Background: The retina is an accessible extension of the central nervous system, yet the protein-coding architecture linking retinal structure, visual function, and major blinding diseases remains poorly defined. Methods: We analyzed whole-exome sequencing data from 356,982 participants and performed exome-wide gene-based tests of rare coding variants and single-variant analyses of common coding variants. Rare-variant findings were further evaluated in an independent All of Us cohort (N = 245,388). We then assessed cross-phenotype pleiotropy, functional validation and biological characterization, clinical relevance, and exploratory analyses of brain-related and systemic traits. Findings: We identified 22 significant rare-variant gene-based associations involving 16 genes, including 12 novel genes, 10 of which were independently supported in the All of Us cohort. Single-variant analyses identified 243 independent common coding variants in 126 genes, including 24 novel genes. CFI, C3, and RIOX1 showed associations across retinal structure, visual function, and disease phenotypes, supporting cross-domain pleiotropy. Among the disease-associated genes, the novel gene FYB2 was selected for experimental validation because it showed expression support in retinal pigment epithelium (RPE)-related contexts and clinical relevance in Cox analyses. FYB2 knockdown aggravated barrier dysfunction in human induced RPE (iRPE) cells, supporting a potential role in diabetic retinopathy. Interpretation: These findings define the protein-coding architecture of retinal phenotypes and support shared genetic links across retinal structure, visual function, and disease. The identified genes provide targets for mechanistic investigation in blinding retinal disorders.

Precision medicine faces a critical challenge in translating high-dimensional omics data into robust disease predictions across diverse populations. Current approaches often fail under distribution shifts, partly due to their inability to encode complex biological feature dependencies. We present OmicFormer, a Transformer-based architecture that embeds two complementary statistical priors, i.e., feature–label associations and feature–feature dependencies, directly into its representation learning. This design captures local and long-range omic interactions often missed by conventional methods. Analyzing 500,000 UK Biobank participants, OmicFormer significantly outperforms strong baselines across 450 disease and 900 trait prediction tasks, with substantial gains spanning diverse metabolic, neurological, cardiovascular, and gastrointestinal conditions, alongside enhanced prediction of circulating metabolites, bone density traits, and retinal imaging biomarkers. Crucially, OmicFormer demonstrates robust generalization, achieving a substantial improvement over tree-based methods in an independent proteomics cohort across 19 diseases (GNPC, N = 7,289), and outperforming tree-based models across 50 multi-site neuroimaging sites (N = 4,728) for autism and schizophrenia classification. By explicitly embedding statistical structure, OmicFormer provides an interpretable and generalizable foundation for omics-based precision medicine.

The clinical utility of plasma phosphorylated tau 217 (p-tau217) levels as a diagnostic tool for cognitive disorders in real-world settings remains uncertain. In this cross-sectional retrospective study, 738 consecutive subjects presenting with cognitive complaints at a memory clinic underwent diagnostic evaluation incorporating plasma p-tau217 alongside standard clinical assessments (“Tier 2”: routine neuroimaging and cognitive testing) and advanced biomarker testing (“Tier 1”: amyloid PET imaging or CSF analysis). Diagnostic impact and confidence were assessed using clinician surveys. Incorporating plasma p-tau217 into the routine “Tier 2” diagnostic workup significantly influenced clinical decision-making, leading to diagnostic revisions in 30.2% of cases and improving overall diagnostic confidence from 65.1% to 79.4%. Plasma p-tau217 demonstrated high diagnostic accuracy for Alzheimer’s disease (AD), with a sensitivity of 0.88 and specificity of 0.84. Notably, its specificity was superior to “Tier 2” diagnoses, whereas its sensitivity did not differ significantly. Plasma p-tau217 also exhibited non-inferior performance to “Tier 1” approaches in the diagnostic distribution of AD, though with marginally lower diagnostic confidence. For non-AD diagnoses, plasma p-tau217 significantly impacted the classification of synucleinopathies and performed comparably to “Tier 1” methods. In conclusion, plasma p-tau217 meaningfully impacted diagnostic precision and confidence in subjects presenting to a real-world memory clinic, offering a minimally invasive, cost-effective, and accessible alternative to state-of-the-art examinations. As the global population ages and cognitive disorders become more prevalent, the integration of plasma p-tau217 into routine practice has the potential to streamline diagnostic workflows and enhance the efficient utilization of medical resources.

INTRODUCTION:Lecanemab, the first disease-modifying therapy for Alzheimer's disease (AD), mitigates pathology primarily by clearing amyloid plaques, but its impact on peripheral immunity remains unclear. METHODS:To assess Lecanemab's immunological effects, we performed antibody array analysis of serum and single-cell RNA sequencing of peripheral blood mononuclear cells collected from healthy controls and patients with AD at baseline, 3, and 6 months post-treatment. RESULTS:Lecanemab restored multiple serum chemokines to healthy levels in patients with AD. Compared to controls, baseline samples from patients with AD showed altered frequencies and functions of naïve and unswitched memory (UswM) B cells. Lecanemab treatment corrected the abnormal naïve and UswM B cell proportions and rebuilt their functional homeostasis by alleviating chronic inflammation and reversing the dysregulation of key pathways including immune response, NF-κB, RAGE, and cell adhesion. DISCUSSION:These findings uncover a novel peripheral immunomodulatory mechanism of Lecanemab, offering new insights into AD therapeutics.

BACKGROUND:Hydromethylthionine mesylate (HMTM) targets tau pathology and also has tau-independent symptomatic activity. A traditional randomised placebo-controlled trial (RCT) was precluded by loss of blinding due to urinary colouration and therapeutic activity at the minimum dose required to maintain blinding. OBJECTIVE:To evaluate the efficacy of HMTM in participants with mild cognitive impairment (MCI) and mild to moderate dementia due to Alzheimer's disease (AD). METHODS:Because a traditional RCT was not feasible without loss of blinding, we compared HMTM 16 mg/day in TRx-237-039 with propensity score matched true placebo controls from the FDA-sponsored Critical Path for AD (CPAD) database with the same inclusion/exclusion criteria (protocol TRx-237-080). We also compared HMTM 16 mg/day with matched natural history controls from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and with a meta-analysis of placebo arms from trials in comparable populations in analyses specified prior to the 104-week database lock of TRx-237-039. PARTICIPANTS:Propensity score matching yielded 127 pairs (HMTM n = 127; CPAD placebo n = 127) in the CPAD comparison, and 189 pairs in the ADNI comparison. A total of 218 receiving HMTM 16 mg/day were compared with meta-analytic controls (n = 1805-8567). INTERVENTION:HMTM 16mg/day MEASUREMENTS: Primary outcomes in TRx-237-080 were change from baseline to 78 weeks in ADAS-Cog13 and whole brain volume (WBV). CDR-Sum of Boxes (CDR-SB) and CDR-Global were analysed at 104 weeks. ADAS-cog11 and WBV were analysed in ADNI comparisons, and ADAS-cog11, ADCS-ADL23, CDR-SB and WBV were analysed in meta-analytic comparisons. RESULTS:Compared with matched CPAD placebo, HMTM 16 mg/day produced statistically significant differences in change on ADAS-Cog13 (p < 0.0001) and WBV at 78 (primary; p < 0.0001) and 104 weeks (p < 0.0001), and CDR-SB differed significantly overall (104-weeks; p < 0.001) and in MCI (p = 0.007). The odds of progressing to a more advanced CDR-Global stage were lower with HMTM (overall OR 0.31) and particularly in MCI (OR 0.15) versus CPAD placebo. Clinical and brain atrophy outcomes were similarly statistically significant in comparisons with ADNI case-matched natural history data and in meta-analytic comparisons. CONCLUSION:Comparisons of HMTM treatment with CPAD, ADNI, and meta-analytic controls provide evidence consistent with clinical benefit HMTM. It has the potential to offer an accessible oral treatment option which could be delivered with minimal patient/physician burden.

​Background: Previous studies have identified a relationship between white matter hyperintensities (WMH) and cardiovascular disease (CVD). However, these studies were often limited by small sample sizes and insufficient adjustment for confounding variables. To address these limitations, this article investigated the association between WMH and CVD in a large-scale, prospective cohort.Methods: 12,912 participants are included from the UK Biobank who were free of CVD at baseline. The cohort had a mean age of 55.79 years, and 52.3% were female (n=6759). Incident CVD was defined using the following ICD 10 (International Classification of Diseases, 10th revision) codes:I20-I25, I50, I60-I64. Measurements of caculated WMH volume from UK biobank were obtained. Cox models, linear regression model and logistic regression model are fit to primary outcome and showed adjusted HR for the log-transformed WMH volume (WMHv) and quatiles of WMHv.Results: During a follow-up of 2.3 years, 809 participants developed CVD events. In comparison with participants in the lowest quartile of WMHv, the risk of incident CVD was distinctly increased in those with higher quartile (HR:1.889, 95% CI: 1.557-2.291, p <0.001). The median WMHv was higher than those without CVD events (7613.48mm3 versus 5013.80 mm3). After adjusting for potential confounders, the highest quartile of WMH volume remained independently associated with an increased risk of incident CVD events (HR 1.77, 95%CI 1.36-2.31, p<0.001).Conclusion: The findings suggest that baseline WMH volume is independently associated with the future risk of CVD in the UK Biobank cohort, supporting its potential utility as a predictor for CVD events

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder marked by cognitive decline, memory loss, and behavioral disturbances, eventually leading to dementia and severely diminishing quality of life. With the global population aging, AD has become an unprecedented challenge for society and families. Recent advances in the development of amyloid-beta (Aβ)-targeting monoclonal anti-bodies, like lecanemab and donanemab, provided hope for slowing or even halting disease progression. However, these treatments have not yet achieved the ultimate goal of reversing cognitive deterioration and restoring normal function. The complexity of AD stems from multiple contributing factors, with Aβ deposition and tau protein tangles being central to its pathology, while genetic predispositions, aging, and systemic factors further drive disease progression. Addressing AD by targeting a single factor has proven insufficient, highlighting the need for a comprehensive understanding of its multifaceted mechanisms. This review explores the latest advances in AD mechanistic research and therapeutic development, focusing on key areas such as amyloid precursor protein (APP) metabolism, Aβ dynamics, Aβ antibody immunotherapy, tau protein dysfunction, genetic influences, aging mechanisms, and systemic factors. By critically examining these aspects, we aim to provide insights that support more holistic approaches to AD diagnosis and treatment, ultimately laying the groundwork for innovative strategies to combat this debilitating disease.

The diagnostic efficacy of tau positron emission tomography (PET) relative to amyloid PET in memory clinic settings remains underexplored. In this study, 1008 consecutive Han Chinese individuals with cognitive impairments were recruited between March 2020 and July 2025. They underwent comprehensive clinical assessment, amyloid PET, and tau PET. Diagnostic reclassification, accuracy, and changes in diagnostic confidence were evaluated across cognitive stages and etiologic categories. For Alzheimer's disease (AD) diagnosis, tau PET demonstrated high sensitivity (97.8%) and specificity (95.4%), showing substantial agreement with amyloid PET. Moreover, tau PET provided additional information beyond that offered by amyloid PET alone. In patients clinically diagnosed with non-AD disorders, tau PET led to a significantly higher rate of diagnostic change compared with amyloid PET (p = 0.018). After a diagnosis based on amyloid PET results, the subsequent addition of tau PET information still resulted in a diagnostic change in 10.0% of patients, highlighting its incremental value. Furthermore, tau PET significantly improved diagnostic confidence (from 68.6% to 84.2%), and the confidence gained from tau PET was significantly greater than that from amyloid PET alone within non-AD diseases. Our findings suggest tau PET offers diagnostic performance for AD comparable to that of amyloid PET, while providing additional value in detecting non-AD tauopathies and clarifying complex cases. These observations support the potential for tau PET to complement existing diagnostic workflows in memory clinic settings.

BACKGROUND AND OBJECTIVES:The 2024 revised criteria introduced an integrated framework for staging Alzheimer disease (AD) across biological and clinical dimensions. However, how this criterion characterizes clinical and biological features in populations outside the original development setting, particularly in non-Western and specialized clinic settings, remains insufficiently described. METHODS:We consecutively enrolled 1,214 memory clinic participants who underwent both amyloid-PET and tau-PET imaging. Among amyloid-positive (A+) individuals, clinical stages (0-6) and biological stages (A-D) were assigned according to the 2024 criteria. Participants were classified as typical (concordant stages), susceptible (clinical > biological), or resilient (clinical < biological). Plasma p-tau217 was measured in 379 A+ participants. Associations were examined using generalized linear models adjusted for relevant covariates, with false discovery rate correction for multiple comparisons. RESULTS:Among the 1,214 participants, 818 were amyloid positive, comprising 412 (50.4%) typical, 330 (40.3%) susceptible, and 76 (9.3%) resilient individuals. Plasma p-tau217 increased progressively across advancing tau PET stages (p for trend <0.001), with significantly higher levels in stage D (1.31 pg/mL) compared with stage A (0.56 pg/mL, q = 0.029) and stage C (0.95 pg/mL, q = 0.013). Compared with the typical group, resilient individuals demonstrated superior cognitive performance (e.g., Mini-Mental State Examination [MMSE]: β = 8.56, q < 0.001), higher educational attainment (>9 years: 74.1% vs 45.1%, q = 0.020), and greater AD-signature regional cortical thickness (t = 3.033, q = 0.005) and volume (t = 3.209, q = 0.003). Conversely, susceptible individuals exhibited inferior cognitive scores (e.g., MMSE: β = -8.29, q < 0.001), reduced cortical thickness (t = -2.872, q = 0.005), and volume (t = -2.751, q = 0.007) in AD-signature regions and a numerically higher burden of vascular risk factors. DISCUSSION:In a large cohort from a specialized tertiary memory clinic, clinical-biological stage discordance under the 2024 AD criteria was common among amyloid-positive individuals. Distinct cognitive, educational, biomarker, and neuroanatomic profiles characterized susceptible, typical, and resilient subgroups. In addition, plasma p-tau217 showed a stepwise increase across tau PET stages, supporting its utility as a blood-based marker of tau pathology severity. These findings support the clinical relevance of the revised staging framework and may inform future refinements of AD diagnostic guidelines. CLASSIFICATION OF EVIDENCE:This study provides Class II evidence that the 2024 revised biological and clinical criteria for AD demonstrate clinical utility in a Chinese cohort from a specialized tertiary memory clinic.

The retina is an accessible extension of the central nervous system, yet the protein-coding architecture linking retinal structure, visual function, and major blinding diseases remains poorly defined. Here, using large-scale whole-exome sequencing data from 356,982 UK Biobank participants, exome-wide gene-based tests of rare coding variants and single-variant analyses of common coding variants are performed. A total of 22 significant rare-variant gene-based associations involving 16 genes are identified, including 12 novel genes, 10 of which are independently supported in the All of Us cohort (N = 245,388). Single-variant analyses identify 243 independent common coding variants in 126 genes, including 24 novel genes. CFI, C3, and RIOX1 show associations across retinal structure, visual function, and disease phenotypes, supporting cross-domain pleiotropy. Among the disease-associated genes, the novel gene FYB2 is prioritized for experimental validation, supported by retinal pigment epithelium (RPE)-related expression evidence and clinical relevance in Cox analyses. FYB2 knockdown aggravates barrier dysfunction in human induced RPE (iRPE) cells, supporting a potential role in diabetic retinopathy. These findings define the protein-coding architecture of retinal phenotypes, and support shared genetic links across retinal structure, visual function, and disease. The identified genes provide candidate targets for mechanistic investigation in blinding retinal disorders.

The relationships between exogenous hormones and dementia, as well as cognitive function in females, remains debated. This study aimed to investigate the associations of exogenous hormone exposure (oral contraceptives [OC] and hormone replacement therapy [HRT]) with incident dementia risk, cognitive function and changes in brain structures. Multivariate Cox proportional hazard regression models were used to assess the associations between exogenous hormone exposure and dementia incidence. Linear regression models were employed to explore the relationships of exogenous hormone exposure and cognitive performances. Mediation models were conducted to explore the underlying mechanisms driven by brain structures. A total of 233,896 female participants from the UK Biobank were included. In fully adjusted models, OC use was associated with reduced risks of all-cause dementia (ACD) (HR [95% CI], 0.806 [0.724-0.897]), Alzheimer's disease (AD) (HR [95%CI], 0.767 [0.659-0.893]) and vascular dementia (VaD) (HR [95%CI], 0.735 [0.578-0.934]). HRT was associated with decreased risks of ACD (HR [95%CI], 0.897 [0.811-0.992]) and AD (HR [95%CI], 0.804 [0.696-0.928]). Duration of OC use showed a non-linear (J-shaped) association with the risks of ACD and AD. In addition, brain structures, including the bilateral pallidum and left thalamus proper were identified as potential mediators in the relationships between the duration of OC use and cognitive performance. To summarize, exogenous hormone use is associated with reduced dementia risk and better cognitive function, with pallidum and thalamus possibly mediating the associations.

Organ-specific aging clocks hold great potential in reflecting organ health. In vivo imaging is inherently organ-specific and delineates structural and functional characteristics more objectively. However, there is no systematic evaluation of imaging-based aging clocks. We utilized 1777 imaging-derived phenotypes (IDPs) from 11,000 healthy participants and assessed the organ-specific biological age of seven organs. The organ-specific age gap was primarily associated with incident diseases and mortality related to corresponding organs. The top-contributing IDPs to organ-specific biological age emerged as biomarkers for incident disease predictions, achieving an area under the curve (AUC) greater than 0.8 for dementia (AUC = 0.82). Subsequent proteomic analysis revealed 966 shared and 507 organ-specific molecular signatures for the aging of different organs. Finally, we identified key modifiable factors and 14 drug targets for organ-specific aging. The imaging-based aging clocks demonstrate organ-specificity at both macro and micro scales, which could promote personalized intervention and treatment of organ aging.

OBJECTIVES:The underlying etiology of epilepsy remains incompletely understood. Hence, a systematic and agnostic mapping of modifiable risk factors is essential to our comprehension and prevention of this complex neurological condition. STUDY DESIGN:Prospective cohort study using data from the UK Biobank. Here, we used data from 401,335 UK Biobank non-epileptic adults, of whom 3,600 developed epilepsy throughout 15.4 years of follow-up. METHODS:An exposure-wide association study (EWAS) was first conducted, significant variables were further combined to generate composite scores for different domains, and joint associations with epilepsy were evaluated. Mendelian randomisation analyses were used to assess causality, and the potential impact of modifying these risk factors on the epilepsy burden was estimated using the population attributable fraction (PAF). RESULTS:First, we conducted EWAS on 196 factors, of these, 27 (13.8%) factors across six domains were significantly associated with epilepsy. Mendelian randomisation analyses further supported a potential causal effect with long-standing disability (βIVW = 0.077, PIVW = 0.008). When considered jointly, we estimated that up to 27.7%-51.5% of epilepsy cases could be prevented. The strongest associations were observed for the absence of sedative-hypnotic or opioid use (PAF: 21.1%/11.0%) across all subgroups, which likely reflected underlying comorbidities such as insomnia or chronic pain. A favourable profile of all domains was associated with a lower epilepsy risk even among participants with high genetic risk. CONCLUSIONS:Our findings underscore the necessity of adopting a comprehensive perspective on the cause of epilepsy. Prioritizing risk factors based on age and genetic risk stratification may helpachieve precise and efficient prevention.

A major challenge in RNA therapeutics for central nervous system disorders is the lack of delivery systems capable of crossing the blood-brain barrier (BBB) while achieving cell-type-specific targeting. Herein, we develop an engineered exosomal siRNA delivery platform for systemic, neuron-targeted RNA transport to the brain. The platform leverages exosomes derived from an immortalized mouse hippocampal neuronal cell line as a biomimetic and functionally privileged material source, enhancing neuronal uptake and intracellular delivery efficiency. Through surface functionalization with a rabies virus glycoprotein-derived peptide, the system enables receptor-mediated BBB transcytosis and programmable siRNA loading. In human cortical organoids, the platform achieves efficient cytosolic delivery and robust gene silencing in neurons, demonstrating high delivery precision and bioavailability. As a proof of concept, targeting receptor-interacting protein kinase 3 (RIPK3) modulates necroptosis, a key pathway in inflammatory neurodegeneration. In transgenic mouse models, systemic administration suppresses RIPK3/MLKL signaling, reduces neuronal loss, and alleviates neuroinflammation and tau-associated pathology. Transcriptomic analyses further indicate stabilization of neuronal homeostasis across vulnerable brain regions. Collectively, the study establishes a modular and programmable exosomal RNA delivery platform and highlights age-defined, cell-derived biomaterials as a generalizable strategy for overcoming delivery barriers in neurological diseases.

Background: Health economic evidence comparing diagnostic strategies for Alzheimer’s disease (AD) in China is lacking. This study aimed to assess the cost-effectiveness of AD cerebrospinal fluid (CSF) biomarkers versus Amyloid β positron emission tomography (Aβ-PET) across the diagnostic and treatment pathway of AD in China. Methods: A decision-tree model was developed to simulate the diagnostic pathway, with an attached Markov model for five-year treatment. Findings: Compared with Aβ-PET, CSF biomarkers achieved similar diagnostic accuracy and quality-adjusted life years at lower costs. From the societal perspective, CSF biomarkers reduced the cost per timely and accurate diagnosis by 34.9% and total costs by $282. From the healthcare system perspective, reductions were 32.9% and $224 respectively. CSF biomarkers was cost-effective from both perspectives. Interpretation: CSF biomarkers is a cost-effective alternative to Aβ-PET for early AD diagnosis in China, providing significant economic and clinical benefits. Its adoption could improve diagnostic access and reduce the overall burden of AD.

As Nature Aging celebrates its fifth anniversary, the journal asks some of the researchers who contributed to the journal early on to reflect on the past and the future of aging and age-related disease research, the impact of the field on human health now and in the future, and what challenges need to be addressed to ensure sustained progress.

Accurate measurement of brain vascular pathology is essential for understanding its role in cognitive aging. Here we classified participants using the amyloid-tau-neurodegeneration framework in a multicenter cohort and identified cerebrospinal fluid brain endothelial-derived small extracellular vesicles (c-BEEVs) as a sensitive biomarker, which correlated with vascular risk factors and the severity of small-vessel disease. c-BEEVs showed high diagnostic performance for vascular cognitive impairment and, when combined with p-tau181, effectively distinguished vascular cognitive impairment from Alzheimer's disease. In individuals with mixed Alzheimer's disease and vascular pathology, c-BEEVs were the earliest indicators of abnormalities. It predicted cognitive decline in participants without p-tau181 pathology. To investigate the mechanistic role of c-BEEVs, we established a hypertension mouse model with elevated c-BEEVs and cognitive deficits. Brain endothelial-specific knockdown of extracellular vesicle secretion alleviated cognitive and synaptic impairment. These findings position c-BEEVs as a promising biomarker for brain vascular pathology and highlight their role in neurovascular dysfunction.

Social isolation and loneliness are increasingly recognized as detrimental risk factors for brain health. Here, utilizing data from 383,421 participants in the UK Biobank, we identify significant associations between social isolation, loneliness, and the incidence of 11 neurological and psychiatric disorders, including major depressive disorder (MDD), schizophrenia, bipolar disorder, anxiety disorders, sleep disorders, dementia, Alzheimer's disease, Parkinson's disease, stroke, multiple sclerosis, and epilepsy employing Cox regression models. Furthermore, using Mendelian randomization analysis we find evidence for putative relationships from social isolation and loneliness to MDD, schizophrenia, sleep disorders, and epilepsy. We also observe significant associations between social isolation, loneliness, and worse cognitive and emotional performance, as well as alterations in brain structures. Additionally, mediation analyses indicate that peripheral inflammatory and biochemical markers partially mediated the links from social isolation and loneliness to neurological and psychiatric disorders.

Population proteomics is emerging as a new framework for equitable precision medicine. By studying protein variation across populations, this field bridges population genomics and conventional proteomics to capture the functional molecular states through which genetic ancestry, environmental exposures and other contextual factors shape human health. Here, we discuss how recent advances are moving population proteomics beyond biomarker discovery toward equitable clinical translation through cross-population validation, mechanistic multiomics and global research infrastructures. Its ultimate promise is not to classify populations as fixed biological categories, but to make human diversity measurable, interpretable and clinically actionable for equitable precision medicine. This Comment discusses how recent advances are moving population proteomics beyond biomarker discovery toward equitable clinical translation through cross-population validation, mechanistic multiomics and global research infrastructures.