Background:Hippocampal atrophy is a core marker of neurodegeneration in dementia, particularly in Alzheimer's disease (AD). However, most studies focus on total hippocampal volume loss and overlook hemispheric asymmetry, which may reflect distinct biological processes. While tau pathology is closely linked to medial temporal lobe degeneration, it remains unclear whether tau is associated with asymmetric patterns of hippocampal atrophy. Methods:We analyzed 483 cognitively unimpaired participants from the Alzheimer's Disease Neuroimaging Initiative (ADNI) with baseline cerebrospinal fluid (CSF) phosphorylated tau (p-tau181) and amyloid-β (Aβ42) measurements and longitudinal structural MRI data over 10 years of follow-up. Total hippocampal volume and hemispheric asymmetry, defined as the absolute difference between left and right hippocampal volumes (|L-R|), were quantified at each visit. Linear mixed-effects models assessed associations between baseline CSF biomarkers and longitudinal changes in hippocampal asymmetry, adjusting for demographic factors, APOE ε4 status, and baseline hippocampal volume. Results:Higher baseline CSF p-tau181 was associated with greater increases in hippocampal asymmetry over time (β = 1.20, SE = 0.43, p = 0.006). This association remained significant after additional adjustment for total hippocampal volume and baseline CSF Aβ42. CSF total tau was highly correlated with p-tau181 (Spearman's ρ =0.98, p < 0.001) and showed comparable associations with hippocampal measures. In contrast, baseline Aβ42 was not associated with subsequent changes in hippocampal asymmetry. Both p-tau181 and Aβ42 were associated with faster decline in total hippocampal volume. In amyloid-stratified analyses, p-tau181 was associated with increasing hippocampal asymmetry only among amyloid-negative individuals, whereas its association with total hippocampal atrophy was observed primarily in amyloid-positive individuals. Conclusions:CSF p-tau181 is associated not only with overall hippocampal atrophy but also with progressive hemispheric asymmetry, suggesting that tau-related neurodegeneration may manifest as both magnitude and imbalance of tissue loss. These findings support hippocampal asymmetry as a complementary neuroimaging marker that may capture non-amyloid-related medial temporal lobe degeneration in cognitively unimpaired older adults.

BACKGROUND:Hippocampal atrophy is a key marker of Alzheimer's disease (AD)- related neurodegeneration; however, hippocampal volume alone may not fully capture heterogeneity in cognitive decline. Left-right hippocampal asymmetry may provide complementary information, but its prognostic value for long-term cognitive decline, particularly in relation to AD pathology, remains unclear. OBJECTIVES:To determine whether hippocampal total volume and left-right hippocampal asymmetry provide complementary and independent information in capturing cognitive decline and clinical progression, and to examine their relationship to AD pathology. DESIGN:Analysis of baseline MRI and longitudinal cognitive data over 10 years in four domains of memory, language, executive, and visuospatial function, using harmonized cognitive data from the Alzheimer's Disease Sequencing Project - Phenotype Harmonization Consortium (ADSP-PHC). SETTING:Participants from ADNI 1, ADNI GO, ADNI 2, and ADNI 3 PARTICIPANTS: A total of 1,142 dementia-free participants from the Alzheimer's Disease Neuroimaging Initiative (ADNI) with available baseline structural MRI, cerebrospinal fluid (CSF) amyloid-β (Aβ42) and phosphorylated tau (p-tau-181), and longitudinal cognitive follow-up. MEASUREMENTS:Total hippocampal volume (left + right) and hemispheric asymmetry (absolute left-right volumetric difference) were modeled simultaneously. Linear mixed-effects models examined associations with baseline performance and longitudinal change across four cognitive domains. Cox proportional hazards models assessed risk of clinical progression to clinical dementia over up to 10 years of follow-up (median follow-up 4 years; median 5 visits per participant). All analyses adjusted for age, sex, education, APOE ε4 status, and CSF biomarkers, with stratification by amyloid status. RESULTS:The study cohort included 546 women (47.8%), with a mean age of 72.54 ± 6.98 years. Smaller total hippocampal volume was consistently associated with worse baseline performance and faster decline across all four cognitive domains, even after adjustment for amyloid and tau. In contrast, greater left-right hippocampal asymmetry was selectively associated with worse performance and faster decline in memory, independent of total hippocampal volume. In amyloid-stratified analyses, total hippocampal volume showed broad associations with cognitive performance across multiple domains in both amyloid-positive and amyloid-negative participants, whereas hippocampal left-right asymmetry demonstrated selective associations with memory performance, which were observed only among amyloid-negative individuals. With respect to clinical progression to dementia, smaller total hippocampal volume was associated with a higher risk of progression in the overall cohort and within both amyloid groups. In contrast, hippocampal asymmetry was associated with progression risk only among amyloid-negative individuals (hazard ratio per SD increase = 1.31, 95% CI: 1.03-1.65). CONCLUSIONS:Hippocampal total volume and asymmetry capture distinct aspects of neurodegeneration, with asymmetry providing additional prognostic information for memory decline and clinical progression in the absence of detectable amyloid pathology.

As the population of older adults continues growing, so will the need for cost-effective approaches to early dementia detection. Deep learning approaches using patient speech samples show promising results. This systematic review examines studies utilizing speech-based deep learning for dementia diagnosis with the objective of identifying best practices for future data-driven dementia research. Studies researching speech-based deep learning for dementia were obtained from PubMed, Wiley Library, Science Direct, IEEE, Web of Science, Google Scholar, and arXiv. 80 studies were reviewed. Studies were analyzed in terms of model architecture and performance, speech features employed, and databases used. We observed that transformer-based approaches were most frequent, achieving an average accuracy of 85.71%, and that linguistic features outperform acoustic features. Our review identified several limitations, including a lack of dataset diversity, inconsistent classification of dementia severity levels across studies, and variability in how sample sizes and model performance metrics (e.g., accuracy, sensitivity, specificity) are reported. These inconsistencies hinder direct comparisons between studies and limit the reproducibility of findings. Still, our findings suggest that incorporation of transformers into current speech-based deep learning models can further improve detection of cognitive impairment. Consideration of our observations in future data-driven dementia research will lead to advancements in the development of diagnostic decision support systems for clinical practice.

The increasing number of affected people with neurodegenerative disorders, like Alzheimer’s disease, has made an urgent demand for cost-effective and accessible screening methods [1] . The Current gold standard diagnostic methods are considered impractical for widespread population screening due to their high costs, invasive nature, and requirement for specialized facilities [1] , [2] . Recent advancements in speech and language analysis provide a new opportunity for an alternative diagnostic method, which identifies cognitive decline through natural language processing methods and language models [3] . Linguistic digital biomarkers that measure lexical diversity, hesitation, and difficulty in retrieving words have demonstrated a strong connection with cognitive decline [1] , [4] . On the other hand, transformer-based language models have been proven effective in capturing linguistic features [2] , [5] . However, large language models have limitations in clinical settings, such as their high demand for computational power, lack of a substantial high-quality dataset, and interpretability challenges. To overcome these limitations, the lightweight transformer architectures were developed to reduce computational cost and time while keeping comparable performance [2] . This study aims to evaluate and compare the diagnostic performance of a lightweight transformer model with machine learning approaches for detection of mild cognitive impairment (MCI) from transcribed language samples. The model could be used for real-world clinical applications.

INTRODUCTION:APOE genotype shows well-established dose-dependent associations with higher amyloid in cognitively unimpaired (CU) adults. In contrast, associations with tau burden and cognition are less well characterized. METHODS:We performed a cross-sectional analysis of harmonized multi-cohort ADSP-PHC data from 4,380 CU participants across 4 cohorts with APOE genotype, amyloid PET, and cognitive data from four domains of memory, language, executive, and visuospatial function, including a subset of 758 with tau PET imaging. RESULTS:APOE ε4 showed a strong dose-dependent association with amyloid burden and amyloid positivity, with the highest levels observed among ε4 homozygotes. Associations between APOE and global tau burden were more modest and appeared to be driven mainly by ε4 homozygotes, while regional analyses showed localized APOE ε4-related associations in medial temporal regions. Independently, higher tau burden was associated with lower memory and language performance. CONCLUSIONS:In CU older adults, APOE ε4 was most strongly associated with amyloid burden, with more modest associations observed for medial temporal tau burden.

INTRODUCTION:Limbic-predominant age-related transactive response DNA-binding protein 43 (TDP-43) encephalopathy neuropathological change (LATE-NC) is a cause of dementia resembling Alzheimer's disease (AD). The 90+ Study found women using hormone replacement therapy (HRT) two to three decades before death had lower odds of LATE-NC. We attempted to replicate this finding in a different cohort. METHODS:Participants (n = 2056) included males (n = 640) and females (n = 1416) aged ≥65 from the Religious Orders Study and Memory and Aging Project with HRT and neuropathology data. We examined the association between HRT and LATE-NC in males and females and between HRT-related and reproductive variables in relation to LATE-NC in females using logistic regression. RESULTS:HRT use within 5 years before or after menopause (odds ratio [OR] = 0.70, 95% confidence interval [CI] = 0.50 to 0.98, p = 0.03) or for 8 to 16 years (OR = 0.44, 95% CI = 0.24 to 0.79, p = 0.006) was associated with lower odds of LATE-NC. DISCUSSION:This finding identifies a potential factor related to LATE risk and highlights the importance of HRT timing and duration for its potential neuroprotective effects.

Cerebral amyloid angiopathy (CAA) is a cerebrovascular disorder characterized by the deposition of amyloid-β (Aβ) in the walls of leptomeningeal and cortical blood vessels that increases risk of intracerebral hemorrhages and progressive cognitive decline. More than 90

INTRODUCTION:Amyloid positron emission tomography (PET) and cerebrospinal fluid (CSF) biomarkers confirm Alzheimer's disease (AD) pathology but are impractical for large-scale screening. Plasma phosphorylated tau at threonine 217 (p-tau217), subjective cognitive concerns, and computerized cognitive testing are non-invasive, scalable, and feasible to implement in large populations. We assessed their separate and combined predictive value for cognitive decline. METHODS:We analyzed 1064 cognitively unimpaired adults (ages 65-85 years) from the Anti-Amyloid Treatment in Asymptomatic Alzheimer's Disease (A4; amyloid-positive) and Longitudinal Evaluation of Amyloid Risk and Neurodegeneration (LEARN; amyloid-negative) studies. Baseline assessments included apolipoprotein E (APOE) ε4 status, hippocampal volume, amyloid PET, plasma p-tau217, Cognitive Function Index (CFI), and Cogstate Computerized Battery (CCB). Cognitive impairment was defined as conversion from a Clinical Dementia Rating Global Score (CDR-GS) of 0 to ≥0.5 over 240 weeks. RESULTS:During the follow-up, 34.1% developed cognitive impairment. Higher p-tau217, higher CFI, and lower CCB were associated with higher odds of converting to CDR-GS >0 across all cohorts. DISCUSSION:P-tau217, CFI, and CCB each independently predict cognitive decline, offering practical, non-invasive tools for early AD risk stratification and trial enrichment.

INTRODUCTION:Plasma glial fibrillary acidic protein (GFAP), a marker of astrocytic activation, has been linked to Alzheimer's disease; however, its prognostic value in cognitively unimpaired (CU) individuals remains unclear. METHODS:We included 949 CU older adults from the A4 preclinical AD trial and its companion LEARN cohort. Baseline plasma GFAP was measured, and associations with cognitive decline (Preclinical Alzheimer's Cognitive Composite [PACC]), Clinical Dementia Rating (CDR) progression, and imaging biomarkers were assessed over 240 weeks. RESULTS:Baseline plasma GFAP was higher in females and in A4 (amyloid-positive) versus LEARN (amyloid-negative) participants. Cross-sectionally, elevated GFAP was associated with lower cognitive performance and greater amyloid burden. Longitudinally, higher GFAP predicted faster cognitive decline, increased risk of CDR progression, AD-related cortical atrophy, and amyloid conversion, with stronger effects in females. DISCUSSION:Plasma GFAP is a prognostic biomarker in CU older adults, predicting cognitive and biological changes, with stronger associations observed in females, highlighting a possible sex-specific vulnerability. HIGHLIGHTS:Elevated plasma glial fibrillary acidic protein (GFAP) predicted faster cognitive decline measured by Preclinical Alzheimer's Cognitive Composite (PACC). GFAP was associated with increased risk of progression to mild cognitive impairment. GFAP predicted conversion to amyloid positivity in amyloid-negative subjects. Higher baseline GFAP was associated with cortical atrophy in Alzheimer's disease (AD) -signature areas. Associations of GFAP with cognition and AD biomarkers were stronger in females.

BackgroundNeuroinflammation actively contributes to the pathophysiology of Alzheimer's disease (AD); however, the value of neuroinflammatory biomarkers for disease-staging or predicting disease progression remains unclear.ObjectiveTo investigate diagnostic and prognostic utility of inflammatory biomarkers in combination with conventional AD biomarkers.MethodsData from 258 participants in the Alzheimer's Disease Neuroimaging Initiative (ADNI) with cerebrospinal fluid (CSF) biomarkers of amyloid-β (Aβ), tau, and inflammation were analyzed. Clinically meaningful cognitive decline (CMCD) was defined as a ≥ 4-point increase on the Alzheimer's Disease Assessment Scale Cognitive Subscore 11. Predictor variables included demographics (D: age, sex, education), APOE4 status (A), inflammatory biomarkers (I), and classic AD biomarkers of Aβ and p-tau181 (C). Models incorporating inflammatory biomarkers assessed their contribution to improving baseline diagnostic classification and 1-year CMCD prediction.ResultsAt 1-year follow-up, 27.1% of participants experienced CMCD. Adding inflammatory biomarkers to models with D and A variables (DA model) improved classification of cognitively normal (CN) versus mild cognitive impairment (MCI) and CN versus Dementia (p < 0.001). Similarly, inflammatory markers enhanced classification in models including C (DAC model), for CN versus MCI (p < 0.01) and CN versus Dementia (p < 0.001). Predictive performance for CMCD was improved in individuals with MCI and dementia in both models (all p < 0.05). In addition, the DAI model outperformed the DAC model in predicting CMCD for MCI and Dementia groups (both p < 0.05).ConclusionsAddition of CSF inflammatory biomarkers to biomarkers of AD improves diagnostic accuracy of clinical disease stage at baseline and add incremental value to AD biomarkers for prediction of cognitive decline.

Objective:To evaluate the predictive utility of baseline plasma biomarkers and neuropsychological measures in identifying cognitively unimpaired older adults at risk of cognitive and functional decline over five years. Background:The clinical and biological heterogeneity observed in Alzheimer's disease (AD) complicates design of trials and the identification of appropriate participants. Identifying practical tools to define more homogenous subgroups could enhance clinical trial enrichment and improve early detection. Methods:We analyzed data from the Anti-Amyloid Treatment in Asymptomatic Alzheimer's Disease (A4) trial and its companion Evaluation of Amyloid Risk and Neurodegeneration (LEARN) observational study. The sample included 866 cognitively unimpaired, amyloid-positive individuals from the A4 trial (comprising participants randomized to receive Solanezumab or placebo) and 343 cognitively unimpaired, amyloid-negative individuals from LEARN. Cognitive/functional decline was defined as an increase of ≥0.5 in Clinical Dementia Rating-Global Score (CDR-GS) during a 240-week period. Using multiple logistic regression models, we evaluated the predictive value of demographic variables, APOE4 status, amyloid PET SUVR, plasma P-tau217, and Alzheimer's Disease Cooperative Study-Preclinical Alzheimer's Cognitive Composite (ADCS-PACC) in three groups: A4-Solanezumab, A4-placebo and LEARN. In a sub-study including 656 participants with available data, we assessed the incremental value of additional plasma biomarkers (Aβ42/Aβ40 ratio, GFAP, and NfL). Results:Both plasma P-tau217 and ADCS-PACC significantly improved predictive performance over a base model with demographics and APOE4. The full model combining all predictor variables yielded the highest AUCs across A4 Solanezumab (0.80 ± 0.06), A4 Placebo (0.80 ± 0.06), and LEARN (0.78 ± 0.08). Adding other plasma biomarkers yielded small but consistent improvements in AUC (1-3%). Conclusions:Plasma P-tau217 and ADCS-PACC, individually and in combination, improved prediction of cognitive/functional decline in asymptomatic older adults. Predictive models incorporating these scalable and non-invasive measures improved clinical trial enrichment and earlier identification of at-risk individuals preclinical AD.

Limbic-predominant age-related TDP-43 encephalopathy neuropathologic change (LATE-NC) denotes TDP-43 deposition in older age and is consequential for cognitive function. Currently there is no way to identify LATE-NC during life. Some forms of TDP-43 deposition in younger age, related to frontotemporal dementia (FTD), are associated with pronounced asymmetrical atrophy of the temporal lobe. Given the similar underlying proteinopathy of TDP-43 in both LATE-NC and FTD, we hypothesized LATE-NC would be associated with asymmetrical hippocampal atrophy. We included participants from The 90+ Study with both MRI and autopsy data. All participants were assessed for LATE-NC, Alzheimer’s disease neuropathologic change (ADNC), and hippocampal sclerosis of aging (HS-A). 3D-T1w MRIs were segmented using FreeSurfer. We examined hippocampal asymmetry, and minimum and average hippocampal volumes (across left and right), in relation to LATE-NC stages. We defined asymmetry as asym = log(100* abs(left. hipp. vol.-right. hipp.vol)/(ave. hipp. vol)) . We fit multiple linear regression models for each outcome, accounting for HS-A presence and ADNC severity. Lastly, we examined prediction performance using area under the curve (AUC) for LATE-NC stage 2 or higher, considering dementia status alone or including asymmetry, minimum, or average hippocampal volume. All models were adjusted for age at death, sex, education, and intracranial volume. Table-1 displays participant characteristics (N = 104), with N = 37 (36%) having LATE-NC. I ncreasing LATE-NC stage was associated with increasing hippocampal asymmetry and decreasing minimum and average hippocampal volumes ( Figure-1 ). LATE-NC stages 2 and 3 showed significant associations with hippocampal asymmetry ( Figure-2 ). HS-A trended towards an association with, but ADNC was not associated with, hippocampal asymmetry ( Figure-2 ). Both asymmetry and minimum hippocampal volume appeared to be more strongly related to LATE-NC compared to average hippocampal volume ( Figure-2 ). AUC analysis indicated that, for predicting LATE-NC stage 2 or greater, the model which included asymmetry (AUC = 0.79) outperformed both the model with dementia status alone (AUC = 0.64, P = 0.005) and the models with average (AUC = 0.65, P = 0.007) and minimum (AUC = 0.69, P = 0.014) hippocampal volumes. LATE-NC is associated with asymmetrical hippocampal atrophy in a stage-dependent fashion, independent of HS-A and ADNC. Using measures of hippocampal asymmetry may lead to better identification of LATE-NC during life.

Cognitive impairment and dementia in late life are often due to co-occurrence of multiple neuropathologic changes (NC) leading to multiple etiology dementia (MED). Our aim was to assess the contribution of these NCs by first studying the shape of cognitive trajectories in the absence of pathology and then to capture the association between longitudinal change in cognition and multiple common age-related NCs in an oldest-old cohort. 415 participants from The 90+ Study with longitudinal evaluations and autopsy data were included. The following criteria were used for NC presence: moderate or severe likelihood of Alzheimer’s disease NC (ADNC) according to NIA-AA criteria; hippocampal or cortical limbic-predominant age-related TDP-43 encephalopathy (LATE-NC); presence of hippocampal sclerosis; limbic/neocortical Lewy bodies disease (LBD); moderate/severe scoring for cerebral amyloid angiopathy (CAA), atherosclerosis, and arteriolosclerosis, and ≥2 microvascular lesions (MVLs). We considered longitudinal scores of the clinical dementia rating sum of boxes (CDR-SB), global cognition measured by mini mental state examination (MMSE), and composite scores for 6 cognitive domains (Figure 1). We quantified the time-varying relationship between each of clinical and cognitive measures, as outcome, and the neuropathologic changes, as independent variables, using varying-coefficient mixed effects models that can identify nonlinear relationships and account for within-subject repeated measures. The models were adjusted for age at death, sex, and education. Table 1 summarizes the demographic and neuropathologic characteristics of the participants. Mean age at death was 97.2, 68% were female, and 50% were college graduate. Most NCs were significantly more frequent in the dementia group. Figure 1 shows the trajectories of CDR-SB and various cognitive domains in a hypothetical individual with no NCs. Attention and executive function showed early decline while CDR and cognitive domains showed an NC independent decline around five years before death. Figure 2 demonstrates the significant associations of NCs with decline in CDR-SB and cognitive domains. LATE-NC was the strongest predictor of CDR and cognitive outcomes closely followed by ADNC, HS, LBD, and MVL. Our results demonstrate the contribution of various NCs to trajectories of cognitive outcomes in the oldest old and highlight the importance of non-Alzheimer’s NCs in this age group.

As cognitive impairment becomes a growing global concern, the need for early diagnosis has become urgent. Traditional diagnostic methods, such as advanced neuroimaging techniques and cerebrospinal fluid biomarker tests [10] , [11] , are costly and not feasible for large-scale screening. Instead, this study aims to develop a novel, multimodal architecture which analyzes speech samples to earlier detect and accurately classify cognitive impairment [3] – [5] ; we chose to analyze speech as it is non-invasive, easily accessible, and efficient. Linguistic markers, such as grammar errors and vocabulary richness, could serve as indicators of cognitive decline. Beyond linguistic features, acoustic features, such as speech rate, energy, and voice quality, hold promise for detecting abnormal signs in speech related to cognitive decline. This study aims to combine linguistic and acoustic features into a multimodal deep learning model to detect cognitive impairment in the elderly.

The increasing prevalence of cognitive impairment and dementia threatens global health, necessitating the development of accessible tools for detection of cognitive impairment. This study explores using a transformer-based approach to detect cognitive impairment using acoustic markers of spontaneous speech. Recordings of unstructured interviews from baseline visits were obtained from participants of The 90+ Study, a longitudinal study of individuals older than 90 years. For this analysis, participants without dementia at baseline who had stable cognitive diagnoses over the subsequent two assessments were included. Of 120 total participants (Table 1), 78 had normal cognition (NC) and 42 had cognitive impairment with no dementia (CIND). From baseline interviews, 1,273 segmented 20-second denoised audio samples were generated. Samples from 80% of participants trained our models while the remaining 20%, unseen data, were test data. After filtering from Mel Spectrogram, Mel-frequency cepstral coefficients (MFCC), spectral centroid (SC), and Chroma representations, 154 acoustic features were extracted. Features were transformed from 2D images into 1D string representations for interpretation by transformer-based language models, namely Bidirectional Encoder Representations from Transformers (BERT) and Generative Pre-Trained Transformer-2 (GPT-2). Transformer-generated feature embeddings were inputted into Random Forest (RF) and Logistic Regression (LR) models. Soft voting ensemble approach combined classifier predictions to maximize prediction power for final classification (Figure 1). To reduce the risk of overfitting, 5-fold cross-validation determined evaluation metric averages for baseline models, including random forest and logistic regression. Our test data comprised 142, 20-second samples from 16 NC and 113, 20-second samples from 9 CIND. Our model achieved 100% AUC with either BERT or GPT-2 feature embeddings. We also investigated F1 scores: BERT reached 99.56% and GPT-2 achieved 100% (Table 2). Our approach converts acoustic features from 2D images into 1D patterned value representations interpretable by transformer-based language models. This renders acoustic markers suitable for detection of cognitive impairment based on language model interpretation. Both GPT-2 and BERT models outperformed baseline models. The model will be validated with external data and increasing sample size in the next step.

Recent studies in the younger old have identified associations between cognitive performance and markers of inflammation, including interleukin-6 (IL-6) and erythrocyte sedimentation rate (ESR). These associations remain unexplored in the oldest old (age 90+), an age group most vulnerable to dementia. In addition, no studies have examined if amyloid burden impacts these associations. This study aims to: (1) examine the associations between inflammatory markers (IL-6, ESR) and cognitive performance in individuals age 90+ and (2) to examine if amyloid burden impacts these associations. Participants with at least one plasma measure of inflammation and amyloid burden measured by positron emission tomography (PET) were selected from The 90+ Study (n = 112 for IL-6, n = 123 for ESR). Cognitive measures included (1) global cognitive score defined as the average of the standardized scores of Mini-Mental State Examination (MMSE) and Modified MMSE (3MS), and (2) memory score, that was the average of the standardized scores of California Verbal Learning Test (CVLT) and memory items of 3MS. PET amyloid burden was measured by standardized uptake value ratio (SUVR). We examined the associations between IL-6 and ESR (independent variables) with cognitive measures (dependent variable) using linear regression models. Two models were implemented: first, without adjustment and second, with adjustment for amyloid. Both models were also adjusted for age at blood draw, sex, and education (college/no college). We additionally adjusted for amyloid burden and years between blood draw and PET scan in the second model. Mean age at blood draw was 94.7±2.8 and most participants were female (63.4%) and white (92.7%) (Table 1). In model 1, for both IL-6 and ESR, higher levels were significantly associated with lower global cognitive scores (Table 2). Only IL-6 was significantly associated with memory score (Table 2). Amyloid adjustment in model 2 had little effect on our results (Table 2). This study suggests that systemic inflammation could contribute to cognitive performance in individuals aged 90 and older. Additionally, this influence may be linked to a process not attributed to amyloid. Further research is warranted to explore the role of inflammation in non-Alzheimer’s disease dementias.

INTRODUCTION:Alzheimer's disease (AD) has a long preclinical phase in which individuals may accumulate amyloid beta (Aβ) and tau pathology without noticeable cognitive impairment. Subjective cognitive impairment reports can provide early insights into cognitive decline. METHODS:In the A4 Study, 339 cognitively unimpaired, Aβ-positive individuals underwent tau positron emission tomography imaging. Tau status was classified based on medial temporal lobe tau standardized uptake value ratios (tauMTL). Participants and study partners assessed cognitive changes using the 15-item Cognitive Function Index (CFI) questionnaire. We explored the relationship among tauMTL, hippocampal volume (HVa), and CFI reports. RESULTS:Higher tauMTL was associated with participant-reported concerns about memory and navigation, and with study partner-reported difficulty remembering appointments. Lower HVa showed a marginal association with participant-reported driving difficulty. DISCUSSION:These findings support the utility of participant- and study partner-reported concerns as early indicators of preclinical AD pathology, with potential value for early detection and trial enrichment strategies. Highlights:Higher tau in the medial temporal lobe (tauMTL) was linked to participant-reported memory and orientation decline such as needing reminders or getting lost.Higher tauMTL was associated with increased memory-related concerns, such as needing help with appointments and asking repetitive questions.Lower hippocampal volume was associated with spatial memory and navigation such as driving difficulties and greater memory decline as reported by study partners.

Spontaneous speech is easily obtainable and has the potential to become an accessible and low-cost marker for cognitive function. The time-consuming and labor-intensive nature of speech analysis has been a major obstacle to utilizing this promising tool. This study uses a novel transformer-based methodology to explore associations between spontaneous speech language features and global cognition. Speech recordings were obtained from participants with clinical diagnoses of mild cognitive impairment (MCI), dementia, and cognitively unimpaired, from the Alzheimer’s Disease Research Center (ADRC) at University of California, Irvine. Audio samples were denoised and transcribed using our transformer-based model (Figure 1). The model conducted association analyses between global cognition, measured by Montreal Cognitive Assessment (MoCA) scores, and 5 linguistic features pre-specified based on popularity in language analysis. Features were extracted from transcripts via Bidirectional Encoder Representations from Transformers (BERT), which assigns each word a special “token” (tokenizing), translates “tokens” into a numerical format for computer comprehension (encoding), and generates unique identifiers (embeddings) that mathematically capture meanings and relationships between words. A linear regression (LR) model was then trained on power-transformed BERT-extracted linguistic features and its performance was tested against both untransformed and power-transformed data without BERT feature extraction using the same linguistic features. In a cohort comprising 73 healthy controls and 12 individuals with MCI or dementia, our analysis, utilizing pre-specified linguistic features and leveraging BERT for enhancing feature extraction, revealed that vocabulary richness ( p = 0.009), average word length ( p = 0.005), and semantic coherence ( p = 0.047) were significantly associated with MoCA scores (Table 1). The non-BERT untransformed model found no significant associations with MoCA while the non-BERT power-transformed model found only one significant association: average word length ( p = 0.016). Our study's novel approach of employing BERT for linguistic feature extraction in association analysis increased the number of pre-selected speech features significantly associated with global cognition, namely vocabulary richness, average word length, and semantic coherence. This improvement in association detection highlights the potential for better deep-learning dementia detection methods and might lead to increased utility of spontaneous speech as an easily obtainable and scalable cognitive measure.

Limbic-predominant age-related TDP-43 encephalopathy neuropathologic change (LATE-NC) is a common cause of dementia in older age. LATE-NC was first coined in 2019 with proposed staging criteria of TDP-43 progressing from amygdala (stage 1), to hippocampus (stage 2), to middle frontal gyrus (stage 3). Criteria were updated in 2023 to further categorize stage 1 to either TDP-43 inclusions in amygdala alone (stage 1a) or hippocampus alone (stage 1b). We applied this updated LATE-NC staging criteria to data from participants in the National Alzheimer’s Coordinating Center (NACC) and examined associations with clinical diagnosis and other neuropathologic changes (NCs). We selected participants in NACC with regional TDP-43 assessments, excluding those with frontotemporal dementia-related and rare NCs, or with a non-Alzheimer’s disease (AD) etiologic clinical diagnosis. LATE-NC stages were assigned according to updated criteria. We performed logistic regressions with LATE-NC stages as predictors and clinical diagnosis (dementia, cognitive impairment, memory impairment), other neuropathologic changes (ADNC, Lewy bodies, hippocampal sclerosis of aging (HS-A), and vascular NCs), or gross atrophy at autopsy (cortical, hippocampal, frontal/temporal lobar) as outcomes. We also examined the association of LATE-NC stages with dementia and memory impairment while accounting for other NCs. Of N = 1365 participants, LATE-NC was present in 519 (37%): 31% stage 1 (78% 1a, 22% 1b), 56% stage 2, 13% stage 3 (Table-1). Participants with stage 1a were younger at death compared to higher stages. Stage 1a and 1b had similar associations with cognitive problems (Figure-1). However, while other stages were associated with ADNC and Lewy bodies, stage 1b was not. Meanwhile stage 1b was associated with HS-A more strongly than stage 1a and to a similar degree as stage 2. We observed expected associations of higher LATE-NC stages with dementia and other NCs (Figure-1). Lastly, LATE-NC stage 1b was significant (and stage 1a trended) for associations with dementia and memory impairment, while higher LATE-NC stages had associations eclipsed in strength only by highest level ADNC (Figure-2). These findings highlight the utility of the updated LATE-NC staging criteria in general, and stage 1 subtypes in particular, in capturing broad associations with cognitive impairment and specific associations with other neuropathologic changes.

Hippocampal sclerosis of ageing (HS-A)-severe cell loss and gliosis in the hippocampal formation-is a neuropathologic change (NC) that affects up to 20% of elderly persons with dementia. The aetiology of HS-A is heterogeneous, but HS-A is strongly associated with limbic-predominant age-related TDP-43 encephalopathy NC (LATE-NC). Other NCs have also been implicated in relation to HS-A, but these associations have been inconsistent across previous studies. Also, because LATE-NC and HS-A are so strongly associated, it is important to adjust for LATE-NC when examining associations between other NCs and HS-A. The goal of this study was to examine associations of other common NCs with HS-A, both before and after adjusting for LATE-NC. We analysed the National Alzheimer's Coordinating Center (NACC) neuropathology dataset and examined associations of Alzheimer's disease NC (ADNC), Lewy bodies (LB) and cerebrovascular NCs, with HS-A, adjusting for LATE-NC in multiple ways. We used Bayesian multilevel logistic regression models with monotonic modelling for ordinal predictors and report the odds ratios (OR) or average OR across levels (aOR), along with 95% credibility intervals (CI) as well as expected frequencies of HS-A for selected models and predictor levels. Of n = 1933 autopsy participants included (average age at death of 83 years, 51.3% women), HS-A was present in 278 (14.4%). LATE-NC was strongly associated with HS-A (aOR = 3.7, 95% CI = 2.8, 5.0). While ADNC showed a modest association with HS-A in models where LATE-NC was not included as a predictor (aOR = 1.4, CI = 1.1, 1.8), this association was reduced when adjusting for LATE-NC (aOR = 1.11, CI = 0.9, 1.5); results were similar for the ADNC-related A/B/C scores and limbic LBs. However, several cerebrovascular NCs were similarly associated with HS-A both without adjusting for LATE-NC [atherosclerosis aOR = 1.4, arteriolosclerosis aOR = 1.6, white matter rarefaction (WMR) aOR = 1.4] and with adjusting for LATE-NC (atherosclerosis aOR = 1.4, arteriolosclerosis aOR = 1.5, WMR aOR = 1.3). In a combined model, LATE-NC was strongly associated with HS-A, but global cerebrovascular NCs, as well as APOE-ε4 (increased odds) and education (decreased odds), were also associated with HS-A. Predicted HS-A frequency for predictor levels of no LATE-NC or global cerebrovascular NCs was 1.5% (CI = 0.6%, 3.1%), while it was 94.5% (CI = 84%, 99.5%) for LATE-NC stage 3 and severe global cerebrovascular NC levels. LATE-NC is likely the most important cause of HS-A. While ADNC seems to be associated with HS-A through its association with LATE-NC, the association of cerebrovascular NCs with HS-A independent of LATE-NC underlines the importance of vascular factors in the aetiology of HS-A.