[Skip to Content]
Sign In
Individual Sign In
Create an Account
Institutional Sign In
OpenAthens Shibboleth
[Skip to Content Landing]
Figure 1.
Progression During the Follow-up Period
Progression During the Follow-up Period

A, Progression from unimpaired status to amnestic mild cognitive impairment (aMCI) in amyloid-positive vs amyloid-negative participants. B, Progression from mild cognitive impairment (MCI) to Alzheimer disease (AD) dementia in amyloid-positive vs amyloid-negative participants. C, Progression from aMCI to AD dementia in amyloid-positive vs amyloid-negative participants. D, Progression from no cognitive impairment or aMCI to AD dementia in amyloid-positive vs amyloid-negative participants. Models were adjusted for age, sex, and education using time from baseline as the time scale. At 3 years, the percentage free of aMCI was 88.0% of amyloid-positive participants who were initially cognitively unimpaired vs 96.2% of amyloid-negative participants who were initially cognitively unimpaired; the percentage free of AD dementia was 73.9% of amyloid-positive people who initially had MCI vs 90.8% of amyloid-negative people with the same condition; the corresponding rates for aMCI were 68.3% vs 90.8%, respectively, after censoring.

Figure 2.
Changes in Global Cognitive Domain and Memory Domain z Scores
Changes in Global Cognitive Domain and Memory Domain z Scores

A and B, Slopes for global cognitive domain z scores and memory domain z scores for participants who were amyloid positive and cognitively unimpaired; amyloid negative and cognitively unimpaired; amyloid positive with MCI; and amyloid negative with MCI. All slopes were centered on age, sex, and education. Slopes for all groups declined significantly except the global z score for participants who were amyloid negative and cognitively unimpaired and the memory z score for amyloid negative participants with MCI; the memory z score for amyloid-negative participants without cognitive impairment increased significantly.

Table 1.  
Characteristics of Participants at Baseline and During Follow-upa
Characteristics of Participants at Baseline and During Follow-upa
Table 2.  
Prevalence of Imaging Biomarkers by Cognitive Status, 10-Year Age Groups, and Sex
Prevalence of Imaging Biomarkers by Cognitive Status, 10-Year Age Groups, and Sex
Table 3.  
Risk of Progression to Incident Amnestic MCI and AD Dementia in Amyloid-Positive Individuals (Compared With Amyloid-Negative Individuals)
Risk of Progression to Incident Amnestic MCI and AD Dementia in Amyloid-Positive Individuals (Compared With Amyloid-Negative Individuals)
1.
Dubois  B, Feldman  HH, Jacova  C,  et al.  Research criteria for the diagnosis of Alzheimer’s disease: revising the NINCDS-ADRDA criteria.  Lancet Neurol. 2007;6(8):734-746.PubMedGoogle ScholarCrossref
2.
Dubois  B, Hampel  H, Feldman  HH,  et al; Proceedings of the Meeting of the International Working Group (IWG) and the American Alzheimer’s Association on “The Preclinical State of AD”; July 23, 2015; Washington DC, USA.  Preclinical Alzheimer’s disease: Definition, natural history, and diagnostic criteria.  Alzheimers Dement. 2016;12(3):292-323.PubMedGoogle ScholarCrossref
3.
Sperling  RA, Aisen  PS, Beckett  LA,  et al.  Toward defining the preclinical stages of Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease.  Alzheimers Dement. 2011;7(3):280-292.PubMedGoogle ScholarCrossref
4.
Albert  MS, DeKosky  ST, Dickson  D,  et al.  The diagnosis of mild cognitive impairment due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease.  Alzheimers Dement. 2011;7(3):270-279.PubMedGoogle ScholarCrossref
5.
Knopman  DS, Jack  CR  Jr, Wiste  HJ,  et al.  Short-term clinical outcomes for stages of NIA-AA preclinical Alzheimer disease.  Neurology. 2012;78(20):1576-1582.PubMedGoogle ScholarCrossref
6.
Petersen  RC, Aisen  P, Boeve  BF,  et al.  Mild cognitive impairment due to Alzheimer disease in the community.  Ann Neurol. 2013;74(2):199-208.PubMedGoogle Scholar
7.
Jack  CR  Jr, Wiste  HJ, Weigand  SD,  et al.  Age-specific population frequencies of cerebral β-amyloidosis and neurodegeneration among people with normal cognitive function aged 50-89 years: a cross-sectional study.  Lancet Neurol. 2014;13(10):997-1005.PubMedGoogle ScholarCrossref
8.
Jansen  WJ, Ossenkoppele  R, Knol  DL,  et al; Amyloid Biomarker Study Group.  Prevalence of cerebral amyloid pathology in persons without dementia: a meta-analysis.  JAMA. 2015;313(19):1924-1938.PubMedGoogle ScholarCrossref
9.
Ossenkoppele  R, Jansen  WJ, Rabinovici  GD,  et al; Amyloid PET Study Group.  Prevalence of amyloid PET positivity in dementia syndromes: a meta-analysis.  JAMA. 2015;313(19):1939-1949.PubMedGoogle ScholarCrossref
10.
Petersen  RC, Wiste  HJ, Weigand  SD,  et al.  Association of elevated amyloid levels with cognition and biomarkers in cognitively normal people from the community.  JAMA Neurol. 2016;73(1):85-92.PubMedGoogle ScholarCrossref
11.
van Harten  AC, Smits  LL, Teunissen  CE,  et al.  Preclinical AD predicts decline in memory and executive functions in subjective complaints.  Neurology. 2013;81(16):1409-1416.PubMedGoogle ScholarCrossref
12.
Hansson  O, Zetterberg  H, Buchhave  P, Londos  E, Blennow  K, Minthon  L.  Association between CSF biomarkers and incipient Alzheimer’s disease in patients with mild cognitive impairment: a follow-up study.  Lancet Neurol. 2006;5(3):228-234.PubMedGoogle ScholarCrossref
13.
Landau  SM, Mintun  MA, Joshi  AD,  et al; Alzheimer’s Disease Neuroimaging Initiative.  Amyloid deposition, hypometabolism, and longitudinal cognitive decline.  Ann Neurol. 2012;72(4):578-586.PubMedGoogle ScholarCrossref
14.
Petersen  RC, Aisen  PS, Beckett  LA,  et al.  Alzheimer’s Disease Neuroimaging Initiative (ADNI): clinical characterization.  Neurology. 2010;74(3):201-209.PubMedGoogle ScholarCrossref
15.
Roberts  RO, Geda  YE, Knopman  DS,  et al.  The Mayo Clinic Study of Aging: design and sampling, participation, baseline measures and sample characteristics.  Neuroepidemiology. 2008;30(1):58-69.PubMedGoogle ScholarCrossref
16.
Petersen  RC, Roberts  RO, Knopman  DS,  et al; The Mayo Clinic Study of Aging.  Prevalence of mild cognitive impairment is higher in men.  Neurology. 2010;75(10):889-897.PubMedGoogle ScholarCrossref
17.
St Sauver  JL, Grossardt  BR, Yawn  BP,  et al.  Data resource profile: the Rochester Epidemiology Project (REP) medical records-linkage system.  Int J Epidemiol. 2012;41(6):1614-1624.PubMedGoogle ScholarCrossref
18.
Ivnik  RJ, Malec  JF, Smith  GE,  et al.  Mayo’s older Americans normative studies: WAIS-R norms for ages 56 to 97.  Clin Neuropsychol. 1992.Google Scholar
19.
Petersen  RC.  Mild cognitive impairment as a diagnostic entity.  J Intern Med. 2004;256(3):183-194.PubMedGoogle ScholarCrossref
20.
American Psychiatric Association.  Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994.
21.
McKhann  G, Drachman  D, Folstein  M, Katzman  R, Price  D, Stadlan  EM.  Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease.  Neurology. 1984;34(7):939-944.PubMedGoogle ScholarCrossref
22.
McKhann  GM, Knopman  DS, Chertkow  H,  et al.  The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease.  Alzheimers Dement. 2011;7(3):263-269.PubMedGoogle ScholarCrossref
23.
Jack  CR  Jr, Wiste  HJ, Knopman  DS,  et al.  Rates of β-amyloid accumulation are independent of hippocampal neurodegeneration.  Neurology. 2014;82(18):1605-1612.PubMedGoogle ScholarCrossref
24.
Senjem  ML, Gunter  JL, Shiung  MM, Petersen  RC, Jack  CR  Jr.  Comparison of different methodological implementations of voxel-based morphometry in neurodegenerative disease.  Neuroimage. 2005;26(2):600-608.PubMedGoogle ScholarCrossref
25.
Jack  CR  Jr, Bennett  DA, Blennow  K,  et al.  A/T/N: An unbiased descriptive classification scheme for Alzheimer disease biomarkers.  Neurology. 2016;87(5):539-547.PubMedGoogle ScholarCrossref
26.
Jack  CR  Jr, Wiste  HJ, Weigand  SD,  et al.  Defining imaging biomarker cut points for brain aging and Alzheimer’s disease.  Alzheimers Dement. 2017;13(3):205-216.PubMedGoogle ScholarCrossref
27.
Kessler  RC, Little  RJ, Groves  RM.  Advances in strategies for minimizing and adjusting for survey nonresponse.  Epidemiol Rev. 1995;17(1):192-204.PubMedGoogle ScholarCrossref
28.
D’Agostino  RB  Jr, Rubin  DB.  Estimating and using propensity scores with partially missing data.  J Am Stat Assoc. 2000;95(451):749-759.Google ScholarCrossref
29.
Little  RJA.  Survey nonresponse adjustments for estimates of means.  Int Stat Rev. 1986;54:139-157.Google ScholarCrossref
30.
Roberts  RO, Geda  YE, Knopman  DS,  et al.  The incidence of MCI differs by subtype and is higher in men: the Mayo Clinic Study of Aging.  Neurology. 2012;78(5):342-351.PubMedGoogle ScholarCrossref
31.
Nordberg  A, Carter  SF, Rinne  J,  et al.  A European multicentre PET study of fibrillar amyloid in Alzheimer’s disease.  Eur J Nucl Med Mol Imaging. 2013;40(1):104-114.PubMedGoogle ScholarCrossref
32.
Rowe  CC, Bourgeat  P, Ellis  KA,  et al.  Predicting Alzheimer disease with β-amyloid imaging: results from the Australian imaging, biomarkers, and lifestyle study of ageing.  Ann Neurol. 2013;74(6):905-913.PubMedGoogle ScholarCrossref
33.
Vos  SJ, Xiong  C, Visser  PJ,  et al.  Preclinical Alzheimer’s disease and its outcome: a longitudinal cohort study.  Lancet Neurol. 2013;12(10):957-965.PubMedGoogle ScholarCrossref
34.
Donohue  MC, Sperling  RA, Petersen  R, Sun  CK, Weiner  MW, Aisen  PS; Alzheimer’s Disease Neuroimaging Initiative.  Association between elevated brain amyloid and subsequent cognitive decline among cognitively normal persons.  JAMA. 2017;317(22):2305-2316.PubMedGoogle ScholarCrossref
Original Investigation
August 2018

Prevalence and Outcomes of Amyloid Positivity Among Persons Without Dementia in a Longitudinal, Population-Based Setting

Author Affiliations
  • 1Department of Health Sciences Research, Mayo Clinic, Rochester, Minnesota
  • 2Department of Neurology, Mayo Clinic, Rochester, Minnesota
  • 3Department of Psychiatry and Psychology and Neurology, Mayo Clinic, Scottsdale, Arizona
  • 4Department of Psychiatry and Psychology, Mayo Clinic, Rochester, Minnesota
  • 5Real World Data Science, F. Hoffmann-La Roche Ltd, Basel, Switzerland
  • 6Medical Affairs, F. Hoffmann-La Roche Ltd, Basel, Switzerland
  • 7Department of Radiology, Mayo Clinic, Rochester, Minnesota
JAMA Neurol. 2018;75(8):970-979. doi:10.1001/jamaneurol.2018.0629
Key Points

Question  What are clinical outcomes of amyloid positivity in persons without dementia?

Findings  In this study, the risk of amnestic mild cognitive impairment (aMCI) was elevated 2.3-fold in amyloid-positive vs amyloid-negative participants without cognitive impairment. The risk of developing clinically diagnosed Alzheimer disease dementia was elevated 1.9-fold in amyloid-positive participants vs amyloid-negative participants with mild cognitive impairment and 2.6-fold in amyloid-positive vs amyloid-negative participants with no cognitive impairment or amnestic mild cognitive impairment.

Meaning  Our estimates of progression for amyloid-positive, cognitively unimpaired status to amnestic mild cognitive impairment may provide reliable and valid information from a population-based study for estimating the outcome of a successful anti-amyloid therapeutic agent on a general North American population.

Abstract

Importance  Brain amyloid deposition is a marker of Alzheimer disease (AD) pathology. The population-based prevalence and outcomes of amyloid positivity in a population without dementia are important for understanding the trajectory of amyloid positivity to clinically significant outcomes and for designing AD prevention trials.

Objective  To determine prevalence and outcomes of amyloid positivity in a population without dementia.

Design, Setting, and Participants  In the prospective, population-based Mayo Clinic Study of Aging in Olmsted County, Minnesota, participants without dementia were randomly selected from the county population and were clinically and cognitively evaluated at baseline and every 15 months from August 1, 2008, to September 18, 2018. They were also invited to undergo carbon11–Pittburgh compound B positron emission tomography (PET) imaging.

Exposures  Amyloid positivity (defined as a standardized uptake value ratio >1.42 on PET).

Main Outcomes and Measures  Prevalence of amyloid positivity in the Olmsted County population without dementia and risk of progression from no cognitive impairment (ie, normal cognition for age) to incident amnestic MCI (aMCI) and from MCI or aMCI to incident AD dementia.

Results  Of 3894 participants, 1671 underwent PET imaging and were included in the study; 2198 did not undergo imaging, and 25 were excluded for other reasons. The mean (SD) age of participants was 71.3 (9.8) years; 892 (53.4%) were men, and 179 (10.7%) had prevalent MCI. The prevalence of amyloid positivity without cognitive impairment in the population without dementia increased from 2.7% (95% CI, 0.5% to 4.9%) in persons aged 50 to 59 years to 41.3% (95% CI, 33.4% to 49.2%) in those aged 80 to 89 years at baseline. Prevalence of amyloid-positive MCI in the population without dementia increased from 0% in persons aged 50 to 59 years to 16.4% (95% CI, 10.3% to 22.5%) in those aged 80 to 89 years. The incident aMCI risk increased more than 2-fold in participants without cognitive impairment who were amyloid positive vs those who were amyloid negative (hazard ratio [HR], 2.26; 95% CI, 1.52 to 3.35; P < .001). The risk of AD dementia was 1.86 (95% CI, 0.89 to 3.88; P = .10) for amyloid-positive participants with MCI vs amyloid-negative participants with MCI, 1.63 (95% CI, 0.78 to 3.41; P = .20) for participants with aMCI who were amyloid positive vs amyloid negative, and 2.56 (95% CI, 1.35 to 4.88; P = .004) for amyloid-positive participants who were either without cognitive impairment or had aMCI vs those who were amyloid negative. Global cognitive and memory domain z scores declined significantly in amyloid-positive individuals during follow-up. The mean (SD) follow-up time from baseline was 3.7 (1.9) years to incident aMCI and 3.8 (2.0) years to incident AD dementia.

Conclusions and Relevance  Population-based prevalence of amyloid-positive status and progression rates of amyloid positivity provide valid information for designing AD prevention trials and assessing the public health outcomes of AD prevention and interventions.

Introduction

The National Institutes on Aging–Alzheimer’s Association and the International Working Group has established stages of preclinical Alzheimer disease (AD) and defined mild cognitive impairment (MCI) due to AD (and its subset, prodromal AD) with elevated brain amyloid as the key biomarker for AD pathology.1-4 Our group has published frequencies of the National Institute on Aging–Alzheimer’s Association biomarker groups in Mayo Clinic Study of Aging (MCSA) participants who either had no cognitive impairment or had amnestic MCI (aMCI).5-7 Whereas frequencies describe the proportion of individuals in a study who have a condition, prevalence represents the number of individuals in a defined population with a disease or condition. Prevalence of amyloid positivity in the population without dementia provides reliable and valid estimates of the burden of AD pathology, which are useful for sample size estimations in designing primary and secondary AD prevention trials in persons without dementia.

Despite this, there are relatively few population-based estimates of amyloid positivity in the population without dementia. Jansen et al8 reported the prevalence of amyloid positivity in persons without dementia from a meta-analysis of clinic-based, volunteer-based studies in hospital departments and population-based studies. Because of differences in study designs and recruitment strategies, the estimates are potentially prone to bias. Ossenkoppele et al9 estimated amyloid positivity prevalence among persons with dementia, but because participants had dementia prior to the initiation of the study, the estimates cannot be used in estimating sample sizes for primary and secondary prevention AD trials. Our first objective was therefore to estimate the prevalence of amyloid positivity in a population without dementia.

Current AD trials are focused on primary and secondary prevention in persons without dementia. Population-based studies on outcomes of preclinical AD, MCI due to AD, and prodromal AD are informative for estimating effect sizes, rates of progression, duration of trials, and interpreting the efficacy of clinical trial interventions. Our group has published studies on the progression in participants with preclinical AD in a small sample during a very short duration.5,10 Van Harten et al11 examined the predictive role of preclinical AD for cognitive decline in memory clinic participants, and Hansson et al12 assessed progression from biomarker-positive MCI (using tau protein in cerebrospinal fluid or amyloid β) to AD dementia in a small sample of memory clinic patients. Clinic-based studies, however, typically overestimate rates of progression because participants may have more advanced disease than the general population. Studies of cognitive decline have been conducted among Alzheimer’s Disease Neuroimaging Initiative (ADNI) participants.13 However, the ADNI cohort is a highly select convenience sample focused on the symptomatic phase of the disease, and it excludes participants with clinical evidence of cerebrovascular disease.14 Therefore ADNI estimates may not represent the general population without dementia. Thus, the second objective of this study was to estimate outcomes of amyloid positivity in a large, prospective, population-based cohort.

Methods

The MCSA cohort was established in 2004 to identify risk factors for MCI and dementia, and the design has been published in detail.15,16 Briefly, researchers enumerated the Olmsted County, Minnesota, population aged 70 to 89 years on October 1, 2004, using the Rochester Epidemiology Project, a medical records–linkage system.17 Participants were then selected for the study using an age-stratified and sex-stratified random sampling scheme. Exclusion criteria included terminal illness, hospice admission, or dementia. Recruitment of people aged 50 to 69 years began in 2012 to determine the timing of onset of brain pathology and cognitive impairment. Recruitment of persons with dementia began in 2014 to allow potential imaging of these participants.

The institutional review boards of the Mayo Clinic and the Olmsted Medical Center approved all study protocols. Written informed consent was obtained from all participants.

Evaluation of Participants

A study coordinator assessed sociodemographic factors, memory, and the Beck Depression Inventory (BDI) in participants and administered the Clinical Dementia Rating Scale and Functional Activities Questionnaire to a study partner (defined as someone who knew the participant well). A physician assessed global cognition and performed a neurological examination to identify potential causes of cognitive impairment. Participants underwent cognitive testing to assess performance in memory (Wechsler Memory Scale–Revised Logical Memory II and Visual Reproduction II Tests and the Auditory Verbal Learning Test, all of which were implemented with delayed recall); attention-executive function (Trail Making Test Part B, the Digit Symbol Substitution Test from Wechsler Adult Intelligent Scale–Revised); language (Boston Naming Test, category fluency); and visuospatial skills (Block Design and Picture Completion Tests from Wechsler Adult Intelligent Scale–Revised). Raw scores from each test were transformed into age-adjusted scores using normative data, averaged, and scores were scaled to compute domain z scores.18 In addition, the raw test scores at each evaluation were averaged and scaled to create a domain z score to allow comparisons across domains; the 4 domain z scores were averaged and scaled to calculate a global cognitive z score.

Diagnostic Assessment

Mild cognitive impairment was diagnosed by a combination of (1) cognitive concern by participant, study partner, and/or physician, (2) impairment in 1 or more cognitive domain, (3) essentially normal functional activities, and (4) absence of dementia.19 The diagnosis was characterized as amnestic MCI (aMCI) if impairment occurred in the memory domain and nonamnestic if impairment occurred only in 1 or more nonmemory domains. Dementia (including AD dementia) was based on published criteria.20-22 Persons were considered cognitively unimpaired if they did not meet criteria for MCI (whether amnestic or nonamnestic) or dementia.

Other Covariates

Weight and height were measured at each evaluation, and apolipoprotein E (ApoE) ɛ4 status was determined. To obtain information on potential causes of nonparticipation bias at baseline, the medical records of participants and nonparticipants were abstracted for vascular diseases (ie, diabetes, hypertension, dyslipidemia, coronary artery disease, atrial fibrillation, congestive cardiac failure, and stroke).

Prospective Follow-up

Participants were evaluated every 15 months to identify incident MCI and dementia. Trained nurses reviewed the medical records of participants lost to follow-up for incident physician-diagnosed AD dementia.

Imaging

Participants underwent magnetic resonance imaging and a computed tomographic scan for attenuation correction. Amyloid positron emission tomography (PET) imaging was initiated in 2008 and consisted of four 5-minute dynamic frames acquired 40 to 60 minutes after injection of carbon11-Pittsburgh compound B.23 The regions of interest were derived from automatically labeled regions of interest from an MRI template.24,25 An amyloid PET standardized uptake value ratio (SUVR) was calculated from the voxel number weighted average of the median uptake in the parietal, temporal, prefrontal, orbitofrontal, anterior and posterior cingulate, and precuneus regions of interest, referenced to the cerebellar gray crus region. Amyloid positivity was defined as SUVR greater than 1.42.26 The first PET result on or after August 1, 2008, was considered the baseline for each participant.

Statistical Analyses

We compared the characteristics of participants without cognitive impairment who were positive or negative for amyloid and participants with MCI who were positive or negative for amyloid using χ2 (categorical) and Wilcoxon rank sum (continuous) tests. We estimated the proportions of amyloid-positive participants who were cognitively unimpaired or had MCI in the sample without dementia. To adjust for nonparticipation bias in the MCSA cohort, we used logistic regression models to assess the effects of age, sex, and education on (1) probability of participation and (2) probability of undergoing imaging tests. We combined these 2 sets of probabilities, then used the reciprocal of the probability for each individual as weights (inverse probability weighting [IPW]) to adjust the proportions of amyloid positivity for nonparticipation bias.27-29 Inclusion of vascular diseases in the IPWs did not make any difference, so these were not included in final IPWs. We then standardized the IPW-adjusted proportions to the Olmsted County population without dementia aged 50 to 89 years in 2010 by age or both age and sex, where applicable, to derive estimates of population-based prevalence. We estimated prevalence of amyloid positivity by ApoE ε4 allele, 20-year age group, and cognitive status in both sexes combined and each sex separately.

We used proportional hazards models to assess risk of progression and calculated hazard ratios (HRs) and 95% CIs from (1) amyloid-positive, cognitively unimpaired status (preclinical AD) to aMCI or AD dementia status; (2) from any amyloid-positive MCI status (ie, MCI due to AD) to AD dementia status; (3) from amyloid-positive aMCI status (ie, prodromal AD) to AD dementia status; and (4) from either amyloid-positive, cognitively unimpaired status or aMCI status (ie, preclinical or prodromal AD) to AD dementia, comparing these with the corresponding amyloid-negative reference categories using time from baseline as the time scale. Model 1 was unadjusted; model 2 was adjusted for age, sex, and education; and model 3 was adjusted for age, sex, and education and additionally for ApoE ε4 genotype. Adjustment for potential confounding by vascular diseases and depressive symptoms (BDI score ≥13 points) did not differ from estimates adjusted for age, sex, or education and therefore only the estimates without adjustment for vascular disease and depressive symptoms are presented. Because of the higher prevalence of amyloid positivity in women participants, we performed post hoc analyses by sex.

We assessed declines in global cognitive domain and memory domain z scores by amyloid status during the follow-up period using mixed–effects models centered on age, sex, and education, including random intercept and slope terms. All analyses were considered significant at a level of P < .05.

Results
Baseline Characteristics

Of 1671 participants without dementia who underwent amyloid imaging between August 1, 2008, and September 18, 2017, the mean (SD) age was 71.3 (9.9) years; 892 (53.4%) were men, 470 (28.3%) were ε4 carriers, and 179 (10.7%) had prevalent MCI (Table 1). The ApoE ε4 allele was more frequently found in people who were amyloid positive than in those who were amyloid negative, whether without cognitive impairment (amyloid positive: 182 of 434 [41.9%] vs amyloid negative: 221 of 1058 [20.9%]) or with MCI (amyloid positive: 56 of 110 [50.9%]) vs amyloid negative: 11 of 69 [16%]; P < .05). Global cognitive and memory z scores were lowest in participants who were amyloid-positive and had MCI. Compared with participants with amyloid imaging, nonparticipants (n = 2198) were older (nonparticipants: mean [SD] age, 72.5 (8.8) years vs participants: 70.8 (9.9) years; P < .001) ; were more frequently women (nonparticipants: 1134 of 2198 [51.6%] vs participants: 779 of 1671 [46.6%]; P = .002); had lower mean years of education (nonparticipants: mean [SD], 14.2 (2.7) years vs participants: 14.7 (2.6) years; P < .001); were more frequently found to have hypertension (nonparticipants: 1554 of 2198 [70.7%] vs participants: 1008 of 1671 [60.4%]; P < .001); were more frequently found to have strokes (nonparticipants: 85 of 2198 [3.9%] vs participants: 42 of 1671 [2.5%]; P = .02); and had lower global cognitive z scores (nonparticipants: mean (SD), −0.156 [1.01] vs participants: 0.20 [0.96]; P < .001). They did not differ in regard to frequencies of ApoE ε4 allele (nonparticipants: 594 of 2198 [28.2%] vs participants: 470 of 1671 [28.3%]; P = .92), depressive symptoms (nonparticipants: 158 of 2198 [7.3%] vs participants: 121 of 1671 [7.3%]; P = .96) or prevalent MCI (nonparticipants: 217 of 2198 [9.9%] vs participants: 179 of 1671 [10.7%]; P = .39).

The IPW-adjusted, population-standardized prevalence of amyloid-positive, cognitively unimpaired status in the population without dementia in both sexes combined increased from 2.7% (95% CI, 0.5% to 4.9%) in the age group from 50 to 59 years to 41.3% (95% CI, 33.4% to 49.2%) in participants aged 80 to 89 years (Table 2). The prevalence of amyloid-positive MCI in the population without dementia in both sexes combined increased from 0% at ages 50 to 59 years to 16.4% (95% CI, 10.3% to 22.5%) at ages 80 to 89 years. The total prevalence of amyloid positivity across all ages was 22.0% (18.9% to 30.5%). Age-specific estimates by sex followed the same pattern, with a higher total prevalence of amyloid positivity in women (25.6%; 95% CI, 20.8% to 30.5%) compared with men (17.8%; 95% CI, 14.2% to 21.4%). The age-specific proportions of amyloid positivity in participants without cognitive impairment and participants with MCI (percentages in Table 2) were comparable with prevalence. However, the proportions in ages 50 to 89 years combined were higher than prevalence.

When stratified by ApoE ε4 status, the prevalence of amyloid positivity was higher in ε4 carriers and older participants (eFigure in the Supplement). The prevalence of amyloid-positive participants without cognitive impairment among ε4 carriers vs noncarriers was 18.5% (95% CI, 14.0% to 23.1%) vs 7.5% (95% CI, 5.3% to 9.6%), respectively, in the age group 50 to 69 years, and 53.1% (95% CI, 46.7% to 59.6%) vs 32.2% (95% CI, 26.6% to 37.9%), respectively, in ages 70 to 89 years. The prevalence of amyloid-positive participants with MCI in ε4 carriers and noncarriers was 2.2% (95% CI, 0.2% to 4.2%) and 0.3% (95% CI, 0.0% to 0.7%), respectively, in the age group 50 to 69 years, and 20.4% (95% CI, 14.9% to 25.8%) and 10.5% (95% CI, 5.9% to 15.0%), respectively, in the age group 70 to 89 years. By sex, the prevalence of amyloid-positive ε4 carriers without cognitive impairment and noncarriers was higher in women (29.5% [95% CI, 24.6% to 34.4%] in ε4 carriers and 17.1% [95% CI, 13.8% to 21.6%] in noncarriers) than in men (24.3% [95% CI, 18.3% to 30.3%] in ε4 carriers and 10.8% [95% CI, 9.0% to 12.7%] in noncarriers) (eTable in the Supplement). The prevalence of amyloid-positive participants with MCI in ε4 carriers was higher in women than in men (women: 8.7% [95% CI, 5.3% to 12.1%] vs men: 3.1% [95% CI, 1.4% to 4.9%]), but there was no sex difference in noncarriers and no interaction of ApoE ε4 with sex.

Clinical Outcomes During Follow-up

During the follow-up period, 1519 participants had 1 or more evaluations: 1377 among participants without cognitive impairment and 142 in participants with MCI; 152 had no follow-up visits (Table 1). In this period, a higher proportion of the 137 deaths occurred in amyloid-positive participants: 54 of 434 amyloid-positive participants without cognitive impairment (13.4%) died vs 52 of 1058 amyloid-negative, cognitively unimpaired participants (4.9%), and 19 of 110 amyloid-positive participants with MCI (17.3%) died vs 8 of 69 amyloid-negative participants with MCI (4.9%).

During a mean (SD) follow-up period of 3.7 (1.9) years, 63 of 434 amyloid-positive participants without cognitive impairment (14.5%) vs 44 of 1058 amyloid-negative participants without cognitive impairment (4.2%) developed incident aMCI. During a mean (SD) follow-up of 3.8 (2.0) years, 3 of 1058 amyloid-negative individuals without cognitive impairment (0.3%); 14 of 434 amyloid-positive individuals without cognitive impairment (3.2%); 10 of 69 amyloid-negative individuals with MCI (14.5%); and 36 of 110 amyloid-positive individuals with MCI (32.7%) developed AD dementia.

Incident aMCI

In unadjusted models, all HRs of incident aMCI were elevated more than 2-fold (Table 3). In models adjusted by age, sex, and education, the HR was 2.26 (95% CI, 1.52 to 3.35; P < .001) for both sexes combined (Figure 1A), 1.96 (95% CI, 1.05 to 3.67; P = .04) in women, and 2.42 (95% CI, 1.46 to 4.04; P < .001) in men. The estimates were slightly attenuated with adjustment for ApoE ε4 (Table 3).

Incident AD Dementia

In unadjusted models, the HR of AD dementia was significantly elevated for amyloid-positive participants with MCI (preclinical AD) or aMCI (prodromal AD) and for amyloid-positive participants with either no cognitive impairment or MCI (preclinical and prodromal AD) for both sexes combined, in women, and in men (Table 3). In models adjusted for age, sex, and education, HRs were nonsignificantly elevated for amyloid-positive participants with MCI (Figure 1B) and aMCI (Figure 1C), and significantly elevated for amyloid-positive participants who were either cognitively unimpaired or had aMCI (2.56; 95% CI, 1.35 to 4.88; P = .004; Figure 1D). By sex, HRs were higher in women (HR, 5.27; 95% CI, 1.52 to 18.32; P = .01) than in men (HR, 1.69; 95% CI, 0.77 to 3.68; P = .19). With adjustment for ApoE ε4 status, the HRs were attenuated but remained elevated, consistent with the role of amyloid as a mediator of the association between the ApoE ɛ4 allele and AD dementia. Given the elevated HRs, the lack of significant differences observed in models adjusted for age, sex, and education is likely because of lack of power.

Progression rates (per 1000 person-years) were 42.0 for amyloid-positive individuals without cognitive impairment vs 12.4 for amyloid-negative individuals without cognitive impairment. Consistent with the concept of amyloid-positive aMCI as prodromal AD, progression rates to AD dementia were higher for amyloid-positive participants with aMCI (130.0) compared with amyloid-positive participants with MCI (including nonamnestic MCI; 103.2), and higher for amyloid-negative participants with aMCI (54.9) compared with all amyloid-negative participants with MCI (39.3).

Cognitive Decline

In mixed-effects models, global cognitive and memory domain z scores declined more for amyloid-positive individuals than for amyloid-negative participants. Slopes for global cognitive z scores were greater in amyloid-positive individuals without cognitive impairment (−0.076; 95% CI, −0.086 to −0.065) than in amyloid-negative individuals without cognitive impairment (0.005; 95% CI,−0.002 to 0.012) and amyloid-positive individuals with MCI (−0.180; 95% CI, −0.204 to −0.156) vs amyloid-negative individuals with MCI (−0.049; 95% CI, −0.079 to −0.019; Figure 2A). Similarly, slopes for memory domain z scores declined more greatly for amyloid-positive individuals without cognitive impairment (−0.048; 95% CI, −0.062 to −0.035) vs amyloid-negative individuals without cognitive impairment (0.037; 95% CI, 0.027 to 0.046) and declines were also greater in amyloid-positive individuals with MCI (−0.126; 95% CI, −0.156 to −0.096) than in amyloid-negative individuals with MCI (−0.025; 95% CI, −0.063 to 0.013; Figure 2B).

Discussion

Among study participants without dementia, the prevalence of amyloid positivity increased with age and was higher in women than in men. Prevalence estimates for ages 50 to 89 years combined were lower than observed proportions (unadjusted or unweighted). Hazard ratios were significantly elevated for the progression of amyloid-positive participants without cognitive impairment to aMCI and the progression of amyloid-positive participants or participants with aMCI to probable AD dementia, and were nonsignificantly elevated for the progression of amyloid-positive individuals from MCI and aMCI to AD dementia. Progression rates were higher for amyloid-positive participants than for amyloid-negative participants with no cognitive impairment or with aMCI and for amyloid-positive participants than for amyloid-negative participants with aMCI. Global cognitive and memory domain z scores declined more in amyloid-positive vs amyloid-negative participants. These findings are consistent with amyloid as a biomarker of AD dementia.

Valid and reliable population-based estimates of prevalence and outcomes are important for estimating what the impact of an approved anti-amyloid drug would be in a general population. Estimation of effect sizes from a population-based, sample without dementia may lead to more realistic expectations for anti-amyloid agents. In contrast, estimates of effect sizes derived from clinic-based studies and volunteer participants are subject to selection bias and may lead to inflated estimates of time to event. Such studies may overestimate progression rates, because these participants may have a higher likelihood of a family history of dementia or may be more symptomatic and therefore may progress at a higher rate than the general population. On the other hand, clinical trial cohorts are likely to have less comorbidity than a population-based study such as this one. This study focused on persons without dementia because current primary and secondary AD prevention trials are designed to prevent progression in persons without dementia. The unadjusted proportions in the age group 50 to 89 years were lower than prevalence because of differences in participation rates by age, sex, and education or because of oversampling of the people in the age group 80 to 89 years, who likely have more pathology, to ensure adequate power to detect significant differences in prospective studies.

Despite the nonsignificant associations by sex for incident AD dementia, the higher prevalence of amyloid positivity in women and sex differences in risk of progression for amyloid-positive individuals who are either cognitively unimpaired or have MCI may have implications for reporting clinical trial results by sex. Although we did not find a significant interaction of amyloid positivity with sex, higher rates of progression in amyloid-positive women with MCI compared with men may be partly because of the higher prevalence of amyloid positivity in women, particularly among women ε4 carriers compared with men. This is consistent with amyloid as a mediator of the effects of ApoE ε4 and the reduced HRs after adjustment for ApoE ε4 allele in the models. The HRs for progression from amyloid-positive, cognitively unimpaired status to aMCI were higher in men than in women, consistent with our previous study on incident MCI.30 We speculate that perhaps other factors (eg, vascular conditions) may play a greater role in progression at the preclinical AD stage in men, whereas amyloid may play a greater role in progression from preclinical or prodromal AD to AD dementia in women.5,6 These preliminary findings need to be validated in larger population-based, prospective studies. However, they suggest that potential sex differences should be taken into account in clinical trial design.

Our prevalence estimates are consistent with studies8,9,31 reporting an increase in amyloid positivity with age and ApoE ε4 allele carriage. Jansen et al8 reported a 10% to 12% prevalence of amyloid positivity in persons aged 50 to 59 years who had no cognitive impairment, which is higher than the 2.4% prevalence observed in this study, and that could be because of the large proportion of volunteers or to more accurate characterization from using both PET and cerebrospinal fluid amyloid test results in the previous study. Landau et al13 reported frequencies of amyloid positivity of 29% in persons without cognitive impairment, 42% in people with early MCI, and 62% in those with late MCI among participants of the ADNI study. Consistent with the highly selective ADNI cohort, these estimates are much higher than the population-based estimates in this study.

These prospective findings suggest that clinic-based studies and studies based primarily on volunteers may overestimate progression. The Australian Imaging, Biomarkers, and Lifestyle Study of Aging, which is based on volunteers and has a much higher proportion of ApoE ε4 carriers than the general population, reported32 a 4.5-fold increased risk of progression from cognitively unimpaired status to MCI or AD in persons who were amyloid positive compared with amyloid-negative persons. Vos et al33 reported HRs ranging from 4.6 to 33.8 for progression from preclinical AD stages 1 to 3 to symptomatic AD dementia among 311 volunteers in a Washington University School of Medicine study. It is unclear if tau and cerebrospinal fluid amyloid makers could have contributed to better characterization of amyloid-positive individuals. Similarly, the clinic-based Amsterdam Dementia Cohort of 132 patients with subjective memory complaints reported declines in memory and cognitive z scores that were higher than in those in the MCSA cohort.11 Nordberg et al31 reported increased progression to probable AD dementia in amyloid-positive participants with MCI in an European multicenter study. In 445 amyloid-positive participants without cognitive impairment of the ADNI study, Donohue et al34 reported longitudinal declines in the Preclinical Alzheimer Cognitive Composite, Mini–Mental State Examination, and Clinical Dementia Rating sum of boxes scores; however, lack of a decline in logical memory raises questions about possible selection bias. In a comparative study on frequencies and outcomes of biomarkers in the MCSA cohort (n = 126) vs the ADNI-1 cohort (n = 58) by Petersen et al,6 ADNI-1 participants had significantly higher frequencies of ApoE ε4 allele (ADNI-1: 32 of 58 [55%] vs MSCA: 49 of 126 [38.9%]), amyloid-positive neurodegeneration (ADNI-1: 32 of 58 [55%] vs MSCA: 54 of 126 [42.9%]), and higher rates of progression from aMCI to dementia (ADNI-1: 14 of 58 [29%] vs MSCA: 14 of 26 [15%]). Despite the small numbers in some of these studies, higher estimates compared with the present findings show the issues worth noting when using information from clinic-based and volunteer studies, including the ADNI study, to design clinical trials.

The strengths of our study include the large sample size, high prevalence of amyloid positivity, and population-based, natural history design. The prospective design and high retention of participants provide reliable and valid information on longitudinal outcomes. The medical records linkage system enabled adjustment for nonparticipation bias and identification of incident dementia outcomes in persons lost to follow-up.

Limitations

Potential limitations of our study include nonparticipation bias, which was mitigated by IPW adjustments. By design, prevalence of amyloid positivity are generalizable to the population without dementia only. Neurodegeneration was not considered, but the frequencies and outcomes of the National Institutes on Aging–Alzheimer’s Association stages in MCSA participants who has no cognitive impairment or had aMCI have been published.5,6,10 The present cohort was 98% white; given the homogeneity of the cohort, the findings may not be generalizable to studies with participants of races/ethnicities with a higher prevalence of risk factors, who therefore may have higher rates of progression.

Conclusions

The study findings may be informative for design of primary and secondary AD prevention trials. The sex differences in outcomes remain to be validated.

Back to top
Article Information

Corresponding Author: Rosebud O. Roberts, MD, ChB, MS, Department of Health Sciences Research, Mayo Clinic, 200 First St, SW, Rochester, MN 55905 (roberts.rosebud@mayo.edu).

Accepted for Publication: January 25, 2018.

Published Online: April 30, 2018. doi:10.1001/jamaneurol.2018.0629

Author Contributions: Dr Roberts had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Roberts, Kremers, Mielke, Coloma, Schauble, Lowe, Petersen.

Acquisition, analysis, or interpretation of data: Roberts, Aakre, Kremers, Vassilaki, Knopman, Alhurani, Geda, Machulda, Lowe, Jack, Petersen.

Drafting of the manuscript: Roberts.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Roberts, Aakre, Kremers.

Obtained funding: Roberts, Coloma, Schauble, Lowe, Jack, Petersen.

Administrative, technical, or material support: Roberts, Coloma, Schauble, Lowe, Jack, Petersen.

Study supervision: Roberts, Lowe.

Conflict of Interest Disclosures: Dr Roberts reports receiving research funding from the National Institutes of Health, Roche, and Biogen. Dr Kremers reports receiving research funding from AstraZeneca, Biogen, and Roche. Dr Vassilaki reports receiving research funding from Roche. Dr Coloma reports receiving personal fees from and owning shares of F. Hoffmann-La Roche Ltd. Dr Geda reports receiving funding from Roche. Dr Petersen reports receiving personal fees as a consultant for Roche, Inc. No other disclosures were reported.

Funding/Support: This study was funded by the National Institutes of Health (grants P50 AG016574, U01 AG006786, R01 AG041851, and R01 AG011378), F. Hoffmann-La Roche Ltd, the Elsie and Marvin Dekelboum Family Foundation, the GHR Foundation, the Mayo Foundation for Medical Education and Research, and the Rochester Epidemiology Project (grant R01 AG034676).

Role of the Funder/Sponsor: The funders and sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript and decision to submit the manuscript. As authors, Drs Coloma and Schauble (who are Roche employees) reviewed the manuscript prior to submission.

Additional Contributions: We thank Dana Swenson-Dravis, MA; Josie Williams, MA; and other Mayo Clinic Study of Aging staff for their work; Mary J. Dugdale, RN, and Connie J. Fortner, RN, for abstraction of medical record data; and Sondra L. Buehler, AA, for administrative assistance. These contributors were paid Mayo Clinic Study of Aging staff and were not additionally compensated for their contributions to this study.

References
1.
Dubois  B, Feldman  HH, Jacova  C,  et al.  Research criteria for the diagnosis of Alzheimer’s disease: revising the NINCDS-ADRDA criteria.  Lancet Neurol. 2007;6(8):734-746.PubMedGoogle ScholarCrossref
2.
Dubois  B, Hampel  H, Feldman  HH,  et al; Proceedings of the Meeting of the International Working Group (IWG) and the American Alzheimer’s Association on “The Preclinical State of AD”; July 23, 2015; Washington DC, USA.  Preclinical Alzheimer’s disease: Definition, natural history, and diagnostic criteria.  Alzheimers Dement. 2016;12(3):292-323.PubMedGoogle ScholarCrossref
3.
Sperling  RA, Aisen  PS, Beckett  LA,  et al.  Toward defining the preclinical stages of Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease.  Alzheimers Dement. 2011;7(3):280-292.PubMedGoogle ScholarCrossref
4.
Albert  MS, DeKosky  ST, Dickson  D,  et al.  The diagnosis of mild cognitive impairment due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease.  Alzheimers Dement. 2011;7(3):270-279.PubMedGoogle ScholarCrossref
5.
Knopman  DS, Jack  CR  Jr, Wiste  HJ,  et al.  Short-term clinical outcomes for stages of NIA-AA preclinical Alzheimer disease.  Neurology. 2012;78(20):1576-1582.PubMedGoogle ScholarCrossref
6.
Petersen  RC, Aisen  P, Boeve  BF,  et al.  Mild cognitive impairment due to Alzheimer disease in the community.  Ann Neurol. 2013;74(2):199-208.PubMedGoogle Scholar
7.
Jack  CR  Jr, Wiste  HJ, Weigand  SD,  et al.  Age-specific population frequencies of cerebral β-amyloidosis and neurodegeneration among people with normal cognitive function aged 50-89 years: a cross-sectional study.  Lancet Neurol. 2014;13(10):997-1005.PubMedGoogle ScholarCrossref
8.
Jansen  WJ, Ossenkoppele  R, Knol  DL,  et al; Amyloid Biomarker Study Group.  Prevalence of cerebral amyloid pathology in persons without dementia: a meta-analysis.  JAMA. 2015;313(19):1924-1938.PubMedGoogle ScholarCrossref
9.
Ossenkoppele  R, Jansen  WJ, Rabinovici  GD,  et al; Amyloid PET Study Group.  Prevalence of amyloid PET positivity in dementia syndromes: a meta-analysis.  JAMA. 2015;313(19):1939-1949.PubMedGoogle ScholarCrossref
10.
Petersen  RC, Wiste  HJ, Weigand  SD,  et al.  Association of elevated amyloid levels with cognition and biomarkers in cognitively normal people from the community.  JAMA Neurol. 2016;73(1):85-92.PubMedGoogle ScholarCrossref
11.
van Harten  AC, Smits  LL, Teunissen  CE,  et al.  Preclinical AD predicts decline in memory and executive functions in subjective complaints.  Neurology. 2013;81(16):1409-1416.PubMedGoogle ScholarCrossref
12.
Hansson  O, Zetterberg  H, Buchhave  P, Londos  E, Blennow  K, Minthon  L.  Association between CSF biomarkers and incipient Alzheimer’s disease in patients with mild cognitive impairment: a follow-up study.  Lancet Neurol. 2006;5(3):228-234.PubMedGoogle ScholarCrossref
13.
Landau  SM, Mintun  MA, Joshi  AD,  et al; Alzheimer’s Disease Neuroimaging Initiative.  Amyloid deposition, hypometabolism, and longitudinal cognitive decline.  Ann Neurol. 2012;72(4):578-586.PubMedGoogle ScholarCrossref
14.
Petersen  RC, Aisen  PS, Beckett  LA,  et al.  Alzheimer’s Disease Neuroimaging Initiative (ADNI): clinical characterization.  Neurology. 2010;74(3):201-209.PubMedGoogle ScholarCrossref
15.
Roberts  RO, Geda  YE, Knopman  DS,  et al.  The Mayo Clinic Study of Aging: design and sampling, participation, baseline measures and sample characteristics.  Neuroepidemiology. 2008;30(1):58-69.PubMedGoogle ScholarCrossref
16.
Petersen  RC, Roberts  RO, Knopman  DS,  et al; The Mayo Clinic Study of Aging.  Prevalence of mild cognitive impairment is higher in men.  Neurology. 2010;75(10):889-897.PubMedGoogle ScholarCrossref
17.
St Sauver  JL, Grossardt  BR, Yawn  BP,  et al.  Data resource profile: the Rochester Epidemiology Project (REP) medical records-linkage system.  Int J Epidemiol. 2012;41(6):1614-1624.PubMedGoogle ScholarCrossref
18.
Ivnik  RJ, Malec  JF, Smith  GE,  et al.  Mayo’s older Americans normative studies: WAIS-R norms for ages 56 to 97.  Clin Neuropsychol. 1992.Google Scholar
19.
Petersen  RC.  Mild cognitive impairment as a diagnostic entity.  J Intern Med. 2004;256(3):183-194.PubMedGoogle ScholarCrossref
20.
American Psychiatric Association.  Diagnostic and Statistical Manual of Mental Disorders. 4th ed. Washington, DC: American Psychiatric Association; 1994.
21.
McKhann  G, Drachman  D, Folstein  M, Katzman  R, Price  D, Stadlan  EM.  Clinical diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group under the auspices of Department of Health and Human Services Task Force on Alzheimer’s Disease.  Neurology. 1984;34(7):939-944.PubMedGoogle ScholarCrossref
22.
McKhann  GM, Knopman  DS, Chertkow  H,  et al.  The diagnosis of dementia due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease.  Alzheimers Dement. 2011;7(3):263-269.PubMedGoogle ScholarCrossref
23.
Jack  CR  Jr, Wiste  HJ, Knopman  DS,  et al.  Rates of β-amyloid accumulation are independent of hippocampal neurodegeneration.  Neurology. 2014;82(18):1605-1612.PubMedGoogle ScholarCrossref
24.
Senjem  ML, Gunter  JL, Shiung  MM, Petersen  RC, Jack  CR  Jr.  Comparison of different methodological implementations of voxel-based morphometry in neurodegenerative disease.  Neuroimage. 2005;26(2):600-608.PubMedGoogle ScholarCrossref
25.
Jack  CR  Jr, Bennett  DA, Blennow  K,  et al.  A/T/N: An unbiased descriptive classification scheme for Alzheimer disease biomarkers.  Neurology. 2016;87(5):539-547.PubMedGoogle ScholarCrossref
26.
Jack  CR  Jr, Wiste  HJ, Weigand  SD,  et al.  Defining imaging biomarker cut points for brain aging and Alzheimer’s disease.  Alzheimers Dement. 2017;13(3):205-216.PubMedGoogle ScholarCrossref
27.
Kessler  RC, Little  RJ, Groves  RM.  Advances in strategies for minimizing and adjusting for survey nonresponse.  Epidemiol Rev. 1995;17(1):192-204.PubMedGoogle ScholarCrossref
28.
D’Agostino  RB  Jr, Rubin  DB.  Estimating and using propensity scores with partially missing data.  J Am Stat Assoc. 2000;95(451):749-759.Google ScholarCrossref
29.
Little  RJA.  Survey nonresponse adjustments for estimates of means.  Int Stat Rev. 1986;54:139-157.Google ScholarCrossref
30.
Roberts  RO, Geda  YE, Knopman  DS,  et al.  The incidence of MCI differs by subtype and is higher in men: the Mayo Clinic Study of Aging.  Neurology. 2012;78(5):342-351.PubMedGoogle ScholarCrossref
31.
Nordberg  A, Carter  SF, Rinne  J,  et al.  A European multicentre PET study of fibrillar amyloid in Alzheimer’s disease.  Eur J Nucl Med Mol Imaging. 2013;40(1):104-114.PubMedGoogle ScholarCrossref
32.
Rowe  CC, Bourgeat  P, Ellis  KA,  et al.  Predicting Alzheimer disease with β-amyloid imaging: results from the Australian imaging, biomarkers, and lifestyle study of ageing.  Ann Neurol. 2013;74(6):905-913.PubMedGoogle ScholarCrossref
33.
Vos  SJ, Xiong  C, Visser  PJ,  et al.  Preclinical Alzheimer’s disease and its outcome: a longitudinal cohort study.  Lancet Neurol. 2013;12(10):957-965.PubMedGoogle ScholarCrossref
34.
Donohue  MC, Sperling  RA, Petersen  R, Sun  CK, Weiner  MW, Aisen  PS; Alzheimer’s Disease Neuroimaging Initiative.  Association between elevated brain amyloid and subsequent cognitive decline among cognitively normal persons.  JAMA. 2017;317(22):2305-2316.PubMedGoogle ScholarCrossref
×