Knopman, David S.

AbstractBackgroundTau phosphorylated at threonine 217 (pTau217) is one of the most promising blood‐based biomarkers of Alzheimer’s disease (AD). This study compares the performance of two plasma pTau217 immunoassays on the Simoa platform for detection of abnormal amyloid‐PET.MethodPlasma samples from 112 individuals without cognitive impairment and 114 with mild cognitive impairment (MCI) or AD related dementia were evaluated. Testing was performed on the Quanterix Simoa® HD‐X Analyzer using the ALZpath Simoa® pTau217 and the Simoa® ACC pTau217 (Janssen antibodies) commercial assays. Amyloid‐PET imaging was performed using 11C‐Pittsburgh Compound B and a standard uptake value ratio (SUVR) cutoff of ≥1.52 (≥25 Centiloid) was used for abnormal (n= 147) vs. normal (n= 79) classification. Passing‐Bablok regression analysis and Spearman’s correlation coefficient (rho) were used to estimate bias and correlation between the assays, respectively. Correlation of each assay with Amyloid‐PET SUVR was evaluated. Amyloid‐PET status as the outcome and plasma pTau217 as predictor was used to estimate the area under the receiver operating characteristic (ROC) curve (AUC).ResultThe pTau217 concentrations between the two assays showed an excellent correlation (Spearman’s rho= 0.94 [95% CI= 0.93 to 0.96, p<0.0001]). However, the Passing‐Bablok regression analysis equation for ALZpath (x‐axis) and ACC (y‐axis) was y= 0.1324x ‐ 0.0019, showing that measured pTau217 concentrations with the ALZpath assay were higher (∼87%) than with the ACC assay (concentrations ranged 0.057 to 3.5 pg/mL and 0.003 to 0.5 pg/mL, respectively). Correlations between amyloid‐PET SUVR and the two pTau217 immunoassays were almost identical with Spearman’s rho= 0.70 (95% CI= 0.62 to 0.76, p<0.0001) for ALZpath and 0.69 (95% CI= 0.61 to 0.75, p<0.0001) for ACC. The AUC for detection of an abnormal amyloid‐PET among the MCI/AD group was 0.88 (95% CI= 0.81 to 0.95) for ALZpath, and 0.89 (95% CI= 0.83 to 0.96) for ACC.ConclusionAnalytically, the two Simoa® pTau217 assays correlate well with each other, but results are not interchangeable due to a significant proportional bias. Clinically, the performance of the assays was largely equivalent for the detection of abnormal amyloid‐PET by AUC analysis based on their almost identical correlations with amyloid‐PET SUVR.

AbstractBackgroundAlzheimer’s disease (AD) patients have decline in cognitive domains including memory, language, visuospatial, and/or executive function and brain pathology including amyloid‐β and tau deposition, neurodegeneration, and frequent vascular co‐pathologies detectable by neuroimaging and/or cerebrospinal fluid biomarkers. However, molecular disease mechanisms are complex and heterogeneous. It is necessary to develop cost‐effective blood‐based biomarkers reflecting brain molecular perturbations in AD. We identified blood‐based gene and co‐expression network level changes associated with AD/mild cognitive impairment (MCI) diagnosis and AD‐related phenotypes.MethodWe performed differential gene expression and weighted gene co‐expression network analysis, followed by meta‐analysis, using blood transcriptome data of 391 participants from the Mayo Clinic Study of Aging and 654 participants from the Alzheimer's Disease Neuroimaging Initiative. The neuroimaging phenotypes include microhemorrhages, infarcts, amyloid burden, hippocampal volume, and white matter hyperintensities. The cognitive phenotypes include standardized cognitive subtest scores and composite scores for memory, language, visuospatial, and executive function.ResultFive out of 18 modules(M) are significantly associated with diagnosis or cognition (FDR‐adjusted p<0.05). M1 and M15 both positively associates with memory, M1 positively associated with language and M15 with visuospatial function. M1 and M15 are enriched in differentially expressed genes (DEGs) associated with language and executive function, respectively. M2 negatively associates with logical memory delayed recall scores(LMDR), memory, executive, and language functions and is enriched in DEGs for these phenotypes. M8 negatively associates with memory, language and executive functions and is enriched in DEGs for memory and language. M12 positively associates with LMDR. M1 and M15 are down‐regulated while M2 and M8 are up‐regulated in AD/MCI patients. Cell‐type enrichment analysis showed M2 is enriched in monocytes and neutrophils; M8 in monocytes; M15 in B cells (FDR <0.05). Gene ontology terms enriched in these modules indicated broad consistency with their cell types.ConclusionWe identified five modules significantly associated with AD/MCI or cognitive phenotypes using blood transcriptome data. These findings nominate blood transcriptome changes and their enriched biological processes as potential pathomechanisms in cognitive decline and AD/MCI development. We aim to investigate these blood transcripts as potential biomarkers for AD or AD‐related phenotypes and therapeutic targets through additional replication and experimental validation studies.

AbstractBackgroundOlder adults with dual cognitive and mobility impairments have higher financial costs and poorer quality of life than adults with either impairment alone. Blood biomarkers of Alzheimer Disease (AD) pathology (Aβ42, Aβ40 and p‐tau181) and neurodegeneration (neurofilament light (NFL) and glial fibrillary acidic protein (GFAP)) may identify individuals at risk for both mobility and cognitive impairment and provide novel insights into mechanistic underpinnings.MethodBlood biomarkers (SiMoA Quanterix N4PE, p‐tau181 single‐plex) were available in a subsample of 1751 ARIC study participants at Visit 5 (V5, 2011‐13, mean age 76.2 years (5.3), 41% men, 28% Black); Aβ42 was reversed and adjusted for Aβ40 (‐Aβ42|40) for consistent directions of associations (higher=worse). Usual gait speed and cognitive status were assessed over 8 years at V5, V6 (2016‐17), and V7 (2017‐18). Participants were categorized by V5 impairment: mobility‐only (gait speed <0.8m/s, n=430), cognitive‐only (expert‐adjudicated dementia, n=38), dual (n=54), or neither (referent, n=1229). Multinomial regression models examined relative risk ratios (RRR) of V5 log‐transformed biomarkers with prevalent impairments and, among those with neither impairment at V5, incident impairments through V7. All models adjusted for V5 age, sex, self‐reported race (Black or White), renal function, and body mass index.ResultHigher levels of all biomarkers were associated with greater risk of prevalent impairment in mobility, cognitive, or dual domains (Figure A). Over 8‐years of follow‐up, incident impairments included 471 participants with only mobility impairments, 39 with only cognitive impairments, 55 with dual impairments, and 168 died without developing any impairment. All biomarkers were associated with incident dual impairment per IQR higher biomarker: p‐tau181 (RRR=2.54; 95% CI: 1.45, 4.44), ‐Aβ42|40 (RRR=1.66; 1.31, 2.09), NfL (RRR=2.69; 1.22, 5.91) and GFAP (RRR=2.13; 1.07, 4.25). P‐tau181, ‐Aβ42|40, and GFAP were also associated with incident cognitive‐only impairment, while P‐tau181, NFL and GFAP were associated with incident mobility‐only impairment. (Figure B).ConclusionBlood biomarkers of brain health may be useful tools for earlier identification of those at risk for cognitive, mobility, or dual declines, facilitating earlier targeted interventions. Potential differences in AD‐pathology versus neurodegeneration blood biomarker relationships with outcomes of mobility and cognition may help to identify more targeted interventions for both.

AbstractBackgroundChronic medical conditions increase dementia risk, and the mechanism may be captured computationally by measures of accelerated brain aging. We hypothesized that quantifying the impact of chronic conditions on GM and WM aging will aid in measuring the differential impact of each condition on brain aging and shed light on the mechanisms through which they increase dementia risk. We specifically focused on microstructural brain changes with Neurite Orientation Dispersion and Density Imaging (NODDI) based on advanced diffusion MRI ‐ neurite density index (NDI), free water fraction (FWF), and orientation dispersion index (ODI).MethodWe identified 1123 participants [913 cognitively unimpaired (CU), 114 with MCI, 96 with AD] from the Mayo Clinic Study of Aging (MCSA) and Mayo ADRC with multi‐shell diffusion MRI. MCSA participants had chronic medical conditions’ diagnosis (based on US Dept of HHS) extracted from the electronic health care records. Using the BorutaSHAP algorithm on NDI, ODI, and ISOVF features from GM and WM separately, we constructed two distinct brain age models for GM and WM. First, we evaluated the usefulness of the predicted age gap (PAG), which is the difference between predicted age by model and chronological age, for both WM and GM NODDI models in CU/MCI/AD. We then developed separate GM and WM models for each condition using samples from condition‐negative participants to compare PAG of GM and WM for that condition.ResultThere was greater PAG in GM (avg = ) than WM (avg = 9.15 yrs) in MCI/AD participants (Fig. 1A). Each condition shifted the GM PAG by 0.10 to 3.56 years and WM PAG by 0.38 to 3.72 years with stroke and congestive heart failure showing the greater PAG overall for GM and WM PAD respectively (Fig. 2). COPD, stroke, chronic kidney disease, hypertension had comparatively greater GM PAG than WM PAG; while hyperlipidemia, congestive heart failure and coronary artery disease had greater WM PAG than GM PAG (Fig. 1B).ConclusionChronic conditions are associated with increased GM and WM Brain aging models can aid in quantifying the impact of chronic conditions on dementia risk and understanding mechanisms.

AbstractBackgroundWeakened white matter (WM) integrity is highly associated with dementia risk. Still, not everyone with WM changes develops dementia, suggesting the important role modifiable lifestyle factors may have in reducing dementia risk. We investigated how social relationships in mid‐life may modify the association between WM integrity and incident dementia risk within race and sex subgroups.MethodIn 1,679 Atherosclerosis Risk in Community (ARIC) study participants who were dementia‐free in mid‐life, psychosocial health was assessed via self‐reported questionnaires and participants were classified as having strong or poor social relationships in mid‐life (visit 2: 1990‐1992). Through diffusion tensor imaging, WM integrity was evaluated with 3T brain MRI (visit 5: 2011 – 2013); fractional anisotropy (FA) and mean diffusivity (MD) were estimated. Incident dementia cases were identified from visit 5 through December 31, 2019, with ongoing surveillance. Relative contributions of social relationships and WM integrity to incident dementia were evaluated in Cox‐proportional hazard regression models with race and race‐sex as interaction terms, and in stratified models (White Females, Black Females, White Males, Black Males).ResultSignificant three‐way interactions were observed between race, WM integrity, and mid‐life social relationships (p‐interaction<0.01). In Black participants with strong mid‐life social relationships, higher MD (hazard ratio (HR): 0.93, 95% CI: 0.63 – 1.38) and lower FA (HR: 1.25, 95% CI: 0.84 – 1.86) were not associated with the risk of dementia, although in Black participants with poor social relationships, higher MD remained a significant risk factor for dementia. However, across all White participants, higher MD and lower FA were associated with the risk of dementia, regardless of mid‐life social relationships (Table 1; Figure 1). Analysis of three‐way interactions between race‐sex, WM integrity, and mid‐life social relationships suggested that effect modification by mid‐life social relationships was most evident in Black Females, respective to other groups (Table 2; Figure 1).ConclusionThe contribution of WM integrity to dementia risk may be lower in Black participants with strong social relationships in mid‐life, particularly in Black Females. Further understanding of how contextual social factors may influence brain health in diverse populations should be examined in future studies and intervention efforts.

AbstractBackgroundThe higher prevalence and incidence of later life dementia among older Black Americans compared to older White Americans is incompletely understood and understudied. HATS is designed as a companion to the Mayo Clinic Study of Aging (MCSA) with a focus on identification of modifiable cardiovascular contributions to brain health among Black older adults living in urban areas in the upper Midwest.MethodHATS is enrolling 300 U.S. Black Minnesotans from the Twin Cities area aged 55 to 90 years without advanced dementia. Enrollment started 4/2023 and is expected to be complete by the end of 2025. Recruitment is primarily community‐based, led by two community engagement partners, the Linc and the Hue‐MAN Partnership. Electronic health record‐based recruitment will begin in 2024. Data collection includes medical comorbidities, structural and social determinants of health, cardiovascular health, functional status, physical performance, specific and non‐specific blood‐based biomarkers of Alzheimer’s disease, and comprehensive assessment of cognitive and neuropsychiatric status. Brain MRIs will be obtained on a subset of participants. A core component of HATS is distribution of meaningful health information back to participants and their community, guided by input from the HATS community engagement partners. The HATS cohort will serve as an initial effort to develop age and education adjusted normative cognitive test data for urban Black older adults in the upper Midwest.ResultAs of January 22 2024, HATS has enrolled 40 participants with a mean age of 70 years (range 51‐90) and mean education of 15 years. Mean informant‐reported Everyday Cognition scale‐12 score is 1.1 (range 1‐1.7), mean Short Test of Mental Status (STMS) score is 31 (range 24‐37), and CDR sum of boxes scores range from 0‐4 consistent with enrollment of participants with normal and mildly impaired cognition.ConclusionHATS is a novel longitudinal cohort study that is well‐positioned to advance understanding of risk factors for dementia among Black Americans living in urban areas in the upper Midwest. Recruitment is ongoing with a focus on enrolling older adults with a wide range of cognitive function. Ongoing community engagement is a central feature of HATS.

AbstractBackgroundMidlife vascular risk factors are associated with an increased risk of dementia. However, the overall contribution of modifiable vascular risk factors in midlife and late‐life to dementia remains unclear. In this study, we quantified population attributable fractions, which account for risk factor prevalence and strength of relative risks, of incident dementia from vascular risk factors measured in midlife and early late‐life.MethodWe used 33 years of prospective cohort data in the Atherosclerosis Risk in Communities Study. Incident dementia (community surveillance and algorithmic diagnosis with expert clinical adjudication) was ascertained from baseline (1987‐1989) through December 31, 2020. We quantified population attributable fractions of incident dementia by age 80 from the combination of hypertension (systolic blood pressure ≥130, diastolic blood pressure ≥80, or use of medication), diabetes (fasting glucose ≥126, non‐fasting glucose ≥200, self‐reported physician diagnosis of diabetes, or use of medication for diabetes/insulin), and self‐reported current cigarette smoking (at least one risk factor versus none) measured at different ages (45‐54 [n = 7,731], 55‐64 [n = 12,273]; 65‐74 [n = 6,660]). We also computed attributable fractions stratified by sex, apolipoprotein ε4 genotype, and self‐reported race.ResultThe prevalence of having one or more of the vascular risk factors was 61.6% (95% CI: 60.5, 62.7%) at age 45‐54 and increased to 77.5% (95% CI: 76.5, 78.5%) by age 65‐74. Population attributable fractions of dementia increased from 22.7% (95% CI: 15.3, 30.1%) for vascular risk factors measured at age 45‐54 to 45.4% (95% CI: 32.5, 58.3%) for vascular risk factors assessed at age 65‐74 (see Figure). Attributable fractions were larger in females, participants without apolipoprotein ε4 alleles, and Black participants (Figure).ConclusionBetween 23% to 45% of incident dementia cases by age 80 were attributed to a combination of hypertension, diabetes, or current smoking up through age 74. Attributable fractions for the combined impact of vascular risk varied by key demographics. Public health strategies targeting modifiable vascular risk factors could reduce dementia risk and increase healthy lifespan.

AbstractBackgroundCortical structural changes in the temporal lobes including global and hippocampal volume loss and decrease in cortical thickness are well‐associated findings linked to cognitive decline in MCI and dementia patients. Neurite orientation dispersion and density imaging (NODDI) is a multishell diffusion MRI method enabling assessment of tissue microstructural integrity at the axonal and dendritic level. We wished to determine whether changes in temporal cortex orientation dispersion index (ODI) & neurite density index (NDI) correlate with decreases in memory function.MethodThe MCALT_ADIR122 cortical atlas was used for ROI designation. NODDI imaging was obtained on a 3T Siemens Prisma scanner with 2 mm isotropic resolution. Participants (ages 60‐95 years) were cognitively unimpaired (CU) or had mild cognitive impairment (MCI). CU & MCI subjects were enrolled in the Mayo Clinic Study of Aging (MCSA) and underwent neuropsychological testing including 3 memory tests from which a composite memory z score was calculated. A generalized additive regression model was used to estimate z scores as a smooth function of ODI, NDI and cortical thickness or volume with adjustment for age, gender, education, MCSA cycle number, WMH and intracranial volume.ResultThe Table displays demographic data. Figures 1 and 2 show the relationship of NDI and ODI measures to memory z scores across participants in individual ROIs both before and after adjustment for cortical thickness (“structural MRI”) as well as for confounds listed above. Both higher cortical thickness and higher NDI values in the hippocampus and entorhinal cortex were associated with significantly higher memory z‐scores (p<0.001) (Figure 1). Higher ODI values in addition to greater cortical thickness in temporal lobe regions, including within the hippocampus, parahippocampal gyrus, entorhinal cortex and inferior temporal gyrus, were associated with higher memory z‐scores (Figure 2). These significant associations (p <=0.01) were maintained after adjusting for cortical thickness in all these regions except entorhinal cortex for ODI.ConclusionNODDI‐derived measures of microstructural integrity (NDI & ODI) within the temporal cortex may provide information about memory decline, beyond what cortical thickness analysis provides, and may be potentially useful for monitoring therapeutic effects of AD immunotherapy in the future.

AbstractBackgroundResearch subjects can potentially be re‐identified from de‐identified MRI, CT, and PET brain scans with up to 98% accuracy using Microsoft Azure’s cloud‐based commercial facial recognition software. This showed the need to “de‐face” publicly shared research brain scans. Subsequently, Microsoft has begun restricting its face recognition services, intending to prevent misuse. This study tests a variety of popular open‐source computer vision and facial recognition software packages to measure the re‐identification risk with only unrestricted free software.MethodThis study used brain MRI from Siemens 3T scanners and face photographs from 182 participants (ages 30‐90+) in the Mayo Clinic Study of Aging. Each participant had photos taken with an iPad, which were cropped loosely around the head and converted to black and white. A total of 81 renders (2D reconstruction images; Figure 1) were made from each 3D T2‐FLAIR MRI, simulating various camera and lighting positions. These face renders approximate black and white photos, allowing use with software packages developed for matching faces/features between photographs. Open‐source Python packages (OpenCV, DeepFace, face_recognition, scikit‐learn) and their included pre‐trained models (designed for photographs) were used to match participant photographs to the labelled library of MRI‐based face renders. A correct match was recorded when the top‐scoring render or set of renders automatically chosen by the algorithm was from the MRI of the correct participant.ResultFeature matching with ORB and SWIFT (both from OpenCV) produced no correct matches, despite a prior publication reporting moderate success. The best match rates (Figure 2) were 58‐59% accuracy using face_recognition or DeepFace packages with support vector machines (SVMs). Without SVMs, these packages performed modestly worse (51%‐55%).ConclusionWhile the highest match rates obtained by widely available open‐source packages (roughly 60%) are much lower than those of a limited‐access commercial product (Microsoft Azure, with 98%), they show that actors with only access to no‐cost readily available software could still re‐identify a research subject from a brain scan most of the time. This further highlights the need for image refacing software, such as mri_reface, to protect the privacy and identifies of research volunteers.

AbstractBackgroundSeveral programs are available for automatically “de‐facing” brain images. These are designed to prevent the potential use of face imagery to re‐identify research participants. Previous works have compared available software and mri_reface is frequently among the top methods. The “Minimally Invasive DeFacing of MRI images” (MiDeFace) program was more recently added to FreeSurfer, but minimal validation of MiDeFace has been published.MethodWe ran mri_reface 0.3.3 and MiDeFace 7.3.2 on two datasets: A) 207 participants with face photographs and 3D T1‐weighted and 3D T2‐weighted‐FLAIR MRI from the Mayo Clinic Study of Aging (all unimpaired, Siemens); and B) 300 participants from ADNI3 with 3D T1‐weighted MRI (50 unimpaired + 50 with clinical AD, from each of 3 vendors). For dataset A, we tested the rate at which Microsoft Azure face recognition matched each participant’s photographs with the correct MRI, comparing unmodified images, those from mri_reface, and those from MiDeFace. For dataset B, we compared measured regional brain volume and cortical thickness from original vs. de‐faced images with SPM12, FreeSurfer 7.4.1, and the FSL‐UK Biobank pipeline. For both datasets, images were visually rated to assess removal of the face and retention of brain tissue.ResultAutomated face recognition match rates were lower with mri_reface (17/207 8.2%) than with MiDeFace (24/207 11.6%), indicating stronger identity protection with mri_reface, but both were greatly reduced vs. unmodified images (203/207 98.1%). Example images with both programs are shown in Figure 1. From visual ratings (Table 1), rates of complete success (modification of the entire face while retaining the entire brain) were much higher with mri_reface, but the great majority of failures from MiDeFace were subtle and occurred in relatively less recognizable face regions. The effects of both programs on downstream imaging measurements (Figure 2) were small and very comparable, but MiDeFace was favored mildly for SPM12 and for FSL, while mri_reface was more favored for FreeSurfer.ConclusionAltogether, mri_reface more reliably removed the complete face, leading to smaller rates of potential re‐identification (8.2% vs 11.6%). Effects on downstream image measurements were small and comparable with both methods.