Within the Alzheimer disease (AD) spectrum, metabolic alterations occur in addition to proteinopathies. While hypometabolism is frequently observed in the late symptomatic stages, characterizing the metabolic changes during early AD stages might aid in both understanding its pathophysiology and in patient selection for future treatments. In this study, we conduct an exploratory, data-driven analysis aimed at better understanding the timing and specificity of the metabolic changes in early AD. Using kinetic analysis of [18F]FDG PET brain imaging data coming from 223 individuals, including 33 preclinical individuals and 19 symptomatic individuals, we find new evidence of a more complex spatiotemporal trajectory of glucose metabolism in early AD, which includes a paradoxical regional increase in glucose phosphorylation during preclinical AD. These findings suggest that the pathologic phases of AD parallel changes in brain glucose metabolism, which is readily assessable with [18F]FDG PET imaging. Moreover, they may indicate that metabolic interventions may work differently during preclinical AD as compared to the early symptomatic phase.

Abstract INTRODUCTION Positron emission tomography (PET) without usable or accompanying magnetic resonance imaging (MRI) is typically excluded in quantitative analyses of Alzheimer's disease, potentially limiting study generalizability. We investigated participant features predicting data exclusion in magnetic resonance (MR)‐dependent analyses and evaluated an existing MR‐free PET pipeline to quantify these missing data. METHODS Imaging, clinical, cognitive, and sociodemographic data were analyzed for 2119 individuals in a multi‐site cohort. Agreement between MR‐dependent and MR‐free Centiloids (CL) assessed using intra‐class correlations and features predicting data exclusion were examined using logistic regressions. RESULTS MR‐free and MR‐dependent CLs generally agreed, but MR‐free CLs underestimated MR‐dependent cross‐sectionally and longitudinally. Approximately 19.5% (n = 405) of our cohort would have been excluded in MR‐dependent analyses. Age and cerebrovascular comorbidities were consistent exclusion features across multiple sites. DISCUSSION Data exclusion in imaging studies is not entirely random. Flexible quantification methods like MR‐free PET could supplement traditional methods to improve generalizability in large, multi‐site studies.

INTRODUCTION:The Clinical Dementia Rating (CDR) scale typically is administered in person, but use of telephone-based, informant-only assessments increased during the coronavirus disease 2019 (COVID-19) pandemic. The correspondence of informant-only assessments with in-person ratings remains unclear. METHODS:We analyzed 1,140 paired in-person and telephone assessments from the Knight Alzheimer's Disease Research Center Memory and Aging Project conducted within 12 months. Agreement for global CDR and CDR Sum of Boxes (CDR-SB) was examined using kappa statistics, intraclass correlation coefficients (ICCs), and Bland-Altman methods; classification performance of telephone global CDR score ≥ 0.5 was assessed. RESULTS:Agreement for the global CDR was moderate (κ = 0.53, 95% CI: 0.47-0.59). CDR-SB demonstrated moderate reliability (ICC = 0.69, 95% confidence interval [CI]: 0.65-0.73). Telephone CDR-SB scores averaged 0.40 points higher than in-person scores, with wide limits of agreement. Sensitivity was 64.2% and specificity 91.9%. DISCUSSION:Telephone CDR-SB shows moderate concordance with in-person CDR-SB but shows consistent score inflation, which may limit clinical staging utility.

INTRODUCTION:We recently identified a plasma-based seven-protein model with strong performance for Alzheimer's disease (AD) classification. Here, we evaluated whether these proteins, alone or combined with plasma phosphorylated tau 217 (p-tau217), predict progression from cognitively unimpaired to symptomatic AD. METHODS:Using longitudinal data from Knight-ADRC (Alzheimer's Disease Research Center) with replication in Alzheimer's Disease Neuroimaging Initiative (ADNI), we modeled time to progression using Cox regression. Models included p-tau217, the seven-protein panel, and their combination. RESULTS:The p-tau217 alone showed similar progression prediction (hazard ratio [HR] = 4.08) than the seven-protein model (HR = 4.85). Integrating the seven-protein model with ptau217 significantly improved risk, identifying a high-risk group (HR = 11.15) with two intermediate-risk groups. Simplified models retained prognostic value, with top-performing ratios, Complexin-2/Synaptic vesicle membrane protein VAT-1 homolog (CPLX2/VAT1) and Acetylcholinesterase/Neuronal pentraxin receptor (ACHE/NPTXR) also lead to a significantly better risk stratification than p-tau217 alone. DISCUSSION:Integrating p-tau217 with targeted plasma proteins enables graded risk stratification and identifies individuals at highest risk of progression, supporting clinically scalable approaches for early risk assessment.

Detection of Alzheimer's disease (AD) before the development of clinical symptoms is critical for enabling the use of new treatments. Circular RNAs (circRNAs) are highly stable non-coding RNAs enriched in the brain that can cross the blood-brain barrier. Here, analyzing blood data from 1,221 individuals with AD and healthy individuals, we identified 34 circRNAs associated with AD status. A predictive model including these 34 circRNAs was comparable to plasma phosphorylated Tau-217 (pTau217) in classifying AD based on biomarker-confirmed (amyloid-β and Tau) status and replicated in independent samples from the Knight-Alzheimer Disease Research Center (n = 551: 76 AD, 475 cognitively unimpaired) and preclinical A4 (n = 1,767) cohorts. Classification of biomarker-confirmed status by blood circRNAs (area under the curve (AUC) = 0.945) had a higher predictive ability than plasma pTau217 (AUC = 0.877) and was further improved in the integrated model (circRNA+pTau217 AUC = 0.977). This model showed high AD specificity with low predictive power for Parkinson's disease, frontotemporal dementia, and other neurodegenerative diseases. In the Knight-Alzheimer Disease Research Center discovery cohort, these circRNAs (hazard ratio = 2.92) outperformed pTau217 (hazard ratio = 1.81) and amyloid-positron emission tomography when predicting progression to symptomatic AD. Although prospective validation in larger cohorts is needed, these results propose blood circRNAs as potential biomarkers for AD diagnosis and disease progression.

BACKGROUND AND OBJECTIVES:To examine whether time from amyloid positivity (amyloid time), a continuous measure of biological Alzheimer's disease (AD) progression, is associated with differences in navigation-related driving behaviors among older adults. METHODS:This longitudinal cohort study was drawn from the DRIVES participants who had at least one PET Pittsburgh compound-B (PiB) scan. AD timeline was assessed using amyloid time, estimated with the Sampled Iterative Local Approximation (SILA) approach. SILA aligns individuals relative to the timing of amyloid positivity. Naturalistic driving data were continuously collected using in-vehicle data loggers. Navigation-related behaviors were quantified using trip-chaining and entropy metrics. Linear mixed-effects models examined associations between amyloid time and longitudinal differences in trip chaining behaviors, adjusting for demographic factors. Growth mixture models were used to explore latent trajectory classes, and logistic regression was used to explore demographic characteristics, self-reported medical history, and medication use as potential risk and resilience factors associated with membership in trajectory classes. RESULTS:Greater amyloid time was associated with higher counts and proportions of chained trips and less entropy, indicating more predictable driving patterns over time. Growth mixture modeling identified two distinct trajectory classes for both trip chaining and entropy. Insomnia was associated with a faster increase in trip chaining, whereas psychiatric conditions, such as having a diagnosis of depression or anxiety, were associated with a slower decline in entropy. DISCUSSION:Amyloid time is associated with gradual differences in real-world navigation behaviors among cognitively normal older adults. More trip chaining and less entropy may reflect compensatory planning and reduced driving space in the context of early AD pathology. These findings highlight the utility of naturalistic driving data as ecologically valid, scalable markers of early functional change and underscore the importance of continuous biomarker measures for capturing disease progression prior to clinical symptom onset.

INTRODUCTION:Recruitment and retention remains a major challenge in Alzheimer's disease research. This study examined the impact of describing potential financial compensation during recruitment on participant enrollment to a longitudinal cohort. METHODS:Participant recruitment calls (N = 337) were randomized to either a compensation-mentioned group (n = 170) or a control group (n = 167). An intention-to-treat logistic regression assessed the effect of compensation on enrollment. RESULTS:Of 320 analyzed, 124 (38.75%) enrolled. The intervention group's consent rate was lower than the control group's in intention-to-treat (-9.72 points; p = 0.074), per-protocol (-12.72 points; p = 0.026), and complier average causal effect analyses (-11.36 points; p = 0.72). DISCUSSION:Disclosing compensation during recruitment may reduce enrollment, potentially due to perceptions that compensation conflicts with altruistic motives. However, this was observed in a highly educated sample; compensation may affect those with lower levels of education and socioeconomic status differently by helping offset participation burden, warranting further investigation.

Concentration-based fluid biomarkers represent an informative and cost-effective way to detect and monitor Alzheimer's disease (AD) pathology. However, non-AD-related interindividual variation in biofluids can also affect biomarker concentrations. Here, we investigated whether normalization of CSF and plasma biomarkers to reference proteins, such as amyloid-β40 (Aβ40) and non-phosphorylated mid-region tau (np-tau), improves their robustness and reliability of representing AD pathology load. Using the Swedish BioFINDER-2 cohort [n = 1702, 50.7% male, mean (standard deviation) age 68.4 (12.2) years], we compared the associations between tau/Aβ-PET load and fluid biomarkers alone versus biomarkers in a ratio with a reference protein (Aβ40 or np-tau) in univariate linear regression models. Fluid biomarkers included CSF and plasma measures of p-tau217, p-tau181, p-tau205, np-tau181-190, np-tau195-210, np-tau212-221, Aβ42 and Aβ40; CSF MTBR-tau243, SNAP-25, neurogranin, YKL-40 and sTREM2; and plasma eMTBR-tau243. Biomarkers were measured with mass spectrometry assays and/or immunoassays. In addition, we performed validation and extended analyses, comparing, for example, group-level diagnostic differences and longitudinal biomarker trajectories, in three independent prospective cohorts [BioFINDER-1, Knight Alzheimer Disease Research Center (ADRC) and Translational Biomarkers in Aging and Dementia (TRIAD)] and in an Italian multiple sclerosis cohort. CSF Aβ40 normalization significantly strengthened the associations of several core CSF AD biomarkers, including CSF MTBR-tau243, p-tau isoforms and synaptic biomarkers, with tau-PET (ΔR2 = 0.064-0.24) and Aβ-PET (ΔR2 = 0.016-0.28). Normalization to CSF np-tau mainly improved concordance with Aβ-PET (ΔR2 = -0.0059 to 0.19). The strongest association with tau-PET was observed for MTBR-tau243/Aβ40 (R2 = 0.78, compared with 0.65 for non-normalized MTBR-tau243), and with Aβ-PET for p-tau217/np-tau (R2 = 0.65, compared with 0.46 for non-normalized p-tau217). Plasma biomarker associations with tau-PET improved when using normalization to plasma Aβ40 or np-tau (ΔR2 = 0.004-0.14), with the strongest effect for eMTBR-tau243/np-tau (R2 = 0.72 versus 0.60). Associations with Aβ-PET were enhanced with np-tau normalization (ΔR2 = 0.018-0.16, strongest for p-tau217/np-tau: R2 = 0.62 versus 0.53). The results were replicated in Knight ADRC and TRIAD. Furthermore, longitudinal analyses showed that Aβ40 normalization typically reduced interindividual rather than intra-individual variability over time. Normalization did not enhance group-level differences in inflammatory CSF biomarkers in AD, nor did it improve biomarker associations in the multiple sclerosis cohort. In conclusion, normalization of CSF and plasma biomarkers to reference proteins, such as Aβ40 or np-tau, enhances their association with brain tau and Aβ pathology, making already high-performing AD fluid biomarkers even more accurate.

Chronic psychological stress has been implicated as a risk factor for Alzheimer’s disease (AD), potentially through cortisol-mediated acceleration of disease progression. However, the molecular pathways underlying this relationship remain poorly understood. Epigenetic regulation of the glucocorticoid and mineralocorticoid receptor genes (NR3C1 and NR3C2), which encode receptors for cortisol, may play an important role, but has not been examined in relation to AD progression. Therefore, this study investigated associations between DNA methylation of NR3C1/NR3C2 and AD-related phenotypes, including cognition, brain amyloid-β (Aβ) burden, and regional brain volumes. These associations were examined in two independent cohorts of cognitively unimpaired individuals with accumulating brain Aβ (n = 89–298 across outcomes) using linear regression and meta-analyses. The study also explored whether DNA methylation within NR3C1 and NR3C2 interacted with depression symptoms to influence relationships with AD-related phenotypes. While only nominal associations were observed in direct analyses, stronger associations emerged in interaction with depressive symptoms. Interaction analyses showed that relationships between DNA methylation and AD-related phenotypes (cognition, hippocampal volume and ventricular expansion) differed depending on the presence of depression symptoms. Consistent patterns across cohorts were observed, with associations primarily evident among individuals with clinically relevant depressive symptoms. One site (NR3C1 cg24052866) was associated with cognitive decline, one (NR3C1 cg08845721) with cross-sectional hippocampal volume, and eight (NR3C1 cg21979215, cg16594263; NR3C2 cg27460943, cg17253842, cg04867484, cg10993059, cg25672354, cg27234800) with ventricular expansion. These exploratory findings suggest epigenetic variation within cortisol receptor genes may influence AD-related neurodegeneration in a depression-dependent manner.

BackgroundMidlife obesity is considered one of the top modifiable risk factors for dementia and Alzheimer's disease (AD). However, body mass index (BMI) on its own does not fully represent obesity-associated risks and it is crucial to disentangle the role of body adiposity and its localization.ObjectiveTo investigate the relationship of MRI-derived body adiposity metrics with AD-related pathology at midlife.MethodsNinety-seven cognitively normal midlife individuals underwent brain amyloid and tau PET, body MRI, and metabolic and cognitive assessments. Key measures included hepatic fat fraction, visceral (VAT) and subcutaneous adipose tissue (SAT) volumes, and thigh muscle and adiposity. The correlation between adiposity/metabolic measurements and amyloid/tau pathologies was investigated.ResultsThe average age of participants was 49.8 years, 65.3% were female and 53.6% had obesity. Amyloid PET burden in Centiloids correlated with VAT (rho = 0.36, p = 0.002), BMI (rho = 0.33, p = 0.002), SAT (rho = 0.33, p = 0.002), and insulin resistance (IR) (rho = 0.34, p = 0.003) in females and Whites, lower high-density lipoprotein (HDL) cholesterol (rho = -0.36, p = 0.002) irrespective of sex and race, and lower MMSE scores (rho = -0.57, p = 0.043) in only in African-Americans, after correction for age, sex, and education. There was no evidence that HDL nor IR mediated VAT-related amyloid. VAT/SAT ratio was significantly associated with mean cortical tau SUVR (β = 0.138, p = 0.030) after adjustment for age, sex, education, and amyloid.ConclusionsAmong fat depots in our study, visceral fat was more strongly correlated to amyloid pathology, and this association is present even independent from BMI. Also, higher visceral compared to subcutaneous fat is related to higher tau pathology.

As individuals age, they are more likely to show increased functional MRI (fMRI) activity, particularly in frontal regions. This has been interpreted as a compensatory mechanism, yet the very need to draw upon such resources indicates increasing failures of brain systems. Almost all work addressing theoretical models explaining these patterns has been cross-sectional, with minimal work testing how elevated fMRI activity predicts future cognitive trajectories. Further, although often viewed as ageing, there is increasing evidence that subtle fMRI changes may represent the earliest manifestation of neurodegenerative conditions such as Alzheimer's disease. One hundred nine individuals completed a Stroop colour/word task during fMRI data acquisition. Analyses focused on differences between trials where the colour and word were incongruent (the word red written in blue) relative to congruent trials (the word blue written in blue). At baseline, participants also underwent amyloid positron emission tomography imaging and APOE genotyping. Individuals had longitudinal clinical follow-up (mean 6.8 years) with 15 individuals reaching the threshold of clinically defined cognitive impairment. Across the entire cohort, several brain regions, including dorsolateral prefrontal cortex, anterior cingulate cortex, and lateral temporal and parietal regions, were more active on conflict trials. At the individual level, increases in activity were related to changes in reaction time, with those experiencing greater conflict having greater evoked activity. Further, individuals who later developed dementia had greater activity at baseline than their peers who remained cognitively normal despite there being no differences in accuracy (t = 0.23, P = 0.82) or reaction time (t = 0.94, P = 0.35) in the Stroop task nor differences in mini-mental state examination (t = 0.06, P = 0.95) or a neuropsychological composite (t = -0.99, P = 0.32). Individuals who progressed were more likely to be amyloid positive (χ 2 = 26.71, P = 0.000002) and carriers of the APOE ε4 allele (χ 2 = 4.81, P = 0.03). The current work suggests that, although compensatory in nature in the short term, increased activation of frontal and parietal control regions during attentional control tasks is indicative of underlying declines in brain health. The larger implications are 2-fold. Undetected neurodegenerative disease pathology biases our understanding of what constitutes healthy ageing. Further, alterations in brain function occur well in advance of clinically detectable cognitive change, emphasizing the need to intervene with disease modifying therapies early in the disease course.

INTRODUCTION With the clinical availability of amyloid-targeting treatments (ATTs), accurate and timely biomarker-based diagnosis of Alzheimer's disease (AD) has become increasingly important. Three AD biomarker modalities are commonly available in clinical practice: amyloid positron emission tomography (PET), cerebrospinal fluid (CSF) tests, and blood tests. METHODS We investigated the use and agreement of different biomarker modalities in a memory clinic. Further, we calculated the time until ATT initiation for patients who underwent a single test versus multiple biomarker tests. RESULTS The blood test agreed with amyloid PET in nine of 11 patients and CSF tests in all 14 patients. The median time from first clinic visit to ATT initiation was 4.7 months in 209 patients who underwent a single test and 8.1 months in 12 patients who underwent multiple tests. DISCUSSION Performing multiple biomarker tests delays initiation of ATT and should be restricted to patients with uncertain amyloid status following the first biomarker test.

The agreement between plasma Aβ42/40 and Aβ-PET is approximately 75%, with a large portion of discrepancies due to positive plasma with negative PET results. Questions remain about whether these reflect brain Aβ changes detectable in plasma before PET-detectable. We aimed to examine these cases over 11 years to assess the risk and timing of progression to Aβ-PET positivity. Cognitively unimpaired participants from large-scale longitudinal studies of AIBL, OASIS, and ADNI underwent baseline Aβ-PET and plasma Aβ42/40 analysis by IPMS, followed by 1-7 additional PET scans every 1.5-3 years. Aβ-PET was quantified to Centiloid (CL) using the SPM pipeline. Individuals with baseline Aβ-PET < 20 CL ( n = 507) were included, with those < 5 CL classified as PET-, and 5-20 CL as PET Low . Plasma -/+ was based on the Aβ42/40 Youden's Index threshold (0.119) corresponding to Aβ-PET status. We used Kaplan-Meier method and Cox proportional hazards analysis to assess the risk of progression to PET+ (> 20 CL). Plasma+/PET- (< 5 CL) individuals were at higher risk than Plasma-/PET- of progressing to PET+ (hazard ratio (HR): 3.90 [95% CI: 2.00-7.61], p <0.001), even after matching the groups’ baseline CL values (HR: 3.43 [1.43-8.26], p = 0.010), or adjusting for age, sex, APOE ε4 and baseline CL (HR: 2.48 [1.22 - 5.07], p = 0.013) (Figure 1A). Plasma+/PET- accumulated brain Aβ ∼8 times faster than Plasma-/PET- (1.14 CL/year vs. 0.15 CL/year respectively, p <0.001). Plasma+/PET- progressors became PET+, on average, 2 years earlier than Plasma-/PET- progressors. Plasma+/PET Low group had faster decline in survival probability than Plasma-/PET Low (HR: 20.82 [11.28 – 38.42], p <0.001 vs. HR: 6.67 [3.51 – 12.65], p <0.001) (Figure 1B) but this was driven by higher CL in the Plasma+ group. Cognitively unimpaired individuals with abnormal plasma Aβ42/40 but negative Aβ-PET face a significantly increased risk of future positive Aβ-PET. This provides supporting evidence that brain Aβ pathology can be detected in plasma with IPMS before it is PET-detectable. Whether this also applies to plasma Aβ42/40 immunoassays warrants investigation.

Importance:Biological staging of Alzheimer disease (AD) can improve diagnostic accuracy and prognostic assessment. However, existing approaches often require invasive procedures and specialized infrastructure, limiting their routine clinical use. Objective:To develop and validate a plasma-based biological staging model that closely mirrors an amyloid and tau positron emission tomography (PET)-based staging system. Design, Setting, and Participants:This observational longitudinal study included the research BioFINDER-2 cohort as the main cohort and the Knight Alzheimer Disease Research Center (ADRC) cohort as an independent validation cohort, with data acquired from November 2019 to January 2025 and from September 2007 to March 2020, respectively. Participants ranged from cognitively unimpaired (CU) to experiencing mild cognitive impairment (MCI) and dementia with AD or other neurodegenerative diseases. All individuals had both plasma biomarkers available. A Knight ADRC subsample had neuropathological data available. Data were analyzed from December 2024 to July 2025. Exposures:Plasma endogenously cleaved microtubule-binding region tau containing residue 243 (eMTBR-tau243) and phosphorylated tau at threonine 217 expressed as a percentage of its unphosphorylated form (%p-tau217). Main Outcome and Measures:The primary outcomes were PET-based biological staging according to the revised Alzheimer's Association criteria, clinical criteria, and biomarker levels assessed cross-sectionally and longitudinally. Results:The BioFINDER-2 main cohort included 872 participants, with an additional 156 participants in the Knight ADRC cohort for independent validation. In the BioFINDER-2 cohort, participants included 383 CU participants, 182 participants with MCI, 151 with AD dementia, and 156 participants with non-AD neurogenerative diseases (mean [SD] age, 72.8 [9.2] years; 438 women [50.2%]; 516 [59.2%] apolipoprotein E [APOE-E4] carriers). A plasma-based model with %p-tau217 and eMTBR-tau243 was derived, which demonstrated high concordance with established PET-based stages (C index, 0.91; 95% CI, 0.90-0.92) and was associated with clinical stage within the AD continuum (C index, 0.84; 95% CI, 0.82-0.86). In the validation cohort, concordance with PET-based stages was high (C index, 0.91; 95% CI, 0.87-0.94), as was concordance with clinical stage (C index, 0.86; 95% CI, 0.82-0.89). Concordance between fluid- and PET-based staging models was higher with this study's %p-tau217 and eMTBR-tau243 model than with a staging model using %p-tau217 alone, especially improving the intermediate or C stage (A+TMOD+) classification. In the neuropathology subsample, plasma staging was strongly aligned with autopsy-confirmed AD pathology using the Alzheimer Disease Neuropathologic Change scale (area under the curve, 0.96; 95% CI, 0.91-1.00). Conclusions and Relevance:In this longitudinal study, a plasma-based biological staging model incorporating %p-tau217 and eMTBR-tau243 showed strong concordance with PET-based staging, correlated with clinical severity and biomarker progression, and aligned with neuropathological categorization. This scalable, minimally invasive approach could facilitate broader implementation of biological staging in clinical practice and could enhance participant selection for disease-modifying treatments and clinical trials.

Predicting the likelihood of developing Alzheimer’s disease (AD) dementia in at-risk individuals is important for the design of and optimal recruitment for clinical trials of disease-modifying therapies. Machine learning (ML) has been shown to excel in this task; however, there remains a lack of models developed specifically for the preclinical AD population, who display early signs of abnormal brain amyloidosis but remain cognitively unimpaired. Here, we trained and evaluated ML classifiers to predict whether individuals with preclinical AD will progress to mild cognitive impairment or dementia within multiple fixed time windows, ranging from one to five years. Models were trained on regional imaging features extracted from amyloid positron emission tomography and magnetic resonance imaging pooled across seven independent sites and from two amyloid radiotracers ([18F]-florbetapir and [11C]-Pittsburgh-compound-B). Out-of-sample generalizability was evaluated via a leave-one-site-out and leave-one-tracer-out cross-validation. Classifiers achieved an out-of-sample receiver operating characteristic area-under-the-curve of 0.66 or greater when applied to all except one hold-out sites and 0.72 or greater when applied to each hold-out radiotracer. Additionally, when applying our models in a retroactive cohort enrichment analysis on A4 clinical trial data, we observed increased statistical power of detecting differences in amyloid accumulation between placebo and treatment arms after enrichment by ML stratifications. As emerging investigations of new disease-modifying therapies for AD increasingly focus on asymptomatic, preclinical populations, our findings underscore the potential applicability of ML-based patient stratification for recruiting more homogeneous cohorts and improving statistical power for detecting treatment effects for future clinical trials.

It is becoming common to measure cognition repeatedly to capture dynamic fluctuations over time. It is currently unknown the extent to which daily deviations in cognitive performance have implications for real-world behaviors or merely reflect short-term change of limited scope, such as assessment artifacts. Daily driving behavior was captured in 160 older adults using GPS tracking across 19,628 individual “trips”, defined as ignition start to ignition off. Automated processing pipelines quantified various aspects of each trip including distance traveled and frequency of “adverse” events such as speeding or hard braking. Participants completed daily assessments of cognition using a remote testing platform. Multilevel negative binomial models were constructed to relate daily cognitive performance to adverse driving events and adaptive behavior such as avoidance of nighttime driving, controlling for global ability, participant age, and distance travelled. Rigorous multi-model inference using AIC-based model averaging indicated that daily cognitive deviations received substantial model support (79-98%) in predicting adverse events and nighttime driving, relative to models that did not include a daily parameter. In contrast, global cognitive ability did not substantially improve prediction of either outcome. These results show that deviations in cognitive performance are not merely “noise”, as they have measurable implications for real-world outcomes. Importantly, it was not global performance that was important but deviations from typical performance. Thus, single-shot assessments of cognition may not be sufficient to make determinations about driving safety. More generally, our findings highlight the utility of repeated cognitive assessments in understanding how cognitive fluctuations may impact real-world behaviors.

White matter microstructural changes play a crucial role in cognitive decline in aging and neurodegenerative disorders including Alzheimer’s disease (AD). However, the processes underlying white matter microstructural changes and the molecular pathways leading to these changes in AD remain largely unknown. AD involves cortical and juxtacortical microstructural changes, with free water fraction (FWF) as a potential imaging marker. We measured FWF using diffusion magnetic resonance imaging in 68 juxtacortical regions of 153 cognitively normal controls and 194 patients with AD as evidenced by elevated amyloid PET. We estimated the expression of 15,633 genes in the same regions using transcriptomic data from the Allen Human Brain Atlas. The biological processes and cell types associated with the linked genes were evaluated. Mediation analysis was used to examine whether FWF mediates the association between APOE ε4 status and cognitive performance. Gene ontological analyses revealed that these genes were enriched for biological processes relating to lipid metabolic process, ensheathment of neurons, and synaptic signaling and were predominantly expressed in oligodendrocytes, GABAergic neurons, and pyramidal neurons from the hippocampus CA region. These ontological enrichment results were replicated in two additional datasets. Furthermore, mediation analyses revealed a domain-specific role of FWF in the association between APOE ε4 status and cognitive performance. Our findings provide mechanistic insights into regional juxtacortical microstructural changes in AD, particularly the processes involving lipid metabolism, offering potential therapeutic targets.

INTRODUCTION:Cerebral amyloidosis is a defining feature of Alzheimer's disease (AD), yet the molecular heterogeneity among amyloidbeta-positive (Aβ+) individuals remains poorly defined. We aimed to map the proteomic correlates of cerebral amyloidosis and link them to clinical variability within Aβ+ individuals. METHODS:We integrated quantitative amyloid PET with large-scale plasma proteomics (∼7000 proteins; SomaScan version 4.1) in Knight Alzheimer's Disease Research Center and Bio-Hermes cohorts (n = 1429). Proteome-wide association analyses identified proteins associated with amyloid load, followed by unsupervised clustering and pathway enrichment analyses. RESULTS:We identified 454 amyloid-associated proteins, of which 54 replicated cross-cohort. A derived 54-protein proteomic score correlated with amyloid burden, AD biomarkers, and clinical severity. Pathway analyses of clinically distinct protein clusters revealed coordinated enrichment of intracellular signaling, immune, and proteostasis modules. DISCUSSION:These findings delineate the circulating proteomic signature of cerebral amyloidosis and support plasma proteomics as a complementary approach to phosphorylated tau at threonine 217 and amyloid PET for biological stratification and characterization of disease heterogeneity in AD.

Deep learning models for neuroimaging have largely been developed for individual tasks, limiting knowledge transfer across applications. Here we introduce GenFAR, a modular deep learning framework that learns general, clinically informed features from brain MRIs. We trained this modular architecture on 49,246 individuals across 11 cohorts, using 17 diverse classification and regression tasks spanning cognition, clinical, diagnosis, demographics, and biomarkers. This yields aggregated, focused feature sets that capture rich, clinically- and biologically-relevant brain representations. We developed a sequential learning approach where tasks progressively build on previously learned representations. Through an analysis of 5,000 task sequences, we identified an optimal sequence length of six tasks and introduced a Donor Score metric to quantify each task's contribution to downstream performance. This analysis revealed five consistently strong donor tasks (Age, AD/MCI, MMSE, Hypertension, Hyperlipidemia) that formed the base of our sequential model. We demonstrated the utility of our learned representation, in various tasks beyond those included in the training set, to serve as the foundation for specialized secondary predictors. We further showed that using the learned feature representation can substantially increase the sample efficiency of secondary deep learning training tasks and models, as well as improve their accuracy.

Although mounting evidence supports that neighborhood greenspaces (e.g., trees and parks) may benefit cognitive health in older age, fewer studies have investigated benefits to brain health measured via magnetic resonance imaging (MRI). We used data on 892 older adults without dementia from three Alzheimer's Disease Research Centers, examining whether neighborhood greenspace (i.e., greenness and percentage park space) is associated with white matter hyperintensity (WMH) and hippocampal volumes from MRI. We also examined whether associations varied by sex, racial group, urbanicity, or apolipoprotein E genotype (genetic risk factor for Alzheimer's disease). Linear regression models that accounted for neighborhood clustering controlled for demographics, research center, cognitive function, neighborhood deprivation, and comorbidities (e.g., hypertension, diabetes, obesity). Interaction terms (e.g., greenness×sex) were added to the models to evaluate potential effect modification. Living in greener neighborhoods was associated with fewer WMH and greater hippocampal volume in the overall sample. In stratified analyses, the beneficial greenness-WMH association was restricted to Black (not White) participants (i.e., significant interaction). In contrast, an adverse association between park space and lower hippocampal volumes was detected among Black (not White) participants (i.e., significant interaction). Overall, this study suggests associations between living in neighborhoods with more greenspace and MRI biomarkers of lower cerebrovascular and dementia risk, although results were mixed for Black individuals. Our findings need replication in other cohorts that are ethnoracially diverse and that represent different geographic regions, and future studies are needed explain the counterintuitive associations between park space and hippocampal volume among Black older adults.