Aging Brain Study Research Articles

Estimating an individual's placement on a theoretical continuum using longitudinal cognitive trajectories: Relationships with longitudinal amyloid and Tau‐PET

AbstractBackgroundTheoretical models of the Alzheimer’s disease (AD) continuum posit a curvilinear trajectory of declining cognition, with accumulating AD pathology catalyzing an accelerated rate of decline. Assuming this continuum can be modelled, individuals can be placed along the curve based on their longitudinal cognitive performance. In this way, we can reduce multi‐dimensional data to a single parameter (simplification/parsimony) and remove non‐linearity in the cognitive data.Method317 participants from the Harvard Aging Brain Study (72 years (7.6); Female:61.5%) underwent longitudinal neuropsychological assessments (follow‐up: 5.9years (2.1) range:1.7‐9.1) and PIB‐PET imaging. A subset of participants (n=230) also underwent FTP‐PET imaging. Cognition was measured using the Preclinical Alzheimer Cognitive Composite (PACC). Using iterative non‐linear least squares optimization, we defined a curvilinear function (incorporating linear, quadratic and exponential components) that best described the group level PACC trajectory of all participants (Figure 1). Each subject’s longitudinal PACC trajectory was then located on the resulting curve using trust‐region‐reflective non‐linear least squares optimization, solving for an individual’s time offset (toffset). We examined associations between toffset, age, sex, apoe4 status, cross‐sectional PIB‐PET and FTP‐PET. We also examined toffset in relation to changing PIB‐PET and FTP‐PET.ResultIndividuals with a more advanced toffset had higher levels of entorhinal (EC) and inferior temporal (IT) FTP‐PET (r=0.42, p=<<0.001; Figure 2A&B.), and also showed higher PiB‐PET signal (r=0.36, p=<<0.001; Figure 2C.) PIB accumulation appears earlier in toffset followed by EC TAU and finally IT TAU, suggesting that a biomarker cascade is being captured by toffset (Figure 3). FTP‐PET and PIB‐PET did not interact to predict toffset. There was no effect of age or APOE4, or an interaction between the two, on toffset.ConclusionUsing toffset, estimating an individual’s position along a hypothetical clinical continuum can inform the unique contributions of AD biomarkers, and test hypotheses about temporal sequences of events. This data‐reduction approach is superior to that of pulling cognitive slopes, as the tpffest removes complex non‐linearities inherent in the data (Figure 4). We also illustrate that an interaction between FTP‐PET and PIB‐PET does not predict toffset, suggesting a common upstream mechanism that may underlie both Aβ and tau to drive decline, at least during the preclinical stages of disease.

Longitudinal increase in depressive symptoms in relation to neurodegeneration in clinically normal older adults: Findings from the Harvard Aging Brain Study

AbstractBackgroundEvidence suggests that depressive symptoms may be early behavioral changes preceding the development of Alzheimer’s Disease (AD). However, the neurobiological trajectory of depressive symptoms in reference to neurodegeneration of the hippocampus—a structure implicated in both mood and memory regulation—remains poorly understood in preclinical AD. In the current study, we examined the impact of baseline hippocampal volume (HV) both independently, and interactively with amyloid‐ß, on longitudinal change in depressive symptoms in clinically normal (CN) older adults.Method270 Harvard Aging Brain Study participants (age:73.5 ±6.1; 58% females) completed annual assessment with the 30‐item Geriatric Depression Scale (GDS) (average follow‐up=4.28 annual visits). All participants were cognitively unimpaired without clinically significant depression at study entry (mean GDS=3.09±2.86; range: (0‐12)). All underwent structural MRI scans and amyloid‐ß (11C‐Pittsburgh Compound B) PET at study year 1. In the primary linear mixed model, the association between dependent variable GDS (for each annual study visit) and HV was examined, with a random intercept and slope for each participant, covariates: age, sex, and education; and the interaction of predictors with time. In a secondary model, HV X cortical amyloid, X time was additionally included as a predictor.ResultIn the primary model, lower baseline HV (ß=‐0.0006; t=‐1.909; 95 % CI (‐1.2 x 10‐3, 1.7 x10‐5); p=0.057), but not its interaction with time (HV X time:(ß =‐0.000; t=‐1.12; 95% CI (‐7x10‐5, 6x10‐5); p=0.903) was marginally associated with higher GDS. In the secondary model, baseline HV and cortical amyloid were not interactive predictors of longitudinal GDS (HV X cortical amyloid X time: (ß =‐0.0001; t=‐0.90; 95% CI (‐4x10‐4, 0.0002); p=0.364).ConclusionLower baseline hippocampal volume is marginally associated with greater depressive symptoms; however, it did not independently, or in synergy with cortical amyloid, predict worsening depressive symptoms over time. These preliminary findings suggest that hippocampal atrophy may not be a predominant indicator of accelerating depression in CN older adults. Of note, our sample was non‐depressed at baseline. Future work in samples with non‐restricted behavioral symptoms and incorporating longitudinal AD biomarkers is needed to better understand neurobiological trajectory and identify targets for treatment at the earliest stages of AD.

Tracking the origin of tau spread in the brain

AbstractBackgroundAutopsy studies report that the locus coeruleus (LC) is the first site to accumulate tau, after which it spreads topographically to the entorhinal cortex (EC) and limbic regions. Once tau has reached the cortex, its spread is assumed to occur via connectivity and to be facilitated by Aβ. We aimed to examine whether lower LC integrity is related to tau accumulation in the EC, followed by spreading of tau to cingulate regions. In addition, we examined whether Aβ facilitates tau accumulation beyond these initial regions.MethodOne‐hundred‐and‐five individuals (Figure 1) from the Harvard Aging Brain Study underwent 3T‐MRI (including LC imaging),Aβ and tau‐PET imaging and neuropsychological assessments. Assessments up to 8 years prior to LC‐imaging were included (median:5.02 years). LC intensity (normalized to pontine tegmentum) was regressed onto each vertex of the cortical mantle of the PET data. Robust regressions and mediation analyses related LC intensity to tau, Aβ and cognition (composites and subtests). LME models examined interactions between LC intensity and Aβ on cognitive changes. Covariates were age, sex and education. Results were compared to LME models of tau spread via connectivity in a previously examined extended sample (n=252).ResultLC intensity was associated with medial and lateral temporal tau (peak:EC (t=4.69)). These associations became more widespread with increasing Aβ (Figure 2). This is consistent with results in the extended sample showing that spread from the EC to the posterior cingulate cortex via structural connections was greater in individuals with elevated Aβ (Figure 3). LC intensity was specifically positively associated with cross‐sectional memory (p=0.024) (Figure 4A‐B) and this was mediated by entorhinal tau (mediation: p=0.007, Figure 4C‐D). Finally, lower LC intensity predicted Aβ‐associated retrospective memory decline (Figure 5, p=0.0009). In the extended sample, lower connectivity and higher downstream‐connected tau accrual was associated with greater prospective memory decline in elevated Aβ persons (Figure 3).ConclusionLC intensity measures are associated with initial cortical Alzheimer’s disease (AD) pathology. Elevated Aβ facilitates tau spread to regions outside the medial temporal lobe and accelerates AD‐related cognitive decline. These results mimic autopsy findings of the anatomy of tau spreading and can advance early detection.

Distinct contributions of longitudinal tau and amyloid to decline in various cognitive domains in preclinical AD

AbstractBackgroundWe sought to assess the preclinical contributions of longitudinal Aβ and tau accumulation at different baseline Aβ levels to decline across multiple cognitive domains.Method124 clinically normal older adults from the Harvard Aging Brain Study with longitudinal Aβ (PIB), tau (FTP) and cognition (PACC) were included (median follow‐up=8.0±1.0 years). The following linear mixed models examined neocortical PIB and inferior temporal (IT) FTP change effects on PACC and its component tasks and included age, sex, APOE and education as covariates: 1) Time*PIB level t0*PIB change t0‐3 on cognition t0‐8, with post‐hoc regional analyses. 2) Time*PIB level t3*FTP change t3‐8 on cognition t3‐8, because FTP was introduced at t3. Mediation analyses assessed whether the effect of PIB change t0‐3 on subsequent cognitive decline t3‐8 was mediated by IT‐FTP at t3. ResultIncreasing PIB change‐t0‐3 was associated with declining cognition‐t0‐8 across all tasks at higher PIB‐t0 levels (Figure 1). The Digit Symbol Substitution Test (DSST) also declined with increasing PIB change at low PIB‐t0 levels (β=‐1.99, p=.002). FTP change‐t3‐5 was associated with declines at high PIB‐t3 levels for all PACC components except DSST (Figure 2). Only free recall (Free and Cued Selective Reminding Test; FCsrt‐FR) exhibited FTP change‐related decline in low PIB adults (β=‐1.70, p=.02). Mediation analyses demonstrated that FTP‐t3 mediated the effect of PIB change t0‐3 on cognitive decline t3‐8 for all PACC components except DSST, which remained associated only with PIB change (Figure 3). Regional analyses revealed different patterns of increasing PIB change associated with FCsrt‐FR and DSST decline (Figure 4).ConclusionTwo different patterns of cognitive decline were observed: 1) A tau‐mediated pattern of steeper decline across multiple domains in individuals with high Aβ. Free recall declined starting at low Aβ levels, in association with tau and accumulating Aβ in regions important for memory retrieval. 2) A decline on the DSST associated with accumulating Aβ, but not IT tau, even in those with low Aβ. DSST involves processing speed and executive function. The broader pattern of Aβ accumulation was associated with DSST decline suggests it may reflect a general cognitive slowing in response to Aβ‐mediated changes in synaptic transmission.

Evaluating pre‐screening tools for older Latino recruitment into preclinical Alzheimer’s disease studies

AbstractBackgroundRecruitment for preclinical Alzheimer’s disease clinical trials and observational studies in the Latino population continues to increase as researchers aim to diversify their samples. The Telephone Interview for Cognitive Status modified version (TICS‐m) is a pre‐screening tool that has shown to have high diagnostic validity for identification of dementia in mostly Caucasian samples. The TICS‐m cut‐off currently used for screening out cognitive impairment in older individuals is ≥26. We aimed to 1) determine whether currently used TICS‐m cut‐offs are appropriate for pre‐screening clinically normal older Latinos, and 2) assess the relationships among the TICS‐m score and outcome measures typically used in clinical trials, such as Mini Mental State Exam (MMSE), Logical Memory Delayed Recall, and Clinical Dementia Rating (CDR).MethodsTwenty‐nine older Latino adults (mean age= 68.7, mean education=15.6, and mean TICS‐m= 33.4/40) from the Harvard Aging Brain Study (HABS) were included in this study. Participants completed the pre‐screening TICS‐m over the phone, followed by an in‐clinic screening neuropsychological battery, which included the MMSE, Logical Memory Delayed Recall, and CDR, administered in Spanish. Linear regression models controlling for age, sex, and education were used to assess the relationships between TICS‐m score and performance on the neuropsychological measures.ResultsTICS‐m scores were positively correlated with years of education (r=0.54, p=0.002). No significant associations were observed among TICS‐m scores and other cognitive measures, after adjusting by age, sex, and education (MMSE: r=0.04, p=0.82; Logical Memory Delayed: r=0.10, p=0.62; and CDR sum of boxes: r=‐0.03, p=0.86).ConclusionPreliminary results suggest that in this sample of largely cognitively normal Latino adults, TICS‐m score is not associated with performance in other cognitive outcome measures typically used in clinical and observational trials for preclinical Alzheimer’s disease. These findings also suggest that the TICS‐m may not accurately reflect actual cognitive function in older Latino participants. However, we do not know if we are inadvertently screen‐failing participants using TICS‐m as a pre‐screening tool. Further research is needed to assess whether this pre‐screening tool is adequate in the pre‐screening process for preclinical Alzheimer’s disease in older Latino adults.

Association of tau tangle burden with depressive symptoms in community‐dwelling older adults: A longitudinal study

AbstractBackgroundDepressive symptoms are thought to be among the earliest behavioral changes in preclinical Alzheimer's disease (AD). However, the association between AD pathology and depressive symptoms in preclinical AD remains poorly understood. In the current study we examined the impact of pathology burden (amyloid and tau) on longitudinal change in depressive symptoms in cognitively unimpaired adults.Method252 adults (age: 70.8 ±8.2; 61% females) from the Harvard Aging Brain Study completed annual assessment with the 30‐item Geriatric Depression Scale (GDS) (average follow‐up=5.87 annual visits). All were cognitively normal, without clinically significant depression at study entry (mean GDS=3.19±3.12; range: (0‐16)). All underwent tau (18F‐Flortaucipir) and amyloid‐β(11C‐Pittburgh Compound B) PET at study year 3 or 4 to derive measures of inferior temporal (IT) and entorhinal cortex (EC) tau and cortical amyloid. A mixed model was employed with dependent variable GDS (for each annual study visit), a random intercept and slope for each participant, and fixed predictors: tau (IT or EC, at study year 3/4), cortical amyloid (at year closest to tau), time (time of tau PET), covariates: age, sex, education; and the interaction of predictors with time.ResultBaseline tau burden was associated with increasing GDS over time. In the model with predictor EC tau, EC tau X time was a significant predictor of GDS (β=0.48; t=3.80; 95% CI (0.23, 0.73); p=0.0002). Findings were similar for IT tau:(β = 0.71; t=3.67; 95% CI (0.33, 1.1); p=0.0002). By contrast, amyloid was not a significant predictor of longitudinal GDS in tau models: (β=‐0.14; t=‐1.58; 95% CI(‐0.23, ‐0.08); p=0.11).ConclusionBaseline tau, but not amyloid, is independently associated with increasing depressive symptoms over time. Future research incorporating longitudinal tau PET and individuals with more severe depressive symptoms is needed to delineate the temporal sequence of rising depressive symptoms and increasing tauopathy in preclinical AD.

Sex, tau, and cortical thinning in the temporal lobe: Findings from the Harvard Aging Brain Study

AbstractBackgroundRecent work suggests regional tau burden is proximal to cortical thinning in clinically‐normal older adults. Clinically‐normal older females show higher levels of temporal tau burden than males, but it remains to be seen whether the association between tau pathology and cortical thinning is significantly impacted by biological sex. We hypothesized that females with higher tau burden would show faster cortical thinning than their male counterparts.Method123 clinically‐normal older adults from the Harvard Aging Brain Study (clinical dementia rating = 0; age (SD) = 73.6(6.1); female = 58%) underwent cross‐sectional Pittsburgh Compound‐B (PiB)‐PET, Flortaucipir (FTP)‐PET and longitudinal MRI (2.95 years; range = 1.3‐5.3 years post‐tau scan). FTP‐PET was introduced after onset of the study, so the first FTP‐PET scan was used as ‘baseline’. We selected temporal tau ROIs based on our previous results (HABS & ADNI) showing sex differences in tau signal (e.g. entorhinal, inferior temporal, fusiform, middle temporal). We conducted iterative linear mixed models examining the interaction of sex*time*tauROI on cortical thinning across 34 ROIs, covarying for baseline age*time. To assess the impact of PiB on this relationship, we also examined PiB+ and PiB− groups separately (based on HABS PiB DVR cut‐off = 1.186).ResultFemales with higher inferior temporal tau burden showed significantly faster cortical thinning than males in the entorhinal cortex only, t = −2.40, p = 0.02 (see Figure 1). This result was significant in the PiB+ group (n = 33: t = −2.28, p = 0.03), but not PiB− (n = 90: t = −1.60, p = 0.11), suggesting this relationship was driven by a sex effect in PiB+ individuals (see Figure 2). We also found a similar association using fusiform tau (p = 0.03).ConclusionIn the context of high inferior temporal tau burden, females exhibit significantly faster entorhinal cortical thinning than males. This result suggests that sex may moderate the effect of tau burden on cortical thinning, and for a given level of tau burden, females may be more susceptible to cortical atrophy in temporal lobe structures. We did not adjust for multiple comparisons, so replication is required in another cohort. This relationship between tau pathology and entorhinal neurodegeneration in females provides further insight into how sex‐related cortical structural vulnerability may correspond to pathology over time.

Depression accelerates longitudinal progression of hippocampal atrophy and cognitive decline in preclinical Alzheimer’s disease

AbstractBackgroundAlthough depression is very common in the preclinical stage of Alzheimer’s disease (AD), it is still unclear whether depression moderates the progression of AD or not. In the present study, we aimed to investigate the moderation of depression on the relationship between β‐amyloid (Aβ) deposition and progression of hippocampal atrophy and cognitive decline in cognitively normal (CN) older adults.MethodTwo‐hundred eighty nine CN older adults (55–90 years; 48.4% female; 14.2% Aβ+) from an ongoing prospective cohort study, the Korean Brain Aging Study for Early Diagnosis and Prediction of Alzheimer’s Disease (KBASE), were followed up for 24 months. All participants underwent annual clinical assessments on depressive symptoms and cognitive function, and biennial [11C] Pittsburgh compound B‐PET and three‐dimensional MRI scans. Depression was defined based on the following: 1) depression diagnosed by a psychiatrist, 2) Geriatric Depression Scale score ≥ 11, or 3) the presence of any depressive symptom on Neuropsychiatric Inventory.ResultAmong the subjects, 73.4% (n = 212) were followed up for two years and twelve converted to mild cognitive impairment. Association between Aβ and depression was not observed at baseline. There was a significant Aβ positivity x Depression x Time interaction effect on hippocampal volume (F = 4.252, p = 0.042) and verbal memory (F = 7.488, p = 0.007) even after controlling for age, sex, educational level, APOE genotype, subjective memory complaint, and baseline white matter hyper intensity volume. Aβ positive individuals with depression showed faster decline of hippocampal atrophy and verbal memory decline than those in other subgroups for two years.ConclusionThe findings indicate that the presence of depression may accelerate hippocampal atrophy and episodic memory loss in older adults with preclinical AD.

Spatial navigation and verbal memory are influenced by the combined effects of APOE and BDNF Val66Met polymorphisms in mild cognitive impairment

AbstractBackgroundThe apolipoprotein E (APOE) ɛ4 allele is associated with an increased risk of Alzheimer’s disease (AD), episodic memory deficits and less accurate spatial navigation in non‐demented older adults. The brain‐derived neurotrophic factor (BDNF) Val66Met polymorphism is also related to the increased risk of AD and more pronounced memory impairment but its role in APOE ɛ4‐related spatial navigation deficits has not been established. Our aim was to examine the influence of combination of APOE and BDNF Val66Met polymorphisms on spatial navigation and other cognitive functions in individuals with amnestic mild cognitive impairment (aMCI).MethodIn total, 116 older adults with aMCI from the Czech Brain Aging Study were stratified based on APOE and BDNF Val66Met polymorphisms into four groups: ɛ4–/BDNF Val/Val (n = 29), ɛ4–/BDNF Met (n = 11), ɛ4+/BDNF Val/Val (n = 52), and ɛ4+/BDNF Met (n = 24). All participants underwent comprehensive neuropsychological examination, brain MRI and spatial navigation testing of egocentric (body‐centred), allocentric (world‐centred), and allocentric delayed navigation in a real‐space human analogue of the Morris water maze.ResultThe ɛ4+/BDNF Met group although similar to other groups in age, gender, education, depressive symptoms and global cognitive function had less accurate egocentric navigation performance (ps ≤ .045). The differences in allocentric and allocentric delayed navigation performance were not significant (ps ≥ .651). The ɛ4+/BDNFMet group had lower verbal memory performance than the ɛ4–/BDNF Val/Val group (p = .007) and similar performance to the ɛ4–/BDNF Met (p = .071) and ɛ4+/BDNF Val/Val (p = .240) groups. There were no between‐group differences in other cognitive domains including attention and working memory, nonverbal memory, executive function, language and visuospatial function (ps ≥ .268).ConclusionThe aMCI carriers with the most risky combination of APOE ɛ4 and BDNF Met polymorphisms had more pronounced egocentric spatial navigation impairment than the other polymorphism groups and more pronounced memory deficits than the carriers with the least risky combination of APOE ɛ3 and BDNF Val/Val polymorphisms. These findings indicate that aMCI individuals with two risky gene polymorphisms are particularly susceptible to hippocampus‐dependent episodic memory deficits and even more to non‐hippocampus‐dependent egocentric spatial navigation impairment that may reflect more advanced underlying pathological changes.

Social support moderates the influence of Alzheimer’s disease–related neurodegeneration on cognitive decline

AbstractBackgroundSocial support (SS) have been associated with decreased rate of cognitive decline in Alzheimer’s disease (AD). However, it is still unclear whether SS can moderate the influence of AD‐related neurodegeneration on cognitive decline. We aimed to test the hypothetical moderation effect of SS on the relationship between AD‐related brain atrophy and cognitive decline.MethodTotal 200 participants from the Korean Brain Aging Study for Early Diagnosis & Prediction of Alzheimer’s disease (KBASE), an ongoing prospective cohort study, were included for this analysis. Of the 200 participants, 130 were MCI and 70 were early AD. All participants underwent comprehensive clinical and neuropsychological assessment, multi‐modal neuroimaging including magnetic resonance imaging, and apolipoprotein E genotyping. AD‐signature cortical thickness (AD‐CT) was defined as a mean cortical thickness of several brain regions known to be sensitive to AD‐specific neurodegeneration. The Medical Outcome Study – Social Support Survey (MOS‐SSS), a 19‐iem brief, multi‐dimensional, self‐report social support survey, were administered to all participants to assess overall SS indices. Global cognitive performance was measured by the total score of the Consortium to Establish a Registry for Alzheimer’s Disease neuropsychological test battery (CERAD‐TS).ResultThere was a significant interaction between AD‐CT and SS on CERAD‐TS. Subgroup analyses demonstrated that the association between AD‐CT and CERAD‐TS was weaker in high SS group than in low SS group.ConclusionOur results suggest that SS may delay the progression of cognitive decline by moderating the influence of AD‐related neurodegeneration.

Ratio of serum proBDNF to BDNF and its association with cognitive performance and brain morphometry in mild cognitive impairment

AbstractBackgroundBrain‐derived neurotrophic factor (BDNF) is the neurotrophin well‐known for its role in neurogenesis, synaptic plasticity and enhancement of long‐term potentiation (LTP). Its precursor proBDNF exhibits opposite biological function – inhibits LTP and increases long‐term depression (LTD) in the brain. Altered proBDNF/BDNF ratio in favor of the pro‐form may lead to enhancement of LTD and reduction of LTP in hippocampal neurons with consequent synaptic dysfunction and memory deficits. To explore this hypothesis, we investigated the associations of serum proBDNF, mature BDNF and its ratio (proBDNF/BDNF) with cognitive performance and structural changes of AD‐related brain regions in individuals with amnestic mild cognitive impairment (aMCI).Method95 individuals with aMCI (age = 68.5 ± 7.3; MMSE = 26.1 ± 2.4) were recruited from the Czech Brain Aging Study. All participants underwent clinical and laboratory evaluation (with APOE and BDNF genotyping), comprehensive neuropsychological examination and brain MRI (1.5T). FreeSurfer 5.3 software was used to measure left and right hippocampal volume and left and right entorhinal and parahippocampal thickness. Serum levels of proBDNF and BDNF were measured by sandwich enzyme‐linked immunosorbent assay (ELISA).ResultA negative correlation was found between serum proBDNF/BDNF ratio and verbal memory (r = ‐0. 25, p = 0.04) and executive function (r = ‐0. 30, p = 0.02) but not with other cognitive composite scores (ps > 0.26). A trend of positive correlation was found between serum BDNF levels and verbal memory domain (r = 0.23, p = 0.08). No other correlation between BDNF and proBDNF levels and cognitive performance was significant. Correlations of proBDNF and BDNF levels and its ratio with volume and thickness of AD‐related brain regions were not significant.ConclusionHigher proBDNF/BDNF ratio is associated with lower cognitive performance in verbal memory and executive function but not with structural brain changes in aMCI individuals. Our findings suggest that proBDNF/BDNF ratio may be a more reliable blood biomarker of cognitive change than BDNF and proBDNF levels alone.

Microstructural alterations of regional white matter tracts predict tau deposition in the Alzheimer’s disease signature regions

AbstractBackgroundWe first aimed to investigate the association between regional white matter (WM) microstructural alterations with tau deposition in Alzheimer’s disease (AD) signature cortical regions. We also tried to develop models to predict pathological tau deposition positivity with regional WM microstructural alterations.MethodOne‐hundred nine older adults with normal cognition or mild cognitive impairment recruited from the Korean Brain Aging Study for the Early diagnosis and prediction of Alzheimer’s disease (KBASE) cohort were included in the analyses. All Participants underwent structural MRI (T1), diffusion tensor imaging, PiB‐PET and Tau‐PET (AV‐1451) scans. AD signature cortical regions for tau deposition included the angular gyrus, posterior cingulate, and inferior temporal cortical regions (Jack et al. 2017). Pathological tau positivity was determined by the presence of significantly high deposition in the Braak regions III‐VI according to the ADNI cutoff (Baker et al. 2017).ResultMean diffusivity (MD) of the posterior corona radiata, fornix, and posterior thalamic radiation were significantly associated with tau deposition in AD signature regions (F(1,107) = 13.48, p < 0.005; F(1,107) = 17.38, p < 0.001; F(1,107) = 15.92, p < 0.001, respectively) (Fig 1). Fractional anisotropy (FA) of the bilateral posterior thalamic radiation was also significantly associated with tau deposition in AD signature regions (F(1,107) = 17.14, p < 0.005) (Fig 1). Pathological tau positivity prediction model based on FA value of the posterior thalamic radiation, MD values of the posterior corona radiata and fornix had area under the curve (AUC) of receiver operating characteristics curve of 0.73 [0.62‐0.84] (Fig. 2a). With additional information on amyloid beta positivity, the AUC increased to 0.86 [0.78‐0.95] (Fig. 2b).ConclusionThe findings indicate that microstructural alterations of the specific regional WM tracts closely linked to AD‐signature cortical regions are strongly correlated with tau deposition in the same cortical regions. Moreover, the findings also support propagation of tauopathy through WM tracts. The information on regional WM alteration may be used for the prediction of pathological tau deposition.

Regional microstructural alteration of the corpus callosum in preclinical Alzheimer’s disease

AbstractBackgroundWhile previous studies consistently reported regional microstructural changes of the cerebral white matter (WM) in Alzheimer’s disease (AD) dementia and mild cognitive impairment (MCI), little information is available for such changes in the preclinical stage of AD. We first aimed to investigate regional microstructural alteration of the WM in cognitively normal (CN) older adults with amyloid‐beta (Aβ) deposition. We also tried to develop a prediction model for Aβ positivity in CN using information on WM microstructural alteration and clinical variables that can be easily obtained in memory clinics.MethodSeventy‐five CN older adults from the Korean Brain Aging Study for the Early diagnosis and prediction of Alzheimer’s disease (KBASE) cohort were included in the analyses. All Participants underwent comprehensive clinical and neuropsychological assessment, T1 MRI, diffusion tensor imaging (DTI), and [11C]PiB‐PET. All CN subjects were divided into Aβ positive (CN+) and negative (CN‐) based on the level of cerebral Aβ retention on [11C]PiB‐PET. Multivariate stepwise logistic regression analyses were conducted to develop Aβ positivity prediction models.ResultWhen compared to CN‐, CN+ showed high mean diffusivity (MD) value of the genu of the corpus callosum (CC_genu) (F(1,73) = 6.357, p = 0.014, Figure 1). The area under the curve (AUC) of the receiver operating characteristics curve for the Aβ positivity prediction model only using MD of the CC_genu was 0.653 [confidence interval (CI) = 0.511‐0.794]. Model with MD of the CC_genu, age, and word list recall and constructional recall test scores were finally selected. The AUC of the final model was 0.851 [CI = 0.765‐0.937] (Figure 2).ConclusionThe findings indicate that microstructural alteration of the CC_genu occurs even in the stage of preclinical AD. When taken together with episodic memory scores commonly available in clinical practice, information about such regional WM microstructural alteration can be used to screen preclinical AD.

Mild behavioral impairment is associated with atrophy in Alzheimer's disease‐related regions in non‐demented older adults

AbstractBackgroundMild behavioral impairment (MBI) is a recently developed diagnostic concept describing late‐onset persistent neuropsychiatric symptoms (NPS) in older adults without dementia. Few studies to date investigated associations of MBI with structural brain changes. Our aim was to explore structural correlates of NPS in a non‐demented memory clinic sample using the Mild Behavioral Impairment Checklist (MBI‐C) that has been developed to measure MBI.Method116 non‐demented older adults from the Czech Brain Aging Study presenting with subjective cognitive concerns underwent a neurological, comprehensive neuropsychological examination and brain MRI and were classified as subjective cognitive decline (n=37, MMSE=29.30±0.88) or mild cognitive impairment (n=79, MMSE=26.82±2.51). The Czech version of MBI‐C was administered to participants´ informants. 1.5T MRI with T1‐weighted 3‐dimensional images and FreeSurfer 5.3. algorithm was used to measure a priori selected brain regions: thicknesses of the orbitofrontal cortex (OFC), anterior cingulate cortex (ACC), posterior cingulate cortex (PCC) and entorhinal cortex (ERC) and hippocampal volume (HV) adjusted for total intracranial volume. The volumetric measures were averaged between the two hemispheres. The associations of the MBI‐C total score and five domain scores with the aforementioned regions were assessed within the whole sample using covariate‐adjusted Spearman Rank correlations.ResultSignificant associations of MBI‐C total score were found with HV (rho=‐.225, p=.016), ERC (rho=‐.284, p=.002) and OFC (rho=‐.214, p=.023) while controlling for sex, age and education. Both latter associations remained significant after controlling also for MMSE. We found associations of HV with both motivation (rho=‐.248, p=.008) and impulse dyscontrol scores (rho=‐.240, p=.011), ERC with impulse dyscontrol (rho=‐.368, p<.001) and OFC with motivation score (rho=‐.201, p=.033). No other associations were observed.ConclusionNPS (particularly in motivation and impulse dyscontrol domains) detected by the MBI‐C in non‐demented older adults are associated with structural changes in typical Alzheimer´s disease‐related regions. This provides supportive evidence for MBI as an early manifestation of AD.

Telomere length, brain tau deposition and cognitive decline: Two‐year follow‐up study

AbstractBackgroundTelomere length (TL) is known to be associated with aging and various age‐related diseases. However, previous studies on the relationship between the TL and Alzheimer’s disease (AD) yielded inconsistent results, and the association between TL and in vivo AD pathologies such as cerebral beta‐amyloid (Aβ) and tau deposition remains unclear. We investigated whether TL is related to in vivo AD pathologies and predicts future cognitive decline.MethodTotal 132 older adults recruited from the Korean Brain Aging Study for the Early Diagnosis and Prediction of Alzheimer’s Disease were included for analysis. For all participants, absolute TL values (aTL) were measured by a quantitative PCR method using DNA extracted from whole blood at baseline, as aTL can provide more direct comparison between experiments and laboratories compared to traditional TL assays. Global PiB retention measured 11C‐PiB‐PET and global tau deposition in the regions corresponding to Braak stages measured by 18F‐AV‐1451 PET were used in this study. For clinical outcome, conversion of clinical diagnosis and change of MMSE score assessed at baseline and 2‐year follow‐up was used.ResultAmong 132 older adults, 71 of them were cognitively normal and 51 of them were cognitively impaired (39 MCI and 22 AD dementia) at baseline and total 20 individuals converted to MCI or AD dementia during 2 year follow‐up period. No significant group difference in aTL was observed between diagnostic groups at baseline. However, aTL at baseline was significantly longer in converters (N=20) compared to non‐converters (N=112) even after adjusting age, sex, APOE4 positivity and baseline diagnosis. In addition, the difference of MMSE score between baseline and 2‐year follow‐up showed positive association with baseline aTL after adjusting age, sex, education, APOE4 and baseline diagnosis. With regard to in vivo AD pathologies, baseline aTL had a significant positive association with global tau deposition, but not with global Aβ deposition.ConclusionOur finding suggests that longer TL relates to greater tau deposition in brain and more rapid cognitive decline in older adults. Further studies to elucidate mechanism underlying the association of TL with tau deposition and cognitive decline will be necessary.

Diabetes mellitus, HBA1C, and in vivo Alzheimer’s disease pathologies

AbstractBackgroundAlthough previous studies have reported associations of diabetes mellitus (DM) with cognitive impairment or dementia, the exact pathophysiological links underlying the associations remain incompletely understood. In this context, we aimed to investigate the associations of DM or glycated hemoglobin (HbA1c) with in vivo Alzheimer’s disease (AD) pathologies such as cerebral beta‐amyloid protein (Aβ) deposition and tau deposition.MethodTotal 136 participants (71 cognitively normal, 31 mild cognitive impairment, and 34 AD dementia) from the Korean Brain Aging Study for Early Diagnosis & Prediction of Alzheimer’s Disease (KBASE), an ongoing prospective cohort study, were included for this analysis. All participants underwent comprehensive clinical and neuropsychological assessment, 11C labelled Pittsburgh Compound B (PiB) positron emission tomography (PET), 11F AV‐1451 PET, magnetic resonance imaging, apolipoprotein E genotyping, and HbA1c measurement.ResultNeither the presence of DM nor HbA1c level was associated with global Aβ deposition after controlling for age, sex, education, vascular risk score (except DM), apolipoprotein e4 positivity, and clinical diagnosis. In contrast, higher HbA1c was significantly associated with lower AD‐signature region tau deposition and the presence of DM showed a trend level association with decreased tau deposition. Additionally, we found significant Aβ deposition x HbA1c (or the presence of DM) interaction effect on tau deposition. While higher Aβ deposition was significantly related to increased tau deposition in lower HbA1c (or DM negative) subgroup, there was no such relationship between Aβ and tau deposition in higher HbA1c (or DM positive) one. Sensitivity analyses including only cognitively normal individuals also showed similar results.ConclusionOur results indicate that DM or higher blood glucose may not contribute to cognitive impairment or dementia via AD‐specific pathologies, especially amyloid deposition. Therefore, other non‐AD pathophysiological contributions including vascular pathologies should be considered. Although being very cautious, high blood glucose may relate to delayed brain tau deposition.

Serum cortisol, cerebral amyloidosis, and neurodegeneration: A longitudinal study

AbstractBackgroundAlthough elevated blood cortisol has been repeatedly associated with Alzheimer`s disease (AD) and related cognitive decline, the pathological links underlying the association remain poorly understood. In this context, we aimed to investigate the relationship between serum cortisol and AD neuroimaging markers including amyloid beta protein (Aβ) deposition, tau deposition, AD‐signature region cerebral glucose metabolism (AD‐CM), AD‐signature region cortical thickness (AD‐CT), and adjusted hippocampal volume (HVa).MethodThis study was part of the of the Korean Brain Aging Study for Early Diagnosis and Prediction of Alzheimer’s Disease (KBASE), an ongoing prospective cohort study. A total of 503 middle‐aged and elderly adults, including cognitively normal adults and individuals with mild cognitive impairment and AD dementia, underwent comprehensive clinical and neuropsychological assessments, 11C‐Pittsburgh compound B (PiB)‐PET, 18F‐deoxyglucose (FDG)‐PET, MRI, and serum cortisol level quantification at baseline and 2 years later. A subset of subjects also received 18F‐AV‐1451 PET during the follow‐up evaluation.ResultAt baseline, a significant positive association was found between serum cortisol level and global cerebral Aβ deposition after controlling for the effects of potential confounders (i.e., age, sex, APOE4 status, cognitive status, vascular risk, and depression). In addition, cortisol level showed significant inverse association with AD‐CM even after controlling for global Aβ retention as well as the confounders. There was no significant relationship between cortisol and cerebral tau deposition. In regard of longitudinal changes of AD markers, baseline cortisol level also had significant positive association with increase of cerebral Aβ deposition over 2 years, but not with changes of any neurodegeneration markers.ConclusionThe findings indicate that high serum cortisol contribute to AD processes through pathological cerebral Aβ deposition. Additionally, cortisol appears to directly affect neurodegeneration in AD‐signature brain regions via Aβ‐independent mechanisms.

Cognitive worry in cognitively normal older adults is associated with decreased memory binding, hippocampal volume and parahippocampal thickness

AbstractBackgroundWorry associated with subjective cognitive decline (cognitive worry) in cognitively normal (CN) older adults has been established as a feature that increases the likelihood of preclinical Alzheimer´s disease (AD). Previously, it has been associated with a higher risk of progression to mild cognitive impairment and dementia. We aimed to examine whether presence of cognitive worry reflects poorer memory performance and decreased volumes or cortical thicknesses in AD‐related regions in CN older adults.MethodIn total, 85 CN older adults reporting cognitive complaints were recruited from the Czech Brain Aging Study and classified into two groups by subjective evaluation of presence/absence of cognitive worry (WORRY+/‐, n = 36 and n = 49, respectively). All participants underwent a comprehensive neuropsychological examination, including the Rey Auditory Verbal Learning Test (RAVLT), Logical Memory I (LM I) and the challenging Memory Binding Test (MBT). 1.5T MRI with T1‐weighted 3‐dimensional images and FreeSurfer 5.3. algorithm was used to measure left and right hippocampal volume (HVL, HVR), left and right entorhinal and parahippocampal thickness (ERCL, ERCR, PHCL, PHCR). Individuals with significant vascular‐related white matter hyperintensities were not included. A one‐way ANCOVA assessed the between‐groups differences, controlling for sex, age and education.ResultThe WORRY+ did not differ from the WORRY‐ group in the RAVLT or LM I scores, but performed significantly lower in the MBT scores evaluating both free and cued recall in immediate and delayed conditions (ps ≤ 0.012). Consistently, the WORRY+ group had significantly decreased adjusted HVR, decreased PHCL and PHCR (ps ≤ 0.019). No other between‐group differences were found in medial temporal lobe measures.ConclusionCognitive worry in CN older adults is associated with both poorer memory performance in the challenging test of memory binding and decreased hippocampal volume and parahippocampal thickness even cross‐sectionally. Given that cortical thinning and hippocampal atrophy on MRI, together with poor cognitive performance on challenging tests, are detectable in late stages of preclinical AD, cognitive worry may be a risk factor for imminent functional impairment and disease progression.

Genetic associations of in vivo pathology influence Alzheimer\u2019s disease susceptibility

IntroductionAlthough the heritability of sporadic Alzheimer’s disease (AD) is estimated to be 60–80%, addressing the genetic contribution to AD risk still remains elusive. More specifically, it remains unclear whether genetic variants are able to affect neurodegenerative brain features that can be addressed by in vivo imaging techniques.MethodsTargeted sequencing analysis of the coding and UTR regions of 132 AD susceptibility genes was performed. Neuroimaging data using 11C-Pittsburgh Compound B positron emission tomography (PET), 18F-fluorodeoxyglucose PET, and MRI that are available from the KBASE (Korean Brain Aging Study for Early Diagnosis and Prediction of Alzheimer’s disease) cohort were acquired. A total of 557 participants consisted of 336 cognitively normal (CN) adults, 137 mild cognitive impairment (MCI), and 84 AD dementia (ADD) groups.ResultsWe called 5391 high-quality single nucleotide variants (SNVs) on AD susceptibility genes and selected significant associations between variants and five in vivo AD pathologies: (1) amyloid β (Aβ) deposition, (2) AD-signature region cerebral glucose metabolism (AD-Cm), (3) posterior cingulate cortex (PCC) cerebral glucose metabolism (PCC-Cm), (4) AD-signature region cortical thickness (AD-Ct), and (5) hippocampal volume (Hv). The association analysis for common variants (allele frequency (AF) > 0.05) yielded several novel loci associated with Aβ deposition (PIWIL1-rs10848087), AD-Cm (NME8-rs2722372 and PSEN2-rs75733498), AD-Ct (PSEN1-rs7523) and, Hv (CASS4-rs3746625). Meanwhile, in a gene-based analysis for rare variants (AF < 0.05), cases carrying rare variants in LPL, FERMT2, NFAT5, DSG2, and ITPR1 displayed associations with the neuroimaging features. Exploratory voxel-based brain morphometry between the variant carriers and non-carriers was performed subsequently. Finally, we document a strong association of previously reported APOE variants with the in vivo AD pathologies and demonstrate that the variants exert a causal effect on AD susceptibility via neuroimaging features.ConclusionsThis study provides novel associations of genetic factors to Aβ accumulation and AD-related neurodegeneration to influence AD susceptibility.

Defining the Lowest Threshold for Amyloid-PET to Predict Future Cognitive Decline and Amyloid Accumulation.

As clinical trials move toward earlier intervention, we sought to redefine the β-amyloid (Aβ)-PET threshold based on the lowest point in a baseline distribution that robustly predicts future Aβ accumulation and cognitive decline in 3 independent samples of clinically normal individuals. Sequential Aβ cutoffs were tested to identify the lowest cutoff associated with future change in cognition (Preclinical Alzheimer Cognitive Composite [PACC]) and Aβ-PET in clinically normal participants from the Harvard Aging Brain Study (n = 342), Australian Imaging, Biomarker and Lifestyle study of aging (n = 157), and Alzheimer's Disease Neuroimaging Initiative (n = 356). Within samples, cutoffs derived from future Aβ-PET accumulation and PACC decline converged on the same inflection point, beyond which trajectories diverged from normal. Across samples, optimal cutoffs fell within a short range (Centiloid 15-18.5). These optimized thresholds can help to inform future research and clinical trials targeting early Aβ. Threshold convergence raises the possibility of contemporaneous early changes in Aβ and cognition. This study provides Class II evidence that among clinically normal individuals a specific Aβ-PET threshold is predictive of cognitive decline.