The relationship between cerebral amyloidosis as assessed by Pittsburgh compound B (PiB) positron emission tomography (PET) scan and serum levels of DHA. A, Box plot of PiB index units in patients with and without APOE ε4 allele. B, Cerebral amyloid load (as indexed by PiB index units) in participants grouped according to the quartiles of their serum DHA levels (percentage of total fatty acids). C, The association of serum DHA levels with PiB index by APOE genotype. Serum DHA explained 10% of the variability in cerebral amyloidosis (R2 = 0.1), whereas both serum DHA and APOE together explained 23% of cerebral amyloidosis (R2 = 0.23). The groups were compared by Wilcoxon t test or analysis of variance. The relationship was modeled using a linear regression. Serum DHA units are the percentage of total fatty acids.
aP < .05.
A, The correlation of hippocampal volume with serum DHA levels in 52 participants from the Aging Brain Study. Serum DHA (percentage of total fatty acids) was associated with hippocampal volume. Serum DHA explained 14% of the variability in hippocampal volumes (R2 = 0.14; P = .005). B, An association between serum DHA and left entorhinal volume was observed and explained 24% of the variability in hippocampal volume (R2 = 0.24; P < .001). The relationships were modeled using a linear regression. Brain volume is reported as a ratio of intracranial volume. Serum DHA units are reported as a percentage of total fatty acids.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Yassine HN, Feng Q, Azizkhanian I, Rawat V, Castor K, Fonteh AN, Harrington MG, Zheng L, Reed BR, DeCarli C, Jagust WJ, Chui HC. Association of Serum Docosahexaenoic Acid With Cerebral Amyloidosis. JAMA Neurol. 2016;73(10):1208–1216. doi:10.1001/jamaneurol.2016.1924
Copyright 2016 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
Is there an association between serum docosahexaenoic acid levels and cerebral amyloidosis?
In this study of 61 older adults without cognitive impairment who underwent amyloid Pittsburgh Compound B positron emission tomography imaging, participants at the lowest quartile of serum docosahexaenoic acid levels had significantly more cerebral amyloidosis.
Limited seafood intake may increase the risk of brain amyloid deposition and Alzheimer disease.
Higher dietary intake of the essential fatty acid docosahexaenoic (DHA) has been associated with better cognitive performance in several epidemiological studies. Animal and in vitro studies also indicate that DHA prevents amyloid deposition in the brain.
To determine the association between serum DHA levels, cerebral amyloidosis, and the volumes of brain areas affected by Alzheimer disease.
Design, Settings, and Participants
Cross-sectional analysis of serum DHA levels together with measures of amyloid deposition (Pittsburgh Compound B index), brain volumes, and neuropsychological testing scores from 61 participants in the Aging Brain Study. The study was conducted between June 2008 and May 2013, and the data were analyzed between October 2015 and April 2016. Linear models were adjusted for age, sex, years of education, and apolipoprotein E status.
Main Outcomes and Measures
Serum DHA levels with cerebral amyloidosis measured using PIB PET.
Samples were available from 61 Aging Brain Study participants (41 women and 20 men) who underwent amyloid PET imaging. The mean (SD) age of the participants was 77 (6) years and ranged from 67 to 88 years. Serum DHA levels (percentage of total fatty acids) were 23% lower in participants with cerebral amyloidosis than those without (0.97 vs 1.25, P = .007) and were inversely correlated with brain amyloid load (r = −0.32, P = .01) independent of age, sex, apolipoprotein E genotype, and years of education. Moreover, greater serum DHA levels were positively associated with brain volume in several subregions affected by AD, in particular the left subiculum (r = 0.38, P = .005) and the left entorhinal volumes (r = 0.51, P = .001). Serum DHA levels were also associated with nonverbal memory scores (r = 0.28, P = .03).
Conclusions and Relevance
In this study, serum DHA levels were associated with pathogenesis of cerebral amyloidosis and with preservation of entorhinal and hippocampal volumes. These findings suggest an important role for DHA metabolism in brain amyloid deposition during the preclinical or early symptomatic stages of Alzheimer disease.
Alzheimer disease (AD) is the most common form of dementia and affects 9.7% of people older than 70 years in the United States.1 There is mounting evidence that ω-3 polyunsaturated fatty acid (ω-3 PUFAs) supplementation and their levels in the blood are associated with the risk of developing AD.2 Reduced levels of ω-3 PUFAs have been observed in the brains, plasma, serum, and erythrocyte membranes of people with AD.3-5 Of special interest is docosahexaenoic acid (DHA), which has neuroprotective qualities and is enriched in gray and white matter of the brain.6-9 Docosahexaenoic acid accounts for 30% to 40% of fatty acids in cortical gray matter and is more concentrated at synapses where it may play a role in synaptic plasticity.10 Both intake and levels of ω-3 PUFAs on erythrocytes membranes have been associated with enhanced gray matter brain volumes, particularly within the hippocampus.10,11 In preclinical studies, DHA is associated with preventing amyloid plaque formation. For example, mice expressing human familial-related AD transgenes accumulate less β amyloid (Aβ) when receiving DHA supplementation.12 The mechanisms linking DHA intake and reduced Aβ plaque formation involve decreased amyloid production through regulating β- and γ-secretase expression, de novo cholesterol synthesis,13 and enhanced amyloid clearance by brain microglial cells.14
Amyloid positron emission tomography (PET) imaging, by virtue of measuring amyloid burden irrespective of clinical condition, provides direct insight into cerebral amyloidosis. Our goal was to determine whether greater serum DHA levels were associated with less cerebral amyloidosis, larger brain volumes, and better cognitive measures among patients participating in the Aging Brain Study.
Participants were recruited for a multisite research program, the Aging Brain Study, designed to assess vascular contributions to cognitive impairment with and without AD. We recruited individuals with no or mild cognitive impairment, targeting individuals with substantial vascular disease risk factors, cardiovascular events, and vascular brain injury. Inclusion criteria have previously been described in detail.15 The study was approved by the institutional review boards at University of California, Berkeley; University of California, Davis; University of California, San Francisco; and University of Southern California. Written informed consent was obtained from all participants following institutional review board–approved protocols.
Cerebral Aβ was measured with PET using the tracer11 C-labeled Pittsburgh Compound B (PiB), which is specifically retained in fibrillar Aβ plaques. The PiB radiotracer was synthesized at Lawrence Berkeley National Laboratory using a previously published protocol.16 Pittsburgh Compound B–PET imaging was conducted using a Siemens ECAT HR scanner in 3-dimensional acquisition mode (Siemens). Pittsburgh Compound B (10-15 mCi) was injected as a bolus into an antecubital vein, after which dynamic acquisition frames were obtained for 90 minutes over progressively longer intervals.17
Pittsburgh Compound B data were preprocessed using Statistical Parametric Mapping 8 (SPM8; http://www.fil.ion.ucl.ac.uk/spm/). Frames 6 through 35, as well as the sum of frames 1 to 5, were realigned to frame 17. Realigned frames reflecting the first 20 minutes of acquisition (frames 1-23) were then averaged and used to guide coregistration with the T1-weighted magnetic resonance image. Distribution volume ratio images were generated from PiB frames corresponding to 35 to 90 minutes postinjection and quantified using Logan graphical analysis and the participant’s gray matter cerebellar reference region. Cerebral Aβ was quantified using a global PiB index that averages PiB signal in brain regions with amyloidosis. Positron emission tomography scans were used to measure PiB retention as previously described.15,17 Distribution volume ratio values were extracted from regions of interest vulnerable to early Aβ deposition, which include the frontal cortex (anterior to the precentral gyrus), lateral parietal cortex, lateral temporal cortex, posterior cingulate, and precuneus. The occipital cortex was also examined owing to its susceptibility to cerebral amyloid angiopathy. Regions of interest were defined using the Desikan-Killiany atlas and the semiautomated FreeSurfer processing stream. A global measure of PiB uptake (Global PiB Index) was generated in each subject’s native space by averaging the mean distribution volume ratio value of these regions of interest. This Global PiB Index served as the primary dependent variable. For purposes of describing the sample (but not for purposes of data analysis), PiB positivity was calculated using a baseline established in 11 young adults (mean [SD] age, 24.5 [3.4] years) who underwent PiB-PET imaging using the same acquisition and processing procedures described in the previous section. Pittsburgh Compound B uptake was determined using distribution volume ratio values from the Global PiB Index. Values 2 SDs above the young average were established as defining values of PiB positivity. Therefore, in our study, participants with a Global PiB Index of 1.08 or higher were determined to be a priori to be PiB-positive, indicating significant amyloidosis.
Total serum DHA was measured by gas chromatography mass spectrometry as previously described18 in fasting serum samples. Deuterated-DHA standard (50 µL at 2 ng/µL) in ethanol was added to the samples prior to lipid extraction, which was performed using the modified Bligh and Dyer procedure. Acid hydrolysis was performed on the lipid extracts isolated from serum followed by derivatization to the pentafluorobenzyl ester as previously described and transferred to GC vials. Solutions were made by diluting the derivatized extracts into 100-µL dodecane for a final deuterated DHA standard concentration of 1 ng/µL. Then, 1 µL of sample was injected into the 7890A gas chromatography system coupled with a 7000 mass spectrometry Triple Quad (Agilent Technologies). Gas chromatography was performed using a Zebron ZB-1 Mass Spectrometry Capillary Gas Chromatography Column as previously described (Phenomenex). Single-ion monitoring was used to measure the carboxylate ions (m/z) for the deuterated (m/z = 332 for d5-DHA) and nondeuterated DHA (m/z = 327). The sum of total fatty acids was obtained by adding the absolute amounts of 15 fatty acids greater than 14 carbon detected in all samples. These included C14:0, C16:0, C18:0, C18:2, C18:1, C18:3n-3, C18:3n-6, C20:4n-6, C20:3n-3, C20:2n-6, C22:6n-3, C22:4n-6, C22:5n-3, C22:2n-6, and C22:3n-3. Docosahexaenoic acid levels were then expressed as a percentage of the total fatty acid for each participant. The samples were run in duplicate. The intra-assay coefficient of variation was 1.5%.
All participants underwent magnetic resonance imaging using a 3-T scanner (Magnetom Trio System; Siemens) with an 8-channel head coil. Acquired images included a T1-weighted, volumetric, magnetization-prepared rapid gradient-echo image (repetition time, 2500 milliseconds; echo time, 2.94 or 2.98 milliseconds; inversion time, 1100 milliseconds) and a fluid-attenuated inversion recovery image (repetition time, 5000 milliseconds; echo time, 403 milliseconds; inversion time, 1700 milliseconds; 1.0×1.0 mm2 in-plane resolution with 1.00-mm thickness). Regional brain areas were normalized to measures of total intracranial volumes. Hippocampal volumes passed quality control measures in 52 of the 61 participants.
All participants received a standardized neuropsychological test battery from which linear measures of global cognition, verbal and nonverbal memory, and executive function were derived using item-response theory, as described previously.15 Scale development used 400 elderly individuals with cognitive function ranging from normal to having dementia. Donor items for the global measure came from the first 2 learning trials of the Memory Assessment Scale List Learning Test, Wechsler Memory Scale–Revised Digit Span total raw score, letter fluency, and animal category fluency. The verbal memory scale combined short-delayed free recall, short-delayed cued recall, and immediate recall on learning trials 1 and 3 of the Memory Assessment Scale List Learning Test. The nonverbal memory scale was derived from items in the Biber Serial Design Learning Test. Donor items for executive function were the Initiation-Perseveration subscale of the Mattis Dementia Rating Scale, letter fluency, backward Digit Span, and backward Spatial Span. Each scale was transformed to a mean of 100 and an SD of 15.
Means (SDs) for normally distributed data or median (25th, 75th percentile) for nonnormally distributed data were computed. Pearson or Spearman coefficients for nonnormally distributed data were used to correlate serum DHA levels with PiB index, brain volumes, and measures of cognition. The groups were also compared using a linear regression model to allow adjusting for apolipoprotein E (APOE) group, age, sex, years of education, and clinical dementia rating score and potential interactions. Years of education were used as an index of socioeconomic status. The Statistical Program R, version 3.2.3 was used (R Programming). Significance was defined by P < .05.
Samples were available from 61 Aging Brain Study participants (41 women and 20 men) who underwent amyloid PET imaging. The mean (SD) age of the participants was 77 (6) years and ranged from 67 to 88 years. Thirty participants had a clinical dementia rating score of 0, while 29 participants had a score of 0.5, and 2 participants presented with a clinical dementia rating score greater than 0.5. Because the Aging Brain Study focused on vascular contributions to cognitive impairment, many participants were overweight with prediabetes or treated diabetes. Participants displayed the characteristic diabetic dyslipidemia of elevated triglyceride levels and low high-density lipoprotein (HDL) cholesterol. Apolipoprotein E genotype was available for 59 of the 61 participants with 13 participants (22%) having the APOE ε4 allele. The positive amyloid group was defined a priori by PiB index greater than 1.08. Carriers of the APOE ε4 allele had significantly more cerebral amyloidosis by PiB index: mean (SD) ε4 carriers, 1.23 (0.21) vs mean (SD) ε4 noncarriers, 1.09 (0.14); P = .03 (Figure 1A). Serum DHA levels (percentage of total fatty acids) were 23% lower in participants with cerebral amyloidosis than those without (0.97 vs 1.25, P = .007) and were inversely correlated with brain amyloid load (r = −0.32, P = .01) independent of age, sex, apolipoprotein E genotype, and years of education. The lowest quartile of serum DHA was significantly associated with greater cerebral amyloidosis compared with the other groups (model: R2 = 0.2, P = .005, with estimates in the first quartile differing from the remaining quartiles; Figure 1B) and serum DHA levels were inversely associated with this index (r = −0.32, P = .01, Figure 1C) in both APOE ε4 carriers and noncarriers.
The study was divided into 2 subgroups on the basis of positive or negative PiB index. The participant characteristics in these 2 groups are summarized in Table 1. Serum DHA levels (percentage of total fatty acids) and APOE group were the only independent predictors of cerebral amyloidosis. This association of serum DHA with PiB index persisted after adjusting for APOE genotype. The estimates, standard errors, and P values for the univariate and multivariate model of cerebral amyloidosis as the dependent variable, with serum DHA levels and APOE ε4 status, are summarized in Table 2. We previously reported that PiB index was associated with low-density lipoprotein (LDL) and HDL cholesterol in 74 participants of the Aging Brain Study.17 In this study (n = 61), the relationships of PiB index with HDL cholesterol (P = .93) and LDL cholesterol (P = .75) were not significant. The levels of HDL cholesterol were weakly correlated with levels of serum DHA levels (r = 0.21; P = .09).
Serum DHA levels (percentage of total fatty acids) were associated with hippocampal and entorhinal brain volumes, 2 regions affected in AD. Greater serum DHA levels were associated with larger hippocampal volume measurements (r = 0.38; P = .005; n = 52; Figure 2A). Within the hippocampal subfields, the strongest association with serum DHA levels was observed with the volume of the left subiculum (r = 0.41; P = .002). Serum DHA levels also correlated with the right CA4 (r = 0.31; P = .02), left CA4 (r = 0.33; P = .01), and presubicular (r = 0.34; P = .01) volumes. Several other regional brain volumes outside of the hippocampus were also correlated with serum DHA. The strongest correlations were observed with the left entorhinal volume (r = 0.51; P < .001; Figure 2B) and left inferior temporal volume (r = 0.3; P = .03). Serum DHA levels were also associated with the volumes of the percuneus on both the right (r = 0.36; P = .008) and left (r = 0.33; P = .01) hemispheres and with total cortical gray matter (r = 0.33; P = .01). There was no significant association between DHA and white matter volume (P = .39) or white matter lesions (P = .39). There was no association between serum DHA levels and the primary motor cortex area (P = .76). The primary motor cortex area is relatively spared in AD pathology.
Measures of global cognition, executive function, and verbal and nonverbal memory were assessed in relation to serum DHA levels. A significant association was observed between serum DHA levels and nonverbal memory (r = 0.28; P = .03). This association persisted after adjusting for age (P = .02) but not after adjusting for APOE genotype (P = .10). Serum DHA levels were not associated with measures of global cognition (r = 0.16; P = .22), executive functions (r = 0.12; P = .38), or verbal memory scores (r = 0.09; P = .50).
Lower serum DHA levels were associated with greater cerebral amyloidosis. This association was driven by participants in the lowest quartile of serum DHA and was independent of APOE genotype and traditional vascular risk factors. Moreover, higher serum DHA levels correlated with relative preservation of brain volumes in regions typically involved in AD, particularly the left subiculum of the hippocampus and the left entorhinal area.
Measurements of DHA in this study are reasonably consistent with those previously reported in the literature. The mean (SD) of DHA levels (expressed as a percentage of total fatty acid) was calculated to be 1.14 (0.43). In comparison with these results, a group of participants representing the population of Costa Rica had a mean percentage plasma DHA of 1.4919 and the Nurses’ Health Study20 calculated a mean of 1.5 in women averaging 60 years old. While our measures of DHA as a percentage of total fatty acids were lower than previously published findings, our gas chromatography mass spectrometry assay is highly sensitive and can measure fatty acids that were not detected in previous studies. A more sensitive assay likely accounts for the lower DHA levels that we report. Additionally, some published studies were conducted with plasma while our measurements were in serum, which may have a different lipid composition.
The Aging Brain Study was designed to determine the contributions of vascular risk factors to cognitive decline. We reported that cerebral amyloidosis was associated with low HDL and elevated LDL cholesterol.17 We also reported the combined effects of vascular risk factors and cerebral amyloidosis on cortical thickness.21 Moreover, in the participants with positive PiB scans, higher HDL cholesterol levels were associated with relative preservation of cortical thickness. In this study, the association of HDL and LDL cholesterol with PiB index was not significant (owing to the smaller sample size). High-density lipoprotein particles are highly enriched in phospholipids that transport esterified DHA among a large species of lipids22 and serve as one of the DHA stores in circulation. Thus, the analyses of serum DHA levels are consistent with and extend our previous findings based on HDL cholesterol.
The direction of the relationship between DHA levels and brain volumes from this study is consistent with the literature. In agreement with our findings, greater levels of ω-3 PUFAs in blood were associated with larger brain volumes.10,11,23 In the Women’s Health Initiative Memory Study cohort, Pottala et al10 reported that in 1111 postmenopausal women, an increase in red blood cell eicosapentaenoic acid and DHA levels (ω-3 index) by 1 SD was significantly associated with greater brain volume 8 years later.10 In another study of 65 healthy male and female participants between 50 and 75 years of age, participants randomly allocated to a 26-week trial of ω-3 supplementation had significantly less loss of gray matter brain volumes compared with the placebo arm.23 Additionally, a 3-year retrospective study showed that gray matter and hippocampal volumes were also maintained in a group of 799 elderly people taking ω-3–rich fish oil supplementation who presented with normal cognition, mild cognitive impairment, or AD.11 However, fish oil supplementation was only associated with larger brain volume outcomes in APOE ε4 noncarriers.11 In a separate trial, benefits from the use of high-dose vitamin B supplementation (folic acid, B6, and B12) on brain volume was demonstrated only in participants with greater ω-3 levels in plasma.24 This suggests a complex relationship between the metabolism of ω-3 and vitamin B supplementation. Consumption of the Mediterranean diet enriched with ω-3 and antioxidant-rich nuts (with presumably higher levels of DHA) compared with a saturated fatty diet was also associated with less brain atrophy.2 Combined, these studies indicate a potential role for diets rich in ω-3 on brain volumes.
Although serum DHA may simply be a marker of a healthy diet rather than the driver of an effect on amyloidosis or neurodegeneration, there is evidence of antiamyloidogenic effects of DHA in vitro and in animal studies.12-14 By decreasing β- and γ-secretase expression and stabilizing α-secretase stability, DHA is associated with reduced amyloidogenic processing and increased nonamyloidogenic processing in vitro.13 Docosahexaenoic acid also influences Aβ formation by reducing de novo cholesterol synthesis via 3-hydroxy-3-methylglutaryl-coA reductase activity.13 Furthermore, DHA treatment reduces distribution of cholesterol to raft fractions of the plasma membrane, reducing the membrane environment that favors excess amyloid production.13 β amyloid clearance by microglia in the brain is also increased after DHA supplementation.14 Therefore, greater serum DHA could indicate higher DHA concentrations in the brain, increasing resilience to the toxic effects of soluble Aβ and preventing the accumulation of fibrillary amyloid measured by PiB.
The positive associations between serum DHA and preservation of volume in the medial temporal lobe are novel and intriguing. The entorhinal cortex and subiculum are the earliest regions to develop neurofibrillary tangles and tauopathy in normal aging and AD. These findings support future studies investigating the relationship between DHA and tauopathy. In contrast, there was no association between DHA levels and the primary motor cortex, an area spared in AD. This suggests that DHA confers some resilience against the AD disease process rather than a generalized effect throughout the brain. This regional specificity might be explained by higher turnover of DHA in areas of the brain involved in synapse and memory formation. This is supported by observations that the hippocampus has a high concentration of APOE receptors.25 The hippocampus depends on APOE lipoprotein particles for DHA uptake, where it plays a role in synaptic plasticity and memory formation.26,27
Docosahexaenoic acid is not effective in the treatment of symptomatic AD,28,29 but several studies suggest that DHA supplementation slows or prevents cognitive decline in cognitively healthy older adults.30-32 One potential factor that can influence the effect of DHA supplementation on cognitive outcomes and brain volumes is APOE genotype. There is evidence that APOE ε4 proteins in the brain are hypolipidated (carry less lipids) compared with APOE ε2 or APOE ε3.33 Transgenic knock-in mice with the human ε4 allele had lower delivery of 14C-labeled DHA to the brain compared with mice with human ε2 and ε3 alleles.27 In these mice, AD pathology was reversed by DHA supplementation.34 In humans, the APOE ε4 allele worsens the severity of AD and may modulate the response to ω-3 supplementation. The major studies investigating ω-3 and cognitive outcomes are summarized in Table 3. Several studies suggest a timing by APOE genotype effect for ω-3 intervention on cognitive decline. For example, in 2 studies that included participants with AD,11,28 no cognitive benefit was observed in carriers of the ε4 allele. In these 2 studies, benefit from ω-3 use was evident only in ε4-negative participants. In contrast, 3 studies in participants without AD at presentation (Multidomain Alzheimer Preventive Trial [MAPT] trial,30 Chicago Health and Aging Project,39 and the Chicago Brain Autopsy Study42) suggest cognitive benefit in ε4 carriers with DHA supplementation, greater ω-3 levels, or weekly seafood consumption. However, as evidenced by Table 3, there is a significant heterogeneity in the response to ω-3 supplementation.
One additional factor explaining this heterogeneity in response is the etiology of dementia. The relationship between amyloidosis and dementia is complex. Dementia can present without amyloid pathology48,49 and cerebral amyloidosis can present in older adults without dementia.50 The use of PiB imaging can help identify participants at risk of amyloid-associated dementia, who could potentially benefit from ω-3 supplementation. Based on our study findings, we anticipate that cognitively healthy older participants at the lowest quartile of serum DHA levels (persons with limited seafood consumption) would be a reasonable target for an ω-3 supplementation trial. Indeed, a preplanned analysis of the MAPT trial30 (presented at Clinical Trials in Alzheimer Disease Conference, San Diego, California, 2015) indicated cognitive benefit in participants at the lowest quartile of red blood cell DHA who received 800 mg/d of DHA supplementation.
This study has some limitations. First, we were limited by a small sample size and were likely underpowered to detect significant associations between serum DHA and some measures of brain volume and cognitive scores. Second, the study participants were enriched for vascular risk factors and vascular brain injury, which is present in 15% to 46% of the elderly population based on evidence of infarcts at autopsy.51 A study enriched with individuals with extensive amyloidosis might present with different results28 in light of evidence that significant cerebral amyloidosis can limit the delivery of DHA to the brain.52,53
We report an association between greater serum DHA levels and less cerebral amyloidosis as well as higher volumes of several subregions of the brain affected in AD in cognitively healthy older adults. These findings stimulate interest toward understanding the complexity of brain ω-3 fatty acid metabolism, with particular focus on the mechanisms regulating their delivery to the brain and their involvement in neuronal survival, function, and synaptogenesis.
Corresponding Author: Hussein N. Yassine, MD, Division of Endocrinology, Department of Medicine, University of Southern California, 2250 Alcazar St, Room 210, Los Angeles, CA 90033 (firstname.lastname@example.org).
Accepted for Publication: April 29, 2016.
Published Online: August 8, 2016. doi:10.1001/jamaneurol.2016.1924
Author Contributions: Dr Yassine 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: Yassine.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Yassine.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Yassine, Zheng.
Obtained funding: Yassine, Harrington, Reed, Jagust, Chui.
Administrative, technical, or material support: Yassine, Feng, Azizkhanian, Rawat, Castor, DeCarli, Jagust, Chui.
Study supervision: Yassine, Fonteh, Harrington, Jagust.
Conflict of Interest Disclosures: Dr Jagust has served as a consultant to Genentech, Novartis, Banner Alzheimer Institute, and Bioclinica. No other disclosures are reported.
Funding/Support: Dr Yassine was supported by grant K23HL107389 from the National Institutes of Health. Drs Castor and Fonteh and Mr Harrington were supported by the LK Whittier Foundation and Huntington Medical Research Institutes. This work was also supported by grants AG012435 (Drs Chui, Jagust, Reed, DeCarli, and Zheng), AG031563 (Dr Reed), AG05142 (Dr Chui), and AG10129 (Dr DeCarli) from the National Institutes of Health.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, or interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.