Abstract
Introduction:
Large longitudinal biomarkers database focusing on middle age is needed for AD prevention.
Methods:
Data for cerebrospinal fluid (CSF) analytes, molecular imaging of cerebral fibrillar β-amyloid with positron emission tomography (PET), magnetic resonance imaging (MRI)-based brain structures, and clinical/cognitive outcomes were harmonized across eight AD biomarker studies. Statistical power was estimated.
Results:
The harmonized database included 7779 participants with clinical/cognitive data: 3542 were 18~65 years at baseline, 5865 had longitudinal cognitive data for a median of 4.7 years 2473 participated in the CSF studies (906 had longitudinal data), 2496 participated in the MRI studies (1283 had longitudinal data), and 1498 participated in the PET amyloid studies (849 had longitudinal data). The database provides adequate power for detecting early biomarker changes, and demonstrates the feasibility of AD prevention trials on middle-aged individuals.
Conclusions:
The harmonized database is an optimum resource to design AD prevention trials decades prior to symptomatic onset.
Keywords: Alzheimer disease, biomarkers, cerebrospinal fluid (CSF), amyloid imaging with positron emission tomography (PET) using the [11 C] benzothiazole tracer, Pittsburgh Compound-B (PIB), magnetic resonance imaging (MRI) volumetrics, preclinical stages, prevention trials
1. Background
Accumulating research suggests that the neuropathological processes of Alzheimer disease (AD) begin decades prior to symptom onset [1-4]. The natural history of this long asymptomatic phase, termed preclinical AD, is important to understand because it likely offers an optimal time window for prevention trials of anti-AD therapies. Because individuals in the preclinical stage of AD have no (or minimal) cognitive impairment, biomarkers of different modalities are important for tracking the progression of preclinical AD. To date, most human biomarker studies, including magnetic resonance imaging (MRI)-based regional brain volumes, cerebrospinal fluid (CSF) analytes, and molecular imaging of cerebral fibrillar β-amyloid with positron emission tomography (PET) using the [11 C] benzothiazole tracer, Pittsburgh Compound-B (PIB), have been focused on either elderly individuals older than 65 years or those who are already symptomatic. This age window, however, may very likely miss the crucial interval during which the neuropathophysiological processes of AD are initiated [5-6]. To fully understand the natural history of AD from the preclinical to symptomatic stages and the temporal orderings of pathological changes in the brain, a high-quality and longitudinal biomarker database of individuals younger than 65 years, preferably starting from young adulthood, is necessary. Analyses of high-quality longitudinal AD biomarker data are also crucial for participant recruitment from secondary prevention trials to primary prevention trials of AD [7-8].
This study describes the development of a large database on individuals 18~65 years old at baseline, as well as those older than 65 years with and without AD dementia, by harmonizing cross-sectional and longitudinal data on CSF biomarkers, PET amyloid imaging markers, MRI structure measures, and clinical/cognitive outcomes collected from eight high-quality longitudinal biomarkers studies of AD. This study further assesses the statistical power of the harmonized data in detecting early changes of biomarkers and in designing future AD prevention trials.
2. Methods
2.1. Participants
Participants are from eight ongoing longitudinal biomarker studies of AD: the Washington University (WU) Adult Children Study (ACS), the Johns Hopkins University (JHU) Biomarkers for Older Controls at Risk for Dementia (BIOCARD) Study, the Wisconsin Registry for Alzheimer's Prevention (WRAP), the Australian Imaging, Biomarkers and Lifestyle (AIBL) Study, the WU Dominantly Inherited Alzheimer Network (DIAN), the WU Healthy Aging and Senile Dementia (HASD) program project, and the WU Knight Alzheimer’s Disease Research Center (ADRC) and Wisconsin ADRC (WADRC). All studies have focused primarily on the asymptomatic phase of AD, and/or recruited young and middle-aged participants at risk for AD and followed longitudinally with assessments of AD biomarkers, cognition and everyday function. All participants have given written informed consent, and agreed to data sharing. The protocol of the current study is approved by the Institutional Review Board of WU School of Medicine.
A summary of participant recruitment, inclusion/exclusion criteria, and specific aims for all eight studies is in Table 1. Briefly, the WU ACS [9] has enrolled community-dwelling volunteers aged 43 to 75 year since 2004, stratified by both family history (FH) for late-onset AD and ages at baseline (43-54, 55-64, 65-75 years). Eligibility criteria are cognitive normality at baseline (defined as a Clinical Dementia Rating (CDR, [10]) of 0), and willingness to complete all biomarker study procedures. Exclusion criteria include conditions that would preclude biomarker assessments and/or confound cognitive assessment or membership in families with a known causative mutation for AD. Individuals are assessed every two years for cognition, CSF, and imaging (amyloid PET and MRI) biomarkers until symptom onset and then annually for cognition; participants age 65 years or older are assessed annually for cognition. The WU Knight ADRC and its affiliated program project, Healthy Aging and Senile Dementia (HASD), were established in 1985 and 1984, respectively, and have since enrolled volunteers mostly older than 65 years. The WU ADRC and HASD implement identical research protocols as the ACS for clinical assessments, CSF and imaging procedures [11]. The cognitive battery of the WU ADRC and HASD is not identical to that of the ACS, but overlaps significantly. A total of 3131 participants from the WU ACS, ADRC, and HASD are included in the database.
Table 1.
Design characteristics of the eight studies
| Cohort | Sponsor | Year of Initiation |
Main Objectives | Minimum Baseline Age (y) |
Major Inclusion Criteria |
|---|---|---|---|---|---|
| HASD+ADRC | WU | 1984,1985 | To study natural history of preclinical and symptomatic AD on clinical, cognitive, CSF biomarkers, and imaging outcomes | 65 | Community-dwelling volunteers with and without AD |
| BIOCARD | JH | 1995 | To understand and predict progression from normal cognition to MCI and to dementia, particularly AD | 40 | Cognitively normal individuals who were first degree relatives of patients with dementia |
| WRAP | Wisconsin | 2001 | To identify early cognitive decline and characterize midlife factors for AD risk | 36 | Cognitive normal middle aged individuals whose parent(s) had AD or whose parents lived to old age with no signs of AD or other serious memory problems |
| WADRC | Wisconsin | 2009 | To identify early signs of AD and its causes and to better understand why AD occurs and finds ways to prevent, slow, or halt its progression. | 43 | Community-dwelling volunteers with MCI or AD dementia, and old as well as middle-aged controls |
| ACS | WU | 2005 | To identify antecedent biomarkers for AD from middle age | 43 | Cognitively normal individuals of age 45~74 years at baseline, whose parents had AD before 80 or never developed AD and lived past 70 years of age |
| AIBL | Australia | 2006 | To assess longitudinal relation of biomarkers and risk factors to cognitive function | 55 | Community-dwelling volunteers with and without AD |
| DIAN | WU | 2008 | To compare autosomal dominant AD between mutation carriers and non-mutation carriers | 18 | family members with at least 50% risk for the 3 autosomal dominant AD-causing gene mutations in Presenilin1 or 2, or amyloid precursor protein) |
The BIOCARD study [12], initiated in 1995 at the National Institutes of Health(NIH), enrolls cognitively normal (CN) middle age individuals. About three-quarters of the participants have a first-degree relative with AD dementia. The study was stopped in 2005 and re-initiated at JHU in 2009. Since then, the annual visits have included a neuropsychological battery, history of symptom onset, CDRs, a physical and neurological examination, medication use, and behavioral and mood assessments, similar to what were given at the NIH. MRI scans and CSF were collected approximately every 2 years while the study was at the NIH. A total of 318 CN participants at baseline who were re-enrolled in 2009 are included in the database.
WRAP enrolls CN volunteers with a FH of probable AD dementia starting late 2001 and those without FH starting in 2004 [13]. Participants are predominantly aged 40-65 years with good health. Participation includes baseline, four years later and then serial visits every two years with clinical/cognitive assessments, medical condition, and lifestyle histories. The WADRC enrolls both CN individuals and those with AD dementia or Mild Cognitive Impairment (MCI, [14]), and a similar middle-aged cohort to WRAP for enrollment in prevention trials. Biomarker data including CSF, MRI, PET PIB of amyloid imaging are acquired from subsets of participants in the WRAP and the WADRC. A total of 2385 participants from the WRAP (n=1555) and the WADRC (n=830) are included in the database.
The AIBL study [15] was initiated in 2006 by the Australian Commonwealth Scientific Industrial and Research Organization. It enrolls volunteers age 55 years or older and excludes individuals with a history of comorbidities which could affect cognition. Initial assessment classifies individuals as CN, MCI, or AD dementia, and further by the CDR. Longitudinal assessments occur at 18-month intervals, including an extensive cognitive battery, CSF biomarkers, health/lifestyle characteristics, medical history and medication use. A subset of participants undergo amyloid PET PIB and MRI imaging. A total of 1441 participants from the AIBL are included in the database.
The DIAN, launched in 2008, is an international registry of adult children (≥18 years) of individuals who had a highly penetrant mutation for AD in amyloid precursor protein, presenilin 1 or 2 [16]. The DIAN enrolls CN and symptomatic mutation carriers and non-carrier family members as controls. Participants are assessed longitudinally with clinical and cognitive batteries, MRI structural and PET amyloid imaging protocols, and CSF collection. Asymptomatic individuals are assessed every two years, but annually if they are within three years of their estimated parental age of symptom onset. Symptomatic participants are annually evaluated unless cognitive decline prevents participation. A total of 504 participants from the DIAN are included in the database.
2.2. Harmonization of Clinical and Cognitive Data
The clinical assessment protocols of all eight studies are largely consistent with that of the National Alzheimer Coordinating Center Uniform Data Set (UDS) [17], which include standard diagnostic criteria for detection of dementia and its differential diagnosis [18]. The presence or absence of dementia and, when present, its severity are operationalized with the CDR by all studies. Individuals in the AIBL, BIOCARD and WRAP and WADRC studies are classified as CN, MCI, and AD dementia by a clinical review panel. However, across studies, the designation of MCI is inconsistent. Hence, a common definition of MCI is adopted, based on a published algorithm that requires a CDR of 0.5 and either at least two memory test scores or at least two non-memory test scores one standard deviation (SD) below the age-adjusted mean [19].
The individual cognitive batteries were detailed in Xiong et al. [9] for WU ACS, Ellis et al. [15] for AIBL, Albert et al. [12] for BIOCARD, and Johnson et al. [13] and Dowling et al. [20] for WRAP and WADRC, Morris et al. [11] for WU HASD and ADRC, and Storandt et al. [21] for DIAN. All include tests of episodic memory, semantic knowledge, working memory, executive function/attention, visuospatial ability, speed of processing, psychomotor speed, and general cognition. Most cognitive tests are shared by at least two studies and included in the harmonized database (see Supplemental Table 1), e.g., the Mini-Mental State Examination (MMSE), California Verbal Learning Test, Wechsler Memory Scale (WMS)-III or WMS-R Logical Memory I and II, WMS-III or WMS-R Verbal Paired Associates I and II, Boston Naming Test, Trailmaking Tests A & B, Wechsler Adult Intelligence Scale (WAIS)-III or WAIS-R Digit Span & Digit Symbol, Letter-number sequencing, Wechsler Abbreviated Scale of Intelligence (Vocabulary & Similarities, Block design), the Stroop task, Free & Cued Selective Reminding, Clock Drawing, Benton Line Orientation, Rey Complex Figure Test, Rey Auditory Verbal Learning Test, Letter and Category Fluency (CFL), and Animal Naming (60 sec) [9,11-13,15,20-21]. A cognitive composite is created using Logical Memory Delayed Recall, Boston Naming, Animal Fluency, MMSE, and WAIS-R Digit Symbol, by averaging the z-scores using the overall means(SD) from the raw baseline scores.
2.3. CSF Collection, Processing, and Database Harmonization
The WU ACS, ADRC, HASD, and WRAP/WADRC collect CSF at 8 a.m. after overnight fasting. Samples are gently inverted to avoid possible gradient effects, briefly centrifuged at low speed, and aliquoted into polypropylene tubes prior to freezing at −80°C [22]. CSF samples from the WU ACS, ADRC, and HASD are analyzed for total tau (Tau), and tau phosphorylated at threonine 181 (Ptau), and Aβ1-42 (Aβ42) by enzyme-linked immunosorbant assay (INNOTEST; Fujirebio, Malverne, PA ). BIOCARD, AIBL, and DIAN collect and analyze the CSF specimens similar to the Alzheimer Disease Neuroimaging Initiative (ADNI, [23]) protocol, i.e., using the xMAP-based AlzBio3 kit (Fujirebio, Malverne, PA ) run on the Bioplex 200 system. Details for BIOCARD CSF analysis were described previously [24]. The WRAP and WADRC use the same CSF protocol as described in [25], and all samples have been processed by Henrik Zetterberg and Kaj Blennow, University of Gothenburg, Sweden, also using enzyme-linked immunosorbant assay. Given that variations in CSF collections and analyte assays may affect the analyses of CSF biomarkers, variables on CSF collections and analyte assays are added in the integrated database to allow sensitivity and adjusted analyses. Additionally, CSF samples in WU ACS, HASD, ADRC, and DIAN have been re-processed with the same Elecsys® immunoassays on the automated Roche cobas e 601 analyzer [26].
2.4. MRI and PET Brain Imaging and Database Harmonization
MRI and PET imaging with PIB for the AIBL study are detailed in [27]. 3D T1 (Siemens 1.5T and 3T scanners) Magnetization Prepared Rapid Acquisition Gradient Echo (MPRAGE) and a T2 turbospin echo and Fluid-attenuated inversion recovery (FLAIR) sequence MRI are acquired for screening and co-registration with the PET images. For the WU ACS, ADRC, HASD, and DIAN studies, consistent protocols are implemented on 3T scanners using a 12-channel head coil equipped with parallel imaging including four major components: structural MRI, diffusion tensor imaging (DTI) and arterial spin labeling (ASL), for a total of approximately 60 minutes of MR imaging. The MRI protocol provides high gray-white matter contrast (1mm × 1mm × 1 mm) MPRAGE T1-weighted volume acquisitions [28]. PET PIB imaging is conducted using a scanner (CTI) in a darkened, quiet room. Details are in [28]. For WRAP, the MRI and PET procedures are described in [29]. Briefly, the procedure involves a detailed 3T (GE) MRI protocol involving a T1w volume, FLAIR 3D volume, resting-state Blood oxygen level-dependent (BOLD) scan, DTI, and pseudo-continuous arterial spin labeling, and a dynamic 70 min [C11] PIB PET scan. For BIOCARD, the MRI protocol at the NIH included a Fast Spin Echo Sequence, Spoiled Gradient Recalled Echo (SPGR) sequences, and FLAIR imaging collected on a 1.5 T GE scanner [30]. Starting in 2015, MRI images have been collected every other year on a Philips 3T scanner, and the protocol includes MPRAGE, FLAIR, DTI and resting BOLD sequences.
To harmonize the imaging measures, raw MRI and PIB PET scans from WU ACS, ADRC, HASD, AIBL, WRAP and WADRC are sent to WU NeuroImaging Lab for central reprocessing. Structural MRI processing steps include motion correction, if applicable, averaging across scans, and atlas transformation. Regional volumes and cortical thickness are obtained via the FreeSurfer image analysis suite (Athinoula A. Martinos Center for Biomedical Imaging, Charlestown, MA) [31], version 5.3. Determination of the regions-of-interest (ROIs) and pipelines are described in [28,32]. FreeSurfer quality control (QC) criteria include (1) dural inclusion, gray matter exclusion, sulcus inclusion, cerebellum inclusion; (2) cerebellum, subcortical and hippocampus segmentation exclusion; (3) white matter exclusion, and (4) lateral ventricle segmentation. Analyses are performed on adjusted volumetric measures after regressing for the effect of scanner platform and with adjustment for head size [32]. PIB amyloid deposition in the ROIs is determined using FreeSurfer, and a standardized uptake value ratio (SUVR) with correction for partial volume effects is calculated [32]. The cerebellum is chosen as the reference region. The mean cortical SUVR is calculated from FreeSurfer regions within the prefrontal cortex, precuneus, and temporal cortex [28].
2.5. Database QC, Missing Data, and Standardization
To harmonize data across studies, multiple teleconferences were held with study-specific data managers to implement final QC and standardization. Final QC includes double checks on missing data and the ranges. The final database standardization includes a uniform database dictionary, and standard codes/formats for biomarkers and cognitive tests, missing data, CSF collection/assay/lots/plates, imaging acquisition (scanners, tracers) and processing parameters, and demographics, using approaches consistent with the STRATOS initiative [33].
3. Results
3.1. The Harmonized Database
The integrated database contained 7779 participants who went through clinical and cognitive assessments and/or biomarker assessments. Table 2 presents the baseline demographics and clinical status, as a function of baseline age. Of the 7779 participants, 3542 (45.53%) were 65 years or younger at baseline, 4751 (61.07%) were female, 2987 (38.40%) were APOE4 positive, 303 carried penetrant mutations in one of three genes for AD, 4640 (59.65%) had more than 12 years of education, and 3768 (48.44%) had a family history of dementia. A total of 5172 (66.49%) participants were CDR 0 at baseline. During the longitudinal follow-up, a total of 894 participants received a CDR 0.5 (or higher) and a diagnosis of MCI or AD.
Table 2.
Baseline characteristics of the integrated database (all individuals with a minimum of baseline clinical or cognitive assessments were included)
| Characteristics | All | Age at Baseline | ||||
|---|---|---|---|---|---|---|
| (18, 45] | (45, 55] | (55, 65] | (65, 75] | (75, 105] | ||
| N (%) | 7779 | 544(6.99) | 1179(15.16) | 1819(23.38) | 2147(27.60) | 2090(26.87) |
| Age: mean/SD (range) | 65.59/14.00 (18.00-103.81) | 35.58/7.46 (18.00-45.00) | 50.75/2.88 (45.02-55.00) | 60.29/2.90 (55.00-65.00) | 69.96/2.90 (65.01-75.00) | 81.89 (75.00-103.81) |
| Sex: N (%) | ||||||
| Female | 4751(61.07) | 328(60.29) | 824(69.89) | 1168(64.21) | 1189(55.38) | 1242(59.43) |
| Male | 3025(38.89) | 215(39.52) | 355(30.11) | 650(35.73) | 957(44.57) | 848(40.57) |
| missing | 3 (0.04) | 1(0.18) | 0(0) | 1(0.05) | 1(0.05) | 0(0) |
| Race: N (%) | ||||||
| White | 5382(69.19) | 467(85.85) | 1034(87.70) | 1359(74.71) | 1259(58.64) | 1263(60.43) |
| Black | 740(9.51) | 10(1.84) | 92(7.80) | 164(9.02) | 226(10.53) | 248(11.87) |
| Other | 202(2.60) | 59(10.85) | 49(4.16) | 44(2.42) | 34(1.58) | 16(0.77) |
| unknown | 1455(18.70) | 8(1.47) | 4(0.34) | 252(13.85) | 628(29.25) | 563(26.94) |
| Family history: N (%): | ||||||
| YES | 3768(48.44) | 522(95.96) | 909(77.10) | 1015(55.80) | 849(39.54) | 473(22.63) |
| NO | 3187(40.97) | 20(3.68) | 255(21.63) | 687(37.77) | 1018(47.41) | 1207(57.75) |
| missing | 824(10.59) | 2(0.37) | 15(1.27) | 117(6.43) | 280(13.04) | 410(19.62) |
| Education (year): N (%) | ||||||
| ≤12 | 2232(28.69) | 133(24.45) | 178(15.10) | 380(20.89) | 707(32.93) | 834(39.90) |
| >12 | 4640(59.65) | 383(70.40) | 890(75.49) | 1264(69.49) | 1190(55.43) | 913(43.68) |
| missing | 907(11.66) | 0(0) | 111(9.41) | 175(9.62) | 250(11.64) | 343(16.41) |
| Clinical diagnosis | ||||||
| Normal | 5172(66.49) | 484(88.97) | 1088(92.28) | 1552(85.32) | 1297(60.41) | 751(35.93) |
| MCI | 627(8.06) | 8(1.47) | 10(0.85) | 59(3.24) | 231(10.76) | 319(15.26) |
| Other | 164(2.11) | 14(2.57) | 9(0.76) | 40(2.20) | 66(3.07) | 35(1.67) |
| AD | 1729(22.23) | 34(6.25) | 60(5.09) | 144(7.92) | 533(24.83) | 958(45.84) |
| Missing | 87(1.12) | 4(0.74) | 12(1.02) | 24(1.32) | 20(0.93) | 27(1.29) |
| CDR: N (%) | ||||||
| 0 | 5172(66.49) | 484(88.97) | 1088(92.28) | 1552(85.32) | 1297(60.41) | 751(35.93) |
| 0.5 | 1574(20.23) | 43(7.90) | 56(4.75) | 159(8.74) | 588(27.39) | 728(34.83) |
| ≥1 | 1004(12.91) | 15(2.76) | 24(2.04) | 100(5.50) | 260(12.11) | 605(28.95) |
| missing | 29(0.37) | 2(0.37) | 11(0.93) | 8(0.44) | 2(0.09) | 6(0.29) |
| APOE4: N (%) | ||||||
| positive | 2987(38.40) | 158(29.04) | 503(42.66) | 637(35.02) | 880(40.99) | 776(37.13) |
| negative | 4354(55.97) | 381(70.04) | 656(55.64) | 1111(61.08) | 1108(51.61) | 1084(51.87) |
| missing | 438(5.63) | 5(0.92) | 20(1.70) | 71(3.90) | 159(7.41) | 230(11.00) |
| CSF: N (%) | 2473(31.79) | 358(65.81) | 414(35.11) | 586(32.22) | 750(34.93) | 365(17.46) |
| PET PIB: N (%) | 1498(19.26) | 350(64.34) | 267(22.65) | 345(18.97) | 354(16.49) | 182(8.71) |
| MRI: N (%) | 2496(32.09) | 389(71.51) | 406(34.44) | 529(29.08) | 762(35.49) | 410(19.62) |
Table 2 and 3 present the amount of cross-sectional and longitudinal data on CSF, neuroimaging, and cognitive markers as a function of baseline age and CDR. A total of 5865 individuals participated in the longitudinal cognitive assessments for a median of 4.7 years. Of these, 2775 were 65 years or younger at baseline. A total of 2473 participated in the CSF studies. Of these, 1358 were 65 years or younger at baseline, and 906 received longitudinal CSF assessments for a median of 3.4 years. A total of 2496 individuals participated in the MRI studies. Of these, 1324 were 65 years or younger at baseline, and 1283 received longitudinal MRI assessments for a median of 4.0 years. Finally, a total of 1498 individuals participated in the PET amyloid PIB studies. Of these, 962 were 65 years or younger at baseline, and 849 received longitudinal PET PIB assessments for a median of 3.4 years. Supplemental Table 2 presents study-specific demographics, sample size, and years of follow-up for each biomarker modality. Supplemental Table 3 presents study-specific mean (SD) for baseline PET PIB cortical mean SUVR, MRI hippocampal volume, CSF biomarkers, and a cognitive composite as a function of baseline age, suggesting largely comparable biomarker values across studies within the same age windows (and with same assay types for CSF markers).
Table 3:
Sample sizes (N), Median (Range) of Longitudinal Follow-up (in Years) of CSF and Neuroimaging Biomarkers and Cognition
| All | Age at Baseline | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| (18. 45] | (45, 55] | (55, 65] | (65, 75] | (75, 105] | 0 | 0.5 | ≥1 | ||
| Cognition | |||||||||
| N | 5865 | 354 | 949 | 1472 | 1683 | 1407 | 4148 | 1144 | 559 |
| median (range) | 4.70(0.34-27.60) | 5.08(1.00-16.00) | 9.00(0.93-24.58) | 7.00(0.88-27.60) | 4.53(0.51-24.93) | 3.30(0.34-17.92) | 6.01(0.51-27.60) | 3.29(0.55-25.83) | 2.05(0.47-13.45) |
| CSF | |||||||||
| N | 906 | 169 | 220 | 235 | 229 | 53 | 800 | 88 | 18 |
| median (range) | 3.37(0.46-12.99) | 3.01(0.92-10.10) | 3.42(0.69-12.99) | 3.24(0.46-12.40) | 4.36(1.00-11.14) | 2.91(0.61-12.01) | 3.53(0.46-12.99) | 2.86(0.61-12.17) | 1.27(1.00-5.75) |
| PIB-PET | |||||||||
| N | 849 | 182 | 169 | 210 | 200 | 88 | 716 | 109 | 24 |
| median (range) | 3.42(0.82-9.61) | 3.19(0.90-8.72) | 3.37(0.82-9.61) | 3.68(1.01-9.52) | 4.33(0.92-8.93) | 3.28(1.03-6.50) | 3.92(0.92-9.61) | 2.61(0.90-7.04) | 1.57(0.82-5.77) |
| MRI | |||||||||
| N | 1283 | 209 | 220 | 262 | 405 | 187 | 1025 | 216 | 42 |
| median (range) | 4.00(0.36-17.19) | 3.18(0.92-10.24) | 4.12(0.96-13.31) | 4.17(0.80-17.19) | 4.24(0.62-13.65) | 3.39(0.36-11.81) | 4.17(0.62-17.19) | 2.98(0.36-15.16) | 1.71(0.94-8.60) |
3.2. Statistical Power to Detect Early Changes of AD
Table 4 presents the annual rates of longitudinal change for all major AD biomarkers as a function of baseline age detectable by the harmonized database with at least 80% statistical power. These power analyses excluded the 303 mutation carriers from DIAN. These power analyses assumed the same variance parameters estimated from the WU ACS in a random intercept and random slope model [34], and were based on a standard normal test for testing whether each rate of change differs from zero at a significance level of 5% [35].
Table 4:
Annual longitudinal changes (in magnitude) detectable with at least 80% power by the harmonized database (Note that different markers may increase or decrease within a specific age interval)
| Baseline age [18, 45] |
Baseline age (45, 55] |
Baseline age (55, 65] |
Baseline age (65, 75] |
Baseline age (75, 105]b |
|
|---|---|---|---|---|---|
| CSF Aβ42: pg/mL/year | 41.3 | 15.1 | 15.7 | 13.6 | 36.8 |
| CSF Tau: pg/mL/year | 2.30 | 1.05 | 1.63 | 2.73 | 6.11 |
| CSF Ptau: pg/mL/year | 0.210 | 0.107 | 0.199 | 0.249 | 0.564 |
| PET PIB Mean Cortical SUVR | 0.011 | 0.008 | 0.011 | 0.011 | 0.019 |
| PET PIB Precuneus SUVR | 0.014 | 0.010 | 0.013 | 0.014 | 0.025 |
| MRI Hippocampal Volume in mm3 | 39.2 | 16.7 | 12.1 | 15.4 | 28.9 |
| Cognitive Compositea | 0.0077 | 0.0034 | 0.0041 | 0.0078 | 0.011 |
The cognitive composite is the averaged z-scores from Logical Memory Delayed Recall, Animal Naming, and Trailmaking A & B.
The large numbers in the age group of 75-105 years reflect the fact that the sample sizes within this age group are much smaller than the other age groups (see Table 3) and hence have limited statistical power, and therefore can only detect a much larger rate of annual change.
3.3. Design of Future Prevention Trials
We consider a future two-arm (treatment, placebo) 1:1 primary prevention trial on individuals younger than 75 years with annual assessments of 5 years to test the efficacy on two amyloid biomarkers (PET PIB SUVR, CSF Aβ42). Table 5 presents, for each biomarker and each baseline age interval, the total sample size needed to detect 25%, 50%, 75% and 100% improvement in the rate of change for the treatment in comparison to the placebo. These power analyses were based on a standard normal test with 80% power and a significance level of 5% [35]. The annual rate of change for the placebo and the variance parameters were estimated from the WU ACS database in a random intercept and random slope model [34].
Table 5.
Total sample sizes for a future 1:1 prevention trial needed to detect 25%, 50%, 75% and 100% improvement on the rate of change for the treatment arm (in comparison to the placebo arm) with at least 80% power
| Percentage of improvement for treatment on the rate of change |
Marker | Age at Baseline (in Years) | ||
|---|---|---|---|---|
| (45, 55] | (55, 65] | (65, 75] | ||
| 25% | CSF Aβ42 | 1346 | 2190 | 908 |
| PET PIB Mean Cortical SUVR | 1986 | 870 | 368 | |
| 50% | CSF Aβ42 | 338 | 548 | 228 |
| PET PIB Mean Cortical SUVR | 498 | 218 | 92 | |
| 75% | CSF Aβ42 | 150 | 244 | 102 |
| PET PIB Mean Cortical SUVR | 222 | 98 | 42 | |
| 100% | CSF Aβ42 | 86 | 138 | 58 |
| PET PIB Mean Cortical SUVR | 126 | 56 | 24 | |
3.4. Data Sharing and Confidentiality
The integrated database is housed behind a firewall to ensure HIPPA compliancey. An internet website allows investigators to query and request the data for analyses. The website is posted on a Linux Apache web server within WU Division of Biostatistics, created with HTML, PHP URL=https://biostat.wustl.edu/adrc/test/xiong/pbs/. Investigators are required to apply for a user name and password to query the data and submit data requests. A review committee of all study principal investigators will review data requests for approval. Upon approval, data will be shared in a secure HIPPA-compliant manner. To ensure scientific rigor and reproducibility, authors of all publications using the database will be required to submit their final analysis data set and computing codes back for archive purposes.
4. Discussion
A biomarker model of AD has recently been hypothesized in which the most widely validated biomarkers of AD pathology become abnormal and progressively change in an ordered manner [1,3,16]. There is now scientific evidence that Aβ accumulation and deposition in the brain is a very early pathological process in AD, detectable by PET imaging of amyloid plaques and CSF Aβ42 concentration. Neurofibrillary tangles (NFTs), neuronal death, and medial-temporal lobe atrophy on MRI appear to begin during the preclinical phase of AD, and by the time of early symptoms, neuronal cell death is already significant in the CA1 region of the hippocampus and layer II of the entorhinal cortex [2]. However, NFTs and neuronal cell death may not peak until the later clinical stages of AD. The acceleration of tau aggregation and neurodegeneration, detectable by increases in CSF Tau and Ptau [22,23], may mark the transition just prior to symptom onset. By the time of early symptoms, significant neuronal loss and structural changes have already occurred in the brain [1,4,16,22].
The statistical test of the hypothesized biomarker model of AD require a large and high-quality longitudinal biomarker database that focuses on the age window when the initiation of the neurodegenerative process of AD likely begins. Our earlier reports on the WU ACS alone [1, 9] and others [4, 27, 29] suggested that biomarker changes may first appear around age 50 for late-onset AD, depending on factors such as APOE4 status. Results from individual studies, however, suffer from a lack of statistical power to definitively pinpoint the earliest possible AD biomarker signatures. To our knowledge, the harmonized database represents the largest longitudinal CSF, PIB and MRI imaging dataset for asymptomatic individuals younger than 65 years. In fact, the database can allow estimates of the longitudinal rate of changes starting from 18 years, which will likely capture the very initial changes in biomarkers during the decades-long preclinical stage, and when compared across biomarkers, offer critical information about the temporal cascade of these changes. For example, the annual rates of changes detectable by the harmonized database, as presented in Table 4, are as small as 15.1 pg/mL in CSF Aβ42 and 0.008 in mean cortical PIB SUVR for individuals from 45~55 years, suggesting sufficient statistical power to detect subtle biomarker changes.
Major secondary prevention RCTs, the A4 trial (https://clinicaltrials.gov/ct2/show/NCT02008357), the DIAN TU trials (https://clinicaltrials.gov/ct2/show/NCT01760005, [7]), and the Alzheimer's Prevention Initiative (API) trial (https://clinicaltrials.gov/ct2/show/NCT01998841), are currently ongoing. All employ biomarkers or genetics as the subject inclusion/exclusion criteria or secondary efficacy endpoints. Primary prevention trials are currently planned, employing imaging and CSF biomarkers as the primary efficacy endpoints [8]. The harmonized database provides an optimal resource for designing such trials when mean cortical PIB SUVR or CSF Aβ42 are used as the primary efficacy endpoints. Power analyses in Table 5 indicate that prevention trials on individuals younger than 65 years are feasible.
The harmonized database also has limitations. First, the database presents unique analytic challenges regarding heterogeneity in cohort characteristics such as family history of AD and research protocols across studies. Valid statistical analyses must analyze study-to-study variation with appropriate meta-analytic models by treating studies as a fixed or random effect, and assess the heterogeneity across studies, and include important covariates such as family history of AD. Especially for CSF biomarker analysis, we recommend analyses be conducted on data obtained from the same assay type, and further adjust for the effect of variables related to CSF collection. Second, the longitudinal follow-ups in biomarkers do not capture the full spectrum of within-individual progression. Third, the large database remains a collection of convenience samples, likely not representative of the real world population. For instance, the combined cohort is predominantly well-educated Caucasians, findings from the database may not generalize to other populations.
Supplementary Material
We integrate and harmonize longitudinal biomarkers and cognitive databases from 8 high-quality longitudinal studies of AD to investigate the natural history of preclinical AD from young adulthood and to design future prevention trials;
The harmonized database provides excellent statistical power for detecting very early biomarker changes on middle age individuals;
The harmonized database demonstrates the feasibility of powering future AD prevention trials on middle age individuals.
5. Acknowledgements
This study was supported by National Institute on Aging (NIA) grant R01 AG053550 (Dr. Xiong) and NIA grant P50 AG005681, P01AG026276, and P01 AG0399131 (Dr. Morris), UF1AG032438 (Dr. Bateman), U19-AGO33655 and R01 AG059869 (Albert), R01-AG027161 and R01 AG021155 (Johnson), P50AG033514 (Asthana), and Australian Commonwealth Scientific Industrial Research Organization (Masters).The AIBL study (www.AIBL.csiro.au) was supported by the Alzheimer’s Association (US), the Alzheimer’s Drug Discovery Foundation, an Anonymous foundation, the Science and Industry Endowment Fund, the Dementia Collaborative Research Centres, the Victorian Government’s Operational Infrastructure Support program, the McCusker Alzheimer’s Research Foundation, the National Health and Medical Research Council, and the Yulgilbar Foundation, plus numerous commercial interactions that supported data collection and analysis. Image processing was supported in part by the Neuroimaging Informatics and Analysis Center (1P30NS098577) and R01 EB009352.
Data collection and sharing for this project was supported by The Dominantly Inherited Alzheimer’s Network (DIAN, UF1AG032438) funded by the National Institute on Aging (NIA), the German Center for Neurodegenerative Diseases (DZNE), Raul Carrea Institute for Neurological Research (FLENI), Partial support by the Research and Development Grants for Dementia from Japan Agency for Medical Research and Development, AMED, and the Korea Health Technology R&D Project through the Korea Health Industry Development Institute (KHIDI). This manuscript has been reviewed by DIAN Study investigators for scientific content and consistency of data interpretation with previous DIAN Study publications. We acknowledge the altruism of the participants and their families and contributions of the DIAN research and support staff at each of the participating sites for their contributions to this study
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
References
- [1].Xiong C, Jasielec MS, Weng H, Fagan AM, Benzinger TL, Head D, et al. Longitudinal relationships among biomarkers for Alzheimer disease in the Adult Children Study. Neurology. 2016;86:1499–506. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Price JL, Morris JC. Tangles and plaques in nondemented aging and "preclinical" Alzheimer's disease. Ann Neurol,1999; 45(3):358–368. [DOI] [PubMed] [Google Scholar]
- [3].Sperling RA, Aisen PS, Beckett LA, Bennett DA, Craft S, Fagan AM, Iwatsubo T, Jack CR Jr, Kaye J, Montine TJ, Park DC, Reiman EM, Rowe CC, Siemers E, Stern Y, Yaffe K, Carrillo MC, Thies B, Morrison-Bogorad M, Wagster MV, Phelps CH. 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. May; 7(3):280–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Villemagne VL, Burnham S, Bourgeat P, et al. , for the Australian Imaging Biomarkers and Lifestyle Research Group. Amyloid β deposition, neurodegeneration, and cognitive decline in sporadic Alzheimer’s disease: a prospective cohort study. Lancet Neurology 2013; 12:357–367. [DOI] [PubMed] [Google Scholar]
- [5].Bischof GN, Rodrigue KM, Kennedy KM, Devous MD, Park DC. Amyloid deposition in younger adults is linked to episodic memory performance. Neurology. 2016;87(24):2562–2566. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Gonneaud J, Arenaza-Urquijo EM, Mézenge F, Landeau B, Gaubert M, Bejanin A, de Flores R, Wirth M, Tomadesso C, Poisnel G, Abbas A, Desgranges B, Chételat G (2017) Increased florbetapir binding in the temporal neocortex from age 20 to 60 years. Neurology 89, 2438–2446. [DOI] [PubMed] [Google Scholar]
- [7].Bateman RJ, Benzinger TL, Berry S, Clifford DB, Duggan C, Fagan AM, et al. The DIAN-TU Next Generation Alzheimer's prevention trial: Adaptive design and disease progression model. Alzheimer's & Dementia. 2017;13:8–19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].McDade E, Bateman RJ. Stop Alzheimer's before it starts. Nature. 2017; 12;547(7662):153–155. [DOI] [PubMed] [Google Scholar]
- [9].Xiong C, Roe CM, Buckles V, Fagan AM, Holtzman D, Balota D, Duchek J, Storandt M, Mintun M, Grant E, Snyder AZ, Head D, Benzinger TLS, Mettenburg J, Csernansky J, Morris JC. Role of family history for Alzheimer biomarker abnormalities in the adult children study. Archives of Neurology 2011; 68(10):1313–1319. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Morris JC. The Clinical Dementia Rating (CDR): current version and scoring rules. Neurology, 1993;43(11):2412–2414. [DOI] [PubMed] [Google Scholar]
- [11].Morris JC, Storandt M, Miller JP, McKeel DW, Price JL, Rubin EH, Berg L. Mild cognitive impairment represents early-stage Alzheimer disease. Arch Neurol. 2001;58(3):397–405. [DOI] [PubMed] [Google Scholar]
- [12].Albert M, Soldan A, Gottesman R, McKhann G, Sacktor N, Farrington L, et al. Cognitive changes preceding clinical symptom onset of mild cognitive impairment and relationship to ApoE genotype. Current Alzheimer Research, 2014;11(8):773–784. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Johnson SC, Koscik RL, Jonaitis EM, Clark LR, Mueller KD, Berman SE, et al. The Wisconsin Registry for Alzheimer’s Prevention: A review of findings and current directions. Alzheimer’s & Dementia: Diagnosis, Assessment & Disease Monitoring, 2018;10:130–142. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Petersen RC, Doody R, Kurz A, Mohs RC, Morris JC, Rabins PV, et al. Current concepts in mild cognitive impairment. Arch Neurol, 2001;58(12):1985–1992. [DOI] [PubMed] [Google Scholar]
- [15].Ellis KA, Bush AI, Darby D, De Fazio D, Foster J, Hudson P, et al. The Australian Imaging, Biomarkers and Lifestyle (AIBL) study of aging: methodology and baseline characteristics of 1112 individuals recruited for a longitudinal study of Alzheimer's disease. Int Psychogeriatr, 2009;21(4):672–687. [DOI] [PubMed] [Google Scholar]
- [16].Bateman RJ, Xiong C, Benzinger TL, Fagan AM, Goate A, Fox NC, et al. Dominantly Inherited Alzheimer Network. Clinical and biomarker changes in dominantly inherited Alzheimer's disease. N Engl J Med, 2012; 367(9):795–804. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Morris JC, Weintraub S, Chui HC, et al. The uniform data set (UDS): Clinical and cognitive variables and descriptive data from Alzheimer disease centers. Alz Dis Assoc Dis. 2006;20(4):210–216. [DOI] [PubMed] [Google Scholar]
- [18].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: 263–269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Gross AL, Hassenstab JJ, Johnson SC, Clark LR, Resnick SM, Kitner-Triolo M, et al. A classification algorithm for predicting progression from normal cognition to mild cognitive impairment across five cohorts: The preclinical AD consortium. Alzheimer's & Dementia: Diagnosis, Assessment & Disease Monitoring, 2017; 8:147–155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Dowling NM, Hermann B, La Rue A, Sager MA. Latent structure and factorial invariance of a neuropsychological test battery for the study of preclinical Alzheimer's disease. Neuropsychology. 2010; 24(6):742–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Storandt M, Balota DA, Aschenbrenner AJ, Morris JC. Clinical and psychological characteristics of the initial cohort of the Dominantly Inherited Alzheimer Network (DIAN). Neuropsychology. 2014. January; 28(1):19–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Fagan AM, Roe CM, Xiong C, Mintun MA, Morris JC, Holtzman DM. Cerebrospinal fluid tau/beta-amyloid(42) ratio as a prediction of cognitive decline in nondemented older adults. Arch Neurol. 2007; 64(3):343–9. [DOI] [PubMed] [Google Scholar]
- [23].Shaw LM, Vanderstichele H, Knapik-Czajka M, Clark CM, Aisen PS, Petersen RC, Blennow K, Soares H, Simon A, Lewczuk P, Dean R, Siemers E, Potter W, Lee VM, Trojanowski JQ; Alzheimer's Disease Neuroimaging Initiative. Cerebrospinal fluid biomarker signature in Alzheimer's disease neuroimaging initiative subjects. Ann Neurol. 2009; 65(4):403–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Moghekar A, Li S, Lu Y, Li M, Wang MC, Albert M, O'Brien R, BIOCARD Research Team. CSF biomarker changes precede symptom onset of mild cognitive impairment. Neurology, 2013. November 12; 81(20): 1753–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [25].Bendlin BB, Carlsson CM, Johnson SC, Zetterberg H, Blennow K, Willette AA, Okonkwo OC, Sodhi A, Ries ML, Birdsill AC, Alexander AL, Rowley HA, Puglielli L, Asthana S, Sager MA. CSF T-Tau²42 predicts white matter microstructure in healthy adults at risk for Alzheimer's disease. PLoS One. 2012; 7(6):e37720. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Bittner T, Zetterberg H, Teunissen CE, Ostlund RE Jr., Militello M, Andreasson U, et al. Technical performance of a novel, fully automated electrochemiluminescence immunoassay for the quantitation of beta-amyloid (1-42) in human cerebrospinal fluid. Alzheimer's & dementia, 2016;12:517–26. [DOI] [PubMed] [Google Scholar]
- [27].Pike KE, Savage G, Villemagne VL, Ng S, Moss SA, Maruff P, Mathis CA, Klunk WE, Masters CL, Rowe CC. Beta-amyloid imaging and memory in non-demented individuals: evidence for preclinical Alzheimer's disease. Brain. 2007. November; 130(Pt11):2837–44. [DOI] [PubMed] [Google Scholar]
- [28].Gordon BA, Blazey T, Su Y, et al. Longitudinal beta-Amyloid Deposition and Hippocampal Volume in Preclinical Alzheimer Disease and Suspected Non-Alzheimer Disease Pathophysiology. JAMA Neurol. 2016;73(10):1192–1200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [29].Johnson SC, Christian BT, Okonkwo OC, Oh JM, Harding S, Xu G, Hillmer AT, Wooten DW, Murali D, Barnhart TE, Hall LT, Racine AM, Klunk WE, Mathis CA, Bendlin BB, Gallagher CL, Carlsson CM, Rowley HA, Hermann BP, Dowling NM, Asthana S, Sager MA. Amyloid burden and neural function in people at risk for Alzheimer's Disease. Neurobiol Aging. 2014. March; 35(3):576–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [30].Pettigrew C, Soldan A, Zhu Y, et al. Cortical thickness in relation to clinical symptom onset in preclinical AD. Neuroimage Clin. 2016;12:116–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Fischl B FreeSurfer. Neuroimage. 2012;62(2):774–781. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32].Su Y, D'Angelo GM, Vlassenko AG, et al. Quantitative analysis of PiB-PET with FreeSurfer ROIs. PLoS One. 2013;8(11):e73377. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [33].Sauerbrei W, Abrahamowicz M, Altman DG, le Cessie S, Carpenter J, on behalf of the STRATOS initiative. STRengthening Analytical Thinking for Observational Studies: the STRATOS initiative. Stat Med. 2014. December 30; 33(30): 5413–5432 [DOI] [PMC free article] [PubMed] [Google Scholar]
- [34].Laird NM, Ware JH. Random-effects models for longitudinal data. Biometrics. 1982. December; 38(4):963–74. [PubMed] [Google Scholar]
- [35].Xiong C, Zhu K, Yu K, Miller JP. Statistical Modeling in Biomedical Research: Longitudinal Data Analysis in Handbook of Statistics: Epidemiology and Medical Statistics, Edited by Rao CR, Philip Miller J, and Rao DC. Amsterdam: Elsevier B.V., 2008. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.