOE alleles (n = 20; ɛ4/ɛ4; Table 5).

Table 5

Association between PRS ${\bar{s}}^{APOE}$ and baseline and longitudinal change in cognition in Aβ ^high CN older adults

	α	p	β	p
All (n = 247)
Global	–0.2638	0.2699	–0.0179	0.6176
Verbal Episodic Memory	–0.2397	0.3108	–0.0291	0.4008
AIBL-PACC	–0.3604	0.5939	–0.0091	0.9262
APOEɛn/ɛn (n = 141)
Global	–0.2751	0.3121	–0.0191	0.6402
Verbal Episodic Memory	–0.2423	0.3786	–0.0476	0.2293
AIBL-PACC	–0.1195	0.8833	0.1093	0.2916
APOEɛn/ɛ4 (n = 86)
Global	–0.1393	0.7365	–0.0813	0.2348
Verbal Episodic Memory	–0.1318	0.7413	–0.0630	0.3384
AIBL-PACC	–0.0024	0.9983	–0.3232	0.1145
APOEɛ4/ɛ4 (n = 20)
Global	–1.8195	0.2066	0.2975	0.1858
Verbal Episodic Memory	–1.9313	0.1758	0.3645	0.1079
AIBL-PACC	–5.3797	0.1674	–0.0343	0.9406

Group slopes for models assessing the association between PRS and baseline (α) and longitudinal (β) change in cognitive composite measures in all cognitively normal (CN) older adults with high Aβ (Aβ^high; n = 247), and by number of APOE ɛ4 alleles carried. AIBL-PACC controlled for age. PRS ${\bar{s}}^{APOE}$ , polygenic risk score without APOE genotype.

Distribution of APOE genotypes within PRS quartiles

Investigation into the distribution of APOE genotypes between quartiles revealed a pattern of increasing PRS with increasing number of APOE ɛ4 alleles (Fig. 7). Quartile 4 contained all ɛ4 homozygotes (n = 20; Fig. 7), which made up 30.8% of the quartile. In addition, 50.0% of ɛ4 heterozygotes were also present in quartile 4 (66.2% of quartile 4), and 1.4% of samples with no ɛ4 alleles (3.1% of quartile 4; Fig. 7). The additional ɛ4 heterozygotes were present in quartile 3 and 2, 44.2% (62.3% of quartile 3) and 5.8% (8.1% of quartile 2) respectively (Fig. 7). Further, the remaining participants with no ɛ4 alleles were allocated to quartile 3, 2, and 1 with 16.3% (37.7% of quartile 3), 40.4% (91.9% of quartile 2) and 41.8% (100% of quartile 1), respectively (Fig. 7).

Fig. 7.

Distribution of APOE genotypes within PRS ${\bar{c}}^{APOE}$ quartiles. Spread and sum of APOE genotypes within PRS quartiles. No APOE ɛ4 alleles (ɛ3/ɛ2, ɛ3/ɛ3; black triangles), one APOE ɛ4 allele (ɛ4/ɛ2, ɛ4/ɛ3; dark grey squares), two APOE alleles (ɛ4/ɛ4; light grey circles). Lines within quartile groups represent median and interquartile range. PRS ${\bar{c}}^{APOE}$ ^AP ^OE, polygenic risk score with APOE genotype.

DISCUSSION

The primary aim of the current study was to investigate the utility of current AD-risk-weighted methods for PRS calculation in predicting rates of cognitive decline in preclinical AD. However, the current study first sought to replicate previous associations with AD risk and biomarkers. Investigation into the PRS’ ability to predict AD found a marked improvement with the inclusion of APOE, although there was no improvement above that of APOE alone. Suggesting that the admixture of additional genes in an AD-risk-weighted PRS offers little above and beyond that of APOE genotype alone. When evaluating AD biomarker associations relating to the PRSs with and without APOE, significant positive correlations were observed with brain Aβ burden both in the combined group and when considering only CN older adults, which is consistent with a previous study [14]. With respect to CSF biomarkers there were slightly different associations observed for the PRSs, dependent on the inclusion of APOE. Both PRSs were inversely correlated with Aβ₄₂:total-tau ratio. However, the inclusion of APOE (PRS ${\bar{c}}^{APOE}$ ) resulted in a PRS significantly inversely correlated with CSF Aβ₄₂ but not significantly correlated with either CSF total-tau or phospho-tau. The exclusion of APOE (PRS ${\bar{s}}^{APOE}$ ) resulted in no significant correlation with CSF Aβ₄₂ but a significant positive correlation with both CSF total-tau and phospho-tau. This validates previous associations of PRSs, including APOE, with CSF Aβ₄₂ [25], and, without APOE, with measures of CSF tau [23]. These differences in CSF biomarker associations due to the presence/absence of APOE require further investigation in larger sample sizes. However, the association of both PRSs with brain Aβ burden suggests that the admixture of different genetic variants may influence AD biomarker associations above and beyond APOE alone.

The association between PRSs and cognition was evaluated in a CN older adult at risk (preclinical) sample based on Aβ brain imaging. A significant association between the PRS ${\bar{c}}^{APOE}$ and both baseline and longitudinal measures of global cognition, verbal episodic memory, and the PACC was observed. In addition, when the PRS ${\bar{c}}^{APOE}$ was partitioned into quartiles the upper quartile of polygenic risk was found to have significantly increased rates of cognitive decline when compared to all other quartiles in all measures of cognition. The ability of this study to observe an association between the PRS including APOE and cognitive decline in CN older adults, where others have not [10 , 28], is likely due to the participants included and the cognitive measures utilized. Specifically, higher levels of brain Aβ has previously been associated with increased rates of cognitive decline [36]. Therefore, the selection of Aβ ^high CN older adults maximized the likelihood of observing statistically significant rates of cognitive decline. Additionally, the cognitive composite scores evaluated, and in particular verbal episodic memory, characterize one of the earliest aspects of preclinical decline [57], allowing for observation of cognitive change in those considered cognitively normal. However, when comparing baseline and longitudinal cognition to APOE alone, the group slopes showed negligible difference between the highest quartile of risk and the presence of at least one APOE ɛ4 allele. This suggests the additional genetic variants in the PRS do not result in observable differences in the rate of decline compared to APOE ɛ4 alone. However, baseline cognition was observably worse in the highest quartile of risk. This observation is tempered by the fact that the highest quartile of risk includes all ɛ4 homozygotes and a smaller proportion of ɛ4 heterozygotes. When considered against ɛ4 homozygotes and heterozygotes separately the highest quartile of risk presents with baseline cognitive performance that falls between these two groups. Further, when APOE was removed from the calculation of the PRS, no association between the PRS ${\bar{s}}^{APOE}$ and any measure of cognition was observed in all Aβ ^high CN older adults. Finally, when partitioning by number of APOE ɛ4 alleles (0, 1, or 2) no association with cognitive performance was observed. This loss of association with measures of cognitive performance after the exclusion of APOE has been previously reported [15, 17], though this study is the first to report this finding in Aβ ^high CN older adults.

The findings reported herein suggest that APOE is a strong determinant of the ability of a PRS to predict cognitive decline with the initial PRS associations with cognition possibly due to the over-representation of APOE ɛ4 carriers at the higher end of the PRS range—particularly in the upper quartile of the PRS, which contained 100% of the APOE ɛ4 homozygotes. When considered collectively the AD-risk-associated weighting applied to APOE ɛ4 has a strong influence within the score, most likely saturating the effects of additional genes present. Few studies have investigated the specific influence of APOE on PRS performance in this regard, and no studies have assessed this specifically in Aβ ^high CN older adults. One study by Mormino and colleagues was successful in calculating a PRS without the influence of APOE that was significantly associated with baseline and longitudinal cognition [14]. That study observed a significant association between their PRS, and memory and executive function in older participants without dementia enrolled in the Alzheimer’s Disease Neuroimaging Initiative (ADNI) study [14]. In contrast with the study presented here, the PRS evaluated by Mormino et al. had a far less stringent statistical significance level (p < 0.01 as opposed to genome-wide significance, p = 5×10 ^- ⁸) for inclusion of SNPs, as reported by the International Genomics of Alzheimer’s Project (IGAP) meta-analysis [5]. This resulted in the inclusion of 16,123 SNPs, which were weighted by their log odds ratio [14]. In addition, that study investigated a conservative 18 SNP (p-value cut-off at 5×10 ^- ⁸) PRS, and similar to the current study reported no association with cognitive performance in the absence of APOE. In another study, Desikan and colleagues developed a polygenic hazard score through the combination of GWAS data and disease incidence rates from population-based epidemiology [35]. The resulting score included 33 SNPs and had the ability to predict age at onset in the absence of APOE [35]. With the inclusion of additional information into the weighting of a polygenic risk score this study demonstrated the ability of a conservative PRS, without APOE, to be able to predict an outcome [35]. Therefore, it is proposed that, at least within the AIBL Study, the prediction of preclinical decline in cognition using a conservative PRS weighted by AD risk offers little above and beyond that of APOE genotype alone.

Our study confirms recent findings that suggest that an AD-risk-weighted PRS has limited utility in terms of assessing AD progression [58]. In addition to the heavy weighting placed on APOE, there are potentially factors that contribute to this reduced utility. An AD-risk based approach to PRS construction is potentially biased as it excludes both “lesser” significant variants, but perhaps more pertinent, it would exclude variants that while not strong AD-risk factors are associated with cognitive performance. One such example is the brain-derived neurotrophic factor (BDNF), the Val66Met variant of which has previously been associated with cognitive decline in Aβ ^high CN older adults [59, 60]. This supports the notion that the genetic architecture that underpins AD-risk is likely different from that which underpins AD-progression, particularly at the preclinical stage. This being said in addition to APOE, several genes included within the PRS investigated in this current study have been individually associated with cognitive change. For example, studies investigating the SNPs, or in linkage with those presented here, within the genes CR1 [15 , 62], ABCA7 [15, 61], CLU [62], BIN1 [63], and NME8 [64], have been associated with cognition and memory. Therefore, an optimized PRS with increased utility for the prediction of cognitive decline at the preclinical stages of AD is still likely to include several AD-risk associated genetic variants that were included in the current PRS assessed herein in addition to cognition-associated risk variants. However, it is likely that the effect sizes derived from associations with risk for developing AD will be substantially different from those associated with the rate at which cognition declines. As such, the combination of phenotype-associated variants and the development of phenotype appropriate effect size weightings may be more efficacious and thus warrants further investigation.

While the inclusion of brain Aβ burden and extended longitudinal follow-up periods are strengths of this study, the following limitations are acknowledged. Participants in the AIBL Study are voluntarily involved; this results in high levels of cognitive performance that are possibly not representative of the general population and difficult to replicate. Further, while the use of composite domain specific measures of cognition is becoming more common, different results could be yielded by alterations in the specific composition of cognitive composites used. Finally, while the length (7.5 years) of longitudinal follow-up data presented here is a clear strength of the study, it is acknowledged that different results could be yielded from the analysis of different durations of longitudinal follow-up, and particularly those of a shorter duration. Finally, the use of AD-risk, as both a basis of genetic variant inclusion and weighting, in the calculation of a PRS for preclinical cognitive decline is itself a potential limitation as it excludes genetic variants associated with cognition and utilizes weightings that are not specific for the rates of decline.

CONCLUSION

In conclusion, this study reports that AD-risk-weighted PRSs are associated with cognitive decline in a cognitively normal at-risk population; however, this association is dependent upon the inclusion of APOE genotype in the score. These results validate those reported in previous studies investigating cross-sectional and longitudinal AD cohorts. This suggests that the utility of an AD-risk-weighted PRS, above that of APOE genotype alone, is perhaps negligible, particularly for the prediction of early stage decline in the absence of APOE genotype. As half of AD cases do not carry an APOE ɛ4 allele, it is important to investigate APOE independent predictors of decline. Lessons from other studies point to the inclusion of phenotype specific weightings to account for the effects of genes in addition to APOE, thus further investigation of this premise is warranted.

Footnotes

ACKNOWLEDGMENTS

Funding for the AIBL 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 AIBL study has 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 (CRC) for Mental Health (Grant ID:20100104), an Australian Government Initiative. We thank all those who took part as subjects in the study for their commitment and dedication to helping advance research into the early detection and causation of AD. We kindly thank all AIBL Research Group members (http://aibl.csiro.au/about/aibl-research-team/). All data and samples used in this study are derived from the Australian Imaging, Biomarkers and Lifestyle (AIBL) Study of Ageing. AIBL data can be accessed through an Expression of Interest procedure, for more information please see .

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Gatz

, Reynolds

, Fratiglioni

, Johansson

, Mortimer

, Berg

, Fiske

, Pedersen

(2006) Role of genes and environments for explaining Alzheimer disease. Arch Gen Psychiatry 63, 168–174.

Ridge

, Mukherjee

, Crane

, Kauwe

, Alzheimer’s Disease Genetics Consortium (2013) Alzheimer’s disease: Analyzing the missing heritability. PLoS One 8, e79771.

Karch

, Cruchaga

, Goate

(2014) Alzheimer’s disease genetics: From the bench to the clinic. Neuron 83, 11–26.

Bertram

, Lange

, Mullin

, Parkinson

, Hsiao

, Hogan

, Schjeide

, Hooli

, Divito

, Ionita

, Jiang

, Laird

, Moscarillo

, Ohlsen

, Elliott

, Wang

, Hu-Lince

, Ryder

, Murphy

, Wagner

, Blacker

, Becker

, Tanzi

(2008) Genome-wide association analysis reveals putative Alzheimer’s disease susceptibility loci in addition to APOE. Am J Hum Genet 83, 623–632.

Lambert

, Ibrahim-Verbaas

, Harold

, Naj

, Sims

, Bellenguez

, DeStafano

, Bis

, Beecham

, Grenier-Boley

, Russo

, Thorton-Wells

, Jones

, Smith

, Chouraki

, Thomas

, Ikram

, Zelenika

, Vardarajan

, Kamatani

, Lin

, Gerrish

, Schmidt

, Kunkle

, Dunstan

, Ruiz

, Bihoreau

, Choi

, Reitz

, Pasquier

, Cruchaga

, Craig

, Amin

, Berr

, Lopez

, De Jager

, Deramecourt

, Johnston

, Evans

, Lovestone

, Letenneur

, Moron

, Rubin-sztein

, Eiriksdottir

, Sleegers

, Goate

, Fievet

, Huentelman

, Gill

, Brown

, Kamboh

, Keller

, Barberger-Gateau

, McGuiness

, Larson

, Green

, Myers

, Dufouil

, Todd

, Wallon

, Love

, Rogaeva

, Gallacher

, St George-Hyslop P, Clarimon

, Lleo

, Bayer

, Tsuang

, Yu

, Tsolaki

, Bossu

, Spalletta

, Proitsi

, Collinge

, Sorbi

, Sanchez-Garcia

, Fox

, Hardy

, Deniz Naranjo

, Bosco

, Clarke

, Brayne

, Galimberti

, Mancuso

, Matthews

, European Alzheimer’s Disease Initiative (EADI); Genetic and Environmental Risk in Alzheimer’s Disease; Alzheimer’s Disease Genetic Consortium; Cohorts for Heart and Aging Research in Genomic Epidemiology, Moebus

, Mecocci

, Del Zompo

, Maier

, Hampel

, Pilotto

, Bullido

, Panza

, Caffarra

, Nacmias

, Gilbert

, Mayhaus

, Lannefelt

, Hakonarson

, Pichler

, Carrasquillo

, Ingelsson

, Beekly

, Alvarez

, Zou

, Valladares

, Younkin

, Coto

, Hamilton-Nelson

, Gu

, Razquin

, Pastor

, Mateo

, Owen

, Faber

, Jonsson

, Combarros

, O'Donovan MC, Cantwell

, Soininen

, Blacker

, Mead

, Mosley

Jr , Bennett

, Harris

, Fratiglioni

, Holmes

, de Bruijn

, Passmore

, Montine

, Bettens

, Rotter

, Brice

, Morgan

, Foroud

, Kukull

, Hannequin

, Powell

, Nalls

, Ritchie

, Lunetta

, Kauwe

, Boerwinkle

, Riemenschneider

, Boada

, Hiltuenen

, Martin

, Schmidt

, Rujescu

, Wang

, Dartigues

, Mayeux

, Tzourio

, Hofman

, Nothen

, Graff

, Psaty

, Jones

, Haines

, Holmans

, Lathrop

, Pericak-Vance

, Launer

, Farrer

, van Duijn

, Van Broeckhoven

, Moskv-ina

, Seshadri

, Williams

, Schellenberg

, Amouyel

(2013) Meta-analysis of 74,046 individuals identifies 11 new susceptibility loci for Alzheimer’s disease. Nat Genet 45, 1452–1458.

Hollingworth

, Harold

, Sims

, Gerrish

, Lambert

, Carrasquillo

, Abraham

, Hamshere

, Pahwa

, Moskvina

, Dowzell

, Jones

, Stretton

, Thomas

, Richards

, Ivanov

, Widdowson

, Chapman

, Lovestone

, Powell

, Proitsi

, Lupton

, Brayne

, Rubinsztein

, Gill

, Lawlor

, Lynch

, Brown

, Passmore

, Craig

, McGuinness

, Todd

, Holmes

, Mann

, Smith

, Beaumont

, Warden

, Wilcock

, Love

, Kehoe

, Hooper

, Vardy

, Hardy

, Mead

, Fox

, Rossor

, Collinge

, Maier

, Jessen

, Rüther E, Schürmann B, Heun

, Kölsch H, van den Bussche

, Heuser

, Kornhuber

, Wiltfang

, Dichgans

, Frolich

, Hampel

, Gallacher

, Hüll M, Rujescu

, Giegling

, Goate

, Kauwe

, Cruchaga

, Nowotny

, Morris

, Mayo

, Sleegers

, Bettens

, Engelborghs

, De Deyn

, Van Broeckhoven

, Livingston

, Bass

, Gurling

, McQuillin

, Gwilliam

, Deloukas

, Al-Chalabi

, Shaw

, Tsolaki

, Singleton

, Guerreiro

, Mühleisen TW, Nothen

, Moebus

, Jockel

, Klopp

, Wichmann

, Pankratz

, Sando

, Aasly

, Bar-cikowska

, Wszolek

, Dickson

, Graff-Radford

, Petersen RC; Alzheimer’s Disease Neuroimaging Initiative, van Duijn

, Breteler

, Ikram

, DeStefano

, Fitzpatrick

, Lopez

, Launer

, Seshadri S; CHARGE consortium, Berr

, Campion

, Epelbaum

, Dartigues

, Tzourio

, Alperovitch

, Lathrop M; EADI1 consortium, Feulner

, Friedrich

, Riehle

, Krawczak

, Schreiber

, Mayhaus

, Nicolhaus

, Wagenpfeil

, Steinberg

, Stefansson

, Snaedal

, Bjornsson

, Jons-son

, Chouraki

, Genier-Boley

, Hiltunen

, Soininen

, Combarros

, Zelenika

, Delepine

, Bullido

, Pasquier

, Mateo

, Frank-Garcia

, Porcellini

, Hanon

, Coto

, Alvarez

, Bosco

, Siciliano

, Mancuso

, Panza

, Solfrizzi

, Nacmias

, Sorbi

, Bossü P, Piccardi

, Aro-sioB, Annoni

, SeripaD, Pilotto

, ScarpiniE, Galimberti D, Brice A, Hannequin

, Licastro

, Jones

, Holmans

, Jonsson

, Riemenschneider

, Morgan

, Younkin

, Owen

, O'Donovan M, Amouyel

, Williams

(2011) Common variants at ABCA7, MS4A6A/MS4A4E, EPHA1, CD33 and CD2AP are associated with Alzheimer’s disease. Nat Genet 43, 429–435.

Naj

, Jun

, Beecham

, Wang

, Vardarajan

, Buros

, Gallins

, Buxbaum

, Jarvik

, Crane

, Larson

, Bird

, Boeve

, Graff-Radford

, De Jager

, Evans

, Schneider

, Carrasquillo

, Ertekin-Taner

, Younkin

, Cruchaga

, Kauwe

, Nowotny

, Kramer

, Hardy

, Huentelman

, Myers

, Barmada

, Demirci

, Baldwin

, Green

, Rogaeva

, St George-Hyslop P, Arnold

, Barber

, Beach

, Bigio

, Bowen

, Boxer

, Burke

, Cairns

, Carlson

, Carney

, Carroll

, Chui

, Clark

, Corneveaux

, Cotman

, Cummings

, DeCarli

, DeKosky

, Diaz-Arrastia

, Dick

, Dickson

, Ellis

, Faber

, Fallon

, Farlow

, Ferris

, Frosch

, Galasko

, Ganguli

, Gearing

, Geschwind

, Ghetti

, Gilbert

, Gilman

, Giordani

, Glass

, Growdon

, Hamilton

, Harrell

, Head

, Honig

, Hulette

, Hyman

, Jicha

, Jin

, Johnson

, Karlawish

, Karydas

, Kaye

, Kim

, Koo

, Kowall

, Lah

, Levey

, Lieberman

, Lopez

, Mack

, Marson

, Martiniuk

, Mash

, Masliah

, McCormick

, McCurry

, McDavid

, McKee

, Mesulam

, Miller

, Parisi

, Perl

, Peskind

, Petersen

, Poon

, Quinn

, Rajbhandary

, Raskind

, Reisberg

, Ringman

, Roberson

, Rosenberg

, Sano

, Schneider

, Seeley

, Shelanski

, Slifer

, Smith

, Sonnen

, Spina

, Stern

, Tanzi

, Trojanowski

, Troncoso

, Van Deerlin

, Vinters

, Vonsattel

, Weintraub

, Welsh-Bohmer

, Williamson

, Woltjer

, Cantwell

, Dombroski

, Beekly

, Lunetta

, Martin

, Kamboh

, Saykin

, Reiman

, Bennett

, Morris

, Montine

, Goate

, Blacker

, Tsuang

, Hakonarson

, Kukull

, Foroud

, Haines

, Mayeux

, Pericak-Vance

, Farrer

, Schellenberg

(2011) Common variants at MS4A4/MS4A6E, CD2AP, CD33 and EPHA1 are associated with late-onset Alzheimer’s disease. Nat Genet 43, 436–441.

Harold

, Abraham

, Hollingworth

, Sims

, Gerrish

, Hamshere

, Pahwa

, Moskvina

, Dowzell

, Williams

, Jones

, Thomas

, Stretton

, Morgan

, Lovestone

, Powell

, Proitsi

, Lupton

, Brayne

, Rubinsztein

, Gill

, Lawlor

, Lynch

, Morgan

, Brown

, Passmore

, Craig

, McGuinness

, Todd

, Holmes

, Mann

, Smith

, Love

, Kehoe

, Hardy

, Mead

, Fox

, Rossor

, Collinge

, Maier

, Jessen

, Schurmann

, Heun

, van den Bussche

, Heuser

, Korn-huber

, Wiltfang

, Dichgans

, Frolich

, Hampel

, Hull

, Rujescu

, Goate

, Kauwe

, Cruchaga

, Nowotny

, Morris

, Mayo

, Sleegers

, Bettens

, Engelborghs

, De Deyn

, Van Broeckhoven

, Livingston

, Bass

, Gurling

, McQuillin

, Gwilliam

, Deloukas

, Al-Chalabi

, Shaw

, Tsolaki

, Singleton

, Guerreiro

, Muhleisen

, Nothen

, Moebus

, Jockel

, Klopp

, Wichmann

, Carrasquillo

, Pankratz

, Younkin

, Holmans

, O'Donovan M, Owen

, Williams

(2009) Genome-wide association study identifies variants at CLU and PICALM associated with Alzheimer’s disease. Nat Genet 41, 1088–1093.

Sims

, Williams

(2016) Defining the genetic architecture of Alzheimer’s disease: Where next. Neurodegener Dis 16, 6–11.

10.

Bressler

, Mosley

, Penman

, Gottesman

, Windham

, Knopman

, Wruck

, Boerwinkle

(2017) Genetic variants associated with risk of Alzheimer’s disease contribute to cognitive change in midlife: The Atherosclerosis Risk in Communities Study. Am J Med Genet B Neuropsychiatr Genet 174, 269–282.

11.

Gui

, Jiang

, Cherny

, Sham

, Xu

, Liu

, Jin

, Zhu

, Zhang

, Thomas

, Cheng

, Lam

(2014) Influence of Alzheimer’s disease genes on cognitive decline: The Guangzhou Biobank Cohort Study. Neurobiol Aging 35, 2422e2423–2428.

12.

Harrison

, Bookheimer

(2016) Neuroimaging genetic risk for Alzheimer’s disease in preclinical individuals: From candidate genes to polygenic approaches. Biol Psychiatry Cogn Neurosci Neuroimaging 1, 14–23.

13.

Sleegers

, Bettens

, De Roeck

, Van Cauwenberghe

, Cuyvers

, Verheijen

, Struyfs

, Van Dongen

, Vermeulen

, Engelborghs

, Vandenbulcke

, Vanden-berghe

, De Deyn

, Van Broeckhoven

, consortium

(2015) A 22-single nucleotide polymorphism Alzheimer’s disease risk score correlates with family history, onset age, and cerebrospinal fluid Abeta42. Alzheimers Dement 11, 1452–1460.

14.

Mormino

, Sperling

, Holmes

, Buckner

, De Jager

, Smoller

, Sabuncu

, Alzheimer’s Disease Neuroimaging Initiative (2016) Polygenic risk of Alzheimer disease is associated with early- and late-life processes. Neurology 87, 481–488.

15.

Andrews

, Das

, Cherbuin

, Anstey

, Easteal

(2016) Association of genetic risk factors with cognitive decline: The PATH through life project. Neurobiol Aging 41, 150–158.

16.

Adams

, de Bruijn

, Hofman

, Uitterlinden

, van Duijn

, Vernooij

, Koudstaal

, Ikram

(2015) Genetic risk of neurodegenerative diseases is associated with mild cognitive impairment and conversion to dementia. Alzheimers Dement 11, 1277–1285.

17.

Carrasquillo

, Crook

, Pedraza

, Thomas

, Pankratz

, Allen

, Nguyen

, Malphrus

, Ma

, Bisceglio

, Roberts

, Lucas

, Smith

, Ivnik

, Machulda

, Graff-Radford

, Petersen

, Younkin

, Ertekin-Taner

(2015) Late-onset Alzheimer’s risk variants in memory decline, incident mild cognitive impairment, and Alzheimer’s disease. Neurobiol Aging 36, 60–67.

18.

Chouraki

, Reitz

, Maury

, Bis

, Bellenguez

, Yu

, Jakobsdottir

, Mukherjee

, Adams

, Choi

, Larson

, Fitzpatrick

, Uitterlinden

, de Jager

, Hofman

, Gudnason

, Vardarajan

, Ibrahim-Verbaas

, van der Lee

, Lopez

, Dartigues

, Berr

, Amouyel

, Bennett

, van Duijn

, DeStefano

, Launer

, Ikram

, Crane

, Lambert

, Mayeux

, Seshadri

, International Genomics of Alzheimer’s Project (2016) Evaluation of a genetic risk score to improve risk prediction for Alzheimer’s disease. J Alzheimers Dis 53, 921–932.

19.

Foley

, Tansey

, Caseras

, Lancaster

, Bracht

, Parker

, Hall

, Williams

, Linden

(2017) Multimodal brain imaging reveals structural differences in Alzheimer’s disease polygenic risk carriers: A study in healthy young adults. Biol Psychiatry 81, 154–161.

20.

Habes

, Janowitz

, Erus

, Toledo

, Resnick

, Doshi

, van der Auwera

, Wittfeld

, Hegenscheid

, Hosten

, Biffar

, Homuth

, Volzke

, Grabe

, Hoffmann

, Davatzikos

(2016) Advanced brain aging: Relationship with epidemiologic and genetic risk factors, and overlap with Alzheimer disease atrophy patterns. Transl Psychiatry 6, e775.

21.

Harrison

, Mahmood

, Lau

, Karacozoff

, Burggren

, Small

, Bookheimer

(2016) An Alzheimer’s disease genetic risk score predicts longitudinal thinning of hippocampal complex subregions in healthy older adults. eNeuro 3, ENEURO.0098-16.2016.

22.

Lacour

, Espinosa

, Louwersheimer

, Heilmann

, Hernandez

, Wolfsgruber

, Fernandez

, Wagner

, Rosende-Roca

, Mauleon

, Moreno-Grau

, Vargas

, Pijnenburg

, Koene

, Rodriguez-Gomez

, Ortega

, Ruiz

, Holstege

, Sotolongo-Grau

, Kornhuber

, Peters

, Frolich

, Hull

, Ruther

, Wiltfang

, Scherer

, Riedel-Heller

, Alegret

, Nothen

, Scheltens

, Wagner

, Tarraga

, Jessen

, Boada

, Maier

, van der Flier

, Becker

, Ramirez

, Ruiz

(2017) Genome-wide significant riskfactors for Alzheimer’s disease: Role in progression to dementia due to Alzheimer’s disease among subjects with mild cognitive impairment. Mol Psychiatry 22, 153–160.

23.

Louwersheimer

, Wolfsgruber

, Espinosa

, Lacour

, Heilmann-Heimbach

, Alegret

, Hernandez

, Rosende-Roca

, Tarraga

, Boada

, Kornhuber

, Peters

, Frolich

, Hull

, Ruther

, Wiltfang

, Scherer

, Riedel-Heller

, Jessen

, Nothen

, Maier

, Koene

, Scheltens

, Holstege

, Wagner

, Ruiz

, van der Flier

, Becker

, Ramirez

(2016) Alzheimer’s disease risk variants modulate endophenotypes in mild cognitive impairment. Alzheimers Dement 12, 872–881.

24.

Lupton

, Strike

, Hansell

, Wen

, Mather

, Armstrong

, Thalamuthu

, McMahon

, de Zubicaray

, Assareh

, Simmons

, Proitsi

, Powell

, Montgomery

, Hibar

, Westman

, Tsolaki

, Kloszewska

, Soininen

, Mecocci

, Velas

, Lovestone

, Alzheimer’s Disease Neuroimaging I, Brodaty

, Ames

, Trollor

, Martin

, Thompson

, Sachdev

, Wright

(2016) The effect of increased genetic risk for Alzheimer’s disease on hippocampal and amygdala volume. Neurobiol Aging 40, 68–77.

25.

Martiskainen

, Helisalmi

, Viswanathan

, Kurki

, Hall

, Herukka

, Sarajarvi

, Natunen

, Kurki-nen

, Huovinen

, Makinen

, Laitinen

, Koivisto

, Mattila

, Lehtimaki

, Remes

, Leinonen

, Haapasalo

, Soininen

, Hiltunen

(2015) Effects of Alzheimer’s disease-associated risk loci on cerebrospinal fluid biomarkers and disease progression: A polygenic risk score approach. J Alzheimers Dis 43, 565–573.

26.

Rodriguez-Rodriguez

, Sanchez-Juan

, Vazquez-Higuera

, Mateo

, Pozueta

, Berciano

, Cervantes

, Alcolea

, Martinez-Lage

, Clarimon

, Lleo

, Pastor

, Combarros

(2013) Genetic risk score predicting accelerated progression from mild cognitive impairment to Alzheimer’s disease. J Neural Transm (Vienna) 120, 807–812.

27.

Sabuncu

, Buckner

, Smoller

, Lee

, Fischl

, Sperling

,Alzheimer’s Disease Neuroimaging Initiative (2012) The association between a polygenic Alzheimer score and cortical thickness in clinically normal subjects. Cereb Cortex 22, 2653–2661.

28.

Darst

, Koscik

, Racine

, Oh

, Krause

, Carlsson

, Zetterberg

, Blennow

, Christian

, Bendlin

, Okonkwo

, Hogan

, Hermann

, Sager

, Asthana

, Johnson

, Engelman

(2017) Pathway-specific polygenic risk scores as predictors of amyloid-beta deposition and cognitive function in a sample at increased risk for Alzheimer’s disease. J Alzheimers Dis 55, 473–484.

29.

Biffi

, Anderson

, Desikan

, Sabuncu

, Cortellini

, Schmansky

, Salat

, Rosand

, Alzheimer’s Disease Neuroimaging Initiative (2010) Genetic variation and neuroimaging measures in Alzheimer disease. Arch Neurol 67, 677–685.

30.

Chauhan

, Adams

, Bis

, Weinstein

, Yu

, Togl-hofer

, Smith

, van der Lee

, Gottesman

, Thomson

, Wang

, Yang

, Niessen

, Lopez

, Becker

, Phan

, Beare

, Arfanakis

, Fleischman

, Vernooij

, Mazoyer

, Schmidt

, Srikanth

, Knopman

, Jack

Jr , Amouyel

, Hofman

, DeCarli

, Tzourio

, van Duijn

, Bennett

, Schmidt

, Longstreth

Jr , Mosley

, Fornage

, Launer

, Seshadri

, Ikram

, Debette

(2015) Association of Alzheimer’s disease GWAS loci with MRI markers of brain aging. Neurobiol Aging 36, 1765e1767–1716.

31.

Marden

, Mayeda

, Walter

, Vivot

, Tchetgen Tchetgen

, Kawachi

, Glymour

(2016) Using an Alzheimer disease polygenic risk score to predict memory decline in black and white americans over 14 years of follow-up. Alzheimer Dis Assoc Disord 30, 195–202.

32.

Escott-Price

, Myers

, Huentelman

, Hardy

(2017) Polygenic risk score analysis of pathologically confirmed alzheimer disease. Ann Neurol 82, 311–314.

33.

Escott-Price

, Sims

, Bannister

, Harold

, Vronskaya

, Majounie

, Badarinarayan

Gerad

Perades , consortia

, Morgan

, Passmore

, Holmes

, Powell

, Brayne

, Gill

, Mead

, Goate

, Cruchaga

, Lambert

, van Duijn

, Maier

, Ramirez

, Holmans

, Jones

, Hardy

, Seshadri

, Schellenberg

, Amouyel

, Williams

(2015) Common polygenic variation enhances risk prediction for Alzheimer’s disease. Brain 138, 3673–3684.

34.

Voyle

, Patel

, Folarin

, Newhouse

, Johnston

, Visser

, Dobson

, Kiddle

, EDAR and DESCRIPA study groups and the Alzheimer’s Disease Neuroimaging Initiative (2017) Genetic risk as a marker of amyloid-beta and tau burden in cerebrospinal fluid. J Alzheimers Dis 55, 1417–1427.

35.

Desikan

, Fan

, Wang

, Schork

, Cabral

, Cup-ples

, Thompson

, Besser

, Kukull

, Holland

, Chen

, Brewer

, Karow

, Kauppi

, Witoe-lar

, Karch

, Bonham

, Yokoyama

, Rosen

, Miller

, Dillon

, Wilson

, Hess

, Pericak-Vance

, Haines

, Farrer

, Mayeux

, Hardy

, Goate

, Hyman

, Schellenberg

, McEvoy

, Andreassen

, Dale

(2017) Genetic assessment of age-associated Alzheimer disease risk: Development and validation of a polygenic hazard score. PLoS Med 14, e1002258.

36.

Villemagne

, Pike

, Chetelat

, Ellis

, Mulligan

, Bourgeat

, Ackermann

, Jones

, Szoeke

, Salvado

, Martins

, O'Keefe

, Mathis

, Klunk

, Ames

, Masters

, Rowe

(2011) Longitudinal assessment of Abeta and cognition in aging and Alzheimer disease. Ann Neurol 69, 181–192.

37.

Burnham

, Bourgeat

, Dore

, Savage

, Brown

, Laws

, Maruff

, Salvado

, Ames

, Martins

, Masters

, Rowe

, Villemagne

, AIBL Research Group (2016) Clinical and cognitive trajectories in cognitively healthy elderly individuals with suspected non-Alzheimer’s disease pathophysiology (SNAP) or Alzheimer’s disease pathology: A longitudinal study. Lancet Neurol 15, 1044–1053.

38.

Villemagne

, Burnham

, Bourgeat

, Brown

, Ellis

, Salvado

, Szoeke

, Macaulay

, Martins

, Maruff

, Ames

, Rowe

, Masters

, Australian Imaging Biomarkers and Lifestyle (AIBL) Research Group (2013) Amyloid beta deposition, neurodegeneration, and cognitive decline in sporadic Alzheimer’s disease: A prospective cohort study. Lancet Neurol 12, 357–367.

39.

Sperling

, Mormino

, Johnson

(2014) The evolution of preclinical Alzheimer’s disease: Implications for prevention trials. Neuron 84, 608–622.

40.

Ellis

, Bush

, Darby

, De Fazio

, Foster

, Hudson

, Lautenschlager

, Lenzo

, Martins

, Maruff

, Masters

, Milner

, Pike

, Rowe

, Savage

, Szoeke

, Taddei

, Villemagne

, Woodward

, Ames

, AIBL Research Group (2009) 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. Int Psychogeriatr 21, 672–687.

41.

Winblad

, Palmer

, Kivipelto

, Jelic

, Fratiglioni

, Wahlund

, Nordberg

, Backman

, Albert

, Almkvist

, Arai

, Basun

, Blennow

, de Leon

, DeCarli

, Erkinjuntti

, Giacobini

, Graff

, Hardy

, Jack

, Jorm

, Ritchie

, van Duijn

, Visser

, Petersen

(2004) Mild cognitive impairment-beyond controversies, towards a consensus: Report of the International Working Group on Mild Cognitive Impairment. J Intern Med 256, 240–246.

42.

McKhann

, Drachman

, Folstein

, Katzman

, Price

, Stadlan

(1984) Clinical diagnosis of Alzheimer’s disease: Report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services TaskForce on Alzheimer’s Disease. Neurology 34,939–944.

43.

Burnham

, Raghavan

, Wilson

, Baker

, Ropacki

, Novak

, Ames

, Ellis

, Martins

, Maruff

, Masters

, Romano

, Rowe

, Savage

, Macaulay

, Narayan

, Alzheimer’s Disease Neuroimaging Initiative; AIBL Research Group (2015) Novel statistically-derived composite measures for assessing the efficacy of disease-modifying therapies in prodromal Alzheimer’s disease trials: An AIBL study. J Alzheimers Dis 46, 1079–1089.

44.

Donohue

, Sperling

, Salmon

, Rentz

, Raman

, Thomas

, Weiner

, Aisen

, Australian Imaging, Biomarkers, and Lifestyle Flagship Study of Ageing; Alzheimer’s Disease Neuroimaging Initiative; Alzheimer’s Disease Cooperative Study (2014) The preclinical Alzheimer cognitive composite: Measuring amyloid-related decline. JAMA Neurol 71, 961–970.

45.

Burnham

, Raghavan

, Wilson

, Baker

, Ropacki

, Novak

, Ames

, Ellis

, Martins

, Maruff

, Masters

, Romano

, Rowe

, Savage

, Twyman

, Macaulay

, Narayan

, The AIBL Research Group (2014) Comparison of three normative data correction approaches: A cross-sectional evaluation in the AIBL study. Alzheimers Dement 10, P4–293.

46.

Rowe

, Ellis

, Rimajova

, Bourgeat

, Pike

, Jones

, Fripp

, Tochon-Danguy

, Morandeau

, O'Keefe

, Price

, Raniga

, Robins

, Acosta

, Lenzo

, Szoeke

, Salvado

, Head

, Martins

, Masters

, Ames

, Villemagne

(2010) Amyloid imaging results from the Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging. Neurobiol Aging 31, 1275–1283.

47.

Clark

, Schneider

, Bedell

, Beach

, Bilker

, Mintun

, Pontecorvo

, Hefti

, Carpenter

, Flitter

, Krautkramer

, Kung

, Coleman

, Doraiswamy

, Fleisher

, Sabbagh

, Sadowsky

, Reiman

, Zehntner

, Skovronsky

, AV45-A07 Study Group (2011) Use of florbetapir-PET for imaging beta-amyloid pathology. JAMA 305, 275–283.

48.

Vandenberghe

, Van Laere

, Ivanoiu

, Salmon

, Bastin

, Triau

, Hasselbalch

, Law

, Andersen

, Korner

, Minthon

, Garraux

, Nelissen

, Bormans

, Buckley

, Owenius

, Thurfjell

, Farrar

, Brooks

(2010) 18F-flutemetamol amyloid imaging in Alzheimer disease and mild cognitive impairment: A phase 2 trial. Ann Neurol 68, 319–329.

49.

Bourgeat

, Villemagne

, Dore

, Brown

, Macaulay

, Martins

, Masters

, Ames

, Ellis

, Rowe

, Salvado

, Fripp

, AIBL Research Group (2015) Comparison of MR-less PiB SUVR quantification methods. Neurobiol Aging 36(Suppl 1), S159–166.

50.

Villemagne

, Dore

, Yates

, Brown

, Mulligan

, Bourgeat

, Veljanoski

, Rainey-Smith

, Ong

, Rem-bach

, Williams

, Burnham

, Laws

, Salvado

, Taddei

, Macaulay

, Martins

, Ames

, Masters

, Rowe

(2014) En attendant centiloid. Adv Res 2,723–729.

51.

Rowe

, Bourgeat

, Ellis

, Brown

, Lim

, Mulligan

, Jones

, Maruff

, Woodward

, Price

, Robins

, Tochon-Danguy

, O'Keefe

, Pike

, Yates

, Szoeke

, Salvado

, Macaulay

, O'Meara

, Head

, Cobiac

, Savage

, Martins

, Masters

, Ames

, Villemagne

(2013) Predicting Alzheimer disease with beta-amyloidimaging: Results from the Australian imaging, biomarkers, and lifestyle study of ageing. Ann Neurol 74, 905–913.

52.

, Villemagne

, Doecke

, Rembach

, Sarros

, Varghese

, McGlade

, Laughton

, Pertile

, Fowler

, Rumble

, Trounson

, Taddei

, Rainey-Smith

, Laws

, Robertson

, Evered

, Silbert

, Ellis

, Rowe

, Macaulay

, Darby

, Martins

, Ames

, Masters

, Collins

, AIBL Research Group (2015) Alzheimer’s disease normative cerebrospinal fluid biomarkers validated in PET amyloid-beta characterized subjects from the Australian Imaging, Biomarkers and Lifestyle (AIBL) study. J Alzheimers Dis 48, 175–187.

53.

Vanderstichele

, Bibl

, Engelborghs

, Le Bastard

, Lewczuk

, Molinuevo

, Parnetti

, Perret-Liaudet

, Shaw

, Teunissen

, Wouters

, Blennow

(2012) Standardization of preanalytical aspects of cerebrospinal fluid biomarker testing for Alzheimer’s disease diagnosis: A consensus paper from the Alzheimer’s Biomarkers Standardization Initiative. Alzheimers Dement 8, 65–73.

54.

Porter

, Burnham

, Dore

, Savage

, Bourgeat

, Bege-mann

, Milicic

, Ames

, Bush

, Maruff

, Masters

, Rowe

, Rainey-Smith

, Martins

, Groth

, Verdile

, Villemagne

, Laws

(2018) KIBRA is associated with accelerated cognitive decline and hippocampal atrophy in APOE epsilon4-positive cognitively normal adults with high Abeta-amyloid burden. Sci Rep 8, 2034.

55.

Guerreiro

, Wojtas

, Bras

, Carrasquillo

, Rogaeva

, Majounie

, Cruchaga

, Sassi

, Kauwe

, Younkin

, Hazrati

, Collinge

, Pocock

, Lashley

, Williams

, Lambert

, Amouyel

, Goate

, Rademakers

, Morgan

, Powell

, St George-Hyslop

, Singleton

, Hardy

, Alzheimer Genetic Analysis Group (2013) TREM2 variants in Alzheimer’s disease. N Engl J Med 368, 117–127.

56.

RStudio Team (2015) RStudio: Integrated Development for R. RStudio I, Boston, MA.

57.

Derby

, Burns

, Wang

, Katz

, Zimmerman

, L'Italien

, Guo

, Berman

, Lipton

(2013) Screening for predementia AD: Time-dependent operating characteristics of episodic memory tests. Neurology 80, 1307–1314.

58.

Del-Aguila

, Fernandez

, Schindler

, Ibanez

, Dem-ing

, Ma

, Saef

, Black

, Budde

, Norton

, Chasse

, Alzheimer’s Disease Neuroimaging Initiative, Harari

, Goate

, Xiong

, Morris

, Cruchaga

(2018) Assessment of the genetic architecture of Alzheimer’s disease risk in rate of memory decline. J Alzheimers Dis 62, 745–756.

59.

Lim

, Villemagne

, Laws

, Ames

, Pietrzak

, Ellis

, Harrington

, Bourgeat

, Salvado

, Darby

, Snyder

, Bush

, Martins

, Masters

, Rowe

, Nathan

, Maruff

, Australian

Imaging

, Biomark-ers and Lifestyle (AIBL) Research Group(2013) BDNF Val66Met, Abeta amyloid, and cognitive decline in preclin-ical Alzheimer’s disease. Neurobiol Aging 34, 2457–2464.

60.

Lim

, Villemagne

, Laws

, Pietrzak

, Snyder

, Ames

, Ellis

, Harrington

, Rembach

, Martins

, Rowe

, Masters

, Maruff

(2015) APOE and BDNF polymorphisms moderate amyloid beta-related cognitive decline in preclinical Alzheimer’s disease. Mol Psychiatry 20, 1322–1328.

61.

Chung

, Kim

, You

, Jang

, Kim

, Lee

(2014) CR1, ABCA7, and APOE genes affect the features of cognitive impairment in Alzheimer’s disease. J Neurol Sci 339, 91–96.

62.

Pedraza

, Allen

, Jennette

, Carrasquillo

, Crook

, Serie

, Pankratz

, Palusak

, Nguyen

, Malphrus

, Ma

, Bisceglio

, Roberts

, Lucas

, Ivnik

, Smith

, Graff-Radford

, Petersen

, Younkin

, Ertekin-Taner

(2014) Evaluation of memory endophenotypes for association with CLU, CR1, and PICALM variants in black and white subjects. Alzheimers Dement 10, 205–213.

63.

Greenbaum

, Ravona-Springer

, Lubitz

, Schmeidler

, Cooper

, Sano

, Silverman

, Heymann

, Beeri

(2016) Potential contribution of the Alzheimer’s disease risk locus BIN1 to episodic memory performance in cognitively normal Type 2 diabetes elderly. Eur Neuropsychopharmacol 26, 787–795.

64.

Liu

, Yu

, Wang

, Hao

, Yang

, Jiang

, Zhu

, Cao

, Zhang

, Tan

(2014) Association between NME8 locus polymorphism and cognitive decline, cerebrospinal fluid and neuroimaging biomarkers in Alzheimer’s disease. PLoS One 9, e114777.